]>
Commit | Line | Data |
---|---|---|
2f1a1aea | 1 | // recursive descent parser for systemtap scripts |
e937a1d2 | 2 | // Copyright (C) 2005-2015 Red Hat Inc. |
77a5c1f9 | 3 | // Copyright (C) 2006 Intel Corporation. |
5811366a | 4 | // Copyright (C) 2007 Bull S.A.S |
92585d32 | 5 | // Copyright (C) 2014 Peter Kjellstrom <cap@nsc.liu.se> |
69c68955 FCE |
6 | // |
7 | // This file is part of systemtap, and is free software. You can | |
8 | // redistribute it and/or modify it under the terms of the GNU General | |
9 | // Public License (GPL); either version 2, or (at your option) any | |
10 | // later version. | |
2f1a1aea | 11 | |
2b066ec1 | 12 | #include "config.h" |
2f1a1aea FCE |
13 | #include "staptree.h" |
14 | #include "parse.h" | |
177a8ead | 15 | #include "session.h" |
3f99432c | 16 | #include "util.h" |
aa389a19 | 17 | #include "stringtable.h" |
3f99432c | 18 | |
2b066ec1 | 19 | #include <iostream> |
eacb10ce | 20 | |
2b066ec1 | 21 | #include <fstream> |
2f1a1aea | 22 | #include <cctype> |
9c0c0e46 | 23 | #include <cstdlib> |
29e64872 | 24 | #include <cassert> |
9c0c0e46 FCE |
25 | #include <cerrno> |
26 | #include <climits> | |
57b73400 | 27 | #include <sstream> |
f74fb737 | 28 | #include <cstring> |
3f99432c | 29 | #include <cctype> |
eacb10ce | 30 | #include <iterator> |
5d46f7cb | 31 | #include <unordered_set> |
eacb10ce | 32 | |
7a468d68 FCE |
33 | extern "C" { |
34 | #include <fnmatch.h> | |
35 | } | |
2f1a1aea FCE |
36 | |
37 | using namespace std; | |
38 | ||
c18f07f8 JS |
39 | |
40 | class lexer | |
41 | { | |
42 | public: | |
fee28e5c | 43 | bool ate_comment; // current token follows a comment |
b5477cd9 | 44 | bool ate_whitespace; // the most recent token followed whitespace |
534aad8b | 45 | bool saw_tokens; // the lexer found tokens (before preprocessing occurred) |
f8405ea5 | 46 | bool check_compatible; // whether to gate features on session.compatible |
534aad8b | 47 | |
b5477cd9 | 48 | token* scan (); |
f086fe02 | 49 | lexer (istream&, const string&, systemtap_session&, bool); |
c18f07f8 | 50 | void set_current_file (stapfile* f); |
101b0805 | 51 | void set_current_token_chain (const token* tok); |
f8405ea5 | 52 | inline bool has_version (const char* v) const; |
c18f07f8 | 53 | |
5775f11f | 54 | unordered_set<interned_string> keywords; |
85c97fc2 | 55 | static unordered_set<string> atwords; |
c18f07f8 JS |
56 | private: |
57 | inline int input_get (); | |
58 | inline int input_peek (unsigned n=0); | |
aa389a19 | 59 | void input_put (const string&, const token*); |
c18f07f8 | 60 | string input_name; |
47d349b1 | 61 | string input_contents; // NB: being a temporary, no need to interned_string optimize this object |
aa389a19 | 62 | const char *input_pointer; // index into input_contents; NB: recompute if input_contents changed! |
c18f07f8 | 63 | const char *input_end; |
aa389a19 FCE |
64 | unsigned cursor_suspend_count; |
65 | unsigned cursor_suspend_line; | |
66 | unsigned cursor_suspend_column; | |
c18f07f8 JS |
67 | unsigned cursor_line; |
68 | unsigned cursor_column; | |
69 | systemtap_session& session; | |
70 | stapfile* current_file; | |
101b0805 | 71 | const token* current_token_chain; |
c18f07f8 JS |
72 | }; |
73 | ||
74 | ||
75 | class parser | |
76 | { | |
77 | public: | |
f8405ea5 | 78 | parser (systemtap_session& s, const string& n, istream& i, unsigned flags=0); |
c18f07f8 JS |
79 | ~parser (); |
80 | ||
f8405ea5 JS |
81 | stapfile* parse (); |
82 | probe* parse_synthetic_probe (const token* chain); | |
83 | stapfile* parse_library_macros (); | |
c18f07f8 JS |
84 | |
85 | private: | |
86 | typedef enum { | |
87 | PP_NONE, | |
88 | PP_KEEP_THEN, | |
89 | PP_SKIP_THEN, | |
90 | PP_KEEP_ELSE, | |
91 | PP_SKIP_ELSE, | |
92 | } pp_state_t; | |
93 | ||
534aad8b SM |
94 | struct pp1_activation; |
95 | ||
fe410f52 SM |
96 | struct pp_macrodecl : public macrodecl { |
97 | pp1_activation* parent_act; // used for param bindings | |
98 | virtual bool is_closure() { return parent_act != 0; } | |
99 | pp_macrodecl () : macrodecl(), parent_act(0) { } | |
534aad8b SM |
100 | }; |
101 | ||
c18f07f8 JS |
102 | systemtap_session& session; |
103 | string input_name; | |
c18f07f8 | 104 | lexer input; |
f8405ea5 | 105 | bool errs_as_warnings; |
c18f07f8 | 106 | bool privileged; |
7b5b30a8 | 107 | bool user_file; |
c18f07f8 JS |
108 | parse_context context; |
109 | ||
534aad8b SM |
110 | // preprocessing subordinate, first pass (macros) |
111 | struct pp1_activation { | |
112 | const token* tok; | |
113 | unsigned cursor; // position within macro body | |
114 | map<string, pp_macrodecl*> params; | |
534aad8b | 115 | |
fe410f52 | 116 | macrodecl* curr_macro; |
534aad8b | 117 | |
bdf7707b JS |
118 | pp1_activation (const token* tok, macrodecl* curr_macro) |
119 | : tok(tok), cursor(0), curr_macro(curr_macro) { } | |
534aad8b SM |
120 | ~pp1_activation (); |
121 | }; | |
122 | ||
fe410f52 | 123 | map<string, macrodecl*> pp1_namespace; |
534aad8b SM |
124 | vector<pp1_activation*> pp1_state; |
125 | const token* next_pp1 (); | |
ed891cf3 | 126 | const token* scan_pp1 (bool ignore_macros); |
534aad8b SM |
127 | const token* slurp_pp1_param (vector<const token*>& param); |
128 | const token* slurp_pp1_body (vector<const token*>& body); | |
129 | ||
130 | // preprocessing subordinate, final pass (conditionals) | |
c18f07f8 | 131 | vector<pair<const token*, pp_state_t> > pp_state; |
b5477cd9 | 132 | const token* scan_pp (); |
c18f07f8 JS |
133 | const token* skip_pp (); |
134 | ||
135 | // scanning state | |
b5477cd9 SM |
136 | const token* next (); |
137 | const token* peek (); | |
c18f07f8 | 138 | |
731a5359 MW |
139 | // Advance past and throw away current token after peek () or next (). |
140 | void swallow (); | |
141 | ||
a07a2c28 | 142 | const token* systemtap_v_seen; |
c18f07f8 JS |
143 | const token* last_t; // the last value returned by peek() or next() |
144 | const token* next_t; // lookahead token | |
145 | ||
731a5359 MW |
146 | // expectations, these swallow the token |
147 | void expect_known (token_type tt, string const & expected); | |
47d349b1 | 148 | void expect_unknown (token_type tt, interned_string & target); |
47d349b1 | 149 | void expect_unknown2 (token_type tt1, token_type tt2, interned_string & target); |
731a5359 MW |
150 | |
151 | // convenience forms, these also swallow the token | |
152 | void expect_op (string const & expected); | |
891b96e6 | 153 | interned_string expect_op_any (initializer_list<const char*> expected); |
731a5359 MW |
154 | void expect_kw (string const & expected); |
155 | void expect_number (int64_t & expected); | |
b1f2b0e8 | 156 | void expect_ident_or_keyword (interned_string & target); |
731a5359 MW |
157 | |
158 | // convenience forms, which return true or false, these don't swallow token | |
c18f07f8 JS |
159 | bool peek_op (string const & op); |
160 | bool peek_kw (string const & kw); | |
161 | ||
731a5359 MW |
162 | // convenience forms, which return the token |
163 | const token* expect_kw_token (string const & expected); | |
b1f2b0e8 | 164 | const token* expect_ident_or_atword (interned_string & target); |
731a5359 | 165 | |
7ac01ea0 | 166 | void print_error (const parse_error& pe, bool errs_as_warnings = false); |
c18f07f8 JS |
167 | unsigned num_errors; |
168 | ||
169 | private: // nonterminals | |
170 | void parse_probe (vector<probe*>&, vector<probe_alias*>&); | |
f41e297c JS |
171 | void parse_private (vector<vardecl*>&, vector<probe*>&, |
172 | string const&, vector<functiondecl*>&); | |
173 | void parse_global (vector<vardecl*>&, vector<probe*>&, | |
174 | string const&); | |
175 | void do_parse_global (vector<vardecl*>&, vector<probe*>&, | |
176 | string const&, const token*, bool); | |
177 | void parse_functiondecl (vector<functiondecl*>&, string const&); | |
178 | void do_parse_functiondecl (vector<functiondecl*>&, const token*, | |
179 | string const&, bool); | |
c18f07f8 | 180 | embeddedcode* parse_embeddedcode (); |
380d759b FL |
181 | vector<probe_point*> parse_probe_points (); |
182 | vector<probe_point*> parse_components (); | |
183 | vector<probe_point*> parse_component (); | |
d24f1ff4 SM |
184 | literal_string* consume_string_literals (const token*); |
185 | literal_string* parse_literal_string (); | |
c18f07f8 JS |
186 | literal* parse_literal (); |
187 | block* parse_stmt_block (); | |
188 | try_block* parse_try_block (); | |
189 | statement* parse_statement (); | |
190 | if_statement* parse_if_statement (); | |
191 | for_loop* parse_for_loop (); | |
192 | for_loop* parse_while_loop (); | |
193 | foreach_loop* parse_foreach_loop (); | |
194 | expr_statement* parse_expr_statement (); | |
195 | return_statement* parse_return_statement (); | |
196 | delete_statement* parse_delete_statement (); | |
197 | next_statement* parse_next_statement (); | |
198 | break_statement* parse_break_statement (); | |
199 | continue_statement* parse_continue_statement (); | |
200 | indexable* parse_indexable (); | |
b1f2b0e8 | 201 | const token *parse_hist_op_or_bare_name (hist_op *&hop, interned_string &name); |
0fb0cac9 JS |
202 | target_symbol *parse_target_symbol (); |
203 | cast_op *parse_cast_op (); | |
204 | atvar_op *parse_atvar_op (); | |
8cc799a5 | 205 | expression* parse_entry_op (const token* t); |
c18f07f8 | 206 | expression* parse_defined_op (const token* t); |
0a7eb12d | 207 | expression* parse_const_op (const token* t); |
3689db05 | 208 | expression* parse_perf_op (const token* t); |
c18f07f8 JS |
209 | expression* parse_expression (); |
210 | expression* parse_assignment (); | |
211 | expression* parse_ternary (); | |
212 | expression* parse_logical_or (); | |
213 | expression* parse_logical_and (); | |
214 | expression* parse_boolean_or (); | |
215 | expression* parse_boolean_xor (); | |
216 | expression* parse_boolean_and (); | |
217 | expression* parse_array_in (); | |
93daaca8 | 218 | expression* parse_comparison_or_regex_query (); |
c18f07f8 JS |
219 | expression* parse_shift (); |
220 | expression* parse_concatenation (); | |
221 | expression* parse_additive (); | |
222 | expression* parse_multiplicative (); | |
223 | expression* parse_unary (); | |
224 | expression* parse_crement (); | |
0fb0cac9 | 225 | expression* parse_dwarf_value (); |
c18f07f8 JS |
226 | expression* parse_value (); |
227 | expression* parse_symbol (); | |
228 | ||
0fb0cac9 | 229 | bool peek_target_symbol_components (); |
c18f07f8 JS |
230 | void parse_target_symbol_components (target_symbol* e); |
231 | }; | |
232 | ||
233 | ||
2f1a1aea FCE |
234 | // ------------------------------------------------------------------------ |
235 | ||
c18f07f8 | 236 | stapfile* |
ba48c27a | 237 | parse (systemtap_session& s, const string& n, istream& i, unsigned flags) |
c18f07f8 | 238 | { |
ba48c27a | 239 | parser p (s, n, i, flags); |
f8405ea5 | 240 | return p.parse (); |
c18f07f8 JS |
241 | } |
242 | ||
c18f07f8 | 243 | stapfile* |
f8405ea5 | 244 | parse (systemtap_session& s, const string& name, unsigned flags) |
c18f07f8 | 245 | { |
4cd32d8c JS |
246 | ifstream i(name.c_str(), ios::in); |
247 | if (i.fail()) | |
248 | { | |
249 | cerr << (file_exists(name) | |
250 | ? _F("Input file '%s' can't be opened for reading.", name.c_str()) | |
251 | : _F("Input file '%s' is missing.", name.c_str())) | |
252 | << endl; | |
253 | return 0; | |
254 | } | |
255 | ||
f8405ea5 JS |
256 | parser p (s, name, i, flags); |
257 | return p.parse (); | |
c18f07f8 JS |
258 | } |
259 | ||
fe410f52 | 260 | stapfile* |
f8405ea5 | 261 | parse_library_macros (systemtap_session& s, const string& name) |
fe410f52 SM |
262 | { |
263 | ifstream i(name.c_str(), ios::in); | |
264 | if (i.fail()) | |
265 | { | |
266 | cerr << (file_exists(name) | |
267 | ? _F("Input file '%s' can't be opened for reading.", name.c_str()) | |
268 | : _F("Input file '%s' is missing.", name.c_str())) | |
269 | << endl; | |
270 | return 0; | |
271 | } | |
272 | ||
f8405ea5 JS |
273 | parser p (s, name, i); |
274 | return p.parse_library_macros (); | |
fe410f52 SM |
275 | } |
276 | ||
101b0805 | 277 | probe* |
aa389a19 | 278 | parse_synthetic_probe (systemtap_session &s, istream& i, const token* tok) |
101b0805 | 279 | { |
d026d78c | 280 | parser p (s, tok ? tok->location.file->name : "<synthetic>", i); |
f8405ea5 | 281 | return p.parse_synthetic_probe (tok); |
101b0805 JS |
282 | } |
283 | ||
c18f07f8 | 284 | // ------------------------------------------------------------------------ |
bb2e3076 | 285 | |
f8405ea5 | 286 | parser::parser (systemtap_session& s, const string &n, istream& i, unsigned flags): |
f086fe02 | 287 | session (s), input_name (n), input (i, input_name, s, !(flags & pf_no_compatible)), |
f8405ea5 | 288 | errs_as_warnings(flags & pf_squash_errors), privileged (flags & pf_guru), |
7b5b30a8 FL |
289 | user_file (flags & pf_user_file), context(con_unknown), systemtap_v_seen(0), |
290 | last_t (0), next_t (0), num_errors (0) | |
4cd32d8c JS |
291 | { |
292 | } | |
2f1a1aea FCE |
293 | |
294 | parser::~parser() | |
295 | { | |
2f1a1aea FCE |
296 | } |
297 | ||
d7f3e0c5 GH |
298 | static string |
299 | tt2str(token_type tt) | |
300 | { | |
301 | switch (tt) | |
302 | { | |
303 | case tok_junk: return "junk"; | |
304 | case tok_identifier: return "identifier"; | |
305 | case tok_operator: return "operator"; | |
306 | case tok_string: return "string"; | |
307 | case tok_number: return "number"; | |
308 | case tok_embedded: return "embedded-code"; | |
6e213f58 | 309 | case tok_keyword: return "keyword"; |
d7f3e0c5 GH |
310 | } |
311 | return "unknown token"; | |
312 | } | |
82919855 | 313 | |
0323ed4d WC |
314 | ostream& |
315 | operator << (ostream& o, const source_loc& loc) | |
316 | { | |
a704a23b | 317 | o << loc.file->name << ":" |
0323ed4d WC |
318 | << loc.line << ":" |
319 | << loc.column; | |
320 | ||
321 | return o; | |
322 | } | |
323 | ||
56099f08 FCE |
324 | ostream& |
325 | operator << (ostream& o, const token& t) | |
326 | { | |
d7f3e0c5 | 327 | o << tt2str(t.type); |
56099f08 | 328 | |
6e213f58 | 329 | if (t.type != tok_embedded && t.type != tok_keyword) // XXX: other types? |
56099f08 | 330 | { |
24cb178f FCE |
331 | o << " '"; |
332 | for (unsigned i=0; i<t.content.length(); i++) | |
333 | { | |
334 | char c = t.content[i]; | |
335 | o << (isprint (c) ? c : '?'); | |
336 | } | |
337 | o << "'"; | |
56099f08 | 338 | } |
56099f08 | 339 | |
dff50e09 | 340 | o << " at " |
0323ed4d | 341 | << t.location; |
56099f08 FCE |
342 | |
343 | return o; | |
344 | } | |
345 | ||
346 | ||
dff50e09 | 347 | void |
7ac01ea0 | 348 | parser::print_error (const parse_error &pe, bool errs_as_warnings) |
2f1a1aea | 349 | { |
16fc963f | 350 | const token *tok = pe.tok ? pe.tok : last_t; |
7ac01ea0 | 351 | session.print_error(pe, tok, input_name, errs_as_warnings); |
2f1a1aea FCE |
352 | num_errors ++; |
353 | } | |
354 | ||
355 | ||
2f1a1aea | 356 | |
c434ec7e FCE |
357 | |
358 | template <typename OPERAND> | |
359 | bool eval_comparison (const OPERAND& lhs, const token* op, const OPERAND& rhs) | |
360 | { | |
361 | if (op->type == tok_operator && op->content == "<=") | |
362 | { return lhs <= rhs; } | |
363 | else if (op->type == tok_operator && op->content == ">=") | |
364 | { return lhs >= rhs; } | |
365 | else if (op->type == tok_operator && op->content == "<") | |
366 | { return lhs < rhs; } | |
367 | else if (op->type == tok_operator && op->content == ">") | |
368 | { return lhs > rhs; } | |
369 | else if (op->type == tok_operator && op->content == "==") | |
370 | { return lhs == rhs; } | |
371 | else if (op->type == tok_operator && op->content == "!=") | |
372 | { return lhs != rhs; } | |
373 | else | |
f0454224 | 374 | throw PARSE_ERROR (_("expected comparison operator"), op); |
c434ec7e FCE |
375 | } |
376 | ||
377 | ||
534aad8b SM |
378 | // Here, we perform on-the-fly preprocessing in two passes. |
379 | ||
380 | // First pass - macro declaration and expansion. | |
381 | // | |
382 | // The basic form of a declaration is @define SIGNATURE %( BODY %) | |
383 | // where SIGNATURE is of the form macro_name (a, b, c, ...) | |
384 | // and BODY can obtain the parameter contents as @a, @b, @c, .... | |
385 | // Note that parameterless macros can also be declared. | |
386 | // | |
3932c705 | 387 | // Macro definitions may not be nested. |
534aad8b SM |
388 | // A macro is available textually after it has been defined. |
389 | // | |
390 | // The basic form of a macro invocation | |
391 | // for a parameterless macro is @macro_name, | |
392 | // for a macro with parameters is @macro_name(param_1, param_2, ...). | |
393 | // | |
26718dbe SM |
394 | // NB: this means that a parameterless macro @foo called as @foo(a, b, c) |
395 | // leaves its 'parameters' alone, rather than consuming them to result | |
396 | // in a "too many parameters error". This may be useful in the unusual | |
397 | // case of wanting @foo to expand to the name of a function. | |
534aad8b SM |
398 | // |
399 | // Invocations of unknown macros are left unexpanded, to allow | |
400 | // the continued use of constructs such as @cast, @var, etc. | |
401 | ||
fe410f52 | 402 | macrodecl::~macrodecl () |
534aad8b SM |
403 | { |
404 | delete tok; | |
405 | for (vector<const token*>::iterator it = body.begin(); | |
406 | it != body.end(); it++) | |
407 | delete *it; | |
408 | } | |
409 | ||
410 | parser::pp1_activation::~pp1_activation () | |
411 | { | |
412 | delete tok; | |
fe410f52 | 413 | if (curr_macro->is_closure()) return; // body is shared with an earlier declaration |
534aad8b SM |
414 | for (map<string, pp_macrodecl*>::iterator it = params.begin(); |
415 | it != params.end(); it++) | |
416 | delete it->second; | |
417 | } | |
418 | ||
419 | // Grab a token from the current input source (main file or macro body): | |
420 | const token* | |
421 | parser::next_pp1 () | |
422 | { | |
423 | if (pp1_state.empty()) | |
424 | return input.scan (); | |
425 | ||
426 | // otherwise, we're inside a macro | |
427 | pp1_activation* act = pp1_state.back(); | |
428 | unsigned& cursor = act->cursor; | |
429 | if (cursor < act->curr_macro->body.size()) | |
430 | { | |
431 | token* t = new token(*act->curr_macro->body[cursor]); | |
0002fc51 | 432 | t->chain = new token(*act->tok); // mark chained token |
534aad8b SM |
433 | cursor++; |
434 | return t; | |
435 | } | |
436 | else | |
437 | return 0; // reached end of macro body | |
438 | } | |
439 | ||
440 | const token* | |
ed891cf3 | 441 | parser::scan_pp1 (bool ignore_macros = false) |
534aad8b SM |
442 | { |
443 | while (true) | |
444 | { | |
445 | const token* t = next_pp1 (); | |
446 | if (t == 0) // EOF or end of macro body | |
447 | { | |
448 | if (pp1_state.empty()) // actual EOF | |
449 | return 0; | |
450 | ||
451 | // Exit macro and loop around to look for the next token. | |
452 | pp1_activation* act = pp1_state.back(); | |
453 | pp1_state.pop_back(); delete act; | |
454 | continue; | |
455 | } | |
456 | ||
457 | // macro definition | |
ed891cf3 MC |
458 | // PR18462 don't catalog preprocessor-disabled macros |
459 | if (t->type == tok_operator && t->content == "@define" && !ignore_macros) | |
534aad8b SM |
460 | { |
461 | if (!pp1_state.empty()) | |
f0454224 | 462 | throw PARSE_ERROR (_("'@define' forbidden inside macro body"), t); |
534aad8b SM |
463 | delete t; |
464 | ||
465 | // handle macro definition | |
466 | // (1) consume macro signature | |
3932c705 | 467 | t = input.scan(); |
534aad8b | 468 | if (! (t && t->type == tok_identifier)) |
f0454224 | 469 | throw PARSE_ERROR (_("expected identifier"), t); |
47d349b1 | 470 | string name = t->content; |
534aad8b SM |
471 | |
472 | // check for redefinition of existing macro | |
473 | if (pp1_namespace.find(name) != pp1_namespace.end()) | |
78ab2de3 | 474 | { |
f0454224 | 475 | parse_error er (ERR_SRC, _F("attempt to redefine macro '@%s' in the same file", name.c_str ()), t); |
78ab2de3 SM |
476 | |
477 | // Also point to pp1_namespace[name]->tok, the site of | |
478 | // the original definition: | |
f0454224 | 479 | er.chain = new PARSE_ERROR (_F("macro '@%s' first defined here", |
78ab2de3 SM |
