regex: Update GRegexError with newer PCRE error codes

Update the GRegexError enum with new PCRE errors up to PCRE 8.30.

https://bugzilla.gnome.org/show_bug.cgi?id=677579

3 files changed, 181 insertions, 55 deletions

diff --git a/glib/gregex.c b/glib/gregex.c
index fd8a9f597..a1383a459 100644
--- a/glib/gregex.c
+++ b/glib/gregex.c
@@ -267,13 +267,6 @@ translate_compile_error (gint *errcode, const gchar **errmsg)
     case G_REGEX_ERROR_UNRECOGNIZED_ESCAPE:
       *errmsg = _("unrecognized character follows \\");
       break;
-    case 137:
-      /* A number of Perl escapes are not handled by PCRE.
-       * Therefore it explicitly raises ERR37.
-       */
-      *errcode = G_REGEX_ERROR_UNRECOGNIZED_ESCAPE;
-      *errmsg = _("case-changing escapes (\\l, \\L, \\u, \\U) are not allowed here");
-      break;
     case G_REGEX_ERROR_QUANTIFIERS_OUT_OF_ORDER:
       *errmsg = _("numbers out of order in {} quantifier");
       break;
@@ -292,16 +285,12 @@ translate_compile_error (gint *errcode, const gchar **errmsg)
     case G_REGEX_ERROR_NOTHING_TO_REPEAT:
       *errmsg = _("nothing to repeat");
       break;
-    case G_REGEX_ERROR_UNRECOGNIZED_CHARACTER:
-      *errmsg = _("unrecognized character after (?");
-      break;
-    case 124:
-      *errcode = G_REGEX_ERROR_UNRECOGNIZED_CHARACTER;
-      *errmsg = _("unrecognized character after (?<");
+    case 111: /* internal error: unexpected repeat */
+      *errcode = G_REGEX_ERROR_INTERNAL;
+      *errmsg = _("unexpected repeat");
       break;
-    case 141:
-      *errcode = G_REGEX_ERROR_UNRECOGNIZED_CHARACTER;
-      *errmsg = _("unrecognized character after (?P");
+    case G_REGEX_ERROR_UNRECOGNIZED_CHARACTER:
+      *errmsg = _("unrecognized character after (? or (?-");
       break;
     case G_REGEX_ERROR_POSIX_NAMED_CLASS_OUTSIDE_CLASS:
       *errmsg = _("POSIX named classes are supported only within a class");
@@ -309,17 +298,6 @@ translate_compile_error (gint *errcode, const gchar **errmsg)
     case G_REGEX_ERROR_UNMATCHED_PARENTHESIS:
       *errmsg = _("missing terminating )");
       break;
-    case 122:
-      *errcode = G_REGEX_ERROR_UNMATCHED_PARENTHESIS;
-      *errmsg = _(") without opening (");
-      break;
-    case 129:
-      *errcode = G_REGEX_ERROR_UNMATCHED_PARENTHESIS;
-      /* translators: '(?R' and '(?[+-]digits' are both meant as (groups of)
-       * sequences here, '(?-54' would be an example for the second group.
-       */
-      *errmsg = _("(?R or (?[+-]digits must be followed by )");
-      break;
     case G_REGEX_ERROR_INEXISTENT_SUBPATTERN_REFERENCE:
       *errmsg = _("reference to non-existent subpattern");
       break;
@@ -327,11 +305,23 @@ translate_compile_error (gint *errcode, const gchar **errmsg)
       *errmsg = _("missing ) after comment");
       break;
     case G_REGEX_ERROR_EXPRESSION_TOO_LARGE:
-      *errmsg = _("regular expression too large");
+      *errmsg = _("regular expression is too large");
       break;
     case G_REGEX_ERROR_MEMORY_ERROR:
       *errmsg = _("failed to get memory");
       break;
+    case 122: /* unmatched parentheses */
+      *errcode = G_REGEX_ERROR_UNMATCHED_PARENTHESIS;
+      *errmsg = _(") without opening (");
+      break;
+    case 123: /* internal error: code overflow */
+      *errcode = G_REGEX_ERROR_INTERNAL;
+      *errmsg = _("code overflow");
+      break;
+    case 124: /* "unrecognized character after (?<\0 */
+      *errcode = G_REGEX_ERROR_UNRECOGNIZED_CHARACTER;
+      *errmsg = _("unrecognized character after (?<");
+      break;
     case G_REGEX_ERROR_VARIABLE_LENGTH_LOOKBEHIND:
       *errmsg = _("lookbehind assertion is not fixed length");
       break;
@@ -344,6 +334,13 @@ translate_compile_error (gint *errcode, const gchar **errmsg)
     case G_REGEX_ERROR_ASSERTION_EXPECTED:
       *errmsg = _("assertion expected after (?(");
       break;
+    case 129:
+      *errcode = G_REGEX_ERROR_UNMATCHED_PARENTHESIS;
+      /* translators: '(?R' and '(?[+-]digits' are both meant as (groups of)
+       * sequences here, '(?-54' would be an example for the second group.
