+++ /dev/null
-/**************************************************************************
-* Copyright (C) 2003 by Hideki Ikemoto , (c)2004 by 421 *
-* ikemo@wakaba.jp *
-* *
-* This program is free software; you can redistribute it and/or modify *
-* it under the terms of the GNU General Public License as published by *
-* the Free Software Foundation; either version 2 of the License, or *
-* (at your option) any later version. *
-***************************************************************************/
-
-/*--------------------------------------*/
-/* miscellaneous utilities for parsing */
-
-#include <qmutex.h>
-#include <qregexp.h>
-#include <qdatetime.h>
-#include <dom/html_document.h>
-#include <dom/html_element.h>
-#include <dom/dom_text.h>
-#include <klocale.h>
-#include <kdebug.h>
-
-#include "parsemisc.h"
-#include "datinfo.h" /* struct RESDAT is defined. */
-#include "kita-utf8.h"
-#include "kita-utf16.h"
-#include "boardmanager.h"
-#include "kita_misc.h"
-
-#define KITA_RESDIGIT 4
-
-using namespace Kita;
-
-
-QString ParseMisc::m_weekstr[ 7 ];
-QString ParseMisc::m_colonstr;
-QString ParseMisc::m_colonnamestr;
-
-
-/*------------------------------------------------------------*/
-/*------------------------------------------------------------*/
-
-/* parsing functions */
-
-
-
-/*------------------------------------------*/
-/* */
-/* Parsing Engine for Title */
-/* */
-/*------------------------------------------*/
-
-/*
- struct RESDAT resdat should be parsed by parseResDat before
- calling this function. struct RESDAT is defined in datinfo.h
-
- If mode = PARSEMODE_DOM, titlenode is DOM tree of title node.
- titletext is ignored.
-
- If mode = PARSEMODE_HTML, titletext is HTML text of title.
- DOM tree is not created. Both hdoc and titlenode are ignored.
-
- If mode = PARSEMODE_TEXT, titletext is plain text of title.
- DOM tree is not created. Both hdoc and titlenode are ignored.
-*/
-
-/* public */
-void ParseMisc::parseTITLEdat(
-
- /* input */
- int mode, /* mode */
- DOM::HTMLDocument& hdoc, /* root node of DOM document*/
- int num, /* number of res */
- bool showMailAddress,
- const RESDAT& resdat, /* RESDAT is defined in datinfo.h */
-
- /* output */
- DOM::Element &titlenode, /* DOM tree of title */
- QString& titletext /* HTML or plain text of title */
-
-)
-{
-
- /*-----------------------------*/
- /* init */
- /*-----------------------------*/
-
- if ( !resdat.parsed ) {
- titletext = QString::null;
- return ;
- }
-
- const QString& str_name = resdat.name;
- const QString& str_address = resdat.address;
- const QString& str_id = resdat.id;
-
- unsigned int i;
- DOM::Element tmpelm;
- QString linkstr, linkstr2, linkurl, addstr, tmpstr;
-
- if ( m_colonstr == QString::null ) {
- m_colonstr = utf8ToUnicode( KITAUTF8_COLON );
- m_colonnamestr = utf8ToUnicode( KITAUTF8_NAME );
- }
-
- titletext = QString::null;
-
-
- /*-----------------------------*/
- /* number */
- /*-----------------------------*/
-
- linkstr = QString( "%1" ).arg( num );
- linkurl = "#write" + linkstr;
-
- switch ( mode ) {
-
- case PARSEMODE_DOM:
-
- tmpelm = titlenode.appendChild( hdoc.createElement( "TD" ) );
- tmpelm.setAttribute( "class", "res_title_number" );
- tmpelm = tmpelm.appendChild( hdoc.createElement( "A" ) );
- {
- /* set anchor id = number */
- tmpelm.setAttribute( "href", linkurl );
- tmpelm.appendChild( hdoc.createTextNode( linkstr ) );
- }
- break;
-
- case PARSEMODE_HTML:
- titletext += "<td class=\"pop_res_title_number\"><a href=\"" + linkurl + "\">";
- titletext += linkstr;
- titletext += "</a>";
- break;
-
- case PARSEMODE_TEXT:
- titletext += linkstr;
- break;
-
- }
-
-
- /*-----------------------------*/
- /* name & mail address */
- /*-----------------------------*/
-
- DOM::Element namenode;
- linkurl = QString::null;
-
- /* parse name & address strings */
