]>
Commit | Line | Data |
---|---|---|
2f1a1aea | 1 | // recursive descent parser for systemtap scripts |
1e6e9ec1 | 2 | // Copyright (C) 2005-2011 Red Hat Inc. |
77a5c1f9 | 3 | // Copyright (C) 2006 Intel Corporation. |
5811366a | 4 | // Copyright (C) 2007 Bull S.A.S |
69c68955 FCE |
5 | // |
6 | // This file is part of systemtap, and is free software. You can | |
7 | // redistribute it and/or modify it under the terms of the GNU General | |
8 | // Public License (GPL); either version 2, or (at your option) any | |
9 | // later version. | |
2f1a1aea | 10 | |
2b066ec1 | 11 | #include "config.h" |
2f1a1aea FCE |
12 | #include "staptree.h" |
13 | #include "parse.h" | |
177a8ead | 14 | #include "session.h" |
3f99432c FCE |
15 | #include "util.h" |
16 | ||
2b066ec1 | 17 | #include <iostream> |
eacb10ce | 18 | |
2b066ec1 | 19 | #include <fstream> |
2f1a1aea | 20 | #include <cctype> |
9c0c0e46 | 21 | #include <cstdlib> |
29e64872 | 22 | #include <cassert> |
9c0c0e46 FCE |
23 | #include <cerrno> |
24 | #include <climits> | |
57b73400 | 25 | #include <sstream> |
f74fb737 | 26 | #include <cstring> |
3f99432c | 27 | #include <cctype> |
eacb10ce FCE |
28 | #include <iterator> |
29 | ||
7a468d68 FCE |
30 | extern "C" { |
31 | #include <fnmatch.h> | |
32 | } | |
2f1a1aea FCE |
33 | |
34 | using namespace std; | |
35 | ||
c18f07f8 JS |
36 | |
37 | class lexer | |
38 | { | |
39 | public: | |
c5be7511 | 40 | bool ate_comment; // the most recent token followed a comment |
c18f07f8 JS |
41 | token* scan (bool wildcard=false); |
42 | lexer (istream&, const string&, systemtap_session&); | |
43 | void set_current_file (stapfile* f); | |
44 | ||
45 | private: | |
46 | inline int input_get (); | |
47 | inline int input_peek (unsigned n=0); | |
48 | void input_put (const string&, const token*); | |
49 | string input_name; | |
50 | string input_contents; | |
51 | const char *input_pointer; // index into input_contents | |
52 | const char *input_end; | |
53 | unsigned cursor_suspend_count; | |
54 | unsigned cursor_suspend_line; | |
55 | unsigned cursor_suspend_column; | |
56 | unsigned cursor_line; | |
57 | unsigned cursor_column; | |
58 | systemtap_session& session; | |
59 | stapfile* current_file; | |
60 | static set<string> keywords; | |
61 | }; | |
62 | ||
63 | ||
64 | class parser | |
65 | { | |
66 | public: | |
67 | parser (systemtap_session& s, istream& i, bool p); | |
68 | parser (systemtap_session& s, const string& n, bool p); | |
69 | ~parser (); | |
70 | ||
71 | stapfile* parse (); | |
72 | ||
73 | private: | |
74 | typedef enum { | |
75 | PP_NONE, | |
76 | PP_KEEP_THEN, | |
77 | PP_SKIP_THEN, | |
78 | PP_KEEP_ELSE, | |
79 | PP_SKIP_ELSE, | |
80 | } pp_state_t; | |
81 | ||
82 | systemtap_session& session; | |
83 | string input_name; | |
84 | istream* free_input; | |
85 | lexer input; | |
86 | bool privileged; | |
87 | parse_context context; | |
88 | ||
89 | // preprocessing subordinate | |
90 | vector<pair<const token*, pp_state_t> > pp_state; | |
91 | const token* scan_pp (bool wildcard=false); | |
92 | const token* skip_pp (); | |
93 | ||
94 | // scanning state | |
95 | const token* last (); | |
96 | const token* next (bool wildcard=false); | |
97 | const token* peek (bool wildcard=false); | |
98 | ||
a07a2c28 | 99 | const token* systemtap_v_seen; |
c18f07f8 JS |
100 | const token* last_t; // the last value returned by peek() or next() |
101 | const token* next_t; // lookahead token | |
102 | ||
103 | // expectations | |
104 | const token* expect_known (token_type tt, string const & expected); | |
105 | const token* expect_unknown (token_type tt, string & target); | |
106 | const token* expect_unknown2 (token_type tt1, token_type tt2, | |
107 | string & target); | |
108 | ||
109 | // convenience forms | |
110 | const token* expect_op (string const & expected); | |
111 | const token* expect_kw (string const & expected); | |
112 | const token* expect_number (int64_t & expected); | |
113 | const token* expect_ident (string & target); | |
114 | const token* expect_ident_or_keyword (string & target); | |
115 | bool peek_op (string const & op); | |
116 | bool peek_kw (string const & kw); | |
117 | ||
118 | void print_error (const parse_error& pe); | |
119 | unsigned num_errors; | |
120 | ||
121 | private: // nonterminals | |
122 | void parse_probe (vector<probe*>&, vector<probe_alias*>&); | |
123 | void parse_global (vector<vardecl*>&, vector<probe*>&); | |
124 | void parse_functiondecl (vector<functiondecl*>&); | |
125 | embeddedcode* parse_embeddedcode (); | |
126 | probe_point* parse_probe_point (); | |
127 | literal* parse_literal (); | |
128 | block* parse_stmt_block (); | |
129 | try_block* parse_try_block (); | |
130 | statement* parse_statement (); | |
131 | if_statement* parse_if_statement (); | |
132 | for_loop* parse_for_loop (); | |
133 | for_loop* parse_while_loop (); | |
134 | foreach_loop* parse_foreach_loop (); | |
135 | expr_statement* parse_expr_statement (); | |
136 | return_statement* parse_return_statement (); | |
137 | delete_statement* parse_delete_statement (); | |
138 | next_statement* parse_next_statement (); | |
139 | break_statement* parse_break_statement (); | |
140 | continue_statement* parse_continue_statement (); | |
141 | indexable* parse_indexable (); | |
142 | const token *parse_hist_op_or_bare_name (hist_op *&hop, string &name); | |
143 | target_symbol *parse_target_symbol (const token* t); | |
8cc799a5 | 144 | expression* parse_entry_op (const token* t); |
c18f07f8 JS |
145 | expression* parse_defined_op (const token* t); |
146 | expression* parse_expression (); | |
147 | expression* parse_assignment (); | |
148 | expression* parse_ternary (); | |
149 | expression* parse_logical_or (); | |
150 | expression* parse_logical_and (); | |
151 | expression* parse_boolean_or (); | |
152 | expression* parse_boolean_xor (); | |
153 | expression* parse_boolean_and (); | |
154 | expression* parse_array_in (); | |
155 | expression* parse_comparison (); | |
156 | expression* parse_shift (); | |
157 | expression* parse_concatenation (); | |
158 | expression* parse_additive (); | |
159 | expression* parse_multiplicative (); | |
160 | expression* parse_unary (); | |
161 | expression* parse_crement (); | |
162 | expression* parse_value (); | |
163 | expression* parse_symbol (); | |
164 | ||
165 | void parse_target_symbol_components (target_symbol* e); | |
166 | }; | |
167 | ||
168 | ||
2f1a1aea FCE |
169 | // ------------------------------------------------------------------------ |
170 | ||
c18f07f8 JS |
171 | stapfile* |
172 | parse (systemtap_session& s, istream& i, bool pr) | |
173 | { | |
174 | parser p (s, i, pr); | |
175 | return p.parse (); | |
176 | } | |
177 | ||
178 | ||
179 | stapfile* | |
180 | parse (systemtap_session& s, const string& n, bool pr) | |
181 | { | |
182 | parser p (s, n, pr); | |
183 | return p.parse (); | |
184 | } | |
185 | ||
186 | // ------------------------------------------------------------------------ | |
bb2e3076 FCE |
187 | |
188 | ||
177a8ead FCE |
189 | parser::parser (systemtap_session& s, istream& i, bool p): |
190 | session (s), | |
24cb178f | 191 | input_name ("<input>"), free_input (0), |
213bee8f | 192 | input (i, input_name, s), privileged (p), |
a07a2c28 | 193 | context(con_unknown), systemtap_v_seen(0), last_t (0), next_t (0), num_errors (0) |
2f1a1aea FCE |
194 | { } |
195 | ||
177a8ead FCE |
196 | parser::parser (systemtap_session& s, const string& fn, bool p): |
197 | session (s), | |
2f1a1aea | 198 | input_name (fn), free_input (new ifstream (input_name.c_str(), ios::in)), |
213bee8f | 199 | input (* free_input, input_name, s), privileged (p), |
a07a2c28 | 200 | context(con_unknown), systemtap_v_seen(0), last_t (0), next_t (0), num_errors (0) |
2f1a1aea FCE |
201 | { } |
202 | ||
203 | parser::~parser() | |
204 | { | |
205 | if (free_input) delete free_input; | |
206 | } | |
207 | ||
d7f3e0c5 GH |
208 | static string |
209 | tt2str(token_type tt) | |
210 | { | |
211 | switch (tt) | |
212 | { | |
213 | case tok_junk: return "junk"; | |
214 | case tok_identifier: return "identifier"; | |
215 | case tok_operator: return "operator"; | |
216 | case tok_string: return "string"; | |
217 | case tok_number: return "number"; | |
218 | case tok_embedded: return "embedded-code"; | |
6e213f58 | 219 | case tok_keyword: return "keyword"; |
d7f3e0c5 GH |
220 | } |
221 | return "unknown token"; | |
222 | } | |
82919855 | 223 | |
0323ed4d WC |
224 | ostream& |
225 | operator << (ostream& o, const source_loc& loc) | |
226 | { | |
a704a23b | 227 | o << loc.file->name << ":" |
0323ed4d WC |
228 | << loc.line << ":" |
229 | << loc.column; | |
230 | ||
231 | return o; | |
232 | } | |
233 | ||
56099f08 FCE |
234 | ostream& |
235 | operator << (ostream& o, const token& t) | |
236 | { | |
d7f3e0c5 | 237 | o << tt2str(t.type); |
56099f08 | 238 | |
6e213f58 | 239 | if (t.type != tok_embedded && t.type != tok_keyword) // XXX: other types? |
56099f08 | 240 | { |
24cb178f FCE |
241 | o << " '"; |
242 | for (unsigned i=0; i<t.content.length(); i++) | |
243 | { | |
244 | char c = t.content[i]; | |
245 | o << (isprint (c) ? c : '?'); | |
246 | } | |
247 | o << "'"; | |
56099f08 | 248 | } |
56099f08 | 249 | |
dff50e09 | 250 | o << " at " |
0323ed4d | 251 | << t.location; |
56099f08 FCE |
252 | |
253 | return o; | |
254 | } | |
255 | ||
256 | ||
dff50e09 | 257 | void |
2f1a1aea FCE |
258 | parser::print_error (const parse_error &pe) |
259 | { | |
1b1b4ceb | 260 | string align_parse_error (" "); |
2677d2fb | 261 | cerr << _("parse error: ") << pe.what () << endl; |
2f1a1aea | 262 | |
177a8ead FCE |
263 | if (pe.tok) |
264 | { | |
2677d2fb | 265 | cerr << _("\tat: ") << *pe.tok << endl; |
1b1b4ceb | 266 | session.print_error_source (cerr, align_parse_error, pe.tok); |
177a8ead | 267 | } |
2f1a1aea | 268 | else |
177a8ead FCE |
269 | { |
270 | const token* t = last_t; | |
271 | if (t) | |
1b1b4ceb | 272 | { |
2677d2fb | 273 | cerr << _("\tsaw: ") << *t << endl; |
1b1b4ceb RA |
274 | session.print_error_source (cerr, align_parse_error, t); |
275 | } | |
177a8ead | 276 | else |
ce0f6648 | 277 | cerr << _("\tsaw: ") << input_name << " EOF" << endl; |
177a8ead | 278 | } |
2f1a1aea FCE |
279 | |
280 | // XXX: make it possible to print the last input line, | |
281 | // so as to line up an arrow with the specific error column | |
282 | ||
283 | num_errors ++; | |
284 | } | |
285 | ||
286 | ||
dff50e09 | 287 | const token* |
2f1a1aea FCE |
288 | parser::last () |
289 | { | |
290 | return last_t; | |
291 | } | |
292 | ||
293 | ||
c434ec7e FCE |
294 | |
295 | template <typename OPERAND> | |
296 | bool eval_comparison (const OPERAND& lhs, const token* op, const OPERAND& rhs) | |
297 | { | |
298 | if (op->type == tok_operator && op->content == "<=") | |
299 | { return lhs <= rhs; } | |
300 | else if (op->type == tok_operator && op->content == ">=") | |
301 | { return lhs >= rhs; } | |
302 | else if (op->type == tok_operator && op->content == "<") | |
303 | { return lhs < rhs; } | |
304 | else if (op->type == tok_operator && op->content == ">") | |
305 | { return lhs > rhs; } | |
306 | else if (op->type == tok_operator && op->content == "==") | |
307 | { return lhs == rhs; } | |
308 | else if (op->type == tok_operator && op->content == "!=") | |
309 | { return lhs != rhs; } | |
310 | else | |
2677d2fb | 311 | throw parse_error (_("expected comparison operator"), op); |
c434ec7e FCE |
312 | } |
313 | ||
314 | ||
177a8ead FCE |
315 | // Here, we perform on-the-fly preprocessing. |
316 | // The basic form is %( CONDITION %? THEN-TOKENS %: ELSE-TOKENS %) | |
44ce8ed5 FCE |
317 | // where CONDITION is: kernel_v[r] COMPARISON-OP "version-string" |
318 | // or: arch COMPARISON-OP "arch-string" | |
db135493 | 319 | // or: systemtap_v COMPARISON-OP "version-string" |
561079c8 | 320 | // or: CONFIG_foo COMPARISON-OP "config-string" |
717a457b | 321 | // or: CONFIG_foo COMPARISON-OP number |
4227f98d | 322 | // or: CONFIG_foo COMPARISON-OP CONFIG_bar |
5811366a FCE |
323 | // or: "string1" COMPARISON-OP "string2" |
324 | // or: number1 COMPARISON-OP number2 | |
44ce8ed5 | 325 | // The %: ELSE-TOKENS part is optional. |
177a8ead FCE |
326 | // |
327 | // e.g. %( kernel_v > "2.5" %? "foo" %: "baz" %) | |
c434ec7e | 328 | // e.g. %( arch != "i?86" %? "foo" %: "baz" %) |
561079c8 | 329 | // e.g. %( CONFIG_foo %? "foo" %: "baz" %) |
177a8ead FCE |
330 | // |
331 | // Up to an entire %( ... %) expression is processed by a single call | |
332 | // to this function. Tokens included by any nested conditions are | |
333 | // enqueued in a private vector. | |
334 | ||
335 | bool eval_pp_conditional (systemtap_session& s, | |
336 | const token* l, const token* op, const token* r) | |
337 | { | |
44ce8ed5 | 338 | if (l->type == tok_identifier && (l->content == "kernel_v" || |
db135493 FCE |
339 | l->content == "kernel_vr" || |
340 | l->content == "systemtap_v")) | |
44ce8ed5 | 341 | { |
db135493 | 342 | if (! (r->type == tok_string)) |
2677d2fb | 343 | throw parse_error (_("expected string literal"), r); |
db135493 | 344 | |
44ce8ed5 | 345 | string target_kernel_vr = s.kernel_release; |
197a4d62 | 346 | string target_kernel_v = s.kernel_base_release; |
db135493 | 347 | string target; |
dff50e09 | 348 | |
db135493 FCE |
349 | if (l->content == "kernel_v") target = target_kernel_v; |
350 | else if (l->content == "kernel_vr") target = target_kernel_vr; | |