+       */
+      *errmsg = _("(?R or (?[+-]digits must be followed by )");
+      break;
     case G_REGEX_ERROR_UNKNOWN_POSIX_CLASS_NAME:
       *errmsg = _("unknown POSIX class name");
       break;
@@ -359,9 +356,20 @@ translate_compile_error (gint *errcode, const gchar **errmsg)
     case G_REGEX_ERROR_SINGLE_BYTE_MATCH_IN_LOOKBEHIND:
       *errmsg = _("\\C not allowed in lookbehind assertion");
       break;
+    case 137: /* PCRE does not support \\L, \\l, \\N{name}, \\U, or \\u\0 */
+      /* A number of Perl escapes are not handled by PCRE.
+       * Therefore it explicitly raises ERR37.
+       */
+      *errcode = G_REGEX_ERROR_UNRECOGNIZED_ESCAPE;
+      *errmsg = _("escapes \\L, \\l, \\N{name}, \\U, and \\u are not supported");
+      break;
     case G_REGEX_ERROR_INFINITE_LOOP:
       *errmsg = _("recursive call could loop indefinitely");
       break;
+    case 141: /* unrecognized character after (?P\0 */
+      *errcode = G_REGEX_ERROR_UNRECOGNIZED_CHARACTER;
+      *errmsg = _("unrecognized character after (?P");
+      break;
     case G_REGEX_ERROR_MISSING_SUBPATTERN_NAME_TERMINATOR:
       *errmsg = _("missing terminator in subpattern name");
       break;
@@ -383,6 +391,14 @@ translate_compile_error (gint *errcode, const gchar **errmsg)
     case G_REGEX_ERROR_INVALID_OCTAL_VALUE:
       *errmsg = _("octal value is greater than \\377");
       break;
+    case 152: /* internal error: overran compiling workspace */
+      *errcode = G_REGEX_ERROR_INTERNAL;
+      *errmsg = _("overran compiling workspace");
+      break;
+    case 153: /* internal error: previously-checked referenced subpattern not found */
+      *errcode = G_REGEX_ERROR_INTERNAL;
+      *errmsg = _("previously-checked referenced subpattern not found");
+      break;
     case G_REGEX_ERROR_TOO_MANY_BRANCHES_IN_DEFINE:
       *errmsg = _("DEFINE group contains more than one branch");
       break;
@@ -390,44 +406,77 @@ translate_compile_error (gint *errcode, const gchar **errmsg)
       *errmsg = _("inconsistent NEWLINE options");
       break;
     case G_REGEX_ERROR_MISSING_BACK_REFERENCE:
-      *errmsg = _("\\g is not followed by a braced name or an optionally "
-                 "braced non-zero number");
+      *errmsg = _("\\g is not followed by a braced, angle-bracketed, or quoted name or "
+                  "number, or by a plain number");
       break;
-    case 111:
-      *errcode = G_REGEX_ERROR_INTERNAL;
-      *errmsg = _("unexpected repeat");
+    case G_REGEX_ERROR_INVALID_RELATIVE_REFERENCE:
+      *errmsg = _("a numbered reference must not be zero");
       break;
-    case 123:
-      *errcode = G_REGEX_ERROR_INTERNAL;
-      *errmsg = _("code overflow");
+    case G_REGEX_ERROR_BACKTRACKING_CONTROL_VERB_ARGUMENT_FORBIDDEN:
+      *errmsg = _("an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)");
       break;
-    case 152:
-      *errcode = G_REGEX_ERROR_INTERNAL;
-      *errmsg = _("overran compiling workspace");
+    case G_REGEX_ERROR_UNKNOWN_BACKTRACKING_CONTROL_VERB:
+      *errmsg = _("(*VERB) not recognized");
       break;
-    case 153:
-      *errcode = G_REGEX_ERROR_INTERNAL;
