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

[DEV] corection of the RegExp properties

Browse Source
This commit is contained in:
Edouard DUPIN 2014-10-02 23:22:51 +02:00
parent 455ffb47e6
commit d162f2a70d
2 changed files with 48 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

43
etk/RegExp.h
View File

@ -208,6 +208,9 @@ class FindProperty {
} }
void setPositionStop(int64_t _newPos) { void setPositionStop(int64_t _newPos) {
m_positionStop = _newPos; m_positionStop = _newPos;
if (m_positionStop < m_positionStart) {
TK_CRITICAL("set volontary a stop position before end : " << this);
}
} }
uint32_t getMultiplicity() const { uint32_t getMultiplicity() const {
return m_multiplicity; return m_multiplicity;
@ -420,18 +423,20 @@ template<class CLASS_TYPE> class NodeValue : 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) {
TK_REG_DEBUG("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " Value{" << Node<CLASS_TYPE>::m_multipleMin << "," << Node<CLASS_TYPE>::m_multipleMax << "} : " << (char)m_data[0]); TK_REG_DEBUG("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " Value{" << Node<CLASS_TYPE>::m_multipleMin << "," << Node<CLASS_TYPE>::m_multipleMax << "} : " << (char)m_data[0]);
TK_REG_DEBUG(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << createString(Node<CLASS_TYPE>::m_regExpData));
TK_REG_DEBUG_3(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " Value " << _property); TK_REG_DEBUG_3(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " Value " << _property);
if (m_data.size() == 0) { if (m_data.size() == 0) {
TK_ERROR("No data inside type elemTypeValue"); TK_ERROR("No data inside type elemTypeValue");
_property.setStatus(parseStatusNone); _property.setStatus(parseStatusNone);
return; return;
} }
if ( _property.getPositionStop() < 0 if (_property.getStatus() != parseStatusPartial) {
&& Node<CLASS_TYPE>::m_multipleMin == 0 if (Node<CLASS_TYPE>::m_multipleMin == 0) {
&& _property.getMultiplicity() == 0) { _property.setPositionStop(_property.getPositionStart());
_property.setPositionStop(_property.getPositionStart()); _property.setStatus(parseStatusPartial);
_property.setStatus(parseStatusPartial); TK_REG_DEBUG("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " ==> partial (minSize=0)");
return; return;
}
} }
bool tmpFind = true; bool tmpFind = true;
int32_t findLen = 0; int32_t findLen = 0;
@ -531,13 +536,15 @@ template<class CLASS_TYPE> class NodeRangeValue : 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 = 0; int32_t findLen = 0;
TK_REG_DEBUG("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << getDescriptiveName() << "{" << Node<CLASS_TYPE>::m_multipleMin << "," << Node<CLASS_TYPE>::m_multipleMax << "}"); TK_REG_DEBUG("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << getDescriptiveName() << "{" << Node<CLASS_TYPE>::m_multipleMin << "," << Node<CLASS_TYPE>::m_multipleMax << "}");
TK_REG_DEBUG(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << createString(Node<CLASS_TYPE>::m_regExpData));
TK_REG_DEBUG_3(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << getDescriptiveName() << " " << _property); TK_REG_DEBUG_3(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << getDescriptiveName() << " " << _property);
if ( _property.getPositionStop() < 0 if (_property.getStatus() != parseStatusPartial) {
&& Node<CLASS_TYPE>::m_multipleMin == 0 if (Node<CLASS_TYPE>::m_multipleMin == 0) {
&& _property.getMultiplicity() == 0) { _property.setPositionStop(_property.getPositionStart());
_property.setPositionStop(_property.getPositionStart()); _property.setStatus(parseStatusPartial);
_property.setStatus(parseStatusPartial); TK_REG_DEBUG("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " ==> partial (minSize=0)");
return; return;
}
} }
char32_t tmpVal = _data[_currentPos]; char32_t tmpVal = _data[_currentPos];
bool find = false; bool find = false;
@ -695,6 +702,7 @@ template<class CLASS_TYPE> class NodeSOL : public Node<CLASS_TYPE> {
int32_t findLen = 0; int32_t findLen = 0;
bool tmpFind = false; bool tmpFind = false;
TK_REG_DEBUG("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " SOL{" << Node<CLASS_TYPE>::m_multipleMin << "," << Node<CLASS_TYPE>::m_multipleMax << "}"); TK_REG_DEBUG("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " SOL{" << Node<CLASS_TYPE>::m_multipleMin << "," << Node<CLASS_TYPE>::m_multipleMax << "}");
TK_REG_DEBUG(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << createString(Node<CLASS_TYPE>::m_regExpData));
// TODO : is it really what I want ... (maybe next ellement will be requested... (check if previous element is \r or \n // TODO : is it really what I want ... (maybe next ellement will be requested... (check if previous element is \r or \n
while ( _property.getMultiplicity() < Node<CLASS_TYPE>::m_multipleMax while ( _property.getMultiplicity() < Node<CLASS_TYPE>::m_multipleMax
&& tmpFind == true && tmpFind == true
@ -1036,10 +1044,10 @@ template<class CLASS_TYPE> class NodePTheseElem : public Node<CLASS_TYPE> {
continue; continue;
} }
} }
tmpCurrentPos = prop.getPositionStop();
if (prop.getPositionStart() > prop.getPositionStop()) { if (prop.getPositionStart() > prop.getPositionStop()) {
TK_CRITICAL("Very bad case ... : " << prop); TK_CRITICAL("Very bad case ... : " << prop);
} }
tmpCurrentPos = prop.getPositionStop();
_property.m_subProperty.push_back(prop); _property.m_subProperty.push_back(prop);
TK_REG_DEBUG(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (elem=" << iii << "/" << m_subNode.size() << ") === OK === find : " << prop); TK_REG_DEBUG(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (elem=" << iii << "/" << m_subNode.size() << ") === OK === find : " << prop);
prop.reset(); prop.reset();
@ -1150,12 +1158,21 @@ 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) {
TK_REG_DEBUG("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) {" << Node<CLASS_TYPE>::m_multipleMin << "," << Node<CLASS_TYPE>::m_multipleMax << "}"); TK_REG_DEBUG("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) {" << Node<CLASS_TYPE>::m_multipleMin << "," << Node<CLASS_TYPE>::m_multipleMax << "}");
TK_REG_DEBUG(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << createString(Node<CLASS_TYPE>::m_regExpData));
TK_REG_DEBUG_2(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) data='" << autoStr(std::string(_data, _currentPos, _lenMax-_currentPos)) << "'"); TK_REG_DEBUG_2(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) data='" << autoStr(std::string(_data, _currentPos, _lenMax-_currentPos)) << "'");
TK_REG_DEBUG_3(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) input property=" << _property); TK_REG_DEBUG_3(" " << 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);
return; return;
} }
if (_property.getStatus() != parseStatusPartial) {
if (Node<CLASS_TYPE>::m_multipleMin == 0) {
_property.setStatus(parseStatusPartial);
_property.setPositionStop(_property.getPositionStart());
TK_REG_DEBUG("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " ==> partial (minSize=0)");
return;
}
}
bool haveSubPartial = false; bool haveSubPartial = false;
for (int64_t iii=_property.m_subProperty.size()-1; iii>=0; --iii) { for (int64_t iii=_property.m_subProperty.size()-1; iii>=0; --iii) {
if (_property.m_subProperty[iii].getStatus() == parseStatusPartial) { if (_property.m_subProperty[iii].getStatus() == parseStatusPartial) {