480 | name.c_str()), pp1_namespace[name]->tok); |
481 | throw er; | |
482 | } | |
483 | ||
484 | // XXX: the above restriction was mostly necessary due to | |
485 | // wanting to leave open the possibility of | |
486 | // statically-scoped semantics in the future. | |
26718dbe SM |
487 | |
488 | // XXX: this cascades into further parse errors as the | |
489 | // parser tries to parse the remaining definition... (e.g. | |
490 | // it can't tell that the macro body isn't a conditional, | |
491 | // that the uses of parameters aren't nonexistent | |
492 | // macros.....) | |
534aad8b | 493 | if (name == "define") |
f0454224 | 494 | throw PARSE_ERROR (_("attempt to redefine '@define'"), t); |
17f56d2a | 495 | if (input.atwords.count(name)) |
534aad8b SM |
496 | session.print_warning (_F("macro redefines built-in operator '@%s'", name.c_str()), t); |
497 | ||
fe410f52 SM |
498 | macrodecl* decl = (pp1_namespace[name] = new macrodecl); |
499 | decl->tok = t; | |
534aad8b SM |
500 | |
501 | // determine if the macro takes parameters | |
4ac28d7e | 502 | bool saw_params = false; |
3932c705 SM |
503 | t = input.scan(); |
504 | if (t && t->type == tok_operator && t->content == "(") | |
4ac28d7e SM |
505 | { |
506 | saw_params = true; | |
507 | do | |
508 | { | |
509 | delete t; | |
510 | ||
511 | t = input.scan (); | |
512 | if (! (t && t->type == tok_identifier)) | |
f0454224 | 513 | throw PARSE_ERROR(_("expected identifier"), t); |
47d349b1 | 514 | decl->formal_args.push_back(t->content); |
4ac28d7e SM |
515 | delete t; |
516 | ||
517 | t = input.scan (); | |
518 | if (t && t->type == tok_operator && t->content == ",") | |
519 | { | |
520 | continue; | |
521 | } | |
522 | else if (t && t->type == tok_operator && t->content == ")") | |
523 | { | |
524 | delete t; | |
525 | t = input.scan(); | |
526 | break; | |
527 | } | |
528 | else | |
529 | { | |
f0454224 | 530 | throw PARSE_ERROR (_("expected ',' or ')'"), t); |
4ac28d7e SM |
531 | } |
532 | } | |
533 | while (true); | |
534 | } | |
534aad8b SM |
535 | |
536 | // (2) identify & consume macro body | |
3932c705 | 537 | if (! (t && t->type == tok_operator && t->content == "%(")) |
4ac28d7e SM |
538 | { |
539 | if (saw_params) | |
f0454224 | 540 | throw PARSE_ERROR (_("expected '%('"), t); |
4ac28d7e | 541 | else |
f0454224 | 542 | throw PARSE_ERROR (_("expected '%(' or '('"), t); |
4ac28d7e | 543 | } |
3932c705 | 544 | delete t; |
534aad8b | 545 | |
3932c705 SM |
546 | t = slurp_pp1_body (decl->body); |
547 | if (!t) | |
f0454224 | 548 | throw PARSE_ERROR (_("incomplete macro definition - missing '%)'"), decl->tok); |
3932c705 | 549 | delete t; |
534aad8b SM |
550 | |
551 | // Now loop around to look for a real token. | |
552 | continue; | |
553 | } | |
554 | ||
555 | // (potential) macro invocation | |
556 | if (t->type == tok_operator && t->content[0] == '@') | |
557 | { | |
7371cd19 | 558 | const string& name = t->content.substr(1); // strip initial '@' |
534aad8b SM |
559 | |
560 | // check if name refers to a real parameter or macro | |
fe410f52 | 561 | macrodecl* decl; |
534aad8b SM |
562 | pp1_activation* act = pp1_state.empty() ? 0 : pp1_state.back(); |
563 | if (act && act->params.find(name) != act->params.end()) | |
564 | decl = act->params[name]; | |
fe410f52 SM |
565 | else if (!(act && act->curr_macro->context == ctx_library) |
566 | && pp1_namespace.find(name) != pp1_namespace.end()) | |
534aad8b | 567 | decl = pp1_namespace[name]; |
fe410f52 SM |
568 | else if (session.library_macros.find(name) |
569 | != session.library_macros.end()) | |
570 | decl = session.library_macros[name]; | |
534aad8b SM |
571 | else // this is an ordinary @operator |
572 | return t; | |
573 | ||
bdf7707b JS |
574 | // handle macro invocation, taking ownership of t |
575 | pp1_activation *new_act = new pp1_activation(t, decl); | |
fe410f52 | 576 | unsigned num_params = decl->formal_args.size(); |
534aad8b SM |
577 | |
578 | // (1a) restore parameter invocation closure | |
fe410f52 | 579 | if (num_params == 0 && decl->is_closure()) |
534aad8b SM |
580 | { |
581 | // NB: decl->parent_act is always safe since the | |
582 | // parameter decl (if any) comes from an activation | |
583 | // record which deeper in the stack than new_act. | |
584 | ||
585 | // decl is a macro parameter which must be evaluated in | |
586 | // the context of the original point of invocation: | |
fe410f52 | 587 | new_act->params = ((pp_macrodecl*)decl)->parent_act->params; |
534aad8b SM |
588 | goto expand; |
589 | } | |
590 | ||
591 | // (1b) consume macro parameters (if any) | |
592 | if (num_params == 0) | |
593 | goto expand; | |
594 | ||
595 | // for simplicity, we do not allow macro constructs here | |
596 | // -- if we did, we'd have to recursively call scan_pp1() | |
597 | t = next_pp1 (); | |
39566df2 | 598 | if (! (t && t->type == tok_operator && t->content == "(")) |
534aad8b SM |
599 | { |
600 | delete new_act; | |
f0454224 | 601 | throw PARSE_ERROR (_NF |
534aad8b | 602 | ("expected '(' in invocation of macro '@%s'" |
f499dee5 | 603 | " taking %d parameter", |
534aad8b | 604 | "expected '(' in invocation of macro '@%s'" |
f499dee5 | 605 | " taking %d parameters", |
52c2652f | 606 | num_params, name.c_str(), num_params), t); |
534aad8b SM |
607 | } |
608 | ||
609 | // XXX perhaps parse/count the full number of params, | |
610 | // so we can say "expected x, found y params" on error? | |
611 | for (unsigned i = 0; i < num_params; i++) | |
612 | { | |
613 | delete t; | |
614 | ||
615 | // create parameter closure | |
fe410f52 | 616 | string param_name = decl->formal_args[i]; |
534aad8b | 617 | pp_macrodecl* p = (new_act->params[param_name] |
fe410f52 SM |
618 | = new pp_macrodecl); |
619 | p->tok = new token(*new_act->tok); | |
620 | p->parent_act = act; | |
534aad8b SM |
621 | // NB: *new_act->tok points to invocation, act is NULL at top level |
622 | ||
623 | t = slurp_pp1_param (p->body); | |
624 | ||
625 | // check correct usage of ',' or ')' | |
626 | if (t == 0) // hit unexpected EOF or end of macro | |
627 | { | |
628 | // XXX could we pop the stack and continue parsing | |
629 | // the invocation, allowing macros to construct new | |
630 | // invocations in piecemeal fashion?? | |
631 | const token* orig_t = new token(*new_act->tok); | |
632 | delete new_act; | |
f0454224 | 633 | throw PARSE_ERROR (_("could not find end of macro invocation"), orig_t); |
534aad8b SM |
634 | } |
635 | if (t->type == tok_operator && t->content == ",") | |
636 | { | |
637 | if (i + 1 == num_params) | |
638 | { | |
639 | delete new_act; | |
f0454224 | 640 | throw PARSE_ERROR (_F("too many parameters for macro '@%s' (expected %d)", name.c_str(), num_params), t); |
534aad8b SM |
641 | } |
642 | } | |
643 | else if (t->type == tok_operator && t->content == ")") | |
644 | { | |
645 | if (i + 1 != num_params) | |
646 | { | |
647 | delete new_act; | |
f0454224 | 648 | throw PARSE_ERROR (_F("too few parameters for macro '@%s' (expected %d)", name.c_str(), num_params), t); |
534aad8b SM |
649 | } |
650 | } | |
651 | else | |
652 | { | |
653 | // XXX this is, incidentally, impossible | |
654 | delete new_act; | |
f0454224 | 655 | throw PARSE_ERROR(_("expected ',' or ')' after macro parameter"), t); |
534aad8b SM |
656 | } |
657 | } | |
658 | ||
659 | delete t; | |
660 | ||
661 | // (2) set up macro expansion | |
662 | expand: | |
663 | pp1_state.push_back (new_act); | |
664 | ||
665 | // Now loop around to look for a real token. | |
666 | continue; | |
667 | } | |
668 | ||
669 | // Otherwise, we have an ordinary token. | |
670 | return t; | |
671 | } | |
672 | } | |
673 | ||
36fa6eb3 | 674 | // Consume a single macro invocation's parameters, heeding nesting |
534aad8b SM |
675 | // brackets and stopping on an unbalanced ')' or an unbracketed ',' |
676 | // (and returning the final separator token). | |
677 | const token* | |
678 | parser::slurp_pp1_param (vector<const token*>& param) | |
679 | { | |
680 | const token* t = 0; | |
681 | unsigned nesting = 0; | |
682 | do | |
683 | { | |
684 | t = next_pp1 (); | |
685 | ||
686 | if (!t) | |
687 | break; | |
36fa6eb3 FCE |
688 | // [ needed in case macro paramater is used as prefix for array-deref operation |
689 | if (t->type == tok_operator && (t->content == "(" || t->content == "[")) | |
534aad8b | 690 | ++nesting; |
36fa6eb3 | 691 | else if (nesting && t->type == tok_operator && (t->content == ")" || t->content == "]")) |
534aad8b SM |
692 | --nesting; |
693 | else if (!nesting && t->type == tok_operator | |
694 | && (t->content == ")" || t->content == ",")) | |
695 | break; | |
696 | param.push_back(t); | |
697 | } | |
698 | while (true); | |
699 | return t; // report ")" or "," or NULL | |
700 | } | |
701 | ||
702 | ||
703 | // Consume a macro declaration's body, heeding nested %( %) brackets. | |
704 | const token* | |
705 | parser::slurp_pp1_body (vector<const token*>& body) | |
706 | { | |
707 | const token* t = 0; | |
708 | unsigned nesting = 0; | |
709 | do | |
710 | { | |
711 | t = next_pp1 (); | |
712 | ||
713 | if (!t) | |
714 | break; | |
715 | if (t->type == tok_operator && t->content == "%(") | |
716 | ++nesting; | |
717 | else if (nesting && t->type == tok_operator && t->content == "%)") | |
718 | --nesting; | |
719 | else if (!nesting && t->type == tok_operator && t->content == "%)") | |
720 | break; | |
721 | body.push_back(t); | |
722 | } | |
723 | while (true); | |
724 | return t; // report final "%)" or NULL | |
725 | } | |
726 | ||
fe410f52 SM |
727 | // Used for parsing .stpm files. |
728 | stapfile* | |
f8405ea5 | 729 | parser::parse_library_macros () |
fe410f52 SM |
730 | { |
731 | stapfile* f = new stapfile; | |
732 | input.set_current_file (f); | |
733 | ||
734 | try | |
735 | { | |
ed891cf3 | 736 | const token* t = scan_pp (); |
fe410f52 SM |
737 | |
738 | // Currently we only take objection to macro invocations if they | |
739 | // produce a non-whitespace token after being expanded. | |
740 | ||
741 | // XXX should we prevent macro invocations even if they expand to empty?? | |
742 | ||
743 | if (t != 0) | |
ed891cf3 | 744 | throw PARSE_ERROR (_F("unexpected token in library macro file '%s'", input_name.c_str()), t); |
fe410f52 SM |
745 | |
746 | // We need to first check whether *any* of the macros are duplicates, | |
747 | // then commit to including the entire file in the global namespace | |
748 | // (or not). Yuck. | |
749 | for (map<string, macrodecl*>::iterator it = pp1_namespace.begin(); | |
750 | it != pp1_namespace.end(); it++) | |
751 | { | |
752 | string name = it->first; | |
753 | ||
754 | if (session.library_macros.find(name) != session.library_macros.end()) | |
755 | { | |
f0454224 JL |
756 | parse_error er(ERR_SRC, _F("duplicate definition of library macro '@%s'", name.c_str()), it->second->tok); |
757 | er.chain = new PARSE_ERROR (_F("macro '@%s' first defined here", name.c_str()), session.library_macros[name]->tok); | |
78ab2de3 | 758 | print_error (er); |
fe410f52 | 759 | |
78ab2de3 | 760 | delete er.chain; |
fe410f52 SM |
761 | delete f; |
762 | return 0; | |
763 | } | |
764 | } | |
765 | ||
766 | } | |
767 | catch (const parse_error& pe) | |
768 | { | |
7ac01ea0 | 769 | print_error (pe, errs_as_warnings); |
fe410f52 SM |
770 | delete f; |
771 | return 0; | |
772 | } | |
773 | ||
774 | // If no errors, include the entire file. Note how this is outside | |
775 | // of the try-catch block -- no errors possible. | |
776 | for (map<string, macrodecl*>::iterator it = pp1_namespace.begin(); | |
777 | it != pp1_namespace.end(); it++) | |
778 | { | |
779 | string name = it->first; | |
780 | ||
781 | session.library_macros[name] = it->second; | |
782 | session.library_macros[name]->context = ctx_library; | |
fe410f52 SM |
783 | } |
784 | ||
785 | return f; | |
786 | } | |
787 | ||
534aad8b SM |
788 | // Second pass - preprocessor conditional expansion. |
789 | // | |
177a8ead | 790 | // The basic form is %( CONDITION %? THEN-TOKENS %: ELSE-TOKENS %) |
44ce8ed5 FCE |
791 | // where CONDITION is: kernel_v[r] COMPARISON-OP "version-string" |
792 | // or: arch COMPARISON-OP "arch-string" | |
db135493 | 793 | // or: systemtap_v COMPARISON-OP "version-string" |
2e6dd9d0 | 794 | // or: systemtap_privilege COMPARISON-OP "privilege-string" |
561079c8 | 795 | // or: CONFIG_foo COMPARISON-OP "config-string" |
717a457b | 796 | // or: CONFIG_foo COMPARISON-OP number |
4227f98d | 797 | // or: CONFIG_foo COMPARISON-OP CONFIG_bar |
5811366a FCE |
798 | // or: "string1" COMPARISON-OP "string2" |
799 | // or: number1 COMPARISON-OP number2 | |
44ce8ed5 | 800 | // The %: ELSE-TOKENS part is optional. |
177a8ead FCE |
801 | // |
802 | // e.g. %( kernel_v > "2.5" %? "foo" %: "baz" %) | |
c434ec7e | 803 | // e.g. %( arch != "i?86" %? "foo" %: "baz" %) |
561079c8 | 804 | // e.g. %( CONFIG_foo %? "foo" %: "baz" %) |
177a8ead FCE |
805 | // |
806 | // Up to an entire %( ... %) expression is processed by a single call | |
807 | // to this function. Tokens included by any nested conditions are | |
808 | // enqueued in a private vector. | |
809 | ||
810 | bool eval_pp_conditional (systemtap_session& s, | |
811 | const token* l, const token* op, const token* r) | |
812 | { | |
44ce8ed5 | 813 | if (l->type == tok_identifier && (l->content == "kernel_v" || |
db135493 FCE |
814 | l->content == "kernel_vr" || |
815 | l->content == "systemtap_v")) | |
44ce8ed5 | 816 | { |
db135493 | 817 | if (! (r->type == tok_string)) |
f0454224 | 818 | throw PARSE_ERROR (_("expected string literal"), r); |
db135493 | 819 | |
44ce8ed5 | 820 | string target_kernel_vr = s.kernel_release; |
197a4d62 | 821 | string target_kernel_v = s.kernel_base_release; |
db135493 | 822 | string target; |
dff50e09 | 823 | |
db135493 FCE |
824 | if (l->content == "kernel_v") target = target_kernel_v; |
825 | else if (l->content == "kernel_vr") target = target_kernel_vr; | |
826 | else if (l->content == "systemtap_v") target = s.compatible; | |
827 | else assert (0); | |
7a468d68 | 828 | |
47d349b1 | 829 | string query = r->content; |
7a468d68 FCE |
830 | bool rhs_wildcard = (strpbrk (query.c_str(), "*?[") != 0); |
831 | ||
44ce8ed5 FCE |
832 | // collect acceptable strverscmp results. |
833 | int rvc_ok1, rvc_ok2; | |
7a468d68 | 834 | bool wc_ok = false; |
44ce8ed5 FCE |
835 | if (op->type == tok_operator && op->content == "<=") |
836 | { rvc_ok1 = -1; rvc_ok2 = 0; } | |
837 | else if (op->type == tok_operator && op->content == ">=") | |
838 | { rvc_ok1 = 1; rvc_ok2 = 0; } | |
839 | else if (op->type == tok_operator && op->content == "<") | |
840 | { rvc_ok1 = -1; rvc_ok2 = -1; } | |
841 | else if (op->type == tok_operator && op->content == ">") | |
842 | { rvc_ok1 = 1; rvc_ok2 = 1; } | |
843 | else if (op->type == tok_operator && op->content == "==") | |
7a468d68 | 844 | { rvc_ok1 = 0; rvc_ok2 = 0; wc_ok = true; } |
44ce8ed5 | 845 | else if (op->type == tok_operator && op->content == "!=") |
7a468d68 | 846 | { rvc_ok1 = -1; rvc_ok2 = 1; wc_ok = true; } |
44ce8ed5 | 847 | else |
f0454224 | 848 | throw PARSE_ERROR (_("expected comparison operator"), op); |
7a468d68 FCE |
849 | |
850 | if ((!wc_ok) && rhs_wildcard) | |
f0454224 | 851 | throw PARSE_ERROR (_("wildcard not allowed with order comparison operators"), op); |
7a468d68 FCE |
852 | |
853 | if (rhs_wildcard) | |
854 | { | |
855 | int rvc_result = fnmatch (query.c_str(), target.c_str(), | |
856 | FNM_NOESCAPE); // spooky | |
857 | bool badness = (rvc_result == 0) ^ (op->content == "=="); | |
858 | return !badness; | |
859 | } | |
860 | else | |
861 | { | |
862 | int rvc_result = strverscmp (target.c_str(), query.c_str()); | |
863 | // normalize rvc_result | |
864 | if (rvc_result < 0) rvc_result = -1; | |
865 | if (rvc_result > 0) rvc_result = 1; | |
866 | return (rvc_result == rvc_ok1 || rvc_result == rvc_ok2); | |
867 | } | |
44ce8ed5 | 868 | } |
2e6dd9d0 SM |
869 | else if (l->type == tok_identifier && l->content == "systemtap_privilege") |
870 | { | |
871 | string target_privilege = | |
cba5b802 SM |
872 | pr_contains(s.privilege, pr_stapdev) ? "stapdev" |
873 | : pr_contains(s.privilege, pr_stapsys) ? "stapsys" | |
874 | : pr_contains(s.privilege, pr_stapusr) ? "stapusr" | |
2e6dd9d0 SM |
875 | : "none"; /* should be impossible -- s.privilege always one of above */ |
876 | assert(target_privilege != "none"); | |
877 | ||
878 | if (! (r->type == tok_string)) | |
f0454224 | 879 | throw PARSE_ERROR (_("expected string literal"), r); |
47d349b1 | 880 | string query_privilege = r->content; |
2e6dd9d0 SM |
881 | |
882 | bool nomatch = (target_privilege != query_privilege); | |
883 | ||
884 | bool result; | |
885 | if (op->type == tok_operator && op->content == "==") | |
886 | result = !nomatch; | |
887 | else if (op->type == tok_operator && op->content == "!=") | |
888 | result = nomatch; | |
889 | else | |
f0454224 | 890 | throw PARSE_ERROR (_("expected '==' or '!='"), op); |
cba5b802 | 891 | /* XXX perhaps allow <= >= and similar comparisons */ |
2e6dd9d0 SM |
892 | |
893 | return result; | |
894 | } | |
92585d32 PK |
895 | else if (l->type == tok_identifier && l->content == "guru_mode") |
896 | { | |
897 | if (! (r->type == tok_number)) | |
898 | throw PARSE_ERROR (_("expected number"), r); | |
899 | int64_t lhs = (int64_t) s.guru_mode; | |
47d349b1 | 900 | int64_t rhs = lex_cast<int64_t>(r->content); |
92585d32 PK |
901 | if (!((rhs == 0)||(rhs == 1))) |
902 | throw PARSE_ERROR (_("expected 0 or 1"), op); | |
903 | if (!((op->type == tok_operator && op->content == "==") || | |
904 | (op->type == tok_operator && op->content == "!="))) | |
905 | throw PARSE_ERROR (_("expected '==' or '!='"), op); | |
906 | ||
907 | return eval_comparison (lhs, op, rhs); | |
908 | } | |
44ce8ed5 FCE |
909 | else if (l->type == tok_identifier && l->content == "arch") |
910 | { | |
911 | string target_architecture = s.architecture; | |
912 | if (! (r->type == tok_string)) | |
f0454224 | 913 | throw PARSE_ERROR (_("expected string literal"), r); |
47d349b1 | 914 | string query_architecture = r->content; |
dff50e09 | 915 | |
7a468d68 FCE |
916 | int nomatch = fnmatch (query_architecture.c_str(), |
917 | target_architecture.c_str(), | |
918 | FNM_NOESCAPE); // still spooky | |
919 | ||
561079c8 FCE |
920 | bool result; |
921 | if (op->type == tok_operator && op->content == "==") | |
922 | result = !nomatch; | |
923 | else if (op->type == tok_operator && op->content == "!=") | |
924 | result = nomatch; | |
925 | else | |
f0454224 | 926 | throw PARSE_ERROR (_("expected '==' or '!='"), op); |
561079c8 | 927 | |
d9677d7b DS |
928 | return result; |
929 | } | |
930 | else if (l->type == tok_identifier && l->content == "runtime") | |
931 | { | |
932 | if (! (r->type == tok_string)) | |
f0454224 | 933 | throw PARSE_ERROR (_("expected string literal"), r); |
d9677d7b | 934 | |
47d349b1 | 935 | string query_runtime = r->content; |
d9677d7b DS |
936 | string target_runtime; |
937 | ||
938 | target_runtime = (s.runtime_mode == systemtap_session::dyninst_runtime | |
939 | ? "dyninst" : "kernel"); | |
940 | int nomatch = fnmatch (query_runtime.c_str(), | |
941 | target_runtime.c_str(), | |
942 | FNM_NOESCAPE); // still spooky | |
943 | ||
944 | bool result; | |
945 | if (op->type == tok_operator && op->content == "==") | |
946 | result = !nomatch; | |
947 | else if (op->type == tok_operator && op->content == "!=") | |
948 | result = nomatch; | |
949 | else | |
f0454224 | 950 | throw PARSE_ERROR (_("expected '==' or '!='"), op); |
d9677d7b | 951 | |
561079c8 FCE |
952 | return result; |
953 | } | |
f5a34c5a | 954 | else if (l->type == tok_identifier && l->content.starts_with("CONFIG_")) |
561079c8 | 955 | { |
717a457b MW |
956 | if (r->type == tok_string) |
957 | { | |
47d349b1 FCE |
958 | string lhs = s.kernel_config[l->content]; // may be empty |
959 | string rhs = r->content; | |
561079c8 | 960 | |
717a457b | 961 | int nomatch = fnmatch (rhs.c_str(), lhs.c_str(), FNM_NOESCAPE); // still spooky |
561079c8 | 962 | |
717a457b MW |
963 | bool result; |
964 | if (op->type == tok_operator && op->content == "==") | |
965 | result = !nomatch; | |
