some general filter parser cleanup

apps/opencs/model/filter/parser.cpp

@@ -1,7 +1,11 @@
 
 #include "parser.hpp"
 
+#include <cctype>
 #include <stdexcept>
+#include <sstream>
+
+#include <components/misc/stringops.hpp>
 
 #include "booleannode.hpp"
 
@@ -12,65 +16,239 @@ namespace CSMFilter
         enum Type
         {
             Type_EOS,
-            Type_None
+            Type_None,
+            Type_String,
+            Type_Number,
+            Type_Open,
+            Type_Close,
+            Type_OpenSquare,
+            Type_CloseSquare,
+            Type_Comma,
+            Type_Keyword_True, ///< \attention Keyword enums must be arranged continuously.
+            Type_Keyword_False,
+            Type_Keyword_And,
+            Type_Keyword_Or,
+            Type_Keyword_Not
         };
 
         Type mType;
+        std::string mString;
+        double mNumber;
 
         Token (Type type);
+
+        Token (const std::string& string);
+
+        Token (double number);
+
+        operator bool() const;
     };
 
     Token::Token (Type type) : mType (type) {}
+
+    Token::Token (const std::string& string) : mType (Type_String), mString (string) {}
+
+    Token::Token (double number) : mType (Type_Number), mNumber (number) {}
+
+    Token::operator bool() const
+    {
+        return mType!=Type_None;
+    }
+
+    bool operator== (const Token& left, const Token& right)
+    {
+        if (left.mType!=right.mType)
+            return false;
+
+        switch (left.mType)
+        {
+            case Token::Type_String: return left.mString==right.mString;
+            case Token::Type_Number: return left.mNumber==right.mNumber;
+
+            default: return true;
+        }
+    }
 }
 
-CSMFilter::Token CSMFilter::Parser::getNextToken (const std::string& filter, int& index) const
+CSMFilter::Token CSMFilter::Parser::getStringToken()
 {
-    if (index>=static_cast<int> (filter.size()))
+    std::string string;
+
+    int size = static_cast<int> (mInput.size());
+
+    for (; mIndex<size; ++mIndex)
+    {
+        char c = mInput[mIndex];
+
+        if (std::isalpha (c) || c=='_' || (!string.empty() && std::isdigit (c)) || c=='"' ||
+            (!string.empty() && string[0]=='"'))
+            string += c;
+        else
+            break;
+
+        if (c=='"' && string.size()>1)
+        {
+            ++mIndex;
+            break;
+        }
+    };
+
+    if (!string.empty())
+    {
+        if (string[0]=='"' && (string[string.size()-1]!='"' || string.size()<2) )
+        {
+            error();
+            return Token (Token::Type_None);
+        }
+
+        if (string[0]!='"' && string[string.size()-1]=='"')
+        {
+            error();
+            return Token (Token::Type_None);
+        }
+    }
+
+    return checkKeywords (string);
+}
+
+CSMFilter::Token CSMFilter::Parser::getNumberToken()
+{
+    std::string string;
+
+    int size = static_cast<int> (mInput.size());
+
+    bool hasDecimalPoint = false;
+    bool hasDigit = false;
+
+    for (; mIndex<size; ++mIndex)
+    {
+        char c = mInput[mIndex];
+
+        if (std::isdigit (c))
+        {
+            string += c;
+            hasDigit = true;
+        }
+        else if (c=='.' && !hasDecimalPoint)
+        {
+            string += c;
+            hasDecimalPoint = true;
+        }
+        else if (string.empty() && c=='-')
+            string += c;
+        else
+            break;
+    }
+
+    if (!hasDigit)
+    {
+        error();
+        return Token (Token::Type_None);
+    }
+
+    float value;
+    std::istringstream stream (string.c_str());
+    stream >> value;
+
+    return value;
+}
+
+CSMFilter::Token CSMFilter::Parser::checkKeywords (const Token& token)
+{
+    static const char *sKeywords[] =
+    {
+        "true", "false",
+        "and", "or", "not",
+        0
+    };
+
+    std::string string = Misc::StringUtils::lowerCase (token.mString);
+
+    for (int i=0; sKeywords[i]; ++i)
+        if (sKeywords[i]==string)
+            return Token (static_cast<Token::Type> (i+Token::Type_Keyword_True));
+
+    return token;
+}
+
+CSMFilter::Token CSMFilter::Parser::getNextToken()
+{
+    int size = static_cast<int> (mInput.size());
+
+    char c = 0;
+
+    for (; mIndex<size; ++mIndex)
+    {
+        c = mInput[mIndex];
+
+        if (c!=' ')
+            break;
+    }
+
+    if (mIndex>=size)
         return Token (Token::Type_EOS);
 