-      *errmsg = _("previously-checked referenced subpattern not found");
+    case G_REGEX_ERROR_NUMBER_TOO_BIG:
+      *errmsg = _("number is too bug");
+      break;
+    case G_REGEX_ERROR_MISSING_SUBPATTERN_NAME:
+      *errmsg = _("missing subpattern name after (?&");
+      break;
+    case G_REGEX_ERROR_MISSING_DIGIT:
+      *errmsg = _("digit expected after (?+");
+      break;
+    case G_REGEX_ERROR_EXTRA_SUBPATTERN_NAME:
+      *errmsg = _("different names for subpatterns of the same number are not allowed");
+      break;
+    case G_REGEX_ERROR_BACKTRACKING_CONTROL_VERB_ARGUMENT_REQUIRED:
+      *errmsg = _("(*MARK) must have an argument");
       break;
-    case 116:
+    case G_REGEX_ERROR_INVALID_CONTROL_CHAR:
+      *errmsg = _( "\\c must be followed by an ASCII character");
+      break;
+    case G_REGEX_ERROR_MISSING_NAME:
+      *errmsg = _("\\k is not followed by a braced, angle-bracketed, or quoted name");
+      break;
+    case G_REGEX_ERROR_NOT_SUPPORTED_IN_CLASS:
+      *errmsg = _("\\N is not supported in a class");
+      break;
+    case G_REGEX_ERROR_TOO_MANY_FORWARD_REFERENCES:
+      *errmsg = _("too many forward references");
+      break;
+    case G_REGEX_ERROR_NAME_TOO_LONG:
+      *errmsg = _("name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)");
+      break;
+
+    case 116: /* erroffset passed as NULL */
       /* This should not happen as we never pass a NULL erroffset */
       g_warning ("erroffset passed as NULL");
       *errcode = G_REGEX_ERROR_COMPILE;
       break;
-    case 117:
+    case 117: /* unknown option bit(s) set */
       /* This should not happen as we check options before passing them
        * to pcre_compile2() */
       g_warning ("unknown option bit(s) set");
       *errcode = G_REGEX_ERROR_COMPILE;
       break;
-    case 132:
-    case 144:
-    case 145:
-      /* These errors should not happen as we are using an UTF8-enabled PCRE
+    case 132: /* this version of PCRE is compiled without UTF support */
+    case 144: /* invalid UTF-8 string */
+    case 145: /* support for \\P, \\p, and \\X has not been compiled */
+    case 167: /* this version of PCRE is not compiled with Unicode property support */
+    case 173: /* disallowed Unicode code point (>= 0xd800 && <= 0xdfff) */
+    case 174: /* invalid UTF-16 string */
+      /* These errors should not happen as we are using an UTF-8 and UCP-enabled PCRE
        * and we do not check if strings are valid */
+    case 164: /* ] is an invalid data character in JavaScript compatibility mode */
+      /* This should not happen as we don't use PCRE_JAVASCRIPT_COMPAT */
       g_warning ("%s", *errmsg);
       *errcode = G_REGEX_ERROR_COMPILE;
       break;
+    case 170: /* internal error: unknown opcode in find_fixedlength() */
+      *errcode = G_REGEX_ERROR_INTERNAL;
+      break;
+
     default:
       *errcode = G_REGEX_ERROR_COMPILE;
     }
diff --git a/glib/gregex.h b/glib/gregex.h
index ae6a5a138..4964c662e 100644
--- a/glib/gregex.h
+++ b/glib/gregex.h
@@ -107,9 +107,28 @@ G_BEGIN_DECLS
  *     This error is never raised. Since: 2.16 Deprecated: 2.34
  * @G_REGEX_ERROR_INCONSISTENT_NEWLINE_OPTIONS: Inconsistent newline options.