966 | else if (op->type == tok_operator && op->content == "!=") | |
967 | result = nomatch; | |
968 | else | |
f0454224 | 969 | throw PARSE_ERROR (_("expected '==' or '!='"), op); |
dff50e09 | 970 | |
717a457b MW |
971 | return result; |
972 | } | |
973 | else if (r->type == tok_number) | |
974 | { | |
7371cd19 JS |
975 | const string& lhs_string = s.kernel_config[l->content]; |
976 | const char* startp = lhs_string.c_str (); | |
717a457b MW |
977 | char* endp = (char*) startp; |
978 | errno = 0; | |
979 | int64_t lhs = (int64_t) strtoll (startp, & endp, 0); | |
980 | if (errno == ERANGE || errno == EINVAL || *endp != '\0') | |
f0454224 | 981 | throw PARSE_ERROR ("Config option value not a number", l); |
717a457b | 982 | |
47d349b1 | 983 | int64_t rhs = lex_cast<int64_t>(r->content); |
717a457b MW |
984 | return eval_comparison (lhs, op, rhs); |
985 | } | |
4227f98d | 986 | else if (r->type == tok_identifier |
f5a34c5a | 987 | && r->content.starts_with( "CONFIG_")) |
4227f98d MW |
988 | { |
989 | // First try to convert both to numbers, | |
990 | // otherwise threat both as strings. | |
7371cd19 JS |
991 | const string& lhs_string = s.kernel_config[l->content]; |
992 | const string& rhs_string = s.kernel_config[r->content]; | |
993 | const char* startp = lhs_string.c_str (); | |
4227f98d MW |
994 | char* endp = (char*) startp; |
995 | errno = 0; | |
996 | int64_t val = (int64_t) strtoll (startp, & endp, 0); | |
997 | if (errno != ERANGE && errno != EINVAL && *endp == '\0') | |
998 | { | |
999 | int64_t lhs = val; | |
7371cd19 | 1000 | startp = rhs_string.c_str (); |
4227f98d MW |
1001 | endp = (char*) startp; |
1002 | errno = 0; | |
1003 | int64_t rhs = (int64_t) strtoll (startp, & endp, 0); | |
1004 | if (errno != ERANGE && errno != EINVAL && *endp == '\0') | |
1005 | return eval_comparison (lhs, op, rhs); | |
1006 | } | |
1007 | ||
7371cd19 | 1008 | return eval_comparison (lhs_string, op, rhs_string); |
4227f98d | 1009 | } |
717a457b | 1010 | else |
f0454224 | 1011 | throw PARSE_ERROR (_("expected string, number literal or other CONFIG_... as right side operand"), r); |
dff50e09 | 1012 | } |
c434ec7e | 1013 | else if (l->type == tok_string && r->type == tok_string) |
5811366a | 1014 | { |
47d349b1 FCE |
1015 | string lhs = l->content; |
1016 | string rhs = r->content; | |
c434ec7e FCE |
1017 | return eval_comparison (lhs, op, rhs); |
1018 | // NB: no wildcarding option here | |
1019 | } | |
1020 | else if (l->type == tok_number && r->type == tok_number) | |
1021 | { | |
47d349b1 FCE |
1022 | int64_t lhs = lex_cast<int64_t>(l->content); |
1023 | int64_t rhs = lex_cast<int64_t>(r->content); | |
c434ec7e | 1024 | return eval_comparison (lhs, op, rhs); |
7a468d68 | 1025 | // NB: no wildcarding option here |
5811366a FCE |
1026 | } |
1027 | else if (l->type == tok_string && r->type == tok_number | |
1028 | && op->type == tok_operator) | |
f0454224 | 1029 | throw PARSE_ERROR (_("expected string literal as right value"), r); |
5811366a FCE |
1030 | else if (l->type == tok_number && r->type == tok_string |
1031 | && op->type == tok_operator) | |
f0454224 | 1032 | throw PARSE_ERROR (_("expected number literal as right value"), r); |
c434ec7e | 1033 | |
177a8ead | 1034 | else |
f0454224 | 1035 | throw PARSE_ERROR (_("expected 'arch', 'kernel_v', 'kernel_vr', 'systemtap_v',\n" |
d9677d7b DS |
1036 | " 'runtime', 'systemtap_privilege', 'CONFIG_...', or\n" |
1037 | " comparison between strings or integers"), l); | |
177a8ead FCE |
1038 | } |
1039 | ||
1040 | ||
5811366a | 1041 | // Only tokens corresponding to the TRUE statement must be expanded |
177a8ead | 1042 | const token* |
b5477cd9 | 1043 | parser::scan_pp () |
177a8ead FCE |
1044 | { |
1045 | while (true) | |
1046 | { | |
e92f2566 JS |
1047 | pp_state_t pp = PP_NONE; |
1048 | if (!pp_state.empty()) | |
1049 | pp = pp_state.back().second; | |
1050 | ||
1051 | const token* t = 0; | |
1052 | if (pp == PP_SKIP_THEN || pp == PP_SKIP_ELSE) | |
1053 | t = skip_pp (); | |
1054 | else | |
534aad8b | 1055 | t = scan_pp1 (); |
e92f2566 JS |
1056 | |
1057 | if (t == 0) // EOF | |
177a8ead | 1058 | { |
e92f2566 JS |
1059 | if (pp != PP_NONE) |
1060 | { | |
1061 | t = pp_state.back().first; | |
1062 | pp_state.pop_back(); // so skip_some doesn't keep trying to close this | |
ce0f6648 | 1063 | //TRANSLATORS: 'conditional' meaning 'conditional preprocessing' |
f0454224 | 1064 | throw PARSE_ERROR (_("incomplete conditional at end of file"), t); |
e92f2566 | 1065 | } |
177a8ead FCE |
1066 | return t; |
1067 | } | |
1068 | ||
e92f2566 JS |
1069 | // misplaced preprocessor "then" |
1070 | if (t->type == tok_operator && t->content == "%?") | |
f0454224 | 1071 | throw PARSE_ERROR (_("incomplete conditional - missing '%('"), t); |
e92f2566 JS |
1072 | |
1073 | // preprocessor "else" | |
1074 | if (t->type == tok_operator && t->content == "%:") | |
1075 | { | |
1076 | if (pp == PP_NONE) | |
f0454224 | 1077 | throw PARSE_ERROR (_("incomplete conditional - missing '%('"), t); |
e92f2566 | 1078 | if (pp == PP_KEEP_ELSE || pp == PP_SKIP_ELSE) |
f0454224 | 1079 | throw PARSE_ERROR (_("invalid conditional - duplicate '%:'"), t); |
1d94e4e5 SM |
1080 | // XXX: here and elsewhere, error cascades might be avoided |
1081 | // by dropping tokens until we reach the closing %) | |
e92f2566 JS |
1082 | |
1083 | pp_state.back().second = (pp == PP_KEEP_THEN) ? | |
1084 | PP_SKIP_ELSE : PP_KEEP_ELSE; | |
1085 | delete t; | |
1086 | continue; | |
1087 | } | |
1088 | ||
1089 | // preprocessor close | |
1090 | if (t->type == tok_operator && t->content == "%)") | |
1091 | { | |
1092 | if (pp == PP_NONE) | |
f0454224 | 1093 | throw PARSE_ERROR (_("incomplete conditional - missing '%('"), t); |
e92f2566 | 1094 | delete pp_state.back().first; |
a07a2c28 | 1095 | delete t; //this is the closing bracket |
e92f2566 JS |
1096 | pp_state.pop_back(); |
1097 | continue; | |
1098 | } | |
dff50e09 | 1099 | |
177a8ead FCE |
1100 | if (! (t->type == tok_operator && t->content == "%(")) // ordinary token |
1101 | return t; | |
1102 | ||
1103 | // We have a %( - it's time to throw a preprocessing party! | |
1104 | ||
2d7881bf PP |
1105 | bool result = false; |
1106 | bool and_result = true; | |
1107 | const token *n = NULL; | |
1108 | do { | |
1109 | const token *l, *op, *r; | |
534aad8b SM |
1110 | l = scan_pp1 (); |
1111 | op = scan_pp1 (); | |
1112 | r = scan_pp1 (); | |
2d7881bf | 1113 | if (l == 0 || op == 0 || r == 0) |
f0454224 | 1114 | throw PARSE_ERROR (_("incomplete condition after '%('"), t); |
2d7881bf PP |
1115 | // NB: consider generalizing to consume all tokens until %?, and |
1116 | // passing that as a vector to an evaluator. | |
1117 | ||
1118 | // Do not evaluate the condition if we haven't expanded everything. | |
1119 | // This may occur when having several recursive conditionals. | |
1120 | and_result &= eval_pp_conditional (session, l, op, r); | |
a07a2c28 LB |
1121 | if(l->content=="systemtap_v") |
1122 | systemtap_v_seen=r; | |
1123 | ||
1124 | else | |
1125 | delete r; | |
1126 | ||
2d7881bf PP |
1127 | delete l; |
1128 | delete op; | |
2d7881bf PP |
1129 | delete n; |
1130 | ||
534aad8b | 1131 | n = scan_pp1 (); |
2d7881bf PP |
1132 | if (n && n->type == tok_operator && n->content == "&&") |
1133 | continue; | |
1134 | result |= and_result; | |
1135 | and_result = true; | |
1136 | if (! (n && n->type == tok_operator && n->content == "||")) | |
1137 | break; | |
1138 | } while (true); | |
3f847830 FCE |
1139 | |
1140 | /* | |
1141 | clog << "PP eval (" << *t << ") == " << result << endl; | |
1142 | */ | |
1143 | ||
e92f2566 | 1144 | const token *m = n; |
177a8ead | 1145 | if (! (m && m->type == tok_operator && m->content == "%?")) |
f0454224 | 1146 | throw PARSE_ERROR (_("expected '%?' marker for conditional"), t); |
70c743d8 | 1147 | delete m; // "%?" |
177a8ead | 1148 | |
e92f2566 JS |
1149 | pp = result ? PP_KEEP_THEN : PP_SKIP_THEN; |
1150 | pp_state.push_back (make_pair (t, pp)); | |
3f847830 | 1151 | |
e92f2566 JS |
1152 | // Now loop around to look for a real token. |
1153 | } | |
1154 | } | |
3f847830 | 1155 | |
3f847830 | 1156 | |
e92f2566 JS |
1157 | // Skip over tokens and any errors, heeding |
1158 | // only nested preprocessor starts and ends. | |
1159 | const token* | |
1160 | parser::skip_pp () | |
1161 | { | |
1162 | const token* t = 0; | |
1163 | unsigned nesting = 0; | |
1164 | do | |
1165 | { | |
1166 | try | |
1167 | { | |
ed891cf3 | 1168 | t = scan_pp1 (true); |
177a8ead | 1169 | } |
e92f2566 | 1170 | catch (const parse_error &e) |
70c743d8 | 1171 | { |
e92f2566 | 1172 | continue; |
70c743d8 | 1173 | } |
e92f2566 JS |
1174 | if (!t) |
1175 | break; | |
1176 | if (t->type == tok_operator && t->content == "%(") | |
1177 | ++nesting; | |
1178 | else if (nesting && t->type == tok_operator && t->content == "%)") | |
1179 | --nesting; | |
1180 | else if (!nesting && t->type == tok_operator && | |
1181 | (t->content == "%:" || t->content == "%?" || t->content == "%)")) | |
1182 | break; | |
1183 | delete t; | |
177a8ead | 1184 | } |
e92f2566 JS |
1185 | while (true); |
1186 | return t; | |
177a8ead FCE |
1187 | } |
1188 | ||
1189 | ||
2f1a1aea | 1190 | const token* |
b5477cd9 | 1191 | parser::next () |
2f1a1aea FCE |
1192 | { |
1193 | if (! next_t) | |
b5477cd9 | 1194 | next_t = scan_pp (); |
2f1a1aea | 1195 | if (! next_t) |
f0454224 | 1196 | throw PARSE_ERROR (_("unexpected end-of-file")); |
2f1a1aea | 1197 | |
2f1a1aea FCE |
1198 | last_t = next_t; |
1199 | // advance by zeroing next_t | |
1200 | next_t = 0; | |
1201 | return last_t; | |
1202 | } | |
1203 | ||
1204 | ||
1205 | const token* | |
b5477cd9 | 1206 | parser::peek () |
2f1a1aea FCE |
1207 | { |
1208 | if (! next_t) | |
b5477cd9 | 1209 | next_t = scan_pp (); |
2f1a1aea FCE |
1210 | |
1211 | // don't advance by zeroing next_t | |
1212 | last_t = next_t; | |
1213 | return next_t; | |
1214 | } | |
1215 | ||
1216 | ||
731a5359 MW |
1217 | void |
1218 | parser::swallow () | |
1219 | { | |
1220 | // can only swallow something last peeked or nexted token. | |
1221 | assert (last_t != 0); | |
1222 | delete last_t; | |
1223 | // advance by zeroing next_t | |
1224 | last_t = next_t = 0; | |
1225 | } | |
1226 | ||
1227 | ||
d7f3e0c5 GH |
1228 | static inline bool |
1229 | tok_is(token const * t, token_type tt, string const & expected) | |
1230 | { | |
1231 | return t && t->type == tt && t->content == expected; | |
1232 | } | |
1233 | ||
1234 | ||
731a5359 | 1235 | void |
d7f3e0c5 GH |
1236 | parser::expect_known (token_type tt, string const & expected) |
1237 | { | |
1238 | const token *t = next(); | |
57b73400 | 1239 | if (! (t && t->type == tt && t->content == expected)) |
f0454224 | 1240 | throw PARSE_ERROR (_F("expected '%s'", expected.c_str())); |
731a5359 | 1241 | swallow (); // We are done with it, content was copied. |
d7f3e0c5 GH |
1242 | } |
1243 | ||
1244 | ||
a3e980f9 | 1245 | void |
47d349b1 | 1246 | parser::expect_unknown (token_type tt, interned_string & target) |
a3e980f9 FCE |
1247 | { |
1248 | const token *t = next(); | |
1249 | if (!(t && t->type == tt)) | |
1250 | throw PARSE_ERROR (_("expected ") + tt2str(tt)); | |
1251 | target = t->content; | |
1252 | swallow (); // We are done with it, content was copied. | |
1253 | } | |
1254 | ||
d7f3e0c5 | 1255 | |
a3e980f9 | 1256 | void |
47d349b1 | 1257 | parser::expect_unknown2 (token_type tt1, token_type tt2, interned_string & target) |
a3e980f9 FCE |
1258 | { |
1259 | const token *t = next(); | |
1260 | if (!(t && (t->type == tt1 || t->type == tt2))) | |
1261 | throw PARSE_ERROR (_F("expected %s or %s", tt2str(tt1).c_str(), tt2str(tt2).c_str())); | |
1262 | target = t->content; | |
1263 | swallow (); // We are done with it, content was copied. | |
1264 | } | |
1265 | ||
493ee224 | 1266 | |
731a5359 | 1267 | void |
aa389a19 | 1268 | parser::expect_op (string const & expected) |
d7f3e0c5 | 1269 | { |
731a5359 | 1270 | expect_known (tok_operator, expected); |
d7f3e0c5 GH |
1271 | } |
1272 | ||
1efdc9a9 | 1273 | interned_string |
891b96e6 | 1274 | parser::expect_op_any (initializer_list<const char*> expected) |
1efdc9a9 FL |
1275 | { |
1276 | const token *t = next(); | |
891b96e6 JS |
1277 | if (t && t->type == tok_operator) |
1278 | for (auto it = expected.begin(); it != expected.end(); ++it) | |
1279 | if (t->content == *it) | |
1efdc9a9 | 1280 | { |
891b96e6 JS |
1281 | interned_string found = t->content; |
1282 | swallow (); // We are done with it, content was copied. | |
1283 | return found; | |
1efdc9a9 | 1284 | } |
891b96e6 JS |
1285 | |
1286 | string msg; | |
1287 | for (auto it = expected.begin(); it != expected.end(); ++it) | |
1288 | { | |
1289 | if (it != expected.begin()) | |
1efdc9a9 FL |
1290 | msg.append(" "); |
1291 | msg.append(*it); | |
1292 | } | |
891b96e6 | 1293 | throw PARSE_ERROR (_F("expected one of '%s'", msg.c_str())); |
1efdc9a9 | 1294 | } |
d7f3e0c5 | 1295 | |
731a5359 | 1296 | void |
aa389a19 | 1297 | parser::expect_kw (string const & expected) |
d7f3e0c5 | 1298 | { |
731a5359 | 1299 | expect_known (tok_keyword, expected); |
d7f3e0c5 GH |
1300 | } |
1301 | ||
dff50e09 | 1302 | const token* |
aa389a19 | 1303 | parser::expect_kw_token (string const & expected) |
731a5359 MW |
1304 | { |
1305 | const token *t = next(); | |
1306 | if (! (t && t->type == tok_keyword && t->content == expected)) | |
f0454224 | 1307 | throw PARSE_ERROR (_F("expected '%s'", expected.c_str())); |
731a5359 MW |
1308 | return t; |
1309 | } | |
1310 | ||
1311 | void | |
e38723d2 | 1312 | parser::expect_number (int64_t & value) |
57b73400 | 1313 | { |
e38723d2 MH |
1314 | bool neg = false; |
1315 | const token *t = next(); | |
1316 | if (t->type == tok_operator && t->content == "-") | |
1317 | { | |
1318 | neg = true; | |
731a5359 | 1319 | swallow (); |
e38723d2 MH |
1320 | t = next (); |
1321 | } | |
1322 | if (!(t && t->type == tok_number)) | |
f0454224 | 1323 | throw PARSE_ERROR (_("expected number")); |
e38723d2 | 1324 | |
7371cd19 JS |
1325 | const string& s = t->content; |
1326 | const char* startp = s.c_str (); | |
e38723d2 MH |
1327 | char* endp = (char*) startp; |
1328 | ||
1329 | // NB: we allow controlled overflow from LLONG_MIN .. ULLONG_MAX | |
1330 | // Actually, this allows all the way from -ULLONG_MAX to ULLONG_MAX, | |
1331 | // since the lexer only gives us positive digit strings, but we'll | |
1332 | // limit it to LLONG_MIN when a '-' operator is fed into the literal. | |
1333 | errno = 0; | |
1334 | value = (int64_t) strtoull (startp, & endp, 0); | |
1335 | if (errno == ERANGE || errno == EINVAL || *endp != '\0' | |
1336 | || (neg && (unsigned long long) value > 9223372036854775808ULL) | |
1337 | || (unsigned long long) value > 18446744073709551615ULL | |
1338 | || value < -9223372036854775807LL-1) | |
f0454224 | 1339 | throw PARSE_ERROR (_("number invalid or out of range")); |
dff50e09 | 1340 | |
e38723d2 MH |
1341 | if (neg) |
1342 | value = -value; | |
1343 | ||
731a5359 | 1344 | swallow (); // We are done with it, content was parsed and copied into value. |
57b73400 GH |
1345 | } |
1346 | ||
d7f3e0c5 | 1347 | |
dff50e09 | 1348 | const token* |
b1f2b0e8 | 1349 | parser::expect_ident_or_atword (interned_string & target) |
d7f3e0c5 | 1350 | { |
06219d6f SM |
1351 | const token *t = next(); |
1352 | ||
1353 | // accept identifiers and operators beginning in '@': | |
1354 | if (!t || (t->type != tok_identifier | |
1355 | && (t->type != tok_operator || t->content[0] != '@'))) | |
1356 | // XXX currently this is only called from parse_hist_op_or_bare_name(), | |
1357 | // so the message is accurate, but keep an eye out in the future: | |
f0454224 | 1358 | throw PARSE_ERROR (_F("expected %s or statistical operation", tt2str(tok_identifier).c_str())); |
06219d6f | 1359 | |
47d349b1 | 1360 | target = t->content; |
06219d6f | 1361 | return t; |
d7f3e0c5 GH |
1362 | } |
1363 | ||
1364 | ||
731a5359 | 1365 | void |
b1f2b0e8 | 1366 | parser::expect_ident_or_keyword (interned_string & target) |
493ee224 | 1367 | { |
731a5359 | 1368 | expect_unknown2 (tok_identifier, tok_keyword, target); |
493ee224 DS |
1369 | } |
1370 | ||
1371 | ||
dff50e09 | 1372 | bool |
aa389a19 | 1373 | parser::peek_op (string const & op) |
d7f3e0c5 GH |
1374 | { |
1375 | return tok_is (peek(), tok_operator, op); | |
1376 | } | |
1377 | ||
1378 | ||
dff50e09 | 1379 | bool |
aa389a19 | 1380 | parser::peek_kw (string const & kw) |
d7f3e0c5 GH |
1381 | { |
1382 | return tok_is (peek(), tok_identifier, kw); | |
1383 | } | |
1384 | ||
1385 | ||
1386 | ||
f086fe02 MC |
1387 | lexer::lexer (istream& input, const string& in, systemtap_session& s, bool cc): |
1388 | ate_comment(false), ate_whitespace(false), saw_tokens(false), check_compatible(cc), | |
aa389a19 FCE |
1389 | input_name (in), input_pointer (0), input_end (0), cursor_suspend_count(0), |
1390 | cursor_suspend_line (1), cursor_suspend_column (1), cursor_line (1), | |
1391 | cursor_column (1), session(s), current_file (0), current_token_chain (0) | |
eacb10ce | 1392 | { |
aa389a19 | 1393 | getline(input, input_contents, '\0'); |
73ddea86 | 1394 | |
4915b3e4 AJ |
1395 | input_pointer = input_contents.data(); |
1396 | input_end = input_contents.data() + input_contents.size(); | |
66c7d4c1 JS |
1397 | |
1398 | if (keywords.empty()) | |
1399 | { | |
3a7ec735 FCE |
1400 | // NB: adding new keywords is highly disruptive to the language, |
1401 | // in particular to existing scripts that could be suddenly | |
1402 | // broken. If done at all, it has to be s.compatible-sensitive, | |
1403 | // and broadly advertised. | |
66c7d4c1 JS |
1404 | keywords.insert("probe"); |
1405 | keywords.insert("global"); | |
38bf68a8 MC |
1406 | if (has_version("3.0")) |
1407 | keywords.insert("private"); | |
66c7d4c1 JS |
1408 | keywords.insert("function"); |
1409 | keywords.insert("if"); | |
1410 | keywords.insert("else"); | |
1411 | keywords.insert("for"); | |
1412 | keywords.insert("foreach"); | |
1413 | keywords.insert("in"); | |
1414 | keywords.insert("limit"); | |
1415 | keywords.insert("return"); | |
1416 | keywords.insert("delete"); | |
1417 | keywords.insert("while"); | |
1418 | keywords.insert("break"); | |
1419 | keywords.insert("continue"); | |
1420 | keywords.insert("next"); | |
1421 | keywords.insert("string"); | |
1422 | keywords.insert("long"); | |
f4fe2e93 FCE |
1423 | keywords.insert("try"); |
1424 | keywords.insert("catch"); | |
66c7d4c1 | 1425 | } |
2524d1fd SM |
1426 | |
1427 | if (atwords.empty()) | |
1428 | { | |
1429 | // NB: adding new @words is mildly disruptive to existing | |
1430 | // scripts that define macros with the same name, but not | |
1431 | // really. The user will merely receive a warning that they are | |
1432 | // redefining an existing operator. | |
17f56d2a JS |
1433 | |
1434 | // These are inserted without the actual '@', so we can directly check | |
1435 | // proposed macro names without building a string with that prefix. | |
1436 | atwords.insert("cast"); | |
1437 | atwords.insert("defined"); | |
0a7eb12d MC |
1438 | if (has_version("3.1")) |
1439 | atwords.insert("const"); | |
17f56d2a JS |
1440 | atwords.insert("entry"); |
1441 | atwords.insert("perf"); | |
1442 | atwords.insert("var"); | |
1443 | atwords.insert("avg"); | |
1444 | atwords.insert("count"); | |
1445 | atwords.insert("sum"); | |
1446 | atwords.insert("min"); | |
1447 | atwords.insert("max"); | |
1448 | atwords.insert("hist_linear"); | |
1449 | atwords.insert("hist_log"); | |
2524d1fd | 1450 | } |
eacb10ce | 1451 | } |
2f1a1aea | 1452 | |
85c97fc2 | 1453 | unordered_set<string> lexer::atwords; |
66c7d4c1 | 1454 | |
1b1b4ceb RA |
1455 | void |
1456 | lexer::set_current_file (stapfile* f) | |
1457 | { | |
1458 | current_file = f; | |
2203b032 JS |
1459 | if (f) |
1460 | { | |
47d349b1 | 1461 | f->file_contents = input_contents; |
2203b032 JS |
1462 | f->name = input_name; |
1463 | } | |
1b1b4ceb | 1464 | } |
bb2e3076 | 1465 | |
101b0805 JS |
1466 | void |
1467 | lexer::set_current_token_chain (const token* tok) | |
1468 | { | |
1469 | current_token_chain = tok; | |
1470 | } | |
1471 | ||
