WordWithRanges( const lString16 & w, const lString8 & enc, const ldomWord & range )
: word( w ), encoded( enc )
{
wordLower = w;
wordLower.lowercase();
ranges.add( range );
}
void match( const lString8& prefix, LVArray<WordWithRanges *> & result )
{
crtrace dumpstr;
dumpstr << "match with " << prefix;
for( int i=0; i<_words.length(); i++ ) {
if( _words[i]->matchEncoded( prefix ) ) {
result.add( _words[i] );
//dumpstr << " " << i << " " << encoded_words_[i];
};
};
}
void AddFootnoteFragmentToList()
{
if ( footstart==NULL )
return; // no data
if ( footend==NULL )
footend = footstart;
//CRLog::trace("AddFootnoteFragmentToList(%d, %d)", footstart->start, footend->end );
int h = footend->getEnd() - footstart->getStart(); // currentFootnoteHeight();
if ( h>0 && h<page_h ) {
footheight += h;
#ifdef DEBUG_FOOTNOTES
//CRLog::trace("AddFootnoteFragmentToList(%d, %d)", footstart->getStart(), h);
#endif
footnotes.add( LVPageFootNoteInfo( footstart->getStart(), h ) );
}
footstart = footend = NULL;
}
bool setManglingKey(lString16 key) {
if (key.startsWith(lString16(L"urn:uuid:")))
key = key.substr(9);
_fontManglingKey.clear();
_fontManglingKey.reserve(16);
lUInt8 b = 0;
int n = 0;
for (int i=0; i<key.length(); i++) {
int d = hexDigit(key[i]);
if (d>=0) {
b = (b << 4) | d;
if (++n > 1) {
_fontManglingKey.add(b);
n = 0;
b = 0;
}
}
}
return _fontManglingKey.length() == 16;
}
virtual void OnText( const lChar16 * text, int len, lUInt32 flags)
{
int fmt = m_stack.getInt(pi_imgfmt);
if (!fmt)
return;
_fmt = fmt;
for (int i=0; i<len;) {
int d = -1;
do {
d = i<len ? hexDigit(text[i]) : -1;
i++;
} while (d<0 && i<len);
if (_lastDigit>=0 && d>=0) {
_buf.add((lUInt8)((_lastDigit<<4) | d));
_lastDigit = -1;
} else {
if (d>=0)
_lastDigit = d;
}
}
}
bool ImportEpubDocument( LVStreamRef stream, ldomDocument * m_doc, LVDocViewCallback * progressCallback, CacheLoadingCallback * formatCallback )
{
LVContainerRef arc = LVOpenArchieve( stream );
if ( arc.isNull() )
return false; // not a ZIP archive
// check root media type
lString16 rootfilePath = EpubGetRootFilePath(arc);
if ( rootfilePath.empty() )
return false;
EncryptedDataContainer * decryptor = new EncryptedDataContainer(arc);
if (decryptor->open()) {
CRLog::debug("EPUB: encrypted items detected");
}
LVContainerRef m_arc = LVContainerRef(decryptor);
if (decryptor->hasUnsupportedEncryption()) {
// DRM!!!
createEncryptedEpubWarningDocument(m_doc);
return true;
}
m_doc->setContainer(m_arc);
// read content.opf
EpubItems epubItems;
//EpubItem * epubToc = NULL; //TODO
LVArray<EpubItem*> spineItems;
lString16 codeBase;
//lString16 css;
//
{
codeBase=LVExtractPath(rootfilePath, false);
CRLog::trace("codeBase=%s", LCSTR(codeBase));
}
LVStreamRef content_stream = m_arc->OpenStream(rootfilePath.c_str(), LVOM_READ);
if ( content_stream.isNull() )
return false;
lString16 ncxHref;
lString16 coverId;
LVEmbeddedFontList fontList;
EmbeddedFontStyleParser styleParser(fontList);
// reading content stream
{
ldomDocument * doc = LVParseXMLStream( content_stream );
if ( !doc )
return false;
CRPropRef m_doc_props = m_doc->getProps();
lString16 author = doc->textFromXPath( lString16(L"package/metadata/creator"));
lString16 title = doc->textFromXPath( lString16(L"package/metadata/title"));
m_doc_props->setString(DOC_PROP_TITLE, title);
m_doc_props->setString(DOC_PROP_AUTHORS, author );
for ( int i=1; i<50; i++ ) {
ldomNode * item = doc->nodeFromXPath( lString16(L"package/metadata/identifier[") + lString16::itoa(i) + L"]" );
