Mercurial > hg > th-libs

changeset 635:d191ded8a790

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Improve the experimental regex matching debugging macros.
author Matti Hamalainen <ccr@tnsp.org>
date Tue, 21 Jan 2020 07:16:18 +0200
parents c09b068ecb32
children c5ce9a4bfc3e
files Makefile.gen tests.c th_regex.c th_regex.h
diffstat 4 files changed, 49 insertions(+), 40 deletions(-) [+]
line wrap: on
line diff
--- a/Makefile.gen	Tue Jan 21 05:00:53 2020 +0200
+++ b/Makefile.gen	Tue Jan 21 07:16:18 2020 +0200
@@ -9,6 +9,7 @@
 CFLAGS += -DTH_PRINTF_DEBUG=1
 CFLAGS += -DTH_USE_OPT_ARG=1
 #CFLAGS += -DTH_EXPERIMENTAL_REGEX=1
+#CFLAGS += -DTH_EXPERIMENTAL_REGEX_DEBUG=1
 
 THLIBS  = ./
 
--- a/tests.c	Tue Jan 21 05:00:53 2020 +0200
+++ b/tests.c	Tue Jan 21 07:16:18 2020 +0200
@@ -881,6 +881,10 @@
         th_regex_ctx *reg = NULL;
         int res;
 
+#ifdef TH_EXPERIMENTAL_REGEX_DEBUG
+        th_dbg_re_flags = TH_DBG_RE_MATCH;
+#endif
+
 #if 0
         res = th_regex_compile(&reg, "z*k+abba fabboa? k{4} [gz]{1,2} foo(bar|zoo)?");
         if (res != THERR_OK)
@@ -933,7 +937,7 @@
             {
 //                { "zoobar"                       , 1, 0 },
                 { "zoo lol bar"                  , 1, 0 },
-//                { "hoho zoo lol lol bar bar"     , 1, 0 },
+                { "hoho zoo lol lol bar bar"     , 1, 0 },
                 { NULL                           , 0, 0 }
             };
 
--- a/th_regex.c	Tue Jan 21 05:00:53 2020 +0200
+++ b/th_regex.c	Tue Jan 21 07:16:18 2020 +0200
@@ -9,25 +9,17 @@
 #include "th_regex.h"
 
 
-//#define DBG_RE_COMPILE 1
-//#define DBG_RE_FREE 1
-#define DBG_RE_MATCH 1
-
-#if defined(DBG_RE_COMPILE)
-#    define DBG_RE_PRINT_COMPILE(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
+#ifdef TH_EXPERIMENTAL_REGEX_DEBUG
+int th_dbg_re_flags = 0;
+#    define DBG_RE_COMPILE(...) do { if (th_dbg_re_flags & TH_DBG_RE_COMPILE) fprintf(stderr, __VA_ARGS__); } while (0)
+#    define DBG_RE_FREE(...)    do { if (th_dbg_re_flags & TH_DBG_RE_FREE) fprintf(stderr, __VA_ARGS__); } while (0)
+#    define DBG_RE_MATCH(...)   do { if (th_dbg_re_flags & TH_DBG_RE_MATCH) fprintf(stderr, __VA_ARGS__); } while (0)
 #else
-#    define DBG_RE_PRINT_COMPILE(...) do { } while (0)
+#    define DBG_RE_COMPILE(...)
+#    define DBG_RE_FREE(...)
+#    define DBG_RE_MATCH(...)
 #endif
-#if defined(DBG_RE_FREE)
-#    define DBG_RE_PRINT_FREE(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
-#else
-#    define DBG_RE_PRINT_FREE(...) do { } while (0)
-#endif
-#if defined(DBG_RE_MATCH)
-#    define DBG_RE_PRINT_MATCH(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
-#else
-#    define DBG_RE_PRINT_MATCH(...) do { } while (0)
-#endif
+
 
 #ifdef TH_EXPERIMENTAL_REGEX
 
@@ -51,7 +43,7 @@
 };
 
 
-#if defined(DBG_RE_COMPILE) || defined(DBG_RE_FREE) || defined(DBG_RE_MATCH)
+#ifdef TH_EXPERIMENTAL_REGEX_DEBUG
 static const char *re_match_modes[] =
 {
     "ONCE",
@@ -178,7 +170,7 @@
     memcpy(&data->nodes[data->nnodes], node, sizeof(th_regex_node));
     data->nnodes++;
 