351 | else if (l->content == "systemtap_v") target = s.compatible; | |
352 | else assert (0); | |
7a468d68 | 353 | |
7a468d68 FCE |
354 | string query = r->content; |
355 | bool rhs_wildcard = (strpbrk (query.c_str(), "*?[") != 0); | |
356 | ||
44ce8ed5 FCE |
357 | // collect acceptable strverscmp results. |
358 | int rvc_ok1, rvc_ok2; | |
7a468d68 | 359 | bool wc_ok = false; |
44ce8ed5 FCE |
360 | if (op->type == tok_operator && op->content == "<=") |
361 | { rvc_ok1 = -1; rvc_ok2 = 0; } | |
362 | else if (op->type == tok_operator && op->content == ">=") | |
363 | { rvc_ok1 = 1; rvc_ok2 = 0; } | |
364 | else if (op->type == tok_operator && op->content == "<") | |
365 | { rvc_ok1 = -1; rvc_ok2 = -1; } | |
366 | else if (op->type == tok_operator && op->content == ">") | |
367 | { rvc_ok1 = 1; rvc_ok2 = 1; } | |
368 | else if (op->type == tok_operator && op->content == "==") | |
7a468d68 | 369 | { rvc_ok1 = 0; rvc_ok2 = 0; wc_ok = true; } |
44ce8ed5 | 370 | else if (op->type == tok_operator && op->content == "!=") |
7a468d68 | 371 | { rvc_ok1 = -1; rvc_ok2 = 1; wc_ok = true; } |
44ce8ed5 | 372 | else |
2677d2fb | 373 | throw parse_error (_("expected comparison operator"), op); |
7a468d68 FCE |
374 | |
375 | if ((!wc_ok) && rhs_wildcard) | |
2677d2fb | 376 | throw parse_error (_("wildcard not allowed with order comparison operators"), op); |
7a468d68 FCE |
377 | |
378 | if (rhs_wildcard) | |
379 | { | |
380 | int rvc_result = fnmatch (query.c_str(), target.c_str(), | |
381 | FNM_NOESCAPE); // spooky | |
382 | bool badness = (rvc_result == 0) ^ (op->content == "=="); | |
383 | return !badness; | |
384 | } | |
385 | else | |
386 | { | |
387 | int rvc_result = strverscmp (target.c_str(), query.c_str()); | |
388 | // normalize rvc_result | |
389 | if (rvc_result < 0) rvc_result = -1; | |
390 | if (rvc_result > 0) rvc_result = 1; | |
391 | return (rvc_result == rvc_ok1 || rvc_result == rvc_ok2); | |
392 | } | |
44ce8ed5 FCE |
393 | } |
394 | else if (l->type == tok_identifier && l->content == "arch") | |
395 | { | |
396 | string target_architecture = s.architecture; | |
397 | if (! (r->type == tok_string)) | |
2677d2fb | 398 | throw parse_error (_("expected string literal"), r); |
44ce8ed5 | 399 | string query_architecture = r->content; |
dff50e09 | 400 | |
7a468d68 FCE |
401 | int nomatch = fnmatch (query_architecture.c_str(), |
402 | target_architecture.c_str(), | |
403 | FNM_NOESCAPE); // still spooky | |
404 | ||
561079c8 FCE |
405 | bool result; |
406 | if (op->type == tok_operator && op->content == "==") | |
407 | result = !nomatch; | |
408 | else if (op->type == tok_operator && op->content == "!=") | |
409 | result = nomatch; | |
410 | else | |
2677d2fb | 411 | throw parse_error (_("expected '==' or '!='"), op); |
561079c8 FCE |
412 | |
413 | return result; | |
414 | } | |
60d98537 | 415 | else if (l->type == tok_identifier && startswith(l->content, "CONFIG_")) |
561079c8 | 416 | { |
717a457b MW |
417 | if (r->type == tok_string) |
418 | { | |
419 | string lhs = s.kernel_config[l->content]; // may be empty | |
420 | string rhs = r->content; | |
561079c8 | 421 | |
717a457b | 422 | int nomatch = fnmatch (rhs.c_str(), lhs.c_str(), FNM_NOESCAPE); // still spooky |
561079c8 | 423 | |
717a457b MW |
424 | bool result; |
425 | if (op->type == tok_operator && op->content == "==") | |
426 | result = !nomatch; | |
427 | else if (op->type == tok_operator && op->content == "!=") | |
428 | result = nomatch; | |
429 | else | |
2677d2fb | 430 | throw parse_error (_("expected '==' or '!='"), op); |
dff50e09 | 431 | |
717a457b MW |
432 | return result; |
433 | } | |
434 | else if (r->type == tok_number) | |
435 | { | |
436 | const char* startp = s.kernel_config[l->content].c_str (); | |
437 | char* endp = (char*) startp; | |
438 | errno = 0; | |
439 | int64_t lhs = (int64_t) strtoll (startp, & endp, 0); | |
440 | if (errno == ERANGE || errno == EINVAL || *endp != '\0') | |
441 | throw parse_error ("Config option value not a number", l); | |
442 | ||
443 | int64_t rhs = lex_cast<int64_t>(r->content); | |
444 | return eval_comparison (lhs, op, rhs); | |
445 | } | |
4227f98d | 446 | else if (r->type == tok_identifier |
60d98537 | 447 | && startswith(r->content, "CONFIG_")) |
4227f98d MW |
448 | { |
449 | // First try to convert both to numbers, | |
450 | // otherwise threat both as strings. | |
451 | const char* startp = s.kernel_config[l->content].c_str (); | |
452 | char* endp = (char*) startp; | |
453 | errno = 0; | |
454 | int64_t val = (int64_t) strtoll (startp, & endp, 0); | |
455 | if (errno != ERANGE && errno != EINVAL && *endp == '\0') | |
456 | { | |
457 | int64_t lhs = val; | |
458 | startp = s.kernel_config[r->content].c_str (); | |
459 | endp = (char*) startp; | |
460 | errno = 0; | |
461 | int64_t rhs = (int64_t) strtoll (startp, & endp, 0); | |
462 | if (errno != ERANGE && errno != EINVAL && *endp == '\0') | |
463 | return eval_comparison (lhs, op, rhs); | |
464 | } | |
465 | ||
466 | string lhs = s.kernel_config[l->content]; | |
467 | string rhs = s.kernel_config[r->content]; | |
468 | return eval_comparison (lhs, op, rhs); | |
469 | } | |
717a457b | 470 | else |
ce0f6648 | 471 | throw parse_error (_("expected string, number literal or other CONFIG_... as right side operand"), r); |
dff50e09 | 472 | } |
c434ec7e | 473 | else if (l->type == tok_string && r->type == tok_string) |
5811366a | 474 | { |
c434ec7e FCE |
475 | string lhs = l->content; |
476 | string rhs = r->content; | |
477 | return eval_comparison (lhs, op, rhs); | |
478 | // NB: no wildcarding option here | |
479 | } | |
480 | else if (l->type == tok_number && r->type == tok_number) | |
481 | { | |
482 | int64_t lhs = lex_cast<int64_t>(l->content); | |
483 | int64_t rhs = lex_cast<int64_t>(r->content); | |
484 | return eval_comparison (lhs, op, rhs); | |
7a468d68 | 485 | // NB: no wildcarding option here |
5811366a FCE |
486 | } |
487 | else if (l->type == tok_string && r->type == tok_number | |
488 | && op->type == tok_operator) | |
2677d2fb | 489 | throw parse_error (_("expected string literal as right value"), r); |
5811366a FCE |
490 | else if (l->type == tok_number && r->type == tok_string |
491 | && op->type == tok_operator) | |
2677d2fb | 492 | throw parse_error (_("expected number literal as right value"), r); |
c434ec7e | 493 | |
177a8ead | 494 | else |
2677d2fb LB |
495 | throw parse_error (_("expected 'arch' or 'kernel_v' or 'kernel_vr' or 'CONFIG_...'\n" |
496 | " or comparison between strings or integers"), l); | |
177a8ead FCE |
497 | } |
498 | ||
499 | ||
5811366a | 500 | // Only tokens corresponding to the TRUE statement must be expanded |
177a8ead | 501 | const token* |
3f847830 | 502 | parser::scan_pp (bool wildcard) |
177a8ead FCE |
503 | { |
504 | while (true) | |
505 | { | |
e92f2566 JS |
506 | pp_state_t pp = PP_NONE; |
507 | if (!pp_state.empty()) | |
508 | pp = pp_state.back().second; | |
509 | ||
510 | const token* t = 0; | |
511 | if (pp == PP_SKIP_THEN || pp == PP_SKIP_ELSE) | |
512 | t = skip_pp (); | |
513 | else | |
514 | t = input.scan (wildcard); | |
515 | ||
516 | if (t == 0) // EOF | |
177a8ead | 517 | { |
e92f2566 JS |
518 | if (pp != PP_NONE) |
519 | { | |
520 | t = pp_state.back().first; | |
521 | pp_state.pop_back(); // so skip_some doesn't keep trying to close this | |
ce0f6648 | 522 | //TRANSLATORS: 'conditional' meaning 'conditional preprocessing' |
2677d2fb | 523 | throw parse_error (_("incomplete conditional at end of file"), t); |
e92f2566 | 524 | } |
177a8ead FCE |
525 | return t; |
526 | } | |
527 | ||
e92f2566 JS |
528 | // misplaced preprocessor "then" |
529 | if (t->type == tok_operator && t->content == "%?") | |
2677d2fb | 530 | throw parse_error (_("incomplete conditional - missing '%('"), t); |
e92f2566 JS |
531 | |
532 | // preprocessor "else" | |
533 | if (t->type == tok_operator && t->content == "%:") | |
534 | { | |
535 | if (pp == PP_NONE) | |
2677d2fb | 536 | throw parse_error (_("incomplete conditional - missing '%('"), t); |
e92f2566 | 537 | if (pp == PP_KEEP_ELSE || pp == PP_SKIP_ELSE) |
2677d2fb | 538 | throw parse_error (_("invalid conditional - duplicate '%:'"), t); |
e92f2566 JS |
539 | |
540 | pp_state.back().second = (pp == PP_KEEP_THEN) ? | |
541 | PP_SKIP_ELSE : PP_KEEP_ELSE; | |
542 | delete t; | |
543 | continue; | |
544 | } | |
545 | ||
546 | // preprocessor close | |
547 | if (t->type == tok_operator && t->content == "%)") | |
548 | { | |
549 | if (pp == PP_NONE) | |
2677d2fb | 550 | throw parse_error (_("incomplete conditional - missing '%('"), t); |
e92f2566 | 551 | delete pp_state.back().first; |
a07a2c28 | 552 | delete t; //this is the closing bracket |
e92f2566 JS |
553 | pp_state.pop_back(); |
554 | continue; | |
555 | } | |
dff50e09 | 556 | |
177a8ead FCE |
557 | if (! (t->type == tok_operator && t->content == "%(")) // ordinary token |
558 | return t; | |
559 | ||
560 | // We have a %( - it's time to throw a preprocessing party! | |
561 | ||
2d7881bf PP |
562 | bool result = false; |
563 | bool and_result = true; | |
564 | const token *n = NULL; | |
565 | do { | |
566 | const token *l, *op, *r; | |
e92f2566 | 567 | l = input.scan (false); |
2d7881bf PP |
568 | op = input.scan (false); |
569 | r = input.scan (false); | |
570 | if (l == 0 || op == 0 || r == 0) | |
2677d2fb | 571 | throw parse_error (_("incomplete condition after '%('"), t); |
2d7881bf PP |
572 | // NB: consider generalizing to consume all tokens until %?, and |
573 | // passing that as a vector to an evaluator. | |
574 | ||
575 | // Do not evaluate the condition if we haven't expanded everything. | |
576 | // This may occur when having several recursive conditionals. | |
577 | and_result &= eval_pp_conditional (session, l, op, r); | |
a07a2c28 LB |
578 | if(l->content=="systemtap_v") |
579 | systemtap_v_seen=r; | |
580 | ||
581 | else | |
582 | delete r; | |
583 | ||
2d7881bf PP |
584 | delete l; |
585 | delete op; | |
2d7881bf PP |
586 | delete n; |
587 | ||
588 | n = input.scan (); | |
589 | if (n && n->type == tok_operator && n->content == "&&") | |
590 | continue; | |
591 | result |= and_result; | |
592 | and_result = true; | |
593 | if (! (n && n->type == tok_operator && n->content == "||")) | |
594 | break; | |
595 | } while (true); | |
3f847830 FCE |
596 | |
597 | /* | |
598 | clog << "PP eval (" << *t << ") == " << result << endl; | |
599 | */ | |
600 | ||
e92f2566 | 601 | const token *m = n; |
177a8ead | 602 | if (! (m && m->type == tok_operator && m->content == "%?")) |
2677d2fb | 603 | throw parse_error (_("expected '%?' marker for conditional"), t); |
70c743d8 | 604 | delete m; // "%?" |
177a8ead | 605 | |
e92f2566 JS |
606 | pp = result ? PP_KEEP_THEN : PP_SKIP_THEN; |
607 | pp_state.push_back (make_pair (t, pp)); | |
3f847830 | 608 | |
e92f2566 JS |
609 | // Now loop around to look for a real token. |
610 | } | |
611 | } | |
3f847830 | 612 | |
3f847830 | 613 | |
e92f2566 JS |
614 | // Skip over tokens and any errors, heeding |
615 | // only nested preprocessor starts and ends. | |
616 | const token* | |
617 | parser::skip_pp () | |
618 | { | |
619 | const token* t = 0; | |
620 | unsigned nesting = 0; | |
621 | do | |
622 | { | |
623 | try | |
624 | { | |
625 | t = input.scan (); | |
177a8ead | 626 | } |
e92f2566 | 627 | catch (const parse_error &e) |
70c743d8 | 628 | { |
e92f2566 | 629 | continue; |
70c743d8 | 630 | } |
e92f2566 JS |
631 | if (!t) |
632 | break; | |
633 | if (t->type == tok_operator && t->content == "%(") | |
634 | ++nesting; | |
635 | else if (nesting && t->type == tok_operator && t->content == "%)") | |
636 | --nesting; | |
637 | else if (!nesting && t->type == tok_operator && | |
638 | (t->content == "%:" || t->content == "%?" || t->content == "%)")) | |
639 | break; | |
640 | delete t; | |
177a8ead | 641 | } |
e92f2566 JS |
642 | while (true); |
643 | return t; | |
177a8ead FCE |
644 | } |
645 | ||
646 | ||
2f1a1aea | 647 | const token* |
0c218afb | 648 | parser::next (bool wildcard) |
2f1a1aea FCE |
649 | { |
650 | if (! next_t) | |
0c218afb | 651 | next_t = scan_pp (wildcard); |
2f1a1aea | 652 | if (! next_t) |
2677d2fb | 653 | throw parse_error (_("unexpected end-of-file")); |
2f1a1aea | 654 | |
2f1a1aea FCE |
655 | last_t = next_t; |
656 | // advance by zeroing next_t | |
657 | next_t = 0; | |
658 | return last_t; | |
659 | } | |
660 | ||
661 | ||
662 | const token* | |
0c218afb | 663 | parser::peek (bool wildcard) |
2f1a1aea FCE |
664 | { |
665 | if (! next_t) | |
0c218afb | 666 | next_t = scan_pp (wildcard); |
2f1a1aea FCE |
667 | |
668 | // don't advance by zeroing next_t | |
669 | last_t = next_t; | |
670 | return next_t; | |
671 | } | |
672 | ||
673 | ||
d7f3e0c5 GH |
674 | static inline bool |
675 | tok_is(token const * t, token_type tt, string const & expected) | |
676 | { | |
677 | return t && t->type == tt && t->content == expected; | |
678 | } | |
679 | ||
680 | ||
dff50e09 | 681 | const token* |
d7f3e0c5 GH |
682 | parser::expect_known (token_type tt, string const & expected) |
683 | { | |
684 | const token *t = next(); | |
57b73400 | 685 | if (! (t && t->type == tt && t->content == expected)) |
ce0f6648 | 686 | throw parse_error (_F("expected '%s'", expected.c_str())); |
d7f3e0c5 GH |
687 | return t; |
688 | } | |
689 | ||
690 | ||
dff50e09 | 691 | const token* |
d7f3e0c5 GH |
692 | parser::expect_unknown (token_type tt, string & target) |
693 | { | |
694 | const token *t = next(); | |
