Minor improvement to minimum length calculation.

ChangeLog

@@ -28,6 +28,14 @@ increase it substantially for non-anchored patterns.
 8. Allow (*ACCEPT) to be quantified, because an ungreedy quantifier with a zero 
 minimum is potentially useful.
 
+9. Some changes to the way the minimum subject length is handled:
+
+   * When PCRE2_NO_START_OPTIMIZE is set, no minimum length is computed; 
+     pcre2test no longer shows a value (of zero).
+     
+   * When no minimum length is set by the normal scan, but a first and/or last 
+     code unit is recorded, set the minimum to 1 or 2 as appropriate.
+
 
 Version 10.33 16-April-2019
 ---------------------------

doc/pcre2api.3

@@ -1,4 +1,4 @@
-.TH PCRE2API 3 "30 May 2019" "PCRE2 10.34"
+.TH PCRE2API 3 "11 June 2019" "PCRE2 10.34"
 .SH NAME
 PCRE2 - Perl-compatible regular expressions (revised API)
 .sp
@@ -2229,12 +2229,12 @@ segment.
   PCRE2_INFO_MINLENGTH
 .sp
 If a minimum length for matching subject strings was computed, its value is
-returned. Otherwise the returned value is 0. The value is a number of
+returned. Otherwise the returned value is 0. This value is not computed when
-characters, which in UTF mode may be different from the number of code units.
+PCRE2_NO_START_OPTIMIZE is set. The value is a number of characters, which in
-The third argument should point to an \fBuint32_t\fP variable. The value is a
+UTF mode may be different from the number of code units. The third argument
-lower bound to the length of any matching string. There may not be any strings
+should point to an \fBuint32_t\fP variable. The value is a lower bound to the
-of that length that do actually match, but every string that does match is at
+length of any matching string. There may not be any strings of that length that
-least that long.
+do actually match, but every string that does match is at least that long.
 .sp
   PCRE2_INFO_NAMECOUNT
   PCRE2_INFO_NAMEENTRYSIZE
@@ -3848,6 +3848,6 @@ Cambridge, England.
 .rs
 .sp
 .nf
-Last updated: 30 May 2019
+Last updated: 11 June 2019
 Copyright (c) 1997-2019 University of Cambridge.
 .fi

doc/pcre2test.1

@@ -1,4 +1,4 @@
-.TH PCRE2TEST 1 "23 May 2019" "PCRE 10.34"
+.TH PCRE2TEST 1 "11 June 2019" "PCRE 10.34"
 .SH NAME
 pcre2test - a program for testing Perl-compatible regular expressions.
 .SH SYNOPSIS
@@ -695,7 +695,9 @@ options, the line is omitted. "First code unit" is where any match must start;
 if there is more than one they are listed as "starting code units". "Last code
 unit" is the last literal code unit that must be present in any match. This is
 not necessarily the last character. These lines are omitted if no starting or
-ending code units are recorded.
+ending code units are recorded. The subject length line is omitted when 
+\fBno_start_optimize\fP is set because the minimum length is not calculated 
+when it can never be used.
 .P
 The \fBframesize\fP modifier shows the size, in bytes, of the storage frames
 used by \fBpcre2_match()\fP for handling backtracking. The size depends on the
@@ -2060,6 +2062,6 @@ Cambridge, England.
 .rs
 .sp
 .nf
-Last updated: 23 May 2019
+Last updated: 11 June 2019
 Copyright (c) 1997-2019 University of Cambridge.
 .fi

src/pcre2_compile.c

@@ -10151,6 +10151,8 @@ unit. */
 
 if ((re->overall_options & PCRE2_NO_START_OPTIMIZE) == 0)
   {
+  int minminlength = 0;  /* For minimal minlength from first/required CU */
+
   /* If we do not have a first code unit, see if there is one that is asserted
   (these are not saved during the compile because they can cause conflicts with
   actual literals that follow). */
@@ -10158,12 +10160,14 @@ if ((re->overall_options & PCRE2_NO_START_OPTIMIZE) == 0)
   if (firstcuflags < 0)
     firstcu = find_firstassertedcu(codestart, &firstcuflags, 0);
 