-    DBG_RE_PRINT_COMPILE(
+    DBG_RE_COMPILE(
         "node [%" PRIu_SIZE_T " / %" PRIu_SIZE_T "]: "
         "mode=%d, type=%d, rmin=%" PRId_SSIZE_T ", rmax=%" PRId_SSIZE_T "\n",
         data->nnodes, data->nodessize,
@@ -255,7 +247,7 @@
     for (; ctx.pattern[ctx.offs] != 0; ctx.offs++)
     {
         th_regex_char cch = ctx.pattern[ctx.offs];
-        DBG_RE_PRINT_COMPILE("[%" PRIu_SIZE_T "] '%c'\n", ctx.offs, cch);
+        DBG_RE_COMPILE("[%" PRIu_SIZE_T "] '%c'\n", ctx.offs, cch);
         switch (cch)
         {
             case '?':
@@ -461,7 +453,7 @@
         {
             th_regex_node *node = &expr->nodes[n];
 
-            DBG_RE_PRINT_FREE(
+            DBG_RE_FREE(
                 "node [%" PRIu_SIZE_T " / %" PRIu_SIZE_T "]: "
                 "mode=%d, type=%d, rmin=%" PRId_SSIZE_T ", rmax=%" PRId_SSIZE_T "\n",
                 n, expr->nnodes,
@@ -471,23 +463,23 @@
             switch (node->type)
             {
                 case TH_RE_TYPE_SUBEXPR:
-                    DBG_RE_PRINT_FREE("  SUBEXPR: %p vs %p\n", expr, node->match.expr);
+                    DBG_RE_FREE("  SUBEXPR: %p vs %p\n", (void *) expr, (void *) node->match.expr);
                     th_regex_free(node->match.expr);
                     break;
 
                 case TH_RE_TYPE_LIST:
                 case TH_RE_TYPE_LIST_REVERSE:
-                    DBG_RE_PRINT_FREE("  list='%s'\n", node->match.list.chars);
+                    DBG_RE_FREE("  list='%s'\n", node->match.list.chars);
                     th_free(node->match.list.chars);
                     break;
 
-#ifdef DBG_RE_FREE
+#ifdef TH_EXPERIMENTAL_REGEX_DEBUG
                 case TH_RE_TYPE_ANY_CHAR:
-                    DBG_RE_PRINT_FREE("  any char\n");
+                    DBG_RE_FREE("  any char\n");
                     break;
 
                 case TH_RE_TYPE_CHAR:
-                    DBG_RE_PRINT_FREE("  char='%c'\n", node->match.chr);
+                    DBG_RE_FREE("  char='%c'\n", node->match.chr);
                     break;
 #endif
             }
@@ -508,18 +500,18 @@
     th_regex_char cch;
     BOOL ret = FALSE;
 
-    DBG_RE_PRINT_MATCH("    node_START [%s]: '%s': ", re_match_types[node->type], haystack + *offs);
+    DBG_RE_MATCH("    node_START [%s]: '%s': ", re_match_types[node->type], haystack + *offs);
 
     switch (node->type)
     {
         case TH_RE_TYPE_SUBEXPR:
-            DBG_RE_PRINT_MATCH("subexpr ..\n");
+            DBG_RE_MATCH("subexpr ..\n");
             ret = th_regex_do_match_expr(node->match.expr, haystack, offs, flags);
             break;
 
         case TH_RE_TYPE_LIST:
         case TH_RE_TYPE_LIST_REVERSE:
-            DBG_RE_PRINT_MATCH("[%s]\n", node->match.list.chars);
+            DBG_RE_MATCH("[%s]\n", node->match.list.chars);
             ret = FALSE;
             if ((cch = haystack[*offs]) == 0)
                 goto out;
@@ -540,7 +532,7 @@
             break;
 
         case TH_RE_TYPE_ANY_CHAR:
-            DBG_RE_PRINT_MATCH("\n");
+            DBG_RE_MATCH("\n");
             if ((cch = haystack[*offs]) == 0)
             {
                 ret = FALSE;
@@ -552,7 +544,7 @@
             break;
 