  *     Since 2.16
- * @G_REGEX_ERROR_MISSING_BACK_REFERENCE: "\\g" is not followed by a braced
- *     name or an optionally braced non-zero number. Since 2.16
- *
+ * @G_REGEX_ERROR_MISSING_BACK_REFERENCE: "\\g" is not followed by a braced,
+ *      angle-bracketed, or quoted name or number, or by a plain number. Since: 2.16
+ * @G_REGEX_ERROR_INVALID_RELATIVE_REFERENCE: relative reference must not be zero. Since: 2.34
+ * @G_REGEX_ERROR_BACKTRACKING_CONTROL_VERB_ARGUMENT_FORBIDDEN: the backtracing
+ *     control verb used does not allow an argument. Since: 2.34
+ * @G_REGEX_ERROR_UNKNOWN_BACKTRACKING_CONTROL_VERB: unknown backtracing 
+ *     control verb. Since: 2.34
+ * @G_REGEX_ERROR_NUMBER_TOO_BIG: number is too big in escape sequence. Since: 2.34
+ * @G_REGEX_ERROR_MISSING_SUBPATTERN_NAME: Missing subpattern name. Since: 2.34
+ * @G_REGEX_ERROR_EXTRA_SUBPATTERN_NAME: different names for subpatterns of the 
+ *     same number are not allowed. Since: 2.34
+ * @G_REGEX_ERROR_BACKTRACKING_CONTROL_VERB_ARGUMENT_REQUIRED: the backtracing control
+ *     verb requires an argument. Since: 2.34
+ * @G_REGEX_ERROR_INVALID_CONTROL_CHAR: "\\c" must be followed by an ASCII 
+ *     character. Since: 2.34
+ * @G_REGEX_ERROR_MISSING_NAME: "\\k" is not followed by a braced, angle-bracketed, or 
+ *     quoted name. Since: 2.34
+ * @G_REGEX_ERROR_NOT_SUPPORTED_IN_CLASS: "\\N" is not supported in a class. Since: 2.34
+ * @G_REGEX_ERROR_TOO_MANY_FORWARD_REFERENCES: too many forward references. Since: 2.34
+ * @G_REGEX_ERROR_NAME_TOO_LONG: the name is too long in "(*MARK)", "(*PRUNE)", 
+ *     "(*SKIP)", or "(*THEN)". Since: 2.34
+ * 
  * Error codes returned by regular expressions functions.
  *
  * Since: 2.14
@@ -159,7 +178,20 @@ typedef enum
   G_REGEX_ERROR_TOO_MANY_BRANCHES_IN_DEFINE = 154,
   G_REGEX_ERROR_DEFINE_REPETION = 155,
   G_REGEX_ERROR_INCONSISTENT_NEWLINE_OPTIONS = 156,
-  G_REGEX_ERROR_MISSING_BACK_REFERENCE = 157
+  G_REGEX_ERROR_MISSING_BACK_REFERENCE = 157,
+  G_REGEX_ERROR_INVALID_RELATIVE_REFERENCE = 158,
+  G_REGEX_ERROR_BACKTRACKING_CONTROL_VERB_ARGUMENT_FORBIDDEN = 159,
+  G_REGEX_ERROR_UNKNOWN_BACKTRACKING_CONTROL_VERB  = 160,
+  G_REGEX_ERROR_NUMBER_TOO_BIG = 161,
+  G_REGEX_ERROR_MISSING_SUBPATTERN_NAME = 162,