bb2e3076 FCE |
1472 | int |
1473 | lexer::input_peek (unsigned n) | |
1474 | { | |
aa389a19 FCE |
1475 | if (input_pointer + n >= input_end) |
1476 | return -1; // EOF | |
1477 | return (unsigned char)*(input_pointer + n); | |
bb2e3076 FCE |
1478 | } |
1479 | ||
1480 | ||
f8405ea5 JS |
1481 | bool |
1482 | lexer::has_version (const char* v) const | |
1483 | { | |
1484 | return check_compatible | |
1485 | ? strverscmp(session.compatible.c_str(), v) >= 0 | |
1486 | : true; | |
1487 | } | |
1488 | ||
dff50e09 | 1489 | int |
2f1a1aea FCE |
1490 | lexer::input_get () |
1491 | { | |
66c7d4c1 | 1492 | int c = input_peek(); |
bb2e3076 FCE |
1493 | if (c < 0) return c; // EOF |
1494 | ||
aa389a19 | 1495 | ++input_pointer; |
66c7d4c1 | 1496 | |
aa389a19 FCE |
1497 | if (cursor_suspend_count) |
1498 | { | |
1499 | // Track effect of input_put: preserve previous cursor/line_column | |
1500 | // until all of its characters are consumed. | |
1501 | if (--cursor_suspend_count == 0) | |
1502 | { | |
1503 | cursor_line = cursor_suspend_line; | |
1504 | cursor_column = cursor_suspend_column; | |
1505 | } | |
1506 | } | |
3f99432c | 1507 | else |
2f1a1aea | 1508 | { |
3f99432c FCE |
1509 | // update source cursor |
1510 | if (c == '\n') | |
1511 | { | |
1512 | cursor_line ++; | |
1513 | cursor_column = 1; | |
1514 | } | |
1515 | else | |
1516 | cursor_column ++; | |
2f1a1aea | 1517 | } |
2f1a1aea | 1518 | |
eacb10ce | 1519 | // clog << "[" << (char)c << "]"; |
2f1a1aea FCE |
1520 | return c; |
1521 | } | |
1522 | ||
1523 | ||
3f99432c | 1524 | void |
aa389a19 | 1525 | lexer::input_put (const string& chars, const token* t) |
3f99432c | 1526 | { |
aa389a19 FCE |
1527 | size_t pos = input_pointer - input_contents.data(); |
1528 | // clog << "[put:" << chars << " @" << pos << "]"; | |
1529 | input_contents.insert (pos, chars); | |
1530 | cursor_suspend_count += chars.size(); | |
1531 | cursor_suspend_line = cursor_line; | |
1532 | cursor_suspend_column = cursor_column; | |
1533 | cursor_line = t->location.line; | |
1534 | cursor_column = t->location.column; | |
1535 | input_pointer = input_contents.data() + pos; | |
1536 | input_end = input_contents.data() + input_contents.size(); | |
46b5cfb2 AJ |
1537 | } |
1538 | ||
3f99432c | 1539 | |
2f1a1aea | 1540 | token* |
b5477cd9 | 1541 | lexer::scan () |
2f1a1aea | 1542 | { |
fee28e5c | 1543 | ate_comment = false; // reset for each new token |
b5477cd9 | 1544 | ate_whitespace = false; // reset for each new token |
534aad8b SM |
1545 | |
1546 | // XXX be very sure to restore old_saw_tokens if we return without a token: | |
1547 | bool old_saw_tokens = saw_tokens; | |
1548 | saw_tokens = true; | |
1549 | ||
2f1a1aea | 1550 | token* n = new token; |
aa389a19 | 1551 | string token_str; // accumulate here instead of by incremental interning |
2203b032 | 1552 | n->location.file = current_file; |
101b0805 | 1553 | n->chain = current_token_chain; |
2f1a1aea | 1554 | |
9300f661 | 1555 | skip: |
aa389a19 | 1556 | bool suspended = (cursor_suspend_count > 0); |
2f1a1aea FCE |
1557 | n->location.line = cursor_line; |
1558 | n->location.column = cursor_column; | |
1559 | ||
1560 | int c = input_get(); | |
3f99432c | 1561 | // clog << "{" << (char)c << (char)c2 << "}"; |
2f1a1aea FCE |
1562 | if (c < 0) |
1563 | { | |
1564 | delete n; | |
534aad8b | 1565 | saw_tokens = old_saw_tokens; |
2f1a1aea FCE |
1566 | return 0; |
1567 | } | |
1568 | ||
1569 | if (isspace (c)) | |
b5477cd9 SM |
1570 | { |
1571 | ate_whitespace = true; | |
1572 | goto skip; | |
1573 | } | |
2f1a1aea | 1574 | |
66c7d4c1 JS |
1575 | int c2 = input_peek (); |
1576 | ||
3f99432c FCE |
1577 | // Paste command line arguments as character streams into |
1578 | // the beginning of a token. $1..$999 go through as raw | |
1579 | // characters; @1..@999 are quoted/escaped as strings. | |
1580 | // $# and @# expand to the number of arguments, similarly | |
1581 | // raw or quoted. | |
9300f661 | 1582 | if ((c == '$' || c == '@') && (c2 == '#')) |
3f99432c | 1583 | { |
aa389a19 FCE |
1584 | token_str.push_back (c); |
1585 | token_str.push_back (c2); | |
3f99432c | 1586 | input_get(); // swallow '#' |
46b5cfb2 | 1587 | |
9300f661 | 1588 | if (suspended) |
16fc963f | 1589 | { |
10e7c19d | 1590 | n->make_junk(tok_junk_nested_arg); |
16fc963f SM |
1591 | return n; |
1592 | } | |
aa389a19 FCE |
1593 | size_t num_args = session.args.size (); |
1594 | input_put ((c == '$') ? lex_cast (num_args) : lex_cast_qstring (num_args), n); | |
1595 | token_str.clear(); | |
9300f661 | 1596 | goto skip; |
3f99432c | 1597 | } |
9300f661 | 1598 | else if ((c == '$' || c == '@') && (isdigit (c2))) |
3f99432c FCE |
1599 | { |
1600 | unsigned idx = 0; | |
aa389a19 | 1601 | token_str.push_back (c); |
3f99432c FCE |
1602 | do |
1603 | { | |
1604 | input_get (); | |
aa389a19 | 1605 | token_str.push_back (c2); |
3f99432c FCE |
1606 | idx = (idx * 10) + (c2 - '0'); |
1607 | c2 = input_peek (); | |
1608 | } while (c2 > 0 && | |
dff50e09 | 1609 | isdigit (c2) && |
3f99432c | 1610 | idx <= session.args.size()); // prevent overflow |
16fc963f SM |
1611 | if (suspended) |
1612 | { | |
10e7c19d | 1613 | n->make_junk(tok_junk_nested_arg); |
16fc963f SM |
1614 | return n; |
1615 | } | |
3f99432c FCE |
1616 | if (idx == 0 || |
1617 | idx-1 >= session.args.size()) | |
16fc963f | 1618 | { |
10e7c19d | 1619 | n->make_junk(tok_junk_invalid_arg); |
16fc963f SM |
1620 | return n; |
1621 | } | |
aa389a19 FCE |
1622 | const string& arg = session.args[idx-1]; |
1623 | input_put ((c == '$') ? arg : lex_cast_qstring (arg), n); | |
1624 | token_str.clear(); | |
9300f661 | 1625 | goto skip; |
3f99432c FCE |
1626 | } |
1627 | ||
b5477cd9 | 1628 | else if (isalpha (c) || c == '$' || c == '@' || c == '_') |
2f1a1aea | 1629 | { |
aa389a19 | 1630 | token_str = (char) c; |
b5477cd9 | 1631 | while (isalnum (c2) || c2 == '_' || c2 == '$') |
2f1a1aea | 1632 | { |
3f99432c | 1633 | input_get (); |
aa389a19 | 1634 | token_str.push_back (c2); |
3f99432c | 1635 | c2 = input_peek (); |
6e213f58 | 1636 | } |
5775f11f | 1637 | n->content = token_str; |
213bee8f | 1638 | |
5775f11f | 1639 | if (n->content[0] == '@') |
dd90d565 | 1640 | // makes it easier to detect illegal use of @words: |
06219d6f | 1641 | n->type = tok_operator; |
5775f11f JS |
1642 | else if (keywords.count(n->content)) |
1643 | n->type = tok_keyword; | |
1644 | else | |
1645 | n->type = tok_identifier; | |
dff50e09 | 1646 | |
2f1a1aea FCE |
1647 | return n; |
1648 | } | |
1649 | ||
3a20432b | 1650 | else if (isdigit (c)) // positive literal |
2f1a1aea | 1651 | { |
2f1a1aea | 1652 | n->type = tok_number; |
aa389a19 | 1653 | token_str = (char) c; |
9c0c0e46 | 1654 | |
66c7d4c1 | 1655 | while (isalnum (c2)) |
2f1a1aea | 1656 | { |
9c0c0e46 FCE |
1657 | // NB: isalnum is very permissive. We rely on strtol, called in |
1658 | // parser::parse_literal below, to confirm that the number string | |
1659 | // is correctly formatted and in range. | |
1660 | ||
66c7d4c1 | 1661 | input_get (); |
aa389a19 | 1662 | token_str.push_back (c2); |
66c7d4c1 | 1663 | c2 = input_peek (); |
2f1a1aea | 1664 | } |
aa389a19 | 1665 | |
47d349b1 | 1666 | n->content = token_str; |
2f1a1aea FCE |
1667 | return n; |
1668 | } | |
1669 | ||
1670 | else if (c == '\"') | |
1671 | { | |
1672 | n->type = tok_string; | |
1673 | while (1) | |
1674 | { | |
1675 | c = input_get (); | |
1676 | ||
3f99432c | 1677 | if (c < 0 || c == '\n') |
2f1a1aea | 1678 | { |
10e7c19d | 1679 | n->make_junk(tok_junk_unclosed_quote); |
16fc963f | 1680 | return n; |
2f1a1aea FCE |
1681 | } |
1682 | if (c == '\"') // closing double-quotes | |
1683 | break; | |
3f99432c | 1684 | else if (c == '\\') // see also input_put |
dff50e09 | 1685 | { |
aa389a19 FCE |
1686 | c = input_get(); |
1687 | switch (c) | |
7d46afb8 | 1688 | { |
ef8a6134 | 1689 | case 'x': |
f8405ea5 | 1690 | if (!has_version("2.3")) |
ef8a6134 | 1691 | goto the_default; |
c92d3b42 | 1692 | /* FALLTHROUGH */ |
7d46afb8 GH |
1693 | case 'a': |
1694 | case 'b': | |
1695 | case 't': | |
1696 | case 'n': | |
1697 | case 'v': | |
1698 | case 'f': | |
1699 | case 'r': | |
f03954fd | 1700 | case '0' ... '7': // NB: need only match the first digit |
7d46afb8 | 1701 | case '\\': |
7d46afb8 | 1702 | // Pass these escapes through to the string value |
dff50e09 | 1703 | // being parsed; it will be emitted into a C literal. |
c7c8d469 FCE |
1704 | // XXX: PR13371: perhaps we should evaluate them here |
1705 | // (and re-quote them during translate.cxx emission). | |
aa389a19 | 1706 | token_str.push_back ('\\'); |
7d46afb8 | 1707 | |
3f99432c | 1708 | // fall through |
ef8a6134 | 1709 | default: the_default: |
aa389a19 FCE |
1710 | token_str.push_back (c); |
1711 | break; | |
7d46afb8 | 1712 | } |
2f1a1aea FCE |
1713 | } |
1714 | else | |
aa389a19 | 1715 | token_str.push_back (c); |
2f1a1aea | 1716 | } |
47d349b1 | 1717 | n->content = token_str; |
2f1a1aea FCE |
1718 | return n; |
1719 | } | |
1720 | ||
1721 | else if (ispunct (c)) | |
1722 | { | |
bb2e3076 | 1723 | int c3 = input_peek (1); |
2f1a1aea | 1724 | |
3a20432b FCE |
1725 | // NB: if we were to recognize negative numeric literals here, |
1726 | // we'd introduce another grammar ambiguity: | |
1727 | // 1-1 would be parsed as tok_number(1) and tok_number(-1) | |
1728 | // instead of tok_number(1) tok_operator('-') tok_number(1) | |
1729 | ||
66c7d4c1 | 1730 | if (c == '#') // shell comment |
2f1a1aea FCE |
1731 | { |
1732 | unsigned this_line = cursor_line; | |
bb2e3076 FCE |
1733 | do { c = input_get (); } |
1734 | while (c >= 0 && cursor_line == this_line); | |
fee28e5c | 1735 | ate_comment = true; |
b5477cd9 | 1736 | ate_whitespace = true; |
2f1a1aea FCE |
1737 | goto skip; |
1738 | } | |
66c7d4c1 | 1739 | else if ((c == '/' && c2 == '/')) // C++ comment |
63a7c90e FCE |
1740 | { |
1741 | unsigned this_line = cursor_line; | |
bb2e3076 FCE |
1742 | do { c = input_get (); } |
1743 | while (c >= 0 && cursor_line == this_line); | |
fee28e5c | 1744 | ate_comment = true; |
b5477cd9 | 1745 | ate_whitespace = true; |
63a7c90e FCE |
1746 | goto skip; |
1747 | } | |
1748 | else if (c == '/' && c2 == '*') // C comment | |
1749 | { | |
66c7d4c1 JS |
1750 | (void) input_get (); // swallow '*' already in c2 |
1751 | c = input_get (); | |
63a7c90e | 1752 | c2 = input_get (); |
bb2e3076 | 1753 | while (c2 >= 0) |
63a7c90e | 1754 | { |
66c7d4c1 JS |
1755 | if (c == '*' && c2 == '/') |
1756 | break; | |
63a7c90e FCE |
1757 | c = c2; |
1758 | c2 = input_get (); | |
63a7c90e | 1759 | } |
fee28e5c | 1760 | ate_comment = true; |
b5477cd9 | 1761 | ate_whitespace = true; |
bb2e3076 | 1762 | goto skip; |
63a7c90e | 1763 | } |
54dfabe9 FCE |
1764 | else if (c == '%' && c2 == '{') // embedded code |
1765 | { | |
1766 | n->type = tok_embedded; | |
1767 | (void) input_get (); // swallow '{' already in c2 | |
66c7d4c1 | 1768 | c = input_get (); |
aa389a19 | 1769 | c2 = input_get (); |
66c7d4c1 | 1770 | while (c2 >= 0) |
54dfabe9 | 1771 | { |
66c7d4c1 | 1772 | if (c == '%' && c2 == '}') |
46b5cfb2 | 1773 | { |
47d349b1 | 1774 | n->content = token_str; |
46b5cfb2 AJ |
1775 | return n; |
1776 | } | |
ebbf9df4 FCE |
1777 | if (c == '}' && c2 == '%') // possible typo |
1778 | session.print_warning (_("possible erroneous closing '}%', use '%}'?"), n); | |
aa389a19 FCE |
1779 | token_str.push_back (c); |
1780 | c = c2; | |
1781 | c2 = input_get(); | |
54dfabe9 | 1782 | } |
72cdb9cd | 1783 | |
10e7c19d | 1784 | n->make_junk(tok_junk_unclosed_embedded); |
46b5cfb2 | 1785 | return n; |
54dfabe9 | 1786 | } |
2f1a1aea | 1787 | |
bb2e3076 FCE |
1788 | // We're committed to recognizing at least the first character |
1789 | // as an operator. | |
2f1a1aea | 1790 | n->type = tok_operator; |
aa389a19 | 1791 | token_str = (char) c; |
2f1a1aea | 1792 | |
bb2e3076 | 1793 | // match all valid operators, in decreasing size order |
66c7d4c1 JS |
1794 | if ((c == '<' && c2 == '<' && c3 == '<') || |
1795 | (c == '<' && c2 == '<' && c3 == '=') || | |
1796 | (c == '>' && c2 == '>' && c3 == '=')) | |
82919855 | 1797 | { |
aa389a19 FCE |
1798 | token_str.push_back (c2); |
1799 | token_str.push_back (c3); | |
46b5cfb2 | 1800 | input_get (); // c2 |
46b5cfb2 | 1801 | input_get (); // c3 |
bb2e3076 | 1802 | } |
66c7d4c1 JS |
1803 | else if ((c == '=' && c2 == '=') || |
1804 | (c == '!' && c2 == '=') || | |
1805 | (c == '<' && c2 == '=') || | |
1806 | (c == '>' && c2 == '=') || | |
93daaca8 SM |
1807 | (c == '=' && c2 == '~') || |
1808 | (c == '!' && c2 == '~') || | |
66c7d4c1 JS |
1809 | (c == '+' && c2 == '=') || |
1810 | (c == '-' && c2 == '=') || | |
1811 | (c == '*' && c2 == '=') || | |
1812 | (c == '/' && c2 == '=') || | |
1813 | (c == '%' && c2 == '=') || | |
1814 | (c == '&' && c2 == '=') || | |
1815 | (c == '^' && c2 == '=') || | |
1816 | (c == '|' && c2 == '=') || | |
1817 | (c == '.' && c2 == '=') || | |
1818 | (c == '&' && c2 == '&') || | |
1819 | (c == '|' && c2 == '|') || | |
1820 | (c == '+' && c2 == '+') || | |
1821 | (c == '-' && c2 == '-') || | |
1822 | (c == '-' && c2 == '>') || | |
1823 | (c == '<' && c2 == '<') || | |
1824 | (c == '>' && c2 == '>') || | |
177a8ead | 1825 | // preprocessor tokens |
66c7d4c1 JS |
1826 | (c == '%' && c2 == '(') || |
1827 | (c == '%' && c2 == '?') || | |
1828 | (c == '%' && c2 == ':') || | |
1829 | (c == '%' && c2 == ')')) | |
bb2e3076 | 1830 | { |
aa389a19 | 1831 | token_str.push_back (c2); |
bb2e3076 | 1832 | input_get (); // swallow other character |
dff50e09 | 1833 | } |
2f1a1aea | 1834 | |
47d349b1 | 1835 | n->content = token_str; |
2f1a1aea FCE |
1836 | return n; |
1837 | } | |
1838 | ||
1839 | else | |
1840 | { | |
1841 | n->type = tok_junk; | |
e3795795 FCE |
1842 | ostringstream s; |
1843 | s << "\\x" << hex << setw(2) << setfill('0') << c; | |
47d349b1 | 1844 | n->content = s.str(); |
10e7c19d JS |
1845 | // signal parser to emit "expected X, found junk" type error |
1846 | n->make_junk(tok_junk_unknown); | |
2f1a1aea FCE |
1847 | return n; |
1848 | } | |
1849 | } | |
1850 | ||
16fc963f SM |
1851 | // ------------------------------------------------------------------------ |
1852 | ||
1853 | void | |
10e7c19d | 1854 | token::make_junk (token_junk_type junk) |
16fc963f SM |
1855 | { |
1856 | type = tok_junk; | |
10e7c19d JS |
1857 | junk_type = junk; |
1858 | } | |
1859 | ||
1860 | // ------------------------------------------------------------------------ | |
1861 | ||
1862 | string | |
1863 | token::junk_message(systemtap_session& session) const | |
1864 | { | |
1865 | switch (junk_type) | |
1866 | { | |
1867 | case tok_junk_nested_arg: | |
1868 | return _("invalid nested substitution of command line arguments"); | |
1869 | ||
1870 | case tok_junk_invalid_arg: | |
1871 | return _F("command line argument out of range [1-%lu]", | |
1872 | (unsigned long) session.args.size()); | |
1873 | ||
1874 | case tok_junk_unclosed_quote: | |
1875 | return _("Could not find matching closing quote"); | |
1876 | ||
1877 | case tok_junk_unclosed_embedded: | |
1878 | return _("Could not find matching '%}' to close embedded function block"); | |
1879 | ||
1880 | default: | |
1881 | return _("unknown junk token"); | |
1882 | } | |
16fc963f | 1883 | } |
2f1a1aea FCE |
1884 | |
1885 | // ------------------------------------------------------------------------ | |
1886 | ||
1887 | stapfile* | |
f8405ea5 | 1888 | parser::parse () |
2f1a1aea FCE |
1889 | { |
1890 | stapfile* f = new stapfile; | |
1b1b4ceb | 1891 | input.set_current_file (f); |
56099f08 FCE |
1892 | |
1893 | bool empty = true; | |
1894 | ||
2f1a1aea FCE |
1895 | while (1) |
1896 | { | |
1897 | try | |
1898 | { | |
a07a2c28 | 1899 | systemtap_v_seen = 0; |
2f1a1aea | 1900 | const token* t = peek (); |
534aad8b | 1901 | if (! t) // nice clean EOF, modulo any preprocessing that occurred |
2f1a1aea FCE |
1902 | break; |
1903 | ||
56099f08 | 1904 | empty = false; |
6e213f58 DS |
1905 | if (t->type == tok_keyword && t->content == "probe") |
1906 | { | |
1907 | context = con_probe; | |
1908 | parse_probe (f->probes, f->aliases); | |
1909 | } | |
38bf68a8 MC |
1910 | else if (t->type == tok_keyword && t->content == "private") |
1911 | { | |
1912 | context = con_unknown; | |
1913 | parse_private (f->globals, f->probes, f->name, f->functions); | |
1914 | } | |
6e213f58 DS |
1915 | else if (t->type == tok_keyword && t->content == "global") |
1916 | { | |
1917 | context = con_global; | |
38bf68a8 | 1918 | parse_global (f->globals, f->probes, f->name); |
6e213f58 DS |
1919 | } |
1920 | else if (t->type == tok_keyword && t->content == "function") | |
1921 | { | |
1922 | context = con_function; | |
38bf68a8 | 1923 | parse_functiondecl (f->functions, f->name); |
6e213f58 | 1924 | } |
54dfabe9 | 1925 | else if (t->type == tok_embedded) |
6e213f58 DS |
1926 | { |
1927 | context = con_embedded; | |
1928 | f->embeds.push_back (parse_embeddedcode ()); | |
1929 | } | |
2f1a1aea | 1930 | else |
6e213f58 DS |
1931 | { |
1932 | context = con_unknown; | |
42eed2a0 | 1933 | throw PARSE_ERROR (_("expected 'probe', 'global', 'private', 'function', or '%{'")); |
6e213f58 | 1934 | } |
2f1a1aea FCE |
1935 | } |
1936 | catch (parse_error& pe) | |
1937 | { | |
7ac01ea0 | 1938 | print_error (pe, errs_as_warnings); |
16fc963f SM |
1939 | |
1940 | // XXX: do we want tok_junk to be able to force skip_some behaviour? | |
cd7116b8 | 1941 | if (pe.skip_some) // for recovery |
46954f1d FCE |
1942 | // Quietly swallow all tokens until the next keyword we can start parsing from. |
1943 | while (1) | |
1944 | try | |
1945 | { | |
cd7116b8 FCE |
1946 | { |
1947 | const token* t = peek (); | |
1948 | if (! t) | |
1949 | break; | |
46954f1d | 1950 | if (t->type == tok_keyword && t->content == "probe") break; |
42eed2a0 | 1951 | else if (t->type == tok_keyword && t->content == "private") break; |
46954f1d FCE |
1952 | else if (t->type == tok_keyword && t->content == "global") break; |
1953 | else if (t->type == tok_keyword && t->content == "function") break; | |
1954 | else if (t->type == tok_embedded) break; | |
731a5359 | 1955 | swallow (); // swallow it |
cd7116b8 | 1956 | } |
46954f1d FCE |
1957 | } |
1958 | catch (parse_error& pe2) | |
1959 | { | |
1960 | // parse error during recovery ... ugh | |
1961 | print_error (pe2); | |
1962 | } | |
177a8ead | 1963 | } |
2f1a1aea FCE |
1964 | } |
1965 | ||
4bc2b5cd | 1966 | if (empty && user_file) |
56099f08 | 1967 | { |
534aad8b SM |
1968 | // vary message depending on whether file was *actually* empty: |
1969 | cerr << (input.saw_tokens | |
1970 | ? _F("Input file '%s' is empty after preprocessing.", input_name.c_str()) | |
4cd32d8c | 1971 | : _F("Input file '%s' is empty.", input_name.c_str())) |
534aad8b | 1972 | << endl; |
56099f08 | 1973 | delete f; |
2203b032 | 1974 | f = 0; |
56099f08 FCE |
1975 | } |
1976 | else if (num_errors > 0) | |
2f1a1aea | 1977 | { |
52c2652f | 1978 | cerr << _NF("%d parse error.", "%d parse errors.", num_errors, num_errors) << endl; |
2f1a1aea | 1979 | delete f; |
2203b032 | 1980 | f = 0; |
2f1a1aea | 1981 | } |
dff50e09 | 1982 | |
2203b032 | 1983 | input.set_current_file(0); |
2f1a1aea FCE |
1984 | return f; |
1985 | } | |
1986 | ||
1987 | ||
101b0805 | 1988 | probe* |
f8405ea5 | 1989 | parser::parse_synthetic_probe (const token* chain) |
101b0805 JS |
1990 | { |
1991 | probe* p = NULL; | |
1992 | stapfile* f = new stapfile; | |
1993 | f->synthetic = true; | |
1994 | input.set_current_file (f); | |
1995 | input.set_current_token_chain (chain); | |
1996 | ||
1997 | try | |
1998 | { | |
1999 | context = con_probe; | |
2000 | parse_probe (f->probes, f->aliases); | |
2001 | ||
2002 | if (f->probes.size() != 1 || !f->aliases.empty()) | |
2003 | throw PARSE_ERROR (_("expected a single synthetic probe")); | |
2004 | p = f->probes[0]; | |
2005 | } | |
2006 | catch (parse_error& pe) | |
2007 | { | |
2008 | print_error (pe, errs_as_warnings); | |
2009 | } | |
2010 | ||
2011 | // TODO check for unparsed tokens? | |
2012 | ||
2013 | input.set_current_file(0); | |
2014 | input.set_current_token_chain(0); | |
2015 | return p; | |
2016 | } | |
2017 | ||
2018 | ||
20c6c071 | 2019 | void |