- parseBODYdatText( PARSEMODE_TEXT, str_name, linkstr );
- if ( str_address != QString::null ) parseBODYdatText( PARSEMODE_TEXT, str_address, addstr );
- else addstr = QString::null;
-
- tmpstr = " " + m_colonnamestr;
- switch ( mode ) {
-
- case PARSEMODE_DOM:
- tmpelm = titlenode.appendChild( hdoc.createElement( "TD" ) );
- tmpelm.setAttribute( "class", "res_title_name" );
- tmpelm.appendChild( hdoc.createTextNode( tmpstr ) );
- namenode = tmpelm.appendChild( hdoc.createElement( "B" ) );
- break;
-
- case PARSEMODE_HTML:
- titletext += "<td class=\"pop_res_title_name\">" + tmpstr;
- break;
- case PARSEMODE_TEXT:
- titletext += tmpstr;
- break;
- }
-
- /* show name with mail address, or show name only */
- if ( showMailAddress || addstr == QString::null ) {
-
- const QChar * chpt = linkstr.unicode();
- unsigned int length = linkstr.length();
-
- unsigned int pos;
- int refNum[ 2 ];
- i = 0;
-
- bool ancChain = TRUE;
- /* ancChain is chain for anchor. For examle, if anchor "2"
- appeared, ancChain is set to TRUE. Moreover, if next strings
- are "=5", anchor for 5 is also set. Thus, we can obtain anchors
- for strings "2=5" as follows:
-
- <a href="#2">2</a><a href="#5">=5</a>
- */
-
- /* show name */
- for ( ;; ) {
-
- linkurl = QString::null;
- linkstr2 = QString::null;
-
- /* get strings & anchor for digits */
- if ( showMailAddress && ancChain ) {
-
- if ( ( ancChain = parseResAnchor( chpt + i, length - i, linkstr2, refNum, pos ) ) ) {
-
- linkurl = QString( "#%1" ).arg( refNum[ 0 ] );
- if ( refNum[ 1 ] ) linkurl += QString( "-%1" ).arg( refNum[ 1 ] );
- }
-
- i += pos;
- } else { /* get strings for non-digits */
-
- while ( i < length ) linkstr2 += chpt[ i++ ];
- }
-
- if ( linkstr2 != QString::null ) {
-
- switch ( mode ) {
-
- case PARSEMODE_DOM:
-
- if ( linkurl != QString::null ) { /* create anchor */
-
- tmpelm = namenode.appendChild( hdoc.createElement( "A" ) );
- {
- tmpelm.setAttribute( "href", linkurl );
- tmpelm.appendChild( hdoc.createTextNode( linkstr2 ) );
- }
-
-
- } else {
-
- tmpelm = namenode.appendChild( hdoc.createElement( "SPAN" ) );
- {
- tmpelm.setAttribute( "style", "color: green" );
- tmpelm.appendChild( hdoc.createTextNode( linkstr2 ) );
- }
- }
-
- break;
-
- case PARSEMODE_HTML:
- case PARSEMODE_TEXT:
-
- if ( mode == PARSEMODE_HTML
- && linkurl != QString::null ) { /* create anchor */
-
- titletext += "<a href=\"" + linkurl + "\">";
- titletext += linkstr2;
- titletext += "</a>";
-
- } else titletext += linkstr2;
-
- break;
- }
- }
-
- if ( i >= linkstr.length() ) break;
-
- } /* for(;;) */
-
-
- /* show mail address */
- switch ( mode ) {
-
- case PARSEMODE_DOM:
-
- if ( showMailAddress && addstr != QString::null ) {
- tmpstr = QString( " [" ) + addstr + "]";
- namenode.appendChild( hdoc.createTextNode( tmpstr ) );
- }
-
- break;
-
- case PARSEMODE_HTML:
- case PARSEMODE_TEXT:
-
- if ( showMailAddress && addstr != QString::null ) {
- tmpstr = QString( " [" ) + addstr + "]";
- titletext += tmpstr;
- }
-
- break;
- }
-
- } /* if ( showMailAddress || addstr == QString::null ) */
-
- /* don't show mail address (i.e. showMailAddress == FALSE) */
- else {
-
- linkurl = QString( "mailto:" ) + addstr;
-
- switch ( mode ) {
-
- case PARSEMODE_DOM:
- tmpelm = namenode.appendChild( hdoc.createElement( "A" ) );
- {
- tmpelm.setAttribute( "href", linkurl );
- tmpelm.setAttribute( "title", addstr );
- tmpelm.appendChild( hdoc.createTextNode( linkstr ) );
- }
- break;
-
- case PARSEMODE_HTML:
-
- titletext += "<a href=\"" + linkurl;
- titletext += " title=\"" + addstr + "\">";
- titletext += linkstr;
- titletext += "</a>";
-
- break;
-
- case PARSEMODE_TEXT:
- titletext += linkstr;
- break;
- }
- }
-
-
- /*-----------------------------*/
- /* date & ID */
- /*-----------------------------*/
-
- /* show date */
