crtxcr/qswiki
1
0
Garpu 0
Kode Masalah 21 Tarik Permintaan 1 Rilis Kegiatan

Parser: Seperate parseDynamcis(), fix headline extraction with tags inside them

Telusuri Sumber
This commit is contained in:
Albert S. 2022-03-29 22:35:45 +02:00
orang tua 15e4f081cc
melakukan fbca85e5ed
3 mengubah file dengan 19 tambahan dan 3 penghapusan
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
iparser.h
Melihat File

@ -23,6 +23,10 @@ class IParser
}
virtual std::string parse(const PageDao &pagedao, UrlProvider &provider, const std::string &content,
const std::function<std::string(std::string_view, std::string_view)> &callback) const = 0;
virtual std::string parseDynamics(
const std::string &content,
const std::function<std::string(std::string_view, std::string_view)> &callback) const = 0;
virtual std::vector<std::string> extractCategories(const std::string &content) const = 0;
virtual ~IParser(){};

15
parser.cpp
Melihat File

@ -30,7 +30,8 @@ SOFTWARE.
std::vector<Headline> Parser::extractHeadlines(const std::string &content) const
{
std::vector<Headline> result;
std::string reg = R"(\[h(1|2|3)\](.*?)\[/h\1\])";
std::string reg = R"(\[h(1|2|3)\](\[.*?\])*(.*?)\[.*?\]*\[\/h\1\])";
std::regex headerfinder(reg);
auto begin = std::sregex_iterator(content.begin(), content.end(), headerfinder);
auto end = std::sregex_iterator();
@ -40,7 +41,7 @@ std::vector<Headline> Parser::extractHeadlines(const std::string &content) const
auto smatch = *it;
Headline h;
h.level = utils::toUInt(smatch.str(1));
h.title = smatch.str(2);
h.title = smatch.str(3);
result.push_back(h);
}
return result;
@ -122,7 +123,7 @@ std::string Parser::parse(const PageDao &pagedao, UrlProvider &provider, const s
std::string result;
// we don't care about commands, but we nevertheless replace them with empty strings
std::regex tagfinder(
R"(\[(b|i|u|li||ul|ol|link|wikilink|h\d|cmd:rename|cmd:redirect|cmd:pagetitle|category|dynamic:postlist|dynamic:includepage)*?\]((\s|\S)*?)\[/\1])");
R"(\[(b|i|u|li||ul|ol|link|wikilink|h\d|cmd:rename|cmd:redirect|cmd:pagetitle|category|dynamic:postlist|dynamic:includepage|dynamic:getvar|dynamic:setvar)*?\]((\s|\S)*?)\[/\1])");
result = utils::regex_callback_replacer(
tagfinder, content,
[&](std::smatch &match)
@ -150,3 +151,11 @@ std::string Parser::parse(const PageDao &pagedao, UrlProvider &provider, const s
result = utils::strreplace(result, "\r\n", "<br>");
return result;
}
std::string Parser::parseDynamics(const std::string &content,
const std::function<std::string(std::string_view, std::string_view)> &callback) const
{
std::regex tagfinder(R"(\[(dynamic:\w+)*?\]((\s|\S)*?)\[/\1])");
return utils::regex_callback_replacer(tagfinder, content,
[&](std::smatch &match) { return callback(match.str(1), match.str(2)); });
}

3
parser.h
Melihat File

@ -15,6 +15,9 @@ class Parser : public IParser
virtual std::string parse(
const PageDao &pagedao, UrlProvider &provider, const std::string &content,
const std::function<std::string(std::string_view, std::string_view)> &callback) const override;
std::string parseDynamics(
const std::string &content,
const std::function<std::string(std::string_view, std::string_view)> &callback) const override;
using IParser::IParser;
};