695 | if (!(t && t->type == tt)) | |
2677d2fb | 696 | throw parse_error (_("expected ") + tt2str(tt)); |
d7f3e0c5 GH |
697 | target = t->content; |
698 | return t; | |
699 | } | |
700 | ||
701 | ||
dff50e09 | 702 | const token* |
493ee224 DS |
703 | parser::expect_unknown2 (token_type tt1, token_type tt2, string & target) |
704 | { | |
705 | const token *t = next(); | |
706 | if (!(t && (t->type == tt1 || t->type == tt2))) | |
707 | throw parse_error ("expected " + tt2str(tt1) + " or " + tt2str(tt2)); | |
708 | target = t->content; | |
709 | return t; | |
710 | } | |
711 | ||
712 | ||
dff50e09 | 713 | const token* |
d7f3e0c5 GH |
714 | parser::expect_op (std::string const & expected) |
715 | { | |
716 | return expect_known (tok_operator, expected); | |
717 | } | |
718 | ||
719 | ||
dff50e09 | 720 | const token* |
d7f3e0c5 GH |
721 | parser::expect_kw (std::string const & expected) |
722 | { | |
f4fe2e93 | 723 | return expect_known (tok_keyword, expected); |
d7f3e0c5 GH |
724 | } |
725 | ||
dff50e09 | 726 | const token* |
e38723d2 | 727 | parser::expect_number (int64_t & value) |
57b73400 | 728 | { |
e38723d2 MH |
729 | bool neg = false; |
730 | const token *t = next(); | |
731 | if (t->type == tok_operator && t->content == "-") | |
732 | { | |
733 | neg = true; | |
734 | t = next (); | |
735 | } | |
736 | if (!(t && t->type == tok_number)) | |
2677d2fb | 737 | throw parse_error (_("expected number")); |
e38723d2 MH |
738 | |
739 | const char* startp = t->content.c_str (); | |
740 | char* endp = (char*) startp; | |
741 | ||
742 | // NB: we allow controlled overflow from LLONG_MIN .. ULLONG_MAX | |
743 | // Actually, this allows all the way from -ULLONG_MAX to ULLONG_MAX, | |
744 | // since the lexer only gives us positive digit strings, but we'll | |
745 | // limit it to LLONG_MIN when a '-' operator is fed into the literal. | |
746 | errno = 0; | |
747 | value = (int64_t) strtoull (startp, & endp, 0); | |
748 | if (errno == ERANGE || errno == EINVAL || *endp != '\0' | |
749 | || (neg && (unsigned long long) value > 9223372036854775808ULL) | |
750 | || (unsigned long long) value > 18446744073709551615ULL | |
751 | || value < -9223372036854775807LL-1) | |
2677d2fb | 752 | throw parse_error (_("number invalid or out of range")); |
dff50e09 | 753 | |
e38723d2 MH |
754 | if (neg) |
755 | value = -value; | |
756 | ||
757 | return t; | |
57b73400 GH |
758 | } |
759 | ||
d7f3e0c5 | 760 | |
dff50e09 | 761 | const token* |
d7f3e0c5 GH |
762 | parser::expect_ident (std::string & target) |
763 | { | |
764 | return expect_unknown (tok_identifier, target); | |
765 | } | |
766 | ||
767 | ||
dff50e09 | 768 | const token* |
493ee224 DS |
769 | parser::expect_ident_or_keyword (std::string & target) |
770 | { | |
771 | return expect_unknown2 (tok_identifier, tok_keyword, target); | |
772 | } | |
773 | ||
774 | ||
dff50e09 | 775 | bool |
d7f3e0c5 GH |
776 | parser::peek_op (std::string const & op) |
777 | { | |
778 | return tok_is (peek(), tok_operator, op); | |
779 | } | |
780 | ||
781 | ||
dff50e09 | 782 | bool |
d7f3e0c5 GH |
783 | parser::peek_kw (std::string const & kw) |
784 | { | |
785 | return tok_is (peek(), tok_identifier, kw); | |
786 | } | |
787 | ||
788 | ||
789 | ||
66c7d4c1 | 790 | lexer::lexer (istream& input, const string& in, systemtap_session& s): |
c5be7511 | 791 | ate_comment(false), input_name (in), input_pointer (0), input_end (0), |
9300f661 JS |
792 | cursor_suspend_count(0), cursor_suspend_line (1), cursor_suspend_column (1), |
793 | cursor_line (1), cursor_column (1), | |
66c7d4c1 | 794 | session(s), current_file (0) |
eacb10ce | 795 | { |
66c7d4c1 | 796 | getline(input, input_contents, '\0'); |
2203b032 | 797 | |
66c7d4c1 JS |
798 | input_pointer = input_contents.data(); |
799 | input_end = input_contents.data() + input_contents.size(); | |
800 | ||
801 | if (keywords.empty()) | |
802 | { | |
803 | keywords.insert("probe"); | |
804 | keywords.insert("global"); | |
805 | keywords.insert("function"); | |
806 | keywords.insert("if"); | |
807 | keywords.insert("else"); | |
808 | keywords.insert("for"); | |
809 | keywords.insert("foreach"); | |
810 | keywords.insert("in"); | |
811 | keywords.insert("limit"); | |
812 | keywords.insert("return"); | |
813 | keywords.insert("delete"); | |
814 | keywords.insert("while"); | |
815 | keywords.insert("break"); | |
816 | keywords.insert("continue"); | |
817 | keywords.insert("next"); | |
818 | keywords.insert("string"); | |
819 | keywords.insert("long"); | |
f4fe2e93 FCE |
820 | keywords.insert("try"); |
821 | keywords.insert("catch"); | |
66c7d4c1 | 822 | } |
eacb10ce | 823 | } |
2f1a1aea | 824 | |
66c7d4c1 JS |
825 | set<string> lexer::keywords; |
826 | ||
1b1b4ceb RA |
827 | void |
828 | lexer::set_current_file (stapfile* f) | |
829 | { | |
830 | current_file = f; | |
2203b032 JS |
831 | if (f) |
832 | { | |
833 | f->file_contents = input_contents; | |
834 | f->name = input_name; | |
835 | } | |
1b1b4ceb | 836 | } |
bb2e3076 FCE |
837 | |
838 | int | |
839 | lexer::input_peek (unsigned n) | |
840 | { | |
66c7d4c1 JS |
841 | if (input_pointer + n >= input_end) |
842 | return -1; // EOF | |
843 | return (unsigned char)*(input_pointer + n); | |
bb2e3076 FCE |
844 | } |
845 | ||
846 | ||
dff50e09 | 847 | int |
2f1a1aea FCE |
848 | lexer::input_get () |
849 | { | |
66c7d4c1 | 850 | int c = input_peek(); |
bb2e3076 FCE |
851 | if (c < 0) return c; // EOF |
852 | ||
66c7d4c1 JS |
853 | ++input_pointer; |
854 | ||
3f99432c | 855 | if (cursor_suspend_count) |
9300f661 JS |
856 | { |
857 | // Track effect of input_put: preserve previous cursor/line_column | |
858 | // until all of its characters are consumed. | |
859 | if (--cursor_suspend_count == 0) | |
860 | { | |
861 | cursor_line = cursor_suspend_line; | |
862 | cursor_column = cursor_suspend_column; | |
863 | } | |
864 | } | |
3f99432c | 865 | else |
2f1a1aea | 866 | { |
3f99432c FCE |
867 | // update source cursor |
868 | if (c == '\n') | |
869 | { | |
870 | cursor_line ++; | |
871 | cursor_column = 1; | |
872 | } | |
873 | else | |
874 | cursor_column ++; | |
2f1a1aea | 875 | } |
2f1a1aea | 876 | |
eacb10ce | 877 | // clog << "[" << (char)c << "]"; |
2f1a1aea FCE |
878 | return c; |
879 | } | |
880 | ||
881 | ||
3f99432c | 882 | void |
9300f661 | 883 | lexer::input_put (const string& chars, const token* t) |
3f99432c | 884 | { |
66c7d4c1 JS |
885 | size_t pos = input_pointer - input_contents.data(); |
886 | // clog << "[put:" << chars << " @" << pos << "]"; | |
887 | input_contents.insert (pos, chars); | |
eacb10ce | 888 | cursor_suspend_count += chars.size(); |
9300f661 JS |
889 | cursor_suspend_line = cursor_line; |
890 | cursor_suspend_column = cursor_column; | |
891 | cursor_line = t->location.line; | |
892 | cursor_column = t->location.column; | |
66c7d4c1 JS |
893 | input_pointer = input_contents.data() + pos; |
894 | input_end = input_contents.data() + input_contents.size(); | |
3f99432c FCE |
895 | } |
896 | ||
897 | ||
2f1a1aea | 898 | token* |
3f847830 | 899 | lexer::scan (bool wildcard) |
2f1a1aea | 900 | { |
c5be7511 | 901 | ate_comment = false; // reset for each new token |
2f1a1aea | 902 | token* n = new token; |
2203b032 | 903 | n->location.file = current_file; |
2f1a1aea | 904 | |
9300f661 JS |
905 | skip: |
906 | bool suspended = (cursor_suspend_count > 0); | |
2f1a1aea FCE |
907 | n->location.line = cursor_line; |
908 | n->location.column = cursor_column; | |
909 | ||
910 | int c = input_get(); | |
3f99432c | 911 | // clog << "{" << (char)c << (char)c2 << "}"; |
2f1a1aea FCE |
912 | if (c < 0) |
913 | { | |
914 | delete n; | |
915 | return 0; | |
916 | } | |
917 | ||
918 | if (isspace (c)) | |
919 | goto skip; | |
920 | ||
66c7d4c1 JS |
921 | int c2 = input_peek (); |
922 | ||
3f99432c FCE |
923 | // Paste command line arguments as character streams into |
924 | // the beginning of a token. $1..$999 go through as raw | |
925 | // characters; @1..@999 are quoted/escaped as strings. | |
926 | // $# and @# expand to the number of arguments, similarly | |
927 | // raw or quoted. | |
9300f661 | 928 | if ((c == '$' || c == '@') && (c2 == '#')) |
3f99432c | 929 | { |
9300f661 JS |
930 | n->content.push_back (c); |
931 | n->content.push_back (c2); | |
3f99432c | 932 | input_get(); // swallow '#' |
9300f661 | 933 | if (suspended) |
2677d2fb | 934 | throw parse_error (_("invalid nested substitution of command line arguments"), n); |
9300f661 JS |
935 | size_t num_args = session.args.size (); |
936 | input_put ((c == '$') ? lex_cast (num_args) : lex_cast_qstring (num_args), n); | |
937 | n->content.clear(); | |
938 | goto skip; | |
3f99432c | 939 | } |
9300f661 | 940 | else if ((c == '$' || c == '@') && (isdigit (c2))) |
3f99432c | 941 | { |
9300f661 | 942 | n->content.push_back (c); |
3f99432c FCE |
943 | unsigned idx = 0; |
944 | do | |
945 | { | |
946 | input_get (); | |
947 | idx = (idx * 10) + (c2 - '0'); | |
9300f661 | 948 | n->content.push_back (c2); |
3f99432c FCE |
949 | c2 = input_peek (); |
950 | } while (c2 > 0 && | |
dff50e09 | 951 | isdigit (c2) && |
3f99432c | 952 | idx <= session.args.size()); // prevent overflow |
9300f661 | 953 | if (suspended) |
2677d2fb | 954 | throw parse_error (_("invalid nested substitution of command line arguments"), n); |
3f99432c FCE |
955 | if (idx == 0 || |
956 | idx-1 >= session.args.size()) | |
fc3f0ba7 FCE |
957 | throw parse_error (_F("command line argument index %lu out of range [1-%lu]", |
958 | (unsigned long) idx, (unsigned long) session.args.size()), | |
984ad50c | 959 | n); |
9300f661 JS |
960 | const string& arg = session.args[idx-1]; |
961 | input_put ((c == '$') ? arg : lex_cast_qstring (arg), n); | |
962 | n->content.clear(); | |
963 | goto skip; | |
3f99432c FCE |
964 | } |
965 | ||
0c218afb MH |
966 | else if (isalpha (c) || c == '$' || c == '@' || c == '_' || |
967 | (wildcard && c == '*')) | |
2f1a1aea FCE |
968 | { |
969 | n->type = tok_identifier; | |
970 | n->content = (char) c; | |
0c218afb MH |
971 | while (isalnum (c2) || c2 == '_' || c2 == '$' || |
972 | (wildcard && c2 == '*')) | |
2f1a1aea | 973 | { |
3f99432c FCE |
974 | input_get (); |
975 | n->content.push_back (c2); | |
976 | c2 = input_peek (); | |
6e213f58 | 977 | } |
213bee8f | 978 | |
66c7d4c1 | 979 | if (keywords.count(n->content)) |
3f99432c | 980 | n->type = tok_keyword; |
dff50e09 | 981 | |
2f1a1aea FCE |
982 | return n; |
983 | } | |
984 | ||
3a20432b | 985 | else if (isdigit (c)) // positive literal |
2f1a1aea | 986 | { |
2f1a1aea | 987 | n->type = tok_number; |
9c0c0e46 FCE |
988 | n->content = (char) c; |
989 | ||
66c7d4c1 | 990 | while (isalnum (c2)) |
2f1a1aea | 991 | { |
9c0c0e46 FCE |
992 | // NB: isalnum is very permissive. We rely on strtol, called in |
993 | // parser::parse_literal below, to confirm that the number string | |
994 | // is correctly formatted and in range. | |
995 | ||
66c7d4c1 JS |
996 | input_get (); |
997 | n->content.push_back (c2); | |
998 | c2 = input_peek (); | |
2f1a1aea FCE |
999 | } |
1000 | return n; | |
1001 | } | |
1002 | ||
1003 | else if (c == '\"') | |
1004 | { | |
1005 | n->type = tok_string; | |
1006 | while (1) | |
1007 | { | |
1008 | c = input_get (); | |
1009 | ||
3f99432c | 1010 | if (c < 0 || c == '\n') |
2f1a1aea | 1011 | { |
2677d2fb | 1012 | throw parse_error(_("Could not find matching closing quote"), n); |
2f1a1aea FCE |
1013 | } |
1014 | if (c == '\"') // closing double-quotes | |
1015 | break; | |
3f99432c | 1016 | else if (c == '\\') // see also input_put |
dff50e09 | 1017 | { |
7d46afb8 GH |
1018 | c = input_get (); |
1019 | switch (c) | |
1020 | { | |
1021 | case 'a': | |
1022 | case 'b': | |
1023 | case 't': | |
1024 | case 'n': | |
1025 | case 'v': | |
1026 | case 'f': | |
1027 | case 'r': | |
f03954fd | 1028 | case '0' ... '7': // NB: need only match the first digit |
7d46afb8 | 1029 | case '\\': |
7d46afb8 | 1030 | // Pass these escapes through to the string value |
dff50e09 | 1031 | // being parsed; it will be emitted into a C literal. |
7d46afb8 GH |
1032 | |
1033 | n->content.push_back('\\'); | |
1034 | ||
3f99432c | 1035 | // fall through |
7d46afb8 | 1036 | default: |
7d46afb8 GH |
1037 | n->content.push_back(c); |
1038 | break; | |
1039 | } | |
2f1a1aea FCE |
1040 | } |
1041 | else | |
1042 | n->content.push_back(c); | |
1043 | } | |
1044 | return n; | |
1045 | } | |
1046 | ||
1047 | else if (ispunct (c)) | |
1048 | { | |
bb2e3076 | 1049 | int c3 = input_peek (1); |
2f1a1aea | 1050 | |
3a20432b FCE |
1051 | // NB: if we were to recognize negative numeric literals here, |
1052 | // we'd introduce another grammar ambiguity: | |
1053 | // 1-1 would be parsed as tok_number(1) and tok_number(-1) | |
1054 | // instead of tok_number(1) tok_operator('-') tok_number(1) | |
1055 | ||
66c7d4c1 | 1056 | if (c == '#') // shell comment |
2f1a1aea FCE |
1057 | { |
1058 | unsigned this_line = cursor_line; | |
bb2e3076 FCE |
1059 | do { c = input_get (); } |
1060 | while (c >= 0 && cursor_line == this_line); | |
c5be7511 | 1061 | ate_comment = true; |
2f1a1aea FCE |
1062 | goto skip; |
1063 | } | |
66c7d4c1 | 1064 | else if ((c == '/' && c2 == '/')) // C++ comment |
63a7c90e FCE |
1065 | { |
1066 | unsigned this_line = cursor_line; | |
bb2e3076 FCE |
1067 | do { c = input_get (); } |
1068 | while (c >= 0 && cursor_line == this_line); | |
c5be7511 | 1069 | ate_comment = true; |
63a7c90e FCE |
1070 | goto skip; |
1071 | } | |
