Files
nuwiki/crates/nuwiki-lsp/src/export.rs
T

420 lines
14 KiB
Rust
Raw Normal View History

//! HTML export — pure helpers shared by the Phase 17 `nuwiki.export.*`
//! command handlers. Side-effecting work (reading the template from
//! disk, writing the rendered HTML, copying CSS) lives in `commands.rs`
//! since the dispatcher is the one with a `&Backend` and async context.
//!
//! Design split:
//! - [`output_path_for`] / [`output_url_for`] — name → on-disk path.
//! - [`template_for`] — locate the right template file on disk.
//! - [`fallback_template`] — minimal stand-in when nothing is found.
//! - [`format_date`] — strftime subset (`%Y`/`%m`/`%d`).
//! - [`build_toc_html`] — TOC rendered into the `{{toc}}` variable.
//! - [`compute_root_path`] — relative path from a page's output to
//! `html_path` for stylesheet hrefs in nested pages.
//! - [`is_excluded`] — glob match against `exclude_files`.
//! - [`render_page_html`] — drives [`HtmlRenderer`] with all of the
//! above wired up; returns the final HTML string.
use std::collections::HashMap;
use std::path::{Path, PathBuf};
use nuwiki_core::ast::{BlockNode, DocumentNode, HeadingNode, InlineNode, LinkKind};
use nuwiki_core::date::DiaryDate;
use nuwiki_core::render::{HtmlRenderer, Renderer};
use crate::config::{HtmlConfig, WikiConfig};
/// Map a page name (e.g. `"diary/2026-05-11"`) to its on-disk HTML
/// output path under `html_path`. Honours
/// [`HtmlConfig::html_filename_parameterization`] for URL-safe slugs
/// — when enabled, only the final path segment is slugified (so the
/// `diary/` subdir survives).
pub fn output_path_for(cfg: &HtmlConfig, name: &str) -> PathBuf {
let segments: Vec<String> = name
.split('/')
.map(|s| {
if cfg.html_filename_parameterization {
slugify(s)
} else {
s.to_string()
}
})
.collect();
let mut p = cfg.html_path.clone();
for (i, seg) in segments.iter().enumerate() {
if i + 1 == segments.len() {
p.push(format!("{seg}.html"));
} else {
p.push(seg);
}
}
p
}
/// Locate the template body for a document. Resolution order:
/// 1. `<template_path>/<doc.metadata.template>.<template_ext>` if set.
/// 2. `<template_path>/<template_default>.<template_ext>`.
/// 3. [`fallback_template`].
///
/// The on-disk read is performed by the caller (`commands.rs`) via the
/// [`TemplateSource`] returned here. Keeps this fn pure for tests.
pub fn template_for(cfg: &HtmlConfig, doc: &DocumentNode) -> TemplateSource {
let primary = doc.metadata.template.as_deref().map(|name| {
cfg.template_path
.join(format!("{name}{}", cfg.template_ext))
});
let fallback = cfg
.template_path
.join(format!("{}{}", cfg.template_default, cfg.template_ext));
TemplateSource { primary, fallback }
}
/// Two-tier lookup result. Caller tries `primary` first, then `fallback`,
/// then falls back to [`fallback_template`] when neither file exists.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct TemplateSource {
pub primary: Option<PathBuf>,
pub fallback: PathBuf,
}
/// Minimal valid HTML page used when no template file is found on disk.
/// Includes the same `{{title}}` / `{{content}}` placeholders as a real
/// template so the renderer's substitution still works.
pub fn fallback_template(css_name: &str) -> String {
format!(
r#"<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<title>{{{{title}}}}</title>
<link rel="stylesheet" href="{{{{root_path}}}}{css_name}">
</head>
<body>
{{{{content}}}}
</body>
</html>
"#
)
}
/// Tiny strftime: supports `%Y` (4-digit year), `%m` (2-digit month),
/// `%d` (2-digit day), `%%` (literal percent). Anything else passes
/// through. The diary use case never produces time-of-day output, so
/// `%H`/`%M`/`%S` are intentionally omitted.
pub fn format_date(date: &DiaryDate, fmt: &str) -> String {
let mut out = String::with_capacity(fmt.len() + 4);
let bytes = fmt.as_bytes();
let mut i = 0;
while i < bytes.len() {
if bytes[i] == b'%' && i + 1 < bytes.len() {
match bytes[i + 1] {
b'Y' => out.push_str(&format!("{:04}", date.year)),
b'm' => out.push_str(&format!("{:02}", date.month)),
b'd' => out.push_str(&format!("{:02}", date.day)),
b'%' => out.push('%'),
other => {
out.push('%');
out.push(other as char);
}
}
i += 2;
} else {
out.push(bytes[i] as char);
i += 1;
}
}
out
}
/// Render the page's headings as a nested `<ul>` for the `{{toc}}`
/// template variable. Returns an empty string when the document has
/// no headings.
