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
}
void tidyhtml::parse(std::string x, std::string path)
{
tidySetErrorBuffer(tdoc, NULL);
tidyParseString(tdoc, x.c_str());
tidyCleanAndRepair( tdoc );
tidyRunDiagnostics( tdoc );
tidySaveFile(tdoc, path.c_str());
}
/* {{{ proto boolean tidy_diagnose()
Run configured diagnostics on parsed and repaired markup. */
static PHP_FUNCTION(tidy_diagnose)
{
TIDY_FETCH_OBJECT;
if (obj->ptdoc->initialized && tidyRunDiagnostics(obj->ptdoc->doc) >= 0) {
tidy_doc_update_properties(obj);
RETURN_TRUE;
}
RETURN_FALSE;
}
int main(int argc, char **argv )
{
CURL *curl;
char curl_errbuf[CURL_ERROR_SIZE];
TidyDoc tdoc;
TidyBuffer docbuf = {0};
TidyBuffer tidy_errbuf = {0};
int err;
if ( argc == 2) {
curl = curl_easy_init();
curl_easy_setopt(curl, CURLOPT_URL, argv[1]);
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_WRITEFUNCTION, write_cb);
tdoc = tidyCreate();
tidyOptSetBool(tdoc, TidyForceOutput, yes); /* try harder */
tidyOptSetInt(tdoc, TidyWrapLen, 4096);
tidySetErrorBuffer( tdoc, &tidy_errbuf );
tidyBufInit(&docbuf);
curl_easy_setopt(curl, CURLOPT_WRITEDATA, &docbuf);
err=curl_easy_perform(curl);
if ( !err ) {
err = tidyParseBuffer(tdoc, &docbuf); /* parse the input */
if ( err >= 0 ) {
err = tidyCleanAndRepair(tdoc); /* fix any problems */
if ( err >= 0 ) {
err = tidyRunDiagnostics(tdoc); /* load tidy error buffer */
if ( err >= 0 ) {
dumpNode( tdoc, tidyGetRoot(tdoc), 0 ); /* walk the tree */
fprintf(stderr, "%s\n", tidy_errbuf.bp); /* show errors */
}
}
}
}
else
fprintf(stderr, "%s\n", curl_errbuf);
/* clean-up */
curl_easy_cleanup(curl);
tidyBufFree(&docbuf);
tidyBufFree(&tidy_errbuf);
tidyRelease(tdoc);
return(err);
}
else
printf( "usage: %s <url>\n", argv[0] );
return(0);
}
static html_valid_status_t
html_valid_run (html_valid_t * htv, SV * html,
SV ** output_ptr, SV ** errors_ptr)
{
const char * html_string;
STRLEN html_length;
SV * output;
SV * errors;
TidyBuffer tidy_output = {0};
TidyBuffer tidy_errbuf = {0};
/* First set these up sanely in case the stuff hits the fan. */
* output_ptr = & PL_sv_undef;
* errors_ptr = & PL_sv_undef;
/* Work around bug where allocator sometimes does not get set. */
CopyAllocator (htv->tdoc, & tidy_output);
CopyAllocator (htv->tdoc, & tidy_errbuf);
html_string = SvPV (html, html_length);
CALL_TIDY (tidySetErrorBuffer (htv->tdoc, & tidy_errbuf));
htv->n_mallocs++;
CALL_TIDY (tidyParseString (htv->tdoc, html_string));
CALL_TIDY (tidyCleanAndRepair (htv->tdoc));
CALL_TIDY (tidyRunDiagnostics (htv->tdoc));
CALL_TIDY (tidySaveBuffer (htv->tdoc, & tidy_output));
htv->n_mallocs++;
/* Copy the contents of the buffers into the Perl scalars. */
output = newSVpv ((char *) tidy_output.bp, tidy_output.size);
errors = newSVpv ((char *) tidy_errbuf.bp, tidy_errbuf.size);
/* HTML Tidy randomly segfaults here due to "allocator" not being
set in some cases, hence the above CopyAllocator fix. */
tidyBufFree (& tidy_output);
htv->n_mallocs--;
tidyBufFree (& tidy_errbuf);
htv->n_mallocs--;
/* These are not our mallocs, they are Perl's mallocs, so we don't
increase htv->n_mallocs for these. After we return them, we no
longer take care of these. */
* output_ptr = output;
* errors_ptr = errors;
return html_valid_ok;
}
void TidyNetworkReply::tidyUp() {
QUrl redirect = reply->attribute(QNetworkRequest::RedirectionTargetAttribute).toUrl();
