More experimental convert code evolution.

src/pcre2.h

@@ -187,7 +187,7 @@ ignored for pcre2_jit_match(). */
 #define PCRE2_CONVERT_POSIX_BASIC            0x00000004u
 #define PCRE2_CONVERT_POSIX_EXTENDED         0x00000008u
 #define PCRE2_CONVERT_GLOB                   0x00000010u
-#define PCRE2_CONVERT_GLOB_NO_BACKSLASH      0x00000030u
+#define PCRE2_CONVERT_GLOB_NO_DOT_SPECIAL    0x00000030u
 #define PCRE2_CONVERT_GLOB_NO_WILD_SEPARATOR 0x00000050u
 #define PCRE2_CONVERT_GLOB_NO_STARSTAR       0x00000090u
 #define PCRE2_CONVERT_GLOB_BASIC             0x000000f0u
@@ -496,6 +496,8 @@ PCRE2_EXP_DECL pcre2_convert_context PCRE2_CALL_CONVENTION \
   *pcre2_convert_context_create(pcre2_general_context *); \
 PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
   pcre2_convert_context_free(pcre2_convert_context *); \
+PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
+  pcre2_set_glob_escape(pcre2_convert_context *, uint32_t); \
 PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
   pcre2_set_glob_separator(pcre2_convert_context *, uint32_t);
 
@@ -733,6 +735,7 @@ pcre2_compile are called by application code. */
 #define pcre2_set_compile_extra_options       PCRE2_SUFFIX(pcre2_set_compile_extra_options_)
 #define pcre2_set_compile_recursion_guard     PCRE2_SUFFIX(pcre2_set_compile_recursion_guard_)
 #define pcre2_set_depth_limit                 PCRE2_SUFFIX(pcre2_set_depth_limit_)
+#define pcre2_set_glob_escape                 PCRE2_SUFFIX(pcre2_set_glob_escape_)
 #define pcre2_set_glob_separator              PCRE2_SUFFIX(pcre2_set_glob_separator_)
 #define pcre2_set_heap_limit                  PCRE2_SUFFIX(pcre2_set_heap_limit_)
 #define pcre2_set_match_limit                 PCRE2_SUFFIX(pcre2_set_match_limit_)

src/pcre2.h.in

@@ -187,7 +187,7 @@ ignored for pcre2_jit_match(). */
 #define PCRE2_CONVERT_POSIX_BASIC            0x00000004u
 #define PCRE2_CONVERT_POSIX_EXTENDED         0x00000008u
 #define PCRE2_CONVERT_GLOB                   0x00000010u
-#define PCRE2_CONVERT_GLOB_NO_BACKSLASH      0x00000030u
+#define PCRE2_CONVERT_GLOB_NO_DOT_SPECIAL    0x00000030u
 #define PCRE2_CONVERT_GLOB_NO_WILD_SEPARATOR 0x00000050u
 #define PCRE2_CONVERT_GLOB_NO_STARSTAR       0x00000090u
 #define PCRE2_CONVERT_GLOB_BASIC             0x000000f0u
@@ -496,6 +496,8 @@ PCRE2_EXP_DECL pcre2_convert_context PCRE2_CALL_CONVENTION \
   *pcre2_convert_context_create(pcre2_general_context *); \
 PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
   pcre2_convert_context_free(pcre2_convert_context *); \
+PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
+  pcre2_set_glob_escape(pcre2_convert_context *, uint32_t); \
 PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
   pcre2_set_glob_separator(pcre2_convert_context *, uint32_t);
 
@@ -733,6 +735,7 @@ pcre2_compile are called by application code. */
 #define pcre2_set_compile_extra_options       PCRE2_SUFFIX(pcre2_set_compile_extra_options_)
 #define pcre2_set_compile_recursion_guard     PCRE2_SUFFIX(pcre2_set_compile_recursion_guard_)
 #define pcre2_set_depth_limit                 PCRE2_SUFFIX(pcre2_set_depth_limit_)
+#define pcre2_set_glob_escape                 PCRE2_SUFFIX(pcre2_set_glob_escape_)
 #define pcre2_set_glob_separator              PCRE2_SUFFIX(pcre2_set_glob_separator_)
 #define pcre2_set_heap_limit                  PCRE2_SUFFIX(pcre2_set_heap_limit_)
 #define pcre2_set_match_limit                 PCRE2_SUFFIX(pcre2_set_match_limit_)

src/pcre2_context.c

@@ -189,15 +189,17 @@ return mcontext;
 }
 
 
-/* A default covert context is set up to save having to initialize at run time
+/* A default convert context is set up to save having to initialize at run time
 when no context is supplied to the convert function. */
 
 const pcre2_convert_context PRIV(default_convert_context) = {
   { default_malloc, default_free, NULL },    /* Default memory handling */
 #ifdef _WIN32
-  CHAR_BACKSLASH                             /* Default path separator */
+  CHAR_BACKSLASH,                            /* Default path separator */
-#else                                        /* is OS dependent */
+  CHAR_GRAVE_ACCENT                          /* Default escape character */
-  CHAR_SLASH                                 /* Not Windows */
+#else  /* Not Windows */
+  CHAR_SLASH,                                /* Default path separator */  
+  CHAR_BACKSLASH                             /* Default escape character */
 #endif
   };
 