+  G_REGEX_ERROR_MISSING_DIGIT = 163,
+  G_REGEX_ERROR_EXTRA_SUBPATTERN_NAME = 165,
+  G_REGEX_ERROR_BACKTRACKING_CONTROL_VERB_ARGUMENT_REQUIRED = 166,
+  G_REGEX_ERROR_INVALID_CONTROL_CHAR = 168,
+  G_REGEX_ERROR_MISSING_NAME = 169,
+  G_REGEX_ERROR_NOT_SUPPORTED_IN_CLASS = 171,
+  G_REGEX_ERROR_TOO_MANY_FORWARD_REFERENCES = 172,
+  G_REGEX_ERROR_NAME_TOO_LONG = 175
 } GRegexError;
 
 /**
diff --git a/glib/tests/regex.c b/glib/tests/regex.c
index af34fc45d..6deb50f40 100644
--- a/glib/tests/regex.c
+++ b/glib/tests/regex.c
@@ -2072,9 +2072,7 @@ main (int argc, char *argv[])
   TEST_NEW_FAIL ("a{4,2}", 0, G_REGEX_ERROR_QUANTIFIERS_OUT_OF_ORDER);
   TEST_NEW_FAIL ("a{999999,}", 0, G_REGEX_ERROR_QUANTIFIER_TOO_BIG);
   TEST_NEW_FAIL ("[a-z", 0, G_REGEX_ERROR_UNTERMINATED_CHARACTER_CLASS);
-#if 0
-  TEST_NEW_FAIL ("[\\b]", 0, G_REGEX_ERROR_INVALID_ESCAPE_IN_CHARACTER_CLASS);
-#endif
+  TEST_NEW_FAIL ("(?X)[\\B]", 0, G_REGEX_ERROR_INVALID_ESCAPE_IN_CHARACTER_CLASS);
   TEST_NEW_FAIL ("[z-a]", 0, G_REGEX_ERROR_RANGE_OUT_OF_ORDER);
   TEST_NEW_FAIL ("{2,4}", 0, G_REGEX_ERROR_NOTHING_TO_REPEAT);
   TEST_NEW_FAIL ("a(?u)", 0, G_REGEX_ERROR_UNRECOGNIZED_CHARACTER);
@@ -2084,8 +2082,55 @@ main (int argc, char *argv[])
   TEST_NEW_FAIL ("a)b", 0, G_REGEX_ERROR_UNMATCHED_PARENTHESIS);
   TEST_NEW_FAIL ("a(?R", 0, G_REGEX_ERROR_UNMATCHED_PARENTHESIS);
   TEST_NEW_FAIL ("a(?-54", 0, G_REGEX_ERROR_UNMATCHED_PARENTHESIS);
+  TEST_NEW_FAIL ("(ab\\2)", 0, G_REGEX_ERROR_INEXISTENT_SUBPATTERN_REFERENCE);
   TEST_NEW_FAIL ("a(?#abc", 0, G_REGEX_ERROR_UNTERMINATED_COMMENT);
+  TEST_NEW_FAIL ("(?<=a+)b", 0, G_REGEX_ERROR_VARIABLE_LENGTH_LOOKBEHIND);
+  TEST_NEW_FAIL ("(?(1?)a|b)", 0, G_REGEX_ERROR_MALFORMED_CONDITION);
+  TEST_NEW_FAIL ("(a)(?(1)a|b|c)", 0, G_REGEX_ERROR_TOO_MANY_CONDITIONAL_BRANCHES);
+  TEST_NEW_FAIL ("(?(?i))", 0, G_REGEX_ERROR_ASSERTION_EXPECTED);
   TEST_NEW_FAIL ("a[[:fubar:]]b", 0, G_REGEX_ERROR_UNKNOWN_POSIX_CLASS_NAME);
+  TEST_NEW_FAIL ("[[.ch.]]", 0, G_REGEX_ERROR_POSIX_COLLATING_ELEMENTS_NOT_SUPPORTED);
+  TEST_NEW_FAIL ("\\x{110000}", 0, G_REGEX_ERROR_HEX_CODE_TOO_LARGE);
+  TEST_NEW_FAIL ("^(?(0)f|b)oo", 0, G_REGEX_ERROR_INVALID_CONDITION);
+  TEST_NEW_FAIL ("(?<=\\C)X", 0, G_REGEX_ERROR_SINGLE_BYTE_MATCH_IN_LOOKBEHIND);