22
test/main.cpp
View File

@ -253,6 +253,7 @@ void testRegExp() {
data = "ddfgdfgh"; data = "ddfgdfgh";
etk::RegExp<std::string> reg(".*"); etk::RegExp<std::string> reg(".*");
reg.setMaximize(true); reg.setMaximize(true);
TK_INFO("Parse RegEx : '" << reg.getRegExDecorated() << "'"); TK_INFO("Parse RegEx : '" << reg.getRegExDecorated() << "'");
if (reg.parse(data, 0, data.size()) == true) { if (reg.parse(data, 0, data.size()) == true) {
//if (reg.processOneElement(data, 0, data.size()) == true) { //if (reg.processOneElement(data, 0, data.size()) == true) {
@ -260,19 +261,32 @@ void testRegExp() {
TK_INFO(" ==> '" << std::string(data, reg.start(), reg.stop()-reg.start()) << "'"); TK_INFO(" ==> '" << std::string(data, reg.start(), reg.stop()-reg.start()) << "'");
} }
data = "plop \"\" sdfsdf s\"swdfsqd"; data = "plop \"\" sdfsdf s\"swdfsqd sdfgsdfg \" \" sdfsf";
reg = etk::RegExp<std::string>("\"(\\\\[\\\\\"]|.)*\""); reg = etk::RegExp<std::string>("\"(\\\\[\\\\\"]|.)*\"");
//reg.setMaximize(true); reg.setMaximize(false);
TK_INFO("Parse RegEx : '" << reg.getRegExDecorated() << "'"); TK_INFO("Parse RegEx : '" << reg.getRegExDecorated() << "'");
if (reg.parse(data, 0, data.size()) == true) { if (reg.parse(data, 0, data.size()) == true) {
//if (reg.processOneElement(data, 0, data.size()) == true) { //if (reg.processOneElement(data, 0, data.size()) == true) {
TK_INFO(" match [" << reg.start() << ".." << reg.stop() << "] "); TK_INFO(" match [" << reg.start() << ".." << reg.stop() << "] ");
TK_INFO(" ==> '" << std::string(data, reg.start(), reg.stop()-reg.start()) << "'"); TK_INFO(" ==> '" << std::string(data, reg.start(), reg.stop()-reg.start()) << "'");
} }
//TODO : good : "(\\+|[0-9])*" ==> really bad : "(+|[0-9])*" //TODO : good : "(\\+|[0-9])*" ==> really bad : "(+|[0-9])*"
data = "void limit(const vec2& _origin, const vec2& _size);\n";
reg = etk::RegExp<std::string>("\\@(\\w|_)+[ \\t]*\\(");
reg.setMaximize(false);
TK_INFO("Parse RegEx : '" << reg.getRegExDecorated() << "'");
if (reg.parse(data, 0, data.size()) == true) {
//if (reg.processOneElement(data, 0, data.size()) == true) {
TK_INFO(" match [" << reg.start() << ".." << reg.stop() << "] ");
TK_INFO(" ==> '" << std::string(data, reg.start(), reg.stop()-reg.start()) << "'");
}
data = "void limit const vec2& _origin, const vec2& _size);\n";
if (reg.parse(data, 0, data.size()) == true) {
//if (reg.processOneElement(data, 0, data.size()) == true) {
TK_INFO(" match [" << reg.start() << ".." << reg.stop() << "] ");
TK_INFO(" ==> '" << std::string(data, reg.start(), reg.stop()-reg.start()) << "'");
}
} }
int main(int argc, const char *argv[]) { int main(int argc, const char *argv[]) {