From 9ba50046cf72f6c7ddf3407043d835f631246384 Mon Sep 17 00:00:00 2001 From: Alasdair Kergon Date: Thu, 22 Apr 2010 17:42:38 +0000 Subject: [PATCH] Move regex printing code from test to main tree (may use in debug messages). Yet another optimiser fix attempt. --- libdm/regex/parse_rx.c | 85 ++++++++++++++++++++++++++++++++++++++ old-tests/regex/parse_t.c | 86 +++++---------------------------------- 2 files changed, 95 insertions(+), 76 deletions(-) diff --git a/libdm/regex/parse_rx.c b/libdm/regex/parse_rx.c index b1d362ffb..d0894fe52 100644 --- a/libdm/regex/parse_rx.c +++ b/libdm/regex/parse_rx.c @@ -16,6 +16,85 @@ #include "dmlib.h" #include "parse_rx.h" +#ifdef DEBUG +#include + +static void _regex_print(struct rx_node *rx, int depth, unsigned show_nodes) +{ + int i, numchars; + + if (rx->left) { + if (rx->left->type != CHARSET && (show_nodes || (!((rx->type == CAT || rx->type == OR) && rx->left->type == CAT)))) + printf("("); + + _regex_print(rx->left, depth + 1, show_nodes); + + if (rx->left->type != CHARSET && (show_nodes || (!((rx->type == CAT || rx->type == OR) && rx->left->type == CAT)))) + printf(")"); + } + + /* display info about the node */ + switch (rx->type) { + case CAT: + break; + + case OR: + printf("|"); + break; + + case STAR: + printf("*"); + break; + + case PLUS: + printf("+"); + break; + + case QUEST: + printf("?"); + break; + + case CHARSET: + numchars = 0; + for (i = 0; i < 256; i++) + if (dm_bit(rx->charset, i) && (isprint(i) || i == HAT_CHAR || i == DOLLAR_CHAR)) + numchars++; + if (numchars == 97) { + printf("."); + break; + } + if (numchars > 1) + printf("["); + for (i = 0; i < 256; i++) + if (dm_bit(rx->charset, i)) { + if isprint(i) + printf("%c", (char) i); + else if (i == HAT_CHAR) + printf("^"); + else if (i == DOLLAR_CHAR) + printf("$"); + } + if (numchars > 1) + printf("]"); + break; + + default: + fprintf(stderr, "Unknown type"); + } + + if (rx->right) { + if (rx->right->type != CHARSET && (show_nodes || (!(rx->type == CAT && rx->right->type == CAT) && rx->right->right))) + printf("("); + _regex_print(rx->right, depth + 1, show_nodes); + if (rx->right->type != CHARSET && (show_nodes || (!(rx->type == CAT && rx->right->type == CAT) && rx->right->right))) + printf(")"); + } + + if (!depth) + printf("\n"); +} +#endif /* DEBUG */ + struct parse_sp { /* scratch pad for the parsing process */ struct dm_pool *mem; int type; /* token type, 0 indicates a charset */ @@ -342,10 +421,16 @@ static struct rx_node *_or_term(struct parse_sp *ps) static unsigned _depth(struct rx_node *r, unsigned leftmost) { int count = 1; + int or_count = 0; while (r->type != CHARSET && LEFT(r)) { count++; r = LEFT(r); + /* Stop if we pass another OR */ + if (or_count) + break; + if (r->type == OR) + or_count++; } return count; diff --git a/old-tests/regex/parse_t.c b/old-tests/regex/parse_t.c index 90eb1ca8c..54ebb0bfd 100644 --- a/old-tests/regex/parse_t.c +++ b/old-tests/regex/parse_t.c @@ -14,6 +14,7 @@ */ /* hack - using unexported internal function */ +#define DEBUG #include "regex/parse_rx.c" #include @@ -67,86 +68,12 @@ static void _pretty_print(struct rx_node *rx, int depth) _pretty_print(rx->right, depth + 1); } -static void _regex_print(struct rx_node *rx, int depth) -{ - int i, numchars; - int left_and_right = (rx->left && rx->right); - - if (left_and_right && rx->type == CAT && rx->left->type == OR) - printf("("); - - if (rx->left) - _regex_print(rx->left, depth + 1); - - if (left_and_right && rx->type == CAT && rx->left->type == OR) - printf(")"); - - /* display info about the node */ - switch (rx->type) { - case CAT: - //printf("Cat"); - break; - - case OR: - printf("|"); - break; - - case STAR: - printf("*"); - break; - - case PLUS: - printf("+"); - break; - - case QUEST: - printf("?"); - break; - - case CHARSET: - numchars = 0; - for (i = 0; i < 256; i++) - if (dm_bit(rx->charset, i) && (isprint(i) || i == HAT_CHAR || i == DOLLAR_CHAR)) - numchars++; - if (numchars == 97) { - printf("."); - break; - } - if (numchars > 1) - printf("["); - for (i = 0; i < 256; i++) - if (dm_bit(rx->charset, i)) { - if (isprint(i)) - printf("%c", (char) i); - else if (i == HAT_CHAR) - printf("^"); - else if (i == DOLLAR_CHAR) - printf("$"); - } - if (numchars > 1) - printf("]"); - break; - - default: - fprintf(stderr, "Unknown type"); - } - - if (left_and_right && rx->type == CAT && rx->right->type == OR) - printf("("); - if (rx->right) - _regex_print(rx->right, depth + 1); - if (left_and_right && rx->type == CAT && rx->right->type == OR) - printf(")"); - - if (!depth) - printf("\n"); -} - int main(int argc, char **argv) { struct dm_pool *mem; struct rx_node *rx; int regex_print = 0; + int show_nodes = 0; int regex_arg = 1; if (argc == 3 && !strcmp(argv[1], "-r")) { @@ -155,6 +82,13 @@ int main(int argc, char **argv) argc--; } + if (argc == 3 && !strcmp(argv[1], "-R")) { + regex_print++; + show_nodes++; + regex_arg++; + argc--; + } + if (argc != 2) { fprintf(stderr, "Usage : %s [-r] \n", argv[0]); exit(0); @@ -174,7 +108,7 @@ int main(int argc, char **argv) } if (regex_print) - _regex_print(rx, 0); + _regex_print(rx, 0, show_nodes); else _pretty_print(rx, 0); -- 2.43.5