Allow updating PCRE2_EXRA_ASCII_DIGIT by (?aP)

For consistency with PCRE2_EXTRA_ASCII_POSIX, allow the inline use of (?a) to affect the definition for [:sigit:] and [:xdigit:]
carenas · Oct 8, 2023 · 25e69a5 · 25e69a5
1 parent cecb429
commit 25e69a5
Show file tree

Hide file tree

Showing 5 changed files with 220 additions and 13 deletions.
diff --git a/src/pcre2_compile.c b/src/pcre2_compile.c
@@ -2741,7 +2741,7 @@ the main compiling phase. */
 
 #define PARSE_TRACKED_EXTRA_OPTIONS (PCRE2_EXTRA_CASELESS_RESTRICT| \
   PCRE2_EXTRA_ASCII_BSD|PCRE2_EXTRA_ASCII_BSS|PCRE2_EXTRA_ASCII_BSW| \
-  PCRE2_EXTRA_ASCII_POSIX)
+  PCRE2_EXTRA_ASCII_DIGIT|PCRE2_EXTRA_ASCII_POSIX)
 
 /* States used for analyzing ranges in character classes. The two OK values
 must be last. */
@@ -4233,7 +4233,7 @@ while (ptr < ptrend)
                 }
               if (*ptr == CHAR_P)
                 {
-                *xoptset |= PCRE2_EXTRA_ASCII_POSIX;
+                *xoptset |= (PCRE2_EXTRA_ASCII_DIGIT|PCRE2_EXTRA_ASCII_POSIX);
                 ptr++;
                 break;
                 }
@@ -4251,7 +4251,8 @@ while (ptr < ptrend)
                 }
               }
             *xoptset |= PCRE2_EXTRA_ASCII_BSD|PCRE2_EXTRA_ASCII_BSS|
-                        PCRE2_EXTRA_ASCII_BSW|PCRE2_EXTRA_ASCII_POSIX;
+                        PCRE2_EXTRA_ASCII_BSW|
+                        PCRE2_EXTRA_ASCII_DIGIT|PCRE2_EXTRA_ASCII_POSIX;
             break;
 
             case CHAR_J:  /* Record that it changed in the external options */

diff --git a/testdata/testinput5 b/testdata/testinput5
@@ -2343,11 +2343,19 @@
 \= Expect no match     
     \x{660}\x{660}\x{660}
 
+/\d(?-aD)\d(?aD)\d/utf,ucp,ascii_bsd
+    999
+    9\x{660}9
+
 /\d(?a)\d(?-a)\d/utf,ucp
     \x{660}9\x{660}
 \= Expect no match     
     \x{660}\x{660}\x{660}
 
+/\d(?-aD)\d(?aD)\d/utf,ucp,ascii_bsd
+    999
+    9\x{660}9
+
 # SPACES
 
 />\s+</i,utf
@@ -2435,18 +2443,51 @@
 
 # POSIX 
 
-/[[:digit:]]+/utf,ucp
+/^[[:digit:]]+$/utf,ucp
+    123456
     123\x{660}456
 
-/[[:digit:]]+/utf,ucp,ascii_digit
+/^[[:digit:]]+$/utf,ucp,ascii_digit
+    123456
+\= Expect no match
     123\x{660}456
 
 /[[:digit:]]+/g,utf,ucp,ascii_digit
     123\x{660}456
 
+/(?-aP)[[:digit:]](?aP)[[:digit:]]/utf,ucp,ascii_digit
+    11
+    \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+
+/(?-aP:[[:digit:]])[[:digit:]]/utf,ucp,ascii_digit
+    11
+    \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+
+/(?-aP:[[:digit:]])[[:digit:]]/utf,never_ucp,ascii_digit
+    11
+\= Expect no match
+    \x{ff11}1
+    1\x{ff11}
+
 /[[:digit:]]+/utf,ucp,ascii_posix
     123\x{660}456
 