-  /* Save the data for a first code unit. */
+  /* Save the data for a first code unit. The existence of one means the
+  minimum length must be at least 1. */
 
   if (firstcuflags >= 0)
     {
     re->first_codeunit = firstcu;
     re->flags |= PCRE2_FIRSTSET;
+    minminlength++;
 
     /* Handle caseless first code units. */
 
@@ -10197,12 +10201,32 @@ if ((re->overall_options & PCRE2_NO_START_OPTIMIZE) == 0)
            is_startline(codestart, 0, &cb, 0, FALSE))
     re->flags |= PCRE2_STARTLINE;
 
-  /* Handle the "required code unit", if one is set. In the case of an anchored
+  /* Handle the "required code unit", if one is set. We can increment the
-  pattern, do this only if it follows a variable length item in the pattern. */
+  minimum minimum length only if we are sure this really is a different
+  character, because the count is in characters, not code units. */
 
-  if (reqcuflags >= 0 &&
+  if (reqcuflags >= 0)
-       ((re->overall_options & PCRE2_ANCHORED) == 0 ||
+    {
-        (reqcuflags & REQ_VARY) != 0))
+#if PCRE2_CODE_UNIT_WIDTH == 16
+    if ((re->overall_options & PCRE2_UTF) == 0 ||   /* Not UTF */
+        firstcuflags < 0 ||                         /* First not set */
+        (firstcu & 0xf800) != 0xd800 ||             /* First not surrogate */
+        (reqcu & 0xfc00) != 0xdc00)                 /* Req not low surrogate */
+#elif PCRE2_CODE_UNIT_WIDTH == 8
+    if ((re->overall_options & PCRE2_UTF) == 0 ||   /* Not UTF */
+        firstcuflags < 0 ||                         /* First not set */
+        (firstcu & 0x80) == 0 ||                    /* First is ASCII */
+        (reqcu & 0x80) == 0)                        /* Req is ASCII */
+#endif
+      {
+      minminlength++;
+      }
+
+    /* In the case of an anchored pattern, set up the value only if it follows
+    a variable length item in the pattern. */
+
+    if ((re->overall_options & PCRE2_ANCHORED) == 0 ||
+        (reqcuflags & REQ_VARY) != 0)
       {
       re->last_codeunit = reqcu;
       re->flags |= PCRE2_LASTSET;
@@ -10221,15 +10245,21 @@ if ((re->overall_options & PCRE2_NO_START_OPTIMIZE) == 0)
 #endif
         }
       }
+    }
 
-  /* Finally, study the compiled pattern to set up information such as a bitmap
+  /* Study the compiled pattern to set up information such as a bitmap of
-  of starting code units and a minimum matching length. */
+  starting code units and a minimum matching length. */
 
   if (PRIV(study)(re) != 0)
     {
     errorcode = ERR31;
     goto HAD_CB_ERROR;
     }
+
+  /* If the minimum length set (or not set) by study() is less than the minimum
+  implied by required code units, override it. */
+
+  if (re->minlength < minminlength) re->minlength = minminlength;
   }   /* End of start-of-match optimizations. */
 
 /* Control ends up here in all cases. When running under valgrind, make a

src/pcre2_study.c

@@ -214,9 +214,7 @@ for (;;)
 
     /* Reached end of a branch; if it's a ket it is the end of a nested
     call. If it's ALT it is an alternation in a nested call. If it is END it's
-    the end of the outer call. All can be handled by the same code. If an
+    the end of the outer call. All can be handled by the same code. */
-    ACCEPT was previously encountered, use the length that was in force at that
-    time, and pass back the shortest ACCEPT length. */
 
     case OP_ALT:
     case OP_KET:

src/pcre2test.c

@@ -4704,6 +4704,7 @@ if ((pat_patctl.control & CTL_INFO) != 0)
       }
     }
 
+  if ((FLD(compiled_code, overall_options) & PCRE2_NO_START_OPTIMIZE) == 0)
     fprintf(outfile, "Subject length lower bound = %d\n", minlength);
 
   if (pat_patctl.jit != 0 && (pat_patctl.control & CTL_JITVERIFY) != 0)

testdata/testoutput2

@@ -1480,7 +1480,7 @@ Subject length lower bound = 2
 Capture group count = 0
 First code unit = 'a'
 Last code unit = 'a'
-Subject length lower bound = 1
+Subject length lower bound = 2
 
