/* the entry point */
void html2nodes(const char *htmltext, bool startpage)
{
char *htmlfix = 0;
tdoc = tidyCreate();
if (!startpage)
tidyOptSetInt(tdoc, TidyBodyOnly, yes);
tidySetReportFilter(tdoc, tidyErrorHandler);
// tidySetReportFilter(tdoc, tidyReportFilter);
tidySetCharEncoding(tdoc, (cons_utf8 ? "utf8" : "latin1"));
htmlfix = tidyPreprocess(htmltext);
if (htmlfix) {
tidyParseString(tdoc, htmlfix);
nzFree(htmlfix);
} else
tidyParseString(tdoc, htmltext);
tidyCleanAndRepair(tdoc);
if (debugLevel >= 5) {
traverse_tidycall = printNode;
traverseTidy();
}
/* convert tidy nodes into edbrowse nodes */
traverse_tidycall = convertNode;
traverseTidy();
tidyRelease(tdoc);
} /* html2nodes */
int lua_tidy_SetCharEncoding(lua_State *L)
{
pTidy t = toTidy(L,1);
const char * enc = lua_tostring(L,2);
lua_pushnumber(L, tidySetCharEncoding (t->tdoc, enc));
return 1;
}
QString tidyHtml(QString str, bool& ok) {
#ifdef NO_TIDY
ok = true;
return str;
#else
QString res = str;
ok = false;
static bool isTidyWithIntBodyOnly = isTidyWithIntBodyOnlyCheck();
TidyDoc tdoc = tidyCreate();
TidyBuffer output;
TidyBuffer errbuf;
tidyBufInit(&output);
tidyBufInit(&errbuf);
bool configOk =
tidyOptSetBool(tdoc, TidyXhtmlOut, yes) &&
tidyOptSetBool(tdoc, TidyForceOutput, yes) &&
tidyOptSetBool(tdoc, TidyMark, no) &&
(isTidyWithIntBodyOnly
? tidyOptSetInt(tdoc, TidyBodyOnly, 1)
: tidyOptSetBool(tdoc, TidyBodyOnly, yes)) &&
tidyOptSetInt(tdoc, TidyWrapLen, 0) &&
tidyOptSetInt(tdoc, TidyDoctypeMode, TidyDoctypeOmit);
if (configOk &&
(tidySetCharEncoding(tdoc, "utf8") >= 0) &&
(tidySetErrorBuffer(tdoc, &errbuf) >= 0) &&
(tidyParseString(tdoc, str.toUtf8().data()) >= 0) &&
(tidyCleanAndRepair(tdoc) >= 0) &&
(tidyRunDiagnostics(tdoc) >= 0) &&
(tidySaveBuffer(tdoc, &output) >= 0) &&
(output.bp != 0 && output.size > 0)) {
res = QString::fromUtf8((char*)output.bp, output.size);
ok = true;
}
#ifdef DEBUG_MARKUP
if (errbuf.size > 0) {
QString errStr = QString::fromUtf8((char*)errbuf.bp, errbuf.size);
qDebug() << "\n[DEBUG] MARKUP, libtidy errors and warnings:\n" << errStr;
}
#endif
if (output.bp != 0)
tidyBufFree(&output);
if (errbuf.bp != 0)
tidyBufFree(&errbuf);
tidyRelease(tdoc);
return res.trimmed();
#endif
}
tidyhtml::tidyhtml()
{
this->tdoc = tidyCreate();
tidyOptSetBool( tdoc, TidyXhtmlOut, yes);
tidyOptSetBool( tdoc, TidyForceOutput, yes);
tidyOptSetBool( tdoc, TidyShowWarnings, no);
tidyOptSetBool( tdoc, TidyQuiet, yes);
tidyOptSetInt( tdoc, TidyShowErrors, 0);
tidySetCharEncoding(tdoc, "utf8");
}
QString tidy(QString input)
// take html code and return it converted to xhtml code
{
// the following code is (c) Charles Reitzel and Dave Raggett, see the package tidy
TidyBuffer output = {0};
TidyBuffer errbuf = {0};
QString result;
int rc = -1;
Bool ok;
TidyDoc tdoc = tidyCreate(); // Initialize "document"
ok = tidyOptSetBool( tdoc, TidyXhtmlOut, yes ); // Convert to XHTML
if ( ok ) rc = tidySetErrorBuffer( tdoc, &errbuf ); // Capture diagnostics
tidySetCharEncoding( tdoc, "utf8" );
if ( rc >= 0 ) rc = tidyParseString( tdoc, input.toUtf8().constData() ); // Parse the input
if ( rc >= 0 ) rc = tidyCleanAndRepair( tdoc ); // Tidy it up!
if ( rc >= 0 ) rc = tidyRunDiagnostics( tdoc ); // Kvetch
if ( rc > 1 ) // If error, force output.
rc = ( tidyOptSetBool(tdoc, TidyForceOutput, yes) ? rc : -1 );
if ( rc >= 0 ) rc = tidySaveBuffer( tdoc, &output ); // Pretty Print
if ( rc >= 0 )
{
char* outputstring; // content of the outputfile
// find out length of outputstring
int length=0; // length of outputstring
byte* string=output.bp;
while (*string)
{
string++;
length++;
}
outputstring=(char*)malloc(length);
snprintf(outputstring,length,"%s",output.bp);
result=QString::fromUtf8(outputstring,length);
}
else
printf( "A severe error (\%d) occurred.\\n", rc );
tidyBufFree( &output );
tidyBufFree( &errbuf );
tidyRelease( tdoc );
result=result.replace("ö","ö");
return result;
}
static int php_tidy_parse_string(PHPTidyObj *obj, char *string, uint len, char *enc)
{
TidyBuffer buf;
if(enc) {
if (tidySetCharEncoding(obj->ptdoc->doc, enc) < 0) {
php_error_docref(NULL, E_WARNING, "Could not set encoding '%s'", enc);
return FAILURE;
}
}
obj->ptdoc->initialized = 1;
tidyBufInit(&buf);
tidyBufAttach(&buf, (byte *) string, len);
if (tidyParseBuffer(obj->ptdoc->doc, &buf) < 0) {
php_error_docref(NULL, E_WARNING, "%s", obj->ptdoc->errbuf->bp);
return FAILURE;
}
tidy_doc_update_properties(obj);
return SUCCESS;
}
static void php_tidy_quick_repair(INTERNAL_FUNCTION_PARAMETERS, zend_bool is_file)
{
char *enc = NULL;
size_t enc_len = 0;
zend_bool use_include_path = 0;
TidyDoc doc;
TidyBuffer *errbuf;
zend_string *data, *arg1;
zval *config = NULL;
if (is_file) {
if (zend_parse_parameters(ZEND_NUM_ARGS(), "P|zsb", &arg1, &config, &enc, &enc_len, &use_include_path) == FAILURE) {
RETURN_FALSE;
}
if (!(data = php_tidy_file_to_mem(ZSTR_VAL(arg1), use_include_path))) {
RETURN_FALSE;
}
} else {
if (zend_parse_parameters(ZEND_NUM_ARGS(), "S|zsb", &arg1, &config, &enc, &enc_len, &use_include_path) == FAILURE) {
RETURN_FALSE;
}
data = arg1;
}
if (ZEND_SIZE_T_UINT_OVFL(ZSTR_LEN(data))) {
php_error_docref(NULL, E_WARNING, "Input string is too long");
RETURN_FALSE;
}
doc = tidyCreate();
errbuf = emalloc(sizeof(TidyBuffer));
tidyBufInit(errbuf);
if (tidySetErrorBuffer(doc, errbuf) != 0) {
tidyBufFree(errbuf);
efree(errbuf);
tidyRelease(doc);
php_error_docref(NULL, E_ERROR, "Could not set Tidy error buffer");
}
tidyOptSetBool(doc, TidyForceOutput, yes);
tidyOptSetBool(doc, TidyMark, no);
TIDY_SET_DEFAULT_CONFIG(doc);
if (config) {
TIDY_APPLY_CONFIG_ZVAL(doc, config);
}
if(enc_len) {
if (tidySetCharEncoding(doc, enc) < 0) {
php_error_docref(NULL, E_WARNING, "Could not set encoding '%s'", enc);
RETVAL_FALSE;
}
}
if (data) {
TidyBuffer buf;
tidyBufInit(&buf);
tidyBufAttach(&buf, (byte *) ZSTR_VAL(data), (uint)ZSTR_LEN(data));
if (tidyParseBuffer(doc, &buf) < 0) {
php_error_docref(NULL, E_WARNING, "%s", errbuf->bp);
RETVAL_FALSE;
} else {
if (tidyCleanAndRepair(doc) >= 0) {
TidyBuffer output;
tidyBufInit(&output);
tidySaveBuffer (doc, &output);
FIX_BUFFER(&output);
RETVAL_STRINGL((char *) output.bp, output.size ? output.size-1 : 0);
tidyBufFree(&output);
} else {
RETVAL_FALSE;
}
}
}
if (is_file) {
zend_string_release(data);
}
tidyBufFree(errbuf);
efree(errbuf);
tidyRelease(doc);
}
bool nuiHTML::Load(nglIStream& rStream, nglTextEncoding OverrideContentsEncoding, const nglString& rSourceURL)
{
if (!rSourceURL.IsEmpty())
SetSourceURL(rSourceURL);
int res = -1;
nglTextEncoding encoding = eUTF8;
TidyDoc tdoc = NULL;
{
HTMLStream strm(rStream);
tdoc = tidyCreate();
tidyOptSetBool(tdoc, TidyShowMarkup, no);
tidyOptSetBool(tdoc, TidyShowWarnings, no);
tidyOptSetInt(tdoc, TidyShowErrors, 0);
tidyOptSetBool(tdoc, TidyQuiet, yes);
tidySetCharEncoding(tdoc, "utf8");
TidyInputSource source;
tidyInitSource( &source, &strm, &HTMLStream::TidyGetByte, &HTMLStream::TidyUngetByte, &HTMLStream::TidyEOF);
res = tidyParseSource(tdoc, &source);
if ( res >= 0 )
res = tidyCleanAndRepair(tdoc); // Tidy it up!