@@ -454,6 +456,14 @@ ccontext->glob_separator = separator;
 return 0;
 }
 
+PCRE2_EXP_DEFN int PCRE2_CALL_CONVENTION
+pcre2_set_glob_escape(pcre2_convert_context *ccontext, uint32_t escape)
+{
+if (escape > 255 || (escape != 0 && !ispunct(escape)))
+  return PCRE2_ERROR_BADDATA;
+ccontext->glob_escape = escape;
+return 0;
+}
 
 /* End of pcre2_context.c */
 

src/pcre2_convert.c

@@ -49,8 +49,10 @@ POSSIBILITY OF SUCH DAMAGE.
   PCRE2_CONVERT_POSIX_BASIC|PCRE2_CONVERT_POSIX_EXTENDED)
 
 #define ALL_OPTIONS (PCRE2_CONVERT_UTF|PCRE2_CONVERT_NO_UTF_CHECK| \
-  PCRE2_CONVERT_GLOB_NO_BACKSLASH|PCRE2_CONVERT_GLOB_NO_WILD_SEPARATOR| \
+  PCRE2_CONVERT_GLOB_NO_DOT_SPECIAL| \
-  PCRE2_CONVERT_GLOB_NO_STARSTAR|TYPE_OPTIONS)
+  PCRE2_CONVERT_GLOB_NO_WILD_SEPARATOR| \
+  PCRE2_CONVERT_GLOB_NO_STARSTAR| \
+  TYPE_OPTIONS)
 
 #define DUMMY_BUFFER_SIZE 100
 
@@ -726,7 +728,7 @@ PCRE2_SPTR pattern_start = pattern;
 PCRE2_SPTR pattern_end = pattern + plength;
 PCRE2_UCHAR separator = ccontext->glob_separator;
 PCRE2_UCHAR c;
-BOOL no_backslash = (options & PCRE2_CONVERT_GLOB_NO_BACKSLASH) != 0;
+BOOL no_escape = ccontext->glob_escape == 0;
 BOOL no_wildsep = (options & PCRE2_CONVERT_GLOB_NO_WILD_SEPARATOR) != 0;
 BOOL no_starstar = (options & PCRE2_CONVERT_GLOB_NO_STARSTAR) != 0;
 BOOL in_atomic = FALSE;
@@ -734,6 +736,8 @@ BOOL after_starstar = FALSE;
 BOOL with_escape, is_start;
 int result, len;
 
+(void)utf;  /* Avoid compiler warning */
+
 if (separator >= 128)
   {
   /* Currently only ASCII separators are supported. */
@@ -805,7 +809,7 @@ while (pattern < pattern_end)
         break;
         }
 
-      if (!no_backslash && *pattern == CHAR_BACKSLASH)
+      if (!no_escape && *pattern == ccontext->glob_escape)
         {
         pattern++;
         if (pattern >= pattern_end)
@@ -925,11 +929,11 @@ while (pattern < pattern_end)
     continue;
     }
 
-  if (!no_backslash && c == CHAR_BACKSLASH)
+  if (!no_escape && c == ccontext->glob_escape)
     {
     if (pattern >= pattern_end)
       {
-      result = ERROR_END_BACKSLASH;
+      result = PCRE2_ERROR_CONVERT_SYNTAX;
       break;
       }
     c = *pattern++;

src/pcre2_intmodedep.h

@@ -596,6 +596,7 @@ typedef struct pcre2_real_match_context {
 typedef struct pcre2_real_convert_context {
   pcre2_memctl memctl;
   uint32_t glob_separator;
+  uint32_t glob_escape;
 } pcre2_real_convert_context;
 
