Support \p 1-char char classes without curly braces

jruby · Mar 19, 2015 · 804f09c · 804f09c
1 parent 005b574
commit 804f09c
Show file tree

Hide file tree

Showing 4 changed files with 30 additions and 7 deletions.
diff --git a/src/org/joni/Lexer.java b/src/org/joni/Lexer.java
@@ -956,6 +956,10 @@ private void fetchTokenFor_charProperty() {
                     unfetch();
                 }
             }
+        } else if (syntax.op2EscPCharCharProperty()) {
+            token.type = TokenType.CHAR_PROPERTY;
+            token.setPropNot(c == 'P');
+            token.setPropSingleChar(true);
         } else {
             syntaxWarn(Warnings.INVALID_UNICODE_PROPERTY, (char)c);
         }
@@ -1252,13 +1256,18 @@ private void possessiveCheck() {
     protected final int fetchCharPropertyToCType() {
         mark();
 
-        while (left()) {
-            int last = p;
+        if (token.getPropSingleChar()) {
             fetch();
-            if (c == '}') {
-                return enc.propertyNameToCType(bytes, _p, last);
-            } else if (c == '(' || c == ')' || c == '{' || c == '|') {
-                throw new CharacterPropertyException(ERR_INVALID_CHAR_PROPERTY_NAME, bytes, _p, last);
+            return enc.propertyNameToCType(bytes, _p, p);
+        } else {
+            while (left()) {
+                int last = p;
+                fetch();
+                if (c == '}') {
+                    return enc.propertyNameToCType(bytes, _p, last);
+                } else if (c == '(' || c == ')' || c == '{' || c == '|') {
+                    throw new CharacterPropertyException(ERR_INVALID_CHAR_PROPERTY_NAME, bytes, _p, last);
+                }
             }
         }
         newInternalException(ERR_PARSER_BUG);

diff --git a/src/org/joni/Syntax.java b/src/org/joni/Syntax.java
@@ -282,6 +282,11 @@ public boolean op2OptionECMAScript() {
         return isOp2(OP2_OPTION_ECMASCRIPT);
     }
 
+    public boolean op2EscPCharCharProperty()
+    {
+        return isOp2(OP2_ESC_P_BRACE_CHAR_PROPERTY);
+    }
+
     /**
      * BEHAVIOR
      *
@@ -535,7 +540,8 @@ public boolean warnReduntantNestedRepeat() {
         OP2_OPTION_PERL | OP2_PLUS_POSSESSIVE_REPEAT |
         OP2_PLUS_POSSESSIVE_INTERVAL | OP2_CCLASS_SET_OP |
         OP2_ESC_V_VTAB | OP2_ESC_U_HEX4 |
-        OP2_ESC_P_BRACE_CHAR_PROPERTY ),
+        OP2_ESC_P_BRACE_CHAR_PROPERTY |
+        OP2_ESC_P_CHAR_CHAR_PROPERTY),
 
         ( GNU_REGEX_BV | DIFFERENT_LEN_ALT_LOOK_BEHIND ),
 

diff --git a/src/org/joni/Token.java b/src/org/joni/Token.java
@@ -169,4 +169,11 @@ boolean getPropNot() {
     void setPropNot(boolean not) {
         INT2 = not ? 1 : 0;
     }
+
+    boolean getPropSingleChar() {
+        return INT3 != 0;
+    }
+    void setPropSingleChar(boolean singleChar) {
+        INT3 = singleChar ? 1 : 0;
+    }
 }
diff --git a/src/org/joni/constants/SyntaxProperties.java b/src/org/joni/constants/SyntaxProperties.java
@@ -75,6 +75,7 @@ public interface SyntaxProperties {
     final int OP2_ESC_H_XDIGIT               = (1<<19); /* \h, \H */
     final int OP2_INEFFECTIVE_ESCAPE         = (1<<20); /* \ */
     final int OP2_OPTION_ECMASCRIPT          = (1<<21); /* EcmaScript quirks */
+    final int OP2_ESC_P_CHAR_CHAR_PROPERTY   = (1<<22); /* \pX, \PX */
 
     /* syntax (behavior); */
     final int CONTEXT_INDEP_ANCHORS           = (1<<31); /* not implemented */