meinproc.cpp

00001 #include <config.h>
00002 #include <string.h>
00003 #include <sys/time.h>
00004 #include <unistd.h>
00005 #include <libxml/xmlversion.h>
00006 #include <libxml/xmlmemory.h>
00007 #include <libxml/debugXML.h>
00008 #include <libxml/HTMLtree.h>
00009 #include <libxml/xmlIO.h>
00010 #include <libxml/parserInternals.h>
00011 #include <libxslt/xsltconfig.h>
00012 #include <libxslt/xsltInternals.h>
00013 #include <libxslt/transform.h>
00014 #include <libxslt/xsltutils.h>
00015 #include <qstring.h>
00016 #include <kstandarddirs.h>
00017 #include <kinstance.h>
00018 #include <xslt.h>
00019 #include <qfile.h>
00020 #include <qdir.h>
00021 #include <kcmdlineargs.h>
00022 #include <klocale.h>
00023 #include <kaboutdata.h>
00024 #include <stdlib.h>
00025 #include <kdebug.h>
00026 #include <qtextcodec.h>
00027 #include <qfileinfo.h>
00028 #include <kprocess.h>
00029 #include <qvaluevector.h>
00030 
00031 extern int xmlLoadExtDtdDefaultValue;
00032 
00033 class MyPair {
00034 public:
00035     QString word;
00036     int base;};
00037 
00038 typedef QValueList<MyPair> PairList;
00039 
00040 void parseEntry(PairList &list, xmlNodePtr cur, int base)
00041 {
00042     if ( !cur )
00043         return;
00044 
00045     base += atoi( ( const char* )xmlGetProp(cur, ( const xmlChar* )"header") );
00046     if ( base > 10 ) // 10 is the maximum
00047         base = 10;
00048 
00049     /* We don't care what the top level element name is */
00050     cur = cur->xmlChildrenNode;
00051     while (cur != NULL) {
00052 
00053         if ( cur->type == XML_TEXT_NODE ) {
00054             QString words = QString::fromUtf8( ( char* )cur->content );
00055             QStringList wlist = QStringList::split( ' ',  words.simplifyWhiteSpace() );
00056             for ( QStringList::ConstIterator it = wlist.begin();
00057                   it != wlist.end(); ++it )
00058             {
00059                 MyPair m;
00060                 m.word = *it;
00061                 m.base = base;
00062                 list.append( m );
00063             }
00064         } else if ( !xmlStrcmp( cur->name, (const xmlChar *) "entry") )
00065             parseEntry( list, cur, base );
00066 
00067         cur = cur->next;
00068     }
00069 
00070 }
00071 
00072 static KCmdLineOptions options[] =
00073 {
00074     { "stylesheet <xsl>",  I18N_NOOP( "Stylesheet to use" ), 0 },
00075     { "stdout", I18N_NOOP( "Output whole document to stdout" ), 0 },
00076     { "o", 0, 0 },
00077     { "output <file>", I18N_NOOP("Output whole document to file" ), 0 },
00078     { "htdig", I18N_NOOP( "Create a ht://dig compatible index" ), 0 },
00079     { "check", I18N_NOOP( "Check the document for validity" ), 0 },
00080     { "cache <file>", I18N_NOOP( "Create a cache file for the document" ), 0},
00081     { "srcdir <dir>", I18N_NOOP( "Set the srcdir, for kdelibs" ), 0},
00082     { "param <key>=<value>", I18N_NOOP( "Parameters to pass to the stylesheet" ), 0},
