mdpreview/src/markdown.rs
2026-07-15 11:14:48 +03:00

513 lines
16 KiB
Rust

use pulldown_cmark::{
Alignment, BlockQuoteKind, CodeBlockKind, CowStr, Event, LinkType, Options, Parser, Tag,
TagEnd, TextMergeStream,
};
use std::collections::HashMap;
use std::fmt::Write;
// ========== Класс состояния таблицы ==========
#[derive(Debug, Clone, PartialEq)]
pub enum AlignmentState {
Left,
Center,
Right,
None,
}
impl From<Alignment> for AlignmentState {
fn from(alignment: Alignment) -> Self {
match alignment {
Alignment::Left => AlignmentState::Left,
Alignment::Center => AlignmentState::Center,
Alignment::Right => AlignmentState::Right,
Alignment::None => AlignmentState::None,
}
}
}
impl AlignmentState {
pub fn to_css_class(&self) -> &'static str {
match self {
AlignmentState::Left => "align-left",
AlignmentState::Center => "align-center",
AlignmentState::Right => "align-right",
AlignmentState::None => "",
}
}
}
#[derive(Debug, Clone, Default)]
pub struct TableState {
pub alignments: Vec<AlignmentState>,
pub column_count: usize,
pub is_inside_table: bool,
pub is_inside_head: bool,
pub is_inside_body: bool,
pub current_row: usize,
pub current_column: usize,
pub cell_data: HashMap<(usize, usize), String>,
pub has_header: bool,
}
impl TableState {
pub fn new() -> Self {
Self::default()
}
pub fn start_table(&mut self, alignments: Vec<Alignment>) {
self.alignments = alignments.into_iter().map(AlignmentState::from).collect();
self.column_count = self.alignments.len();
self.is_inside_table = true;
self.is_inside_head = false;
self.is_inside_body = false;
self.current_row = 0;
self.current_column = 0;
self.cell_data.clear();
self.has_header = false;
}
pub fn end_table(&mut self) {
self.is_inside_table = false;
self.is_inside_head = false;
self.is_inside_body = false;
}
pub fn start_head(&mut self) {
if self.is_inside_table {
self.is_inside_head = true;
self.is_inside_body = false;
self.current_row = 0;
self.current_column = 0;
self.has_header = true;
}
}
pub fn end_head(&mut self) {
self.is_inside_head = false;
}
pub fn start_row(&mut self) {
if self.is_inside_table {
if self.is_inside_body {
self.current_row += 1;
}
self.current_column = 0;
}
}
pub fn end_row(&mut self) {
if self.current_column != self.column_count && self.is_inside_table && self.column_count > 0
{
// Исправляем неполные строки, добавляя недостающие ячейки
while self.current_column < self.column_count {
let row = if self.is_inside_head {
0
} else {
self.current_row
};
let col = self.current_column;
self.cell_data.insert((row, col), String::new());
self.current_column += 1;
}
}
}
pub fn start_cell(&mut self) {
if self.is_inside_table {
self.current_column += 1;
}
}
pub fn end_cell(&mut self, data: Option<String>) {
if let Some(text) = data {
let row = if self.is_inside_head {
0
} else {
self.current_row
};
let col = self.current_column - 1;
self.cell_data.insert((row, col), text);
}
}
pub fn get_current_alignment(&self) -> &AlignmentState {
if self.current_column == 0 || self.current_column > self.alignments.len() {
return &AlignmentState::None;
}
let col_index = self.current_column - 1;
&self.alignments[col_index]
}
}
// ========== Основной рендерер ==========
pub struct MarkdownRenderer {
table_state: TableState,
in_code_block: bool,
code_content: String,
}
impl MarkdownRenderer {
pub fn new() -> Self {
Self {
table_state: TableState::new(),
in_code_block: false,
code_content: String::new(),
}
}
pub fn render_start_tag(&mut self, tag: Tag<'_>) -> String {
let mut output = String::new();
match tag {
Tag::Heading { level, .. } => {
output.push_str(format!("<{level}>").as_str());
}
Tag::Strikethrough => {
output.push_str("<strike>");
}
Tag::List(Some(_)) => output.push_str("<ol>"),
Tag::List(None) => output.push_str("<ul>"),
Tag::Item => output.push_str("<li>"),
Tag::Paragraph => output.push_str("<p>"),
Tag::CodeBlock(kind) => {
output.push_str(render_codeblock(kind).as_str());
}
Tag::Emphasis => output.push_str("<em>"),
Tag::Strong => output.push_str("<strong>"),
Tag::BlockQuote(kind) => output.push_str(render_blockquote(kind).as_str()),
Tag::Link {
link_type,
dest_url,
title,
..
} => output.push_str(render_link(link_type, &dest_url, &title).as_str()),
Tag::Image {
link_type,
dest_url,
title,
..