         case TH_RE_TYPE_CHAR:
-            DBG_RE_PRINT_MATCH("'%c'\n", node->match.chr);
+            DBG_RE_MATCH("'%c'\n", node->match.chr);
             if ((cch = haystack[*offs]) == 0)
             {
                 ret = FALSE;
@@ -565,7 +557,7 @@
     }
 
 out:
-    DBG_RE_PRINT_MATCH("    node_DONE  [%s]: match %s\n", re_match_types[node->type], ret ? "YES" : "NO");
+    DBG_RE_MATCH("    node_DONE  [%s]: match %s\n", re_match_types[node->type], ret ? "YES" : "NO");
     return ret;
 }
 
@@ -578,14 +570,14 @@
         const th_regex_node *node = &expr->nodes[n];
         size_t soffs;
 
-        DBG_RE_PRINT_MATCH("  expr [%" PRIu_SIZE_T "/%" PRIu_SIZE_T "]: %s ",
+        DBG_RE_MATCH("  expr [%" PRIu_SIZE_T "/%" PRIu_SIZE_T "]: %s ",
             n, expr->nnodes, re_match_modes[node->mode]);
 
         switch (node->mode)
         {
             case TH_RE_MATCH_ONCE:
                 {
-                    DBG_RE_PRINT_MATCH("\n");
+                    DBG_RE_MATCH("\n");
                     soffs = *offs;
                     if (!th_regex_do_match_node(haystack, &soffs, node, flags))
                         return FALSE;
@@ -599,7 +591,7 @@
                     BOOL done = FALSE;
                     ssize_t count = 0;
 
-                    DBG_RE_PRINT_MATCH("min=%" PRId_SSIZE_T ", max=%" PRId_SSIZE_T "\n", node->repeatMin, node->repeatMax);
+                    DBG_RE_MATCH("min=%" PRId_SSIZE_T ", max=%" PRId_SSIZE_T "\n", node->repeatMin, node->repeatMax);
 
                     do
                     {
@@ -631,7 +623,7 @@
                         }
                     } while (!done);
 
-                    DBG_RE_PRINT_MATCH("RESULT: %" PRId_SSIZE_T " = %s\n", count, done ? "YES" : "NO");
+                    DBG_RE_MATCH("RESULT: %" PRId_SSIZE_T " = %s\n", count, done ? "YES" : "NO");
 
                     if (!done)
                         return FALSE;
@@ -641,13 +633,13 @@
                 break;
 
             case TH_RE_MATCH_ANCHOR_START:
-                DBG_RE_PRINT_MATCH("offs=%" PRIu_SIZE_T "\n", *offs);
+                DBG_RE_MATCH("offs=%" PRIu_SIZE_T "\n", *offs);
                 if (*offs != 0)
                     return FALSE;
                 break;
 
             case TH_RE_MATCH_ANCHOR_END:
-                DBG_RE_PRINT_MATCH("is end=%d\n", haystack[*offs]);
+                DBG_RE_MATCH("is end=%d\n", haystack[*offs]);
                 if (haystack[*offs] != 0)
                     return FALSE;
                 break;
@@ -681,7 +673,7 @@
         BOOL matched;
         size_t coffs = soffs;
 
-        DBG_RE_PRINT_MATCH("\nTRY_MATCH @ startoffs=%" PRIu_SIZE_T ": '%s'\n",
+        DBG_RE_MATCH("\nTRY_MATCH @ startoffs=%" PRIu_SIZE_T ": '%s'\n",
             soffs, haystack + soffs);
 
         if ((matched = th_regex_do_match_expr(expr, haystack, &coffs, flags)))
--- a/th_regex.h	Tue Jan 21 05:00:53 2020 +0200
+++ b/th_regex.h	Tue Jan 21 07:16:18 2020 +0200
@@ -68,6 +68,18 @@
 } th_regex_match_node;
 
 
+#ifdef TH_EXPERIMENTAL_REGEX_DEBUG
+enum
+{
+    TH_DBG_RE_COMPILE = 0x0001,
+    TH_DBG_RE_FREE    = 0x0002,
+    TH_DBG_RE_MATCH   = 0x0004,
+};
+
+extern int th_dbg_re_flags;
+#endif
+
+
 //
 // Functions
 //