if ( res >= 0 )
res = tidyRunDiagnostics(tdoc); // Kvetch
if (OverrideContentsEncoding == eEncodingUnknown)
{
nglString encoding_string(GetEncodingString(tidyGetRoot(tdoc)));
//ascii, latin1, raw, utf8, iso2022, mac, win1252, utf16le, utf16be, utf16, big5 shiftjis
encoding = nuiGetTextEncodingFromString(encoding_string);
}
else
{
encoding = OverrideContentsEncoding;
}
}
char* pStr = NULL;
if (encoding != eUTF8)
{
// Release the doc to create a new one
tidyRelease(tdoc);
nglOMemory omem;
rStream.SetPos(0, eStreamFromStart);
rStream.PipeTo(omem);
nglString decoded;
decoded.Import(omem.GetBufferData(), omem.GetSize(), encoding);
pStr = decoded.Export(eUTF8);
nglIMemory imem(pStr, strlen(pStr));
HTMLStream strm(imem);
tdoc = tidyCreate();
tidySetCharEncoding(tdoc, "utf8");
TidyInputSource source;
tidyInitSource( &source, &strm, &HTMLStream::TidyGetByte, &HTMLStream::TidyUngetByte, &HTMLStream::TidyEOF);
res = tidyParseSource(tdoc, &source);
if ( res >= 0 )
res = tidyCleanAndRepair(tdoc); // Tidy it up!
if ( res >= 0 )
res = tidyRunDiagnostics(tdoc); // Kvetch
}
BuildTree(tdoc, tidyGetRoot(tdoc), eUTF8, mComputeStyle);
tidyRelease(tdoc);
if (pStr)
free(pStr);
return res < 2;
}
int CProxyParse::RunFromMem( wxString content )
{
char *pBuffer;
//http://www.51proxied.com/http_non_anonymous.html
//wxString path = wxT("f:/work/windows/wxUrlRefresh/data/最新透明HTTP代理服务器.htm");
//wxString path1 = wxT("f:/work/windows/wxUrlRefresh/data/result.xml");
wxString data_path = wxGetCwd() + "/data/";
wxString path1 = data_path + "_tmp.xml";
if (!wxDirExists(data_path))
wxMkdir(data_path);
pBuffer = (char*)calloc(content.Length()+1, 1);
wxStrncpy(pBuffer, content, content.Len()+1);
wxLogMessage("Run Tidy!");
TidyBuffer output;
TidyBuffer errbuf;
int rc = -1;
Bool ok;
TidyDoc tdoc = tidyCreate(); // Initialize "document"
tidyBufInit( &output );
tidyBufInit( &errbuf );
//printf( "Tidying:\t\%s\\n", input );
tidySetCharEncoding(tdoc, "utf8");
ok = tidyOptSetBool( tdoc, TidyXhtmlOut, yes ); // Convert to XHTML
if ( ok )
rc = tidySetErrorBuffer( tdoc, &errbuf ); // Capture diagnostics
if ( rc >= 0 )
rc = tidyParseString( tdoc, pBuffer ); // Parse the input
if ( rc >= 0 )
rc = tidyCleanAndRepair( tdoc ); // Tidy it up!
if ( rc >= 0 )
rc = tidyRunDiagnostics( tdoc ); // Kvetch
if ( rc > 1 ) // If error, force output.
rc = ( tidyOptSetBool(tdoc, TidyForceOutput, yes) ? rc : -1 );
if ( rc >= 0 )
rc = tidySaveBuffer( tdoc, &output ); // Pretty Print
if ( rc >= 0 )
{
#ifdef _DEBUG
//if ( rc > 0 )
// WriteAllToFile("f:/work/windows/wxUrlRefresh/data/error.xml", (char*)errbuf.bp, errbuf.size);
WriteAllToFile(path1, (char*)output.bp, output.size);
#endif
}
else
wxLogError("tidyFail");
tidyBufFree( &output );
tidyBufFree( &errbuf );
tidyRelease( tdoc );
if (pBuffer) free(pBuffer);
wxLogMessage("Fetch data!");
// 解析数据
TiXmlDocument doc(path1);
if (doc.LoadFile())
{
// root
CTiXmlProxyVistor vistor(&m_array);
TiXmlElement *pRoot = doc.RootElement();
pRoot->Accept(&vistor);
}
else
{
wxLogMessage("shit");
return -2;
}
return 0;
}
void CCFHtmlTidy::SetTidyProp(const std::string& strParam, int nNumValue, const std::string& /*strNumValue*/, const std::string& strTextValue)
{
TidyDoc formatter = tidyImplToDoc(tidy);
if ("axd" == strParam)
{
tidyOptSetBool(formatter, TidyXmlDecl, yes);
}
else if ("axs" == strParam)
{
tidyOptSetBool(formatter, TidyXmlSpace, yes);
}
else if ("aan" == strParam)
{
tidyOptSetBool(formatter, TidyAnchorAsName, yes);
}
else if ("axp" == strParam)
{
tidyOptSetBool(formatter, TidyXmlPIs, yes);
}
else if ("b" == strParam)
{
tidyOptSetBool(formatter, TidyMakeBare, yes);
}
else if ("c" == strParam)
{
tidyOptSetBool(formatter, TidyMakeClean, yes);
}
else if ("diu" == strParam)
{
tidyOptSetBool(formatter, TidyDecorateInferredUL, yes);
}
else if ("dep" == strParam)
{
tidyOptSetBool(formatter, TidyDropEmptyParas, yes);
}
else if ("dft" == strParam)
{
tidyOptSetBool(formatter, TidyDropFontTags, yes);
}
else if ("dpa" == strParam)
{
tidyOptSetBool(formatter, TidyDropPropAttrs, yes);
}
else if ("ebt" == strParam)
{
tidyOptSetBool(formatter, TidyEncloseBlockText, yes);
}
else if ("et" == strParam)
{
tidyOptSetBool(formatter, TidyEncloseBodyText, yes);
}
else if ("ec" == strParam)
{
tidyOptSetBool(formatter, TidyEscapeCdata, yes);
}
else if ("fb" == strParam)
{
tidyOptSetBool(formatter, TidyFixBackslash, yes);
}
else if ("fbc" == strParam)
{
tidyOptSetBool(formatter, TidyFixComments, yes);
}
else if ("fu" == strParam)
{
tidyOptSetBool(formatter, TidyFixUri, yes);
}
else if ("hc" == strParam)
{
tidyOptSetBool(formatter, TidyHideComments, yes);
}
else if ("he" == strParam)
{
tidyOptSetBool(formatter, TidyHideEndTags, yes);
}
else if ("ic" == strParam)
{
tidyOptSetBool(formatter, TidyIndentCdata, yes);
}
else if ("ix" == strParam)
{
tidyOptSetBool(formatter, TidyXmlTags, yes);
}
else if ("jc" == strParam)
{
tidyOptSetBool(formatter, TidyJoinClasses, yes);
}
else if ("js" == strParam)
{
tidyOptSetBool(formatter, TidyJoinStyles, yes);
}
else if ("la" == strParam)
{
tidyOptSetBool(formatter, TidyLiteralAttribs, yes);
}
else if ("le" == strParam)
{
tidyOptSetBool(formatter, TidyLogicalEmphasis, yes);
}