- tmpstr = " " + m_colonstr + resdat.date + " ";
-
- switch ( mode ) {
-
- case PARSEMODE_DOM:
- tmpelm = titlenode.appendChild( hdoc.createElement( "TD" ) );
- tmpelm.setAttribute( "class", "res_title_date" );
- tmpelm.appendChild( hdoc.createTextNode( tmpstr ) );
- break;
-
- case PARSEMODE_HTML:
- titletext += "<td class=\"pop_res_title_date\">" + tmpstr;
- break;
-
- case PARSEMODE_TEXT:
- titletext += tmpstr;
- break;
-
- }
-
- /* show ID */
- if ( str_id != QString::null ) {
-
- if ( str_id == "???" ) tmpstr = "ID:" + str_id;
- else tmpstr = ":" + str_id;
-
- switch ( mode ) {
-
- case PARSEMODE_DOM:
-
- tmpelm = titlenode.appendChild( hdoc.createElement( "TD" ) );
- tmpelm.setAttribute( "class", "res_title_id" );
- if ( tmpstr.left( 3 ) != "ID:" ) {
- DOM::Element aelm = tmpelm.appendChild( hdoc.createElement( "A" ) );
- {
- aelm.setAttribute( "href", "#idpop" + str_id );
- aelm.appendChild( hdoc.createTextNode( "ID" ) );
- }
- }
-
- tmpelm.appendChild( hdoc.createTextNode( tmpstr ) );
- break;
-
- case PARSEMODE_HTML:
- titletext += "<td class=\"pop_res_title_id\">";
- if ( tmpstr.left( 3 ) != "ID:" )
- titletext += "<a href=\"#idpop" + str_id + "\">ID</a>";
- titletext += tmpstr;
- break;
-
- case PARSEMODE_TEXT:
- if ( tmpstr.left( 3 ) != "ID:" ) titletext += "ID";
- titletext += tmpstr;
- break;
- }
- }
-
-}
-
-
-/*-----------------------------------------------------*/
-/* For convenience.
- mode can be set to PARSEMODE_HTML or PARSEMODE_TEXT */ /* public */
-void ParseMisc::parseTITLEdatText(
-
- /* input */
- int mode, /* mode = PARSEMODE_HTML or PARSEMODE_TEXT */
- int num,
- bool showMailAddress,
- const RESDAT& resdat, /* RESDAT is defined in datinfo.h */
-
- /* output */
- QString& titletext /* HTML or plain text of title */
-
-)
-{
-
- /* dummy */
- DOM::HTMLDocument hdoc;
- DOM::Element titlenode;
-
- parseTITLEdat( mode, hdoc, num, showMailAddress, resdat, titlenode, titletext );
-}
-
-
-
-
-
-/*------------------------------------------*/
-/* */
-/* Parsing Engine for Body */
-/* */
-/*------------------------------------------*/
-
-/*
- If mode = PARSEMODE_DOM, bodynode is DOM tree of body.
- bodytext is ignored.
-
- If mode = PARSEMODE_HTML, bodytext is HTML text of body.
- DOM tree is not created. Both hdoc and bodynode are ignored.
-
- If mode = PARSEMODE_TEXT, bodytext is plain text of body.
- DOM tree is not created. Both hdoc and bodynode are ignored.
-*/
-
-/* public */
-void ParseMisc::parseBODYdat(
-
- /* input */
- int mode, /* mode */
- const QString &rawStr, /* raw strings of body text */
- DOM::HTMLDocument& hdoc, /* root node of DOM document*/
- bool showAA, /* show AA (for KDE3.1x) */
-
- /* output */
- DOM::Element &bodynode, /* DOM tree of body */
- QString& bodytext /* HTML or plain text of body */
-
-)
-{
-
- /*-----------------------------------------*/
- /* init */
-
- unsigned int i, i2, index, pos, length = rawStr.length();
- DOM::Element tmpelm;
- QString linkstr, linkurl;
-
- const QChar *chpt = rawStr.unicode();
- QString lineStr = QString::null;
- bodytext = QString::null;
-
- bool ancChain = FALSE;
- /* ancChain is chain for anchor. For examle, if anchor ">2"
- appeared, ancChain is set to TRUE. Moreover, if next strings
- are "=5", anchor for 5 is also set. Thus, we can obtain anchors
- for strings ">2=5" as follows:
-
- <a href="#2">>2</a><a href="#5">=5</a>
- */
-
-
- /*-----------------------------------------*/
-
- for ( i = index = 0 ; i < length ; i++ ) {
-
- switch ( chpt[ i ].unicode() ) {
-
- case '<':
-
- /* " <br> " */
- if ( chpt[ i + 1 ] == 'b' && chpt[ i + 2 ] == 'r' && chpt[ i + 3 ] == '>' ) {
-
- i2 = i - index;
- if ( i > 0 && chpt[ i - 1 ] == ' ' ) i2--; /* remove space before <br> */
- lineStr += rawStr.mid( index, i2 );
-
- switch ( mode ) {
-
- case PARSEMODE_DOM:
-
- /* add BR node */