aa389a19 FCE |
2020 | parser::parse_probe (vector<probe *> & probe_ret, |
2021 | vector<probe_alias *> & alias_ret) | |
2f1a1aea | 2022 | { |
82919855 | 2023 | const token* t0 = next (); |
6e213f58 | 2024 | if (! (t0->type == tok_keyword && t0->content == "probe")) |
f0454224 | 2025 | throw PARSE_ERROR (_("expected 'probe'")); |
82919855 | 2026 | |
20c6c071 GH |
2027 | vector<probe_point *> aliases; |
2028 | vector<probe_point *> locations; | |
2029 | ||
97266278 LG |
2030 | int epilogue_alias = 0; |
2031 | ||
2f1a1aea FCE |
2032 | while (1) |
2033 | { | |
380d759b | 2034 | vector<probe_point*> pps = parse_probe_points(); |
dff50e09 | 2035 | |
b4ceace2 | 2036 | const token* t = peek (); |
380d759b | 2037 | if (pps.size() == 1 && t |
b4ceace2 FCE |
2038 | && t->type == tok_operator && t->content == "=") |
2039 | { | |
380d759b FL |
2040 | if (pps[0]->optional || pps[0]->sufficient) |
2041 | throw PARSE_ERROR (_("probe point alias name cannot be optional nor sufficient"), pps[0]->components.front()->tok); | |
2042 | aliases.push_back(pps[0]); | |
731a5359 | 2043 | swallow (); |
b4ceace2 FCE |
2044 | continue; |
2045 | } | |
380d759b | 2046 | else if (pps.size() == 1 && t |
97266278 LG |
2047 | && t->type == tok_operator && t->content == "+=") |
2048 | { | |
380d759b FL |
2049 | if (pps[0]->optional || pps[0]->sufficient) |
2050 | throw PARSE_ERROR (_("probe point alias name cannot be optional nor sufficient"), pps[0]->components.front()->tok); | |
2051 | aliases.push_back(pps[0]); | |
97266278 | 2052 | epilogue_alias = 1; |
731a5359 | 2053 | swallow (); |
97266278 LG |
2054 | continue; |
2055 | } | |
b4ceace2 FCE |
2056 | else if (t && t->type == tok_operator && t->content == "{") |
2057 | { | |
380d759b | 2058 | locations.insert(locations.end(), pps.begin(), pps.end()); |
b4ceace2 FCE |
2059 | break; |
2060 | } | |
2f1a1aea | 2061 | else |
f0454224 | 2062 | throw PARSE_ERROR (_("expected probe point specifier")); |
2f1a1aea | 2063 | } |
20c6c071 | 2064 | |
20c6c071 GH |
2065 | if (aliases.empty()) |
2066 | { | |
54dfabe9 FCE |
2067 | probe* p = new probe; |
2068 | p->tok = t0; | |
2069 | p->locations = locations; | |
2070 | p->body = parse_stmt_block (); | |
37ebca01 | 2071 | p->privileged = privileged; |
a07a2c28 | 2072 | p->systemtap_v_conditional = systemtap_v_seen; |
54dfabe9 | 2073 | probe_ret.push_back (p); |
20c6c071 GH |
2074 | } |
2075 | else | |
2076 | { | |
54dfabe9 | 2077 | probe_alias* p = new probe_alias (aliases); |
97266278 LG |
2078 | if(epilogue_alias) |
2079 | p->epilogue_style = true; | |
2080 | else | |
2081 | p->epilogue_style = false; | |
54dfabe9 FCE |
2082 | p->tok = t0; |
2083 | p->locations = locations; | |
2084 | p->body = parse_stmt_block (); | |
37ebca01 | 2085 | p->privileged = privileged; |
a07a2c28 | 2086 | p->systemtap_v_conditional = systemtap_v_seen; |
54dfabe9 | 2087 | alias_ret.push_back (p); |
20c6c071 | 2088 | } |
54dfabe9 | 2089 | } |
20c6c071 | 2090 | |
54dfabe9 FCE |
2091 | |
2092 | embeddedcode* | |
2093 | parser::parse_embeddedcode () | |
2094 | { | |
2095 | embeddedcode* e = new embeddedcode; | |
2096 | const token* t = next (); | |
2097 | if (t->type != tok_embedded) | |
f0454224 | 2098 | throw PARSE_ERROR (_("expected '%{'")); |
24cb178f FCE |
2099 | |
2100 | if (! privileged) | |
f0454224 | 2101 | throw PARSE_ERROR (_("embedded code in unprivileged script; need stap -g"), |
cd7116b8 | 2102 | false /* don't skip tokens for parse resumption */); |
54dfabe9 FCE |
2103 | |
2104 | e->tok = t; | |
47d349b1 | 2105 | e->code = t->content; |
54dfabe9 | 2106 | return e; |
2f1a1aea FCE |
2107 | } |
2108 | ||
2109 | ||
2110 | block* | |
56099f08 | 2111 | parser::parse_stmt_block () |
2f1a1aea FCE |
2112 | { |
2113 | block* pb = new block; | |
2114 | ||
56099f08 FCE |
2115 | const token* t = next (); |
2116 | if (! (t->type == tok_operator && t->content == "{")) | |
f0454224 | 2117 | throw PARSE_ERROR (_("expected '{'")); |
56099f08 FCE |
2118 | |
2119 | pb->tok = t; | |
2b066ec1 | 2120 | |
2f1a1aea FCE |
2121 | while (1) |
2122 | { | |
46954f1d FCE |
2123 | t = peek (); |
2124 | if (t && t->type == tok_operator && t->content == "}") | |
2125 | { | |
731a5359 | 2126 | swallow (); |
46954f1d FCE |
2127 | break; |
2128 | } | |
2129 | pb->statements.push_back (parse_statement ()); | |
2f1a1aea FCE |
2130 | } |
2131 | ||
2132 | return pb; | |
2133 | } | |
2134 | ||
2135 | ||
f4fe2e93 FCE |
2136 | try_block* |
2137 | parser::parse_try_block () | |
2138 | { | |
2139 | try_block* pb = new try_block; | |
2140 | ||
731a5359 | 2141 | pb->tok = expect_kw_token ("try"); |
f4fe2e93 FCE |
2142 | pb->try_block = parse_stmt_block(); |
2143 | expect_kw ("catch"); | |
2144 | ||
2145 | const token* t = peek (); | |
3819d181 | 2146 | if (t != NULL && t->type == tok_operator && t->content == "(") |
f4fe2e93 | 2147 | { |
731a5359 | 2148 | swallow (); // swallow the '(' |
f4fe2e93 FCE |
2149 | |
2150 | t = next(); | |
2151 | if (! (t->type == tok_identifier)) | |
f0454224 | 2152 | throw PARSE_ERROR (_("expected identifier")); |
f4fe2e93 FCE |
2153 | symbol* sym = new symbol; |
2154 | sym->tok = t; | |
a3e980f9 | 2155 | sym->name = t->content; |
f4fe2e93 FCE |
2156 | pb->catch_error_var = sym; |
2157 | ||
2158 | expect_op (")"); | |
2159 | } | |
2160 | else | |
2161 | pb->catch_error_var = 0; | |
2162 | ||
2163 | pb->catch_block = parse_stmt_block(); | |
2164 | ||
2165 | return pb; | |
2166 | } | |
2167 | ||
2168 | ||
2169 | ||
2f1a1aea FCE |
2170 | statement* |
2171 | parser::parse_statement () | |
2172 | { | |
40b71c47 | 2173 | statement *ret; |
2f1a1aea FCE |
2174 | const token* t = peek (); |
2175 | if (t && t->type == tok_operator && t->content == ";") | |
f946b10f | 2176 | return new null_statement (next ()); |
dff50e09 | 2177 | else if (t && t->type == tok_operator && t->content == "{") |
40b71c47 | 2178 | return parse_stmt_block (); // Don't squash semicolons. |
f4fe2e93 FCE |
2179 | else if (t && t->type == tok_keyword && t->content == "try") |
2180 | return parse_try_block (); // Don't squash semicolons. | |
6e213f58 | 2181 | else if (t && t->type == tok_keyword && t->content == "if") |
40b71c47 | 2182 | return parse_if_statement (); // Don't squash semicolons. |
6e213f58 | 2183 | else if (t && t->type == tok_keyword && t->content == "for") |
40b71c47 | 2184 | return parse_for_loop (); // Don't squash semicolons. |
6e213f58 | 2185 | else if (t && t->type == tok_keyword && t->content == "foreach") |
40b71c47 MW |
2186 | return parse_foreach_loop (); // Don't squash semicolons. |
2187 | else if (t && t->type == tok_keyword && t->content == "while") | |
2188 | return parse_while_loop (); // Don't squash semicolons. | |
6e213f58 | 2189 | else if (t && t->type == tok_keyword && t->content == "return") |
40b71c47 | 2190 | ret = parse_return_statement (); |
6e213f58 | 2191 | else if (t && t->type == tok_keyword && t->content == "delete") |
40b71c47 | 2192 | ret = parse_delete_statement (); |
6e213f58 | 2193 | else if (t && t->type == tok_keyword && t->content == "break") |
40b71c47 | 2194 | ret = parse_break_statement (); |
6e213f58 | 2195 | else if (t && t->type == tok_keyword && t->content == "continue") |
40b71c47 | 2196 | ret = parse_continue_statement (); |
6e213f58 | 2197 | else if (t && t->type == tok_keyword && t->content == "next") |
40b71c47 | 2198 | ret = parse_next_statement (); |
2f1a1aea FCE |
2199 | else if (t && (t->type == tok_operator || // expressions are flexible |
2200 | t->type == tok_identifier || | |
2201 | t->type == tok_number || | |
7d902887 FCE |
2202 | t->type == tok_string || |
2203 | t->type == tok_embedded )) | |
40b71c47 | 2204 | ret = parse_expr_statement (); |
54dfabe9 | 2205 | // XXX: consider generally accepting tok_embedded here too |
2f1a1aea | 2206 | else |
f0454224 | 2207 | throw PARSE_ERROR (_("expected statement")); |
40b71c47 MW |
2208 | |
2209 | // Squash "empty" trailing colons after any "non-block-like" statement. | |
2210 | t = peek (); | |
2211 | if (t && t->type == tok_operator && t->content == ";") | |
2212 | { | |
731a5359 | 2213 | swallow (); // Silently eat trailing ; after statement |
40b71c47 MW |
2214 | } |
2215 | ||
2216 | return ret; | |
2f1a1aea FCE |
2217 | } |
2218 | ||
38bf68a8 | 2219 | void |
f41e297c JS |
2220 | parser::parse_private (vector <vardecl*>& globals, vector<probe*>& probes, |
2221 | string const & fname, vector<functiondecl*>& functions) | |
38bf68a8 MC |
2222 | { |
2223 | const token* t = next (); | |
127e4e36 | 2224 | if (! (t->type == tok_keyword && t->content == "private")) |
38bf68a8 MC |
2225 | throw PARSE_ERROR (_("expected 'private'")); |
2226 | swallow (); | |
2227 | t = next (); | |
2228 | if (t->type == tok_keyword && t->content == "function") | |
2229 | { | |
2230 | swallow (); | |
2231 | context = con_function; | |
2232 | do_parse_functiondecl(functions, t, fname, true); | |
2233 | } | |
2234 | else if (t->type == tok_keyword && t->content == "global") | |
2235 | { | |
2236 | swallow (); | |
2237 | context = con_global; | |
2238 | t = next (); | |
127e4e36 | 2239 | if (! (t->type == tok_identifier)) |
38bf68a8 MC |
2240 | throw PARSE_ERROR (_("expected identifier")); |
2241 | do_parse_global(globals, probes, fname, t, true); | |
2242 | } | |
2243 | // The `private <identifier>` is an acceptable shorthand | |
2244 | // for `private global <identifier>` per above. | |
2245 | else if (t->type == tok_identifier) | |
2246 | { | |
2247 | context = con_global; | |
2248 | do_parse_global(globals, probes, fname, t, true); | |
2249 | } | |
2250 | else | |
2251 | throw PARSE_ERROR (_("expected 'function' or identifier")); | |
2252 | } | |
2f1a1aea | 2253 | |
56099f08 | 2254 | void |
f41e297c JS |
2255 | parser::parse_global (vector <vardecl*>& globals, vector<probe*>& probes, |
2256 | string const & fname) | |
2f1a1aea | 2257 | { |
82919855 | 2258 | const token* t0 = next (); |
6e213f58 | 2259 | if (! (t0->type == tok_keyword && t0->content == "global")) |
38bf68a8 | 2260 | throw PARSE_ERROR (_("expected 'global' or 'private'")); |
731a5359 | 2261 | swallow (); |
38bf68a8 MC |
2262 | do_parse_global(globals, probes, fname, 0, false); |
2263 | } | |
82919855 | 2264 | |
38bf68a8 | 2265 | void |
f41e297c JS |
2266 | parser::do_parse_global (vector <vardecl*>& globals, vector<probe*>&, |
2267 | string const & fname, const token* t0, bool priv) | |
38bf68a8 MC |
2268 | { |
2269 | bool iter0 = true; | |
2270 | const token* t; | |
56099f08 FCE |
2271 | while (1) |
2272 | { | |
42eed2a0 MC |
2273 | t = (iter0 && priv) ? t0 : next (); |
2274 | iter0 = false; | |
56099f08 | 2275 | if (! (t->type == tok_identifier)) |
f0454224 | 2276 | throw PARSE_ERROR (_("expected identifier")); |
56099f08 | 2277 | |
c8fbf931 MC |
2278 | string gname = "__global_" + string(t->content); |
2279 | string pname = "__private_" + detox_path(fname) + string(t->content); | |
2280 | string name = priv ? pname : gname; | |
2281 | ||
2b066ec1 | 2282 | for (unsigned i=0; i<globals.size(); i++) |
c8fbf931 MC |
2283 | { |
2284 | if (globals[i]->name == name) | |
f0454224 | 2285 | throw PARSE_ERROR (_("duplicate global name")); |
c8fbf931 MC |
2286 | if ((globals[i]->name == gname) || (globals[i]->name == pname)) |
2287 | throw PARSE_ERROR (_("global versus private variable declaration conflict")); | |
2288 | } | |
38bf68a8 | 2289 | |
24cb178f | 2290 | vardecl* d = new vardecl; |
9fef07ff | 2291 | d->unmangled_name = t->content; |
38bf68a8 | 2292 | d->name = name; |
24cb178f | 2293 | d->tok = t; |
a07a2c28 | 2294 | d->systemtap_v_conditional = systemtap_v_seen; |
24cb178f | 2295 | globals.push_back (d); |
56099f08 | 2296 | |
82919855 | 2297 | t = peek (); |
ef474d24 | 2298 | |
74e6cc92 CM |
2299 | if(t && t->type == tok_operator && t->content == "%") //wrapping |
2300 | { | |
2301 | d->wrap = true; | |
731a5359 | 2302 | swallow (); |
74e6cc92 CM |
2303 | t = peek(); |
2304 | } | |
2305 | ||
ef474d24 JS |
2306 | if (t && t->type == tok_operator && t->content == "[") // array size |
2307 | { | |
2308 | int64_t size; | |
731a5359 | 2309 | swallow (); |
ef474d24 JS |
2310 | expect_number(size); |
2311 | if (size <= 0 || size > 1000000) // arbitrary max | |
f0454224 | 2312 | throw PARSE_ERROR(_("array size out of range")); |
ef474d24 JS |
2313 | d->maxsize = (int)size; |
2314 | expect_known(tok_operator, "]"); | |
2315 | t = peek (); | |
2316 | } | |
2317 | ||
4b5f3e45 | 2318 | if (t && t->type == tok_operator && t->content == "=") // initialization |
ef474d24 JS |
2319 | { |
2320 | if (!d->compatible_arity(0)) | |
f0454224 | 2321 | throw PARSE_ERROR(_("only scalar globals can be initialized")); |
58701b78 | 2322 | d->set_arity(0, t); |
731a5359 | 2323 | next (); // Don't swallow, set_arity() used the peeked token. |
ef474d24 JS |
2324 | d->init = parse_literal (); |
2325 | d->type = d->init->type; | |
2326 | t = peek (); | |
2327 | } | |
4b5f3e45 | 2328 | |
c3799d72 | 2329 | if (t && t->type == tok_operator && t->content == ";") // termination |
950da622 | 2330 | { |
731a5359 | 2331 | swallow (); |
950da622 MW |
2332 | break; |
2333 | } | |
c3799d72 | 2334 | |
4b5f3e45 | 2335 | if (t && t->type == tok_operator && t->content == ",") // next global |
82919855 | 2336 | { |
731a5359 | 2337 | swallow (); |
82919855 FCE |
2338 | continue; |
2339 | } | |
56099f08 | 2340 | else |
82919855 | 2341 | break; |
56099f08 FCE |
2342 | } |
2343 | } | |
2344 | ||
24cb178f | 2345 | void |
f41e297c JS |
2346 | parser::parse_functiondecl (vector<functiondecl*>& functions, |
2347 | string const & fname) | |
56099f08 | 2348 | { |
82919855 | 2349 | const token* t = next (); |
6e213f58 | 2350 | if (! (t->type == tok_keyword && t->content == "function")) |
f0454224 | 2351 | throw PARSE_ERROR (_("expected 'function'")); |
731a5359 | 2352 | swallow (); |
38bf68a8 MC |
2353 | do_parse_functiondecl(functions, t, fname, false); |
2354 | } | |
56099f08 | 2355 | |
38bf68a8 | 2356 | void |
f41e297c JS |
2357 | parser::do_parse_functiondecl (vector<functiondecl*>& functions, const token* t, |
2358 | string const & fname, bool priv) | |
38bf68a8 | 2359 | { |
82919855 | 2360 | t = next (); |
6e213f58 DS |
2361 | if (! (t->type == tok_identifier) |
2362 | && ! (t->type == tok_keyword | |
2363 | && (t->content == "string" || t->content == "long"))) | |
f0454224 | 2364 | throw PARSE_ERROR (_("expected identifier")); |
24cb178f | 2365 | |
c8fbf931 MC |
2366 | string gname = "__global_" + string(t->content); |
2367 | string pname = "__private_" + detox_path(fname) + string(t->content); | |
2368 | string name = priv ? pname : gname; | |
7b5b30a8 | 2369 | name += "__overload_" + lex_cast(session.overload_count[t->content]++); |
38bf68a8 | 2370 | |
24cb178f | 2371 | functiondecl *fd = new functiondecl (); |
9fef07ff | 2372 | fd->unmangled_name = t->content; |
38bf68a8 | 2373 | fd->name = name; |
56099f08 FCE |
2374 | fd->tok = t; |
2375 | ||
2376 | t = next (); | |
6a505121 FCE |
2377 | if (t->type == tok_operator && t->content == ":") |
2378 | { | |
731a5359 | 2379 | swallow (); |
6a505121 | 2380 | t = next (); |
6e213f58 | 2381 | if (t->type == tok_keyword && t->content == "string") |
6a505121 | 2382 | fd->type = pe_string; |
6e213f58 | 2383 | else if (t->type == tok_keyword && t->content == "long") |
6a505121 | 2384 | fd->type = pe_long; |
f0454224 | 2385 | else throw PARSE_ERROR (_("expected 'string' or 'long'")); |
731a5359 | 2386 | swallow (); |
6a505121 FCE |
2387 | |
2388 | t = next (); | |
2389 | } | |
2390 | ||
56099f08 | 2391 | if (! (t->type == tok_operator && t->content == "(")) |
f0454224 | 2392 | throw PARSE_ERROR (_("expected '('")); |
731a5359 | 2393 | swallow (); |
56099f08 FCE |
2394 | |
2395 | while (1) | |
2396 | { | |
2397 | t = next (); | |
2398 | ||
100a540e | 2399 | // permit zero-argument functions |
56099f08 | 2400 | if (t->type == tok_operator && t->content == ")") |
731a5359 MW |
2401 | { |
2402 | swallow (); | |
2403 | break; | |
2404 | } | |
56099f08 | 2405 | else if (! (t->type == tok_identifier)) |
f0454224 | 2406 | throw PARSE_ERROR (_("expected identifier")); |
56099f08 | 2407 | vardecl* vd = new vardecl; |
9fef07ff | 2408 | vd->unmangled_name = vd->name = t->content; |
59093206 CS |
2409 | |
2410 | for (auto it = fd->formal_args.begin() ; it != fd->formal_args.end(); ++it) | |
2411 | { | |
2412 | string param = vd->unmangled_name; | |
2413 | if ((*it)->unmangled_name == param) | |
2414 | throw PARSE_ERROR(_("duplicate parameter names")); | |
2415 | } | |
2416 | ||
56099f08 FCE |
2417 | vd->tok = t; |
2418 | fd->formal_args.push_back (vd); | |
a07a2c28 | 2419 | fd->systemtap_v_conditional = systemtap_v_seen; |
56099f08 FCE |
2420 | |
2421 | t = next (); | |
59093206 | 2422 | |
6a505121 FCE |
2423 | if (t->type == tok_operator && t->content == ":") |
2424 | { | |
731a5359 | 2425 | swallow (); |
6a505121 | 2426 | t = next (); |
6e213f58 | 2427 | if (t->type == tok_keyword && t->content == "string") |
6a505121 | 2428 | vd->type = pe_string; |
6e213f58 | 2429 | else if (t->type == tok_keyword && t->content == "long") |
6a505121 | 2430 | vd->type = pe_long; |
f0454224 | 2431 | else throw PARSE_ERROR (_("expected 'string' or 'long'")); |
731a5359 | 2432 | swallow (); |
6a505121 FCE |
2433 | t = next (); |
2434 | } | |
56099f08 | 2435 | if (t->type == tok_operator && t->content == ")") |
731a5359 MW |
2436 | { |
2437 | swallow (); | |
2438 | break; | |
59093206 | 2439 | } |
56099f08 | 2440 | if (t->type == tok_operator && t->content == ",") |
731a5359 MW |
2441 | { |
2442 | swallow (); | |
2443 | continue; | |
2444 | } | |
56099f08 | 2445 | else |
f0454224 | 2446 | throw PARSE_ERROR (_("expected ',' or ')'")); |
56099f08 FCE |
2447 | } |
2448 | ||
7b5b30a8 FL |
2449 | t = peek(); |
2450 | if (t->type == tok_operator && t->content == ":") | |
2451 | { | |
7b5b30a8 | 2452 | swallow(); |
c0d0d623 FL |
2453 | literal* literal = parse_literal(); |
2454 | fd->priority = dynamic_cast<literal_number*>(literal)->value; | |
7b5b30a8 | 2455 | // reserve priority 0 for user script implementation |
c0d0d623 | 2456 | if (fd->priority < 1) |
7b5b30a8 | 2457 | throw PARSE_ERROR (_("specified priority must be > 0")); |
c0d0d623 | 2458 | delete literal; |
7b5b30a8 FL |
2459 | } |
2460 | else if (user_file) | |
2461 | { | |
2462 | // allow script file implementation override automatically when | |
2463 | // priority not specified | |
2464 | fd->priority = 0; | |
2465 | } | |
2466 | ||
54dfabe9 FCE |
2467 | t = peek (); |
2468 | if (t && t->type == tok_embedded) | |
2469 | fd->body = parse_embeddedcode (); | |
2470 | else | |
2471 | fd->body = parse_stmt_block (); | |
24cb178f FCE |
2472 | |
2473 | functions.push_back (fd); | |
2f1a1aea FCE |
2474 | } |
2475 | ||
380d759b FL |
2476 | vector<probe_point*> |
2477 | parser::parse_probe_points() | |
2f1a1aea | 2478 | { |
380d759b FL |
2479 | vector<probe_point*> pps; |
2480 | while (1) | |
2481 | { | |
2482 | vector<probe_point*> tail = parse_components(); | |
2483 | pps.insert(pps.end(), tail.begin(), tail.end()); | |
2484 | ||
2485 | const token* t = peek(); | |
2486 | if (t && t->type == tok_operator && t->content == ",") | |
2487 | { | |
2488 | swallow(); | |
2489 | continue; | |
2490 | } | |
2491 | ||
2492 | if (t && t->type == tok_operator | |
2493 | && (t->content == "{" || t->content == "=" || | |
2494 | t->content == "+="|| t->content == "}")) | |
2495 | break; | |
2f1a1aea | 2496 | |
380d759b FL |
2497 | throw PARSE_ERROR (_("expected one of ', { } = +='")); |
2498 | } | |
2499 | return pps; | |
2500 | } | |
2501 | ||
2502 | vector<probe_point*> | |
2503 | parser::parse_components() | |
2504 | { | |
2505 | vector<probe_point*> pps; | |
9c0c0e46 | 2506 | while (1) |
2f1a1aea | 2507 | { |
380d759b | 2508 | vector<probe_point*> suffix = parse_component(); |
9c0c0e46 | 2509 | |
380d759b FL |
2510 | // Cartesian product of components |
2511 | if (pps.empty()) | |
2512 | pps = suffix; | |
2513 | else | |
2514 | { | |
2515 | assert(!suffix.empty()); | |
2516 | vector<probe_point*> product; | |
2517 | for (unsigned i = 0; i < pps.size(); i++) | |
2518 | { | |