 /(?=a)a.*/I
 Capture group count = 0
@@ -3406,7 +3406,7 @@ Subject length lower bound = 2
 Capture group count = 0
 May match empty string
 First code unit = 'a'
-Subject length lower bound = 0
+Subject length lower bound = 1
 
 /(?=abc).xyz/Ii
 Capture group count = 0
@@ -3425,7 +3425,7 @@ Subject length lower bound = 4
 Capture group count = 0
 May match empty string
 First code unit = 'a'
-Subject length lower bound = 0
+Subject length lower bound = 1
 
 /(?=.)a/I
 Capture group count = 0
@@ -3436,7 +3436,7 @@ Subject length lower bound = 1
 Capture group count = 1
 First code unit = 'a'
 Last code unit = 'a'
-Subject length lower bound = 1
+Subject length lower bound = 2
 
 /((?=abcda)ab)/I
 Capture group count = 1
@@ -10780,7 +10780,7 @@ Capture group count = 0
 Options: caseless
 First code unit = 'a' (caseless)
 Last code unit = 'a' (caseless)
-Subject length lower bound = 1
+Subject length lower bound = 2
 
 /(abc)\1+/
 
@@ -11254,7 +11254,7 @@ Subject length lower bound = 0
 /z(*ACCEPT)a/I,aftertext
 Capture group count = 0
 First code unit = 'z'
-Subject length lower bound = 0
+Subject length lower bound = 1
     baxzbx
  0: z
  0+ bx
@@ -13572,7 +13572,6 @@ Subject length lower bound = 4
 /abcd/I,no_start_optimize
 Capture group count = 0
 Options: no_start_optimize
-Subject length lower bound = 0
 
 /(|ab)*?d/I
 Capture group count = 1
@@ -13588,7 +13587,6 @@ Subject length lower bound = 1
 /(|ab)*?d/I,no_start_optimize
 Capture group count = 1
 Options: no_start_optimize
-Subject length lower bound = 0
    abd
  0: abd
  1: ab
@@ -14638,7 +14636,7 @@ Named capture groups:
 Options: dupnames
 Starting code units: a b 
 Last code unit = 'c'
-Subject length lower bound = 0
+Subject length lower bound = 1
 
 /ab{3cd/
     ab{3cd
@@ -16436,7 +16434,7 @@ Capture group count = 1
 Max back reference = 1
 First code unit = 'a'
 Last code unit = 'b'
-Subject length lower bound = 1
+Subject length lower bound = 2
     ab
  0: ab
  1: a

testdata/testoutput22-16

@@ -9,7 +9,7 @@ Contains \C
 Options: utf
 First code unit = 'a'
 Last code unit = 'e'
-Subject length lower bound = 0
+Subject length lower bound = 2
     abXde
  0: abXde
 

testdata/testoutput22-8

@@ -9,7 +9,7 @@ Contains \C
 Options: utf
 First code unit = 'a'
 Last code unit = 'e'
-Subject length lower bound = 0
+Subject length lower bound = 2
     abXde
  0: abXde
 

testdata/testoutput5

@@ -474,7 +474,6 @@ Subject length lower bound = 0
 Capture group count = 0
 Compile options: no_start_optimize utf
 Overall options: anchored no_start_optimize utf
-Subject length lower bound = 0
 
 /()()()()()()()()()()
  ()()()()()()()()()()

testdata/testoutput6

@@ -6845,7 +6845,6 @@ No match
 /(abc|def|xyz)/I,no_start_optimize
 Capture group count = 1
 Options: no_start_optimize
-Subject length lower bound = 0
     terhjk;abcdaadsfe
  0: abc
     the quick xyz brown fox