- bodynode.appendChild( hdoc.createTextNode( lineStr ) );
- bodynode.appendChild( hdoc.createElement( "BR" ) );
-
- /* show Ascii Art (for KDE3.1*) */
- if ( showAA ) {
-
- /* put the span node after BR node */
- tmpelm = bodynode.appendChild( hdoc.createElement( "SPAN" ) );
- {
- tmpelm.setAttribute( "style", "color: white" );
- tmpelm.appendChild( hdoc.createTextNode( "" ) );
- }
- }
-
- break;
-
- case PARSEMODE_HTML:
-
- bodytext += lineStr;
- bodytext += "<br>";
- /* show Ascii Art (for KDE3.1*) */
- if ( showAA ) {
- bodytext += "<span style=\"color: white\"></span>";
- }
- break;
-
- case PARSEMODE_TEXT:
-
- bodytext += lineStr;
- bodytext += '\n';
-
- break;
- }
-
- index = i + 4;
- if ( chpt[ index ] == ' ' ) index++; /* remove space after <br> */
- i = index - 1;
- lineStr = QString::null;
- ancChain = FALSE;
- }
-
- /*----------------------------------------*/
-
- /* remove HTML tags <[^>]*> */
- else {
- lineStr += rawStr.mid( index, i - index );
- while ( chpt[ i ] != '>' && i < length ) i++;
- index = i + 1;
- }
-
- break;
-
- /*----------------------------------------*/
-
- case 'h': /* "http://" or "ttp://" or "tp:" */
- case 't':
-
- if ( mode != PARSEMODE_TEXT
- && parseLink( chpt + i, length - i, linkstr, linkurl, pos ) ) {
-
- lineStr += rawStr.mid( index, i - index );
-
- switch ( mode ) {
-
- case PARSEMODE_DOM:
-
- /* create A node */
- bodynode.appendChild( hdoc.createTextNode( lineStr ) );
-
- tmpelm = bodynode.appendChild( hdoc.createElement( "A" ) );
- {
- tmpelm.setAttribute( "href", linkurl );
- tmpelm.appendChild( hdoc.createTextNode( linkstr ) );
- }
-
- break;
-
- case PARSEMODE_HTML:
- bodytext += lineStr;
- bodytext += "<a href=\"" + linkurl + "\">";
- bodytext += linkstr;
- bodytext += "</a>";
- break;
- }
-
- index = i + pos;
- i = index - 1;
- lineStr = QString::null;
- }
-
- break;
-
- /*----------------------------------*/
-
- case '&':
-
- /* > */
- if ( ( mode == PARSEMODE_DOM || mode == PARSEMODE_HTML )
- && chpt[ i + 1 ] == 'g' && chpt[ i + 2 ] == 't' && chpt[ i + 3 ] == ';' )
- ancChain = createResAnchor( mode, rawStr, hdoc, bodynode, bodytext, chpt, i, index, lineStr );
- /* special char */
- else if ( mode == PARSEMODE_DOM || mode == PARSEMODE_TEXT ) {
-
- QString tmpstr;
- tmpstr = parseSpecialChar( chpt + i, pos );
-
- if ( tmpstr != QString::null ) {
- lineStr += rawStr.mid( index, i - index ) + tmpstr;
- index = i + pos;
- i = index - 1;
- }
- }
-
- break;
-
- /*----------------------------------------*/
-
- /* unicode '>' */
- case UTF16_BRACKET:
- if ( mode != PARSEMODE_TEXT )
- ancChain = createResAnchor( mode, rawStr, hdoc, bodynode, bodytext, chpt, i, index, lineStr );
- break;
-
- /*----------------------------------*/
-
- default:
- if ( mode != PARSEMODE_TEXT && ancChain )
- ancChain = createResAnchor( mode, rawStr, hdoc, bodynode, bodytext, chpt, i, index, lineStr );
- }
- }
-
-
- /*---------------------------*/
-
- lineStr += rawStr.mid( index );
-
- switch ( mode ) {
-
- case PARSEMODE_DOM:
- bodynode.appendChild( hdoc.createTextNode( lineStr ) );
- break;
-
- case PARSEMODE_HTML:
- case PARSEMODE_TEXT:
- bodytext += lineStr;
- break;
- }
-
-}
-
-
-/*-----------------------------------------------------*/
-/* For convenience.
- mode can be set to PARSEMODE_HTML or PARSEMODE_TEXT */ /* public */
-void ParseMisc::parseBODYdatText(
-
- /* input */
- int mode, /* mode = PARSEMODE_HTML or PARSEMODE_TEXT */
- const QString &rawStr, /* raw strings of body text */
-
- /* output */
- QString& bodytext /* HTML or plain text of body */
-
-)
-{
-
- /* dummy */
- DOM::HTMLDocument hdoc;
- DOM::Element bodynode;
-
- parseBODYdat( mode, rawStr, hdoc, FALSE, bodynode, bodytext );
-}
-
-
-
-/*--------------------------------------------*/
-/* get HTML from RESDAT */
-/*
- resdat should be parsed by parseResDat before
- calling this function.
