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

[DEBUG] try to correct parsing recursivity error

Browse Source
This commit is contained in:
Edouard DUPIN 2014-07-28 23:56:40 +02:00
parent 3ca9ce7a2c
commit 1c454891b6
2 changed files with 130 additions and 48 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

32
etk/RegExp.cpp
View File

@ -325,8 +325,7 @@ int64_t etk::regexp::getLenOfNormal(const std::vector<char32_t>& _data, int64_t
} }
bool etk::regexp::parseBrace(const std::vector<char32_t>& _data, uint32_t& _min, uint32_t& _max) bool etk::regexp::parseBrace(const std::vector<char32_t>& _data, uint32_t& _min, uint32_t& _max) {
{
//TK_INFO("parse {...} in "; DisplayElem(data); ); //TK_INFO("parse {...} in "; DisplayElem(data); );
int64_t k=0; int64_t k=0;
@ -383,4 +382,33 @@ allIsSet:
return true; return true;
} }
std::string etk::regexp::autoStr(const std::string& _data) {
std::string out;
for (auto &it : _data) {
if (it == '\n') {
out += "\\n";
} else if (it == '\t') {
out += "\\t";
} else if (it == '\r') {
out += "\\r";
} else {
out += it;
}
}
return out;
}
std::string etk::regexp::autoStr(char _data) {
std::string out;
if (_data == '\n') {
out += "\\n";
} else if (_data == '\t') {
out += "\\t";
} else if (_data == '\r') {
out += "\\r";
} else {
out += _data;
}
return out;
}