2519 | if (pps[i]->optional || pps[i]->sufficient || pps[i]->condition) | |
2520 | throw PARSE_ERROR (_("'?', '!' or condition must only be specified in suffix"), | |
2521 | pps[i]->components[0]->tok); | |
2522 | for (unsigned j = 0; j < suffix.size(); j++) | |
2523 | { | |
2524 | probe_point* pp = new probe_point; | |
2525 | pp->components.insert(pp->components.end(), | |
2526 | pps[i]->components.begin(), pps[i]->components.end()); | |
2527 | pp->components.insert(pp->components.end(), | |
2528 | suffix[j]->components.begin(), suffix[j]->components.end()); | |
2529 | pp->optional = suffix[j]->optional; | |
2530 | pp->sufficient = suffix[j]->sufficient; | |
2531 | pp->condition = suffix[j]->condition; | |
2532 | product.push_back(pp); | |
2533 | } | |
2534 | } | |
2535 | for (unsigned i = 0; i < pps.size(); i++) delete pps[i]; | |
2536 | for (unsigned i = 0; i < suffix.size(); i++) delete suffix[i]; | |
2537 | pps = product; | |
2538 | } | |
2539 | ||
2540 | const token* t = peek(); | |
2541 | if (t && t->type == tok_operator && t->content == ".") | |
2542 | { | |
2543 | swallow (); | |
2544 | continue; | |
2545 | } | |
2546 | ||
2547 | // We only fall through here at the end of a probe point (past | |
2548 | // all the dotted/parametrized components). | |
2549 | ||
2550 | if (t && t->type == tok_operator && | |
2551 | (t->content == "?" || t->content == "!")) | |
2552 | { | |
2553 | for (unsigned i = 0; i < pps.size(); i++) | |
2554 | { | |
2555 | if (pps[i]->optional || pps[i]->sufficient) | |
2556 | throw PARSE_ERROR (_("'?' or '!' respecified")); | |
2557 | pps[i]->optional = true; | |
2558 | if (t->content == "!") pps[i]->sufficient = true; | |
2559 | } | |
2560 | // NB: sufficient implies optional | |
2561 | swallow (); | |
2562 | t = peek (); | |
2563 | // fall through | |
2564 | } | |
2565 | ||
2566 | if (t && t->type == tok_keyword && t->content == "if") | |
2567 | { | |
2568 | swallow (); | |
2569 | t = peek (); | |
2570 | if (!(t && t->type == tok_operator && t->content == "(")) | |
2571 | throw PARSE_ERROR (_("expected '('")); | |
2572 | swallow (); | |
2573 | ||
2574 | expression* e = parse_expression(); | |
2575 | for (unsigned i = 0; i < pps.size(); i++) | |
2576 | { | |
2577 | if (pps[i]->condition != 0) | |
2578 | throw PARSE_ERROR (_("condition respecified")); | |
2579 | pps[i]->condition = e; | |
2580 | } | |
2581 | ||
2582 | t = peek (); | |
2583 | if (!(t && t->type == tok_operator && t->content == ")")) | |
2584 | throw PARSE_ERROR (_("expected ')'")); | |
2585 | swallow (); | |
2586 | } | |
2587 | ||
2588 | break; | |
2589 | } | |
2590 | return pps; | |
2591 | } | |
2592 | ||
2593 | vector<probe_point*> | |
2594 | parser::parse_component() | |
2595 | { | |
2596 | const token* t = next (); | |
2597 | if (! (t->type == tok_identifier | |
2598 | // we must allow ".return" and ".function", which are keywords | |
2599 | || t->type == tok_keyword | |
2600 | // we must allow "*", due to being an operator | |
2601 | || (t->type == tok_operator && (t->content == "*" || t->content == "{")))) | |
2602 | throw PARSE_ERROR (_("expected identifier or '*' or '{'")); | |
2603 | ||
2604 | if (t && t->type == tok_operator && t->content == "{") | |
2605 | { | |
2606 | swallow(); | |
2607 | vector<probe_point*> pps = parse_probe_points(); | |
2608 | t = peek(); | |
2609 | if (!(t && t->type == tok_operator && t->content == "}")) | |
2610 | throw PARSE_ERROR (_("expected '}'")); | |
2611 | swallow(); | |
2612 | return pps; | |
2613 | } | |
2614 | else | |
2615 | { | |
b5477cd9 | 2616 | // loop which reconstitutes an identifier with wildcards |
47d349b1 | 2617 | string content = t->content; |
eadd685c | 2618 | bool changed_p = false; |
b5477cd9 SM |
2619 | while (1) |
2620 | { | |
2621 | const token* u = peek(); | |
3819d181 MW |
2622 | if (u == NULL) |
2623 | break; | |
b5477cd9 SM |
2624 | // ensure pieces of the identifier are adjacent: |
2625 | if (input.ate_whitespace) | |
2626 | break; | |
2627 | // ensure pieces of the identifier are valid: | |
2628 | if (! (u->type == tok_identifier | |
2629 | // we must allow arbitrary keywords with a wildcard | |
2630 | || u->type == tok_keyword | |
2631 | // we must allow "*", due to being an operator | |
2632 | || (u->type == tok_operator && u->content == "*"))) | |
2633 | break; | |
2634 | ||
2635 | // append u to t | |
47d349b1 | 2636 | content = content + (string)u->content; |
eadd685c | 2637 | changed_p = true; |
380d759b | 2638 | |
b5477cd9 | 2639 | // consume u |
731a5359 | 2640 | swallow (); |
b5477cd9 | 2641 | } |
eadd685c FCE |
2642 | |
2643 | if (changed_p) | |
2644 | { | |
2645 | // We've already swallowed the first token and we're not | |
2646 | // putting it back; no one else has a copy; so we can | |
2647 | // safely overwrite its content and reuse it. | |
2648 | const_cast<token*>(t)->content = content; | |
2649 | } | |
9c0c0e46 FCE |
2650 | |
2651 | probe_point::component* c = new probe_point::component; | |
a3e980f9 | 2652 | c->functor = t->content; |
f1a0157a | 2653 | c->tok = t; |
380d759b FL |
2654 | vector<probe_point*> pps; |
2655 | probe_point* pp = new probe_point; | |
2656 | pp->components.push_back(c); | |
2657 | pps.push_back(pp); | |
6e3347a9 | 2658 | // NB we may add c->arg soon |
9c0c0e46 FCE |
2659 | |
2660 | t = peek (); | |
a477f3f1 | 2661 | |
6e3347a9 | 2662 | // consume optional parameter |
9c0c0e46 FCE |
2663 | if (t && t->type == tok_operator && t->content == "(") |
2664 | { | |
731a5359 | 2665 | swallow (); // consume "(" |
9c0c0e46 FCE |
2666 | c->arg = parse_literal (); |
2667 | ||
2668 | t = next (); | |
2669 | if (! (t->type == tok_operator && t->content == ")")) | |
f0454224 | 2670 | throw PARSE_ERROR (_("expected ')'")); |
731a5359 | 2671 | swallow (); |
6e3347a9 | 2672 | } |
380d759b | 2673 | return pps; |
2f1a1aea | 2674 | } |
2f1a1aea FCE |
2675 | } |
2676 | ||
d24f1ff4 SM |
2677 | literal_string* |
2678 | parser::consume_string_literals(const token *t) | |
2679 | { | |
0ad0aae2 | 2680 | literal_string *ls = new literal_string (t->content); |
d24f1ff4 SM |
2681 | |
2682 | // PR11208: check if the next token is also a string literal; | |
2683 | // auto-concatenate it. This is complicated to the extent that we | |
2684 | // need to skip intermediate whitespace. | |
2685 | // | |
2686 | // NB for versions prior to 2.0: but don't skip over intervening comments | |
0ad0aae2 JS |
2687 | string concat; |
2688 | bool p_concat = false; | |
d24f1ff4 SM |
2689 | const token *n = peek(); |
2690 | while (n != NULL && n->type == tok_string | |
f8405ea5 | 2691 | && ! (!input.has_version("2.0") && input.ate_comment)) |
d24f1ff4 | 2692 | { |
0ad0aae2 JS |
2693 | if (!p_concat) |
2694 | { | |
2695 | concat = t->content; | |
2696 | p_concat = true; | |
2697 | } | |
2698 | concat.append(n->content.data(), n->content.size()); | |
2699 | next(); // consume the token | |
d24f1ff4 SM |
2700 | n = peek(); |
2701 | } | |
0ad0aae2 JS |
2702 | if (p_concat) |
2703 | ls->value = concat; | |
d24f1ff4 SM |
2704 | return ls; |
2705 | } | |
2706 | ||
2707 | ||
2708 | // Parse a string literal and perform backslash escaping on the contents: | |
2709 | literal_string* | |
2710 | parser::parse_literal_string () | |
2711 | { | |
2712 | const token* t = next (); | |
2713 | literal_string* l; | |
2714 | if (t->type == tok_string) | |
2715 | l = consume_string_literals (t); | |
2716 | else | |
f0454224 | 2717 | throw PARSE_ERROR (_("expected literal string")); |
d24f1ff4 SM |
2718 | |
2719 | l->tok = t; | |
2720 | return l; | |
2721 | } | |
2722 | ||
2723 | ||
2f1a1aea FCE |
2724 | literal* |
2725 | parser::parse_literal () | |
2726 | { | |
2727 | const token* t = next (); | |
56099f08 | 2728 | literal* l; |
2f1a1aea | 2729 | if (t->type == tok_string) |
c5be7511 | 2730 | { |
d24f1ff4 | 2731 | l = consume_string_literals (t); |
c5be7511 | 2732 | } |
16e8f21f | 2733 | else |
9c0c0e46 | 2734 | { |
16e8f21f JS |
2735 | bool neg = false; |
2736 | if (t->type == tok_operator && t->content == "-") | |
2737 | { | |
2738 | neg = true; | |
731a5359 | 2739 | swallow (); |
16e8f21f JS |
2740 | t = next (); |
2741 | } | |
2742 | ||
2743 | if (t->type == tok_number) | |
2744 | { | |
7371cd19 JS |
2745 | const string& s = t->content; |
2746 | const char* startp = s.c_str (); | |
16e8f21f JS |
2747 | char* endp = (char*) startp; |
2748 | ||
2749 | // NB: we allow controlled overflow from LLONG_MIN .. ULLONG_MAX | |
2750 | // Actually, this allows all the way from -ULLONG_MAX to ULLONG_MAX, | |
79e6d33f JS |
2751 | // since the lexer only gives us positive digit strings, but we'll |
2752 | // limit it to LLONG_MIN when a '-' operator is fed into the literal. | |
16e8f21f JS |
2753 | errno = 0; |
2754 | long long value = (long long) strtoull (startp, & endp, 0); | |
16e8f21f | 2755 | if (errno == ERANGE || errno == EINVAL || *endp != '\0' |
79e6d33f | 2756 | || (neg && (unsigned long long) value > 9223372036854775808ULL) |
16e8f21f JS |
2757 | || (unsigned long long) value > 18446744073709551615ULL |
2758 | || value < -9223372036854775807LL-1) | |
f0454224 | 2759 | throw PARSE_ERROR (_("number invalid or out of range")); |
16e8f21f | 2760 | |
79e6d33f JS |
2761 | if (neg) |
2762 | value = -value; | |
2763 | ||
16e8f21f JS |
2764 | l = new literal_number (value); |
2765 | } | |
2766 | else | |
f0454224 | 2767 | throw PARSE_ERROR (_("expected literal string or number")); |
9c0c0e46 | 2768 | } |
56099f08 FCE |
2769 | |
2770 | l->tok = t; | |
2771 | return l; | |
2f1a1aea FCE |
2772 | } |
2773 | ||
2774 | ||
2775 | if_statement* | |
2776 | parser::parse_if_statement () | |
2777 | { | |
2778 | const token* t = next (); | |
6e213f58 | 2779 | if (! (t->type == tok_keyword && t->content == "if")) |
f0454224 | 2780 | throw PARSE_ERROR (_("expected 'if'")); |
56099f08 FCE |
2781 | if_statement* s = new if_statement; |
2782 | s->tok = t; | |
2783 | ||
2784 | t = next (); | |
2f1a1aea | 2785 | if (! (t->type == tok_operator && t->content == "(")) |
f0454224 | 2786 | throw PARSE_ERROR (_("expected '('")); |
731a5359 | 2787 | swallow (); |
2f1a1aea | 2788 | |
2f1a1aea FCE |
2789 | s->condition = parse_expression (); |
2790 | ||
2791 | t = next (); | |
2792 | if (! (t->type == tok_operator && t->content == ")")) | |
f0454224 | 2793 | throw PARSE_ERROR (_("expected ')'")); |
731a5359 | 2794 | swallow (); |
2f1a1aea FCE |
2795 | |
2796 | s->thenblock = parse_statement (); | |
2797 | ||
2798 | t = peek (); | |
6e213f58 | 2799 | if (t && t->type == tok_keyword && t->content == "else") |
2f1a1aea | 2800 | { |
731a5359 | 2801 | swallow (); |
2f1a1aea FCE |
2802 | s->elseblock = parse_statement (); |
2803 | } | |
ed10c639 FCE |
2804 | else |
2805 | s->elseblock = 0; // in case not otherwise initialized | |
2f1a1aea FCE |
2806 | |
2807 | return s; | |
2808 | } | |
2809 | ||
2810 | ||
69c68955 FCE |
2811 | expr_statement* |
2812 | parser::parse_expr_statement () | |
2813 | { | |
2814 | expr_statement *es = new expr_statement; | |
2815 | const token* t = peek (); | |
5e58d11c | 2816 | if (t == NULL) |
f0454224 | 2817 | throw PARSE_ERROR (_("expression statement expected")); |
731a5359 MW |
2818 | // Copy, we only peeked, parse_expression might swallow. |
2819 | es->tok = new token (*t); | |
69c68955 FCE |
2820 | es->value = parse_expression (); |
2821 | return es; | |
2822 | } | |
2823 | ||
2824 | ||
56099f08 FCE |
2825 | return_statement* |
2826 | parser::parse_return_statement () | |
2827 | { | |
2828 | const token* t = next (); | |
6e213f58 | 2829 | if (! (t->type == tok_keyword && t->content == "return")) |
f0454224 | 2830 | throw PARSE_ERROR (_("expected 'return'")); |
6e213f58 | 2831 | if (context != con_function) |
f0454224 | 2832 | throw PARSE_ERROR (_("found 'return' not in function context")); |
56099f08 FCE |
2833 | return_statement* s = new return_statement; |
2834 | s->tok = t; | |
2835 | s->value = parse_expression (); | |
2836 | return s; | |
2837 | } | |
2838 | ||
2839 | ||
2840 | delete_statement* | |
2841 | parser::parse_delete_statement () | |
2842 | { | |
2843 | const token* t = next (); | |
6e213f58 | 2844 | if (! (t->type == tok_keyword && t->content == "delete")) |
f0454224 | 2845 | throw PARSE_ERROR (_("expected 'delete'")); |
56099f08 FCE |
2846 | delete_statement* s = new delete_statement; |
2847 | s->tok = t; | |
2848 | s->value = parse_expression (); | |
2849 | return s; | |
2850 | } | |
2851 | ||
2852 | ||
f3c26ea5 FCE |
2853 | next_statement* |
2854 | parser::parse_next_statement () | |
2855 | { | |
2856 | const token* t = next (); | |
6e213f58 | 2857 | if (! (t->type == tok_keyword && t->content == "next")) |
f0454224 | 2858 | throw PARSE_ERROR (_("expected 'next'")); |
f3c26ea5 FCE |
2859 | next_statement* s = new next_statement; |
2860 | s->tok = t; | |
2861 | return s; | |
2862 | } | |
2863 | ||
2864 | ||
2865 | break_statement* | |
2866 | parser::parse_break_statement () | |
2867 | { | |
2868 | const token* t = next (); | |
6e213f58 | 2869 | if (! (t->type == tok_keyword && t->content == "break")) |
f0454224 | 2870 | throw PARSE_ERROR (_("expected 'break'")); |
f3c26ea5 FCE |
2871 | break_statement* s = new break_statement; |
2872 | s->tok = t; | |
2873 | return s; | |
2874 | } | |
2875 | ||
2876 | ||
2877 | continue_statement* | |
2878 | parser::parse_continue_statement () | |
2879 | { | |
2880 | const token* t = next (); | |
6e213f58 | 2881 | if (! (t->type == tok_keyword && t->content == "continue")) |
f0454224 | 2882 | throw PARSE_ERROR (_("expected 'continue'")); |
f3c26ea5 FCE |
2883 | continue_statement* s = new continue_statement; |
2884 | s->tok = t; | |
2885 | return s; | |
2886 | } | |
2887 | ||
2888 | ||
69c68955 FCE |
2889 | for_loop* |
2890 | parser::parse_for_loop () | |
2891 | { | |
f3c26ea5 | 2892 | const token* t = next (); |
6e213f58 | 2893 | if (! (t->type == tok_keyword && t->content == "for")) |
f0454224 | 2894 | throw PARSE_ERROR (_("expected 'for'")); |
f3c26ea5 FCE |
2895 | for_loop* s = new for_loop; |
2896 | s->tok = t; | |
2897 | ||
2898 | t = next (); | |
2899 | if (! (t->type == tok_operator && t->content == "(")) | |
f0454224 | 2900 | throw PARSE_ERROR (_("expected '('")); |
731a5359 | 2901 | swallow (); |
f3c26ea5 FCE |
2902 | |
2903 | // initializer + ";" | |
2904 | t = peek (); | |
2905 | if (t && t->type == tok_operator && t->content == ";") | |
2906 | { | |
cbfbbf69 | 2907 | s->init = 0; |
731a5359 | 2908 | swallow (); |
f3c26ea5 FCE |
2909 | } |
2910 | else | |
2911 | { | |
2912 | s->init = parse_expr_statement (); | |
2913 | t = next (); | |
2914 | if (! (t->type == tok_operator && t->content == ";")) | |
f0454224 | 2915 | throw PARSE_ERROR (_("expected ';'")); |
731a5359 | 2916 | swallow (); |
f3c26ea5 FCE |
2917 | } |
2918 | ||
2919 | // condition + ";" | |
2920 | t = peek (); | |
2921 | if (t && t->type == tok_operator && t->content == ";") | |
2922 | { | |
2923 | literal_number* l = new literal_number(1); | |
2924 | s->cond = l; | |
2925 | s->cond->tok = next (); | |
2926 | } | |
2927 | else | |
2928 | { | |
2929 | s->cond = parse_expression (); | |
2930 | t = next (); | |
2931 | if (! (t->type == tok_operator && t->content == ";")) | |
f0454224 | 2932 | throw PARSE_ERROR (_("expected ';'")); |
731a5359 | 2933 | swallow (); |
f3c26ea5 | 2934 | } |
dff50e09 | 2935 | |
f3c26ea5 FCE |
2936 | // increment + ")" |
2937 | t = peek (); | |
2938 | if (t && t->type == tok_operator && t->content == ")") | |
2939 | { | |
cbfbbf69 | 2940 | s->incr = 0; |
731a5359 | 2941 | swallow (); |
f3c26ea5 FCE |
2942 | } |
2943 | else | |
2944 | { | |
2945 | s->incr = parse_expr_statement (); | |
2946 | t = next (); | |
2947 | if (! (t->type == tok_operator && t->content == ")")) | |
f0454224 | 2948 | throw PARSE_ERROR (_("expected ')'")); |
731a5359 | 2949 | swallow (); |
f3c26ea5 FCE |
2950 | } |
2951 | ||
2952 | // block | |
2953 | s->block = parse_statement (); | |
2954 | ||
2955 | return s; | |
2956 | } | |
2957 | ||
2958 | ||
2959 | for_loop* | |
2960 | parser::parse_while_loop () | |
2961 | { | |
2962 | const token* t = next (); | |
6e213f58 | 2963 | if (! (t->type == tok_keyword && t->content == "while")) |
f0454224 | 2964 | throw PARSE_ERROR (_("expected 'while'")); |
f3c26ea5 FCE |
2965 | for_loop* s = new for_loop; |
2966 | s->tok = t; | |
2967 | ||
2968 | t = next (); | |
2969 | if (! (t->type == tok_operator && t->content == "(")) | |
f0454224 | 2970 | throw PARSE_ERROR (_("expected '('")); |
731a5359 | 2971 | swallow (); |
f3c26ea5 FCE |
2972 | |
2973 | // dummy init and incr fields | |
cbfbbf69 FCE |
2974 | s->init = 0; |
2975 | s->incr = 0; | |
f3c26ea5 FCE |
2976 | |
2977 | // condition | |
2978 | s->cond = parse_expression (); | |
2979 | ||
f3c26ea5 FCE |
2980 | t = next (); |
2981 | if (! (t->type == tok_operator && t->content == ")")) | |
f0454224 | 2982 | throw PARSE_ERROR (_("expected ')'")); |
731a5359 | 2983 | swallow (); |
dff50e09 | 2984 | |
f3c26ea5 FCE |
2985 | // block |
2986 | s->block = parse_statement (); | |
2987 | ||
2988 | return s; | |
69c68955 FCE |
2989 | } |
2990 | ||
2991 | ||
2992 | foreach_loop* | |
2993 | parser::parse_foreach_loop () | |
2994 | { | |
2995 | const token* t = next (); | |
6e213f58 | 2996 | if (! (t->type == tok_keyword && t->content == "foreach")) |
f0454224 | 2997 | throw PARSE_ERROR (_("expected 'foreach'")); |
69c68955 FCE |
2998 | foreach_loop* s = new foreach_loop; |
2999 | s->tok = t; | |
93484556 | 3000 | s->sort_direction = 0; |
fd5689dc | 3001 | s->sort_aggr = sc_none; |
c261711d | 3002 | s->value = NULL; |
27f21e8c | 3003 | s->limit = NULL; |
69c68955 FCE |
3004 | |
3005 | t = next (); | |
3006 | if (! (t->type == tok_operator && t->content == "(")) | |
f0454224 | 3007 | throw PARSE_ERROR (_("expected '('")); |
731a5359 | 3008 | swallow (); |
69c68955 | 3009 | |
c261711d JS |
3010 | symbol* lookahead_sym = NULL; |
3011 | int lookahead_sort = 0; | |
3012 | ||
3013 | t = peek (); | |
3014 | if (t && t->type == tok_identifier) | |
3015 | { | |
3016 | next (); | |
3017 | lookahead_sym = new symbol; | |
3018 | lookahead_sym->tok = t; | |
a3e980f9 | 3019 | lookahead_sym->name = t->content; |
c261711d JS |
3020 | |
3021 | t = peek (); | |
3022 | if (t && t->type == tok_operator && | |
3023 | (t->content == "+" || t->content == "-")) | |
3024 | { | |
c261711d | 3025 | lookahead_sort = (t->content == "+") ? 1 : -1; |
731a5359 | 3026 | swallow (); |
c261711d JS |
3027 | } |
3028 | ||
3029 | t = peek (); | |
3030 | if (t && t->type == tok_operator && t->content == "=") | |
3031 | { | |
731a5359 | 3032 | swallow (); |
c261711d JS |
3033 | s->value = lookahead_sym; |
3034 | if (lookahead_sort) | |
3035 | { | |
3036 | s->sort_direction = lookahead_sort; | |
3037 | s->sort_column = 0; | |
3038 | } | |
3039 | lookahead_sym = NULL; | |
3040 | } | |
3041 | } | |
3042 | ||
69c68955 FCE |
3043 | // see also parse_array_in |
3044 | ||
3045 | bool parenthesized = false; | |
3046 | t = peek (); | |
c261711d | 3047 | if (!lookahead_sym && t && t->type == tok_operator && t->content == "[") |
69c68955 | 3048 | { |
731a5359 | 3049 | swallow (); |
69c68955 FCE |
3050 | parenthesized = true; |
3051 | } | |
3052 | ||
c261711d JS |
3053 | if (lookahead_sym) |
3054 | { | |
3055 | s->indexes.push_back (lookahead_sym); | |
3056 | if (lookahead_sort) | |
3057 | { | |
3058 | s->sort_direction = lookahead_sort; | |
3059 | s->sort_column = 1; | |
3060 | } | |
3061 | lookahead_sym = NULL; | |
3062 | } | |
3063 | else while (1) | |
69c68955 FCE |
3064 | { |
3065 | t = next (); | |
3066 | if (! (t->type == tok_identifier)) | |
f0454224 | 3067 | throw PARSE_ERROR (_("expected identifier")); |
69c68955 FCE |
3068 | symbol* sym = new symbol; |
3069 | sym->tok = t; | |
a3e980f9 | 3070 | sym->name = t->content; |
69c68955 FCE |
3071 | s->indexes.push_back (sym); |
3072 | ||
93484556 FCE |
3073 | t = peek (); |
3074 | if (t && t->type == tok_operator && | |
3075 | (t->content == "+" || t->content == "-")) | |
3076 | { | |
3077 | if (s->sort_direction) | |
f0454224 | 3078 | throw PARSE_ERROR (_("multiple sort directives")); |
93484556 FCE |
3079 | s->sort_direction = (t->content == "+") ? 1 : -1; |