1072 | else if (c == '/' && c2 == '*') // C comment | |
1073 | { | |
66c7d4c1 JS |
1074 | (void) input_get (); // swallow '*' already in c2 |
1075 | c = input_get (); | |
63a7c90e | 1076 | c2 = input_get (); |
bb2e3076 | 1077 | while (c2 >= 0) |
63a7c90e | 1078 | { |
66c7d4c1 JS |
1079 | if (c == '*' && c2 == '/') |
1080 | break; | |
63a7c90e FCE |
1081 | c = c2; |
1082 | c2 = input_get (); | |
63a7c90e | 1083 | } |
c5be7511 | 1084 | ate_comment = true; |
bb2e3076 | 1085 | goto skip; |
63a7c90e | 1086 | } |
54dfabe9 FCE |
1087 | else if (c == '%' && c2 == '{') // embedded code |
1088 | { | |
1089 | n->type = tok_embedded; | |
1090 | (void) input_get (); // swallow '{' already in c2 | |
66c7d4c1 JS |
1091 | c = input_get (); |
1092 | c2 = input_get (); | |
1093 | while (c2 >= 0) | |
54dfabe9 | 1094 | { |
66c7d4c1 JS |
1095 | if (c == '%' && c2 == '}') |
1096 | return n; | |
54dfabe9 | 1097 | n->content += c; |
66c7d4c1 JS |
1098 | c = c2; |
1099 | c2 = input_get (); | |
54dfabe9 | 1100 | } |
72cdb9cd | 1101 | |
2677d2fb | 1102 | throw parse_error (_("Could not find matching '%}' to close embedded function block"), n); |
54dfabe9 | 1103 | } |
2f1a1aea | 1104 | |
bb2e3076 FCE |
1105 | // We're committed to recognizing at least the first character |
1106 | // as an operator. | |
2f1a1aea | 1107 | n->type = tok_operator; |
66c7d4c1 | 1108 | n->content = c; |
2f1a1aea | 1109 | |
bb2e3076 | 1110 | // match all valid operators, in decreasing size order |
66c7d4c1 JS |
1111 | if ((c == '<' && c2 == '<' && c3 == '<') || |
1112 | (c == '<' && c2 == '<' && c3 == '=') || | |
1113 | (c == '>' && c2 == '>' && c3 == '=')) | |
82919855 | 1114 | { |
66c7d4c1 JS |
1115 | n->content += c2; |
1116 | n->content += c3; | |
bb2e3076 FCE |
1117 | input_get (); input_get (); // swallow other two characters |
1118 | } | |
66c7d4c1 JS |
1119 | else if ((c == '=' && c2 == '=') || |
1120 | (c == '!' && c2 == '=') || | |
1121 | (c == '<' && c2 == '=') || | |
1122 | (c == '>' && c2 == '=') || | |
1123 | (c == '+' && c2 == '=') || | |
1124 | (c == '-' && c2 == '=') || | |
1125 | (c == '*' && c2 == '=') || | |
1126 | (c == '/' && c2 == '=') || | |
1127 | (c == '%' && c2 == '=') || | |
1128 | (c == '&' && c2 == '=') || | |
1129 | (c == '^' && c2 == '=') || | |
1130 | (c == '|' && c2 == '=') || | |
1131 | (c == '.' && c2 == '=') || | |
1132 | (c == '&' && c2 == '&') || | |
1133 | (c == '|' && c2 == '|') || | |
1134 | (c == '+' && c2 == '+') || | |
1135 | (c == '-' && c2 == '-') || | |
1136 | (c == '-' && c2 == '>') || | |
1137 | (c == '<' && c2 == '<') || | |
1138 | (c == '>' && c2 == '>') || | |
177a8ead | 1139 | // preprocessor tokens |
66c7d4c1 JS |
1140 | (c == '%' && c2 == '(') || |
1141 | (c == '%' && c2 == '?') || | |
1142 | (c == '%' && c2 == ':') || | |
1143 | (c == '%' && c2 == ')')) | |
bb2e3076 | 1144 | { |
66c7d4c1 | 1145 | n->content += c2; |
bb2e3076 | 1146 | input_get (); // swallow other character |
dff50e09 | 1147 | } |
2f1a1aea FCE |
1148 | |
1149 | return n; | |
1150 | } | |
1151 | ||
1152 | else | |
1153 | { | |
1154 | n->type = tok_junk; | |
1155 | n->content = (char) c; | |
1156 | return n; | |
1157 | } | |
1158 | } | |
1159 | ||
1160 | ||
1161 | // ------------------------------------------------------------------------ | |
1162 | ||
1163 | stapfile* | |
1164 | parser::parse () | |
1165 | { | |
1166 | stapfile* f = new stapfile; | |
1b1b4ceb | 1167 | input.set_current_file (f); |
56099f08 FCE |
1168 | |
1169 | bool empty = true; | |
1170 | ||
2f1a1aea FCE |
1171 | while (1) |
1172 | { | |
1173 | try | |
1174 | { | |
a07a2c28 | 1175 | systemtap_v_seen = 0; |
2f1a1aea | 1176 | const token* t = peek (); |
56099f08 | 1177 | if (! t) // nice clean EOF |
2f1a1aea FCE |
1178 | break; |
1179 | ||
56099f08 | 1180 | empty = false; |
6e213f58 DS |
1181 | if (t->type == tok_keyword && t->content == "probe") |
1182 | { | |
1183 | context = con_probe; | |
1184 | parse_probe (f->probes, f->aliases); | |
1185 | } | |
1186 | else if (t->type == tok_keyword && t->content == "global") | |
1187 | { | |
1188 | context = con_global; | |
4b5f3e45 | 1189 | parse_global (f->globals, f->probes); |
6e213f58 DS |
1190 | } |
1191 | else if (t->type == tok_keyword && t->content == "function") | |
1192 | { | |
1193 | context = con_function; | |
1194 | parse_functiondecl (f->functions); | |
1195 | } | |
54dfabe9 | 1196 | else if (t->type == tok_embedded) |
6e213f58 DS |
1197 | { |
1198 | context = con_embedded; | |
1199 | f->embeds.push_back (parse_embeddedcode ()); | |
1200 | } | |
2f1a1aea | 1201 | else |
6e213f58 DS |
1202 | { |
1203 | context = con_unknown; | |
2677d2fb | 1204 | throw parse_error (_("expected 'probe', 'global', 'function', or '%{'")); |
6e213f58 | 1205 | } |
2f1a1aea FCE |
1206 | } |
1207 | catch (parse_error& pe) | |
1208 | { | |
1209 | print_error (pe); | |
cd7116b8 | 1210 | if (pe.skip_some) // for recovery |
46954f1d FCE |
1211 | // Quietly swallow all tokens until the next keyword we can start parsing from. |
1212 | while (1) | |
1213 | try | |
1214 | { | |
cd7116b8 FCE |
1215 | { |
1216 | const token* t = peek (); | |
1217 | if (! t) | |
1218 | break; | |
46954f1d FCE |
1219 | if (t->type == tok_keyword && t->content == "probe") break; |
1220 | else if (t->type == tok_keyword && t->content == "global") break; | |
1221 | else if (t->type == tok_keyword && t->content == "function") break; | |
1222 | else if (t->type == tok_embedded) break; | |
1223 | next(); // swallow it | |
cd7116b8 | 1224 | } |
46954f1d FCE |
1225 | } |
1226 | catch (parse_error& pe2) | |
1227 | { | |
1228 | // parse error during recovery ... ugh | |
1229 | print_error (pe2); | |
1230 | } | |
177a8ead | 1231 | } |
2f1a1aea FCE |
1232 | } |
1233 | ||
56099f08 FCE |
1234 | if (empty) |
1235 | { | |
46a1a151 | 1236 | cerr << _F("Input file '%s' is empty or missing.", input_name.c_str()) << endl; |
56099f08 | 1237 | delete f; |
2203b032 | 1238 | f = 0; |
56099f08 FCE |
1239 | } |
1240 | else if (num_errors > 0) | |
2f1a1aea | 1241 | { |
cb9ae1fe | 1242 | cerr << _F(ngettext("%d parse error.", "%d parse errors.", num_errors), num_errors) << endl; |
2f1a1aea | 1243 | delete f; |
2203b032 | 1244 | f = 0; |
2f1a1aea | 1245 | } |
dff50e09 | 1246 | |
2203b032 | 1247 | input.set_current_file(0); |
2f1a1aea FCE |
1248 | return f; |
1249 | } | |
1250 | ||
1251 | ||
20c6c071 | 1252 | void |
54dfabe9 FCE |
1253 | parser::parse_probe (std::vector<probe *> & probe_ret, |
1254 | std::vector<probe_alias *> & alias_ret) | |
2f1a1aea | 1255 | { |
82919855 | 1256 | const token* t0 = next (); |
6e213f58 | 1257 | if (! (t0->type == tok_keyword && t0->content == "probe")) |
2677d2fb | 1258 | throw parse_error (_("expected 'probe'")); |
82919855 | 1259 | |
20c6c071 GH |
1260 | vector<probe_point *> aliases; |
1261 | vector<probe_point *> locations; | |
1262 | ||
1263 | bool equals_ok = true; | |
82919855 | 1264 | |
97266278 LG |
1265 | int epilogue_alias = 0; |
1266 | ||
2f1a1aea FCE |
1267 | while (1) |
1268 | { | |
b4ceace2 | 1269 | probe_point * pp = parse_probe_point (); |
dff50e09 | 1270 | |
b4ceace2 | 1271 | const token* t = peek (); |
dff50e09 | 1272 | if (equals_ok && t |
b4ceace2 FCE |
1273 | && t->type == tok_operator && t->content == "=") |
1274 | { | |
1ad820e3 | 1275 | if (pp->optional || pp->sufficient) |
2677d2fb | 1276 | throw parse_error (_("probe point alias name cannot be optional nor sufficient"), pp->components.front()->tok); |
b4ceace2 FCE |
1277 | aliases.push_back(pp); |
1278 | next (); | |
1279 | continue; | |
1280 | } | |
dff50e09 | 1281 | else if (equals_ok && t |
97266278 LG |
1282 | && t->type == tok_operator && t->content == "+=") |
1283 | { | |
1ad820e3 | 1284 | if (pp->optional || pp->sufficient) |
2677d2fb | 1285 | throw parse_error (_("probe point alias name cannot be optional nor sufficient"), pp->components.front()->tok); |
97266278 LG |
1286 | aliases.push_back(pp); |
1287 | epilogue_alias = 1; | |
1288 | next (); | |
1289 | continue; | |
1290 | } | |
b4ceace2 FCE |
1291 | else if (t && t->type == tok_operator && t->content == ",") |
1292 | { | |
1293 | locations.push_back(pp); | |
1294 | equals_ok = false; | |
1295 | next (); | |
1296 | continue; | |
1297 | } | |
1298 | else if (t && t->type == tok_operator && t->content == "{") | |
1299 | { | |
1300 | locations.push_back(pp); | |
1301 | break; | |
1302 | } | |
2f1a1aea | 1303 | else |
2677d2fb | 1304 | throw parse_error (_("expected probe point specifier")); |
2f1a1aea | 1305 | } |
20c6c071 | 1306 | |
20c6c071 GH |
1307 | if (aliases.empty()) |
1308 | { | |
54dfabe9 FCE |
1309 | probe* p = new probe; |
1310 | p->tok = t0; | |
1311 | p->locations = locations; | |
1312 | p->body = parse_stmt_block (); | |
37ebca01 | 1313 | p->privileged = privileged; |
a07a2c28 | 1314 | p->systemtap_v_conditional = systemtap_v_seen; |
54dfabe9 | 1315 | probe_ret.push_back (p); |
20c6c071 GH |
1316 | } |
1317 | else | |
1318 | { | |
54dfabe9 | 1319 | probe_alias* p = new probe_alias (aliases); |
97266278 LG |
1320 | if(epilogue_alias) |
1321 | p->epilogue_style = true; | |
1322 | else | |
1323 | p->epilogue_style = false; | |
54dfabe9 FCE |
1324 | p->tok = t0; |
1325 | p->locations = locations; | |
1326 | p->body = parse_stmt_block (); | |
37ebca01 | 1327 | p->privileged = privileged; |
a07a2c28 | 1328 | p->systemtap_v_conditional = systemtap_v_seen; |
54dfabe9 | 1329 | alias_ret.push_back (p); |
20c6c071 | 1330 | } |
54dfabe9 | 1331 | } |
20c6c071 | 1332 | |
54dfabe9 FCE |
1333 | |
1334 | embeddedcode* | |
1335 | parser::parse_embeddedcode () | |
1336 | { | |
1337 | embeddedcode* e = new embeddedcode; | |
1338 | const token* t = next (); | |
1339 | if (t->type != tok_embedded) | |
2677d2fb | 1340 | throw parse_error (_("expected '%{'")); |
24cb178f FCE |
1341 | |
1342 | if (! privileged) | |
2677d2fb | 1343 | throw parse_error (_("embedded code in unprivileged script"), |
cd7116b8 | 1344 | false /* don't skip tokens for parse resumption */); |
54dfabe9 FCE |
1345 | |
1346 | e->tok = t; | |
1347 | e->code = t->content; | |
1348 | return e; | |
2f1a1aea FCE |
1349 | } |
1350 | ||
1351 | ||
1352 | block* | |
56099f08 | 1353 | parser::parse_stmt_block () |
2f1a1aea FCE |
1354 | { |
1355 | block* pb = new block; | |
1356 | ||
56099f08 FCE |
1357 | const token* t = next (); |
1358 | if (! (t->type == tok_operator && t->content == "{")) | |
2677d2fb | 1359 | throw parse_error (_("expected '{'")); |
56099f08 FCE |
1360 | |
1361 | pb->tok = t; | |
2b066ec1 | 1362 | |
2f1a1aea FCE |
1363 | while (1) |
1364 | { | |
46954f1d FCE |
1365 | t = peek (); |
1366 | if (t && t->type == tok_operator && t->content == "}") | |
1367 | { | |
1368 | next (); | |
1369 | break; | |
1370 | } | |
1371 | pb->statements.push_back (parse_statement ()); | |
2f1a1aea FCE |
1372 | } |
1373 | ||
1374 | return pb; | |
1375 | } | |
1376 | ||
1377 | ||
f4fe2e93 FCE |
1378 | try_block* |
1379 | parser::parse_try_block () | |
1380 | { | |
1381 | try_block* pb = new try_block; | |
1382 | ||
1383 | pb->tok = expect_kw ("try"); | |
1384 | pb->try_block = parse_stmt_block(); | |
1385 | expect_kw ("catch"); | |
1386 | ||
1387 | const token* t = peek (); | |
1388 | if (t->type == tok_operator && t->content == "(") | |
1389 | { | |
1390 | next (); // swallow the '(' | |
1391 | ||
1392 | t = next(); | |
1393 | if (! (t->type == tok_identifier)) | |
2677d2fb | 1394 | throw parse_error (_("expected identifier")); |
f4fe2e93 FCE |
1395 | symbol* sym = new symbol; |
1396 | sym->tok = t; | |
1397 | sym->name = t->content; | |
1398 | pb->catch_error_var = sym; | |
1399 | ||
1400 | expect_op (")"); | |
1401 | } | |
1402 | else | |
1403 | pb->catch_error_var = 0; | |
1404 | ||
1405 | pb->catch_block = parse_stmt_block(); | |
1406 | ||
1407 | return pb; | |
1408 | } | |
1409 | ||
1410 | ||
1411 | ||
2f1a1aea FCE |
1412 | statement* |
1413 | parser::parse_statement () | |
1414 | { | |
40b71c47 | 1415 | statement *ret; |
2f1a1aea FCE |
1416 | const token* t = peek (); |
1417 | if (t && t->type == tok_operator && t->content == ";") | |
f946b10f | 1418 | return new null_statement (next ()); |
dff50e09 | 1419 | else if (t && t->type == tok_operator && t->content == "{") |
40b71c47 | 1420 | return parse_stmt_block (); // Don't squash semicolons. |
f4fe2e93 FCE |
1421 | else if (t && t->type == tok_keyword && t->content == "try") |
1422 | return parse_try_block (); // Don't squash semicolons. | |
6e213f58 | 1423 | else if (t && t->type == tok_keyword && t->content == "if") |
40b71c47 | 1424 | return parse_if_statement (); // Don't squash semicolons. |
6e213f58 | 1425 | else if (t && t->type == tok_keyword && t->content == "for") |
40b71c47 | 1426 | return parse_for_loop (); // Don't squash semicolons. |
6e213f58 | 1427 | else if (t && t->type == tok_keyword && t->content == "foreach") |
40b71c47 MW |
1428 | return parse_foreach_loop (); // Don't squash semicolons. |
1429 | else if (t && t->type == tok_keyword && t->content == "while") | |
1430 | return parse_while_loop (); // Don't squash semicolons. | |
6e213f58 | 1431 | else if (t && t->type == tok_keyword && t->content == "return") |
40b71c47 | 1432 | ret = parse_return_statement (); |
6e213f58 | 1433 | else if (t && t->type == tok_keyword && t->content == "delete") |
40b71c47 | 1434 | ret = parse_delete_statement (); |
6e213f58 | 1435 | else if (t && t->type == tok_keyword && t->content == "break") |
40b71c47 | 1436 | ret = parse_break_statement (); |