102
etk/RegExp.h
View File

@ -145,6 +145,9 @@ int64_t getLenOfBrace(const std::vector<char32_t>& _data, int64_t _startPos);
int64_t getLenOfNormal(const std::vector<char32_t>& _data, int64_t _startPos); int64_t getLenOfNormal(const std::vector<char32_t>& _data, int64_t _startPos);
//! @not-in-doc //! @not-in-doc
bool parseBrace(const std::vector<char32_t>& _data, uint32_t& _min, uint32_t& _max); bool parseBrace(const std::vector<char32_t>& _data, uint32_t& _min, uint32_t& _max);
//! @not-in-doc
std::string autoStr(const std::string& _data);
std::string autoStr(char _data);
#undef __class__ #undef __class__
@ -160,6 +163,7 @@ class FindProperty {
int64_t m_positionStop; //!< find end position int64_t m_positionStop; //!< find end position
uint32_t m_multiplicity; //!< curent multiplicity of find element uint32_t m_multiplicity; //!< curent multiplicity of find element
enum parseStatus m_status; //!< curent status of parsing enum parseStatus m_status; //!< curent status of parsing
int32_t m_subIndex; //!< dubindex int the upper list ... for (...)
public: public:
std::vector<FindProperty> m_subProperty; //!< list of all sub elements std::vector<FindProperty> m_subProperty; //!< list of all sub elements
public: public:
@ -167,9 +171,17 @@ class FindProperty {
m_positionStart(-1), m_positionStart(-1),
m_positionStop(-1), m_positionStop(-1),
m_multiplicity(0), m_multiplicity(0),
m_status(parseStatusUnknow) { m_status(parseStatusUnknow),
m_subIndex(-1) {
// nothing to do ... // nothing to do ...
} }
void reset() {
m_positionStart = -1;
m_positionStop = -1;
m_multiplicity = 0;
m_status = parseStatusUnknow;
m_subIndex = -1;
}
int64_t getPositionStart() const { int64_t getPositionStart() const {
return m_positionStart; return m_positionStart;
} }
@ -206,6 +218,12 @@ class FindProperty {
enum parseStatus getStatus() const { enum parseStatus getStatus() const {
return m_status; return m_status;
} }
int32_t getSubIndex() const {
return m_subIndex;
}
void setSubIndex(int32_t _newIndex) {
m_subIndex = _newIndex;
}
void display(const std::string& _data, int32_t _level = 0) { void display(const std::string& _data, int32_t _level = 0) {
TK_INFO("prop : " << levelSpace(_level) << " [" TK_INFO("prop : " << levelSpace(_level) << " ["
@ -353,7 +371,12 @@ template<class CLASS_TYPE> class NodeValue : public Node<CLASS_TYPE> {
uint32_t ofset = 0; uint32_t ofset = 0;
int64_t kkk; int64_t kkk;
for (kkk=0; findLen+kkk<_lenMax && kkk < (int64_t)m_data.size(); kkk++) { for (kkk=0; findLen+kkk<_lenMax && kkk < (int64_t)m_data.size(); kkk++) {
TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " check element value : '" << (char)m_data[kkk] << "' ?= '" << (char)_data[_currentPos+findLen+kkk] << "'"); TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel)
<< " check element value : '"
<< etk::regexp::autoStr((char)m_data[kkk])
<< "' ?= '"
<< etk::regexp::autoStr((char)_data[_currentPos+findLen+kkk])
<< "'");
if (m_data[kkk] != (char32_t)_data[_currentPos+findLen+kkk]) { if (m_data[kkk] != (char32_t)_data[_currentPos+findLen+kkk]) {
tmpFind=false; tmpFind=false;
break; break;
@ -430,6 +453,7 @@ 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_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);
char32_t tmpVal = _data[_currentPos]; char32_t tmpVal = _data[_currentPos];
bool find = false; bool find = false;
// Check range // Check range
@ -443,6 +467,7 @@ template<class CLASS_TYPE> class NodeRangeValue : public Node<CLASS_TYPE> {
// Check Value // Check Value
if (find == false) { if (find == false) {
for (auto &it : m_dataList) { for (auto &it : m_dataList) {
TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << getDescriptiveName() << " : '" << autoStr(tmpVal) << "'=?='" << autoStr(it) << "'");
if (tmpVal == it) { if (tmpVal == it) {
find = true; find = true;
break; break;
@ -455,7 +480,9 @@ template<class CLASS_TYPE> class NodeRangeValue : public Node<CLASS_TYPE> {
|| ( find == false || ( find == false
&& m_invert == true) ) { && m_invert == true) ) {
find = true; find = true;
TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << getDescriptiveName() << " : Find (invert=" << m_invert << ")");
} else { } else {
TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << getDescriptiveName() << " : Not find (invert=" << m_invert << ")");
find = false; find = false;
} }
if (find == true) { if (find == true) {
@ -469,23 +496,26 @@ template<class CLASS_TYPE> class NodeRangeValue : public Node<CLASS_TYPE> {
if(_property.getMultiplicity() > Node<CLASS_TYPE>::m_multipleMax) { if(_property.getMultiplicity() > Node<CLASS_TYPE>::m_multipleMax) {
_property.multiplicityDecrement(); _property.multiplicityDecrement();
_property.setStatus(parseStatusFull); _property.setStatus(parseStatusFull);
return;
} else { } else {
_property.setPositionStop(newPosVal); _property.setPositionStop(newPosVal);
if (_currentPos>=_lenMax) { if (_currentPos>=_lenMax) {
_property.setStatus(parseStatusFull); _property.setStatus(parseStatusFull);
return; } else {
if(_property.getMultiplicity() == Node<CLASS_TYPE>::m_multipleMax) {
_property.setStatus(parseStatusFull);
} else { } else {
_property.setStatus(parseStatusPartial); _property.setStatus(parseStatusPartial);
return;
} }
} }
} }
}else {
if (_property.getPositionStop() != -1) { if (_property.getPositionStop() != -1) {
_property.setStatus(parseStatusFull); _property.setStatus(parseStatusFull);
return; } else {
}
_property.setStatus(parseStatusNone); _property.setStatus(parseStatusNone);
}
}
TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " " << getDescriptiveName() << " : out=" << _property);
return; return;
}; };
virtual void display() { virtual void display() {
@ -967,6 +997,7 @@ 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)) << "'");
int findLen = 0; int findLen = 0;
bool error = false; bool error = false;