if (redirect.isValid()) {
redirect.setScheme("tidy");
setAttribute(QNetworkRequest::RedirectionTargetAttribute, QVariant(redirect));
emit finished();
reply->deleteLater();
return;
}
int rc = -1;
Bool ok;
ok = tidyOptSetBool( tdoc, TidyXmlOut, yes ); // Convert to XHTML
if (ok)
ok = tidyOptSetBool(tdoc, TidyQuoteNbsp, no);
//if (ok)
//ok = tidyOptSetValue(tdoc, TidyBlockTags, "header,nav,article,time,section,footer");
if ( ok )
rc = tidySetErrorBuffer( tdoc, &errbuf ); // Capture diagnostics
if ( rc >= 0 )
rc = tidyParseString( tdoc, reply->readAll() ); // 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 ) {
if ( rc > 0 ) {
;//printf( "\nDiagnostics:\n\n%s", errbuf.bp );
}
} else {
;//printf( "A severe error (%d) occurred.\n", rc );
}
open(ReadOnly);
emit readyRead();
emit finished();
reply->deleteLater();
//QTimer::singleShot(0, this, SIGNAL(readyRead()));
//QTimer::singleShot(0, this, SIGNAL(finished()));
}
void HTMLTidy::run() throw( std::runtime_error ) {
TidyBuffer outputBuffer = { 0 };
TidyBuffer errorBuffer = { 0 };
// try to create valid XHTML document for XML parser:
int tidyResult = -1;
if( tidyOptSetBool( handle, TidyXhtmlOut, yes ) ) {
tidyResult = tidySetErrorBuffer( handle, &errorBuffer );
}
if( tidyResult >= 0 ) {
tidyResult = tidyParseString( handle, document.c_str() );
}
if( tidyResult >= 0 ) {
tidyResult = tidyCleanAndRepair( handle );
}
if( tidyResult >= 0 ) {
tidyResult = tidyRunDiagnostics( handle );
}
if( tidyResult > 1 ) {
if( !tidyOptSetBool( handle, TidyForceOutput, yes ) ) {
tidyResult = -1;
}
}
if( tidyResult >= 0 ) {
tidyResult = tidySaveBuffer( handle, &outputBuffer );
}
if( tidyResult > 0 ) {
std::clog << "*********************************" << std::endl;
std::clog << "HTMLTidy: Diagnostics of libtidy:" << std::endl;
std::clog << errorBuffer.bp;
std::clog << "*********************************" << std::endl;
}
else if( tidyResult < 0 ) {
std::stringstream sstrTidyResult;
sstrTidyResult << tidyResult;
throw std::runtime_error( "HTMLTidy: A severe error occured while tidying up the received document ("
+ sstrTidyResult.str()
+ ")."
);
}
resultDocument.reserve( outputBuffer.size ); // avoid frequent (re-)allocations
for( unsigned int i = 0; i < outputBuffer.size; i++ ) {
resultDocument.insert( resultDocument.end(), static_cast< char >( *(outputBuffer.bp + i) ) );
}
tidyBufFree( &outputBuffer );
tidyBufFree( &errorBuffer );
}
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;
}
bool CCFHtmlTidy::TidyMain(const char* pSourceIn, const char* pOptions, std::string &strOut, std::string &strErr)
{
TidyBuffer output;
TidyBuffer errbuf;
int rc = -1;
Bool ok = yes;
TidyDoc tdoc = tidyCreate(); // Initialize "document"
tidyBufInit(&output);
tidyBufInit(&errbuf);
TidyOptionsSet(tidyDocToImpl(tdoc), pOptions);
if (ok)
rc = tidySetErrorBuffer(tdoc, &errbuf); // Capture diagnostics
if (rc >= 0)
rc = tidyParseString(tdoc, pSourceIn); // 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)
{
if (output.bp)
{
strOut = reinterpret_cast<char const*>(output.bp);
}
}
strErr = reinterpret_cast<char const*>(errbuf.bp);
std::string strEmpty = "No warnings or errors were found.\n\n";
if (0 == strEmpty.compare(strErr))
{
strErr.clear();
}
tidyBufFree(&output);
tidyBufFree(&errbuf);
tidyRelease(tdoc);
return true;
}
bool HTidyInterface::formatSource( const char* textIn, CString &strTidy, CString &strMsg )
{