+/(?-aP)[[:digit:]](?aP)[[:digit:]]/utf,ucp,ascii_posix
+    11
+    \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+
+/(?-aP:[[:digit:]])[[:digit:]]/utf,ucp,ascii_posix
+    11
+    \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+
 /^[[:xdigit:]]+$/utf,ucp
     f0
     1A

diff --git a/testdata/testinput7 b/testdata/testinput7
@@ -2355,11 +2355,19 @@
 \= Expect no match     
     \x{660}\x{660}\x{660}
 
+/\d(?-aD)\d(?aD)\d/utf,ucp,ascii_bsd
+    999
+    9\x{660}9
+
 /\d(?a)\d(?-a)\d/utf,ucp
     \x{660}9\x{660}
 \= Expect no match     
     \x{660}\x{660}\x{660}
 
+/\d(?-aD)\d(?aD)\d/utf,ucp,ascii_bsd
+    999
+    9\x{660}9
+
 # SPACES
 
 />\s+</i,utf
@@ -2432,18 +2440,51 @@
 
 # POSIX
 
-/[[:digit:]]+/utf,ucp
+/^[[:digit:]]+$/utf,ucp
+    123456
     123\x{660}456
 
-/[[:digit:]]+/utf,ucp,ascii_digit
+/^[[:digit:]]+$/utf,ucp,ascii_digit
+    123456
+\= Expect no match
     123\x{660}456
 
 /[[:digit:]]+/g,utf,ucp,ascii_digit
     123\x{660}456
 
+/(?-aP)[[:digit:]](?aP)[[:digit:]]/utf,ucp,ascii_digit
+    11
+    \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+
+/(?-aP:[[:digit:]])[[:digit:]]/utf,ucp,ascii_digit
+    11
+    \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+
+/(?-aP:[[:digit:]])[[:digit:]]/utf,never_ucp,ascii_digit
+    11
+\= Expect no match
+    \x{ff11}1
+    1\x{ff11}
+
 /[[:digit:]]+/utf,ucp,ascii_posix
     123\x{660}456
 
+/(?-aP)[[:digit:]](?aP)[[:digit:]]/utf,ucp,ascii_posix
+    11
+    \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+
+/(?-aP:[[:digit:]])[[:digit:]]/utf,ucp,ascii_posix
+    11
+    \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+
 />[[:space:]]+</utf,ucp
     >\x{a0} \x{a0}<
     >\x{a0}\x{a0}\x{a0}<

diff --git a/testdata/testoutput5 b/testdata/testoutput5
@@ -5251,13 +5251,25 @@ No match
     \x{660}\x{660}\x{660}
 No match
 
+/\d(?-aD)\d(?aD)\d/utf,ucp,ascii_bsd
+    999
+ 0: 999
+    9\x{660}9
+ 0: 9\x{660}9
+
 /\d(?a)\d(?-a)\d/utf,ucp
     \x{660}9\x{660}
  0: \x{660}9\x{660}
 \= Expect no match     
     \x{660}\x{660}\x{660}
 No match
 
+/\d(?-aD)\d(?aD)\d/utf,ucp,ascii_bsd
+    999
+ 0: 999
+    9\x{660}9
+ 0: 9\x{660}9
+
 # SPACES
 
 />\s+</i,utf
@@ -5375,23 +5387,73 @@ No match
 
 # POSIX 
 
-/[[:digit:]]+/utf,ucp
+/^[[:digit:]]+$/utf,ucp
+    123456
+ 0: 123456
     123\x{660}456
  0: 123\x{660}456
 
-/[[:digit:]]+/utf,ucp,ascii_digit
+/^[[:digit:]]+$/utf,ucp,ascii_digit
+    123456
+ 0: 123456
+\= Expect no match
     123\x{660}456
- 0: 123
+No match
 