6e213f58 | 1437 | else if (t && t->type == tok_keyword && t->content == "continue") |
40b71c47 | 1438 | ret = parse_continue_statement (); |
6e213f58 | 1439 | else if (t && t->type == tok_keyword && t->content == "next") |
40b71c47 | 1440 | ret = parse_next_statement (); |
2f1a1aea FCE |
1441 | else if (t && (t->type == tok_operator || // expressions are flexible |
1442 | t->type == tok_identifier || | |
1443 | t->type == tok_number || | |
7d902887 FCE |
1444 | t->type == tok_string || |
1445 | t->type == tok_embedded )) | |
40b71c47 | 1446 | ret = parse_expr_statement (); |
54dfabe9 | 1447 | // XXX: consider generally accepting tok_embedded here too |
2f1a1aea | 1448 | else |
2677d2fb | 1449 | throw parse_error (_("expected statement")); |
40b71c47 MW |
1450 | |
1451 | // Squash "empty" trailing colons after any "non-block-like" statement. | |
1452 | t = peek (); | |
1453 | if (t && t->type == tok_operator && t->content == ";") | |
1454 | { | |
1455 | next (); // Silently eat trailing ; after statement | |
1456 | } | |
1457 | ||
1458 | return ret; | |
2f1a1aea FCE |
1459 | } |
1460 | ||
1461 | ||
56099f08 | 1462 | void |
78f6bba6 | 1463 | parser::parse_global (vector <vardecl*>& globals, vector<probe*>&) |
2f1a1aea | 1464 | { |
82919855 | 1465 | const token* t0 = next (); |
6e213f58 | 1466 | if (! (t0->type == tok_keyword && t0->content == "global")) |
2677d2fb | 1467 | throw parse_error (_("expected 'global'")); |
82919855 | 1468 | |
56099f08 FCE |
1469 | while (1) |
1470 | { | |
1471 | const token* t = next (); | |
1472 | if (! (t->type == tok_identifier)) | |
2677d2fb | 1473 | throw parse_error (_("expected identifier")); |
56099f08 | 1474 | |
2b066ec1 FCE |
1475 | for (unsigned i=0; i<globals.size(); i++) |
1476 | if (globals[i]->name == t->content) | |
2677d2fb | 1477 | throw parse_error (_("duplicate global name")); |
dff50e09 | 1478 | |
24cb178f FCE |
1479 | vardecl* d = new vardecl; |
1480 | d->name = t->content; | |
1481 | d->tok = t; | |
a07a2c28 | 1482 | d->systemtap_v_conditional = systemtap_v_seen; |
24cb178f | 1483 | globals.push_back (d); |
56099f08 | 1484 | |
82919855 | 1485 | t = peek (); |
ef474d24 | 1486 | |
74e6cc92 CM |
1487 | if(t && t->type == tok_operator && t->content == "%") //wrapping |
1488 | { | |
1489 | d->wrap = true; | |
1490 | next(); | |
1491 | t = peek(); | |
1492 | } | |
1493 | ||
ef474d24 JS |
1494 | if (t && t->type == tok_operator && t->content == "[") // array size |
1495 | { | |
1496 | int64_t size; | |
1497 | next (); | |
1498 | expect_number(size); | |
1499 | if (size <= 0 || size > 1000000) // arbitrary max | |
2677d2fb | 1500 | throw parse_error(_("array size out of range")); |
ef474d24 JS |
1501 | d->maxsize = (int)size; |
1502 | expect_known(tok_operator, "]"); | |
1503 | t = peek (); | |
1504 | } | |
1505 | ||
4b5f3e45 | 1506 | if (t && t->type == tok_operator && t->content == "=") // initialization |
ef474d24 JS |
1507 | { |
1508 | if (!d->compatible_arity(0)) | |
2677d2fb | 1509 | throw parse_error(_("only scalar globals can be initialized")); |
58701b78 | 1510 | d->set_arity(0, t); |
ef474d24 JS |
1511 | next (); |
1512 | d->init = parse_literal (); | |
1513 | d->type = d->init->type; | |
1514 | t = peek (); | |
1515 | } | |
4b5f3e45 | 1516 | |
c3799d72 AM |
1517 | if (t && t->type == tok_operator && t->content == ";") // termination |
1518 | next(); | |
1519 | ||
4b5f3e45 | 1520 | if (t && t->type == tok_operator && t->content == ",") // next global |
82919855 FCE |
1521 | { |
1522 | next (); | |
1523 | continue; | |
1524 | } | |
56099f08 | 1525 | else |
82919855 | 1526 | break; |
56099f08 FCE |
1527 | } |
1528 | } | |
1529 | ||
1530 | ||
24cb178f FCE |
1531 | void |
1532 | parser::parse_functiondecl (std::vector<functiondecl*>& functions) | |
56099f08 | 1533 | { |
82919855 | 1534 | const token* t = next (); |
6e213f58 | 1535 | if (! (t->type == tok_keyword && t->content == "function")) |
2677d2fb | 1536 | throw parse_error (_("expected 'function'")); |
82919855 | 1537 | |
56099f08 | 1538 | |
82919855 | 1539 | t = next (); |
6e213f58 DS |
1540 | if (! (t->type == tok_identifier) |
1541 | && ! (t->type == tok_keyword | |
1542 | && (t->content == "string" || t->content == "long"))) | |
2677d2fb | 1543 | throw parse_error (_("expected identifier")); |
24cb178f FCE |
1544 | |
1545 | for (unsigned i=0; i<functions.size(); i++) | |
1546 | if (functions[i]->name == t->content) | |
2677d2fb | 1547 | throw parse_error (_("duplicate function name")); |
24cb178f FCE |
1548 | |
1549 | functiondecl *fd = new functiondecl (); | |
56099f08 FCE |
1550 | fd->name = t->content; |
1551 | fd->tok = t; | |
1552 | ||
1553 | t = next (); | |
6a505121 FCE |
1554 | if (t->type == tok_operator && t->content == ":") |
1555 | { | |
1556 | t = next (); | |
6e213f58 | 1557 | if (t->type == tok_keyword && t->content == "string") |
6a505121 | 1558 | fd->type = pe_string; |
6e213f58 | 1559 | else if (t->type == tok_keyword && t->content == "long") |
6a505121 | 1560 | fd->type = pe_long; |
2677d2fb | 1561 | else throw parse_error (_("expected 'string' or 'long'")); |
6a505121 FCE |
1562 | |
1563 | t = next (); | |
1564 | } | |
1565 | ||
56099f08 | 1566 | if (! (t->type == tok_operator && t->content == "(")) |
2677d2fb | 1567 | throw parse_error (_("expected '('")); |
56099f08 FCE |
1568 | |
1569 | while (1) | |
1570 | { | |
1571 | t = next (); | |
1572 | ||
1573 | // permit zero-argument fuctions | |
1574 | if (t->type == tok_operator && t->content == ")") | |
1575 | break; | |
1576 | else if (! (t->type == tok_identifier)) | |
2677d2fb | 1577 | throw parse_error (_("expected identifier")); |
56099f08 FCE |
1578 | vardecl* vd = new vardecl; |
1579 | vd->name = t->content; | |
1580 | vd->tok = t; | |
1581 | fd->formal_args.push_back (vd); | |
a07a2c28 | 1582 | fd->systemtap_v_conditional = systemtap_v_seen; |
56099f08 FCE |
1583 | |
1584 | t = next (); | |
6a505121 FCE |
1585 | if (t->type == tok_operator && t->content == ":") |
1586 | { | |
1587 | t = next (); | |
6e213f58 | 1588 | if (t->type == tok_keyword && t->content == "string") |
6a505121 | 1589 | vd->type = pe_string; |
6e213f58 | 1590 | else if (t->type == tok_keyword && t->content == "long") |
6a505121 | 1591 | vd->type = pe_long; |
2677d2fb | 1592 | else throw parse_error (_("expected 'string' or 'long'")); |
dff50e09 | 1593 | |
6a505121 FCE |
1594 | t = next (); |
1595 | } | |
56099f08 FCE |
1596 | if (t->type == tok_operator && t->content == ")") |
1597 | break; | |
1598 | if (t->type == tok_operator && t->content == ",") | |
1599 | continue; | |
1600 | else | |
2677d2fb | 1601 | throw parse_error (_("expected ',' or ')'")); |
56099f08 FCE |
1602 | } |
1603 | ||
54dfabe9 FCE |
1604 | t = peek (); |
1605 | if (t && t->type == tok_embedded) | |
1606 | fd->body = parse_embeddedcode (); | |
1607 | else | |
1608 | fd->body = parse_stmt_block (); | |
24cb178f FCE |
1609 | |
1610 | functions.push_back (fd); | |
2f1a1aea FCE |
1611 | } |
1612 | ||
1613 | ||
9c0c0e46 FCE |
1614 | probe_point* |
1615 | parser::parse_probe_point () | |
2f1a1aea | 1616 | { |
9c0c0e46 | 1617 | probe_point* pl = new probe_point; |
2f1a1aea | 1618 | |
9c0c0e46 | 1619 | while (1) |
2f1a1aea | 1620 | { |
0c218afb | 1621 | const token* t = next (true); // wildcard scanning here |
6e213f58 DS |
1622 | if (! (t->type == tok_identifier |
1623 | // we must allow ".return" and ".function", which are keywords | |
0c218afb | 1624 | || t->type == tok_keyword)) |
2677d2fb | 1625 | throw parse_error (_("expected identifier or '*'")); |
9c0c0e46 | 1626 | |
9c0c0e46 FCE |
1627 | |
1628 | probe_point::component* c = new probe_point::component; | |
1629 | c->functor = t->content; | |
f1a0157a | 1630 | c->tok = t; |
9c0c0e46 | 1631 | pl->components.push_back (c); |
6e3347a9 | 1632 | // NB we may add c->arg soon |
9c0c0e46 FCE |
1633 | |
1634 | t = peek (); | |
a477f3f1 | 1635 | |
6e3347a9 | 1636 | // consume optional parameter |
9c0c0e46 FCE |
1637 | if (t && t->type == tok_operator && t->content == "(") |
1638 | { | |
1639 | next (); // consume "(" | |
1640 | c->arg = parse_literal (); | |
1641 | ||
1642 | t = next (); | |
1643 | if (! (t->type == tok_operator && t->content == ")")) | |
2677d2fb | 1644 | throw parse_error (_("expected ')'")); |
9c0c0e46 FCE |
1645 | |
1646 | t = peek (); | |
9c0c0e46 | 1647 | } |
9c0c0e46 FCE |
1648 | |
1649 | if (t && t->type == tok_operator && t->content == ".") | |
6e3347a9 FCE |
1650 | { |
1651 | next (); | |
1652 | continue; | |
1653 | } | |
1654 | ||
f1a0157a | 1655 | // We only fall through here at the end of a probe point (past |
6e3347a9 FCE |
1656 | // all the dotted/parametrized components). |
1657 | ||
d898100a FCE |
1658 | if (t && t->type == tok_operator && |
1659 | (t->content == "?" || t->content == "!")) | |
6e3347a9 FCE |
1660 | { |
1661 | pl->optional = true; | |
d898100a FCE |
1662 | if (t->content == "!") pl->sufficient = true; |
1663 | // NB: sufficient implies optional | |
6e3347a9 FCE |
1664 | next (); |
1665 | t = peek (); | |
1666 | // fall through | |
cbbe8080 MH |
1667 | } |
1668 | ||
1669 | if (t && t->type == tok_keyword && t->content == "if") | |
1670 | { | |
1671 | next (); | |
1672 | t = peek (); | |
75686668 | 1673 | if (t && ! (t->type == tok_operator && t->content == "(")) |
2677d2fb | 1674 | throw parse_error (_("expected '('")); |
cbbe8080 MH |
1675 | next (); |
1676 | ||
1677 | pl->condition = parse_expression (); | |
1678 | ||
1679 | t = peek (); | |
75686668 | 1680 | if (t && ! (t->type == tok_operator && t->content == ")")) |
2677d2fb | 1681 | throw parse_error (_("expected ')'")); |
cbbe8080 MH |
1682 | next (); |
1683 | t = peek (); | |
1684 | // fall through | |
6e3347a9 FCE |
1685 | } |
1686 | ||
dff50e09 | 1687 | if (t && t->type == tok_operator |
6e3347a9 FCE |
1688 | && (t->content == "{" || t->content == "," || |
1689 | t->content == "=" || t->content == "+=" )) | |
1690 | break; | |
dff50e09 | 1691 | |
2677d2fb | 1692 | throw parse_error (_("expected one of '. , ( ? ! { = +='")); |
2f1a1aea FCE |
1693 | } |
1694 | ||
1695 | return pl; | |
1696 | } | |
1697 | ||
1698 | ||
1699 | literal* | |
1700 | parser::parse_literal () | |
1701 | { | |
1702 | const token* t = next (); | |
56099f08 | 1703 | literal* l; |
2f1a1aea | 1704 | if (t->type == tok_string) |
c5be7511 JS |
1705 | { |
1706 | literal_string *ls = new literal_string (t->content); | |
1707 | ||
1708 | // PR11208: check if the next token is also a string literal; auto-concatenate it | |
1709 | // This is complicated to the extent that we need to skip intermediate whitespace. | |
1710 | // XXX: but not comments | |
83d4d1a3 MW |
1711 | const token *n = peek(); |
1712 | while (n != NULL && n->type == tok_string && !input.ate_comment) | |
1713 | { | |
1714 | ls->value.append(next()->content); // consume and append the token | |
1715 | n = peek(); | |
1716 | } | |
c5be7511 JS |
1717 | l = ls; |
1718 | } | |
16e8f21f | 1719 | else |
9c0c0e46 | 1720 | { |
16e8f21f JS |
1721 | bool neg = false; |
1722 | if (t->type == tok_operator && t->content == "-") | |
1723 | { | |
1724 | neg = true; | |
1725 | t = next (); | |
1726 | } | |
1727 | ||
1728 | if (t->type == tok_number) | |
1729 | { | |
1730 | const char* startp = t->content.c_str (); | |
1731 | char* endp = (char*) startp; | |
1732 | ||
1733 | // NB: we allow controlled overflow from LLONG_MIN .. ULLONG_MAX | |
1734 | // Actually, this allows all the way from -ULLONG_MAX to ULLONG_MAX, | |
79e6d33f JS |
1735 | // since the lexer only gives us positive digit strings, but we'll |
1736 | // limit it to LLONG_MIN when a '-' operator is fed into the literal. | |
16e8f21f JS |
1737 | errno = 0; |
1738 | long long value = (long long) strtoull (startp, & endp, 0); | |
16e8f21f | 1739 | if (errno == ERANGE || errno == EINVAL || *endp != '\0' |
79e6d33f | 1740 | || (neg && (unsigned long long) value > 9223372036854775808ULL) |
16e8f21f JS |
1741 | || (unsigned long long) value > 18446744073709551615ULL |
1742 | || value < -9223372036854775807LL-1) | |
2677d2fb | 1743 | throw parse_error (_("number invalid or out of range")); |
16e8f21f | 1744 | |
79e6d33f JS |
1745 | if (neg) |
1746 | value = -value; | |
1747 | ||
16e8f21f JS |
1748 | l = new literal_number (value); |
1749 | } | |
1750 | else | |
2677d2fb | 1751 | throw parse_error (_("expected literal string or number")); |
9c0c0e46 | 1752 | } |
56099f08 FCE |
1753 | |
1754 | l->tok = t; | |
1755 | return l; | |
2f1a1aea FCE |
1756 | } |
1757 | ||
1758 | ||
1759 | if_statement* | |
1760 | parser::parse_if_statement () | |
1761 | { | |
1762 | const token* t = next (); | |
6e213f58 | 1763 | if (! (t->type == tok_keyword && t->content == "if")) |
2677d2fb | 1764 | throw parse_error (_("expected 'if'")); |
56099f08 FCE |
1765 | if_statement* s = new if_statement; |
1766 | s->tok = t; | |
1767 | ||
1768 | t = next (); | |
2f1a1aea | 1769 | if (! (t->type == tok_operator && t->content == "(")) |
2677d2fb | 1770 | throw parse_error (_("expected '('")); |
2f1a1aea | 1771 | |
2f1a1aea FCE |
1772 | s->condition = parse_expression (); |
1773 | ||
1774 | t = next (); | |
1775 | if (! (t->type == tok_operator && t->content == ")")) | |
2677d2fb | 1776 | throw parse_error (_("expected ')'")); |
2f1a1aea FCE |
1777 | |
1778 | s->thenblock = parse_statement (); | |
1779 | ||
1780 | t = peek (); | |
6e213f58 | 1781 | if (t && t->type == tok_keyword && t->content == "else") |
2f1a1aea FCE |
1782 | { |
1783 | next (); | |
