Improve support for partial matching in pcre2_substitute (#858)

NWilson · web-flow · commit a8ab3be45e89 · 2025-12-08T16:43:50.000Z
In 10.46 and previously, this was partially possible using PCRE2_SUBSTITUTE_MATCHED, but that capability was removed in 10.47.

This extends and restores a fixed version of the behaviour.
diff --git a/doc/html/pcre2api.html b/doc/html/pcre2api.html
@@ -3925,6 +3925,15 @@ <h2><a name="SEC38" href="#TOC1">CREATING A NEW STRING WITH SUBSTITUTIONS</a></h
 can be used to separate them if necessary.
 </p>
 <p>
+Partial matching is supported, with limitations: if matching succeeds but with a
+partial match, then pcre2_substitute returns PCRE2_ERROR_PARTIAL. When
+partial-matching (either of PCRE2_PARTIAL_HARD or PCRE2_PARTIAL_SOFT is passed),
+then PCRE2_SUBSTITUTE_REPLACEMENT_ONLY must also be set, or else
+PCRE2_ERROR_BADOPTION is returned. Similarly, certain replacement items
+($' and $_) cause PCRE2_ERROR_PARTIALSUBS to be returned when partial-matching,
+even if a complete match is found.
+</p>
+<p>
 The <i>outlengthptr</i> argument of <b>pcre2_substitute()</b> must point to a
 variable that contains the length, in code units, of the output buffer. If the
 function is successful, the value is updated to contain the length in code
diff --git a/doc/pcre2.txt b/doc/pcre2.txt
diff --git a/doc/pcre2api.3 b/doc/pcre2api.3
@@ -3928,6 +3928,14 @@ below)
 .\"
 can be used to separate them if necessary.
 .P
+Partial matching is supported, with limitations: if matching succeeds but with a
+partial match, then pcre2_substitute returns PCRE2_ERROR_PARTIAL. When
+partial-matching (either of PCRE2_PARTIAL_HARD or PCRE2_PARTIAL_SOFT is passed),
+then PCRE2_SUBSTITUTE_REPLACEMENT_ONLY must also be set, or else
+PCRE2_ERROR_BADOPTION is returned. Similarly, certain replacement items
+($' and $_) cause PCRE2_ERROR_PARTIALSUBS to be returned when partial-matching,
+even if a complete match is found.
+.P
 The \fIoutlengthptr\fP argument of \fBpcre2_substitute()\fP must point to a
 variable that contains the length, in code units, of the output buffer. If the
 function is successful, the value is updated to contain the length in code
diff --git a/src/pcre2.h.generic b/src/pcre2.h.generic
@@ -438,6 +438,7 @@ released, the numbers must not be changed. */
 #define PCRE2_ERROR_DIFFSUBSOFFSET    (-73)
 #define PCRE2_ERROR_DIFFSUBSOPTIONS   (-74)
 #define PCRE2_ERROR_BAD_BACKSLASH_K   (-75)
+#define PCRE2_ERROR_PARTIALSUBS       (-76)
 
 
 /* Request types for pcre2_pattern_info() */
diff --git a/src/pcre2.h.in b/src/pcre2.h.in
@@ -438,6 +438,7 @@ released, the numbers must not be changed. */
 #define PCRE2_ERROR_DIFFSUBSOFFSET    (-73)
 #define PCRE2_ERROR_DIFFSUBSOPTIONS   (-74)
 #define PCRE2_ERROR_BAD_BACKSLASH_K   (-75)
+#define PCRE2_ERROR_PARTIALSUBS       (-76)
 