else if ("ll" == strParam)
{
tidyOptSetBool(formatter, TidyLowerLiterals, yes);
}
else if ("n" == strParam)
{
tidyOptSetBool(formatter, TidyNCR, yes);
}
else if ("ne" == strParam)
{
tidyOptSetBool(formatter, TidyNumEntities, yes);
}
else if ("oh" == strParam)
{
tidyOptSetBool(formatter, TidyHtmlOut, yes);
}
else if ("ox" == strParam)
{
tidyOptSetBool(formatter, TidyXhtmlOut, yes);
}
else if ("oxm" == strParam)
{
tidyOptSetBool(formatter, TidyXmlOut, yes);
}
else if ("pe" == strParam)
{
tidyOptSetBool(formatter, TidyPreserveEntities, yes);
}
else if ("qa" == strParam)
{
tidyOptSetBool(formatter, TidyQuoteAmpersand, yes);
}
else if ("qm" == strParam)
{
tidyOptSetBool(formatter, TidyQuoteMarks, yes);
}
else if ("qn" == strParam)
{
tidyOptSetBool(formatter, TidyQuoteNbsp, yes);
}
else if ("rc" == strParam)
{
tidyOptSetBool(formatter, TidyReplaceColor, yes);
}
else if ("ua" == strParam)
{
tidyOptSetBool(formatter, TidyUpperCaseAttrs, yes);
}
else if ("ut" == strParam)
{
tidyOptSetBool(formatter, TidyUpperCaseTags, yes);
}
else if ("wo" == strParam)
{
tidyOptSetBool(formatter, TidyWord2000, yes);
}
else if ("bbb" == strParam)
{
tidyOptSetBool(formatter, TidyBreakBeforeBR, yes);
}
else if ("ia" == strParam)
{
tidyOptSetBool(formatter, TidyIndentAttributes, yes);
}
else if ("m" == strParam)
{
tidyOptSetBool(formatter, TidyShowMarkup, yes);
}
else if ("pw" == strParam)
{
tidyOptSetBool(formatter, TidyPunctWrap, yes);
}
else if ("vs" == strParam)
{
tidyOptSetBool(formatter, TidyVertSpace, yes);
}
else if ("wa" == strParam)
{
tidyOptSetBool(formatter, TidyWrapAsp, yes);
}
else if ("wat" == strParam)
{
tidyOptSetBool(formatter, TidyWrapAttVals, yes);
}
else if ("wj" == strParam)
{
tidyOptSetBool(formatter, TidyWrapJste, yes);
}
else if ("wp" == strParam)
{
tidyOptSetBool(formatter, TidyWrapPhp, yes);
}
else if ("wsl" == strParam)
{
tidyOptSetBool(formatter, TidyWrapScriptlets, yes);
}
else if ("ws" == strParam)
{
tidyOptSetBool(formatter, TidyWrapSection, yes);
}
else if ("ac" == strParam)
{
tidyOptSetBool(formatter, TidyAsciiChars, yes);
}
else if ("sw" == strParam)
{
tidyOptSetBool(formatter, TidyShowWarnings, yes);
}
else if ("fo" == strParam)
{
tidyOptSetBool(formatter, TidyForceOutput, yes);
}
else if ("i" == strParam)
{
tidyOptSetInt(formatter, TidyIndentContent, abs(nNumValue - 2) % 3);
}
else if ("md" == strParam)
{
tidyOptSetInt(formatter, TidyMergeDivs, abs(nNumValue - 2) % 3);
}
else if ("ms" == strParam)
{
tidyOptSetInt(formatter, TidyMergeSpans, abs(nNumValue - 2) % 3);
}
else if ("sbo" == strParam)
{
tidyOptSetInt(formatter, TidyBodyOnly, abs(nNumValue - 2) % 3);
}
else if ("d" == strParam)
{
tidyOptSetInt(formatter, TidyDoctypeMode, nNumValue % 5);
}
else if ("du" == strParam)
{
tidyOptSetValue(formatter, TidyDoctype, strTextValue.c_str());
}
else if ("ra" == strParam)
{
tidyOptSetInt(formatter, TidyDuplicateAttrs, nNumValue % 2);
}
else if ("sa" == strParam)
{
tidyOptSetInt(formatter, TidySortAttributes, nNumValue % 2);
}
else if ("ce" == strParam)
{
tidySetCharEncoding(formatter, TY_(GetEncodingOptNameFromTidyId)(nNumValue));
}
else if ("ie" == strParam)
{
tidySetInCharEncoding(formatter, TY_(GetEncodingOptNameFromTidyId)(nNumValue));
}
else if ("oe" == strParam)
{
tidySetOutCharEncoding(formatter, TY_(GetEncodingOptNameFromTidyId)(nNumValue));
}
else if ("se" == strParam)
{
tidyOptSetInt(formatter, TidyShowErrors, nNumValue);
}
else if ("is" == strParam)
{
tidyOptSetInt(formatter, TidyIndentSpaces, nNumValue);
}
else if ("ts" == strParam)
{
tidyOptSetInt(formatter, TidyTabSize, nNumValue);
}
else if ("w" == strParam)
{
tidyOptSetInt(formatter, TidyWrapLen, nNumValue);
}
else if ("at" == strParam)
{
tidyOptSetValue(formatter, TidyAltText, strTextValue.c_str());
}
else if ("cp" == strParam)
{
tidyOptSetValue(formatter, TidyCSSPrefix, strTextValue.c_str());
}
else if ("nbt" == strParam)
{
tidyOptSetValue(formatter, TidyBlockTags, strTextValue.c_str());
}
else if ("net" == strParam)
{
tidyOptSetValue(formatter, TidyEmptyTags, strTextValue.c_str());
}
else if ("nit" == strParam)
{
tidyOptSetValue(formatter, TidyInlineTags, strTextValue.c_str());
}
else if ("npt" == strParam)
{
tidyOptSetValue(formatter, TidyPreTags, strTextValue.c_str());
}
}
int main( int argc, char** argv )
{
ctmbstr prog = argv[0];
ctmbstr cfgfil = NULL, errfil = NULL, htmlfil = NULL;
TidyDoc tdoc = tidyCreate();
int status = 0;
uint contentErrors = 0;
uint contentWarnings = 0;
uint accessWarnings = 0;
errout = stderr; /* initialize to stderr */
status = 0;
#ifdef TIDY_CONFIG_FILE
if ( tidyFileExists( tdoc, TIDY_CONFIG_FILE) )
{
status = tidyLoadConfig( tdoc, TIDY_CONFIG_FILE );
if ( status != 0 )
fprintf(errout, "Loading config file \"%s\" failed, err = %d\n", TIDY_CONFIG_FILE, status);
}
#endif /* TIDY_CONFIG_FILE */
/* look for env var "HTML_TIDY" */
/* then for ~/.tidyrc (on platforms defining $HOME) */
if ( (cfgfil = getenv("HTML_TIDY")) != NULL )
{
status = tidyLoadConfig( tdoc, cfgfil );
if ( status != 0 )
fprintf(errout, "Loading config file \"%s\" failed, err = %d\n", cfgfil, status);
}
#ifdef TIDY_USER_CONFIG_FILE
else if ( tidyFileExists( tdoc, TIDY_USER_CONFIG_FILE) )
{