1784 | s->elseblock = parse_statement (); | |
1785 | } | |
ed10c639 FCE |
1786 | else |
1787 | s->elseblock = 0; // in case not otherwise initialized | |
2f1a1aea FCE |
1788 | |
1789 | return s; | |
1790 | } | |
1791 | ||
1792 | ||
69c68955 FCE |
1793 | expr_statement* |
1794 | parser::parse_expr_statement () | |
1795 | { | |
1796 | expr_statement *es = new expr_statement; | |
1797 | const token* t = peek (); | |
1798 | es->tok = t; | |
1799 | es->value = parse_expression (); | |
1800 | return es; | |
1801 | } | |
1802 | ||
1803 | ||
56099f08 FCE |
1804 | return_statement* |
1805 | parser::parse_return_statement () | |
1806 | { | |
1807 | const token* t = next (); | |
6e213f58 | 1808 | if (! (t->type == tok_keyword && t->content == "return")) |
2677d2fb | 1809 | throw parse_error (_("expected 'return'")); |
6e213f58 | 1810 | if (context != con_function) |
2677d2fb | 1811 | throw parse_error (_("found 'return' not in function context")); |
56099f08 FCE |
1812 | return_statement* s = new return_statement; |
1813 | s->tok = t; | |
1814 | s->value = parse_expression (); | |
1815 | return s; | |
1816 | } | |
1817 | ||
1818 | ||
1819 | delete_statement* | |
1820 | parser::parse_delete_statement () | |
1821 | { | |
1822 | const token* t = next (); | |
6e213f58 | 1823 | if (! (t->type == tok_keyword && t->content == "delete")) |
2677d2fb | 1824 | throw parse_error (_("expected 'delete'")); |
56099f08 FCE |
1825 | delete_statement* s = new delete_statement; |
1826 | s->tok = t; | |
1827 | s->value = parse_expression (); | |
1828 | return s; | |
1829 | } | |
1830 | ||
1831 | ||
f3c26ea5 FCE |
1832 | next_statement* |
1833 | parser::parse_next_statement () | |
1834 | { | |
1835 | const token* t = next (); | |
6e213f58 | 1836 | if (! (t->type == tok_keyword && t->content == "next")) |
2677d2fb | 1837 | throw parse_error (_("expected 'next'")); |
6e213f58 | 1838 | if (context != con_probe) |
2677d2fb | 1839 | throw parse_error (_("found 'next' not in probe context")); |
f3c26ea5 FCE |
1840 | next_statement* s = new next_statement; |
1841 | s->tok = t; | |
1842 | return s; | |
1843 | } | |
1844 | ||
1845 | ||
1846 | break_statement* | |
1847 | parser::parse_break_statement () | |
1848 | { | |
1849 | const token* t = next (); | |
6e213f58 | 1850 | if (! (t->type == tok_keyword && t->content == "break")) |
2677d2fb | 1851 | throw parse_error (_("expected 'break'")); |
f3c26ea5 FCE |
1852 | break_statement* s = new break_statement; |
1853 | s->tok = t; | |
1854 | return s; | |
1855 | } | |
1856 | ||
1857 | ||
1858 | continue_statement* | |
1859 | parser::parse_continue_statement () | |
1860 | { | |
1861 | const token* t = next (); | |
6e213f58 | 1862 | if (! (t->type == tok_keyword && t->content == "continue")) |
2677d2fb | 1863 | throw parse_error (_("expected 'continue'")); |
f3c26ea5 FCE |
1864 | continue_statement* s = new continue_statement; |
1865 | s->tok = t; | |
1866 | return s; | |
1867 | } | |
1868 | ||
1869 | ||
69c68955 FCE |
1870 | for_loop* |
1871 | parser::parse_for_loop () | |
1872 | { | |
f3c26ea5 | 1873 | const token* t = next (); |
6e213f58 | 1874 | if (! (t->type == tok_keyword && t->content == "for")) |
2677d2fb | 1875 | throw parse_error (_("expected 'for'")); |
f3c26ea5 FCE |
1876 | for_loop* s = new for_loop; |
1877 | s->tok = t; | |
1878 | ||
1879 | t = next (); | |
1880 | if (! (t->type == tok_operator && t->content == "(")) | |
2677d2fb | 1881 | throw parse_error (_("expected '('")); |
f3c26ea5 FCE |
1882 | |
1883 | // initializer + ";" | |
1884 | t = peek (); | |
1885 | if (t && t->type == tok_operator && t->content == ";") | |
1886 | { | |
cbfbbf69 FCE |
1887 | s->init = 0; |
1888 | next (); | |
f3c26ea5 FCE |
1889 | } |
1890 | else | |
1891 | { | |
1892 | s->init = parse_expr_statement (); | |
1893 | t = next (); | |
1894 | if (! (t->type == tok_operator && t->content == ";")) | |
2677d2fb | 1895 | throw parse_error (_("expected ';'")); |
f3c26ea5 FCE |
1896 | } |
1897 | ||
1898 | // condition + ";" | |
1899 | t = peek (); | |
1900 | if (t && t->type == tok_operator && t->content == ";") | |
1901 | { | |
1902 | literal_number* l = new literal_number(1); | |
1903 | s->cond = l; | |
1904 | s->cond->tok = next (); | |
1905 | } | |
1906 | else | |
1907 | { | |
1908 | s->cond = parse_expression (); | |
1909 | t = next (); | |
1910 | if (! (t->type == tok_operator && t->content == ";")) | |
2677d2fb | 1911 | throw parse_error (_("expected ';'")); |
f3c26ea5 | 1912 | } |
dff50e09 | 1913 | |
f3c26ea5 FCE |
1914 | // increment + ")" |
1915 | t = peek (); | |
1916 | if (t && t->type == tok_operator && t->content == ")") | |
1917 | { | |
cbfbbf69 FCE |
1918 | s->incr = 0; |
1919 | next (); | |
f3c26ea5 FCE |
1920 | } |
1921 | else | |
1922 | { | |
1923 | s->incr = parse_expr_statement (); | |
1924 | t = next (); | |
1925 | if (! (t->type == tok_operator && t->content == ")")) | |
2677d2fb | 1926 | throw parse_error (_("expected ')'")); |
f3c26ea5 FCE |
1927 | } |
1928 | ||
1929 | // block | |
1930 | s->block = parse_statement (); | |
1931 | ||
1932 | return s; | |
1933 | } | |
1934 | ||
1935 | ||
1936 | for_loop* | |
1937 | parser::parse_while_loop () | |
1938 | { | |
1939 | const token* t = next (); | |
6e213f58 | 1940 | if (! (t->type == tok_keyword && t->content == "while")) |
2677d2fb | 1941 | throw parse_error (_("expected 'while'")); |
f3c26ea5 FCE |
1942 | for_loop* s = new for_loop; |
1943 | s->tok = t; | |
1944 | ||
1945 | t = next (); | |
1946 | if (! (t->type == tok_operator && t->content == "(")) | |
2677d2fb | 1947 | throw parse_error (_("expected '('")); |
f3c26ea5 FCE |
1948 | |
1949 | // dummy init and incr fields | |
cbfbbf69 FCE |
1950 | s->init = 0; |
1951 | s->incr = 0; | |
f3c26ea5 FCE |
1952 | |
1953 | // condition | |
1954 | s->cond = parse_expression (); | |
1955 | ||
f3c26ea5 FCE |
1956 | t = next (); |
1957 | if (! (t->type == tok_operator && t->content == ")")) | |
2677d2fb | 1958 | throw parse_error (_("expected ')'")); |
dff50e09 | 1959 | |
f3c26ea5 FCE |
1960 | // block |
1961 | s->block = parse_statement (); | |
1962 | ||
1963 | return s; | |
69c68955 FCE |
1964 | } |
1965 | ||
1966 | ||
1967 | foreach_loop* | |
1968 | parser::parse_foreach_loop () | |
1969 | { | |
1970 | const token* t = next (); | |
6e213f58 | 1971 | if (! (t->type == tok_keyword && t->content == "foreach")) |
2677d2fb | 1972 | throw parse_error (_("expected 'foreach'")); |
69c68955 FCE |
1973 | foreach_loop* s = new foreach_loop; |
1974 | s->tok = t; | |
93484556 | 1975 | s->sort_direction = 0; |
c261711d | 1976 | s->value = NULL; |
27f21e8c | 1977 | s->limit = NULL; |
69c68955 FCE |
1978 | |
1979 | t = next (); | |
1980 | if (! (t->type == tok_operator && t->content == "(")) | |
2677d2fb | 1981 | throw parse_error (_("expected '('")); |
69c68955 | 1982 | |
c261711d JS |
1983 | symbol* lookahead_sym = NULL; |
1984 | int lookahead_sort = 0; | |
1985 | ||
1986 | t = peek (); | |
1987 | if (t && t->type == tok_identifier) | |
1988 | { | |
1989 | next (); | |
1990 | lookahead_sym = new symbol; | |
1991 | lookahead_sym->tok = t; | |
1992 | lookahead_sym->name = t->content; | |
1993 | ||
1994 | t = peek (); | |
1995 | if (t && t->type == tok_operator && | |
1996 | (t->content == "+" || t->content == "-")) | |
1997 | { | |
1998 | next (); | |
1999 | lookahead_sort = (t->content == "+") ? 1 : -1; | |
2000 | } | |
2001 | ||
2002 | t = peek (); | |
2003 | if (t && t->type == tok_operator && t->content == "=") | |
2004 | { | |
2005 | next (); | |
2006 | s->value = lookahead_sym; | |
2007 | if (lookahead_sort) | |
2008 | { | |
2009 | s->sort_direction = lookahead_sort; | |
2010 | s->sort_column = 0; | |
2011 | } | |
2012 | lookahead_sym = NULL; | |
2013 | } | |
2014 | } | |
2015 | ||
69c68955 FCE |
2016 | // see also parse_array_in |
2017 | ||
2018 | bool parenthesized = false; | |
2019 | t = peek (); | |
c261711d | 2020 | if (!lookahead_sym && t && t->type == tok_operator && t->content == "[") |
69c68955 FCE |
2021 | { |
2022 | next (); | |
2023 | parenthesized = true; | |
2024 | } | |
2025 | ||
c261711d JS |
2026 | if (lookahead_sym) |
2027 | { | |
2028 | s->indexes.push_back (lookahead_sym); | |
2029 | if (lookahead_sort) | |
2030 | { | |
2031 | s->sort_direction = lookahead_sort; | |
2032 | s->sort_column = 1; | |
2033 | } | |
2034 | lookahead_sym = NULL; | |
2035 | } | |
2036 | else while (1) | |
69c68955 FCE |
2037 | { |
2038 | t = next (); | |
2039 | if (! (t->type == tok_identifier)) | |
2677d2fb | 2040 | throw parse_error (_("expected identifier")); |
69c68955 FCE |
2041 | symbol* sym = new symbol; |
2042 | sym->tok = t; | |
2043 | sym->name = t->content; | |
2044 | s->indexes.push_back (sym); | |
2045 | ||
93484556 FCE |
2046 | t = peek (); |
2047 | if (t && t->type == tok_operator && | |
2048 | (t->content == "+" || t->content == "-")) | |
2049 | { | |
2050 | if (s->sort_direction) | |
2677d2fb | 2051 | throw parse_error (_("multiple sort directives")); |
93484556 FCE |
2052 | s->sort_direction = (t->content == "+") ? 1 : -1; |
2053 | s->sort_column = s->indexes.size(); | |
2054 | next(); | |
2055 | } | |
2056 | ||
69c68955 FCE |
2057 | if (parenthesized) |
2058 | { | |
93484556 | 2059 | t = peek (); |
69c68955 FCE |
2060 | if (t && t->type == tok_operator && t->content == ",") |
2061 | { | |
2062 | next (); | |
2063 | continue; | |
2064 | } | |
2065 | else if (t && t->type == tok_operator && t->content == "]") | |
2066 | { | |
2067 | next (); | |
2068 | break; | |
2069 | } | |
dff50e09 | 2070 | else |
2677d2fb | 2071 | throw parse_error (_("expected ',' or ']'")); |
69c68955 FCE |
2072 | } |
2073 | else | |
2074 | break; // expecting only one expression | |
2075 | } | |
2076 | ||
2077 | t = next (); | |
6e213f58 | 2078 | if (! (t->type == tok_keyword && t->content == "in")) |
2677d2fb | 2079 | throw parse_error (_("expected 'in'")); |
dff50e09 | 2080 | |
d02548c0 | 2081 | s->base = parse_indexable(); |
69c68955 | 2082 | |
93484556 FCE |
2083 | t = peek (); |
2084 | if (t && t->type == tok_operator && | |
2085 | (t->content == "+" || t->content == "-")) | |
2086 | { | |
2087 | if (s->sort_direction) | |
2677d2fb | 2088 | throw parse_error (_("multiple sort directives")); |
93484556 FCE |
2089 | s->sort_direction = (t->content == "+") ? 1 : -1; |
2090 | s->sort_column = 0; | |
2091 | next(); | |
2092 | } | |
2093 | ||
27f21e8c DS |
2094 | t = peek (); |
2095 | if (tok_is(t, tok_keyword, "limit")) | |
2096 | { | |
2097 | next (); // get past the "limit" | |
2098 | s->limit = parse_expression (); | |
2099 | } | |
2100 | ||
69c68955 FCE |
2101 | t = next (); |
2102 | if (! (t->type == tok_operator && t->content == ")")) | |
2103 | throw parse_error ("expected ')'"); | |
2104 | ||
2105 | s->block = parse_statement (); | |
2106 | return s; | |
2107 | } | |
2108 | ||
2109 | ||
2f1a1aea FCE |
2110 | expression* |
2111 | parser::parse_expression () | |
2112 | { | |
2113 | return parse_assignment (); | |
2114 | } | |
2115 | ||
2f1a1aea FCE |
2116 | |
2117 | expression* | |
2118 | parser::parse_assignment () | |
2119 | { | |
2120 | expression* op1 = parse_ternary (); | |
2121 | ||
2122 | const token* t = peek (); | |
82919855 | 2123 | // right-associative operators |
dff50e09 | 2124 | if (t && t->type == tok_operator |
2f1a1aea | 2125 | && (t->content == "=" || |
82919855 | 2126 | t->content == "<<<" || |
2f1a1aea | 2127 | t->content == "+=" || |
bb2e3076 FCE |
2128 | t->content == "-=" || |
2129 | t->content == "*=" || | |
2130 | t->content == "/=" || | |
2131 | t->content == "%=" || | |
2132 | t->content == "<<=" || | |
2133 | t->content == ">>=" || | |
2134 | t->content == "&=" || | |
2135 | t->content == "^=" || | |
2136 | t->content == "|=" || | |
d5d7c2cc | 2137 | t->content == ".=" || |
dff50e09 | 2138 | false)) |
2f1a1aea | 2139 | { |
bb2e3076 | 2140 | // NB: lvalueness is checked during elaboration / translation |
2f1a1aea | 2141 | assignment* e = new assignment; |
56099f08 | 2142 | e->left = op1; |
2f1a1aea | 2143 | e->op = t->content; |
56099f08 | 2144 | e->tok = t; |
2f1a1aea | 2145 | next (); |
82919855 | 2146 | e->right = parse_expression (); |
56099f08 | 2147 | op1 = e; |
2f1a1aea | 2148 | } |
56099f08 FCE |
2149 | |
2150 | return op1; | |
2f1a1aea FCE |
2151 | } |
2152 | ||
2153 | ||
2154 | expression* | |
2155 | parser::parse_ternary () | |
2156 | { | |
2157 | expression* op1 = parse_logical_or (); | |
2158 | ||
2159 | const token* t = peek (); | |
2160 | if (t && t->type == tok_operator && t->content == "?") | |
2161 | { | |
2f1a1aea | 2162 | ternary_expression* e = new ternary_expression; |
56099f08 | 2163 | e->tok = t; |
2f1a1aea | 2164 | e->cond = op1; |
56099f08 FCE |
2165 | next (); |
2166 | e->truevalue = parse_expression (); // XXX | |
2f1a1aea FCE |
2167 | |
2168 | t = next (); | |
2169 | if (! (t->type == tok_operator && t->content == ":")) | |
2677d2fb | 2170 | throw parse_error (_("expected ':'")); |
2f1a1aea | 2171 | |
56099f08 | 2172 | e->falsevalue = parse_expression (); // XXX |
2f1a1aea FCE |
2173 | return e; |
2174 | } | |
2175 | else | |
2176 | return op1; | |
2177 | } | |
2178 | ||
2179 | ||
2180 | expression* | |
2181 | parser::parse_logical_or () | |
2182 | { | |
2183 | expression* op1 = parse_logical_and (); | |
dff50e09 | 2184 | |
2f1a1aea | 2185 | const token* t = peek (); |
56099f08 | 2186 | while (t && t->type == tok_operator && t->content == "||") |
2f1a1aea | 2187 | { |