+    switch (c)
+    {
+        case '(': ++mIndex; return Token (Token::Type_Open);
+        case ')': ++mIndex; return Token (Token::Type_Close);
+        case '[': ++mIndex; return Token (Token::Type_OpenSquare);
+        case ']': ++mIndex; return Token (Token::Type_CloseSquare);
+        case ',': ++mIndex; return Token (Token::Type_Comma);
+    }
+
+    if (c=='"' || c=='_' || std::isalpha (c))
+        return getStringToken();
+
+    if (c=='-' || c=='.' || std::isdigit (c))
+        return getNumberToken();
+
+    error();
     return Token (Token::Type_None);
 }
 
-bool CSMFilter::Parser::isEndState() const
+boost::shared_ptr<CSMFilter::Node> CSMFilter::Parser::parseImp()
 {
-    return mState==State_End || mState==State_UnexpectedCharacter;
+
+
+
+    return boost::shared_ptr<Node>();
 }
 
-CSMFilter::Parser::Parser() : mState (State_Begin) {}
+void CSMFilter::Parser::error()
+{
+    mError = true;
+}
 
-void CSMFilter::Parser::parse (const std::string& filter)
+CSMFilter::Parser::Parser() : mIndex (0), mError (false) {}
+
+bool CSMFilter::Parser::parse (const std::string& filter)
 {
     // reset
-    mState = State_Begin;
     mFilter.reset();
-    int index = 0;
+    mError = false;
+    mInput = filter;
+    mIndex = 0;
 
-    while (!isEndState())
-    {
-        Token token = getNextToken (filter, index);
+    boost::shared_ptr<Node> node = parseImp();
 
-        switch (token.mType)
-        {
-            case Token::Type_None: mState = State_UnexpectedCharacter; break;
-            case Token::Type_EOS: mState = State_End; break;
-        }
-    }
+    if (mError)
+        return false;
 
-    if (mState==State_End && !mFilter)
+    if (node)
+        mFilter = node;
+    else
     {
         // Empty filter string equals to filter "true".
         mFilter.reset (new BooleanNode (true));
     }
-}
 
-CSMFilter::Parser::State CSMFilter::Parser::getState() const
-{
-    return mState;
+    return true;
 }
 
 boost::shared_ptr<CSMFilter::Node> CSMFilter::Parser::getFilter() const
 {
-    if (mState!=State_End)
+    if (mError)
         throw std::logic_error ("No filter available");
 
     return mFilter;

apps/opencs/model/filter/parser.hpp

@@ -11,36 +11,36 @@ namespace CSMFilter
 
     class Parser
     {
-        public:
+            boost::shared_ptr<Node> mFilter;
+            std::string mInput;
+            int mIndex;
+            bool mError;
 
-            enum State
-            {
-                State_Begin,
-                State_UnexpectedCharacter,
-                State_End
-            };
+            Token getStringToken();
 
-        private:
+            Token getNumberToken();
 
-            State mState;
-            boost::shared_ptr<Node> mFilter;
+            Token getNextToken();
 
-            Token getNextToken (const std::string& filter, int& index) const;
+            Token checkKeywords (const Token& token);
+            ///< Turn string token into keyword token, if possible.
 
-            bool isEndState() const;
-            ///< This includes error states.
+            boost::shared_ptr<Node> parseImp();
+            ///< Will return a null-pointer, if there is nothing more to parse.
+
+            void error();
 
         public:
 
             Parser();
 
-            void parse (const std::string& filter);
+            bool parse (const std::string& filter);
             ///< Discards any previous calls to parse
-
-            State getState() const;
+            ///
+            /// \return Success?
 
             boost::shared_ptr<Node> getFilter() const;
-            ///< Throws an exception if getState()!=State_End
+            ///< Throws an exception if the last call to parse did not return true.
     };
 }
 

apps/opencs/view/filter/editwidget.cpp

@@ -10,9 +10,7 @@ CSVFilter::EditWidget::EditWidget (QWidget *parent)
 
 void CSVFilter::EditWidget::textChanged (const QString& text)
 {
-    mParser.parse (text.toUtf8().constData());
-
-    if (mParser.getState()==CSMFilter::Parser::State_End)
+    if (mParser.parse (text.toUtf8().constData()))
     {
         setPalette (mPalette);
         emit filterChanged (mParser.getFilter(), "");