if (!item)
break;
lString16 key = item->getText();
if (decryptor->setManglingKey(key)) {
CRLog::debug("Using font mangling key %s", LCSTR(key));
break;
}
}
CRLog::info("Author: %s Title: %s", LCSTR(author), LCSTR(title));
for ( int i=1; i<20; i++ ) {
ldomNode * item = doc->nodeFromXPath( lString16(L"package/metadata/meta[") + lString16::itoa(i) + L"]" );
if ( !item )
break;
lString16 name = item->getAttributeValue(L"name");
lString16 content = item->getAttributeValue(L"content");
if ( name == L"cover" )
coverId = content;
else if ( name==L"calibre:series" )
m_doc_props->setString(DOC_PROP_SERIES_NAME, content );
else if ( name==L"calibre:series_index" )
m_doc_props->setInt(DOC_PROP_SERIES_NUMBER, content.atoi() );
}
// items
for ( int i=1; i<50000; i++ ) {
ldomNode * item = doc->nodeFromXPath( lString16(L"package/manifest/item[") + lString16::itoa(i) + L"]" );
if ( !item )
break;
lString16 href = item->getAttributeValue(L"href");
lString16 mediaType = item->getAttributeValue(L"media-type");
lString16 id = item->getAttributeValue(L"id");
if ( !href.empty() && !id.empty() ) {
if ( id==coverId ) {
// coverpage file
lString16 coverFileName = codeBase + href;
CRLog::info("EPUB coverpage file: %s", LCSTR(coverFileName));
LVStreamRef stream = m_arc->OpenStream(coverFileName.c_str(), LVOM_READ);
if ( !stream.isNull() ) {
LVImageSourceRef img = LVCreateStreamImageSource(stream);
if ( !img.isNull() ) {
CRLog::info("EPUB coverpage image is correct: %d x %d", img->GetWidth(), img->GetHeight() );
m_doc_props->setString(DOC_PROP_COVER_FILE, coverFileName);
}
}
}
EpubItem * epubItem = new EpubItem;
epubItem->href = href;
epubItem->id = id;
epubItem->mediaType = mediaType;
epubItems.add( epubItem );
// // register embedded document fonts
// if (mediaType == L"application/vnd.ms-opentype"
// || mediaType == L"application/x-font-otf"
// || mediaType == L"application/x-font-ttf") { // TODO: more media types?
// // TODO:
// fontList.add(codeBase + href);
// }
}
if ( mediaType==L"text/css" ) {
lString16 name = LVCombinePaths(codeBase, href);
LVStreamRef cssStream = m_arc->OpenStream(name.c_str(), LVOM_READ);
if (!cssStream.isNull()) {
lString8 cssFile = UnicodeToUtf8(LVReadTextFile(cssStream));
lString16 base = name;
LVExtractLastPathElement(base);
CRLog::trace("style: %s", cssFile.c_str());
styleParser.parse(base, cssFile);
}
}
}
// spine == itemrefs
if ( epubItems.length()>0 ) {
ldomNode * spine = doc->nodeFromXPath( lString16(L"package/spine") );
if ( spine ) {
EpubItem * ncx = epubItems.findById( spine->getAttributeValue(L"toc") ); //TODO
//EpubItem * ncx = epubItems.findById(lString16("ncx"));
if ( ncx!=NULL )
ncxHref = codeBase + ncx->href;
for ( int i=1; i<50000; i++ ) {
ldomNode * item = doc->nodeFromXPath( lString16(L"package/spine/itemref[") + lString16::itoa(i) + L"]" );
if ( !item )
break;
EpubItem * epubItem = epubItems.findById( item->getAttributeValue(L"idref") );
if ( epubItem ) {
// TODO: add to document
spineItems.add( epubItem );
}
}
}
}
delete doc;
}
if ( spineItems.length()==0 )
return false;
#if BUILD_LITE!=1
if ( m_doc->openFromCache(formatCallback) ) {
if ( progressCallback ) {
progressCallback->OnLoadFileEnd( );
}
return true;
}
#endif
lUInt32 saveFlags = m_doc->getDocFlags();
m_doc->setDocFlags( saveFlags );
m_doc->setContainer( m_arc );
ldomDocumentWriter writer(m_doc);
#if 0
m_doc->setNodeTypes( fb2_elem_table );
m_doc->setAttributeTypes( fb2_attr_table );
