hashirama/goldendict-ng
hashirama/goldendict-ng
0
0
Fork 0
mirror of https://github.com/xiaoyifang/goldendict-ng.git synced 2024-11-27 23:34:06 +00:00
Code Issues Actions 1 Projects Releases Wiki Activity

update ZimDictionary::convert(), proper display

Browse source
This commit is contained in:
jjzz 2015-01-07 07:08:13 -05:00
parent 49a24f61b1
commit 6bf6dd46df
1 changed files with 45 additions and 60 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

101
zim.cc
View file

@ -597,93 +597,78 @@ string ZimDictionary::convert( const string & in )
{
QString text = QString::fromUtf8( in.c_str() );
// replace background
text.replace( QRegExp( "<\\s*body\\s*([^>]*)background:([^;\"]*)" ),
QString( "<body \\1background: inherited;" ) );
// pattern of img and script
text.replace( QRegExp( "<\\s*(img|script)\\s*([^>]*)src=(\"|)(\\.\\.|)/" ),
QString( "<\\1 \\2src=\\3bres://%1/").arg( getId().c_str() ) );
// Fix links without '"'
text.replace( QRegExp( "href=(\\.\\.|)/([^\\s>]+)" ), QString( "href=\"\\1/\\2\"" ) );
// pattern <link... href="..." ...>
text.replace( QRegExp( "<\\s*link\\s*([^>]*)href=\"(\\.\\.|)/" ),
QString( "<link \\1href=\"bres://%1/").arg( getId().c_str() ) );
QRegExp linkRegexp1( "<\\s*a\\s*([^>]*)href=\"(?!(http(s|)|ftp)://)(/|)[^\"]*\"\\s*title=\"([^\"]*)\"",
// localize the en.wiki***.com|org series links
text.replace( QRegExp( "<\\s*a\\s+(class=\"external\"\\s+)href=\"http(s|)://en\\.(wiki(pedia|books|news|quote|source|versity)|wiktionary)\\.(org|com)/wiki/" ),
QString( "<a href=\"gdlookup://localhost/" ) );
// pattern <a href="..." ...>, excluding any known protocols such as http://, mailto:, #(comment)
// these links will be translated into local definitions
QRegExp rxLink( "<\\s*a\\s+([^>]*)href=\"(?!(\\w+://|#|mailto:|tel:))(/|)([^\"]*)\"\\s*(title=\"[^\"]*\")?[^>]*>",
Qt::CaseSensitive,
QRegExp::RegExp2 );
QRegExp linkRegexp2( "<\\s*a\\s*([^>]*)href=\"(\\.\\.|)/([^\"]*)\"",
Qt::CaseSensitive,
QRegExp::RegExp2 );
QRegExp linkRegexp3( "\\.(s|)htm(l|)", Qt::CaseInsensitive );
int pos = 0;
while( pos >= 0 )
while( (pos = rxLink.indexIn( text, pos )) >= 0 )
{
pos = linkRegexp1.indexIn( text, pos );
if( pos < 0 )
break;
QStringList list = rxLink.capturedTexts();
QString tag = list[3]; // a url, ex: Precambrian_Chaotian.html
if ( !list[4].isEmpty() ) // a title, ex: title="Precambrian/Chaotian"
tag = list[4].split("\"")[1];
QStringList list = linkRegexp1.capturedTexts();
tag.remove( QRegExp(".*/") ).
remove( QRegExp( "\\.(s|)htm(l|)$", Qt::CaseInsensitive ) ).
replace( "_", "%20" ).
prepend( "<a href=\"gdlookup://localhost/" ).
append( "\" " + list[4] + ">" );
QString tag = QString( "<a href=\"gdlookup://localhost/" );
QString link = list[ 3 ];
int nbeg = link.lastIndexOf( "/" );
if( nbeg < 0 )
nbeg = 0;
else
nbeg += 1;
int nend = link.lastIndexOf( "." );
if( nend < 0 || !link.mid( nend ).contains( linkRegexp3 ) )
nend = -1;
link = link.mid( nbeg, nend < 0 ? -1 : nend - nbeg );
link.replace( QChar( '_' ), "%20", Qt::CaseInsensitive );
tag += link + "\" title=\"" + link + "\"";
text.replace( pos, list[0].length(), tag );
pos += tag.length() + 1;
}
// Occassionally words needs to be displayed in vertical, but <br/> were changed to <br\> somewhere
// proper style: <a href="gdlookup://localhost/Neoptera" ... >N<br/>e<br/>o<br/>p<br/>t<br/>e<br/>r<br/>a</a>
QRegExp rxBR( "(<a href=\"gdlookup://localhost/[^\"]*\"\\s*[^>]*>)\\s*((\\w\\s*&lt;br(\\\\|/|)&gt;\\s*)+\\w)\\s*</a>",
Qt::CaseSensitive,
QRegExp::RegExp2 );
pos = 0;
while( pos >= 0 )
while( (pos = rxBR.indexIn( text, pos )) >= 0 )
{
pos = linkRegexp2.indexIn( text, pos );
if( pos < 0 )
break;
QStringList list = rxBR.capturedTexts();
QString tag = list[2];
tag.replace( QRegExp( "&lt;br( |)(\\\\|/|)&gt;", Qt::CaseInsensitive ) , "<br/>" ).
prepend( list[1] ).
append( "</a>" );
QStringList list = linkRegexp2.capturedTexts();
QString tag = QString( "<a ") + list[ 1 ]
+ "href=\"gdlookup://localhost/";
QString link = list[ 3 ];
int nbeg = link.lastIndexOf( "/" );
if( nbeg <= 0 )
nbeg = 0;
else
nbeg += 1;
int nend = link.lastIndexOf( "." );
if( nend < 0 || !link.mid( nend ).contains( linkRegexp3 ) )
nend = -1;
link = link.mid( nbeg, nend < 0 ? -1 : nend - nbeg );
link.replace( QChar( '_' ), "%20", Qt::CaseInsensitive );
tag += link + "\"";
text.replace( pos, list[0].length(), tag );
pos += tag.length() + 1;
}
// // output all links in the page - only for analysis
// QRegExp rxPrintAllLinks( "<\\s*a\\s+[^>]*href=\"[^\"]*\"[^>]*>",
// Qt::CaseSensitive,
// QRegExp::RegExp2 );
// pos = 0;
// while( (pos = rxPrintAllLinks.indexIn( text, pos )) >= 0 )
// {
// QStringList list = rxPrintAllLinks.capturedTexts();
// qDebug() << "\n--Alllinks--" << list[0];
// pos += list[0].length() + 1;
// }
// Fix outstanding elements
text += "<br style=\"clear:both;\" />";