Remove duplicated scan substring captures

Author: zherczeg

src/pcre2_compile.c

@@ -6504,6 +6504,60 @@ for (;; pptr++)
     cb->assert_depth += 1;
     goto GROUP_PROCESS;
 
+    /* Handle the capture list of scan substring. */
+    case META_CAPTURE_NAME:   /* Capture name */
+    case META_CAPTURE_NUMBER: /* Capture number */
+    if (lengthptr != NULL)
+      {
+      if (!PRIV(compile_parse_scan_substr_args)(&pptr, offset, errorcodeptr, cb, lengthptr))
+        return 0;
+      break;
+      }
+
+    while (TRUE)
+      {
+      int count, index;
+      named_group *ng;
+
+      switch (META_CODE(*pptr))
+        {
+        case META_OFFSET:
+        pptr++;
+        SKIPOFFSET(pptr);
+        continue;
+
+        case META_CAPTURE_NAME:
+        ng = cb->named_groups + pptr[1];
+        pptr += 2;
+        count = 0;
+        index = 0;
+
+        if (!PRIV(compile_find_dupname_details)(ng->name, ng->length, &index,
+          &count, errorcodeptr, cb)) return 0;
+
+        code[0] = OP_DNCREF;
+        PUT2(code, 1, index);
+        PUT2(code, 1 + IMM2_SIZE, count);
+        code += 1 + 2 * IMM2_SIZE;
+        continue;
+
+        case META_CAPTURE_NUMBER:
+        pptr += 2;
+        if (pptr[-1] == 0) continue;
+
+        code[0] = OP_CREF;
+        PUT2(code, 1, pptr[-1]);
+        code += 1 + IMM2_SIZE;
+        continue;
+
+        default:
+        break;
+        }
+
+      break;
+      }
+    --pptr;
+    break;
 
     /* ===================================================================*/
     /* Handle conditional subpatterns. The case of (?(Rdigits) is ambiguous
@@ -6515,7 +6569,6 @@ for (;; pptr++)
     case META_COND_RNUMBER:   /* (?(Rdigits) */
     case META_COND_NAME:      /* (?(name) or (?'name') or ?(<name>) */
     case META_COND_RNAME:     /* (?(R&name) - test for recursion */
-    case META_CAPTURE_NAME:   /* Generic capture name */
     bravalue = OP_COND;
 
     if (lengthptr != NULL)
@@ -6526,10 +6579,7 @@ for (;; pptr++)
       uint32_t *start_pptr = pptr;
       uint32_t length = *(++pptr);
 
-      if (meta == META_CAPTURE_NAME)
-        offset += meta_arg;
-      else
-        GETPLUSOFFSET(offset, pptr);
+      GETPLUSOFFSET(offset, pptr);
       name = cb->start_pattern + offset;
 
       /* In the first pass, the names generated in the pre-pass are available,
@@ -6592,12 +6642,6 @@ for (;; pptr++)
         start_pptr[0] = meta;
         start_pptr[1] = ng->number;
 
-        if (meta == META_CAPTURE_NAME)
-          {
-          code += 1 + IMM2_SIZE;
-          break;
-          }
-
         skipunits = 1 + IMM2_SIZE;
         goto GROUP_PROCESS_NOTE_EMPTY;
         }
@@ -6608,12 +6652,6 @@ for (;; pptr++)
       start_pptr[0] = meta | 1;
       start_pptr[1] = (uint32_t)(ng - cb->named_groups);
 
-      if (meta == META_CAPTURE_NAME)
-        {
-        code += 1 + 2 * IMM2_SIZE;
-        break;
-        }
-
       /* A duplicated name was found. Note that if an R<digits> name is found
       (META_COND_RNUMBER), it is a reference test, not a recursion test. */
       skipunits = 1 + 2 * IMM2_SIZE;
@@ -6639,15 +6677,6 @@ for (;; pptr++)
 
       if (meta_arg == 0)
         {
-        if (meta == META_CAPTURE_NAME)
-          {
-          code[0] = OP_CREF;
-          PUT2(code, 1, pptr[1]);
-          code += 1 + IMM2_SIZE;
-          pptr++;
-          break;
-          }
-
         code[1+LINK_SIZE] = (meta == META_COND_RNAME)? OP_RREF : OP_CREF;
         PUT2(code, 2 + LINK_SIZE, pptr[1]);
         skipunits = 1 + IMM2_SIZE;
@@ -6663,16 +6692,6 @@ for (;; pptr++)
       if (!PRIV(compile_find_dupname_details)(ng->name, ng->length, &index,
             &count, errorcodeptr, cb)) return 0;
 
-      if (meta == META_CAPTURE_NAME)
-        {
-        code[0] = OP_DNCREF;
-        PUT2(code, 1, index);
-        PUT2(code, 1 + IMM2_SIZE, count);
-        code += 1 + 2 * IMM2_SIZE;
-        pptr++;
-        break;
-        }
-
       /* A duplicated name was found. Note that if an R<digits> name is found
       (META_COND_RNUMBER), it is a reference test, not a recursion test. */
 
@@ -6702,12 +6721,8 @@ for (;; pptr++)
     /* Conditional test of a group's being set. */
 
     case META_COND_NUMBER:
-    case META_CAPTURE_NUMBER:
     bravalue = OP_COND;
-    if (meta == META_CAPTURE_NUMBER)
-      offset += meta_arg;
-    else
-      GETPLUSOFFSET(offset, pptr);
+    GETPLUSOFFSET(offset, pptr);
 
     groupnumber = *(++pptr);
     if (groupnumber > cb->bracount)
@@ -6718,14 +6733,6 @@ for (;; pptr++)
       }
     if (groupnumber > cb->top_backref) cb->top_backref = groupnumber;
 
-    if (meta == META_CAPTURE_NUMBER)
-      {
-      code[0] = OP_CREF;
-      PUT2(code, 1, groupnumber);
-      code += 1+IMM2_SIZE;
-      break;
-      }
-
     /* Point at initial ( for too many branches error */
     offset -= 2;
     code[1+LINK_SIZE] = OP_CREF;

src/pcre2_compile.h

@@ -274,6 +274,7 @@ typedef struct {
 #define _pcre2_compile_find_named_group        PCRE2_SUFFIX(_pcre2_compile_find_named_group)
 #define _pcre2_compile_find_dupname_details    PCRE2_SUFFIX(_pcre2_compile_find_dupname_details)
 #define _pcre2_compile_add_name_to_table       PCRE2_SUFFIX(_pcre2_compile_add_name_to_table)
+#define _pcre2_compile_parse_scan_substr_args  PCRE2_SUFFIX(_pcre2_compile_parse_scan_substr_args)
 #define _pcre2_compile_parse_recurse_args      PCRE2_SUFFIX(_pcre2_compile_parse_recurse_args)
 
 
@@ -342,6 +343,11 @@ BOOL PRIV(compile_find_dupname_details)(PCRE2_SPTR name, uint32_t length,
 
 /* Parse the arguments of recurse operations. */
 
+BOOL PRIV(compile_parse_scan_substr_args)(uint32_t **pptr_start,
+  PCRE2_SIZE offset, int *errorcodeptr, compile_block *cb, PCRE2_SIZE *lengthptr);
+
+/* Parse the arguments of recurse operations. */
+
 BOOL PRIV(compile_parse_recurse_args)(uint32_t *pptr_start,
   PCRE2_SIZE offset, int *errorcodeptr, compile_block *cb);