Better handling of character classes within the PCRE parser

aelliott · Mar 28, 2012 · d2b2674 · d2b2674
1 parent 0d3df24
commit d2b2674
Show file tree

Hide file tree

Showing 4 changed files with 69 additions and 3 deletions.
diff --git a/GraphicsItems/bracketexpressiongraphicsitem.cpp b/GraphicsItems/bracketexpressiongraphicsitem.cpp
@@ -77,6 +77,24 @@ void BracketExpressionGraphicsItem::readTokens()
         case T_NOT_WORD_BOUNDARY:
             items.push_back(QString("Not Word Boundary"));
             break;
+        case T_BELL:
+            items.push_back(QString("Bell Character"));
+            break;
+        case T_ESCAPE:
+            items.push_back(QString("Escape Character"));
+            break;
+        case T_FORM_FEED:
+            items.push_back(QString("Form Feed"));
+            break;
+        case T_LINE_FEED:
+            items.push_back(QString("Line Feed"));
+            break;
+        case T_HORIZONTAL_TAB:
+            items.push_back(QString("Tab Character"));
+            break;
+        case T_OCTAL_CHAR:
+            items.push_back(QString("Octal Character"));
+            break;
         default:
             qDebug() << "Unhandled token in bracket expression: " << token->value();
         }

diff --git a/RegexModules/icuparser.cpp b/RegexModules/icuparser.cpp
@@ -198,7 +198,7 @@ void IcuParser::handleToken(RegexpToken token)
 {
     QRegExp rx;
     QRegExp characterClass;
-    QRegExp bracketExpressionLiteral("[^\\\\[\\]()|?*+{}\\.$]|\\\\[^a-zA-Z0-9]");
+    QRegExp bracketExpressionLiteral("[^\\\\[\\]\.]|\\\\[^a-zA-Z0-9]");
     QRegExp bracketExpressionRange(_syntax[T_BRACKET_EXPRESSION_RANGE]);
     QRegExp flagClose;
     Token *openingToken;

diff --git a/RegexModules/pcreparser.cpp b/RegexModules/pcreparser.cpp
@@ -194,7 +194,7 @@ void PcreParser::handleToken(RegexpToken token)
 {
     QRegExp rx;
     QRegExp characterClass;
-    QRegExp bracketExpressionLiteral("[^\\\\[\\]()|?*+{}\\.$]|\\\\[^a-zA-Z0-9]");
+    QRegExp bracketExpressionLiteral("[^\\\\[\\]\.]|\\\\[^a-zA-Z0-9]");
     QRegExp bracketExpressionRange(_syntax[T_BRACKET_EXPRESSION_RANGE]);
     QRegExp flagClose;
     Token *openingToken;
@@ -363,6 +363,54 @@ void PcreParser::handleToken(RegexpToken token)
                 _pos += characterClass.matchedLength();
             }
 
+            characterClass.setPattern(_syntax[T_BELL]);
+            if(matched == false && characterClass.indexIn(_expression, _pos) == _pos)
+            {
+                matched = true;
+                _tokens.push_back(new Token(T_BELL, _expression.mid(_pos, characterClass.matchedLength())));
+                _pos += characterClass.matchedLength();
+            }
+
+            characterClass.setPattern(_syntax[T_ESCAPE]);
+            if(matched == false && characterClass.indexIn(_expression, _pos) == _pos)
+            {
+                matched = true;
+                _tokens.push_back(new Token(T_ESCAPE, _expression.mid(_pos, characterClass.matchedLength())));
+                _pos += characterClass.matchedLength();
+            }
+
+            characterClass.setPattern(_syntax[T_FORM_FEED]);
+            if(matched == false && characterClass.indexIn(_expression, _pos) == _pos)
+            {
+                matched = true;
+                _tokens.push_back(new Token(T_FORM_FEED, _expression.mid(_pos, characterClass.matchedLength())));
+                _pos += characterClass.matchedLength();
+            }
+
+            characterClass.setPattern(_syntax[T_LINE_FEED]);
+            if(matched == false && characterClass.indexIn(_expression, _pos) == _pos)
+            {
+                matched = true;
+                _tokens.push_back(new Token(T_LINE_FEED, _expression.mid(_pos, characterClass.matchedLength())));
+                _pos += characterClass.matchedLength();
+            }
+
+            characterClass.setPattern(_syntax[T_HORIZONTAL_TAB]);
+            if(matched == false && characterClass.indexIn(_expression, _pos) == _pos)
+            {
+                matched = true;
+                _tokens.push_back(new Token(T_HORIZONTAL_TAB, _expression.mid(_pos, characterClass.matchedLength())));
+                _pos += characterClass.matchedLength();
+            }
+
+            characterClass.setPattern(_syntax[T_OCTAL_CHAR]);
+            if(matched == false && characterClass.indexIn(_expression, _pos) == _pos)
+            {
+                matched = true;
+                _tokens.push_back(new Token(T_OCTAL_CHAR, _expression.mid(_pos, characterClass.matchedLength())));
+                _pos += characterClass.matchedLength();
+            }
+
             if(!matched)
             {
                 _tokens.push_back(new Token(T_ERROR, _expression.mid(_pos, 1)));

diff --git a/RegexModules/qtparser.cpp b/RegexModules/qtparser.cpp
@@ -91,7 +91,7 @@ void QtParser::handleToken(RegexpToken token)
 {
     QRegExp groupingClose(_syntax[T_GROUPING_CLOSE]);
     QRegExp assertionClose(_syntax[T_ASSERTION_CLOSE]);
-    QRegExp bracketExpressionLiteral("[^\\\\[\\]()|?*+{}\\.^$]|\\\\[^a-zA-Z0-9]");
+    QRegExp bracketExpressionLiteral("[^\\\\[\\]\.]|\\\\[^a-zA-Z0-9]");
     QRegExp bracketExpressionRange(_syntax[T_BRACKET_EXPRESSION_RANGE]);
     QRegExp bracketExpressionClose(_syntax[T_BRACKET_EXPRESSION_CLOSE]);
     QRegExp characterClass("");