} => output.push_str(render_image(link_type, &dest_url, &title).as_str()),
Tag::Table(alignments) => {
self.table_state.start_table(alignments.clone());
output.push_str("<table>");
}
Tag::TableHead => {
self.table_state.start_head();
output.push_str("<thead>");
}
Tag::TableCell => {
self.table_state.start_cell();
let alignment = self.table_state.get_current_alignment();
let css_class = alignment.to_css_class();
if css_class.is_empty() {
output.push_str("<td>");
} else {
let _ = write!(output, "<td class=\"{css_class}\">");
}
}
Tag::TableRow => {
self.table_state.start_row();
output.push_str("<tr>");
}
Tag::FootnoteDefinition(fref) => {
let _ = write!(output, "<sup><a href=\"#{fref}\" id=\"{fref}\">");
}
Tag::HtmlBlock
| Tag::DefinitionList
| Tag::DefinitionListTitle
| Tag::DefinitionListDefinition
| Tag::Superscript
| Tag::Subscript
| Tag::MetadataBlock(_) => (),
}
output
}
pub fn render_end_tag(&mut self, tag: TagEnd) -> String {
let mut output = String::new();
match tag {
TagEnd::Heading(level) => {
output.push_str(format!("</{level}>").as_str());
}
TagEnd::Strikethrough => {
output.push_str("</strike>");
}
TagEnd::List(true) => output.push_str("</ol>"),
TagEnd::List(false) => output.push_str("</ul>"),
TagEnd::Item => output.push_str("</li>"),
TagEnd::Paragraph => output.push_str("</p>"),
TagEnd::CodeBlock => {
output.push_str("</code></pre></div>");
}
TagEnd::Emphasis => output.push_str("</em>"),
TagEnd::Strong => output.push_str("</strong>"),
TagEnd::BlockQuote(_) => output.push_str("</blockquote>"),
TagEnd::Link => output.push_str("</a>"),
TagEnd::Image => output.push_str("</img>"),
TagEnd::Table => {
self.table_state.end_table();
output.push_str("</table>");
}
TagEnd::TableHead => {
self.table_state.end_head();
output.push_str("</thead>");
if self.table_state.is_inside_table {
output.push_str("<tbody>");
self.table_state.is_inside_body = true;
}
}
TagEnd::TableCell => {
self.table_state.end_cell(None);
output.push_str("</td>");
}
TagEnd::TableRow => {
self.table_state.end_row();
output.push_str("</tr>");
}
TagEnd::FootnoteDefinition => {
output.push_str("</a></sup>");
}
TagEnd::HtmlBlock
| TagEnd::DefinitionList
| TagEnd::DefinitionListTitle
| TagEnd::DefinitionListDefinition
| TagEnd::Superscript
| TagEnd::Subscript
| TagEnd::MetadataBlock(_) => (),
}
output
}
}
// ========== Основная функция ==========
pub fn markdown_to_html(markdown: &str, _file_path: &str) -> String {
let mut options = Options::empty();
options.insert(Options::ENABLE_TABLES);
options.insert(Options::ENABLE_FOOTNOTES);
options.insert(Options::ENABLE_STRIKETHROUGH);
options.insert(Options::ENABLE_TASKLISTS);
options.insert(Options::ENABLE_HEADING_ATTRIBUTES);
options.insert(Options::ENABLE_GFM);
let mut renderer = MarkdownRenderer::new();
let mut html_output = String::new();
let iterator = TextMergeStream::new(Parser::new_ext(markdown, options));
for event in iterator {
match event {
Event::Start(tag) => {
if let Tag::CodeBlock(_) = tag {
renderer.in_code_block = true;
renderer.code_content.clear();
}
html_output.push_str(renderer.render_start_tag(tag).as_str());
}
Event::End(tag) => {
if let TagEnd::CodeBlock = tag {
renderer.in_code_block = false;
html_output.push_str(escape_html(&renderer.code_content).as_str());
}
html_output.push_str(renderer.render_end_tag(tag).as_str());
}
Event::Text(text) => {
if renderer.in_code_block {
renderer.code_content.push_str(&text);
} else {
html_output.push_str(&escape_html(&text));
}
}
Event::TaskListMarker(true) => {
html_output.push_str("<input type=\"checkbox\" checked/>");
}
Event::TaskListMarker(false) => {
html_output.push_str("<input type=\"checkbox\"/>");
}
Event::Code(code) => {
html_output.push_str(format!("<code>{}</code>", escape_html(&code)).as_str());
}
Event::SoftBreak => html_output.push(' '),
Event::HardBreak => html_output.push_str("<br>"),
Event::Html(html) => html_output.push_str(&html),
Event::Rule => html_output.push_str("<hr>"),
Event::FootnoteReference(fref) => {
let _ = write!(
html_output,
"<sup><a href=\"#{fref}\" id=\"{fref}\">{fref}</a></sup>"
);
}
Event::InlineHtml(_) | Event::InlineMath(_) | Event::DisplayMath(_) => (),
}
}
html_output
}