TidyBuffer output;
TidyBuffer errbuf;
int rc = -1;
Bool ok = yes;
TidyDoc tdoc = tidyCreate(); // Initialize "document"
tidyBufInit(&output);
tidyBufInit(&errbuf);
InitTidyDefault(tdoc);
SetTidyConfig(tdoc);
if ( ok )
rc = tidySetErrorBuffer(tdoc, &errbuf); // Capture diagnostics
if ( rc >= 0 )
rc = tidyParseString(tdoc, textIn); // 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 )
{
strTidy = reinterpret_cast< char const* >(output.bp);
}
strMsg = reinterpret_cast< char const* >(errbuf.bp);
CString strEmpty = _T("No warnings or errors were found.\r\n\r\n");
if (0 == strEmpty.Compare(strMsg))
{
strMsg.Empty();
}
tidyBufFree(&output);
tidyBufFree(&errbuf);
tidyRelease(tdoc);
return true;
}
int main(int argc, char **argv )
{
const char* input = "<title>Hello</title><p>World!";
TidyBuffer output = {0};
TidyBuffer errbuf = {0};
int rc = -1;
Bool ok;
// Initialize "document"
TidyDoc tdoc = tidyCreate();
printf( "Tidying:\t%s\n", input );
// Convert to XHTML
ok = tidyOptSetBool( tdoc, TidyXhtmlOut, yes );
if ( ok )
rc = tidySetErrorBuffer( tdoc, &errbuf ); // Capture diagnostics
if ( rc >= 0 )
rc = tidyParseString( tdoc, input ); // 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 )
{
if ( rc > 0 )
printf( "\nDiagnostics:\n\n%s", errbuf.bp );
printf( "\nAnd here is the result:\n\n%s", output.bp );
}
else
printf( "A severe error (%d) occurred.\n", rc );
tidyBufFree( &output );
tidyBufFree( &errbuf );
tidyRelease( tdoc );
return rc;
}
// quick and dirty shortcut function.
int lua_tidy_easyClean ( lua_State *L )
{
TidyBuffer output;
TidyBuffer errbuf;
int rc;
pTidy t;
const char * input;
tidyBufInit(&output);
tidyBufInit(&errbuf);
rc = -1;
t = toTidy(L,1);
input = lua_tostring(L,2);
rc = tidySetErrorBuffer( t->tdoc, &errbuf );
if ( rc >= 0 )
rc = tidyParseString( t->tdoc, input );
if ( rc >= 0 )
rc = tidyCleanAndRepair( t->tdoc );
if ( rc >= 0 )
rc = tidyRunDiagnostics( t->tdoc );
if ( rc >= 0 )
rc = tidySaveBuffer( t->tdoc, &output );
lua_pushlstring(L, (char*)output.bp,output.size);
if ( rc != 0 )
lua_pushlstring(L, (char*)errbuf.bp,errbuf.size);
else
lua_pushnil(L);
lua_pushnumber(L, rc);
tidyBufFree( &output );
tidyBufFree( &errbuf );
return 3;
}
void html_parse(const gchar* html, GSList** objs) {
TidyDoc tdoc = tidyCreate();
TidyBuffer tidy_errbuf = {0};
int err = 0;
tidyOptSetBool(tdoc, TidyForceOutput, yes); /* try harder */
tidyOptSetInt(tdoc, TidyWrapLen, 4096);
tidySetErrorBuffer( tdoc, &tidy_errbuf );
err = tidyParseString(tdoc, html); /* parse the input */
if ( err >= 0 ) {
err = tidyCleanAndRepair(tdoc); /* fix any problems */
if ( err >= 0 ) {
err = tidyRunDiagnostics(tdoc); /* load tidy error buffer */
if ( err >= 0 ) {
html_find_objects(tidyGetHtml(tdoc), objs); /* walk the tree */
}
}
}
}
static PyObject*
elementtidy_fixup(PyObject* self, PyObject* args)
{
int rc;
TidyDoc doc;
TidyBuffer out = {0};
TidyBuffer err = {0};
PyObject* pyout;
PyObject* pyerr;
char* text;
char* encoding = NULL;
if (!PyArg_ParseTuple(args, "s|s:fixup", &text, &encoding))
return NULL;
doc = tidyCreate();
/* options for nice XHTML output */
if (encoding)
/* if an encoding is given, use it for both input and output */
tidyOptSetValue(doc, TidyCharEncoding, encoding);
else
/* if no encoding is given, use default input and utf-8 output */
tidyOptSetValue(doc, TidyOutCharEncoding, "utf8");
tidyOptSetBool(doc, TidyForceOutput, yes);