pub fn build_toc_html(doc: &DocumentNode) -> String {
let headings = collect_headings(doc);
if headings.is_empty() {
return String::new();
}
let min_level = headings.iter().map(|h| h.level).min().unwrap_or(1);
let mut out = String::from("<ul class=\"toc\">");
let mut depth = 0usize;
for h in &headings {
let target = h.level.saturating_sub(min_level) as usize;
while depth < target {
out.push_str("<ul>");
depth += 1;
}
while depth > target {
out.push_str("</ul>");
depth -= 1;
}
let anchor = crate::index::slugify(&h.title);
out.push_str(&format!(
"<li><a href=\"#{anchor}\">{title}</a></li>",
title = escape_html(&h.title)
));
}
while depth > 0 {
out.push_str("</ul>");
depth -= 1;
}
out.push_str("</ul>");
out
}
/// Compute the relative path back to `html_path` from the directory
/// holding the output file for `name`. Used as the `{{root_path}}`
/// prefix so a nested page's `<link>`/`<script>` hrefs still find
/// the shared CSS at the export root.
pub fn compute_root_path(name: &str) -> String {
let depth = name.matches('/').count();
if depth == 0 {
String::new()
} else {
let mut s = String::with_capacity(depth * 3);
for _ in 0..depth {
s.push_str("../");
}
s
}
}
/// True if `name` matches any of the `exclude_files` glob patterns.
/// Supported wildcards: `*` (any segment chars except `/`), `**` (any
/// segments including `/`), `?` (one char). Anything else is literal.
pub fn is_excluded(patterns: &[String], name: &str) -> bool {
patterns.iter().any(|p| glob_match(p, name))
}
/// Pure HTML render pass. The caller has already loaded the template
/// (or falls back). Wires up `{{date}}`, `{{root_path}}`, `{{toc}}`
/// plus arbitrary extras passed via `extra_vars`.
///
/// `wiki_extension` is the wiki's source file extension (e.g. `".wiki"`).
/// When set, it is stripped from wiki/interwiki link targets so a link
/// written with the extension (`[[todo.wiki]]`) exports to `todo.html`,
/// matching how the LSP resolves the same link for in-editor navigation.
pub fn render_page_html(
doc: &DocumentNode,
template: Option<String>,
name: &str,
today: DiaryDate,
cfg: &HtmlConfig,
wiki_extension: Option<&str>,
extra_vars: HashMap<String, String>,
) -> std::io::Result<String> {
let date_str = match doc.metadata.date.as_deref() {
Some(s) => match DiaryDate::parse(s) {
Some(d) => format_date(&d, &cfg.template_date_format),
None => s.to_string(),
},
None => format_date(&today, &cfg.template_date_format),
};
let root_path = compute_root_path(name);
let toc = build_toc_html(doc);
let mut vars: HashMap<String, String> = HashMap::new();
vars.insert("date".into(), date_str);
vars.insert("root_path".into(), root_path);
vars.insert("toc".into(), toc);
vars.insert("css".into(), cfg.css_name.clone());
// vimwiki names the stylesheet placeholder `%wiki_css%`; alias it so stock
// templates resolve. nuwiki's own templates use `{{css}}`.
vars.insert("wiki_css".into(), cfg.css_name.clone());
for (k, v) in extra_vars {
vars.insert(k, v);
}
let mut r = HtmlRenderer::new();
if let Some(ext) = wiki_extension.filter(|e| !e.trim_start_matches('.').is_empty()) {
// Strip the wiki extension from page links before the default resolver
// turns them into `.html` URLs, so `[[todo.wiki]]` → `todo.html` rather
// than `todo.wiki.html`. Only wiki/interwiki targets are touched;
// file:/local: paths keep their literal extension.