+  TEST_NEW_FAIL ("(?!\\w)(?R)", 0, G_REGEX_ERROR_INFINITE_LOOP);
+  TEST_NEW_FAIL ("(?(?<ab))", 0, G_REGEX_ERROR_MISSING_SUBPATTERN_NAME_TERMINATOR);
+  TEST_NEW_FAIL ("(?P<x>eks)(?P<x>eccs)", 0, G_REGEX_ERROR_DUPLICATE_SUBPATTERN_NAME);
+#if 0
+  TEST_NEW_FAIL (?, 0, G_REGEX_ERROR_MALFORMED_PROPERTY);
+  TEST_NEW_FAIL (?, 0, G_REGEX_ERROR_UNKNOWN_PROPERTY);
+#endif
+  TEST_NEW_FAIL ("\\666", G_REGEX_RAW, G_REGEX_ERROR_INVALID_OCTAL_VALUE);
+  TEST_NEW_FAIL ("^(?(DEFINE) abc | xyz ) ", 0, G_REGEX_ERROR_TOO_MANY_BRANCHES_IN_DEFINE);
+#if 0
+  TEST_NEW_FAIL ("a", G_REGEX_NEWLINE_CRLF | G_REGEX_NEWLINE_ANYCRLF, G_REGEX_ERROR_INCONSISTENT_NEWLINE_OPTIONS);
+#endif
+  TEST_NEW_FAIL ("^(a)\\g{3", 0, G_REGEX_ERROR_MISSING_BACK_REFERENCE);
+  TEST_NEW_FAIL ("^(a)\\g{0}", 0, G_REGEX_ERROR_INVALID_RELATIVE_REFERENCE);
+  TEST_NEW_FAIL ("abc(*FAIL:123)xyz", 0, G_REGEX_ERROR_BACKTRACKING_CONTROL_VERB_ARGUMENT_FORBIDDEN);
+  TEST_NEW_FAIL ("a(*FOOBAR)b", 0, G_REGEX_ERROR_UNKNOWN_BACKTRACKING_CONTROL_VERB);
+  TEST_NEW_FAIL ("(?i:A{1,}\\6666666666)", 0, G_REGEX_ERROR_NUMBER_TOO_BIG);
+  TEST_NEW_FAIL ("(?<a>)(?&)", 0, G_REGEX_ERROR_MISSING_SUBPATTERN_NAME);
+  TEST_NEW_FAIL ("(?+-a)", 0, G_REGEX_ERROR_MISSING_DIGIT);
+  TEST_NEW_FAIL ("(?|(?<a>A)|(?<b>B))", 0, G_REGEX_ERROR_EXTRA_SUBPATTERN_NAME);
+  TEST_NEW_FAIL ("a(*MARK)b", 0, G_REGEX_ERROR_BACKTRACKING_CONTROL_VERB_ARGUMENT_REQUIRED);
+  TEST_NEW_FAIL ("^\\c€", 0, G_REGEX_ERROR_INVALID_CONTROL_CHAR);
+  TEST_NEW_FAIL ("\\k", 0, G_REGEX_ERROR_MISSING_NAME);
+  TEST_NEW_FAIL ("a[\\NB]c", 0, G_REGEX_ERROR_NOT_SUPPORTED_IN_CLASS);
+#if 0
+  /* This one should fail but doesn't. Why? */
+  TEST_NEW_FAIL ("(*:0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEFG)XX", 0, G_REGEX_ERROR_NAME_TOO_LONG);
+#endif
+  /* These errors can't really be tested sanely:
+   * G_REGEX_ERROR_EXPRESSION_TOO_LARGE
+   * G_REGEX_ERROR_MEMORY_ERROR
+   * G_REGEX_ERROR_SUBPATTERN_NAME_TOO_LONG
+   * G_REGEX_ERROR_TOO_MANY_SUBPATTERNS
+   * G_REGEX_ERROR_TOO_MANY_FORWARD_REFERENCES
+   *
+   * These errors are obsolete and never raised by PCRE:
+   * G_REGEX_ERROR_DEFINE_REPETION
+   */
 
   /* TEST_MATCH_SIMPLE(pattern, string, compile_opts, match_opts, expected) */
   TEST_MATCH_SIMPLE("a", "", 0, 0, FALSE);