// ========== Вспомогательные функции ==========
fn render_codeblock(block_kind: CodeBlockKind<'_>) -> String {
let header = String::from(
r#"
<div class="code-block-wrapper">
<div class="code-header">
<span class="code-lang">%CODE%</span> <button class="copy-btn" onclick="copyCode(this)">Copy</button>
</div>
<pre>"#,
);
let mut output = String::new();
match block_kind {
CodeBlockKind::Indented => {
output.push_str(&header.replace("%CODE%", "code"));
output.push_str("<code>");
}
CodeBlockKind::Fenced(CowStr::Borrowed("mermaid")) => {
output.push_str(&header.replace("%CODE%", "mermaid"));
output.push_str("<code class=\"language-text mermaid\">");
}
CodeBlockKind::Fenced(CowStr::Borrowed("rawmermaid")) => {
output.push_str(&header.replace("%CODE%", "mermaid"));
output.push_str("<code class=\"language-text\">");
}
CodeBlockKind::Fenced(code_type) => {
output.push_str(&header.replace("%CODE%", &code_type));
output.push_str(format!("<code class=\"language-{code_type}\">").as_str());
}
}
output
}
fn render_blockquote(quote_kind: Option<BlockQuoteKind>) -> String {
let alert_class_and_title = match quote_kind {
None => return "<blockquote>".to_string(),
Some(BlockQuoteKind::Note) => ("note", "Note"),
Some(BlockQuoteKind::Tip) => ("tip", "Tip"),
Some(BlockQuoteKind::Important) => ("important", "Important"),
Some(BlockQuoteKind::Warning) => ("warning", "Warning"),
Some(BlockQuoteKind::Caution) => ("caution", "Caution"),
};
format!(
"<blockquote><div class=\"markdown-alert markdown-alert-{}\"><p class=\"markdown-alert-title\">{}</p>",
alert_class_and_title.0, alert_class_and_title.1
)
}
fn render_link(link_type: LinkType, dest_url: &CowStr<'_>, title: &CowStr<'_>) -> String {
match link_type {
LinkType::Email => format!("<a href=\"mailto:{dest_url}\" title=\"{title}\">"),
_ => format!("<a href=\"{dest_url}\" title=\"{title}\">"),
}
}
fn render_image(_link_type: LinkType, dest_url: &CowStr<'_>, title: &CowStr<'_>) -> String {
format!("<img src=\"{dest_url}\" alt=\"{title}\">")
}
fn escape_html(text: &str) -> String {
text.replace('&', "&amp;")
.replace('<', "&lt;")
.replace('>', "&gt;")
.replace('"', "&quot;")
.replace('\'', "&#39;")
}
// ========== Тесты ==========
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_table_with_alignments() {
let markdown = r#"
| Header 1 | Header 2 | Header 3 |
|----------|:--------:|---------:|
| Left | Center | Right |
| Data 1 | Data 2 | Data 3 |
"#;
let html = markdown_to_html(markdown, "");
// Проверяем, что HTML содержит правильные классы и атрибуты
assert!(html.contains("<table>"));
assert!(html.contains("<thead>"));
assert!(html.contains("<tbody>"));
assert!(html.contains("text-left"));
assert!(html.contains("text-center"));
assert!(html.contains("text-right"));
}
#[test]
fn test_table_state_tracking() {
let mut renderer = MarkdownRenderer::new();
let alignments = vec![Alignment::Left, Alignment::Center, Alignment::Right];
renderer.render_start_tag(Tag::Table(alignments.clone()));
assert!(renderer.table_state.is_inside_table);
assert_eq!(renderer.table_state.column_count, 3);
renderer.render_start_tag(Tag::TableHead);
assert!(renderer.table_state.is_inside_head);
renderer.render_start_tag(Tag::TableRow);
renderer.render_start_tag(Tag::TableCell);
assert_eq!(
renderer.table_state.get_current_alignment(),
&AlignmentState::Left
);
renderer.render_end_tag(TagEnd::TableCell);
renderer.render_start_tag(Tag::TableCell);
assert_eq!(
renderer.table_state.get_current_alignment(),
&AlignmentState::Center
);
renderer.render_end_tag(TagEnd::TableCell);
renderer.render_start_tag(Tag::TableCell);
assert_eq!(
renderer.table_state.get_current_alignment(),
&AlignmentState::Right
);
renderer.render_end_tag(TagEnd::TableCell);
renderer.render_end_tag(TagEnd::TableRow);
renderer.render_end_tag(TagEnd::TableHead);
// Проверяем, что тело было автоматически открыто
renderer.render_start_tag(Tag::TableRow);
renderer.render_start_tag(Tag::TableCell);
assert_eq!(
renderer.table_state.get_current_alignment(),
&AlignmentState::Left
);
renderer.render_end_tag(TagEnd::TableCell);
renderer.render_end_tag(TagEnd::TableRow);
renderer.render_end_tag(TagEnd::Table);
assert!(!renderer.table_state.is_inside_table);
}
}