tris/recoll
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

previous html fix didnt work

Browse Source
This commit is contained in:
dockes 2005-12-06 09:40:18 +00:00
parent d2b54d6af2
commit c8e18ccc81
2 changed files with 10 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
src/internfile/mh_html.cpp
View File

@ -101,8 +101,12 @@ MimeHandlerHtml::mkDoc(RclConfig *conf, const string &,
try { try {
p.parse_html(transcoded); p.parse_html(transcoded);
} catch (bool) { } catch (bool diag) {
pres = p; pres = p;
if (diag == true)
break;
LOGDEB(("textHtmlToDoc: charset [%s] doc charset [%s]\n",
charset.c_str(),pres.doccharset.c_str()));
if (!pres.doccharset.empty() && if (!pres.doccharset.empty() &&
!samecharset(pres.doccharset, pres.ocharset)) { !samecharset(pres.doccharset, pres.ocharset)) {
LOGDEB(("textHtmlToDoc: charset '%s' doc charset '%s'," LOGDEB(("textHtmlToDoc: charset '%s' doc charset '%s',"

6
src/internfile/myhtmlparse.cpp
View File

@ -156,7 +156,8 @@ MyHtmlParser::opening_tag(const string &tag, const map<string,string> &p)
if (val.find("none") != string::npos || if (val.find("none") != string::npos ||
val.find("noindex") != string::npos) { val.find("noindex") != string::npos) {
indexing_allowed = false; indexing_allowed = false;
throw true; LOGDEB1(("myhtmlparse: robots/noindex\n"));
throw false;
} }
} }
} else if ((j = p.find("http-equiv")) != p.end()) { } else if ((j = p.find("http-equiv")) != p.end()) {
@ -175,7 +176,7 @@ MyHtmlParser::opening_tag(const string &tag, const map<string,string> &p)
"differs from announced '%s'\n", "differs from announced '%s'\n",
doccharset.c_str(), doccharset.c_str(),
ocharset.c_str())); ocharset.c_str()));
throw true; throw false;
} }
} }
} }
@ -232,6 +233,7 @@ MyHtmlParser::closing_tag(const string &tag)
break; break;
case 'b': case 'b':
if (tag == "body") { if (tag == "body") {
LOGDEB1(("Myhtmlparse: body close tag found\n"));
throw true; throw true;
} }
if (tag == "blockquote" || tag == "br") pending_space = true; if (tag == "blockquote" || tag == "br") pending_space = true;