00083     { "+xml", I18N_NOOP("The file to transform"), 0},
00084     KCmdLineLastOption // End of options.
00085 };
00086 
00087 
00088 
00089 
00090 int main(int argc, char **argv) {
00091 
00092     // xsltSetGenericDebugFunc(stderr, NULL);
00093 
00094     KAboutData aboutData( "meinproc", I18N_NOOP("XML-Translator" ),
00095     "$Revision: 497203 $",
00096     I18N_NOOP("KDE Translator for XML"));
00097 
00098     KCmdLineArgs::init(argc, argv, &aboutData);
00099     KCmdLineArgs::addCmdLineOptions( options );
00100 
00101     KLocale::setMainCatalogue("kio_help");
00102     KInstance ins("meinproc");
00103     KGlobal::locale();
00104 
00105 
00106     KCmdLineArgs *args = KCmdLineArgs::parsedArgs();
00107     if ( args->count() != 1 ) {
00108         args->usage();
00109         return ( 1 );
00110     }
00111 
00112     // Need to set SRCDIR before calling fillInstance
00113     QString srcdir;
00114     if ( args->isSet( "srcdir" ) )
00115         srcdir = QDir( QFile::decodeName( args->getOption( "srcdir" ) ) ).absPath();
00116     fillInstance(ins,srcdir);
00117 
00118     LIBXML_TEST_VERSION
00119 
00120     QString checkFilename = QFile::decodeName(args->arg( 0 ));
00121     QFileInfo checkFile(checkFilename);
00122     if (!checkFile.exists())
00123     {
00124         kdError() << "File '" << checkFilename << "' does not exist." << endl;
00125         return ( 2 );
00126     }
00127     if (!checkFile.isFile())
00128     {
00129         kdError() << "'" << checkFilename << "' is not a file." << endl;
00130         return ( 2 );
00131     }
00132     if (!checkFile.isReadable())
00133     {
00134         kdError() << "File '" << checkFilename << "' is not readable." << endl;
00135         return ( 2 );
00136     }
00137 
00138     if ( args->isSet( "check" ) ) {
00139         char pwd_buffer[250];
00140         QFileInfo file( QFile::decodeName(args->arg( 0 )) );
00141         if ( !getcwd( pwd_buffer, sizeof(pwd_buffer) ) ) {
00142          kdError() << "getcwd failed." << endl;
00143              return 2;
00144     }
00145 
00146         chdir( QFile::encodeName( file.dirPath( true ) ) );
00147 
00148         QString catalogs;
00149         catalogs += locate( "dtd", "customization/catalog" );
00150         catalogs += " ";
00151         catalogs += locate( "dtd", "docbook/xml-dtd-4.1.2/docbook.cat" );
00152 
00153         setenv( "SGML_CATALOG_FILES", QFile::encodeName( catalogs ).data(), 1);
00154         QString exe;
00155 #if defined( XMLLINT )
00156         exe = XMLLINT;
00157 #endif
00158         if ( ::access( QFile::encodeName( exe ), X_OK ) ) {
00159             exe = KStandardDirs::findExe( "xmllint" );
00160             if (exe.isEmpty())
00161                 exe = locate( "exe", "xmllint" );
00162         }
00163         if ( !::access( QFile::encodeName( exe ), X_OK ) ) {
00164             QString cmd = exe;
00165             cmd += " --catalogs --valid --noout ";
00166             cmd += KProcess::quote(file.fileName());
00167             cmd += " 2>&1";
00168             FILE *xmllint = popen( QFile::encodeName( cmd ), "r");
00169             char buf[ 512 ];
00170             bool noout = true;
00171             unsigned int n;
00172             while ( ( n = fread(buf, 1, sizeof( buf ), xmllint ) ) ) {
00173                 noout = false;
00174                 buf[ n ] = '\0';
00175                 fputs( buf, stderr );
00176             }
00177             pclose( xmllint );
00178             chdir( pwd_buffer );
00179             if ( !noout )
00180                 return 1;
00181         } else {
00182             kdWarning() << "couldn't find xmllint" << endl;
00183         }
00184     }
00185 
00186     xmlSubstituteEntitiesDefault(1);
00187     xmlLoadExtDtdDefaultValue = 1;
00188 
00189     QValueVector<const char *> params;
00190     if (args->isSet( "output" ) ) {
00191         params.append( qstrdup( "outputFile" ) );
00192         params.append( qstrdup( QFile::decodeName( args->getOption( "output" ) ).latin1() ) );
00193     }
00194     {
00195         const QCStringList paramList = args->getOptionList( "param" );
00196         QCStringList::ConstIterator it = paramList.begin();
00197         QCStringList::ConstIterator end = paramList.end();
00198         for ( ; it != end; ++it ) {
00199             const QCString tuple = *it;
00200             const int ch = tuple.find( '=' );
00201             if ( ch == -1 ) {
00202                 kdError() << "Key-Value tuple '" << tuple << "' lacks a '='!" << endl;
00203                 return( 2 );
00204             }
