atria-soft/etk
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

[DEBUG] some correction of the regexpParser

Browse Source
This commit is contained in:
Edouard DUPIN 2014-07-29 15:24:11 +02:00
parent 4719d7d0b8
commit 27d1f599e6
4 changed files with 49 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
etk/RegExp.cpp
View File

@ -412,3 +412,12 @@ std::string etk::regexp::autoStr(char _data) {
} }
return out; return out;
} }
std::string etk::regexp::strTick(int32_t _pos) {
std::string out;
for (int32_t iii=0; iii<_pos; ++iii) {
out += " ";
}
out += "^";
return out;
}

34
etk/RegExp.h
View File

@ -16,8 +16,10 @@
#include <vector> #include <vector>
#include <memory> #include <memory>
#define TK_REG_EXP_DBG_MODE2 TK_HIDDEN
#define TK_REG_EXP_DBG_MODE TK_HIDDEN
//#define TK_REG_EXP_DBG_MODE TK_VERBOSE //#define TK_REG_EXP_DBG_MODE TK_VERBOSE
#define TK_REG_EXP_DBG_MODE TK_DEBUG //#define TK_REG_EXP_DBG_MODE TK_DEBUG
//regular colors //regular colors
#define ETK_BASH_COLOR_BLACK "\e[0;30m" #define ETK_BASH_COLOR_BLACK "\e[0;30m"
@ -148,6 +150,7 @@ bool parseBrace(const std::vector<char32_t>& _data, uint32_t& _min, uint32_t& _m
//! @not-in-doc //! @not-in-doc
std::string autoStr(const std::string& _data); std::string autoStr(const std::string& _data);
std::string autoStr(char _data); std::string autoStr(char _data);
std::string strTick(int32_t _pos);
#undef __class__ #undef __class__
@ -454,6 +457,13 @@ template<class CLASS_TYPE> class NodeRangeValue : public Node<CLASS_TYPE> {
int32_t findLen = 0; int32_t findLen = 0;
TK_REG_EXP_DBG_MODE("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << getDescriptiveName() << "{" << Node<CLASS_TYPE>::m_multipleMin << "," << Node<CLASS_TYPE>::m_multipleMax << "}"); TK_REG_EXP_DBG_MODE("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << getDescriptiveName() << "{" << Node<CLASS_TYPE>::m_multipleMin << "," << Node<CLASS_TYPE>::m_multipleMax << "}");
TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << getDescriptiveName() << " input property :" << _property); TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << getDescriptiveName() << " input property :" << _property);
if ( _property.getPositionStop() < 0
&& Node<CLASS_TYPE>::m_multipleMin == 0
&& _property.getMultiplicity() == 0) {
_property.setPositionStop(_property.getPositionStart());
_property.setStatus(parseStatusPartial);
return;
}
char32_t tmpVal = _data[_currentPos]; char32_t tmpVal = _data[_currentPos];
bool find = false; bool find = false;
// Check range // Check range
@ -511,6 +521,9 @@ template<class CLASS_TYPE> class NodeRangeValue : public Node<CLASS_TYPE> {
}else { }else {
if (_property.getPositionStop() != -1) { if (_property.getPositionStop() != -1) {
_property.setStatus(parseStatusFull); _property.setStatus(parseStatusFull);
} else if (_property.getMultiplicity() == Node<CLASS_TYPE>::m_multipleMin) {
_property.setPositionStop(_property.getPositionStart());
_property.setStatus(parseStatusFull);
} else { } else {
_property.setStatus(parseStatusNone); _property.setStatus(parseStatusNone);
} }
@ -997,7 +1010,7 @@ template<class CLASS_TYPE> class NodePTheseElem : public Node<CLASS_TYPE> {
return _data.size(); return _data.size();
}; };
virtual void parse(const CLASS_TYPE& _data, int64_t _currentPos, int64_t _lenMax, FindProperty& _property) { virtual void parse(const CLASS_TYPE& _data, int64_t _currentPos, int64_t _lenMax, FindProperty& _property) {
TK_REG_EXP_DBG_MODE("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (elem) data to parse : '");// << autoStr(std::string(_data, _currentPos, _lenMax-_currentPos)) << "'"); TK_REG_EXP_DBG_MODE2("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (elem) data to parse : '" << autoStr(std::string(_data, _currentPos, _lenMax-_currentPos)) << "'");
int findLen = 0; int findLen = 0;
bool error = false; bool error = false;
size_t iii = 0; size_t iii = 0;
@ -1024,7 +1037,7 @@ template<class CLASS_TYPE> class NodePTheseElem : public Node<CLASS_TYPE> {
} }
prop.setPositionStart(tmpCurrentPos); prop.setPositionStart(tmpCurrentPos);
while (iii < m_subNode.size()) { while (iii < m_subNode.size()) {
TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (elem=" << iii << "/" << m_subNode.size() << ") data='");// << autoStr(std::string(_data, tmpCurrentPos, _lenMax-tmpCurrentPos)) << "'"); TK_REG_EXP_DBG_MODE2(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (elem=" << iii << "/" << m_subNode.size() << ") data='" << autoStr(std::string(_data, tmpCurrentPos, _lenMax-tmpCurrentPos)) << "'");
m_subNode[iii]->parse(_data, tmpCurrentPos, _lenMax, prop); m_subNode[iii]->parse(_data, tmpCurrentPos, _lenMax, prop);
if (prop.getStatus() == parseStatusNone) { if (prop.getStatus() == parseStatusNone) {
TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (elem=" << iii << "/" << m_subNode.size() << ") ===None==="); TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (elem=" << iii << "/" << m_subNode.size() << ") ===None===");
@ -1156,7 +1169,7 @@ template<class CLASS_TYPE> class NodePThese : public Node<CLASS_TYPE> {
virtual void parse(const CLASS_TYPE& _data, int64_t _currentPos, int64_t _lenMax, FindProperty& _property) { virtual void parse(const CLASS_TYPE& _data, int64_t _currentPos, int64_t _lenMax, FindProperty& _property) {
int32_t findLen = _property.getFindLen(); int32_t findLen = _property.getFindLen();
TK_REG_EXP_DBG_MODE("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) {" << Node<CLASS_TYPE>::m_multipleMin << "," << Node<CLASS_TYPE>::m_multipleMax << "}"); TK_REG_EXP_DBG_MODE("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) {" << Node<CLASS_TYPE>::m_multipleMin << "," << Node<CLASS_TYPE>::m_multipleMax << "}");
//TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) data='" << autoStr(std::string(_data, _currentPos, _lenMax-_currentPos)) << "'"); TK_REG_EXP_DBG_MODE2(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) data='" << autoStr(std::string(_data, _currentPos, _lenMax-_currentPos)) << "'");
TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) input property=" << _property); TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) input property=" << _property);
if (0 == m_subNode.size()) { if (0 == m_subNode.size()) {
_property.setStatus(parseStatusNone); _property.setStatus(parseStatusNone);
@ -1374,9 +1387,8 @@ template<class CLASS_TYPE> class RegExp {
int64_t jjj; int64_t jjj;
// Find the element in the list... // Find the element in the list...
for (jjj=0; jjj<regexp::constConvertionTableSize; jjj++) { for (jjj=0; jjj<regexp::constConvertionTableSize; jjj++) {
if( true == regexp::constConvertionTable[jjj].haveBackSlash if ( regexp::constConvertionTable[jjj].haveBackSlash == true
&& _regexp[iii+1] == (char32_t)regexp::constConvertionTable[jjj].inputValue) && _regexp[iii+1] == (char32_t)regexp::constConvertionTable[jjj].inputValue) {
{
if (regexp::constConvertionTable[jjj].newValue==0) { if (regexp::constConvertionTable[jjj].newValue==0) {
tmpExp.push_back(regexp::constConvertionTable[jjj].specialChar); tmpExp.push_back(regexp::constConvertionTable[jjj].specialChar);
} else { } else {
@ -1387,8 +1399,9 @@ template<class CLASS_TYPE> class RegExp {
} }
// check error : // check error :
if (jjj==regexp::constConvertionTableSize) { if (jjj==regexp::constConvertionTableSize) {
TK_ERROR("Dangerous parse of the \\ " << _regexp[iii+1] << " at element " << iii); TK_WARNING(" parse : " << _regexp);
// TODO : Generate Exeption ... TK_WARNING(" " << etk::regexp::strTick(iii+1));
TK_ERROR("Dangerous parse of the \\x with the value : '" << _regexp[iii+1] << "' at element " << iii);
return; return;
} }
// less one char in the regular expression ... // less one char in the regular expression ...
@ -1616,7 +1629,8 @@ template<class CLASS_TYPE> class RegExp {
regexp::FindProperty prop; regexp::FindProperty prop;
prop.setPositionStart(_startPos); prop.setPositionStart(_startPos);
m_exprRootNode.parse(_SearchIn, _startPos, maxlen, prop); m_exprRootNode.parse(_SearchIn, _startPos, maxlen, prop);
if (prop.getStatus() == regexp::parseStatusFull) { if ( prop.getStatus() == regexp::parseStatusFull
|| prop.getStatus() == regexp::parseStatusPartial ) {
findLen = prop.getFindLen(); findLen = prop.getFindLen();
if ( _escapeChar != 0 if ( _escapeChar != 0
&& _startPos>0) { && _startPos>0) {