 
 /* Request types for pcre2_pattern_info() */
diff --git a/src/pcre2_error.c b/src/pcre2_error.c
@@ -305,7 +305,9 @@ static const unsigned char match_error_texts[] =
   "substitute subject differs from prior match call\0"
   "substitute start offset differs from prior match call\0"
   "substitute options differ from prior match call\0"
+  /* 75 */
   "disallowed use of \\K in lookaround\0"
+  "replacement $' or $_ not supported with partial match\0"
   ;
 
 
diff --git a/src/pcre2_substitute.c b/src/pcre2_substitute.c
@@ -756,6 +756,7 @@ BOOL overflowed = FALSE;
 BOOL use_existing_match;
 BOOL replacement_only;
 BOOL utf = (code->overall_options & PCRE2_UTF) != 0;
+BOOL partial = (options & (PCRE2_PARTIAL_HARD|PCRE2_PARTIAL_SOFT)) != 0;
 PCRE2_UCHAR temp[6];
 PCRE2_UCHAR null_str[1] = { 0xcd };
 PCRE2_SPTR original_subject = subject;
@@ -783,10 +784,16 @@ if (mcontext != NULL)
   substitute_case_callout_data = mcontext->substitute_case_callout_data;
   }
 
-/* Partial matching is not valid. This must come after setting *blength to
-PCRE2_UNSET, so as not to imply an offset in the replacement. */
+/* Partial matching is supported, with limitations. We allow matching in partial
+mode, however, if a partial match is found, the substitution will fail with a
+PCRE2_ERROR_PARTIAL error. Additionally, outputting the after-match text is not
+allowed (PCRE2_ERROR_BADOPTION), and certain replacement items such as $' and $_
+are not supported (PCRE2_ERROR_PARTIALSUBS).
 
-if ((options & (PCRE2_PARTIAL_HARD|PCRE2_PARTIAL_SOFT)) != 0)
+This must come after setting *blength to PCRE2_UNSET, so as not to imply an
+offset in the replacement. */
+
+if (partial && (options & PCRE2_SUBSTITUTE_REPLACEMENT_ONLY) == 0)
   return PCRE2_ERROR_BADOPTION;
 
 /* Validate length and find the end of the replacement. A NULL replacement of
@@ -1125,8 +1132,16 @@ for (;;)
       if (next == CHAR_GRAVE_ACCENT || next == CHAR_APOSTROPHE)
         {
         ++ptr;
+
+        /* (Sanity-check ovector before reading from it.) */
         rc = pcre2_substring_length_bynumber(match_data, 0, &sublength);
-        if (rc < 0) goto PTREXIT; /* (Sanity-check ovector before reading from it.) */
+        /* LCOV_EXCL_START */
+        if (rc < 0)
+          {
+          PCRE2_DEBUG_UNREACHABLE();
+          goto PTREXIT;
+          }
+        /* LCOV_EXCL_STOP */
 
         if (next == CHAR_GRAVE_ACCENT)
           {
@@ -1135,6 +1150,12 @@ for (;;)
           }
         else
           {
+          if (partial)
+            {
+            rc = PCRE2_ERROR_PARTIALSUBS;
+            goto PTREXIT;
+            }
+
           subptr = subject + ovector[1];
           subptrend = subject + length;
           }
@@ -1145,12 +1166,19 @@ for (;;)
         {
         /* Java, .NET support $_ for "entire input string". */
         ++ptr;
+
+        if (partial)
+          {
+          rc = PCRE2_ERROR_PARTIALSUBS;
+          goto PTREXIT;
+          }
+
         subptr = subject;
         subptrend = subject + length;
         goto SUBPTR_SUBSTITUTE;
         }
-      else if (next == CHAR_PLUS &&
-               !(ptr+1 < repend && ptr[1] == CHAR_LEFT_CURLY_BRACKET))
+      if (next == CHAR_PLUS &&
+          !(ptr+1 < repend && ptr[1] == CHAR_LEFT_CURLY_BRACKET))
         {
         /* Perl supports $+ for "highest captured group" (not the same as $^N
         which is mainly only useful inside Perl's match callbacks). We also
diff --git a/testdata/testinput2 b/testdata/testinput2
@@ -8322,4 +8322,33 @@ a)"xI
     foo
     fooBAR
 