status = tidyLoadConfig( tdoc, TIDY_USER_CONFIG_FILE );
if ( status != 0 )
fprintf(errout, "Loading config file \"%s\" failed, err = %d\n", TIDY_USER_CONFIG_FILE, status);
}
#endif /* TIDY_USER_CONFIG_FILE */
/* read command line */
while ( argc > 0 )
{
if (argc > 1 && argv[1][0] == '-')
{
/* support -foo and --foo */
ctmbstr arg = argv[1] + 1;
if ( strcasecmp(arg, "xml") == 0)
tidyOptSetBool( tdoc, TidyXmlTags, yes );
else if ( strcasecmp(arg, "asxml") == 0 ||
strcasecmp(arg, "asxhtml") == 0 )
{
tidyOptSetBool( tdoc, TidyXhtmlOut, yes );
}
else if ( strcasecmp(arg, "ashtml") == 0 )
tidyOptSetBool( tdoc, TidyHtmlOut, yes );
else if ( strcasecmp(arg, "indent") == 0 )
{
tidyOptSetInt( tdoc, TidyIndentContent, TidyAutoState );
if ( tidyOptGetInt(tdoc, TidyIndentSpaces) == 0 )
tidyOptResetToDefault( tdoc, TidyIndentSpaces );
}
else if ( strcasecmp(arg, "omit") == 0 )
tidyOptSetBool( tdoc, TidyHideEndTags, yes );
else if ( strcasecmp(arg, "upper") == 0 )
tidyOptSetBool( tdoc, TidyUpperCaseTags, yes );
else if ( strcasecmp(arg, "clean") == 0 )
tidyOptSetBool( tdoc, TidyMakeClean, yes );
else if ( strcasecmp(arg, "bare") == 0 )
tidyOptSetBool( tdoc, TidyMakeBare, yes );
else if ( strcasecmp(arg, "raw") == 0 ||
strcasecmp(arg, "ascii") == 0 ||
strcasecmp(arg, "latin0") == 0 ||
strcasecmp(arg, "latin1") == 0 ||
strcasecmp(arg, "utf8") == 0 ||
#ifndef NO_NATIVE_ISO2022_SUPPORT
strcasecmp(arg, "iso2022") == 0 ||
#endif
#if SUPPORT_UTF16_ENCODINGS
strcasecmp(arg, "utf16le") == 0 ||
strcasecmp(arg, "utf16be") == 0 ||
strcasecmp(arg, "utf16") == 0 ||
#endif
#if SUPPORT_ASIAN_ENCODINGS
strcasecmp(arg, "shiftjis") == 0 ||
strcasecmp(arg, "big5") == 0 ||
#endif
strcasecmp(arg, "mac") == 0 ||
strcasecmp(arg, "win1252") == 0 ||
strcasecmp(arg, "ibm858") == 0 )
{
tidySetCharEncoding( tdoc, arg );
}
else if ( strcasecmp(arg, "numeric") == 0 )
tidyOptSetBool( tdoc, TidyNumEntities, yes );
else if ( strcasecmp(arg, "modify") == 0 ||
strcasecmp(arg, "change") == 0 || /* obsolete */
strcasecmp(arg, "update") == 0 ) /* obsolete */
{
tidyOptSetBool( tdoc, TidyWriteBack, yes );
}
else if ( strcasecmp(arg, "errors") == 0 )
tidyOptSetBool( tdoc, TidyShowMarkup, no );
else if ( strcasecmp(arg, "quiet") == 0 )
tidyOptSetBool( tdoc, TidyQuiet, yes );
else if ( strcasecmp(arg, "help") == 0 ||
strcasecmp(arg, "h") == 0 || *arg == '?' )
{
help( prog );
tidyRelease( tdoc );
return 0; /* success */
}
else if ( strcasecmp(arg, "xml-help") == 0)
{
xml_help( );
tidyRelease( tdoc );
return 0; /* success */
}
else if ( strcasecmp(arg, "help-config") == 0 )
{
optionhelp( tdoc );
tidyRelease( tdoc );
return 0; /* success */
}
else if ( strcasecmp(arg, "xml-config") == 0 )
{
XMLoptionhelp( tdoc );
tidyRelease( tdoc );
return 0; /* success */
}
else if ( strcasecmp(arg, "show-config") == 0 )
{
optionvalues( tdoc );
tidyRelease( tdoc );
return 0; /* success */
}
else if ( strcasecmp(arg, "config") == 0 )
{
if ( argc >= 3 )
{
ctmbstr post;
tidyLoadConfig( tdoc, argv[2] );
/* Set new error output stream if setting changed */
post = tidyOptGetValue( tdoc, TidyErrFile );
if ( post && (!errfil || !samefile(errfil, post)) )
{
errfil = post;
errout = tidySetErrorFile( tdoc, post );
}
--argc;
++argv;
}
}
#if SUPPORT_ASIAN_ENCODINGS
else if ( strcasecmp(arg, "language") == 0 ||
strcasecmp(arg, "lang") == 0 )
{
if ( argc >= 3 )
{
tidyOptSetValue( tdoc, TidyLanguage, argv[2] );
--argc;
++argv;
}
}
#endif
else if ( strcasecmp(arg, "output") == 0 ||
strcasecmp(arg, "-output-file") == 0 ||
strcasecmp(arg, "o") == 0 )
{
if ( argc >= 3 )
{
tidyOptSetValue( tdoc, TidyOutFile, argv[2] );
--argc;
++argv;
}
}
else if ( strcasecmp(arg, "file") == 0 ||
strcasecmp(arg, "-file") == 0 ||
strcasecmp(arg, "f") == 0 )
{
if ( argc >= 3 )
{
errfil = argv[2];
errout = tidySetErrorFile( tdoc, errfil );
--argc;
++argv;
}
}
else if ( strcasecmp(arg, "wrap") == 0 ||
strcasecmp(arg, "-wrap") == 0 ||
strcasecmp(arg, "w") == 0 )
{
if ( argc >= 3 )
{
uint wraplen = 0;
int nfields = sscanf( argv[2], "%u", &wraplen );
tidyOptSetInt( tdoc, TidyWrapLen, wraplen );
if (nfields > 0)
{
--argc;
++argv;
}
}
}
else if ( strcasecmp(arg, "version") == 0 ||
strcasecmp(arg, "-version") == 0 ||
strcasecmp(arg, "v") == 0 )
{
version();
tidyRelease( tdoc );
return 0; /* success */
}
else if ( strncmp(argv[1], "--", 2 ) == 0)
{
if ( tidyOptParseValue(tdoc, argv[1]+2, argv[2]) )
{
/* Set new error output stream if setting changed */
ctmbstr post = tidyOptGetValue( tdoc, TidyErrFile );
if ( post && (!errfil || !samefile(errfil, post)) )
{
errfil = post;
errout = tidySetErrorFile( tdoc, post );
}
++argv;
--argc;
}
}
#if SUPPORT_ACCESSIBILITY_CHECKS
else if ( strcasecmp(arg, "access") == 0 )
{
if ( argc >= 3 )
{
uint acclvl = 0;
int nfields = sscanf( argv[2], "%u", &acclvl );
tidyOptSetInt( tdoc, TidyAccessibilityCheckLevel, acclvl );
if (nfields > 0)
{
--argc;
++argv;
}
}
}
#endif
else
{
uint c;
ctmbstr s = argv[1];
while ( (c = *++s) != '\0' )
{
switch ( c )
{
case 'i':
tidyOptSetInt( tdoc, TidyIndentContent, TidyAutoState );
if ( tidyOptGetInt(tdoc, TidyIndentSpaces) == 0 )
tidyOptResetToDefault( tdoc, TidyIndentSpaces );
break;
/* Usurp -o for output file. Anyone hiding end tags?