 /* The real compiled code structure. The type for the blocksize field is

src/pcre2test.c

@@ -400,14 +400,14 @@ typedef struct convertstruct {
 } convertstruct;
 
 static convertstruct convertlist[] = {
-  { "glob",              PCRE2_CONVERT_GLOB },
+  { "glob",                   PCRE2_CONVERT_GLOB },
-  { "glob_basic",        PCRE2_CONVERT_GLOB_BASIC },
+  { "glob_basic",             PCRE2_CONVERT_GLOB_BASIC },
-  { "glob_no_backslash", PCRE2_CONVERT_GLOB_NO_BACKSLASH }, 
+  { "glob_no_dot_special",    PCRE2_CONVERT_GLOB_NO_DOT_SPECIAL }, 
-  { "glob_no_starstar",  PCRE2_CONVERT_GLOB_NO_STARSTAR }, 
+  { "glob_no_starstar",       PCRE2_CONVERT_GLOB_NO_STARSTAR }, 
   { "glob_no_wild_separator", PCRE2_CONVERT_GLOB_NO_WILD_SEPARATOR }, 
-  { "posix_basic",       PCRE2_CONVERT_POSIX_BASIC },
+  { "posix_basic",            PCRE2_CONVERT_POSIX_BASIC },
-  { "posix_extended",    PCRE2_CONVERT_POSIX_EXTENDED },
+  { "posix_extended",         PCRE2_CONVERT_POSIX_EXTENDED },
-  { "unset",             CONVERT_UNSET }};
+  { "unset",                  CONVERT_UNSET }};
 
 #define convertlistcount (sizeof(convertlist)/sizeof(convertstruct))
 
@@ -524,6 +524,7 @@ typedef struct patctl {    /* Structure for pattern modifiers. */
   uint32_t  tables_id;
   uint32_t  convert_type;
   uint32_t  convert_length;
+  uint32_t  convert_glob_escape;
   uint32_t  convert_glob_separator;
   uint32_t  regerror_buffsize;
    uint8_t  locale[LOCALESIZE];
@@ -599,6 +600,7 @@ static modstruct modlist[] = {
   { "callout_none",               MOD_DAT,  MOD_CTL, CTL_CALLOUT_NONE,           DO(control) },
   { "caseless",                   MOD_PATP, MOD_OPT, PCRE2_CASELESS,             PO(options) },
   { "convert",                    MOD_PAT,  MOD_CON, 0,                          PO(convert_type) },
+  { "convert_glob_escape",        MOD_PAT,  MOD_CHR, 0,                          PO(convert_glob_escape) },
   { "convert_glob_separator",     MOD_PAT,  MOD_CHR, 0,                          PO(convert_glob_separator) },
   { "convert_length",             MOD_PAT,  MOD_INT, 0,                          PO(convert_length) },
   { "copy",                       MOD_DAT,  MOD_NN,  DO(copy_numbers),           DO(copy_names) },
@@ -1286,6 +1288,14 @@ are supported. */
   else \
     r = pcre2_set_glob_separator_32(G(a,32),b)
 
+#define PCRE2_SET_GLOB_ESCAPE(r,a,b) \
+  if (test_mode == PCRE8_MODE) \
+    r = pcre2_set_glob_escape_8(G(a,8),b); \
+  else if (test_mode == PCRE16_MODE) \
+    r = pcre2_set_glob_escape_16(G(a,16),b); \
+  else \
+    r = pcre2_set_glob_escape_32(G(a,32),b)
+
 #define PCRE2_SET_HEAP_LIMIT(a,b) \
   if (test_mode == PCRE8_MODE) \
     pcre2_set_heap_limit_8(G(a,8),b); \
@@ -1753,6 +1763,12 @@ the three different cases. */
   else \
     G(pcre2_set_depth_limit_,BITTWO)(G(a,BITTWO),b)
 