 /[[:digit:]]+/g,utf,ucp,ascii_digit
     123\x{660}456
  0: 123
  0: 456
 
+/(?-aP)[[:digit:]](?aP)[[:digit:]]/utf,ucp,ascii_digit
+    11
+ 0: 11
+    \x{ff11}1
+ 0: \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+No match
+
+/(?-aP:[[:digit:]])[[:digit:]]/utf,ucp,ascii_digit
+    11
+ 0: 11
+    \x{ff11}1
+ 0: \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+No match
+
+/(?-aP:[[:digit:]])[[:digit:]]/utf,never_ucp,ascii_digit
+    11
+ 0: 11
+\= Expect no match
+    \x{ff11}1
+No match
+    1\x{ff11}
+No match
+
 /[[:digit:]]+/utf,ucp,ascii_posix
     123\x{660}456
  0: 123
 
+/(?-aP)[[:digit:]](?aP)[[:digit:]]/utf,ucp,ascii_posix
+    11
+ 0: 11
+    \x{ff11}1
+ 0: \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+No match
+
+/(?-aP:[[:digit:]])[[:digit:]]/utf,ucp,ascii_posix
+    11
+ 0: 11
+    \x{ff11}1
+ 0: \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+No match
+
 /^[[:xdigit:]]+$/utf,ucp
     f0
  0: f0

diff --git a/testdata/testoutput7 b/testdata/testoutput7
@@ -3971,13 +3971,25 @@ No match
     \x{660}\x{660}\x{660}
 No match
 
+/\d(?-aD)\d(?aD)\d/utf,ucp,ascii_bsd
+    999
+ 0: 999
+    9\x{660}9
+ 0: 9\x{660}9
+
 /\d(?a)\d(?-a)\d/utf,ucp
     \x{660}9\x{660}
  0: \x{660}9\x{660}
 \= Expect no match     
     \x{660}\x{660}\x{660}
 No match
 
+/\d(?-aD)\d(?aD)\d/utf,ucp,ascii_bsd
+    999
+ 0: 999
+    9\x{660}9
+ 0: 9\x{660}9
+
 # SPACES
 
 />\s+</i,utf
@@ -4076,23 +4088,73 @@ No match
 
 # POSIX
 
-/[[:digit:]]+/utf,ucp
+/^[[:digit:]]+$/utf,ucp
+    123456
+ 0: 123456
     123\x{660}456
  0: 123\x{660}456
 
-/[[:digit:]]+/utf,ucp,ascii_digit
+/^[[:digit:]]+$/utf,ucp,ascii_digit
+    123456
+ 0: 123456
+\= Expect no match
     123\x{660}456
- 0: 123
+No match
 
 /[[:digit:]]+/g,utf,ucp,ascii_digit
     123\x{660}456
  0: 123
  0: 456
 
+/(?-aP)[[:digit:]](?aP)[[:digit:]]/utf,ucp,ascii_digit
+    11
+ 0: 11
+    \x{ff11}1
+ 0: \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+No match
+
+/(?-aP:[[:digit:]])[[:digit:]]/utf,ucp,ascii_digit
+    11
+ 0: 11
+    \x{ff11}1
+ 0: \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+No match
+
+/(?-aP:[[:digit:]])[[:digit:]]/utf,never_ucp,ascii_digit
+    11
+ 0: 11
+\= Expect no match
+    \x{ff11}1
+No match
+    1\x{ff11}
+No match
+
 /[[:digit:]]+/utf,ucp,ascii_posix
     123\x{660}456
  0: 123
 
+/(?-aP)[[:digit:]](?aP)[[:digit:]]/utf,ucp,ascii_posix
+    11
+ 0: 11
+    \x{ff11}1
+ 0: \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+No match
+
+/(?-aP:[[:digit:]])[[:digit:]]/utf,ucp,ascii_posix
+    11
+ 0: 11
+    \x{ff11}1
+ 0: \x{ff11}1
+\= Expect no match
+    1\x{ff11}
+No match
+
 />[[:space:]]+</utf,ucp
     >\x{a0} \x{a0}<
  0: >\x{a0} \x{a0}<