case 'o':
tidyOptSetBool( tdoc, TidyHideEndTags, yes );
break;
*/
case 'u':
tidyOptSetBool( tdoc, TidyUpperCaseTags, yes );
break;
case 'c':
tidyOptSetBool( tdoc, TidyMakeClean, yes );
break;
case 'b':
tidyOptSetBool( tdoc, TidyMakeBare, yes );
break;
case 'n':
tidyOptSetBool( tdoc, TidyNumEntities, yes );
break;
case 'm':
tidyOptSetBool( tdoc, TidyWriteBack, yes );
break;
case 'e':
tidyOptSetBool( tdoc, TidyShowMarkup, no );
break;
case 'q':
tidyOptSetBool( tdoc, TidyQuiet, yes );
break;
default:
unknownOption( c );
break;
}
}
}
--argc;
++argv;
continue;
}
if ( argc > 1 )
{
htmlfil = argv[1];
if ( tidyOptGetBool(tdoc, TidyEmacs) )
tidyOptSetValue( tdoc, TidyEmacsFile, htmlfil );
status = tidyParseFile( tdoc, htmlfil );
}
else
{
htmlfil = "stdin";
status = tidyParseStdin( tdoc );
}
if ( status >= 0 )
status = tidyCleanAndRepair( tdoc );
if ( status >= 0 )
status = tidyRunDiagnostics( tdoc );
if ( status > 1 ) /* If errors, do we want to force output? */
status = ( tidyOptGetBool(tdoc, TidyForceOutput) ? status : -1 );
if ( status >= 0 && tidyOptGetBool(tdoc, TidyShowMarkup) )
{
if ( tidyOptGetBool(tdoc, TidyWriteBack) && argc > 1 )
status = tidySaveFile( tdoc, htmlfil );
else
{
ctmbstr outfil = tidyOptGetValue( tdoc, TidyOutFile );
if ( outfil )
status = tidySaveFile( tdoc, outfil );
else
status = tidySaveStdout( tdoc );
}
}
contentErrors += tidyErrorCount( tdoc );
contentWarnings += tidyWarningCount( tdoc );
accessWarnings += tidyAccessWarningCount( tdoc );
--argc;
++argv;
if ( argc <= 1 )
break;
}
if (!tidyOptGetBool(tdoc, TidyQuiet) &&
errout == stderr && !contentErrors)
fprintf(errout, "\n");
if (contentErrors + contentWarnings > 0 &&
!tidyOptGetBool(tdoc, TidyQuiet))
tidyErrorSummary(tdoc);
if (!tidyOptGetBool(tdoc, TidyQuiet))
tidyGeneralInfo(tdoc);
/* called to free hash tables etc. */
tidyRelease( tdoc );
/* return status can be used by scripts */
if ( contentErrors > 0 )
return 2;
if ( contentWarnings > 0 )
return 1;
/* 0 signifies all is ok */
return 0;
}
void FetchTaskHandler(const FetchTaskMessage &message, const Theron::Address from)
{
std::cout<<"get data.................."<<std::endl;
std::string url=message.fi->url;
char curl_errbuf[CURL_ERROR_SIZE];
CURL *curl = curl_easy_init();
int err;
fetch::FetchResult *result=new fetch::FetchResult();
fetch::FetchInfo fi=*(message.fi);
delete message.fi;
result->type=fetch::UNKNOWN;
result->url=fi.url;
result->pathList=fi.pathList;
result->attMap=fi.attMap;
int errCode=0;
if(curl!=NULL) {
curl_easy_setopt(curl, CURLOPT_URL,url.c_str());
curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errbuf);
// curl_easy_setopt(curl, CURLOPT_NOPROGRESS, 0L);
// curl_easy_setopt(curl, CURLOPT_VERBOSE, 1L);
curl_easy_setopt(curl, CURLOPT_CONNECTTIMEOUT, 10);
curl_easy_setopt(curl, CURLOPT_NOSIGNAL,1);
curl_easy_setopt(curl, CURLOPT_TIMEOUT ,60);
curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, fetch_write);
std::stringstream iss;
curl_easy_setopt(curl, CURLOPT_WRITEDATA, &iss);
err=curl_easy_perform(curl);
if ( !err )
{
std::map<std::string,std::string>::iterator efit=fi.attMap.find("encode");
char *resStr=new char[iss.str().length()*3];
memset(resStr,0,iss.str().length()*3);
if(efit!=fi.attMap.end())
{
UErrorCode error = U_ZERO_ERROR;
ucnv_convert("UTF-8",efit->second.c_str(),resStr, iss.str().length()*3, iss.str().c_str(), iss.str().length(), &error );
}else
{
strcpy(resStr,iss.str().c_str());
}
TidyDoc tdoc;
// TidyBuffer tidy_errbuf = {0};
// TidyBuffer docbuf = {0};
tdoc = tidyCreate();
tidyOptSetInt(tdoc, TidyWrapLen, 4096);
// tidySetErrorBuffer( tdoc, &tidy_errbuf );
tidyOptSetBool( tdoc, TidyXmlOut, yes );
tidyOptSetBool(tdoc, TidyQuoteNbsp, no);
tidyOptSetBool(tdoc, TidyForceOutput, yes); /* try harder */
// tidyOptSetBool( tdoc, TidyXmlDecl, yes );
tidyOptSetBool(tdoc, TidyQuiet, yes);
tidyOptSetBool(tdoc, TidyShowWarnings, no);
tidyOptSetValue(tdoc,TidyDoctype,"omit");
tidyOptSetBool(tdoc, TidyFixBackslash, yes);
tidyOptSetBool(tdoc, TidyMark, no);
tidySetCharEncoding(tdoc,"utf8");
// tidyBufInit(&docbuf);
// err = tidyParseBuffer(tdoc, &docbuf);
err = tidyParseString(tdoc, resStr);
if ( err >= 0 ) {
err = tidyCleanAndRepair(tdoc); /* fix any problems */
if ( err >= 0 ) {
// err=tidyRunDiagnostics(tdoc); /* load tidy error buffer */
// if ( err >= 0 ) {
// std::cout<<"tidy error:"<<tidy_errbuf.bp<<std::endl; /* show errors */
TidyBuffer outbuf = {0};
tidyBufInit(&outbuf);
tidySaveBuffer( tdoc, &outbuf );
std::stringstream hss;
hss<<(char*)outbuf.bp;
tidyBufFree(&outbuf);
result->result=hss.str();
// }
// else
// {
// errCode=-5;
// }
}else
{
errCode=-4;
}
}else
{
errCode=-3;
}
// tidyBufFree(&tidy_errbuf);
// tidyBufFree(&docbuf);
tidyRelease(tdoc);
delete [] resStr;
}else
{
errCode=-2;
}
}else
{
errCode=-1;
}
if(errCode<0)
{
std::stringstream ess;
ess<<errCode;
result->type=fetch::ERROR;
result->result=ess.str();
}
curl_easy_cleanup(curl);
Send(FetchResultMessage(result), from);
}
int CCaHtmlParse::ParseCaHtmlFlights(std::list<SCaLowPriceFlightDetail*> & listFlight, const std::string& strHtmlData, const CStringA & straDCode, const CStringA & straACode, const SCaLowPriceFlightInfo* pLowPriceFlightInfo)
{
TidyDoc doc = tidyCreate();
tidySetCharEncoding(doc,"raw");
tidyParseString(doc,strHtmlData.c_str());