m_doc->setNameSpaceTypes( fb2_ns_table );
#endif
//m_doc->setCodeBase( codeBase );
ldomDocumentFragmentWriter appender(&writer, lString16(L"body"), lString16(L"DocFragment"), lString16::empty_str );
writer.OnStart(NULL);
writer.OnTagOpenNoAttr(L"", L"body");
int fragmentCount = 0;
for ( int i=0; i<spineItems.length(); i++ ) {
if ( spineItems[i]->mediaType==L"application/xhtml+xml" ) {
lString16 name = codeBase + spineItems[i]->href;
appender.addPathSubstitution( name, lString16(L"_doc_fragment_") + lString16::itoa(i) );
}
}
for ( int i=0; i<spineItems.length(); i++ ) {
if ( spineItems[i]->mediaType==L"application/xhtml+xml" ) {
lString16 name = codeBase + spineItems[i]->href;
{
CRLog::debug("Checking fragment: %s", LCSTR(name));
LVStreamRef stream = m_arc->OpenStream(name.c_str(), LVOM_READ);
if ( !stream.isNull() ) {
appender.setCodeBase( name );
lString16 base = name;
LVExtractLastPathElement(base);
//CRLog::trace("base: %s", LCSTR(base));
//LVXMLParser
LVHTMLParser parser(stream, &appender);
if ( parser.CheckFormat() && parser.Parse() ) {
// valid
fragmentCount++;
lString8 headCss = appender.getHeadStyleText();
//CRLog::trace("style: %s", headCss.c_str());
styleParser.parse(base, headCss);
} else {
CRLog::error("Document type is not XML/XHTML for fragment %s", LCSTR(name));
}
}
}
}
}
ldomDocument * ncxdoc = NULL;
if ( !ncxHref.empty() ) {
LVStreamRef stream = m_arc->OpenStream(ncxHref.c_str(), LVOM_READ);
lString16 codeBase = LVExtractPath( ncxHref );
if ( codeBase.length()>0 && codeBase.lastChar()!='/' )
codeBase.append(1, L'/');
appender.setCodeBase(codeBase);
if ( !stream.isNull() ) {
ldomDocument * ncxdoc = LVParseXMLStream( stream );
if ( ncxdoc!=NULL ) {
ldomNode * navMap = ncxdoc->nodeFromXPath( lString16(L"ncx/navMap"));
if ( navMap!=NULL )
ReadEpubToc( m_doc, navMap, m_doc->getToc(), appender );
delete ncxdoc;
}
}
}
writer.OnTagClose(L"", L"body");
writer.OnStop();
CRLog::debug("EPUB: %d documents merged", fragmentCount);
if (!fontList.empty()) {
// set document font list, and register fonts
m_doc->getEmbeddedFontList().set(fontList);
m_doc->registerEmbeddedFonts();
m_doc->forceReinitStyles();
}
if ( fragmentCount==0 )
return false;
#if 0
// set stylesheet
//m_doc->getStyleSheet()->clear();
m_doc->setStyleSheet( NULL, true );
//m_doc->getStyleSheet()->parse(m_stylesheet.c_str());
if ( !css.empty() && m_doc->getDocFlag(DOC_FLAG_ENABLE_INTERNAL_STYLES) ) {
m_doc->setStyleSheet( "p.p { text-align: justify }\n"
"svg { text-align: center }\n"
"i { display: inline; font-style: italic }\n"
"b { display: inline; font-weight: bold }\n"
"abbr { display: inline }\n"
"acronym { display: inline }\n"
"address { display: inline }\n"
"p.title-p { hyphenate: none }\n"
//abbr, acronym, address, blockquote, br, cite, code, dfn, div, em, h1, h2, h3, h4, h5, h6, kbd, p, pre, q, samp, span, strong, var
, false);
m_doc->setStyleSheet( UnicodeToUtf8(css).c_str(), false );
//m_doc->getStyleSheet()->parse(UnicodeToUtf8(css).c_str());
} else {
//m_doc->getStyleSheet()->parse(m_stylesheet.c_str());
//m_doc->setStyleSheet( m_stylesheet.c_str(), false );
}
#endif
#if 0
LVStreamRef out = LVOpenFileStream( L"c:\\doc.xml" , LVOM_WRITE );
if ( !out.isNull() )
m_doc->saveToStream( out, "utf-8" );
#endif
// DONE!
if ( progressCallback ) {
progressCallback->OnLoadFileEnd( );
m_doc->compact();
m_doc->dumpStatistics();
}
return true;
}
void add( const ldomWord & range )
{
ranges.add( range );
}