2f1a1aea | 2188 | logical_or_expr* e = new logical_or_expr; |
56099f08 FCE |
2189 | e->tok = t; |
2190 | e->op = t->content; | |
2f1a1aea | 2191 | e->left = op1; |
56099f08 FCE |
2192 | next (); |
2193 | e->right = parse_logical_and (); | |
2194 | op1 = e; | |
2195 | t = peek (); | |
2f1a1aea | 2196 | } |
56099f08 FCE |
2197 | |
2198 | return op1; | |
2f1a1aea FCE |
2199 | } |
2200 | ||
2201 | ||
2202 | expression* | |
2203 | parser::parse_logical_and () | |
2204 | { | |
bb2e3076 | 2205 | expression* op1 = parse_boolean_or (); |
2f1a1aea FCE |
2206 | |
2207 | const token* t = peek (); | |
56099f08 | 2208 | while (t && t->type == tok_operator && t->content == "&&") |
2f1a1aea | 2209 | { |
2f1a1aea FCE |
2210 | logical_and_expr *e = new logical_and_expr; |
2211 | e->left = op1; | |
56099f08 FCE |
2212 | e->op = t->content; |
2213 | e->tok = t; | |
2214 | next (); | |
bb2e3076 FCE |
2215 | e->right = parse_boolean_or (); |
2216 | op1 = e; | |
2217 | t = peek (); | |
2218 | } | |
2219 | ||
2220 | return op1; | |
2221 | } | |
2222 | ||
2223 | ||
2224 | expression* | |
2225 | parser::parse_boolean_or () | |
2226 | { | |
2227 | expression* op1 = parse_boolean_xor (); | |
2228 | ||
2229 | const token* t = peek (); | |
2230 | while (t && t->type == tok_operator && t->content == "|") | |
2231 | { | |
2232 | binary_expression* e = new binary_expression; | |
2233 | e->left = op1; | |
2234 | e->op = t->content; | |
2235 | e->tok = t; | |
2236 | next (); | |
2237 | e->right = parse_boolean_xor (); | |
2238 | op1 = e; | |
2239 | t = peek (); | |
2240 | } | |
2241 | ||
2242 | return op1; | |
2243 | } | |
2244 | ||
2245 | ||
2246 | expression* | |
2247 | parser::parse_boolean_xor () | |
2248 | { | |
2249 | expression* op1 = parse_boolean_and (); | |
2250 | ||
2251 | const token* t = peek (); | |
2252 | while (t && t->type == tok_operator && t->content == "^") | |
2253 | { | |
2254 | binary_expression* e = new binary_expression; | |
2255 | e->left = op1; | |
2256 | e->op = t->content; | |
2257 | e->tok = t; | |
2258 | next (); | |
2259 | e->right = parse_boolean_and (); | |
2260 | op1 = e; | |
2261 | t = peek (); | |
2262 | } | |
2263 | ||
2264 | return op1; | |
2265 | } | |
2266 | ||
2267 | ||
2268 | expression* | |
2269 | parser::parse_boolean_and () | |
2270 | { | |
2271 | expression* op1 = parse_array_in (); | |
2272 | ||
2273 | const token* t = peek (); | |
2274 | while (t && t->type == tok_operator && t->content == "&") | |
2275 | { | |
2276 | binary_expression* e = new binary_expression; | |
2277 | e->left = op1; | |
2278 | e->op = t->content; | |
2279 | e->tok = t; | |
2280 | next (); | |
56099f08 FCE |
2281 | e->right = parse_array_in (); |
2282 | op1 = e; | |
2283 | t = peek (); | |
2f1a1aea | 2284 | } |
56099f08 FCE |
2285 | |
2286 | return op1; | |
2f1a1aea FCE |
2287 | } |
2288 | ||
2289 | ||
2290 | expression* | |
2291 | parser::parse_array_in () | |
2292 | { | |
ce10591c | 2293 | // This is a very tricky case. All these are legit expressions: |
69c68955 | 2294 | // "a in b" "a+0 in b" "[a,b] in c" "[c,(d+0)] in b" |
ce10591c FCE |
2295 | vector<expression*> indexes; |
2296 | bool parenthesized = false; | |
2f1a1aea FCE |
2297 | |
2298 | const token* t = peek (); | |
69c68955 | 2299 | if (t && t->type == tok_operator && t->content == "[") |
ce10591c FCE |
2300 | { |
2301 | next (); | |
2302 | parenthesized = true; | |
2303 | } | |
2304 | ||
2305 | while (1) | |
2306 | { | |
2307 | expression* op1 = parse_comparison (); | |
2308 | indexes.push_back (op1); | |
2309 | ||
2310 | if (parenthesized) | |
2311 | { | |
2312 | const token* t = peek (); | |
2313 | if (t && t->type == tok_operator && t->content == ",") | |
2314 | { | |
2315 | next (); | |
2316 | continue; | |
2317 | } | |
69c68955 | 2318 | else if (t && t->type == tok_operator && t->content == "]") |
ce10591c FCE |
2319 | { |
2320 | next (); | |
2321 | break; | |
2322 | } | |
dff50e09 | 2323 | else |
2677d2fb | 2324 | throw parse_error (_("expected ',' or ']'")); |
ce10591c FCE |
2325 | } |
2326 | else | |
2327 | break; // expecting only one expression | |
2328 | } | |
2329 | ||
2330 | t = peek (); | |
6e213f58 | 2331 | if (t && t->type == tok_keyword && t->content == "in") |
2f1a1aea | 2332 | { |
2f1a1aea | 2333 | array_in *e = new array_in; |
56099f08 | 2334 | e->tok = t; |
ce10591c FCE |
2335 | next (); // swallow "in" |
2336 | ||
2337 | arrayindex* a = new arrayindex; | |
2338 | a->indexes = indexes; | |
d02548c0 GH |
2339 | a->base = parse_indexable(); |
2340 | a->tok = a->base->get_tok(); | |
ce10591c | 2341 | e->operand = a; |
2f1a1aea FCE |
2342 | return e; |
2343 | } | |
ce10591c FCE |
2344 | else if (indexes.size() == 1) // no "in" - need one expression only |
2345 | return indexes[0]; | |
2f1a1aea | 2346 | else |
2677d2fb | 2347 | throw parse_error (_("unexpected comma-separated expression list")); |
2f1a1aea FCE |
2348 | } |
2349 | ||
2350 | ||
2351 | expression* | |
2352 | parser::parse_comparison () | |
2353 | { | |
bb2e3076 | 2354 | expression* op1 = parse_shift (); |
2f1a1aea FCE |
2355 | |
2356 | const token* t = peek (); | |
dff50e09 | 2357 | while (t && t->type == tok_operator |
553d27a5 FCE |
2358 | && (t->content == ">" || |
2359 | t->content == "<" || | |
2360 | t->content == "==" || | |
2361 | t->content == "!=" || | |
2362 | t->content == "<=" || | |
bb2e3076 | 2363 | t->content == ">=")) |
2f1a1aea FCE |
2364 | { |
2365 | comparison* e = new comparison; | |
2366 | e->left = op1; | |
2367 | e->op = t->content; | |
56099f08 | 2368 | e->tok = t; |
2f1a1aea | 2369 | next (); |
bb2e3076 FCE |
2370 | e->right = parse_shift (); |
2371 | op1 = e; | |
2372 | t = peek (); | |
2373 | } | |
2374 | ||
2375 | return op1; | |
2376 | } | |
2377 | ||
2378 | ||
2379 | expression* | |
2380 | parser::parse_shift () | |
2381 | { | |
2382 | expression* op1 = parse_concatenation (); | |
2383 | ||
2384 | const token* t = peek (); | |
dff50e09 | 2385 | while (t && t->type == tok_operator && |
bb2e3076 FCE |
2386 | (t->content == "<<" || t->content == ">>")) |
2387 | { | |
2388 | binary_expression* e = new binary_expression; | |
2389 | e->left = op1; | |
2390 | e->op = t->content; | |
2391 | e->tok = t; | |
2392 | next (); | |
56099f08 FCE |
2393 | e->right = parse_concatenation (); |
2394 | op1 = e; | |
2395 | t = peek (); | |
2f1a1aea | 2396 | } |
56099f08 FCE |
2397 | |
2398 | return op1; | |
2f1a1aea FCE |
2399 | } |
2400 | ||
2401 | ||
2402 | expression* | |
2403 | parser::parse_concatenation () | |
2404 | { | |
2405 | expression* op1 = parse_additive (); | |
2406 | ||
2407 | const token* t = peek (); | |
2408 | // XXX: the actual awk string-concatenation operator is *whitespace*. | |
2409 | // I don't know how to easily to model that here. | |
56099f08 | 2410 | while (t && t->type == tok_operator && t->content == ".") |
2f1a1aea FCE |
2411 | { |
2412 | concatenation* e = new concatenation; | |
2413 | e->left = op1; | |
2414 | e->op = t->content; | |
56099f08 | 2415 | e->tok = t; |
2f1a1aea | 2416 | next (); |
56099f08 FCE |
2417 | e->right = parse_additive (); |
2418 | op1 = e; | |
2419 | t = peek (); | |
2f1a1aea | 2420 | } |
56099f08 FCE |
2421 | |
2422 | return op1; | |
2f1a1aea FCE |
2423 | } |
2424 | ||
2425 | ||
2426 | expression* | |
2427 | parser::parse_additive () | |
2428 | { | |
2429 | expression* op1 = parse_multiplicative (); | |
2430 | ||
2431 | const token* t = peek (); | |
dff50e09 | 2432 | while (t && t->type == tok_operator |
2f1a1aea FCE |
2433 | && (t->content == "+" || t->content == "-")) |
2434 | { | |
2435 | binary_expression* e = new binary_expression; | |
2436 | e->op = t->content; | |
2437 | e->left = op1; | |
56099f08 | 2438 | e->tok = t; |
2f1a1aea | 2439 | next (); |
56099f08 FCE |
2440 | e->right = parse_multiplicative (); |
2441 | op1 = e; | |
2442 | t = peek (); | |
2f1a1aea | 2443 | } |
56099f08 FCE |
2444 | |
2445 | return op1; | |
2f1a1aea FCE |
2446 | } |
2447 | ||
2448 | ||
2449 | expression* | |
2450 | parser::parse_multiplicative () | |
2451 | { | |
2452 | expression* op1 = parse_unary (); | |
2453 | ||
2454 | const token* t = peek (); | |
dff50e09 | 2455 | while (t && t->type == tok_operator |
2f1a1aea FCE |
2456 | && (t->content == "*" || t->content == "/" || t->content == "%")) |
2457 | { | |
2458 | binary_expression* e = new binary_expression; | |
2459 | e->op = t->content; | |
2460 | e->left = op1; | |
56099f08 | 2461 | e->tok = t; |
2f1a1aea | 2462 | next (); |
56099f08 FCE |
2463 | e->right = parse_unary (); |
2464 | op1 = e; | |
2465 | t = peek (); | |
2f1a1aea | 2466 | } |
56099f08 FCE |
2467 | |
2468 | return op1; | |
2f1a1aea FCE |
2469 | } |
2470 | ||
2471 | ||
2472 | expression* | |
2473 | parser::parse_unary () | |
2474 | { | |
2475 | const token* t = peek (); | |
dff50e09 FCE |
2476 | if (t && t->type == tok_operator |
2477 | && (t->content == "+" || | |
2478 | t->content == "-" || | |
bb2e3076 FCE |
2479 | t->content == "!" || |
2480 | t->content == "~" || | |
2481 | false)) | |
2f1a1aea FCE |
2482 | { |
2483 | unary_expression* e = new unary_expression; | |
2484 | e->op = t->content; | |
56099f08 | 2485 | e->tok = t; |
2f1a1aea | 2486 | next (); |
1cb79a72 | 2487 | e->operand = parse_unary (); |
2f1a1aea FCE |
2488 | return e; |
2489 | } | |
2490 | else | |
bb2e3076 | 2491 | return parse_crement (); |
2f1a1aea FCE |
2492 | } |
2493 | ||
2494 | ||
2495 | expression* | |
2496 | parser::parse_crement () // as in "increment" / "decrement" | |
2497 | { | |
cbfbbf69 FCE |
2498 | // NB: Ideally, we'd parse only a symbol as an operand to the |
2499 | // *crement operators, instead of a general expression value. We'd | |
2500 | // need more complex lookahead code to tell apart the postfix cases. | |
2501 | // So we just punt, and leave it to pass-3 to signal errors on | |
2502 | // cases like "4++". | |
2503 | ||
2f1a1aea | 2504 | const token* t = peek (); |
dff50e09 | 2505 | if (t && t->type == tok_operator |
2f1a1aea FCE |
2506 | && (t->content == "++" || t->content == "--")) |
2507 | { | |
2508 | pre_crement* e = new pre_crement; | |
2509 | e->op = t->content; | |
56099f08 | 2510 | e->tok = t; |
2f1a1aea FCE |
2511 | next (); |
2512 | e->operand = parse_value (); | |
2513 | return e; | |
2514 | } | |
2515 | ||
2516 | // post-crement or non-crement | |
2517 | expression *op1 = parse_value (); | |
dff50e09 | 2518 | |
2f1a1aea | 2519 | t = peek (); |
dff50e09 | 2520 | if (t && t->type == tok_operator |
2f1a1aea FCE |
2521 | && (t->content == "++" || t->content == "--")) |
2522 | { | |
2523 | post_crement* e = new post_crement; | |
2524 | e->op = t->content; | |
56099f08 | 2525 | e->tok = t; |
2f1a1aea FCE |
2526 | next (); |
2527 | e->operand = op1; | |
2528 | return e; | |
2529 | } | |
2530 | else | |
2531 | return op1; | |
2532 | } | |
2533 | ||
2534 | ||
2535 | expression* | |
2536 | parser::parse_value () | |
2537 | { | |
2538 | const token* t = peek (); | |
2539 | if (! t) | |
2677d2fb | 2540 | throw parse_error (_("expected value")); |
2f1a1aea | 2541 | |
7d902887 FCE |
2542 | if (t->type == tok_embedded) |
2543 | { | |
2544 | next (); | |
2545 | if (! privileged) | |
2677d2fb | 2546 | throw parse_error (_("embedded expression code in unprivileged script"), false); |
7d902887 FCE |
2547 | |
2548 | embedded_expr *e = new embedded_expr; | |
2549 | e->tok = t; | |
2550 | e->code = t->content; | |
2551 | return e; | |
2552 | } | |
2553 | ||
2f1a1aea FCE |
2554 | if (t->type == tok_operator && t->content == "(") |
2555 | { | |
2556 | next (); | |
2557 | expression* e = parse_expression (); | |
2558 | t = next (); | |
2559 | if (! (t->type == tok_operator && t->content == ")")) | |
2677d2fb | 2560 | throw parse_error (_("expected ')'")); |
2f1a1aea FCE |
2561 | return e; |
2562 | } | |
03c75a4a JS |
2563 | else if (t->type == tok_operator && t->content == "&") |
2564 | { | |
2565 | next (); | |
d48afc20 | 2566 | return parse_target_symbol (t); |
03c75a4a | 2567 | } |
2f1a1aea FCE |
2568 | else if (t->type == tok_identifier) |
2569 | return parse_symbol (); | |
2570 | else | |
2571 | return parse_literal (); | |
2572 | } | |
2573 | ||
2574 | ||
d02548c0 GH |
2575 | const token * |
2576 | parser::parse_hist_op_or_bare_name (hist_op *&hop, string &name) | |
2577 | { | |
2578 | hop = NULL; | |
2579 | const token* t = expect_ident (name); | |
2580 | if (name == "@hist_linear" || name == "@hist_log") | |
2581 | { | |
2582 | hop = new hist_op; | |
2583 | if (name == "@hist_linear") | |
2584 | hop->htype = hist_linear; | |
2585 | else if (name == "@hist_log") | |
2586 | hop->htype = hist_log; | |
2587 | hop->tok = t; | |
2588 | expect_op("("); | |
2589 | hop->stat = parse_expression (); | |
2590 | int64_t tnum; | |
2591 | if (hop->htype == hist_linear) | |
2592 | { | |
2593 | for (size_t i = 0; i < 3; ++i) | |
2594 | { | |
2595 | expect_op (","); | |
2596 | expect_number (tnum); | |
2597 | hop->params.push_back (tnum); | |
2598 | } | |
2599 | } | |
d02548c0 GH |
2600 | expect_op(")"); |
2601 | } | |
2602 | return t; | |
2603 | } | |
2604 | ||
2605 | ||
2606 | indexable* | |
2607 | parser::parse_indexable () | |
2608 | { | |
2609 | hist_op *hop = NULL; | |
2610 | string name; | |
2611 | const token *tok = parse_hist_op_or_bare_name(hop, name); | |
2612 | if (hop) | |
2613 | return hop; | |
2614 | else | |
2615 | { | |
2616 | symbol* sym = new symbol; | |
2617 | sym->name = name; | |
2618 | sym->tok = tok; | |
2619 | return sym; | |
2620 | } | |
2621 | } | |
2622 | ||