+#define PCRE2_SET_GLOB_ESCAPE(r,a,b) \
+  if (test_mode == G(G(PCRE,BITONE),_MODE)) \
+    r = G(pcre2_set_glob_escape_,BITONE)(G(a,BITONE),b); \
+  else \
+    r = G(pcre2_set_glob_escape_,BITTWO)(G(a,BITTWO),b)
+
 #define PCRE2_SET_GLOB_SEPARATOR(r,a,b) \
   if (test_mode == G(G(PCRE,BITONE),_MODE)) \
     r = G(pcre2_set_glob_separator_,BITONE)(G(a,BITONE),b); \
@@ -1983,6 +1999,7 @@ the three different cases. */
 #define PCRE2_SET_COMPILE_RECURSION_GUARD(a,b,c) \
   pcre2_set_compile_recursion_guard_8(G(a,8),b,c)
 #define PCRE2_SET_DEPTH_LIMIT(a,b) pcre2_set_depth_limit_8(G(a,8),b)
+#define PCRE2_SET_GLOB_ESCAPE(r,a,b) r = pcre2_set_glob_escape_8(G(a,8),b)
 #define PCRE2_SET_GLOB_SEPARATOR(r,a,b) r = pcre2_set_glob_separator_8(G(a,8),b)
 #define PCRE2_SET_HEAP_LIMIT(a,b) pcre2_set_heap_limit_8(G(a,8),b)
 #define PCRE2_SET_MATCH_LIMIT(a,b) pcre2_set_match_limit_8(G(a,8),b)
@@ -2086,6 +2103,7 @@ the three different cases. */
 #define PCRE2_SET_COMPILE_RECURSION_GUARD(a,b,c) \
   pcre2_set_compile_recursion_guard_16(G(a,16),b,c)
 #define PCRE2_SET_DEPTH_LIMIT(a,b) pcre2_set_depth_limit_16(G(a,16),b)
+#define PCRE2_SET_GLOB_ESCAPE(r,a,b) r = pcre2_set_glob_escape_16(G(a,16),b)
 #define PCRE2_SET_GLOB_SEPARATOR(r,a,b) r = pcre2_set_glob_separator_16(G(a,16),b)
 #define PCRE2_SET_HEAP_LIMIT(a,b) pcre2_set_heap_limit_16(G(a,16),b)
 #define PCRE2_SET_MATCH_LIMIT(a,b) pcre2_set_match_limit_16(G(a,16),b)
@@ -2189,6 +2207,7 @@ the three different cases. */
 #define PCRE2_SET_COMPILE_RECURSION_GUARD(a,b,c) \
   pcre2_set_compile_recursion_guard_32(G(a,32),b,c)
 #define PCRE2_SET_DEPTH_LIMIT(a,b) pcre2_set_depth_limit_32(G(a,32),b)
+#define PCRE2_SET_GLOB_ESCAPE(r,a,b) r = pcre2_set_glob_escape_32(G(a,32),b)
 #define PCRE2_SET_GLOB_SEPARATOR(r,a,b) r = pcre2_set_glob_separator_32(G(a,32),b)
 #define PCRE2_SET_HEAP_LIMIT(a,b) pcre2_set_heap_limit_32(G(a,32),b)
 #define PCRE2_SET_MATCH_LIMIT(a,b) pcre2_set_match_limit_32(G(a,32),b)
@@ -2903,13 +2922,14 @@ return yield;
 *************************************************/
 
 /* Must handle UTF-32 strings in utf mode. Yields number of characters printed.
-For printing *MARK strings, a negative length is given.If handed a NULL file,
+For printing *MARK strings, a negative length is given. If handed a NULL file,
 just counts chars without printing. */
 
 static int pchars32(PCRE2_SPTR32 p, int length, BOOL utf, FILE *f)
 {
 int yield = 0;
 (void)(utf);  /* Avoid compiler warning */
+
 if (length < 0) length = p[-1];
 while (length-- > 0)
   {
@@ -5385,6 +5405,21 @@ if (pat_patctl.convert_type != CONVERT_UNSET)
     convert_options |= PCRE2_CONVERT_NO_UTF_CHECK;
 
   CONCTXCPY(con_context, default_con_context);
+
+  if (pat_patctl.convert_glob_escape != 0)
+    {
+    uint32_t escape = (pat_patctl.convert_glob_escape == '0')? 0 :
+      pat_patctl.convert_glob_escape;  
+    PCRE2_SET_GLOB_ESCAPE(rc, con_context, escape);
+    if (rc != 0)
+      {
+      fprintf(outfile, "** Invalid glob escape '%c'\n",
+        pat_patctl.convert_glob_escape);
+      convert_return = PR_SKIP;
+      goto CONVERT_FINISH;
+      }
+    }
+
   if (pat_patctl.convert_glob_separator != 0)
     {
     PCRE2_SET_GLOB_SEPARATOR(rc, con_context, pat_patctl.convert_glob_separator);

testdata/testinput24

@@ -281,12 +281,18 @@
 /??a??/
 
 #pattern convert=unset
-#pattern convert=glob:glob_no_backslash
+#pattern convert=glob,convert_glob_escape=0
 
 /a\b\cd/
 
 /**\/a/
 
+/a`*b/convert_glob_escape=`
+
+/a`*b/convert_glob_escape=0
+
+/a`*b/convert_glob_escape=x
+
 #pattern convert=unset:posix_extended
 
 /a[[:>:]z/

testdata/testoutput24

@@ -429,7 +429,7 @@ No match
 (?s)\A..a..\z
 
 #pattern convert=unset
-#pattern convert=glob:glob_no_backslash
+#pattern convert=glob,convert_glob_escape=0
 
 /a\b\cd/
 (?s)\Aa\\b\\cd\z
@@ -437,6 +437,15 @@ No match
 /**\/a/
 ** Pattern conversion error at offset 2: invalid syntax
 
+/a`*b/convert_glob_escape=`
+(?s)\Aa\*b\z
+
+/a`*b/convert_glob_escape=0
+(?s)\Aa`(*COMMIT)[^/]*?b\z
+
+/a`*b/convert_glob_escape=x
+** Invalid glob escape 'x'
+
 #pattern convert=unset:posix_extended
 
 /a[[:>:]z/