size_t iii = 0; size_t iii = 0;
@ -988,10 +1019,12 @@ template<class CLASS_TYPE> class NodePTheseElem : public Node<CLASS_TYPE> {
// We did not find the element : // We did not find the element :
if (findPartialNode == false) { if (findPartialNode == false) {
_property.m_subProperty.clear(); _property.m_subProperty.clear();
_property.reset();
} }
} }
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)) << "'");
m_subNode[iii]->parse(_data, tmpCurrentPos, _lenMax, prop); m_subNode[iii]->parse(_data, tmpCurrentPos, _lenMax, prop);
if (prop.getStatus() == parseStatusNone) { if (prop.getStatus() == parseStatusNone) {
// rewind the list: // rewind the list:
@ -1016,13 +1049,14 @@ template<class CLASS_TYPE> class NodePTheseElem : public Node<CLASS_TYPE> {
} }
tmpCurrentPos = prop.getPositionStop(); tmpCurrentPos = prop.getPositionStop();
_property.m_subProperty.push_back(prop); _property.m_subProperty.push_back(prop);
TK_REG_EXP_DBG_MODE("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (Elem=" << iii << "/" << m_subNode.size() << ") find : " << prop); TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (elem=" << iii << "/" << m_subNode.size() << ") find : " << prop);
prop.reset();
iii++; iii++;
} }
_property.setStatus(parseStatusFull); _property.setStatus(parseStatusFull);
// Display sub List : // Display sub List :
for (auto &it : _property.m_subProperty) { for (auto &it : _property.m_subProperty) {
TK_REG_EXP_DBG_MODE(" plop : " << it); TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (elem) sub=" << it);
} }
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) {
@ -1118,44 +1152,63 @@ 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 = 0; int32_t findLen = 0;
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)) << "'");
if (0 == m_subNode.size()) { if (0 == m_subNode.size()) {
_property.setStatus(parseStatusNone); _property.setStatus(parseStatusNone);
return; return;
} }
if (_property.getMultiplicity() >= Node<CLASS_TYPE>::m_multipleMax) {
_property.setStatus(parseStatusFull);
return;
}
_property.setStatus(parseStatusFull);
bool tmpFind = true; bool tmpFind = true;
while ( _property.getMultiplicity() < Node<CLASS_TYPE>::m_multipleMax while ( _property.getMultiplicity() <= Node<CLASS_TYPE>::m_multipleMax
&& tmpFind == true) { && tmpFind == true) {
tmpFind = false; tmpFind = false;
for (auto &it : m_subNode) { for (size_t iii=0; iii<m_subNode.size(); ++iii) {
FindProperty prop; FindProperty prop;
prop.reset();
prop.setPositionStart(_currentPos+findLen); prop.setPositionStart(_currentPos+findLen);
int32_t offset = 0; int32_t offset = 0;
do { m_subNode[iii]->parse(_data, _currentPos+findLen+offset, _lenMax, prop);
it->parse(_data, _currentPos+findLen+offset, _lenMax, prop);
offset = prop.getFindLen(); offset = prop.getFindLen();
TK_REG_EXP_DBG_MODE("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) mult=" << _property.getMultiplicity() << " tmp " << prop.getFindLen()); TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) mult=" <<
} while (prop.getStatus() == parseStatusPartial); _property.getMultiplicity() << " tmp " << prop.getFindLen());
if (prop.getStatus() == parseStatusFull) { if ( prop.getStatus() == parseStatusFull
|| prop.getStatus() == parseStatusPartial) {
findLen += prop.getFindLen(); findLen += prop.getFindLen();
prop.setSubIndex(iii);
_property.m_subProperty.push_back(prop); _property.m_subProperty.push_back(prop);
tmpFind = true; tmpFind = true;
} else if (prop.getStatus() == parseStatusPartial) { break;
} }
} }
TK_REG_EXP_DBG_MODE("Parse " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) mult=" << _property.getMultiplicity() << " find " << findLen); if (tmpFind == true) {
_property.multiplicityIncrement(); _property.multiplicityIncrement();
TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) mult=" << _property.getMultiplicity() << " find " << findLen);
if (_property.getMultiplicity() >= Node<CLASS_TYPE>::m_multipleMin) {
_property.setStatus(parseStatusPartial);
break;
}
} else {
TK_TODO("plop");
}
}
for (int64_t iii=_property.m_subProperty.size()-1; iii>=0; --iii) {
if (_property.m_subProperty[iii].getStatus() == parseStatusPartial) {
_property.setStatus(parseStatusPartial);
break;
}
} }
_property.setPositionStop(_property.getPositionStart() + findLen); _property.setPositionStop(_property.getPositionStart() + findLen);
if( _property.getMultiplicity() >= Node<CLASS_TYPE>::m_multipleMin if( _property.getMultiplicity() >= Node<CLASS_TYPE>::m_multipleMin
&& _property.getMultiplicity() <= Node<CLASS_TYPE>::m_multipleMax && _property.getMultiplicity() <= Node<CLASS_TYPE>::m_multipleMax
&& findLen> 0 ) { && findLen> 0 ) {
TK_REG_EXP_DBG_MODE("find " << findLen); TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) find " << findLen);
_property.setStatus(parseStatusFull);
return; return;
} else if( 0 == Node<CLASS_TYPE>::m_multipleMin ) { } else if( 0 == Node<CLASS_TYPE>::m_multipleMin ) {
TK_REG_EXP_DBG_MODE("find size=0"); TK_REG_EXP_DBG_MODE(" " << levelSpace(Node<CLASS_TYPE>::m_nodeLevel) << " (...) find size=0");
_property.setStatus(parseStatusFull);
return; return;
} }
_property.setStatus(parseStatusNone); _property.setStatus(parseStatusNone);
@ -1489,7 +1542,8 @@ template<class CLASS_TYPE> class RegExp {
regexp::FindProperty prop; regexp::FindProperty prop;
prop.setPositionStart(iii); prop.setPositionStart(iii);
m_exprRootNode.parse(_SearchIn, iii, maxlen, prop); m_exprRootNode.parse(_SearchIn, iii, maxlen, prop);
if (prop.getStatus() == regexp::parseStatusFull) { if ( prop.getStatus() == regexp::parseStatusFull
|| prop.getStatus() == regexp::parseStatusPartial ) {
findLen = prop.getFindLen(); findLen = prop.getFindLen();
TK_DEBUG("main search find : " << findLen << " elements"); TK_DEBUG("main search find : " << findLen << " elements");
if ( _escapeChar != 0 if ( _escapeChar != 0