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

dont throw away text even if html is weird

Browse Source
This commit is contained in:
dockes 2006-09-21 05:59:59 +00:00
parent 507ee32fdb
commit e3f89dca7e
1 changed files with 14 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
src/internfile/myhtmlparse.cpp
View File

@ -167,7 +167,9 @@ MyHtmlParser::opening_tag(const string &tag, const map<string,string> &p)
sprintf(ascuxtime, "%ld", (long)mktime(&tm)); sprintf(ascuxtime, "%ld", (long)mktime(&tm));
dmtime = ascuxtime; dmtime = ascuxtime;
} }
} else if (name == "robots") { }
#if 0 // We're not a robot, so we don't care about robots metainfo
else if (name == "robots") {
string val = i->second; string val = i->second;
decode_entities(val); decode_entities(val);
lowercase_term(val); lowercase_term(val);
@ -178,6 +180,7 @@ MyHtmlParser::opening_tag(const string &tag, const map<string,string> &p)
throw false; throw false;
} }
} }
#endif // 0
} else if ((j = p.find("http-equiv")) != p.end()) { } else if ((j = p.find("http-equiv")) != p.end()) {
string hequiv = j->second; string hequiv = j->second;
lowercase_term(hequiv); lowercase_term(hequiv);
@ -332,13 +335,17 @@ MyHtmlParser::closing_tag(const string &tag)
} }
} }
// This gets called when hitting eof. If the <body> is open, do // This gets called when hitting eof.
// something with the text (that is, don't throw up). Else, things are // We used to do:
// too weird, throw an error. We don't get called if the parser finds // > If the <body> is open, do
// a closing body tag (exception gets thrown by closing_tag()) // > something with the text (that is, don't throw up). Else, things are
// > too weird, throw an error. We don't get called if the parser finds
// > a closing body tag (exception gets thrown by closing_tag())
// But we don't throw any more. Whatever text we've extracted up to now is
// better than nothing.
void void
MyHtmlParser::do_eof() MyHtmlParser::do_eof()
{ {
if (!in_body_tag) // if (!in_body_tag)
throw(false); // throw(false);
} }