Small optimizations in pcre2_study.c

ChangeLog

@@ -282,6 +282,8 @@ a factor of the size of the compiling workspace (it currently is).
 81. Check for integer overflow in minimum length calculation and cap it at 
 65535.
 
+82. Small optimizations in code for finding the minimum matching length.
+
 
 Version 10.20 30-June-2015
 --------------------------

doc/pcre2api.3

@@ -1,4 +1,4 @@
-.TH PCRE2API 3 "05 November 2015" "PCRE2 10.21"
+.TH PCRE2API 3 "10 November 2015" "PCRE2 10.21"
 .SH NAME
 PCRE2 - Perl-compatible regular expressions (revised API)
 .sp
@@ -1684,8 +1684,11 @@ value, 0 is returned.
 .sp
   PCRE2_INFO_MATCHEMPTY
 .sp
-Return 1 if the pattern can match an empty string, otherwise 0. The third
+Return 1 if the pattern might match an empty string, otherwise 0. The third
-argument should point to an \fBuint32_t\fP variable.
+argument should point to an \fBuint32_t\fP variable. When a pattern contains 
+recursive subroutine calls it is not always possible to determine whether or 
+not it can match an empty string. PCRE2 takes a cautious approach and returns 1 
+in such cases.
 .sp
   PCRE2_INFO_MATCHLIMIT
 .sp
@@ -3084,6 +3087,6 @@ Cambridge, England.
 .rs
 .sp
 .nf
-Last updated: 05 November 2015
+Last updated: 10 November 2015
 Copyright (c) 1997-2015 University of Cambridge.
 .fi

src/pcre2_study.c

@@ -104,18 +104,21 @@ recurse_check this_recurse;
 register int branchlength = 0;
 register PCRE2_UCHAR *cc = (PCRE2_UCHAR *)code + 1 + LINK_SIZE;
 
+/* If this is a "could be empty" group, its minimum length is 0. */
+
+if (*code >= OP_SBRA && *code <= OP_SCOND) return 0;
+
+/* Skip over capturing bracket number */
+
+if (*code == OP_CBRA || *code == OP_CBRAPOS) cc += IMM2_SIZE;
+
 /* A large and/or complex regex can take too long to process. */
 
 if ((*countptr)++ > 1000) return -1;
 
-/* Skip over capturing bracket number */
+/* Scan along the opcodes for this branch. If we get to the end of the branch,
-
+check the length against that of the other branches. If the accumulated length
-if (*code == OP_CBRA || *code == OP_SCBRA ||
+passes 16-bits, stop. */
-    *code == OP_CBRAPOS || *code == OP_SCBRAPOS) cc += IMM2_SIZE;
-
-/* Scan along the opcodes for this branch. If we get to the end of the
-branch, check the length against that of the other branches. If the accumulated
-length passes 16-bits, stop and return it. */
 
 for (;;)
   {
@@ -1543,24 +1546,28 @@ if ((re->overall_options & PCRE2_ANCHORED) == 0 &&
   if (rc == SSB_DONE) re->flags |= PCRE2_FIRSTMAPSET;
   }
 
-/* Find the minimum length of subject string. */
+/* Find the minimum length of subject string. If it can match an empty string, 
+the minimum length is already known. */
 
-switch(min = find_minlength(re, code, code, utf, NULL, &count))
+if ((re->flags & PCRE2_MATCH_EMPTY) == 0)
   {
-  case -1:  /* \C in UTF mode or (*ACCEPT) or over-complex regex */
+  switch(min = find_minlength(re, code, code, utf, NULL, &count))
-  break;    /* Leave minlength unchanged (will be zero) */
+    {
-
+    case -1:  /* \C in UTF mode or (*ACCEPT) or over-complex regex */
-  case -2:
+    break;    /* Leave minlength unchanged (will be zero) */
-  return 2; /* missing capturing bracket */
+  
-
+    case -2:
-  case -3:
+    return 2; /* missing capturing bracket */
-  return 3; /* unrecognized opcode */
+  
-
+    case -3:
-  default:
+    return 3; /* unrecognized opcode */
-  if (min > UINT16_MAX) min = UINT16_MAX;
+  
-  re->minlength = min;
+    default:
-  break;
+    if (min > UINT16_MAX) min = UINT16_MAX;
-  }
+    re->minlength = min;
+    break;
+    }
+  }   
 
 return 0;
 }

testdata/testoutput15

@@ -252,7 +252,7 @@ Failed: error -52: nested recursion at the same subject position
 /(a|(?R))/I
 Capturing subpattern count = 1
 May match empty string
-Subject length lower bound = 1
+Subject length lower bound = 0
     abcd
  0: a
  1: a
@@ -262,7 +262,7 @@ Failed: error -52: nested recursion at the same subject position
 /(ab|(bc|(de|(?R))))/I
 Capturing subpattern count = 3
 May match empty string
-Subject length lower bound = 2
+Subject length lower bound = 0
     abcd
  0: ab
  1: ab
@@ -272,7 +272,7 @@ Failed: error -52: nested recursion at the same subject position
 /(ab|(bc|(de|(?1))))/I
 Capturing subpattern count = 3
 May match empty string
-Subject length lower bound = 2
+Subject length lower bound = 0
     abcd
  0: ab
  1: ab

testdata/testoutput17

@@ -416,7 +416,7 @@ Failed: error -46: JIT stack limit reached
 /(a|(?R))/I
 Capturing subpattern count = 1
 May match empty string
-Subject length lower bound = 1
+Subject length lower bound = 0
 JIT compilation was successful
     abcd
  0: a (JIT)
@@ -427,7 +427,7 @@ Failed: error -46: JIT stack limit reached
 /(ab|(bc|(de|(?R))))/I
 Capturing subpattern count = 3
 May match empty string
-Subject length lower bound = 2
+Subject length lower bound = 0
 JIT compilation was successful
     abcd
  0: ab (JIT)
@@ -438,7 +438,7 @@ Failed: error -46: JIT stack limit reached
 /(ab|(bc|(de|(?1))))/I
 Capturing subpattern count = 3
 May match empty string
-Subject length lower bound = 2
+Subject length lower bound = 0
 JIT compilation was successful
     abcd
  0: ab (JIT)

testdata/testoutput2

@@ -3960,7 +3960,7 @@ Subject length lower bound = 3
 Capturing subpattern count = 2
 Compile options: <none>
 Overall options: anchored
-Subject length lower bound = 3
+Subject length lower bound = 2
     a=a
  0: a=a
  1: a