TidyNode tnRoot = tidyGetRoot(doc);
TidyNode tFlightTab;
TidyNode tdChild;
int nIndexTd = 0;
CTime tCurrent = CTime::GetCurrentTime();
SCaLowPriceFlightDetail *pfindFlight = NULL;
if (FindNode(tnRoot,"class","CA_table mt_10 clear",tFlightTab))
{
//循环解析结算价,tblPolicy下的每一个子节点即为一条结算价信息
TidyNode trFlight;
int nIndexTr = 0;
BOOL bValid = FALSE;
CStringA straDPortCode = straDCode;
CStringA straAPortCode = straACode;
CStringA straFlightNo("");
CStringA straFlightStartDate("");
CStringA straSaleEndDate("");
CStringA straSaleEndTime("");
CStringA straFlightStartTime("");
UINT uPrice = 0;
UINT uRemainTicket = 0;
for ( trFlight = tidyGetChild(tFlightTab); trFlight; trFlight = tidyGetNext(trFlight) )
{
if (0 == nIndexTr)//跳过表头
{
nIndexTr++;
continue;
}
nIndexTd = 0;
bValid = FALSE;
straFlightNo = "";
straFlightStartDate = "";
straSaleEndDate = "";
straSaleEndTime = "";
straFlightStartTime = "";
uPrice = 0;
uRemainTicket = 0;
for ( tdChild = tidyGetChild(trFlight); tdChild; tdChild = tidyGetNext(tdChild) )
{
switch(nIndexTd)
{
case 0:
{
//选择，是否为disabled
bValid = __IsFlightValid(tdChild);
TRACE(_T("Flight valid:%d-"), bValid);
}
break;
case 1:
{
//日期/航班号
//dumpNode(tdChild, 0);
//TRACE(_T("\r\n"));
__GetFlightNoAndFlightStartDate(straFlightNo, straFlightStartDate, doc, tdChild);
TRACE("date:%s, no:%s-", straFlightStartDate, straFlightNo);
//TRACE("%s\r\n", GetNodeContent(doc, tdChild));
}
break;
case 2:
{
//起降时间
//dumpNode(tdChild, 0);
//TRACE(_T("\r\n"));
//TRACE("%s\r\n", GetNodeContent(doc, tdChild));
__GetFlightStartTime(straFlightStartTime, doc, tdChild);
}
break;
case 3:
{
//机场
//dumpNode(tdChild, 0);
//TRACE(_T("\r\n"));
//TRACE("%s\r\n", GetNodeContent(doc, tdChild));
if (__IsTwoAirPort(straDCode, straACode))
{
__GetAirPortCode(straDPortCode, straAPortCode, doc, tdChild);
if(straDPortCode.IsEmpty())
straDPortCode = straDCode;
if(straAPortCode.IsEmpty())
straAPortCode = straACode;
TRACE("%s->%s-", straDPortCode, straAPortCode);
}
}
break;
case 4:
{
//销售结束日期,时间
//dumpNode(tdChild, 0);
//TRACE(_T("\r\n"));
//TRACE("%s\r\n", GetNodeContent(doc, tdChild));
__GetSaleEndDate(straSaleEndDate, straSaleEndTime, doc, tdChild);
TRACE("sale end date:%s, %s-", straSaleEndDate, straSaleEndTime);
}
break;
case 5:
{
//团购价
//dumpNode(tdChild, 0);
//TRACE(_T("\r\n"));
//TRACE("%s\r\n", GetNodeContent(doc, tdChild));
//CStringA straSetPrice = GetNodeContent(doc, tdChild);
//double fSetPrice = atof(straSetPrice.GetBuffer(0));
//straSetPrice.ReleaseBuffer();
//tidyRelease(doc);
//return fSetPrice;
__GetPriceAndRamainTicket(&uPrice, &uRemainTicket, doc, tdChild);
TRACE("price:%d, remain %d seats", uPrice, uRemainTicket);
}
break;
}
nIndexTd++;
}
TRACE(_T("\r\n"));
//截至日期之后的航班不抓取
//得到起飞日期
int nFlightStartYear = 2014;
int nFlightStartMonth = 12;
int nFlightStartDay = 12;
GetYearMonthDay(straFlightStartDate, &nFlightStartYear, &nFlightStartMonth, &nFlightStartDay);
CTime tStart(nFlightStartYear, nFlightStartMonth, nFlightStartDay, 0, 0, 0);
//if (!m_bGetAllCaTuanFlight)
//{
// if (tStart > m_tGetEndTime)
// continue;
//}
//
//double d6 = pLowPriceFlightInfo->iMinHangPrice * 0.6;
//UINT u6 = (UINT)d6;
////6折以上普通团购退改签要收费（低价申请不受限制），所以不上
//if (uPrice > d6 && CA_TUAN_PRODUCT == pLowPriceFlightInfo->iProductType)
//{
// bValid = FALSE;
// uRemainTicket = 0;
// continue;
//}
//相同日期、时间、班次的航班，只取最低价
BOOL bFind = __findCaFlight(&pfindFlight, straFlightStartDate, straDPortCode, straAPortCode, straFlightNo, listFlight);
if (bFind)
{
int nCurPrice = (int)uPrice;
//当前解析出的这个比上次解析出的便宜
if(pfindFlight->nPrice > nCurPrice)
{
if (uRemainTicket > m_nMinTicketWarnNum)
{
//当前票的数量充足时，用当前票的数量更新上次解析出的数量
pfindFlight->nRemainSeat = uRemainTicket;
pfindFlight->nPrice = nCurPrice;
pfindFlight = NULL;
}
}
else //(pfindFlight->nPrice <= nCurPrice)
{
if(pfindFlight->nRemainSeat <= m_nMinTicketWarnNum)
{
pfindFlight->nRemainSeat = uRemainTicket;
pfindFlight->nPrice = nCurPrice;
pfindFlight = NULL;
}
}
continue;
}
//保存解析出来的航班信息,调用者负责释放内存
if (bValid)
{
SCaLowPriceFlightDetail* pDetail = new SCaLowPriceFlightDetail;
pDetail->straCompany = "CA";
pDetail->straFromCityCode = straDPortCode;
pDetail->straToCityCode = straAPortCode;
pDetail->straFlightNo = straFlightNo;
pDetail->straFromDate = straFlightStartDate;
//由于携程订单进入需要一定的时间，国航下班16：00下班，所以当天的票,第2天12:00之前的票，销售结束时间提前30分钟，
//取销售间隔
int nSaleEndYear = 2014;
int nSaleEndMonth = 12;
int nSaleEndDay = 12;
GetYearMonthDay(straSaleEndDate, &nSaleEndYear, &nSaleEndMonth, &nSaleEndDay);
int nSaleEndHour = 12;
int nSaleEndMin = 0;
GetHourMinSec(straSaleEndTime, &nSaleEndHour, &nSaleEndMin);
CTime tSaleEndDate(nSaleEndYear, nSaleEndMonth, nSaleEndDay, nSaleEndHour, nSaleEndMin, 0);
CTimeSpan tSpan = tSaleEndDate - tCurrent;
//end 取销售间隔
//得到起飞时间
int nFlightStartHour = 12;
int nFlightStartMin = 0;
GetHourMinSec(straFlightStartTime, &nFlightStartHour, &nFlightStartMin);
CTime tFlightStartTime(nFlightStartYear, nFlightStartMonth, nFlightStartDay, nFlightStartHour, nFlightStartMin, 0);
CTime tTimeKey(nFlightStartYear, nFlightStartMonth, nFlightStartDay, 12, 0, 0);
//end 得到起飞时间
//今明两天的、起飞时间在12点之前、且是低价申请的，销售结束时间为 前一天的官网销售结束的前30分钟
if ((CA_TUAN_LOW_PRICE_APPLY_PRODUT == pLowPriceFlightInfo->iProductType) && (1 == tSpan.GetDays()))//明天的的低价申请
{
if(tFlightStartTime <= tTimeKey)//明天12起飞的低价申请, 今天下午3：25前有效（国航4点下班）
{