-
- struct RESDAT is defined in datinfo.h . */ /* public */
-QString ParseMisc::ResDatToHtml( const RESDAT& resdat, int num, bool showAddr )
-{
- QString result, titletext, bodytext;
-
- ParseMisc::parseTITLEdatText( PARSEMODE_HTML, num, showAddr, resdat, titletext );
- ParseMisc::parseBODYdatText( PARSEMODE_HTML, resdat.body, bodytext );
-
- result = QString( "<div class=\"pop_res_block\"><table class=\"pop_res_title\"><tr>" );
- result += titletext;
- result += "</tr></table><div class=\"pop_res_body\">";
- result += bodytext;
- result += "</div></div>";
-
- return result;
-}
-
-
-
-/*--------------------------------------------*/
-/* get HTML from raw data */ /* public */
-
-QString ParseMisc::DatToHtml( const QString& rawData, int num, bool showAddr )
-{
- QString tmpstr;
- RESDAT resdat;
-
- resdat.linestr = rawData;
- resdat.set = TRUE;
- resdat.parsed = FALSE;
- parseResDat( resdat, tmpstr );
-
- return ResDatToHtml( resdat, num, showAddr );
-}
-
-
-
-
-/*-------------------------------------------------------*/
-/*-------------------------------------------------------*/
-
-/* public utils */
-
-
-/*------------------------------------------*/
-/* parsing function for anchor (>>digits) */
-/*------------------------------------------*/ /* public */
-
-/* This fuction parses res anchor.
-
- For example, if cdat = ">12-20", then
-
- linkstr = ">12-20",
- refNum[0] = 12,
- refNum[1] = 20,
- pos (= length of cdat ) = 9,
- ret = TRUE;
-
-*/
-
-bool ParseMisc::parseResAnchor(
-
- /* input */
- const QChar *cdat, const unsigned int length,
-
- /* output */
- QString& linkstr, int* refNum, unsigned int& pos )
-{
-
- struct LocalFunc {
- static bool isHYPHEN( unsigned short c )
- {
-
- /* UTF-16 */
- if ( c == '-'
- || ( c >= 0x2010 && c <= 0x2015 )
- || ( c == 0x2212 )
- || ( c == 0xFF0D ) /* UTF8: 0xEFBC8D */
- ) {
- return TRUE;
- }
-
- return FALSE;
- }
- };
-
- bool ret = FALSE;
- int i;
-
- if ( length == 0 ) return FALSE;
-
- linkstr = QString::null;
- refNum[ 0 ] = 0;
- refNum[ 1 ] = 0;
- pos = 0;
-
- /* check '>' twice */
- for ( i = 0;i < 2;i++ ) {
-
- if ( cdat[ pos ].unicode() == UTF16_BRACKET ) {
- linkstr += cdat[ pos ];
- pos++;
- } else if ( cdat[ pos ] == '&' && cdat[ pos + 1 ] == 'g' /* > */
- && cdat[ pos + 2 ] == 't' && cdat[ pos + 3 ] == ';' ) {
- linkstr += ">";
- pos += 4;
- }
-
- }
-
- /* check ',' */
- if ( !pos ) {
- if ( cdat[ pos ] == ',' || cdat[ pos ].unicode() == UTF16_COMMA ) {
- linkstr += ",";
- pos ++;
- }
- }
-
- /* check '=' */
- if ( !pos ) {
- if ( cdat[ pos ] == '=' || cdat[ pos ].unicode() == UTF16_EQ ) {
- linkstr += "=";
- pos ++;
- }
- }
-
- /* check digits */
- int hyphen = 0;
-
- for ( i = 0 ; i < KITA_RESDIGIT + 1 && pos < length ; i++, pos++ ) {
-
- unsigned short c = cdat[ pos ].unicode();
-
- if ( ( c < UTF16_0 || c > UTF16_9 )
- && ( c < '0' || c > '9' )
- && ( !LocalFunc::isHYPHEN( c )
- || ( i == 0 && LocalFunc::isHYPHEN( c ) )
- || ( hyphen && LocalFunc::isHYPHEN( c ) ) )
- ) break;
-
- linkstr += cdat[ pos ];
-
- if ( LocalFunc::isHYPHEN( c ) ) {
- hyphen = 1;
- i = -1;
- } else {
- if ( c >= UTF16_0 ) c = '0' + cdat[ pos ].unicode() - UTF16_0;
- refNum[ hyphen ] *= 10;
- refNum[ hyphen ] += c - '0';
- }
-
- ret = TRUE;
- }
-
- return ret;
-}
-
-
-
-/* convert strings to positive number. */
-/* if cdat is not number, return -1. */
-
-/* For example, if cdat = "1234", then
- ret = 1234. If cdat = "abcd", then
- ret = -1. */ /* public */
-int ParseMisc::stringToPositiveNum( const QChar *cdat, const unsigned int length )
-{
- int ret = 0;
-
- for ( unsigned int i = 0 ; i < length ; i++ ) {
-
- unsigned short c = cdat[ i ].unicode();
-
- if ( ( c < UTF16_0 || c > UTF16_9 ) && ( c < '0' || c > '9' ) ) return -1;
-
- ret *= 10;
- if ( c >= UTF16_0 ) ret += c - UTF16_0;
- else ret += c - '0';
- }
-
- return ret;
-}
-
-
-
-/*-----------------------------------------------------*/
-/* parsing function for special char (such as ♥ */
-/*-----------------------------------------------------*/ /* public */
-
-/* For example, if cdat = "&", then
-
- pos (= length of cdat) = 5,
- retstr = "&".