let ext = ext.to_string();
r = r.with_link_resolver(move |target| {
let mut t = target.clone();
if matches!(t.kind, LinkKind::Wiki | LinkKind::Interwiki) {
if let Some(p) = t.path.take() {
t.path = Some(crate::index::strip_wiki_extension(&p, Some(&ext)).to_string());
}
}
nuwiki_core::render::html::default_link_resolver(&t)
});
}
if let Some(t) = template {
r = r.with_template(t);
}
r = r.with_vars(vars);
if !cfg.color_dic.is_empty() {
r = r.with_colors(cfg.color_dic.clone());
}
r.render_to_string(doc)
}
// ===== Internals =====
fn slugify(s: &str) -> String {
let mut out = String::with_capacity(s.len());
let mut prev_dash = false;
for ch in s.chars() {
if ch.is_alphanumeric() {
for c in ch.to_lowercase() {
out.push(c);
}
prev_dash = false;
} else if !prev_dash && !out.is_empty() {
out.push('-');
prev_dash = true;
}
}
while out.ends_with('-') {
out.pop();
}
if out.is_empty() {
"untitled".into()
} else {
out
}
}
fn collect_headings(doc: &DocumentNode) -> Vec<HeadingProj> {
let mut out = Vec::new();
for b in &doc.children {
if let BlockNode::Heading(h) = b {
out.push(HeadingProj {
level: h.level,
title: inline_to_text(h),
});
}
}
out
}
struct HeadingProj {
level: u8,
title: String,
}
fn inline_to_text(h: &HeadingNode) -> String {
let mut s = String::new();
for n in &h.children {
push_inline(n, &mut s);
}
s.trim().to_string()
}
fn push_inline(n: &InlineNode, out: &mut String) {
match n {
InlineNode::Text(t) => out.push_str(&t.content),
InlineNode::Bold(b) => b.children.iter().for_each(|c| push_inline(c, out)),
InlineNode::Italic(i) => i.children.iter().for_each(|c| push_inline(c, out)),
InlineNode::BoldItalic(bi) => bi.children.iter().for_each(|c| push_inline(c, out)),
InlineNode::Strikethrough(s) => s.children.iter().for_each(|c| push_inline(c, out)),
InlineNode::Code(c) => out.push_str(&c.content),
InlineNode::Superscript(s) => s.children.iter().for_each(|c| push_inline(c, out)),
InlineNode::Subscript(s) => s.children.iter().for_each(|c| push_inline(c, out)),
InlineNode::Color(c) => c.children.iter().for_each(|c| push_inline(c, out)),
InlineNode::WikiLink(w) => match &w.description {
Some(d) => d.iter().for_each(|c| push_inline(c, out)),
None => {
if let Some(p) = &w.target.path {
out.push_str(p);
}
}
},
InlineNode::ExternalLink(e) => match &e.description {
Some(d) => d.iter().for_each(|c| push_inline(c, out)),
None => out.push_str(&e.url),
},
_ => {}
}
}
fn escape_html(s: &str) -> String {
let mut out = String::with_capacity(s.len());
for ch in s.chars() {
match ch {
'&' => out.push_str("&amp;"),
'<' => out.push_str("&lt;"),
'>' => out.push_str("&gt;"),
'"' => out.push_str("&quot;"),
_ => out.push(ch),
}
}
out
}
/// Minimal glob match for `exclude_files`. Handles `*`, `**`, `?` and
/// literal characters. No character classes — those aren't needed for
/// "exclude foo/*.tmp" style patterns.
fn glob_match(pattern: &str, text: &str) -> bool {
glob_impl(pattern.as_bytes(), text.as_bytes())
}
fn glob_impl(p: &[u8], t: &[u8]) -> bool {
let mut pi = 0;
let mut ti = 0;
let mut star: Option<(usize, usize)> = None;
let mut globstar = false;
while ti < t.len() {
if pi < p.len() && pi + 1 < p.len() && p[pi] == b'*' && p[pi + 1] == b'*' {
globstar = true;
star = Some((pi, ti));
pi += 2;
// Allow optional `/` after globstar.
if pi < p.len() && p[pi] == b'/' {
pi += 1;
}
continue;
} else if pi < p.len() && p[pi] == b'*' {
star = Some((pi, ti));
pi += 1;
globstar = false;
continue;
} else if pi < p.len() && (p[pi] == b'?' || p[pi] == t[ti]) {
pi += 1;
ti += 1;
continue;
} else if let Some((s_pi, s_ti)) = star {
if !globstar && t[s_ti] == b'/' && t[ti] != b'/' {
// single-star can't cross a `/`
return false;
}
pi = if globstar { s_pi + 2 } else { s_pi + 1 };
ti = s_ti + 1;
star = Some((s_pi, ti));
continue;
} else {
return false;
}
}
while pi < p.len() && (p[pi] == b'*' || (p[pi] == b'/' && globstar)) {
pi += 1;
}
pi == p.len()
}
/// `<html_path>/<css_name>` — the absolute path of the CSS file the
/// renderer references via `{{root_path}}{{css}}`.
pub fn css_path(cfg: &HtmlConfig) -> PathBuf {
cfg.html_path.join(&cfg.css_name)
}
/// Minimal default CSS. Used by `nuwiki.export.*` commands when no CSS
/// file exists yet at [`css_path`]. Deliberately tiny — users are
/// expected to replace it with their own.
pub const DEFAULT_CSS: &str =
"body{font-family:sans-serif;max-width:46em;margin:2em auto;padding:0 1em;line-height:1.55}\
h1,h2,h3,h4,h5,h6{font-family:serif;line-height:1.2}\
pre{background:#f4f4f4;padding:.5em;overflow:auto}\
code{background:#f4f4f4;padding:.1em .25em;border-radius:3px}\
table{border-collapse:collapse}\
table td,table th{border:1px solid #ccc;padding:.25em .5em}\
ul.toc{padding-left:1.5em}\n";
/// Convenience wrapper exposed for the dispatcher: convert a wiki config
/// to the absolute output dir.
pub fn html_output_dir(cfg: &WikiConfig) -> &Path {
&cfg.html.html_path
}