pDetail->straSaleEndDate.Format("%d-%02d-%02d", tCurrent.GetYear(), tCurrent.GetMonth(), tCurrent.GetDay());
CTime tSaleEnd(tCurrent.GetYear(), tCurrent.GetMonth(), tCurrent.GetDay(), 15, 25, 0);
pDetail->straSaleEndTime.Format("%02d:%02d:%02d", tSaleEnd.GetHour(), tSaleEnd.GetMinute(), 0);
}
else//明天12后起飞的低价申请，明早可以出票
{
pDetail->straSaleEndDate = straSaleEndDate;
pDetail->straSaleEndTime.Format("%02d:%02d:%02d", nSaleEndHour, nSaleEndMin, 0);
}
}
else if ((CA_TUAN_LOW_PRICE_APPLY_PRODUT == pLowPriceFlightInfo->iProductType) && (tSpan.GetDays() < 1))//今天的的低价申请，今天下午3：30前有效（国航4点下班）
{
pDetail->straSaleEndDate.Format("%d-%02d-%02d", tCurrent.GetYear(), tCurrent.GetMonth(), tCurrent.GetDay());
CTime tSaleEnd(tCurrent.GetYear(), tCurrent.GetMonth(), tCurrent.GetDay(), 15, 30, 0);
pDetail->straSaleEndTime.Format("%02d:%02d:%02d", tSaleEnd.GetHour(), tSaleEnd.GetMinute(), 0);
}
else//普通团购，后天及以后的低价申请
{
pDetail->straSaleEndDate = straSaleEndDate;
pDetail->straSaleEndTime.Format("%02d:%02d:%02d", nSaleEndHour, nSaleEndMin, 0);
}
//政策销售时间到，删除政策
GetYearMonthDay(pDetail->straSaleEndDate, &nSaleEndYear, &nSaleEndMonth, &nSaleEndDay);
int nSaleEndSec = 0;
GetHourMinSec(pDetail->straSaleEndTime, &nSaleEndHour, &nSaleEndMin, &nSaleEndSec);
CTime tPolicyDeleteTime(nSaleEndYear, nSaleEndMonth, nSaleEndDay, nSaleEndHour, nSaleEndMin, nSaleEndSec);
if (tCurrent >= tPolicyDeleteTime)
uRemainTicket = 0;
pDetail->nPrice = uPrice;
pDetail->nProductId = pLowPriceFlightInfo->iProductId;
pDetail->nRemainSeat = uRemainTicket;
pDetail->nProductType = pLowPriceFlightInfo->iProductType;
listFlight.push_back(pDetail);
}
}
}
tidyRelease(doc);
return -1.0;
}
void HTidyInterface::SetTidyControl( TidyDoc tdoc, LPCTSTR lpszTidy, int nPos, int nSize )
{
if (lpszTidy[nPos] != '-' || nSize < 2)
{
return;
}
int nNumValue = nSize;
for (int i = nPos + 1; i < nPos + nSize; ++i)
{
if (!_istalpha(lpszTidy[i]))
{
nNumValue = i - nPos;
break;
}
}
CString strParam(lpszTidy + nPos + 1, nNumValue - 1);
if (nNumValue != nSize)
{
CString strNum(lpszTidy + nPos + nNumValue, nSize - nNumValue);
nNumValue = _ttoi(strNum);
}
else
{
nNumValue = 0;
}
CString strNumValue;
strNumValue.Format(_T("%d"), nNumValue);
CString strNothing;
strNothing.Empty();
if (_T("axd") == strParam)
{
tidyOptSetBool(tdoc, TidyXmlDecl, yes);
}
else if (_T("axs") == strParam)
{
tidyOptSetBool(tdoc, TidyXmlSpace, yes);
}
else if (_T("aan") == strParam)
{
tidyOptSetBool(tdoc, TidyAnchorAsName, yes);
}
else if (_T("axp") == strParam)
{
tidyOptSetBool(tdoc, TidyXmlPIs, yes);
}
else if (_T("b") == strParam)
{
tidyOptSetBool(tdoc, TidyMakeBare, yes);
}
else if (_T("c") == strParam)
{
tidyOptSetBool(tdoc, TidyMakeClean, yes);
}
else if (_T("diu") == strParam)
{
tidyOptSetBool(tdoc, TidyDecorateInferredUL, yes);
}
else if (_T("dep") == strParam)
{
tidyOptSetBool(tdoc, TidyDropEmptyParas, yes);
}
else if (_T("dft") == strParam)
{
tidyOptSetBool(tdoc, TidyDropFontTags, yes);
}
else if (_T("dpa") == strParam)
{
tidyOptSetBool(tdoc, TidyDropPropAttrs, yes);
}
else if (_T("ebt") == strParam)
{
tidyOptSetBool(tdoc, TidyEncloseBlockText, yes);
}
else if (_T("et") == strParam)
{
tidyOptSetBool(tdoc, TidyEncloseBodyText, yes);
}
else if (_T("ec") == strParam)
{
tidyOptSetBool(tdoc, TidyEscapeCdata, yes);
}
else if (_T("fb") == strParam)
{
tidyOptSetBool(tdoc, TidyFixBackslash, yes);
}
else if (_T("fbc") == strParam)
{
tidyOptSetBool(tdoc, TidyFixComments, yes);
}
else if (_T("fu") == strParam)
{
tidyOptSetBool(tdoc, TidyFixUri, yes);
}
else if (_T("hc") == strParam)
{
tidyOptSetBool(tdoc, TidyHideComments, yes);
}
else if (_T("he") == strParam)
{
tidyOptSetBool(tdoc, TidyHideEndTags, yes);
}
else if (_T("ic") == strParam)
{
tidyOptSetBool(tdoc, TidyIndentCdata, yes);
}
else if (_T("ix") == strParam)
{
tidyOptSetBool(tdoc, TidyXmlTags, yes);
}
else if (_T("jc") == strParam)
{
tidyOptSetBool(tdoc, TidyJoinClasses, yes);
}
else if (_T("js") == strParam)
{
tidyOptSetBool(tdoc, TidyJoinStyles, yes);
}
else if (_T("la") == strParam)
{
tidyOptSetBool(tdoc, TidyLiteralAttribs, yes);
}
else if (_T("le") == strParam)
{
tidyOptSetBool(tdoc, TidyLogicalEmphasis, yes);
}
else if (_T("ll") == strParam)
{
tidyOptSetBool(tdoc, TidyLowerLiterals, yes);
}
else if (_T("n") == strParam)
{
tidyOptSetBool(tdoc, TidyNCR, yes);
}
else if (_T("ne") == strParam)
{
tidyOptSetBool(tdoc, TidyNumEntities, yes);
}
else if (_T("oh") == strParam)
{
tidyOptSetBool(tdoc, TidyHtmlOut, yes);
}
else if (_T("ox") == strParam)
{
tidyOptSetBool(tdoc, TidyXhtmlOut, yes);
}
else if (_T("oxm") == strParam)
{
tidyOptSetBool(tdoc, TidyXmlOut, yes);
}
else if (_T("pe") == strParam)
{
tidyOptSetBool(tdoc, TidyPreserveEntities, yes);
}
else if (_T("qa") == strParam)
{
tidyOptSetBool(tdoc, TidyQuoteAmpersand, yes);
}
else if (_T("qm") == strParam)
{
tidyOptSetBool(tdoc, TidyQuoteMarks, yes);
}
else if (_T("qn") == strParam)
{
tidyOptSetBool(tdoc, TidyQuoteNbsp, yes);
}
else if (_T("rc") == strParam)
{
tidyOptSetBool(tdoc, TidyReplaceColor, yes);
}
else if (_T("ua") == strParam)
{
tidyOptSetBool(tdoc, TidyUpperCaseAttrs, yes);
}
else if (_T("ut") == strParam)
{
tidyOptSetBool(tdoc, TidyUpperCaseTags, yes);
}
else if (_T("wo") == strParam)
{
tidyOptSetBool(tdoc, TidyWord2000, yes);
}
else if (_T("bbb") == strParam)
{
tidyOptSetBool(tdoc, TidyBreakBeforeBR, yes);