-*/
-
-QString ParseMisc::parseSpecialChar(
-
- /* input */
- const QChar *cdat,
-
- /* output */
- unsigned int& pos )
-{
-
- struct LocalFunc {
- static int isEqual( const QChar *cdat, const QString& str )
- {
- int i = 0;
- while ( str.at( i ) != '\0' ) {
- if ( *cdat != str.at( i ) ) return 0;
- cdat++;i++;
- }
- return i;
- }
- };
-
- QString retstr = QString::null;
-
- if ( ( pos = LocalFunc::isEqual( cdat , ">" ) ) ) retstr = ">";
- else if ( ( pos = LocalFunc::isEqual( cdat , "<" ) ) ) retstr = "<";
- else if ( ( pos = LocalFunc::isEqual( cdat , " " ) ) ) retstr = " ";
- else if ( ( pos = LocalFunc::isEqual( cdat , "&" ) ) ) retstr = "&";
- else if ( ( pos = LocalFunc::isEqual( cdat , """ ) ) ) retstr = "\"";
-
- else if ( ( pos = LocalFunc::isEqual( cdat , "♥" ) ) )
- retstr = utf8ToUnicode( KITAUTF8_HEART );
-
- else if ( ( pos = LocalFunc::isEqual( cdat , "♦" ) ) )
- retstr = utf8ToUnicode( KITAUTF8_DIA );
-
- else if ( ( pos = LocalFunc::isEqual( cdat , "♣" ) ) )
- retstr = utf8ToUnicode( KITAUTF8_CLUB );
-
- else if ( ( pos = LocalFunc::isEqual( cdat , "♠" ) ) )
- retstr = utf8ToUnicode( KITAUTF8_SPADE );
-
- return retstr;
-}
-
-
-
-/* copied from comment.cpp */ /* public */
-void ParseMisc::parseDateId(
-
- /* input */
- const QString& str,
-
- /* output */
- RESDAT& resdat )
-{
- QRegExp regexp( "(\\d\\d(\\d\\d)?)/(\\d\\d)/(\\d\\d) (\\d\\d):(\\d\\d)(:\\d\\d)?( ID:(.*))?" );
-
- if ( regexp.search( str ) == -1 ) {
- resdat.date = str;
- resdat.id = QString::null;
- return ;
- }
-
- int year = regexp.cap( 1 ).toInt();
- if ( year >= 70 && year < 100 ) {
- year += 1900;
- } else if ( year < 70 ) {
- year += 2000;
- }
-
- if ( m_weekstr[ 0 ] == QString::null ) {
- m_weekstr[ 0 ] = utf8ToUnicode( KITAUTF8_MONDAY );
- m_weekstr[ 1 ] = utf8ToUnicode( KITAUTF8_TUESDAY );
- m_weekstr[ 2 ] = utf8ToUnicode( KITAUTF8_WEDNESDAY );
- m_weekstr[ 3 ] = utf8ToUnicode( KITAUTF8_THURSDAY );
- m_weekstr[ 4 ] = utf8ToUnicode( KITAUTF8_FRIDAY );
- m_weekstr[ 5 ] = utf8ToUnicode( KITAUTF8_SATURDAY );
- m_weekstr[ 6 ] = utf8ToUnicode( KITAUTF8_SUNDAY );
- }
-
- resdat.dateTime = QDateTime( QDate( year, regexp.cap( 3 ).toInt(), regexp.cap( 4 ).toInt() ),
- QTime( regexp.cap( 5 ).toInt(), regexp.cap( 6 ).toInt(), regexp.cap( 7 ).mid( 1 ).toInt() ) );
-
- resdat.date = regexp.cap( 1 ) + "/" + regexp.cap( 3 ) + "/" + regexp.cap( 4 )
- + m_weekstr[ resdat.dateTime.date().dayOfWeek() - 1 ]
- + regexp.cap( 5 ) + ":" + regexp.cap( 6 ) + regexp.cap( 7 );
-
- resdat.id = regexp.cap( 9 );
-}
-
-
-/*--------------------------------------*/
-/* split raw data, then get ID, name,
- date, body text, subject, etc.
