Improve logic for setting starting code units by ignoring certain assertions.

Author: PhilipHazel

ChangeLog

@@ -184,13 +184,18 @@ PCRE2_CASELESS and PCRE2_UCP (but not PCRE2_UTF) were set. Fixed by not trying
 to look for other cases for characters above the Unicode range.
 
 50. In caseless 32-bit mode with UCP (but not UTF) set, the character
-0xffffffff incorrectly matched any character that has more than one other case, 
+0xffffffff incorrectly matched any character that has more than one other case,
 in particular k and s.
 
 51. Fix accept and endanchored interaction in JIT.
 
 52. Fix backreferences with unset backref and non-greedy iterators in JIT.
 
+53. Improve the logic that checks for a list of starting code units -- positive
+lookahead assertions are now ignored if the immediately following item is one
+that sets a mandatory starting character. For example, /a?(?=bc|)d/ used to set
+all of a, b, and d as possible starting code units; now it sets only a and d.
+
 
 Version 10.42 11-December-2022
 ------------------------------
@@ -214,12 +219,12 @@ maximum of 65535 is now silently applied.
 
 5. Merged @carenas patch #175 which fixes #86 - segfault on aarch64 (ARM),
 
-6. The prototype for pcre2_substring_list_free() specified its argument as 
-PCRE2_SPTR * which is a const data type, whereas the yield from 
-pcre2_substring_list() is not const. This caused compiler warnings. I have 
-changed the argument of pcre2_substring_list_free() to be PCRE2_UCHAR ** to 
-remove this anomaly. This might cause new warnings in existing code where a 
-cast has been used to avoid previous ones. 
+6. The prototype for pcre2_substring_list_free() specified its argument as
+PCRE2_SPTR * which is a const data type, whereas the yield from
+pcre2_substring_list() is not const. This caused compiler warnings. I have
+changed the argument of pcre2_substring_list_free() to be PCRE2_UCHAR ** to
+remove this anomaly. This might cause new warnings in existing code where a
+cast has been used to avoid previous ones.
 
 
 Version 10.41 06-December-2022

src/pcre2_study.c

@@ -979,6 +979,7 @@ do
   while (try_next)    /* Loop for items in this branch */
     {
     int rc;
+    PCRE2_SPTR ncode;
     uint8_t *classmap = NULL;
 #ifdef SUPPORT_WIDE_CHARS
     PCRE2_UCHAR xclassflags;
@@ -1110,10 +1111,53 @@ do
       tcode++;
       break;
 
-      /* If we hit a bracket or a positive lookahead assertion, recurse to set
-      bits from within the subpattern. If it can't find anything, we have to
-      give up. If it finds some mandatory character(s), we are done for this
-      branch. Otherwise, carry on scanning after the subpattern. */
+      /* For a positive lookahead assertion, inspect what immediately follows.
+      If the next item is one that sets a mandatory character, skip this
+      assertion. Otherwise, treat it the same as other bracket groups. */
+
+      case OP_ASSERT:
+      case OP_ASSERT_NA:
+      ncode = tcode + GET(tcode, 1);
+      while (*ncode == OP_ALT) ncode += GET(ncode, 1);
+      ncode += 1 + LINK_SIZE;
+      switch(*ncode)
+        {
+        default:
+        break;
+
+        case OP_PROP:
+        if (ncode[1] != PT_CLIST) break;
+        /* Fall through */
+        case OP_ANYNL:
+        case OP_CHAR:
+        case OP_CHARI:
+        case OP_EXACT:
+        case OP_EXACTI:
+        case OP_HSPACE:
+        case OP_MINPLUS:
+        case OP_MINPLUSI:
+        case OP_PLUS:
+        case OP_PLUSI:
+        case OP_POSPLUS:
+        case OP_POSPLUSI:
+        case OP_VSPACE:
+        /* Note that these types will only be present in non-UCP mode. */
+        case OP_DIGIT:
+        case OP_NOT_DIGIT:
+        case OP_WORDCHAR:
+        case OP_NOT_WORDCHAR:
+        case OP_WHITESPACE:
+        case OP_NOT_WHITESPACE:
+        tcode = ncode;
+        continue;   /* With the following opcode */
+        }
+      /* Fall through */
+
+      /* For a group bracket or a positive assertion without an immediately
+      following mandatory setting, recurse to set bits from within the
+      subpattern. If it can't find anything, we have to give up. If it finds
+      some mandatory character(s), we are done for this branch. Otherwise,
+      carry on scanning after the subpattern. */
 
       case OP_BRA:
       case OP_SBRA:
@@ -1125,8 +1169,6 @@ do
       case OP_SCBRAPOS:
       case OP_ONCE:
       case OP_SCRIPT_RUN:
-      case OP_ASSERT:
-      case OP_ASSERT_NA:
       rc = set_start_bits(re, tcode, utf, ucp, depthptr);
       if (rc == SSB_DONE)
         {

testdata/testinput2

@@ -6075,4 +6075,16 @@ a)"xI
 /(?:|(?0).)(?(R)|\z)/
     abcd
 
+/a?(?=b(*COMMIT)c|)d/I
+    bd
+
+/(?=b(*COMMIT)c|)d/I
+    bd
+
+/a?(?=b(*COMMIT)c|)d/I,no_start_optimize
+    bd
+
+/(?=b(*COMMIT)c|)d/I,no_start_optimize
+    bd
+
 # End of testinput2

testdata/testoutput2

@@ -17977,6 +17977,33 @@ No match
     abcd
  0: abcd
 
+/a?(?=b(*COMMIT)c|)d/I
+Capture group count = 0
+Starting code units: a d 
+Last code unit = 'd'
+Subject length lower bound = 1
+    bd
+ 0: d
+
+/(?=b(*COMMIT)c|)d/I
+Capture group count = 0
+First code unit = 'd'
+Subject length lower bound = 1
+    bd
+ 0: d
+
+/a?(?=b(*COMMIT)c|)d/I,no_start_optimize
+Capture group count = 0
+Options: no_start_optimize
+    bd
+No match
+
+/(?=b(*COMMIT)c|)d/I,no_start_optimize
+Capture group count = 0
+Options: no_start_optimize
+    bd
+No match
+
 # End of testinput2
 Error -70: PCRE2_ERROR_BADDATA (unknown error number)
 Error -62: bad serialized data