2623 | ||
2624 | // var, indexable[index], func(parms), printf("...", ...), $var, $var->member, @stat_op(stat) | |
30263a73 | 2625 | expression* parser::parse_symbol () |
2f1a1aea | 2626 | { |
d02548c0 GH |
2627 | hist_op *hop = NULL; |
2628 | symbol *sym = NULL; | |
d7f3e0c5 | 2629 | string name; |
d02548c0 GH |
2630 | const token *t = parse_hist_op_or_bare_name(hop, name); |
2631 | ||
2632 | if (!hop) | |
0fefb486 | 2633 | { |
dff50e09 | 2634 | // If we didn't get a hist_op, then we did get an identifier. We can |
d02548c0 GH |
2635 | // now scrutinize this identifier for the various magic forms of identifier |
2636 | // (printf, @stat_op, and $var...) | |
2637 | ||
30263a73 FCE |
2638 | if (name == "@cast" || (name.size()>0 && name[0] == '$')) |
2639 | return parse_target_symbol (t); | |
9b5af295 | 2640 | |
db135493 FCE |
2641 | // NB: PR11343: @defined() is not incompatible with earlier versions |
2642 | // of stap, so no need to check session.compatible for 1.2 | |
30263a73 FCE |
2643 | if (name == "@defined") |
2644 | return parse_defined_op (t); | |
8cc799a5 JS |
2645 | |
2646 | if (name == "@entry") | |
2647 | return parse_entry_op (t); | |
2648 | ||
9b5af295 | 2649 | else if (name.size() > 0 && name[0] == '@') |
d7f3e0c5 | 2650 | { |
d02548c0 GH |
2651 | stat_op *sop = new stat_op; |
2652 | if (name == "@avg") | |
2653 | sop->ctype = sc_average; | |
2654 | else if (name == "@count") | |
2655 | sop->ctype = sc_count; | |
2656 | else if (name == "@sum") | |
2657 | sop->ctype = sc_sum; | |
2658 | else if (name == "@min") | |
2659 | sop->ctype = sc_min; | |
2660 | else if (name == "@max") | |
2661 | sop->ctype = sc_max; | |
2662 | else | |
2677d2fb | 2663 | throw parse_error(_("unknown statistic operator ") + name); |
d02548c0 GH |
2664 | expect_op("("); |
2665 | sop->tok = t; | |
2666 | sop->stat = parse_expression (); | |
2667 | expect_op(")"); | |
2668 | return sop; | |
2669 | } | |
dff50e09 | 2670 | |
d5e178c1 | 2671 | else if (print_format *fmt = print_format::create(t)) |
d02548c0 | 2672 | { |
d02548c0 | 2673 | expect_op("("); |
b15c465c PP |
2674 | if ((name == "print" || name == "println" || |
2675 | name == "sprint" || name == "sprintln") && | |
3cb17058 | 2676 | (peek_kw("@hist_linear") || peek_kw("@hist_log"))) |
a4636912 GH |
2677 | { |
2678 | // We have a special case where we recognize | |
2679 | // print(@hist_foo(bar)) as a magic print-the-histogram | |
2680 | // construct. This is sort of gross but it avoids | |
2681 | // promoting histogram references to typeful | |
2682 | // expressions. | |
dff50e09 | 2683 | |
1bbeef03 GH |
2684 | hop = NULL; |
2685 | t = parse_hist_op_or_bare_name(hop, name); | |
2686 | assert(hop); | |
dff50e09 | 2687 | |
1bbeef03 GH |
2688 | // It is, sadly, possible that even while parsing a |
2689 | // hist_op, we *mis-guessed* and the user wishes to | |
2690 | // print(@hist_op(foo)[bucket]), a scalar. In that case | |
2691 | // we must parse the arrayindex and print an expression. | |
839325a1 JS |
2692 | // |
2693 | // XXX: This still fails if the arrayindex is part of a | |
2694 | // larger expression. To really handle everything, we'd | |
2695 | // need to push back all the hist tokens start over. | |
dff50e09 | 2696 | |
1bbeef03 GH |
2697 | if (!peek_op ("[")) |
2698 | fmt->hist = hop; | |
2699 | else | |
2700 | { | |
2701 | // This is simplified version of the | |
2702 | // multi-array-index parser below, because we can | |
2703 | // only ever have one index on a histogram anyways. | |
2704 | expect_op("["); | |
2705 | struct arrayindex* ai = new arrayindex; | |
2706 | ai->tok = t; | |
2707 | ai->base = hop; | |
2708 | ai->indexes.push_back (parse_expression ()); | |
2709 | expect_op("]"); | |
2710 | fmt->args.push_back(ai); | |
839325a1 JS |
2711 | |
2712 | // Consume any subsequent arguments. | |
2713 | while (!peek_op (")")) | |
2714 | { | |
2715 | expect_op(","); | |
2716 | expression *e = parse_expression (); | |
2717 | fmt->args.push_back(e); | |
2718 | } | |
1bbeef03 | 2719 | } |
a4636912 | 2720 | } |
d7f3e0c5 | 2721 | else |
d02548c0 | 2722 | { |
3cb17058 JS |
2723 | int min_args = 0; |
2724 | if (fmt->print_with_format) | |
2725 | { | |
2726 | // Consume and convert a format string. Agreement between the | |
2727 | // format string and the arguments is postponed to the | |
2728 | // typechecking phase. | |
2729 | string tmp; | |
2730 | expect_unknown (tok_string, tmp); | |
2731 | fmt->raw_components = tmp; | |
2732 | fmt->components = print_format::string_to_components (tmp); | |
2733 | } | |
2734 | else if (fmt->print_with_delim) | |
2735 | { | |
2736 | // Consume a delimiter to separate arguments. | |
2737 | fmt->delimiter.clear(); | |
2738 | fmt->delimiter.type = print_format::conv_literal; | |
2739 | expect_unknown (tok_string, fmt->delimiter.literal_string); | |
2740 | min_args = 2; | |
2741 | } | |
2742 | else | |
2743 | { | |
2744 | // If we are not printing with a format string, we must have | |
2745 | // at least one argument (of any type). | |
2746 | expression *e = parse_expression (); | |
2747 | fmt->args.push_back(e); | |
2748 | } | |
2749 | ||
2750 | // Consume any subsequent arguments. | |
2751 | while (min_args || !peek_op (")")) | |
2752 | { | |
2753 | expect_op(","); | |
2754 | expression *e = parse_expression (); | |
2755 | fmt->args.push_back(e); | |
2756 | if (min_args) | |
2757 | --min_args; | |
2758 | } | |
d02548c0 GH |
2759 | } |
2760 | expect_op(")"); | |
2761 | return fmt; | |
2762 | } | |
dff50e09 | 2763 | |
d02548c0 GH |
2764 | else if (peek_op ("(")) // function call |
2765 | { | |
2766 | next (); | |
2767 | struct functioncall* f = new functioncall; | |
2768 | f->tok = t; | |
2769 | f->function = name; | |
2770 | // Allow empty actual parameter list | |
2771 | if (peek_op (")")) | |
2772 | { | |
2773 | next (); | |
2774 | return f; | |
2775 | } | |
2776 | while (1) | |
2777 | { | |
2778 | f->args.push_back (parse_expression ()); | |
2779 | if (peek_op (")")) | |
2780 | { | |
2781 | next(); | |
2782 | break; | |
2783 | } | |
2784 | else if (peek_op (",")) | |
2785 | { | |
2786 | next(); | |
2787 | continue; | |
2788 | } | |
2789 | else | |
2677d2fb | 2790 | throw parse_error (_("expected ',' or ')'")); |
d02548c0 GH |
2791 | } |
2792 | return f; | |
2793 | } | |
2794 | ||
2795 | else | |
2796 | { | |
2797 | sym = new symbol; | |
2798 | sym->name = name; | |
2799 | sym->tok = t; | |
d7f3e0c5 | 2800 | } |
0fefb486 | 2801 | } |
dff50e09 FCE |
2802 | |
2803 | // By now, either we had a hist_op in the first place, or else | |
d02548c0 GH |
2804 | // we had a plain word and it was converted to a symbol. |
2805 | ||
70c743d8 | 2806 | assert (!hop != !sym); // logical XOR |
d02548c0 GH |
2807 | |
2808 | // All that remains is to check for array indexing | |
2809 | ||
d7f3e0c5 | 2810 | if (peek_op ("[")) // array |
2f1a1aea FCE |
2811 | { |
2812 | next (); | |
2813 | struct arrayindex* ai = new arrayindex; | |
d02548c0 GH |
2814 | ai->tok = t; |
2815 | ||
2816 | if (hop) | |
2817 | ai->base = hop; | |
2818 | else | |
2819 | ai->base = sym; | |
2820 | ||
2f1a1aea FCE |
2821 | while (1) |
2822 | { | |
2823 | ai->indexes.push_back (parse_expression ()); | |
d7f3e0c5 | 2824 | if (peek_op ("]")) |
dff50e09 FCE |
2825 | { |
2826 | next(); | |
2827 | break; | |
d7f3e0c5 GH |
2828 | } |
2829 | else if (peek_op (",")) | |
2830 | { | |
2831 | next(); | |
2832 | continue; | |
2833 | } | |
2f1a1aea | 2834 | else |
2677d2fb | 2835 | throw parse_error (_("expected ',' or ']'")); |
2f1a1aea FCE |
2836 | } |
2837 | return ai; | |
2838 | } | |
d02548c0 GH |
2839 | |
2840 | // If we got to here, we *should* have a symbol; if we have | |
2841 | // a hist_op on its own, it doesn't count as an expression, | |
2842 | // so we throw a parse error. | |
2843 | ||
2844 | if (hop) | |
2677d2fb | 2845 | throw parse_error(_("base histogram operator where expression expected"), t); |
dff50e09 FCE |
2846 | |
2847 | return sym; | |
2f1a1aea | 2848 | } |
56099f08 | 2849 | |
81931eab | 2850 | |
30263a73 FCE |
2851 | // Parse a @cast or $var. Given head token has already been consumed. |
2852 | target_symbol* parser::parse_target_symbol (const token* t) | |
2853 | { | |
d48afc20 JS |
2854 | bool addressof = false; |
2855 | if (t->type == tok_operator && t->content == "&") | |
2856 | { | |
2857 | addressof = true; | |
2858 | t = next (); | |
2859 | } | |
2860 | ||
30263a73 FCE |
2861 | if (t->type == tok_identifier && t->content == "@cast") |
2862 | { | |
2863 | cast_op *cop = new cast_op; | |
2864 | cop->tok = t; | |
277c21bc | 2865 | cop->name = t->content; |
30263a73 FCE |
2866 | expect_op("("); |
2867 | cop->operand = parse_expression (); | |
2868 | expect_op(","); | |
7f6b80bd | 2869 | expect_unknown(tok_string, cop->type_name); |
30263a73 FCE |
2870 | if (peek_op (",")) |
2871 | { | |
2872 | next(); | |
2873 | expect_unknown(tok_string, cop->module); | |
2874 | } | |
2875 | expect_op(")"); | |
2876 | parse_target_symbol_components(cop); | |
d48afc20 | 2877 | cop->addressof = addressof; |
30263a73 FCE |
2878 | return cop; |
2879 | } | |
2880 | ||
2881 | if (t->type == tok_identifier && t->content[0]=='$') | |
2882 | { | |
2883 | // target_symbol time | |
2884 | target_symbol *tsym = new target_symbol; | |
2885 | tsym->tok = t; | |
277c21bc | 2886 | tsym->name = t->content; |
30263a73 | 2887 | parse_target_symbol_components(tsym); |
d48afc20 | 2888 | tsym->addressof = addressof; |
30263a73 FCE |
2889 | return tsym; |
2890 | } | |
2891 | ||
2677d2fb | 2892 | throw parse_error (_("expected @cast or $var")); |
30263a73 FCE |
2893 | } |
2894 | ||
2895 | ||
2896 | // Parse a @defined(). Given head token has already been consumed. | |
2897 | expression* parser::parse_defined_op (const token* t) | |
2898 | { | |
2899 | defined_op* dop = new defined_op; | |
2900 | dop->tok = t; | |
2901 | expect_op("("); | |
30263a73 | 2902 | // no need for parse_hist_op... etc., as @defined takes only target_symbols as its operand. |
d48afc20 | 2903 | const token* tt = next (); |
30263a73 FCE |
2904 | dop->operand = parse_target_symbol (tt); |
2905 | expect_op(")"); | |
2906 | return dop; | |
2907 | } | |
2908 | ||
2909 | ||
8cc799a5 JS |
2910 | // Parse a @entry(). Given head token has already been consumed. |
2911 | expression* parser::parse_entry_op (const token* t) | |
2912 | { | |
2913 | entry_op* eop = new entry_op; | |
2914 | eop->tok = t; | |
2915 | expect_op("("); | |
2916 | eop->operand = parse_expression (); | |
2917 | expect_op(")"); | |
2918 | return eop; | |
2919 | } | |
2920 | ||
2921 | ||
30263a73 | 2922 | |
81931eab JS |
2923 | void |
2924 | parser::parse_target_symbol_components (target_symbol* e) | |
2925 | { | |
5f36109e JS |
2926 | bool pprint = false; |
2927 | ||
2928 | // check for pretty-print in the form $foo$ | |
277c21bc | 2929 | string &base = e->name; |
5f36109e JS |
2930 | size_t pprint_pos = base.find_last_not_of('$'); |
2931 | if (0 < pprint_pos && pprint_pos < base.length() - 1) | |
2932 | { | |
2933 | string pprint_val = base.substr(pprint_pos + 1); | |
2934 | base.erase(pprint_pos + 1); | |
2935 | e->components.push_back (target_symbol::component(e->tok, pprint_val, true)); | |
2936 | pprint = true; | |
2937 | } | |
2938 | ||
2939 | while (!pprint) | |
81931eab | 2940 | { |
81931eab JS |
2941 | if (peek_op ("->")) |
2942 | { | |
c67847a0 JS |
2943 | const token* t = next(); |
2944 | string member; | |
2945 | expect_ident_or_keyword (member); | |
5f36109e JS |
2946 | |
2947 | // check for pretty-print in the form $foo->$ or $foo->bar$ | |
2948 | pprint_pos = member.find_last_not_of('$'); | |
2949 | string pprint_val; | |
2950 | if (pprint_pos == string::npos || pprint_pos < member.length() - 1) | |
2951 | { | |
2952 | pprint_val = member.substr(pprint_pos + 1); | |
2953 | member.erase(pprint_pos + 1); | |
2954 | pprint = true; | |
2955 | } | |
2956 | ||
2957 | if (!member.empty()) | |
2958 | e->components.push_back (target_symbol::component(t, member)); | |
2959 | if (pprint) | |
2960 | e->components.push_back (target_symbol::component(t, pprint_val, true)); | |
81931eab JS |
2961 | } |
2962 | else if (peek_op ("[")) | |
2963 | { | |
c67847a0 | 2964 | const token* t = next(); |
6fda2dff JS |
2965 | expression* index = parse_expression(); |
2966 | literal_number* ln = dynamic_cast<literal_number*>(index); | |
2967 | if (ln) | |
2968 | e->components.push_back (target_symbol::component(t, ln->value)); | |
2969 | else | |
2970 | e->components.push_back (target_symbol::component(t, index)); | |
81931eab | 2971 | expect_op ("]"); |
81931eab JS |
2972 | } |
2973 | else | |
2974 | break; | |
2975 | } | |
5f36109e JS |
2976 | |
2977 | if (!pprint) | |
2978 | { | |
2979 | // check for pretty-print in the form $foo $ | |
2980 | // i.e. as a separate token, esp. for $foo[i]$ and @cast(...)$ | |
2981 | const token* t = peek(); | |
2982 | if (t->type == tok_identifier && | |
2983 | t->content.find_first_not_of('$') == string::npos) | |
2984 | { | |
2985 | t = next(); | |
2986 | e->components.push_back (target_symbol::component(t, t->content, true)); | |
2987 | pprint = true; | |
2988 | } | |
2989 | } | |
2990 | ||
2991 | if (pprint && (peek_op ("->") || peek_op("["))) | |
ce0f6648 | 2992 | throw parse_error(_("-> and [ are not accepted for a pretty-printing variable")); |
81931eab JS |
2993 | } |
2994 | ||
73267b89 | 2995 | /* vim: set sw=2 ts=8 cino=>4,n-2,{2,^-2,t0,(0,u0,w1,M1 : */ |