tidyOptSetInt(doc, TidyWrapLen, 0);
tidyOptSetBool(doc, TidyQuiet, yes);
tidyOptSetBool(doc, TidyXhtmlOut, yes);
tidyOptSetBool(doc, TidyXmlDecl, yes);
tidyOptSetInt(doc, TidyIndentContent, 0);
tidyOptSetBool(doc, TidyNumEntities, yes);
rc = tidySetErrorBuffer(doc, &err);
if (rc < 0) {
PyErr_SetString(PyExc_IOError, "tidySetErrorBuffer failed");
goto error;
}
rc = tidyParseString(doc, text);
if (rc < 0) {
PyErr_SetString(PyExc_IOError, "tidyParseString failed");
goto error;
}
rc = tidyCleanAndRepair(doc);
if (rc < 0) {
PyErr_SetString(PyExc_IOError, "tidyCleanAndRepair failed");
goto error;
}
rc = tidyRunDiagnostics(doc);
if (rc < 0) {
PyErr_SetString(PyExc_IOError, "tidyRunDiagnostics failed");
goto error;
}
rc = tidySaveBuffer(doc, &out);
if (rc < 0) {
PyErr_SetString(PyExc_IOError, "tidyRunDiagnostics failed");
goto error;
}
pyout = PyString_FromString(out.bp ? out.bp : "");
if (!pyout)
goto error;
pyerr = PyString_FromString(err.bp ? err.bp : "");
if (!pyerr) {
Py_DECREF(pyout);
goto error;
}
tidyBufFree(&out);
tidyBufFree(&err);
tidyRelease(doc);
return Py_BuildValue("NN", pyout, pyerr);
error:
tidyBufFree(&out);
tidyBufFree(&err);
tidyRelease(doc);
return NULL;
}
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;
}
/*!
* \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;
}
bool TidyReader::openFile (const char * szFilename)
{
UT_DEBUGMSG(("using libtidy to parse HTML...\n"));
m_tidy = tidyCreate ();
if (m_tidy == 0) return false;
if (tidyOptSetBool (m_tidy, TidyXhtmlOut, yes) == 0)
{
UT_DEBUGMSG(("tidyOptSetBool failed!\n"));
closeFile ();
return false;
}
#ifndef DEBUG
tidySetErrorBuffer (m_tidy, &m_errbuf);
#endif
int parse_status;
if (m_buffer && m_length)
{
UT_DEBUGMSG(("parse HTML in buffer...\n"));
UT_Byte * buffer = const_cast<UT_Byte *>(m_buffer); // grr.
TidyBuffer inbuf;
tidyBufInit (&inbuf);
tidyBufAttach (&inbuf, buffer, static_cast<unsigned int>(m_length));
parse_status = tidyParseBuffer (m_tidy, &inbuf);
tidyBufDetach (&inbuf);
}
else
{
UT_DEBUGMSG(("parse HTML in file: %s\n",szFilename));
parse_status = tidyParseFile (m_tidy, szFilename);
}
if (parse_status < 0)
{
UT_DEBUGMSG(("tidyParseBuffer/File failed!\n"));
closeFile ();
return false;
}
parse_status = tidyCleanAndRepair (m_tidy);
if (parse_status < 0)
{
UT_DEBUGMSG(("tidyCleanAndRepair failed!\n"));
closeFile ();
return false;
}
parse_status = tidyRunDiagnostics (m_tidy);
if (parse_status < 0)
{
UT_DEBUGMSG(("tidyRunDiagnostics failed!\n"));
closeFile ();
return false;
}
if (parse_status > 1)
{
parse_status = (tidyOptSetBool (m_tidy, TidyForceOutput, yes) ? parse_status : -1);
}
if (parse_status < 0)
{
UT_DEBUGMSG(("tidyOptSetBool failed!\n"));
closeFile ();
return false;
}
parse_status = tidySaveBuffer (m_tidy, &m_outbuf);
if (parse_status < 0)
{
UT_DEBUGMSG(("tidySaveBuffer failed!\n"));
closeFile ();
return false;
}
UT_DEBUGMSG(("tidy succeeded!\n"));
#ifdef DEBUG
fputs ("================================================================\n", stderr);
fputs ((const char *) m_outbuf.bp, stderr);
fputs ("================================================================\n", stderr);
#endif
m_outbuf.next = 0;
return true;
}
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;
}
// returns warnings.
int lua_tidy_runDiagnostics (lua_State *L)
{
pTidy t = toTidy(L,1);
lua_pushnumber(L, tidyRunDiagnostics (t->tdoc));
return 1;
}