-
- (input) "resdat.linestr","resdat.set"
-
- "resdat.linestr" is raw data, and
- "resdat.set" should be set to TRUE
- before calling this.
-
- struct RESDAT is defined in datinfo.h.
- see also datinfo.h. */ /* public */
-bool ParseMisc::parseResDat( RESDAT& resdat, QString& subject )
-{
- if ( resdat.parsed ) return TRUE;
- if ( !resdat.set ) return FALSE;
-
- resdat.parsed = TRUE;
-
- /* split dat */
- QString idstr = "none";
- QStringList list = QStringList::split( "<>", resdat.linestr, true );
-
- if ( list.size() == 5 ) {
-
- resdat.broken = FALSE;
-
- resdat.name = list[ 0 ];
- parseBODYdatText( PARSEMODE_TEXT, resdat.name, resdat.parsedName );
- resdat.address = list[ 1 ];
- parseDateId( list[ 2 ], resdat );
- if ( list[ 3 ].at( 0 ) == ' ' ) resdat.body = list[ 3 ].mid( 1 ); /* remove space after <> */
- else resdat.body = list[ 3 ];
-
- /* get subject */
- if ( list[ 4 ] != QString::null ) {
- subject = list[ 4 ];
- }
-
- } else resdat.broken = TRUE;
-
- return TRUE;
-}
-
-
-
-/*-----------------------------*/
-/* parsing function for link */
-
-/* For example,
-
- cdat = "ttp://foo.com",
-
- then
-
- linkstr = "ttp://foo.com",
- linkurl = "http://foo.com",
- pos (= length of cdat) = 13,
-
- and return TRUE.
- */ /* public */
-bool ParseMisc::parseLink(
-
- /* input */
- const QChar *cdat, const unsigned int length,
-
- /* output */
- QString& linkstr, QString& linkurl, unsigned int& pos
-)
-{
-
- /*-----------------------------*/
-
- linkstr = QString::null;
- linkurl = QString::null;
-
- QString retlinkstr = QString::null;
- QString prefix = QString::null;
-
- if ( isEqual( cdat , "http://" ) ) prefix = "http://";
- else if ( isEqual( cdat , "ttp://" ) ) prefix = "ttp://";
- else if ( isEqual( cdat , "tp://" ) ) prefix = "tp://";
- if ( prefix == QString::null ) return FALSE;
-
- pos = prefix.length();
- while ( cdat[ pos ] >= '!' && cdat[ pos ] <= '~' &&
- cdat[ pos ] != ' ' && cdat[ pos ] != '<' && cdat[ pos ] != '>'
- && pos < length ) {
- retlinkstr += cdat[ pos++ ];
- }
- if ( pos > length ) return FALSE;
-
- if ( retlinkstr != QString::null ) parseBODYdatText( PARSEMODE_TEXT, retlinkstr, linkstr );
- linkurl = "http://" + linkstr;
- linkstr = prefix + linkstr;
-
- return TRUE;
-}
-
-
-
-
-/* convert thread URL, and get reference.
- if mode = 0, output is URL of dat file. If mode = 1, output is URL of read.cgi .
-
-(ex.1)
-
-mode = 0
-url = http://pc5.2ch.net/linux/dat/1069738960.dat#20-30
-->
-return : http://pc5.2ch.net/linux/dat/1069738960.dat
-retstr : 20-30
-
-(ex.2)
-
-mode = 1
-url = http://pc5.2ch.net/linux/dat/1069738960.dat#20-30
-->
-return : http://pc5.2ch.net/test/read.cgi/linux/1069738960
-retstr : 20-30
-
-(ex.3)
-
-mode = 0
-url = http://pc5.2ch.net/test/read.cgi/linux/1069738960/-100
-->
-return : http://pc5.2ch.net/linux/dat/1069738960.dat
-refstr : 1-100
- */ /* public */
-
-QString ParseMisc::convertURL( int mode, /* if 0, output is dat URL. If 1, output is thread URL */
- const KURL& url , QString& refstr )
-{
- /* Is board enrolled ? */
- BoardData* bdata = Kita::BoardManager::getBoardData( url );
- if( bdata == NULL ) return url.prettyURL();
-
- QString urlstr = url.prettyURL();
-
- /* get thread & reference */
- QString thread = QString::null;
- QString refBase = QString::null;;
-
- if( urlstr.contains( "/dat/" ) ){
-
- /* url = (hostname)/(rootPath)/(bbsPath)/dat/(thread_ID).(ext)#(refBase) */
- thread = url.filename().remove( bdata->ext() );
- refBase = url.ref();
- }
- else if( urlstr.contains( bdata->delimiter() ) ) {
-
- QString tmpstr;
- switch( bdata->type() ){
-
- /* url = (hostname)/(rootPath)/(delimiter)/(bbsPath)/(thread_ID)/(refBase) */
- case Board_2ch:
- case Board_JBBS:
