Page MenuHomePhorge
Files F114740

kazv-markdown.cpp
No OneTemporary
Actions

Size
5 KB
Referenced Files
None
Subscribers
None

kazv-markdown.cpp
View Options

/*
* This file is part of kazv.
* SPDX-FileCopyrightText: 2024 tusooa <tusooa@kazv.moe>
* SPDX-License-Identifier: AGPL-3.0-or-later
*/
#include <kazv-defs.hpp>
#include <unordered_set>
#include <boost/algorithm/string.hpp>
#include <boost/regex.hpp>
#include <cmark.h>
#include <QDebug>
#include "kazv-markdown.hpp"
static const std::string serverNameRegexStr = R"aaaa([A-Za-z0-9.-]{1,255}(:[0-9]{1,5})?)aaaa";
static const auto userIdRegex = boost::regex(
// The char before must not be a non-space
// This means either it is the beginning of the string or a space char
R"aaaa((?<!\S)@[a-z0-9._=/+-]+:)aaaa"
+ serverNameRegexStr
// The char after must not be a non-space
// This means either it is the end of the string or a space char
+ R"((?!\S))");
static std::string makeUserLink(const std::string &userId)
{
return "https://matrix.to/#/" + userId;
}
struct MarkdownParser
{
MarkdownParser(std::string markdown)
: m_doc(cmark_parse_document(markdown.data(), markdown.size(), CMARK_OPT_UNSAFE))
{
linkify();
m_html = cmark_render_html(m_doc, CMARK_OPT_UNSAFE);
}
~MarkdownParser()
{
cmark_node_free(m_doc);
free(m_html);
}
void linkify()
{
cmark_iter *iter = cmark_iter_new(m_doc);
for (
cmark_event_type evType = cmark_iter_next(iter);
evType != CMARK_EVENT_DONE;
evType = cmark_iter_next(iter)
) {
cmark_node *cur = cmark_iter_get_node(iter);
maybeLinkifyNode(cur, evType);
}
cmark_iter_free(iter);
}
bool isInsideLink(cmark_node *cur)
{
if (!cur) {
return false;
}
if (cmark_node_get_type(cur) == CMARK_NODE_LINK) {
return true;
}
return isInsideLink(cmark_node_parent(cur));
}
void maybeLinkifyNode(cmark_node *cur, cmark_event_type evType)
{
auto nodeType = cmark_node_get_type(cur);
if (!((nodeType == CMARK_NODE_HTML_BLOCK || nodeType == CMARK_NODE_TEXT)
&& evType == CMARK_EVENT_ENTER) || isInsideLink(cur)) {
return;
}
auto textContent = cmark_node_get_literal(cur);
if (!textContent) {
return;
}
auto contentStr = std::string(textContent);
if (nodeType == CMARK_NODE_HTML_BLOCK) {
auto replacedStr = boost::regex_replace(
contentStr,
userIdRegex,
[this](const boost::smatch &match) {
auto matched = match.str();
m_mentions.insert(matched);
return "<a href=\"" + makeUserLink(matched) + "\">" + matched + "</a>";
}
);
if (contentStr != replacedStr) {
cmark_node_set_literal(cur, replacedStr.data());
}
} else {
boost::smatch match;
auto firstNode = cmark_node_new(CMARK_NODE_TEXT);
cmark_node_set_literal(firstNode, "");
cmark_node_replace(cur, firstNode);
cmark_node_free(cur);
auto lastNode = firstNode;
auto appendTextNode = [&lastNode](const std::string &str) {
auto textNode = cmark_node_new(CMARK_NODE_TEXT);
cmark_node_set_literal(textNode, str.data());
cmark_node_insert_after(lastNode, textNode);
lastNode = textNode;
};
auto appendLinkNode = [&lastNode](const std::string &str, const std::string &linkTarget) {
auto linkNode = cmark_node_new(CMARK_NODE_LINK);
cmark_node_set_url(linkNode, linkTarget.data());
cmark_node_insert_after(lastNode, linkNode);
lastNode = linkNode;
auto textNode = cmark_node_new(CMARK_NODE_TEXT);
cmark_node_set_literal(textNode, str.data());
cmark_node_append_child(linkNode, textNode);
};
while (boost::regex_search(contentStr, match, userIdRegex)) {
if (match.position()) {
auto preMatchStr = match.prefix().str();
appendTextNode(preMatchStr);
}
auto matched = match.str();
m_mentions.insert(matched);
appendLinkNode(matched, makeUserLink(matched));
contentStr = match.suffix().str();
};
if (!contentStr.empty()) {
appendTextNode(contentStr);
}
}
}
std::string getHtml() const
{
auto htmlStr = std::string(m_html);
boost::algorithm::trim(htmlStr);
return htmlStr;
}
immer::flex_vector<std::string> getMentions() const
{
return immer::flex_vector<std::string>(
m_mentions.begin(), m_mentions.end()
);
}
cmark_node *m_doc;
char *m_html;
std::unordered_set<std::string> m_mentions;
};
KazvRichTextParseResult markdownToHtml(const std::string &markdown)
{
auto parser = MarkdownParser(markdown);
return { parser.getHtml(), parser.getMentions() };
}

File Metadata

Mime Type
text/x-c
Expires
Tue, Nov 26, 2:08 PM (1 d, 12 h)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
40365
Default Alt Text
kazv-markdown.cpp (5 KB)

Event Timeline