2
etk/debug.h
View File

@ -42,6 +42,8 @@ namespace etk {
#define TK_TODO(data) do { } while(false) #define TK_TODO(data) do { } while(false)
#endif #endif
#define TK_HIDDEN(data) do { } while(false)
#define TK_ASSERT(cond,data) \ #define TK_ASSERT(cond,data) \
do { \ do { \
if (!(cond)) { \ if (!(cond)) { \

18
test/main.cpp
View File

@ -190,13 +190,15 @@ void testColor() {
void testRegExpSingle(const std::string& _expression, const std::string& _search) { void testRegExpSingle(const std::string& _expression, const std::string& _search) {
etk::RegExp<std::string> expression(_expression); etk::RegExp<std::string> expression(_expression);
TK_INFO("Parse RegEx : " << expression.getRegExDecorated()); TK_INFO("Parse RegEx: " << expression.getRegExDecorated());
TK_INFO(" IN: " << etk::regexp::autoStr(_search));
if (expression.parse(_search, 0, _search.size()) == true) { if (expression.parse(_search, 0, _search.size()) == true) {
TK_INFO(" match [" << expression.start() << ".." << expression.stop() << "] "); TK_INFO(" match [" << expression.start() << ".." << expression.stop() << "] ");
TK_INFO(" ==> '" << std::string(_search, expression.start(), expression.stop() - expression.start()) << "'"); TK_INFO(" ==> '" << etk::regexp::autoStr(std::string(_search, expression.start(), expression.stop() - expression.start())) << "'");
} }
} }
void testRegExp() { void testRegExp() {
//std::string data = " a /* plop */ \n int eee = 22; // error value \nint main(void) {\n return 0;\n}\n"; //std::string data = " a /* plop */ \n int eee = 22; // error value \nint main(void) {\n return 0;\n}\n";
//std::string data = "alpha /* plop */ test"; //std::string data = "alpha /* plop */ test";
@ -209,8 +211,16 @@ void testRegExp() {
//testRegExpSingle("a.*plop(z{2,3}|h+)+r", data); //testRegExpSingle("a.*plop(z{2,3}|h+)+r", data);
//std::string data = "pp \n# plop // qdfqdfsdf \nde"; //std::string data = "pp \n# plop // qdfqdfsdf \nde";
std::string data = "pp \n# plop //\\\n qdfqdfsdf \nde"; //std::string data = "pp \n# plop //\\\n qdfqdfsdf \nde";
testRegExpSingle("#(\\\\\\\\|\\\\\\n|.)*$", data); //std::string data = "p#\ne";
//testRegExpSingle("#(\\\\\\\\|\\\\\\n|.)*$", data);
//testRegExpSingle("#.*$", data);
//std::string data = "p//TODO:\ndse";
//std::string data = "p// TODO:\ndse";
//std::string data = "p// TODO :\ndse";
std::string data = "p// TODO : sdfgsdfsd \ndse";
testRegExpSingle("//[ \\t]*TODO[ \\t]*:.*$", data);
} }
int main(int argc, const char *argv[]) { int main(int argc, const char *argv[]) {