+# --------------
+# Tests for substitution with partial
+# --------------
+
+/(a)b+/
+\= Expect to fail with "bad option"
+    ab\=ph,replace=FOO
+\= Expect to fail with "partial match"
+    ab\=ph,substitute_replacement_only,replace=FOO
+\= Expect success
+    abc\=ph,substitute_replacement_only,replace=FOO
+    zabc\=ph,substitute_replacement_only,replace=>$&|$1|$`<
+\= Expect to fail with PCRE2_ERROR_PARTIALSUBS
+    abc\=ph,substitute_replacement_only,replace=>$_<
+    abc\=ph,substitute_replacement_only,replace=>$'<
+
+/(a)b+/
+\= Expect to fail with "bad option"
+    ab\=ps,replace=FOO
+\= Expect to fail with "partial match"
+    a\=ps,substitute_replacement_only,replace=FOO
+\= Expect success
+    ab\=ps,substitute_replacement_only,replace=FOO
+    abc\=ps,substitute_replacement_only,replace=FOO
+    zabc\=ps,substitute_replacement_only,replace=>$&|$1|$`<
+\= Expect to fail with PCRE2_ERROR_PARTIALSUBS
+    abc\=ps,substitute_replacement_only,replace=>$_<
+    abc\=ps,substitute_replacement_only,replace=>$'<
+
 # End of testinput2
diff --git a/testdata/testoutput2 b/testdata/testoutput2
@@ -23666,6 +23666,52 @@ Failed: error -72: substitute subject differs from prior match call
     fooBAR
  1: X::textY
 
+# --------------
+# Tests for substitution with partial
+# --------------
+
+/(a)b+/
+\= Expect to fail with "bad option"
+    ab\=ph,replace=FOO
+Failed: error -34: bad option value
+\= Expect to fail with "partial match"
+    ab\=ph,substitute_replacement_only,replace=FOO
+Failed: error -2: partial match
+\= Expect success
+    abc\=ph,substitute_replacement_only,replace=FOO
+ 1: FOO
+    zabc\=ph,substitute_replacement_only,replace=>$&|$1|$`<
+ 1: >ab|a|z<
+\= Expect to fail with PCRE2_ERROR_PARTIALSUBS
+    abc\=ph,substitute_replacement_only,replace=>$_<
+Failed: error -76 at offset 3 in replacement: replacement $' or $_ not supported with partial match
+        here: >$_ |<--| <
+    abc\=ph,substitute_replacement_only,replace=>$'<
+Failed: error -76 at offset 3 in replacement: replacement $' or $_ not supported with partial match
+        here: >$' |<--| <
+
+/(a)b+/
+\= Expect to fail with "bad option"
+    ab\=ps,replace=FOO
+Failed: error -34: bad option value
+\= Expect to fail with "partial match"
+    a\=ps,substitute_replacement_only,replace=FOO
+Failed: error -2: partial match
+\= Expect success
+    ab\=ps,substitute_replacement_only,replace=FOO
+ 1: FOO
+    abc\=ps,substitute_replacement_only,replace=FOO
+ 1: FOO
+    zabc\=ps,substitute_replacement_only,replace=>$&|$1|$`<
+ 1: >ab|a|z<
+\= Expect to fail with PCRE2_ERROR_PARTIALSUBS
+    abc\=ps,substitute_replacement_only,replace=>$_<
+Failed: error -76 at offset 3 in replacement: replacement $' or $_ not supported with partial match
+        here: >$_ |<--| <
+    abc\=ps,substitute_replacement_only,replace=>$'<
+Failed: error -76 at offset 3 in replacement: replacement $' or $_ not supported with partial match
+        here: >$' |<--| <
+
 # End of testinput2
 Error -80: PCRE2_ERROR_BADDATA (unknown error number)
 Error -62: bad serialized data

Original file line number	Diff line number	Diff line change
`@@ -305,7 +305,9 @@ static const unsigned char match_error_texts[] =`
`305`	`305`	`"substitute subject differs from prior match call\0"`
`306`	`306`	`"substitute start offset differs from prior match call\0"`
`307`	`307`	`"substitute options differ from prior match call\0"`
	`308`	`+ /* 75 */`
`308`	`309`	`"disallowed use of \\K in lookaround\0"`
	`310`	`+ "replacement $' or $_ not supported with partial match\0"`
`309`	`311`	`;`
`310`	`312`
`311`	`313`