- tmpstr = urlstr.section( bdata->delimiter() + bdata->bbsPath(), 1, 1 );
- thread = tmpstr.section( '/', 1, 1 );
- refBase = tmpstr.section( '/', 2, 2 );
- break;
-
- /* machi BBS */
- /* ex.) If url = http://kanto.machi.to/bbs/read.pl?BBS=kana&KEY=1096716679 ,
- then, thread = 1096716679 */
- case Board_MachiBBS:
- thread = url.queryItem( "KEY" );
- refBase = QString::null;
- break;
- }
- }
-
- if( thread == QString::null ){
- if( mode == 0 ) return bdata->basePath();
- return bdata->cgiBasePath();
- }
-
- if ( refBase != QString::null ){
-
- if ( refBase.at( 0 ) == '-' ) refstr = "1" + refBase;
- else refstr = refBase;
- }
-
- /* create new URL */
- QString newURL;
- if( mode == 0 ) newURL = bdata->basePath() + "dat/" + thread + bdata->ext();
- else {
- newURL = bdata->cgiBasePath();
-
- switch( bdata->type() ){
- case Board_2ch:
- case Board_JBBS:
- newURL += thread;
- break;
-
- case Board_MachiBBS:
- newURL += "&KEY=" + thread;
- break;
- }
- }
-/*
- qDebug( "convertURL (mode %d)", mode );
- qDebug( "input = %s", url.prettyURL().ascii() );
- qDebug( "output = %s", newURL.ascii() );
- qDebug( "thread = %s", thread.ascii() );
- qDebug( "ref = %s", refBase.ascii() );
-*/
- return newURL;
-}
-
-KURL ParseMisc::parseURL( const KURL& url , QString& refstr )
-{
- return convertURL( 0, url, refstr );
-}
-
-/* for convenience */
-KURL ParseMisc::parseURLonly( const KURL& url )
-{
- QString refstr;
- return convertURL( 0, url, refstr );
-}
-
-
-/*--------------------------------------------*/
-/* private functions */
-
-
-/* if cdat == str, return str.length() */ /* private */
-int ParseMisc::isEqual( const QChar *cdat, const QString& str )
-{
- int i = 0;
- while ( str.at( i ) != '\0' ) {
- if ( *cdat != str.at( i ) ) return 0;
- cdat++;i++;
- }
- return i;
-}
-
-
-
-
-
-/*------------------------*/
-/* create node of anchor */
-
-/* This function is called from parseBODYdat
- internally. So, see also parseBODYdat */ /* private */
-bool ParseMisc::createResAnchor(
-
- /* input */
- int mode,
- const QString &rawStr,
- DOM::HTMLDocument& hdoc ,
-
- /* output */
- DOM::Element &bodynode,
- QString& bodytext,
-
- /* internal variables */
- /* They are the same variables that ara used in parseBODYdat. */
- const QChar *chpt, unsigned int &i, unsigned int &index, QString& lineStr
-)
-{
- /*-----------------------*/
-
- QString linkstr, linkurl;
- DOM::Element tmpelm;
- int refNum[ 2 ];
- unsigned int pos;
- unsigned int length = rawStr.length();
-
- /* parse anchor */
- if ( !parseResAnchor( chpt + i, length - i, linkstr, refNum, pos ) ) {
- lineStr += rawStr.mid( index, i - index ) + linkstr;
- index = i + pos;
- i = index - 1;
- return FALSE;
- }
-
- /* create anchor */
- lineStr += rawStr.mid( index, i - index );
- linkurl = QString( "#%1" ).arg( refNum[ 0 ] );
- if ( refNum[ 1 ] ) linkurl += QString( "-%1" ).arg( refNum[ 1 ] );
-
- switch ( mode ) {
-
- case PARSEMODE_DOM:
-
- /* create 'A' element */
- bodynode.appendChild( hdoc.createTextNode( lineStr ) );
-
- tmpelm = bodynode.appendChild( hdoc.createElement( "A" ) );
- {
- tmpelm.setAttribute( "href", linkurl );
- tmpelm.appendChild( hdoc.createTextNode( linkstr ) );
- }
-
- break;
-
- case PARSEMODE_HTML:
- bodytext += lineStr;
- bodytext += "<a href=\"" + linkurl + "\">";
- bodytext += linkstr;
- bodytext += "</a>";
- break;
- }
-
- index = i + pos;
- i = index - 1;
- lineStr = QString::null;
-
- return TRUE;
-}
-
-
-
-
-/*---------------------------------------------*/
-
-/* obsolete */
-
-
-QString ParseMisc::qcpToUnicode( const QString& str )
-{
- return Kita::qcpToUnicode( str );
-}
-
-
-QString ParseMisc::utf8ToUnicode( const QString& str )
-{
- return Kita::utf8ToUnicode( str );
-}
-
OSDN Git Service