3080 | s->sort_column = s->indexes.size(); | |
731a5359 | 3081 | swallow (); |
93484556 FCE |
3082 | } |
3083 | ||
69c68955 FCE |
3084 | if (parenthesized) |
3085 | { | |
93484556 | 3086 | t = peek (); |
69c68955 FCE |
3087 | if (t && t->type == tok_operator && t->content == ",") |
3088 | { | |
731a5359 | 3089 | swallow (); |
69c68955 FCE |
3090 | continue; |
3091 | } | |
3092 | else if (t && t->type == tok_operator && t->content == "]") | |
3093 | { | |
731a5359 | 3094 | swallow (); |
69c68955 FCE |
3095 | break; |
3096 | } | |
dff50e09 | 3097 | else |
f0454224 | 3098 | throw PARSE_ERROR (_("expected ',' or ']'")); |
69c68955 FCE |
3099 | } |
3100 | else | |
3101 | break; // expecting only one expression | |
3102 | } | |
3103 | ||
3104 | t = next (); | |
6e213f58 | 3105 | if (! (t->type == tok_keyword && t->content == "in")) |
f0454224 | 3106 | throw PARSE_ERROR (_("expected 'in'")); |
731a5359 | 3107 | swallow (); |
dff50e09 | 3108 | |
d02548c0 | 3109 | s->base = parse_indexable(); |
69c68955 | 3110 | |
3040bf3a AJ |
3111 | // check if there was an array slice that was specified |
3112 | t = peek(); | |
3113 | if (t && t->type == tok_operator && t->content == "[") | |
3114 | { | |
3115 | swallow(); | |
3116 | while (1) | |
3117 | { | |
3118 | t = peek(); | |
3040bf3a AJ |
3119 | if (t && t->type == tok_operator && t->content == "*") |
3120 | { | |
45af9d1b AJ |
3121 | swallow(); |
3122 | s->array_slice.push_back (NULL); | |
3040bf3a AJ |
3123 | } |
3124 | else | |
45af9d1b | 3125 | s->array_slice.push_back (parse_expression()); |
3040bf3a AJ |
3126 | |
3127 | t = peek (); | |
3128 | if (t && t->type == tok_operator && t->content == ",") | |
3129 | { | |
3130 | swallow (); | |
3131 | continue; | |
3132 | } | |
3133 | else if (t && t->type == tok_operator && t->content == "]") | |
3134 | { | |
3135 | swallow (); | |
3136 | break; | |
3137 | } | |
3138 | else | |
3139 | throw PARSE_ERROR (_("expected ',' or ']'")); | |
3140 | } | |
3141 | } | |
3142 | ||
3143 | ||
fd5689dc FCE |
3144 | // check for atword, see also expect_ident_or_atword, |
3145 | t = peek (); | |
3146 | if (t && t->type == tok_operator && t->content[0] == '@') | |
3147 | { | |
3148 | if (t->content == "@avg") s->sort_aggr = sc_average; | |
3149 | else if (t->content == "@min") s->sort_aggr = sc_min; | |
3150 | else if (t->content == "@max") s->sort_aggr = sc_max; | |
3151 | else if (t->content == "@count") s->sort_aggr = sc_count; | |
3152 | else if (t->content == "@sum") s->sort_aggr = sc_sum; | |
f0454224 | 3153 | else throw PARSE_ERROR(_("expected statistical operation")); |
fd5689dc FCE |
3154 | swallow(); |
3155 | ||
3156 | t = peek (); | |
3157 | if (! (t && t->type == tok_operator && (t->content == "+" || t->content == "-"))) | |
f0454224 | 3158 | throw PARSE_ERROR(_("expected sort directive")); |
fd5689dc FCE |
3159 | } |
3160 | ||
93484556 FCE |
3161 | t = peek (); |
3162 | if (t && t->type == tok_operator && | |
3163 | (t->content == "+" || t->content == "-")) | |
3164 | { | |
3165 | if (s->sort_direction) | |
f0454224 | 3166 | throw PARSE_ERROR (_("multiple sort directives")); |
93484556 FCE |
3167 | s->sort_direction = (t->content == "+") ? 1 : -1; |
3168 | s->sort_column = 0; | |
731a5359 | 3169 | swallow (); |
93484556 FCE |
3170 | } |
3171 | ||
27f21e8c DS |
3172 | t = peek (); |
3173 | if (tok_is(t, tok_keyword, "limit")) | |
3174 | { | |
731a5359 | 3175 | swallow (); // get past the "limit" |
27f21e8c DS |
3176 | s->limit = parse_expression (); |
3177 | } | |
3178 | ||
69c68955 FCE |
3179 | t = next (); |
3180 | if (! (t->type == tok_operator && t->content == ")")) | |
f0454224 | 3181 | throw PARSE_ERROR ("expected ')'"); |
731a5359 | 3182 | swallow (); |
69c68955 FCE |
3183 | |
3184 | s->block = parse_statement (); | |
3185 | return s; | |
3186 | } | |
3187 | ||
3188 | ||
2f1a1aea FCE |
3189 | expression* |
3190 | parser::parse_expression () | |
3191 | { | |
3192 | return parse_assignment (); | |
3193 | } | |
3194 | ||
2f1a1aea FCE |
3195 | |
3196 | expression* | |
3197 | parser::parse_assignment () | |
3198 | { | |
3199 | expression* op1 = parse_ternary (); | |
3200 | ||
3201 | const token* t = peek (); | |
82919855 | 3202 | // right-associative operators |
dff50e09 | 3203 | if (t && t->type == tok_operator |
2f1a1aea | 3204 | && (t->content == "=" || |
82919855 | 3205 | t->content == "<<<" || |
2f1a1aea | 3206 | t->content == "+=" || |
bb2e3076 FCE |
3207 | t->content == "-=" || |
3208 | t->content == "*=" || | |
3209 | t->content == "/=" || | |
3210 | t->content == "%=" || | |
3211 | t->content == "<<=" || | |
3212 | t->content == ">>=" || | |
3213 | t->content == "&=" || | |
3214 | t->content == "^=" || | |
3215 | t->content == "|=" || | |
d5d7c2cc | 3216 | t->content == ".=" || |
dff50e09 | 3217 | false)) |
2f1a1aea | 3218 | { |
bb2e3076 | 3219 | // NB: lvalueness is checked during elaboration / translation |
2f1a1aea | 3220 | assignment* e = new assignment; |
56099f08 | 3221 | e->left = op1; |
47d349b1 | 3222 | e->op = t->content; |
56099f08 | 3223 | e->tok = t; |
2f1a1aea | 3224 | next (); |
82919855 | 3225 | e->right = parse_expression (); |
56099f08 | 3226 | op1 = e; |
2f1a1aea | 3227 | } |
56099f08 FCE |
3228 | |
3229 | return op1; | |
2f1a1aea FCE |
3230 | } |
3231 | ||
3232 | ||
3233 | expression* | |
3234 | parser::parse_ternary () | |
3235 | { | |
3236 | expression* op1 = parse_logical_or (); | |
3237 | ||
3238 | const token* t = peek (); | |
3239 | if (t && t->type == tok_operator && t->content == "?") | |
3240 | { | |
2f1a1aea | 3241 | ternary_expression* e = new ternary_expression; |
56099f08 | 3242 | e->tok = t; |
2f1a1aea | 3243 | e->cond = op1; |
56099f08 FCE |
3244 | next (); |
3245 | e->truevalue = parse_expression (); // XXX | |
2f1a1aea FCE |
3246 | |
3247 | t = next (); | |
3248 | if (! (t->type == tok_operator && t->content == ":")) | |
f0454224 | 3249 | throw PARSE_ERROR (_("expected ':'")); |
731a5359 | 3250 | swallow (); |
2f1a1aea | 3251 | |
56099f08 | 3252 | e->falsevalue = parse_expression (); // XXX |
2f1a1aea FCE |
3253 | return e; |
3254 | } | |
3255 | else | |
3256 | return op1; | |
3257 | } | |
3258 | ||
3259 | ||
3260 | expression* | |
3261 | parser::parse_logical_or () | |
3262 | { | |
3263 | expression* op1 = parse_logical_and (); | |
dff50e09 | 3264 | |
2f1a1aea | 3265 | const token* t = peek (); |
56099f08 | 3266 | while (t && t->type == tok_operator && t->content == "||") |
2f1a1aea | 3267 | { |
2f1a1aea | 3268 | logical_or_expr* e = new logical_or_expr; |
56099f08 | 3269 | e->tok = t; |
47d349b1 | 3270 | e->op = t->content; |
2f1a1aea | 3271 | e->left = op1; |
56099f08 FCE |
3272 | next (); |
3273 | e->right = parse_logical_and (); | |
3274 | op1 = e; | |
3275 | t = peek (); | |
2f1a1aea | 3276 | } |
56099f08 FCE |
3277 | |
3278 | return op1; | |
2f1a1aea FCE |
3279 | } |
3280 | ||
3281 | ||
3282 | expression* | |
3283 | parser::parse_logical_and () | |
3284 | { | |
bb2e3076 | 3285 | expression* op1 = parse_boolean_or (); |
2f1a1aea FCE |
3286 | |
3287 | const token* t = peek (); | |
56099f08 | 3288 | while (t && t->type == tok_operator && t->content == "&&") |
2f1a1aea | 3289 | { |
2f1a1aea FCE |
3290 | logical_and_expr *e = new logical_and_expr; |
3291 | e->left = op1; | |
47d349b1 | 3292 | e->op = t->content; |
56099f08 FCE |
3293 | e->tok = t; |
3294 | next (); | |
bb2e3076 FCE |
3295 | e->right = parse_boolean_or (); |
3296 | op1 = e; | |
3297 | t = peek (); | |
3298 | } | |
3299 | ||
3300 | return op1; | |
3301 | } | |
3302 | ||
3303 | ||
3304 | expression* | |
3305 | parser::parse_boolean_or () | |
3306 | { | |
3307 | expression* op1 = parse_boolean_xor (); | |
3308 | ||
3309 | const token* t = peek (); | |
3310 | while (t && t->type == tok_operator && t->content == "|") | |
3311 | { | |
3312 | binary_expression* e = new binary_expression; | |
3313 | e->left = op1; | |
47d349b1 | 3314 | e->op = t->content; |
bb2e3076 FCE |
3315 | e->tok = t; |
3316 | next (); | |
3317 | e->right = parse_boolean_xor (); | |
3318 | op1 = e; | |
3319 | t = peek (); | |
3320 | } | |
3321 | ||
3322 | return op1; | |
3323 | } | |
3324 | ||
3325 | ||
3326 | expression* | |
3327 | parser::parse_boolean_xor () | |
3328 | { | |
3329 | expression* op1 = parse_boolean_and (); | |
3330 | ||
3331 | const token* t = peek (); | |
3332 | while (t && t->type == tok_operator && t->content == "^") | |
3333 | { | |
3334 | binary_expression* e = new binary_expression; | |
3335 | e->left = op1; | |
47d349b1 | 3336 | e->op = t->content; |
bb2e3076 FCE |
3337 | e->tok = t; |
3338 | next (); | |
3339 | e->right = parse_boolean_and (); | |
3340 | op1 = e; | |
3341 | t = peek (); | |
3342 | } | |
3343 | ||
3344 | return op1; | |
3345 | } | |
3346 | ||
3347 | ||
3348 | expression* | |
3349 | parser::parse_boolean_and () | |
3350 | { | |
3351 | expression* op1 = parse_array_in (); | |
3352 | ||
3353 | const token* t = peek (); | |
3354 | while (t && t->type == tok_operator && t->content == "&") | |
3355 | { | |
3356 | binary_expression* e = new binary_expression; | |
3357 | e->left = op1; | |
47d349b1 | 3358 | e->op = t->content; |
bb2e3076 FCE |
3359 | e->tok = t; |
3360 | next (); | |
56099f08 FCE |
3361 | e->right = parse_array_in (); |
3362 | op1 = e; | |
3363 | t = peek (); | |
2f1a1aea | 3364 | } |
56099f08 FCE |
3365 | |
3366 | return op1; | |
2f1a1aea FCE |
3367 | } |
3368 | ||
3369 | ||
3370 | expression* | |
3371 | parser::parse_array_in () | |
3372 | { | |
ce10591c | 3373 | // This is a very tricky case. All these are legit expressions: |
69c68955 | 3374 | // "a in b" "a+0 in b" "[a,b] in c" "[c,(d+0)] in b" |
ce10591c FCE |
3375 | vector<expression*> indexes; |
3376 | bool parenthesized = false; | |
2f1a1aea FCE |
3377 | |
3378 | const token* t = peek (); | |
69c68955 | 3379 | if (t && t->type == tok_operator && t->content == "[") |
ce10591c | 3380 | { |
731a5359 | 3381 | swallow (); |
ce10591c FCE |
3382 | parenthesized = true; |
3383 | } | |
3384 | ||
3385 | while (1) | |
3386 | { | |
e225e273 AJ |
3387 | t = peek(); |
3388 | if (t && t->type == tok_operator && t->content == "*" && parenthesized) | |
3389 | { | |
45af9d1b AJ |
3390 | swallow(); |
3391 | indexes.push_back(NULL); | |
e225e273 AJ |
3392 | } |
3393 | else | |
3394 | { | |
3395 | expression* op1 = parse_comparison_or_regex_query (); | |
3396 | indexes.push_back (op1); | |
3397 | } | |
ce10591c FCE |
3398 | |
3399 | if (parenthesized) | |
3400 | { | |
3401 | const token* t = peek (); | |
3402 | if (t && t->type == tok_operator && t->content == ",") | |
3403 | { | |
731a5359 | 3404 | swallow (); |
ce10591c FCE |
3405 | continue; |
3406 | } | |
69c68955 | 3407 | else if (t && t->type == tok_operator && t->content == "]") |
ce10591c | 3408 | { |
731a5359 | 3409 | swallow (); |
ce10591c FCE |
3410 | break; |
3411 | } | |
dff50e09 | 3412 | else |
f0454224 | 3413 | throw PARSE_ERROR (_("expected ',' or ']'")); |
ce10591c FCE |
3414 | } |
3415 | else | |
3416 | break; // expecting only one expression | |
3417 | } | |
3418 | ||
3419 | t = peek (); | |
6e213f58 | 3420 | if (t && t->type == tok_keyword && t->content == "in") |
2f1a1aea | 3421 | { |
2f1a1aea | 3422 | array_in *e = new array_in; |
56099f08 | 3423 | e->tok = t; |
731a5359 | 3424 | next (); |
ce10591c FCE |
3425 | |
3426 | arrayindex* a = new arrayindex; | |
3427 | a->indexes = indexes; | |
d02548c0 | 3428 | a->base = parse_indexable(); |
d15d767c | 3429 | a->tok = a->base->tok; |
ce10591c | 3430 | e->operand = a; |
2f1a1aea FCE |
3431 | return e; |
3432 | } | |
ce10591c FCE |
3433 | else if (indexes.size() == 1) // no "in" - need one expression only |
3434 | return indexes[0]; | |
2f1a1aea | 3435 | else |
f0454224 | 3436 | throw PARSE_ERROR (_("unexpected comma-separated expression list")); |
2f1a1aea FCE |
3437 | } |
3438 | ||
3439 | ||
3440 | expression* | |
93daaca8 | 3441 | parser::parse_comparison_or_regex_query () |
2f1a1aea | 3442 | { |
bb2e3076 | 3443 | expression* op1 = parse_shift (); |
2f1a1aea | 3444 | |
557abe61 | 3445 | // XXX precedence -- perhaps a =~ b == c =~ d --> (a =~ b) == (c =~ d) ? |
93daaca8 SM |
3446 | const token *t = peek(); |
3447 | if (t && t->type == tok_operator | |
3448 | && (t->content == "=~" || | |
3449 | t->content == "!~")) | |
3450 | { | |
3451 | regex_query* r = new regex_query; | |
3452 | r->left = op1; | |
47d349b1 | 3453 | r->op = t->content; |
93daaca8 SM |
3454 | r->tok = t; |
3455 | next (); | |
d3bc48f0 | 3456 | r->right = parse_literal_string(); |
93daaca8 SM |
3457 | op1 = r; |
3458 | t = peek (); | |
3459 | } | |
3460 | else while (t && t->type == tok_operator | |
553d27a5 FCE |
3461 | && (t->content == ">" || |
3462 | t->content == "<" || | |
3463 | t->content == "==" || | |
3464 | t->content == "!=" || | |
3465 | t->content == "<=" || | |
bb2e3076 | 3466 | t->content == ">=")) |
2f1a1aea FCE |
3467 | { |
3468 | comparison* e = new comparison; | |
3469 | e->left = op1; | |
47d349b1 | 3470 | e->op = t->content; |
56099f08 | 3471 | e->tok = t; |
2f1a1aea | 3472 | next (); |
bb2e3076 FCE |
3473 | e->right = parse_shift (); |
3474 | op1 = e; | |
3475 | t = peek (); | |
3476 | } | |
3477 | ||
3478 | return op1; | |
3479 | } | |
3480 | ||
3481 | ||
3482 | expression* | |
3483 | parser::parse_shift () | |
3484 | { | |
3485 | expression* op1 = parse_concatenation (); | |
3486 | ||
3487 | const token* t = peek (); | |
dff50e09 | 3488 | while (t && t->type == tok_operator && |
bb2e3076 FCE |
3489 | (t->content == "<<" || t->content == ">>")) |
3490 | { | |
3491 | binary_expression* e = new binary_expression; | |
3492 | e->left = op1; | |
47d349b1 | 3493 | e->op = t->content; |
bb2e3076 FCE |
3494 | e->tok = t; |
3495 | next (); | |
56099f08 FCE |
3496 | e->right = parse_concatenation (); |
3497 | op1 = e; | |
3498 | t = peek (); | |
2f1a1aea | 3499 | } |
56099f08 FCE |
3500 | |
3501 | return op1; | |
2f1a1aea FCE |
3502 | } |
3503 | ||
3504 | ||
3505 | expression* | |
3506 | parser::parse_concatenation () | |
3507 | { | |
3508 | expression* op1 = parse_additive (); | |
3509 | ||
3510 | const token* t = peek (); | |
3511 | // XXX: the actual awk string-concatenation operator is *whitespace*. | |
3512 | // I don't know how to easily to model that here. | |
56099f08 | 3513 | while (t && t->type == tok_operator && t->content == ".") |
2f1a1aea FCE |
3514 | { |
3515 | concatenation* e = new concatenation; | |
3516 | e->left = op1; | |
47d349b1 | 3517 | e->op = t->content; |
56099f08 | 3518 | e->tok = t; |
2f1a1aea | 3519 | next (); |
56099f08 FCE |
3520 | e->right = parse_additive (); |
3521 | op1 = e; | |
3522 | t = peek (); | |
2f1a1aea | 3523 | } |
56099f08 FCE |
3524 | |
3525 | return op1; | |
2f1a1aea FCE |
3526 | } |
3527 | ||
3528 | ||
3529 | expression* | |
3530 | parser::parse_additive () | |
3531 | { | |
3532 | expression* op1 = parse_multiplicative (); | |
3533 | ||
3534 | const token* t = peek (); | |
dff50e09 | 3535 | while (t && t->type == tok_operator |
2f1a1aea FCE |
3536 | && (t->content == "+" || t->content == "-")) |
3537 | { | |
3538 | binary_expression* e = new binary_expression; | |
47d349b1 | 3539 | e->op = t->content; |
2f1a1aea | 3540 | e->left = op1; |
56099f08 | 3541 | e->tok = t; |
2f1a1aea | 3542 | next (); |
56099f08 FCE |
3543 | e->right = parse_multiplicative (); |
3544 | op1 = e; | |
3545 | t = peek (); | |
2f1a1aea | 3546 | } |
56099f08 FCE |
3547 | |
3548 | return op1; | |
2f1a1aea FCE |
3549 | } |
3550 | ||
3551 | ||
3552 | expression* | |
3553 | parser::parse_multiplicative () | |
3554 | { | |
3555 | expression* op1 = parse_unary (); | |
3556 | ||
3557 | const token* t = peek (); | |
dff50e09 | 3558 | while (t && t->type == tok_operator |
2f1a1aea FCE |
3559 | && (t->content == "*" || t->content == "/" || t->content == "%")) |
3560 | { | |
3561 | binary_expression* e = new binary_expression; | |
47d349b1 | 3562 | e->op = t->content; |
2f1a1aea | 3563 | e->left = op1; |
56099f08 | 3564 | e->tok = t; |
2f1a1aea | 3565 | next (); |
56099f08 FCE |
3566 | e->right = parse_unary (); |
3567 | op1 = e; | |
3568 | t = peek (); | |
2f1a1aea | 3569 | } |
56099f08 FCE |
3570 | |
3571 | return op1; | |
2f1a1aea FCE |
3572 | } |
3573 | ||
3574 | ||
3575 | expression* | |
3576 | parser::parse_unary () | |
3577 | { | |
3578 | const token* t = peek (); | |
dff50e09 FCE |
3579 | if (t && t->type == tok_operator |
3580 | && (t->content == "+" || | |
3581 | t->content == "-" || | |
bb2e3076 FCE |
3582 | t->content == "!" || |
3583 | t->content == "~" || | |
3584 | false)) | |
2f1a1aea FCE |
3585 | { |
3586 | unary_expression* e = new unary_expression; | |
47d349b1 | 3587 | e->op = t->content; |
56099f08 | 3588 | e->tok = t; |
2f1a1aea | 3589 | next (); |
1cb79a72 | 3590 | e->operand = parse_unary (); |
2f1a1aea FCE |
3591 | return e; |
3592 | } | |
3593 | else | |
bb2e3076 | 3594 | return parse_crement (); |
2f1a1aea FCE |
3595 | } |
3596 | ||
3597 | ||
3598 | expression* | |
3599 | parser::parse_crement () // as in "increment" / "decrement" | |
3600 | { | |
cbfbbf69 FCE |
3601 | // NB: Ideally, we'd parse only a symbol as an operand to the |
3602 | // *crement operators, instead of a general expression value. We'd | |
3603 | // need more complex lookahead code to tell apart the postfix cases. | |
3604 | // So we just punt, and leave it to pass-3 to signal errors on | |
3605 | // cases like "4++". | |
3606 | ||
2f1a1aea | 3607 | const token* t = peek (); |
dff50e09 | 3608 | if (t && t->type == tok_operator |
2f1a1aea FCE |
3609 | && (t->content == "++" || t->content == "--")) |
3610 | { | |
3611 | pre_crement* e = new pre_crement; | |
47d349b1 | 3612 | e->op = t->content; |
56099f08 | 3613 | e->tok = t; |
2f1a1aea | 3614 | next (); |
0fb0cac9 | 3615 | e->operand = parse_dwarf_value (); |
2f1a1aea FCE |
3616 | return e; |
3617 | } | |
3618 | ||
3619 | // post-crement or non-crement | |
0fb0cac9 | 3620 | expression *op1 = parse_dwarf_value (); |
dff50e09 | 3621 | |
2f1a1aea | 3622 | t = peek (); |
dff50e09 | 3623 | if (t && t->type == tok_operator |
2f1a1aea FCE |
3624 | && (t->content == "++" || t->content == "--")) |
3625 | { | |
3626 | post_crement* e = new post_crement; | |
47d349b1 | 3627 | e->op = t->content; |
56099f08 | 3628 | e->tok = t; |
2f1a1aea FCE |
3629 | next (); |
3630 | e->operand = op1; | |
3631 | return e; | |
3632 | } | |
3633 | else | |
3634 | return op1; | |
3635 | } | |
3636 | ||
3637 | ||
0fb0cac9 JS |
3638 | expression* |
3639 | parser::parse_dwarf_value () | |
3640 | { | |
3641 | expression* expr = NULL; | |
3642 | target_symbol* tsym = NULL; | |
3643 | ||
3644 | // With '&' we'll definitely be making a target symbol of some sort | |
251707c8 JS |
3645 | const token* addrtok = peek_op ("&") ? next () : NULL; |
3646 | bool addressof = (addrtok != NULL); | |
0fb0cac9 JS |
3647 | |
3648 | // First try target_symbol types: $var, @cast, and @var. | |
0fb0cac9 JS |
3649 | const token* t = peek (); |
3650 | if (t && t->type == tok_identifier && t->content[0] == '$') | |
3651 | expr = tsym = parse_target_symbol (); | |
3652 | else if (tok_is (t, tok_operator, "@cast")) | |
3653 | expr = tsym = parse_cast_op (); | |
3654 | else if (tok_is (t, tok_operator, "@var")) | |
3655 | expr = tsym = parse_atvar_op (); | |
f8405ea5 | 3656 | else if (addressof && !input.has_version("2.6")) |
eff66d40 JS |
3657 | // '&' on old version only allowed specific target_symbol types |
3658 | throw PARSE_ERROR (_("expected @cast, @var or $var")); | |
0fb0cac9 | 3659 | else |
eff66d40 | 3660 | // Otherwise just get a plain value of any sort. |
0fb0cac9 JS |
3661 | expr = parse_value (); |
3662 | ||
3663 | // If we had '&' or see any target suffixes, that forces a target_symbol. | |
eff66d40 JS |
3664 | // For compatibility, we only do this starting with 2.6. |
3665 | if (!tsym && (addressof || peek_target_symbol_components ()) | |
f8405ea5 | 3666 | && input.has_version("2.6")) |
0fb0cac9 | 3667 | { |
251707c8 JS |
3668 | autocast_op *cop = new autocast_op; |
3669 | cop->tok = addrtok ?: peek (); | |
0fb0cac9 JS |
3670 | cop->operand = expr; |
3671 | expr = tsym = cop; | |
3672 | } | |
3673 | ||