00205             params.append( qstrdup( tuple.left( ch ) ) );
00206             params.append( qstrdup( tuple.mid( ch + 1 ) )  );
00207         }
00208     }
00209     params.append( NULL );
00210 
00211     bool index = args->isSet( "htdig" );
00212     QString tss = args->getOption( "stylesheet" );
00213     if ( tss.isEmpty() )
00214         tss =  "customization/kde-chunk.xsl";
00215     if ( index )
00216         tss = "customization/htdig_index.xsl" ;
00217 
00218     tss = locate( "dtd", tss );
00219 
00220     if ( index ) {
00221         xsltStylesheetPtr style_sheet =
00222             xsltParseStylesheetFile((const xmlChar *)tss.latin1());
00223 
00224         if (style_sheet != NULL) {
00225 
00226             xmlDocPtr doc = xmlParseFile( QFile::encodeName( args->arg( 0 ) ) );
00227 
00228             xmlDocPtr res = xsltApplyStylesheet(style_sheet, doc, &params[0]);
00229 
00230             xmlFreeDoc(doc);
00231             xsltFreeStylesheet(style_sheet);
00232             if (res != NULL) {
00233                 xmlNodePtr cur = xmlDocGetRootElement(res);
00234                 if (!cur || xmlStrcmp(cur->name, (const xmlChar *) "entry")) {
00235                     fprintf(stderr,"document of the wrong type, root node != entry");
00236                     xmlFreeDoc(res);
00237                     return(1);
00238                 }
00239                 PairList list;
00240                 parseEntry( list, cur, 0 );
00241                 int wi = 0;
00242                 for ( PairList::ConstIterator it = list.begin(); it != list.end();
00243                       ++it, ++wi )
00244                     fprintf( stdout, "w\t%s\t%d\t%d\n", ( *it ).word.utf8().data(),
00245                              1000*wi/list.count(), ( *it ).base );
00246 
00247                 xmlFreeDoc(res);
00248             } else {
00249                 kdDebug() << "couldn't parse document " << args->arg( 0 ) << endl;
00250             }
00251         } else {
00252             kdDebug() << "couldn't parse style sheet " << tss << endl;
00253         }
00254 
00255     } else {
00256         QString output = transform(args->arg( 0 ) , tss, params);
00257         if (output.isEmpty()) {
00258             fprintf(stderr, "unable to parse %s\n", args->arg( 0 ));
00259             return(1);
00260         }
00261 
00262         QString cache = args->getOption( "cache" );
00263         if ( !cache.isEmpty() ) {
00264             if ( !saveToCache( output, cache ) ) {
00265                 kdError() << i18n( "Could not write to cache file %1." ).arg( cache ) << endl;
00266             }
00267             goto end;
00268         }
00269 
00270         if (output.find( "<FILENAME " ) == -1 || args->isSet( "stdout" ) || args->isSet("output") )
00271         {
00272             QFile file;
00273             if (args->isSet( "stdout" ) ) {
00274                 file.open( IO_WriteOnly, stdout );
00275             } else {
00276                 if (args->isSet( "output" ) )
00277                    file.setName( QFile::decodeName(args->getOption( "output" )));
00278                 else
00279                    file.setName( "index.html" );
00280                 file.open(IO_WriteOnly);
00281             }
00282             replaceCharsetHeader( output );
00283 
00284             QCString data = output.local8Bit();
00285             file.writeBlock(data.data(), data.length());
00286             file.close();
00287         } else {
00288             int index = 0;
00289             while (true) {
00290                 index = output.find("<FILENAME ", index);
00291                 if (index == -1)
00292                     break;
00293                 int filename_index = index + strlen("<FILENAME filename=\"");
00294 
00295                 QString filename = output.mid(filename_index,
00296                                               output.find("\"", filename_index) -
00297                                               filename_index);
00298 
00299                 QString filedata = splitOut(output, index);
00300                 QFile file(filename);
00301                 file.open(IO_WriteOnly);
00302                 replaceCharsetHeader( filedata );
00303                 QCString data = fromUnicode( filedata );
00304                 file.writeBlock(data.data(), data.length());
00305                 file.close();
00306 
00307                 index += 8;
00308             }
00309         }
00310     }
00311  end:
00312     xmlCleanupParser();
00313     xmlMemoryDump();
00314     return(0);
00315 }
00316 
KDE Home | KDE Accessibility Home | Description of Access Keys