}
else if (_T("ia") == strParam)
{
tidyOptSetBool(tdoc, TidyIndentAttributes, yes);
}
else if (_T("m") == strParam)
{
tidyOptSetBool(tdoc, TidyShowMarkup, yes);
}
else if (_T("pw") == strParam)
{
tidyOptSetBool(tdoc, TidyPunctWrap, yes);
}
else if (_T("vs") == strParam)
{
tidyOptSetBool(tdoc, TidyVertSpace, yes);
}
else if (_T("wa") == strParam)
{
tidyOptSetBool(tdoc, TidyWrapAsp, yes);
}
else if (_T("wat") == strParam)
{
tidyOptSetBool(tdoc, TidyWrapAttVals, yes);
}
else if (_T("wj") == strParam)
{
tidyOptSetBool(tdoc, TidyWrapJste, yes);
}
else if (_T("wp") == strParam)
{
tidyOptSetBool(tdoc, TidyWrapPhp, yes);
}
else if (_T("wsl") == strParam)
{
tidyOptSetBool(tdoc, TidyWrapScriptlets, yes);
}
else if (_T("ws") == strParam)
{
tidyOptSetBool(tdoc, TidyWrapSection, yes);
}
else if (_T("ac") == strParam)
{
tidyOptSetBool(tdoc, TidyAsciiChars, yes);
}
else if (_T("sw") == strParam)
{
tidyOptSetBool(tdoc, TidyShowWarnings, yes);
}
else if (_T("fo") == strParam)
{
tidyOptSetBool(tdoc, TidyForceOutput, yes);
}
else if (_T("i") == strParam)
{
tidyOptSetInt(tdoc, TidyIndentContent, abs(nNumValue - 2) % 3);
}
else if (_T("md") == strParam)
{
tidyOptSetInt(tdoc, TidyMergeDivs, abs(nNumValue - 2) % 3);
}
else if (_T("ms") == strParam)
{
tidyOptSetInt(tdoc, TidyMergeSpans, abs(nNumValue - 2) % 3);
}
else if (_T("sbo") == strParam)
{
tidyOptSetInt(tdoc, TidyBodyOnly, abs(nNumValue - 2) % 3);
}
else if (_T("d") == strParam)
{
tidyOptSetInt(tdoc, TidyDoctypeMode, nNumValue % 5);
}
else if (_T("ra") == strParam)
{
tidyOptSetInt(tdoc, TidyDuplicateAttrs, nNumValue % 2);
}
else if (_T("sa") == strParam)
{
tidyOptSetInt(tdoc, TidySortAttributes, nNumValue % 2);
}
else if (_T("ce") == strParam)
{
tidySetCharEncoding(tdoc, GetEncodeByIndex(nNumValue));
}
else if (_T("ie") == strParam)
{
tidySetInCharEncoding(tdoc, GetEncodeByIndex(nNumValue));
}
else if (_T("oe") == strParam)
{
tidySetOutCharEncoding(tdoc, GetEncodeByIndex(nNumValue));
}
else if (_T("se") == strParam)
{
tidyOptSetInt(tdoc, TidyShowErrors, nNumValue);
}
else if (_T("is") == strParam)
{
tidyOptSetInt(tdoc, TidyIndentSpaces, nNumValue);
}
else if (_T("ts") == strParam)
{
tidyOptSetInt(tdoc, TidyTabSize, nNumValue);
}
else if (_T("w") == strParam)
{
tidyOptSetInt(tdoc, TidyWrapLen, nNumValue);
}
else if (_T("at") == strParam)
{
tidyOptSetValue(tdoc, TidyAltText, CT2A(m_onlyHtmlXml == SYN_HTML?g_GlobalTidy.m_TidyHtml_at:strNothing));
}
else if (_T("cp") == strParam)
{
tidyOptSetValue(tdoc, TidyCSSPrefix, CT2A(m_onlyHtmlXml == SYN_HTML?g_GlobalTidy.m_TidyHtml_cp:strNothing));
}
else if (_T("nbt") == strParam)
{
tidyOptSetValue(tdoc, TidyBlockTags, CT2A(m_onlyHtmlXml == SYN_HTML?g_GlobalTidy.m_TidyHtml_nbt:strNothing));
}
else if (_T("net") == strParam)
{
tidyOptSetValue(tdoc, TidyEmptyTags, CT2A(m_onlyHtmlXml == SYN_HTML?g_GlobalTidy.m_TidyHtml_net:strNothing));
}
else if (_T("nit") == strParam)
{
tidyOptSetValue(tdoc, TidyInlineTags, CT2A(m_onlyHtmlXml == SYN_HTML?g_GlobalTidy.m_TidyHtml_nit:strNothing));
}
else if (_T("npt") == strParam)
{
tidyOptSetValue(tdoc, TidyPreTags, CT2A(m_onlyHtmlXml == SYN_HTML?g_GlobalTidy.m_TidyHtml_npt:strNothing));
}
}
/*!
* \fn static int TidyHtml(const char *pcSourcePage, string &sDestPage);
* \brief 修补丢失、错误标签
* \param [in]待修补网页字符串
* \param [out]修补后的网页string
* \return 结果码，==0修补正确，<0修补失败
* \date 2011-06-01
* \author nanjunxiao
*/
int Pretreat::TidyHtml(const char *pcSourcePage, std::string &sDestPage)
{
int iReturn = 0;
TidyBuffer errbuf = {0};
TidyDoc tdoc;
tmbstr pBuffer = NULL;
try
{
if ( (pcSourcePage == NULL) || (strlen(pcSourcePage) ==0 ) )
{
//cerr << "TidyHtml 输入页面为空！" << endl;
throw (-1);
}
int iRet = -1;
Bool bOk;
uint uiBufLen;
int iBufSize;
tdoc = tidyCreate();// Initialize "document"
bOk = tidyOptSetBool(tdoc, TidyXhtmlOut, yes);// Convert to XHTML
if (bOk)
{
iRet = tidySetErrorBuffer(tdoc, &errbuf); // Capture diagnostics
}
else
{
throw (-1);
}
if (iRet >= 0)
{
iRet = tidySetCharEncoding(tdoc,"utf8"); //Ensure dealing with gb2312 successfully
}
else
{
throw (-1);
}
if (iRet >= 0)
{
string htmlsrc = pcSourcePage;
iRet = tidyParseString (tdoc, htmlsrc.c_str() ); // Parse the input
}
else
{
throw (-1);
}
if (iRet >= 0)
{
iRet = tidyCleanAndRepair(tdoc); //Tidy it up!
}
else
{
throw (-1);
}
if (iRet >= 0)
{
iRet = tidyRunDiagnostics(tdoc); //Kvetch
}
else
{
throw (-1);
}
if(iRet > 1) // If error, force output.
{
iRet = ( tidyOptSetBool(tdoc, TidyForceOutput, yes) ? iRet : -1 );
}
else if (iRet < 0)
{
throw (-1);
}
if (iRet >= 0)
{
// Pretty Print
iBufSize = 1024 * 1024 * 5;
uiBufLen = iBufSize;
pBuffer = new char [iBufSize];
memset(pBuffer, '\0', iBufSize);
iRet = tidySaveString(tdoc, pBuffer, &uiBufLen);
}
else
{
throw (-1);
}
if (iRet >= 0)
{
sDestPage = pBuffer;
}
else if (iRet == -ENOMEM)
{
//pBuffer 长度不够
//cerr << "TidyHtml pBuffer长度不够！" << endl;
throw (-1);
}
else
{
throw (-1);
}
}
catch(exception &err)
{
//cerr << "TidyHtml HtmlTidy修补页面失败！ " << err.what() << endl;
iReturn = -1;
}
catch(int iThrow)
{
if (iThrow < 0)
{
//cerr << "TidyHtml HtmlTidy修补页面失败！" << endl;
}
iReturn = iThrow;
}
catch(...)
{
//cerr << "TidyHtml HtmlTidy修补页面失败！" << endl;
iReturn = -1;
}
tidyBufFree(&errbuf);
tidyRelease(tdoc);
if (pBuffer != NULL)
{
delete [] pBuffer;
pBuffer = NULL;
}
return iReturn;
}