Parser: Add callback support for unknown "tags"

This commit is contained in:
Albert S. 2022-03-27 08:31:59 +02:00
джерело 44c27ed8b4
коміт a4a45d9add
3 змінених файлів з 23 додано та 7 видалено

@ -2,15 +2,27 @@
#define IPARSER_H #define IPARSER_H
#include <vector> #include <vector>
#include <string_view> #include <string_view>
#include <functional>
#include "headline.h" #include "headline.h"
#include "database/pagedao.h" #include "database/pagedao.h"
#include "urlprovider.h" #include "urlprovider.h"
class IParser class IParser
{ {
protected:
static std::string empty(std::string_view key, std::string_view content)
{
return "";
}
public: public:
virtual std::string extractCommand(std::string cmdname, const std::string &content) const = 0; virtual std::string extractCommand(std::string cmdname, const std::string &content) const = 0;
virtual std::vector<Headline> extractHeadlines(const std::string &content) const = 0; virtual std::vector<Headline> extractHeadlines(const std::string &content) const = 0;
virtual std::string parse(const PageDao &pagedao, UrlProvider &provider, const std::string &content) const = 0; virtual inline std::string parse(const PageDao &pagedao, UrlProvider &provider, const std::string &content) const
{
return parse(pagedao, provider, content, empty);
}
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::vector<std::string> extractCategories(const std::string &content) const = 0; virtual std::vector<std::string> extractCategories(const std::string &content) const = 0;
virtual ~IParser(){}; virtual ~IParser(){};

@ -116,12 +116,13 @@ std::string Parser::processLink(const PageDao &pageDao, UrlProvider &urlProvider
return htmllink.render(); return htmllink.render();
} }
std::string Parser::parse(const PageDao &pagedao, UrlProvider &provider, const std::string &content) const std::string Parser::parse(const PageDao &pagedao, UrlProvider &provider, const std::string &content,
const std::function<std::string(std::string_view, std::string_view)> &callback) const
{ {
std::string result; std::string result;
// we don't care about commands, but we nevertheless replace them with empty strings // we don't care about commands, but we nevertheless replace them with empty strings
std::regex tagfinder( std::regex tagfinder(
R"(\[(b|i|u|li||ul|ol|link|wikilink|h\d|cmd:rename|cmd:redirect|cmd:pagetitle|category)*?\]((\s|\S)*?)\[/\1])"); R"(\[(b|i|u|li||ul|ol|link|wikilink|h\d|cmd:rename|cmd:redirect|cmd:pagetitle|category|dynamic:postlist)*?\]((\s|\S)*?)\[/\1])");
result = utils::regex_callback_replacer( result = utils::regex_callback_replacer(
tagfinder, content, tagfinder, content,
[&](std::smatch &match) [&](std::smatch &match)
@ -129,7 +130,7 @@ std::string Parser::parse(const PageDao &pagedao, UrlProvider &provider, const s
std::string tag = match.str(1); std::string tag = match.str(1);
std::string content = match.str(2); std::string content = match.str(2);
std::string justreplace[] = {"b", "i", "u", "ul", "li", "ol"}; std::string justreplace[] = {"b", "i", "u", "ul", "li", "ol"};
content = parse(pagedao, provider, content); content = parse(pagedao, provider, content, callback);
if(std::find(std::begin(justreplace), std::end(justreplace), tag) != std::end(justreplace)) if(std::find(std::begin(justreplace), std::end(justreplace), tag) != std::end(justreplace))
{ {
return "<" + tag + ">" + content + "</" + tag + ">"; return "<" + tag + ">" + content + "</" + tag + ">";
@ -144,7 +145,7 @@ std::string Parser::parse(const PageDao &pagedao, UrlProvider &provider, const s
{ {
return "<" + tag + " id='" + content + "'>" + content + "</" + tag + ">"; return "<" + tag + " id='" + content + "'>" + content + "</" + tag + ">";
} }
return std::string(""); return callback(tag, content);
}); });
result = utils::strreplace(result, "\r\n", "<br>"); result = utils::strreplace(result, "\r\n", "<br>");
return result; return result;

@ -11,9 +11,12 @@ class Parser : public IParser
std::string extractCommand(std::string cmdname, const std::string &content) const; std::string extractCommand(std::string cmdname, const std::string &content) const;
std::vector<Headline> extractHeadlines(const std::string &content) const override; std::vector<Headline> extractHeadlines(const std::string &content) const override;
std::vector<std::string> extractCategories(const std::string &content) const override; std::vector<std::string> extractCategories(const std::string &content) const override;
std::string parse(const PageDao &pagedao, UrlProvider &provider, const std::string &content) const override; using IParser::parse;
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;
using IParser::IParser; using IParser::IParser;
~Parser(){};
}; };
#endif // PARSER_H #endif // PARSER_H