3674 | if (tsym) | |
3675 | { | |
3676 | // Parse the rest of any kind of target symbol | |
3677 | tsym->addressof = addressof; | |
3678 | parse_target_symbol_components (tsym); | |
3679 | } | |
3680 | ||
3681 | return expr; | |
3682 | } | |
3683 | ||
3684 | ||
2f1a1aea FCE |
3685 | expression* |
3686 | parser::parse_value () | |
3687 | { | |
3688 | const token* t = peek (); | |
3689 | if (! t) | |
f0454224 | 3690 | throw PARSE_ERROR (_("expected value")); |
2f1a1aea | 3691 | |
7d902887 FCE |
3692 | if (t->type == tok_embedded) |
3693 | { | |
7d902887 | 3694 | if (! privileged) |
f0454224 | 3695 | throw PARSE_ERROR (_("embedded expression code in unprivileged script; need stap -g"), false); |
7d902887 FCE |
3696 | |
3697 | embedded_expr *e = new embedded_expr; | |
3698 | e->tok = t; | |
47d349b1 | 3699 | e->code = t->content; |
731a5359 | 3700 | next (); |
7d902887 FCE |
3701 | return e; |
3702 | } | |
3703 | ||
2f1a1aea FCE |
3704 | if (t->type == tok_operator && t->content == "(") |
3705 | { | |
731a5359 | 3706 | swallow (); |
2f1a1aea FCE |
3707 | expression* e = parse_expression (); |
3708 | t = next (); | |
3709 | if (! (t->type == tok_operator && t->content == ")")) | |
f0454224 | 3710 | throw PARSE_ERROR (_("expected ')'")); |
731a5359 | 3711 | swallow (); |
2f1a1aea FCE |
3712 | return e; |
3713 | } | |
06219d6f SM |
3714 | else if (t->type == tok_identifier |
3715 | || (t->type == tok_operator && t->content[0] == '@')) | |
2f1a1aea FCE |
3716 | return parse_symbol (); |
3717 | else | |
3718 | return parse_literal (); | |
3719 | } | |
3720 | ||
3721 | ||
d02548c0 | 3722 | const token * |
b1f2b0e8 | 3723 | parser::parse_hist_op_or_bare_name (hist_op *&hop, interned_string &name) |
d02548c0 GH |
3724 | { |
3725 | hop = NULL; | |
50cc7cd5 | 3726 | const token* t = expect_ident_or_atword (name); |
d02548c0 GH |
3727 | if (name == "@hist_linear" || name == "@hist_log") |
3728 | { | |
3729 | hop = new hist_op; | |
3730 | if (name == "@hist_linear") | |
3731 | hop->htype = hist_linear; | |
3732 | else if (name == "@hist_log") | |
3733 | hop->htype = hist_log; | |
3734 | hop->tok = t; | |
3735 | expect_op("("); | |
3736 | hop->stat = parse_expression (); | |
3737 | int64_t tnum; | |
3738 | if (hop->htype == hist_linear) | |
3739 | { | |
3740 | for (size_t i = 0; i < 3; ++i) | |
3741 | { | |
3742 | expect_op (","); | |
3743 | expect_number (tnum); | |
3744 | hop->params.push_back (tnum); | |
3745 | } | |
3746 | } | |
d02548c0 GH |
3747 | expect_op(")"); |
3748 | } | |
3749 | return t; | |
3750 | } | |
3751 | ||
3752 | ||
3753 | indexable* | |
3754 | parser::parse_indexable () | |
3755 | { | |
3756 | hist_op *hop = NULL; | |
b1f2b0e8 | 3757 | interned_string name; |
d02548c0 GH |
3758 | const token *tok = parse_hist_op_or_bare_name(hop, name); |
3759 | if (hop) | |
3760 | return hop; | |
3761 | else | |
3762 | { | |
3763 | symbol* sym = new symbol; | |
3764 | sym->name = name; | |
3765 | sym->tok = tok; | |
3766 | return sym; | |
3767 | } | |
3768 | } | |
3769 | ||
3770 | ||
0fb0cac9 JS |
3771 | // var, indexable[index], func(parms), printf("...", ...), |
3772 | // @defined, @entry, @stat_op(stat) | |
30263a73 | 3773 | expression* parser::parse_symbol () |
2f1a1aea | 3774 | { |
d02548c0 GH |
3775 | hist_op *hop = NULL; |
3776 | symbol *sym = NULL; | |
b1f2b0e8 | 3777 | interned_string name; |
d02548c0 GH |
3778 | const token *t = parse_hist_op_or_bare_name(hop, name); |
3779 | ||
3780 | if (!hop) | |
0fefb486 | 3781 | { |
dff50e09 | 3782 | // If we didn't get a hist_op, then we did get an identifier. We can |
d02548c0 | 3783 | // now scrutinize this identifier for the various magic forms of identifier |
0fb0cac9 | 3784 | // (printf, @stat_op...) |
9b5af295 | 3785 | |
db135493 FCE |
3786 | // NB: PR11343: @defined() is not incompatible with earlier versions |
3787 | // of stap, so no need to check session.compatible for 1.2 | |
30263a73 FCE |
3788 | if (name == "@defined") |
3789 | return parse_defined_op (t); | |
8cc799a5 | 3790 | |
0a7eb12d MC |
3791 | if (name == "@const") |
3792 | return parse_const_op (t); | |
3793 | ||
8cc799a5 JS |
3794 | if (name == "@entry") |
3795 | return parse_entry_op (t); | |
3796 | ||
3689db05 SC |
3797 | if (name == "@perf") |
3798 | return parse_perf_op (t); | |
3799 | ||
cc9001af | 3800 | if (name.size() > 0 && name[0] == '@') |
d7f3e0c5 | 3801 | { |
d02548c0 GH |
3802 | stat_op *sop = new stat_op; |
3803 | if (name == "@avg") | |
3804 | sop->ctype = sc_average; | |
3805 | else if (name == "@count") | |
3806 | sop->ctype = sc_count; | |
3807 | else if (name == "@sum") | |
3808 | sop->ctype = sc_sum; | |
3809 | else if (name == "@min") | |
3810 | sop->ctype = sc_min; | |
3811 | else if (name == "@max") | |
3812 | sop->ctype = sc_max; | |
3813 | else | |
b1f2b0e8 JS |
3814 | throw PARSE_ERROR(_F("unknown operator %s", |
3815 | name.to_string().c_str())); | |
d02548c0 GH |
3816 | expect_op("("); |
3817 | sop->tok = t; | |
3818 | sop->stat = parse_expression (); | |
3819 | expect_op(")"); | |
3820 | return sop; | |
3821 | } | |
dff50e09 | 3822 | |
d5e178c1 | 3823 | else if (print_format *fmt = print_format::create(t)) |
d02548c0 | 3824 | { |
d02548c0 | 3825 | expect_op("("); |
b15c465c PP |
3826 | if ((name == "print" || name == "println" || |
3827 | name == "sprint" || name == "sprintln") && | |
f34254da | 3828 | (peek_op("@hist_linear") || peek_op("@hist_log"))) |
a4636912 GH |
3829 | { |
3830 | // We have a special case where we recognize | |
3831 | // print(@hist_foo(bar)) as a magic print-the-histogram | |
3832 | // construct. This is sort of gross but it avoids | |
3833 | // promoting histogram references to typeful | |
3834 | // expressions. | |
dff50e09 | 3835 | |
1bbeef03 GH |
3836 | hop = NULL; |
3837 | t = parse_hist_op_or_bare_name(hop, name); | |
3838 | assert(hop); | |
dff50e09 | 3839 | |
1bbeef03 GH |
3840 | // It is, sadly, possible that even while parsing a |
3841 | // hist_op, we *mis-guessed* and the user wishes to | |
3842 | // print(@hist_op(foo)[bucket]), a scalar. In that case | |
3843 | // we must parse the arrayindex and print an expression. | |
839325a1 JS |
3844 | // |
3845 | // XXX: This still fails if the arrayindex is part of a | |
3846 | // larger expression. To really handle everything, we'd | |
3847 | // need to push back all the hist tokens start over. | |
dff50e09 | 3848 | |
1bbeef03 GH |
3849 | if (!peek_op ("[")) |
3850 | fmt->hist = hop; | |
3851 | else | |
3852 | { | |
3853 | // This is simplified version of the | |
3854 | // multi-array-index parser below, because we can | |
3855 | // only ever have one index on a histogram anyways. | |
3856 | expect_op("["); | |
3857 | struct arrayindex* ai = new arrayindex; | |
3858 | ai->tok = t; | |
3859 | ai->base = hop; | |
3860 | ai->indexes.push_back (parse_expression ()); | |
3861 | expect_op("]"); | |
3862 | fmt->args.push_back(ai); | |
839325a1 JS |
3863 | |
3864 | // Consume any subsequent arguments. | |
1efdc9a9 | 3865 | while (expect_op_any ({",", ")"}) != ")") |
839325a1 | 3866 | { |
839325a1 JS |
3867 | expression *e = parse_expression (); |
3868 | fmt->args.push_back(e); | |
3869 | } | |
1bbeef03 | 3870 | } |
a4636912 | 3871 | } |
d7f3e0c5 | 3872 | else |
d02548c0 | 3873 | { |
3cb17058 | 3874 | int min_args = 0; |
80cb29eb | 3875 | bool consumed_arg = false; |
3cb17058 JS |
3876 | if (fmt->print_with_format) |
3877 | { | |
3878 | // Consume and convert a format string. Agreement between the | |
3879 | // format string and the arguments is postponed to the | |
3880 | // typechecking phase. | |
c92d3b42 | 3881 | literal_string* ls = parse_literal_string(); |
47d349b1 | 3882 | fmt->raw_components = ls->value; |
c92d3b42 FCE |
3883 | delete ls; |
3884 | fmt->components = print_format::string_to_components (fmt->raw_components); | |
80cb29eb | 3885 | consumed_arg = true; |
3cb17058 JS |
3886 | } |
3887 | else if (fmt->print_with_delim) | |
3888 | { | |
3889 | // Consume a delimiter to separate arguments. | |
c92d3b42 | 3890 | literal_string* ls = parse_literal_string(); |
d70e3afe | 3891 | fmt->delimiter = ls->value; |
c92d3b42 | 3892 | delete ls; |
80cb29eb JL |
3893 | consumed_arg = true; |
3894 | min_args = 2; // so that the delim is used at least once | |
3cb17058 | 3895 | } |
80cb29eb | 3896 | else if (!fmt->print_with_newline) |
3cb17058 | 3897 | { |
80cb29eb JL |
3898 | // If we are not printing with a format string, nor with a |
3899 | // delim, nor with a newline, then it's either print() or | |
3900 | // sprint(), both of which require at least one argument (of | |
3901 | // any type). | |
3902 | min_args = 1; | |
3cb17058 JS |
3903 | } |
3904 | ||
3905 | // Consume any subsequent arguments. | |
3906 | while (min_args || !peek_op (")")) | |
3907 | { | |
f0a49a41 FL |
3908 | // ')' is not possible here but we want to output a nicer |
3909 | // parser error message. | |
80cb29eb | 3910 | if (consumed_arg) |
1efdc9a9 | 3911 | (void) expect_op_any({",", ")"}); |
3cb17058 JS |
3912 | expression *e = parse_expression (); |
3913 | fmt->args.push_back(e); | |
80cb29eb | 3914 | consumed_arg = true; |
3cb17058 JS |
3915 | if (min_args) |
3916 | --min_args; | |
3917 | } | |
1efdc9a9 | 3918 | expect_op(")"); |
d02548c0 | 3919 | } |
d02548c0 GH |
3920 | return fmt; |
3921 | } | |
dff50e09 | 3922 | |
d02548c0 GH |
3923 | else if (peek_op ("(")) // function call |
3924 | { | |
731a5359 | 3925 | swallow (); |
d02548c0 GH |
3926 | struct functioncall* f = new functioncall; |
3927 | f->tok = t; | |
3928 | f->function = name; | |
3929 | // Allow empty actual parameter list | |
3930 | if (peek_op (")")) | |
3931 | { | |
731a5359 | 3932 | swallow (); |
d02548c0 GH |
3933 | return f; |
3934 | } | |
3935 | while (1) | |
3936 | { | |
3937 | f->args.push_back (parse_expression ()); | |
1efdc9a9 FL |
3938 | interned_string op = expect_op_any({")", ","}); |
3939 | if (op == ")") | |
3940 | break; | |
3941 | else if (op == ",") | |
3942 | continue; | |
d02548c0 GH |
3943 | } |
3944 | return f; | |
3945 | } | |
3946 | ||
3947 | else | |
3948 | { | |
3949 | sym = new symbol; | |
3950 | sym->name = name; | |
3951 | sym->tok = t; | |
d7f3e0c5 | 3952 | } |
0fefb486 | 3953 | } |
dff50e09 FCE |
3954 | |
3955 | // By now, either we had a hist_op in the first place, or else | |
d02548c0 GH |
3956 | // we had a plain word and it was converted to a symbol. |
3957 | ||
70c743d8 | 3958 | assert (!hop != !sym); // logical XOR |
d02548c0 GH |
3959 | |
3960 | // All that remains is to check for array indexing | |
3961 | ||
d7f3e0c5 | 3962 | if (peek_op ("[")) // array |
2f1a1aea | 3963 | { |
731a5359 | 3964 | swallow (); |
2f1a1aea | 3965 | struct arrayindex* ai = new arrayindex; |
d02548c0 GH |
3966 | ai->tok = t; |
3967 | ||
3968 | if (hop) | |
3969 | ai->base = hop; | |
3970 | else | |
3971 | ai->base = sym; | |
3972 | ||
2f1a1aea FCE |
3973 | while (1) |
3974 | { | |
a98c930b AJ |
3975 | if (peek_op("*")) |
3976 | { | |
45af9d1b AJ |
3977 | swallow(); |
3978 | ai->indexes.push_back (NULL); | |
a98c930b AJ |
3979 | } |
3980 | else | |
3981 | ai->indexes.push_back (parse_expression ()); | |
1efdc9a9 FL |
3982 | interned_string op = expect_op_any({"]", ","}); |
3983 | if (op == "]") | |
3984 | break; | |
616c2df5 | 3985 | else if (op == ",") |
1efdc9a9 | 3986 | continue; |
2f1a1aea | 3987 | } |
0fb0cac9 | 3988 | |
2f1a1aea FCE |
3989 | return ai; |
3990 | } | |
d02548c0 GH |
3991 | |
3992 | // If we got to here, we *should* have a symbol; if we have | |
3993 | // a hist_op on its own, it doesn't count as an expression, | |
3994 | // so we throw a parse error. | |
3995 | ||
3996 | if (hop) | |
f0454224 | 3997 | throw PARSE_ERROR(_("base histogram operator where expression expected"), t); |
dff50e09 FCE |
3998 | |
3999 | return sym; | |
2f1a1aea | 4000 | } |
56099f08 | 4001 | |
0fb0cac9 JS |
4002 | // Parse a $var. |
4003 | target_symbol* parser::parse_target_symbol () | |
30263a73 | 4004 | { |
0fb0cac9 JS |
4005 | const token* t = next (); |
4006 | if (t->type == tok_identifier && t->content[0]=='$') | |
d48afc20 | 4007 | { |
0fb0cac9 JS |
4008 | // target_symbol time |
4009 | target_symbol *tsym = new target_symbol; | |
4010 | tsym->tok = t; | |
a3e980f9 | 4011 | tsym->name = t->content; |
0fb0cac9 | 4012 | return tsym; |
d48afc20 JS |
4013 | } |
4014 | ||
0fb0cac9 JS |
4015 | throw PARSE_ERROR (_("expected $var")); |
4016 | } | |
4017 | ||
4018 | ||
4019 | // Parse a @cast. | |
4020 | cast_op* parser::parse_cast_op () | |
4021 | { | |
4022 | const token* t = next (); | |
06219d6f | 4023 | if (t->type == tok_operator && t->content == "@cast") |
30263a73 FCE |
4024 | { |
4025 | cast_op *cop = new cast_op; | |
4026 | cop->tok = t; | |
a3e980f9 | 4027 | cop->name = t->content; |
30263a73 FCE |
4028 | expect_op("("); |
4029 | cop->operand = parse_expression (); | |
4030 | expect_op(","); | |
7f6b80bd | 4031 | expect_unknown(tok_string, cop->type_name); |
0fb0cac9 JS |
4032 | if (cop->type_name.empty()) |
4033 | throw PARSE_ERROR (_("expected non-empty string")); | |
30263a73 FCE |
4034 | if (peek_op (",")) |
4035 | { | |
731a5359 | 4036 | swallow (); |
30263a73 FCE |
4037 | expect_unknown(tok_string, cop->module); |
4038 | } | |
4039 | expect_op(")"); | |
30263a73 FCE |
4040 | return cop; |
4041 | } | |
4042 | ||
0fb0cac9 JS |
4043 | throw PARSE_ERROR (_("expected @cast")); |
4044 | } | |
4045 | ||
30263a73 | 4046 | |
0fb0cac9 JS |
4047 | // Parse a @var. |
4048 | atvar_op* parser::parse_atvar_op () | |
4049 | { | |
4050 | const token* t = next (); | |
06219d6f | 4051 | if (t->type == tok_operator && t->content == "@var") |
cc9001af | 4052 | { |
bd1fcbad YZ |
4053 | atvar_op *aop = new atvar_op; |
4054 | aop->tok = t; | |
a3e980f9 | 4055 | aop->name = t->content; |
cc9001af | 4056 | expect_op("("); |
bd1fcbad YZ |
4057 | expect_unknown(tok_string, aop->target_name); |
4058 | size_t found_at = aop->target_name.find("@"); | |
bfa7e523 | 4059 | if (found_at != string::npos) |
bd1fcbad | 4060 | aop->cu_name = aop->target_name.substr(found_at + 1); |
bfa7e523 | 4061 | else |
bd1fcbad YZ |
4062 | aop->cu_name = ""; |
4063 | if (peek_op (",")) | |
4064 | { | |
4065 | swallow (); | |
4066 | expect_unknown (tok_string, aop->module); | |
4067 | } | |
4068 | else | |
4069 | aop->module = ""; | |
cc9001af | 4070 | expect_op(")"); |
bd1fcbad | 4071 | return aop; |
cc9001af MW |
4072 | } |
4073 | ||
0fb0cac9 | 4074 | throw PARSE_ERROR (_("expected @var")); |
30263a73 FCE |
4075 | } |
4076 | ||
4077 | ||
4078 | // Parse a @defined(). Given head token has already been consumed. | |
4079 | expression* parser::parse_defined_op (const token* t) | |
4080 | { | |
4081 | defined_op* dop = new defined_op; | |
4082 | dop->tok = t; | |
4083 | expect_op("("); | |
0fb0cac9 | 4084 | dop->operand = parse_expression (); |
30263a73 FCE |
4085 | expect_op(")"); |
4086 | return dop; | |
4087 | } | |
4088 | ||
4089 | ||
0a7eb12d MC |
4090 | // Parse a @const(). Given head token has already been consumed. |
4091 | expression* parser::parse_const_op (const token* t) | |
4092 | { | |
4093 | if (! privileged) | |
4094 | throw PARSE_ERROR (_("using @const operator not permitted; need stap -g"), | |
4095 | false /* don't skip tokens for parse resumption */); | |
4096 | ||
a7b0fd27 MC |
4097 | interned_string cnst; |
4098 | embedded_expr *ee = new embedded_expr; | |
4099 | ee->tok = t; | |
0a7eb12d | 4100 | expect_op("("); |
a7b0fd27 MC |
4101 | expect_unknown(tok_string, cnst); |
4102 | if(cnst.empty()) | |
0a7eb12d MC |
4103 | throw PARSE_ERROR (_("expected non-empty string")); |
4104 | expect_op(")"); | |
a7b0fd27 MC |
4105 | ee->code = string("/* pure */ /* unprivileged */ /* stable */ ") + string(cnst); |
4106 | return ee; | |
0a7eb12d MC |
4107 | } |
4108 | ||
4109 | ||
8cc799a5 JS |
4110 | // Parse a @entry(). Given head token has already been consumed. |
4111 | expression* parser::parse_entry_op (const token* t) | |
4112 | { | |
4113 | entry_op* eop = new entry_op; | |
4114 | eop->tok = t; | |
4115 | expect_op("("); | |
4116 | eop->operand = parse_expression (); | |
4117 | expect_op(")"); | |
4118 | return eop; | |
4119 | } | |
4120 | ||
4121 | ||
3689db05 SC |
4122 | // Parse a @perf(). Given head token has already been consumed. |
4123 | expression* parser::parse_perf_op (const token* t) | |
4124 | { | |
4125 | perf_op* pop = new perf_op; | |
4126 | pop->tok = t; | |
4127 | expect_op("("); | |
ace7c23f FCE |
4128 | pop->operand = parse_literal_string (); |
4129 | if (pop->operand->value == "") | |
f0454224 | 4130 | throw PARSE_ERROR (_("expected non-empty string")); |
3689db05 SC |
4131 | expect_op(")"); |
4132 | return pop; | |
4133 | } | |
4134 | ||
4135 | ||
0fb0cac9 JS |
4136 | bool |
4137 | parser::peek_target_symbol_components () | |
4138 | { | |
3ddcf938 JS |
4139 | const token * t = peek (); |
4140 | return t && | |
4141 | ((t->type == tok_operator && (t->content == "->" || t->content == "[")) | |
4142 | || (t->type == tok_identifier && | |
4143 | t->content.find_first_not_of('$') == string::npos)); | |
0fb0cac9 | 4144 | } |
30263a73 | 4145 | |
81931eab JS |
4146 | void |
4147 | parser::parse_target_symbol_components (target_symbol* e) | |
4148 | { | |
5f36109e JS |
4149 | bool pprint = false; |
4150 | ||
4151 | // check for pretty-print in the form $foo$ | |
47d349b1 | 4152 | string base = e->name; |
5f36109e JS |
4153 | size_t pprint_pos = base.find_last_not_of('$'); |
4154 | if (0 < pprint_pos && pprint_pos < base.length() - 1) | |
4155 | { | |
4156 | string pprint_val = base.substr(pprint_pos + 1); | |
4157 | base.erase(pprint_pos + 1); | |
47d349b1 | 4158 | e->name = base; |
5f36109e JS |
4159 | e->components.push_back (target_symbol::component(e->tok, pprint_val, true)); |
4160 | pprint = true; | |
4161 | } | |
4162 | ||
4163 | while (!pprint) | |
81931eab | 4164 | { |
81931eab JS |
4165 | if (peek_op ("->")) |
4166 | { | |
c67847a0 | 4167 | const token* t = next(); |
b1f2b0e8 | 4168 | interned_string member; |
c67847a0 | 4169 | expect_ident_or_keyword (member); |
5f36109e JS |
4170 | |
4171 | // check for pretty-print in the form $foo->$ or $foo->bar$ | |
4172 | pprint_pos = member.find_last_not_of('$'); | |
b1f2b0e8 | 4173 | interned_string pprint_val; |
5f36109e JS |
4174 | if (pprint_pos == string::npos || pprint_pos < member.length() - 1) |
4175 | { | |
4176 | pprint_val = member.substr(pprint_pos + 1); | |
b1f2b0e8 | 4177 | member = member.substr(0, pprint_pos + 1); |
5f36109e JS |
4178 | pprint = true; |
4179 | } | |
4180 | ||
4181 | if (!member.empty()) | |
4182 | e->components.push_back (target_symbol::component(t, member)); | |
4183 | if (pprint) | |
4184 | e->components.push_back (target_symbol::component(t, pprint_val, true)); | |
81931eab JS |
4185 | } |
4186 | else if (peek_op ("[")) | |
4187 | { | |
c67847a0 | 4188 | const token* t = next(); |
6fda2dff JS |
4189 | expression* index = parse_expression(); |
4190 | literal_number* ln = dynamic_cast<literal_number*>(index); | |
4191 | if (ln) | |
4192 | e->components.push_back (target_symbol::component(t, ln->value)); | |
4193 | else | |
4194 | e->components.push_back (target_symbol::component(t, index)); | |
81931eab | 4195 | expect_op ("]"); |
81931eab JS |
4196 | } |
4197 | else | |
4198 | break; | |
4199 | } | |
5f36109e JS |
4200 | |
4201 | if (!pprint) | |
4202 | { | |
4203 | // check for pretty-print in the form $foo $ | |
4204 | // i.e. as a separate token, esp. for $foo[i]$ and @cast(...)$ | |
4205 | const token* t = peek(); | |
3819d181 | 4206 | if (t != NULL && t->type == tok_identifier && |
5f36109e JS |
4207 | t->content.find_first_not_of('$') == string::npos) |
4208 | { | |
4209 | t = next(); | |
47d349b1 | 4210 | e->components.push_back (target_symbol::component(t, t->content, true)); |
5f36109e JS |
4211 | pprint = true; |
4212 | } | |
4213 | } | |
4214 | ||
4215 | if (pprint && (peek_op ("->") || peek_op("["))) | |
f0454224 | 4216 | throw PARSE_ERROR(_("-> and [ are not accepted for a pretty-printing variable")); |
81931eab JS |
4217 | } |
4218 | ||
73267b89 | 4219 | /* vim: set sw=2 ts=8 cino=>4,n-2,{2,^-2,t0,(0,u0,w1,M1 : */ |