@north-light/crouter 0.3.196 → 0.3.198

package/dist/core/human/markdown-html.js

@@ -1,581 +0,0 @@
-// Markdown → HTML for md-dialect page prose. Hand-rolled and dependency-free:
-// crtr ships no markdown-to-HTML dependency, and the only markdown machinery it
-// has (`core/termrender`) renders ANSI terminal lines through a managed Python
-// venv — the wrong thing to invoke inside a daemon HTTP handler.
-//
-// The supported subset is a deliberate CommonMark slice, pinned here because
-// page authors and page hosts both read it as the contract:
-//
-//   Blocks — ATX headings (`#`…`######`), setext headings (`===`/`---`
-//     underlines), fenced code (``` and ~~~, info string → `class="language-x"`),
-//     blockquotes, thematic breaks, bullet lists (`-`, `*`, `+`), ordered lists
-//     (`1.`, `1)`), GFM pipe tables (with `:---:` alignment), raw HTML blocks,
-//     and paragraphs. Lists nest by indentation and render tight or loose.
-//   Inline — backslash escapes, code spans, `**strong**`, `*em*`, `***both***`
-//     (and the `_` forms, intraword-safe), links, images, autolinks
-//     (`<https://…>`, `<a@b.c>`), raw inline HTML, character entities, and
-//     two-space hard breaks.
-//
-// NOT supported (authors get the literal text): reference links and
-// definitions, footnotes, indented (four-space) code blocks, task-list
-// checkboxes, strikethrough, and HTML comments as block separators.
-//
-// Raw HTML passes through verbatim. A page document is authored by the agent
-// that owns the ticket, and the html dialect is whole-document passthrough by
-// design — md prose is not a less-trusted channel than the document carrying it.
-const ENTITY = /^&(?:[a-zA-Z][a-zA-Z0-9]{1,31}|#\d{1,7}|#[xX][0-9a-fA-F]{1,6});/;
-const ASCII_PUNCTUATION = /[!"#$%&'()*+,\-./:;<=>?@[\\\]^_`{|}~]/;
-const AUTOLINK_URI = /^<([a-zA-Z][a-zA-Z0-9+.-]{1,31}:[^<>\u0000-\u0020]*)>/;
-const AUTOLINK_EMAIL = /^<([^\s@<>]+@[^\s@<>]+\.[^\s@<>]+)>/;
-const RAW_INLINE_TAG = /^<\/?[A-Za-z][A-Za-z0-9-]*(?:\s[^<>]*?)?\/?>/;
-const HTML_COMMENT = /^<!--[\s\S]*?-->/;
-const LIST_ITEM = /^( {0,3})([-*+]|\d{1,9}[.)])( +|$)(.*)$/;
-const THEMATIC_BREAK = /^ {0,3}([-*_])(?: *\1){2,} *$/;
-/** Escape text for an HTML text node. */
-export function escapeHtmlText(raw) {
-    return raw.replace(/[&<>]/g, (character) => (character === '&' ? '&amp;' : character === '<' ? '&lt;' : '&gt;'));
-}
-/** Escape a value for a double-quoted HTML attribute. */
-export function escapeHtmlAttribute(raw) {
-    return escapeHtmlText(raw).replace(/"/g, '&quot;');
-}
-/** Render one markdown document fragment as HTML. */
-export function renderMarkdownToHtml(markdown) {
-    return renderBlocks(markdown.replace(/\r\n?/g, '\n').split('\n'), false);
-}
-// ---------------------------------------------------------------------------
-// Blocks
-// ---------------------------------------------------------------------------
-/** `unwrapParagraphs` renders a tight list item: its paragraphs emit inline
- *  content with no `<p>` wrapper, exactly as CommonMark specifies. */
-function renderBlocks(lines, unwrapParagraphs) {
-    const blocks = [];
-    let index = 0;
-    while (index < lines.length) {
-        const line = lines[index];
-        if (line.trim() === '') {
-            index++;
-            continue;
-        }
-        const fence = /^ {0,3}(`{3,}|~{3,})(.*)$/.exec(line);
-        if (fence !== null) {
-            index = pushFencedCode(blocks, lines, index, fence[1], fence[2]);
-            continue;
-        }
-        const heading = /^ {0,3}(#{1,6})(?:[ \t]+(.*?))?[ \t]*$/.exec(line);
-        if (heading !== null) {
-            const level = heading[1].length;
-            blocks.push(`<h${level}>${renderInline((heading[2] ?? '').replace(/[ \t]+#+[ \t]*$/, ''))}</h${level}>`);
-            index++;
-            continue;
-        }
-        if (THEMATIC_BREAK.test(line)) {
-            blocks.push('<hr />');
-            index++;
-            continue;
-        }
-        if (/^ {0,3}>/.test(line)) {
-            index = pushBlockquote(blocks, lines, index);
-            continue;
-        }
-        if (LIST_ITEM.test(line) && !THEMATIC_BREAK.test(line)) {
-            index = pushList(blocks, lines, index);
-            continue;
-        }
-        if (isHtmlBlockStart(line)) {
-            index = pushHtmlBlock(blocks, lines, index);
-            continue;
-        }
-        const tableEnd = pushTable(blocks, lines, index);
-        if (tableEnd !== null) {
-            index = tableEnd;
-            continue;
-        }
-        index = pushParagraph(blocks, lines, index, unwrapParagraphs);
-    }
-    return blocks.join('\n');
-}
-function closesFence(text, opener) {
-    const trimmed = text.trim();
-    return trimmed.length >= opener.length && trimmed.split('').every((character) => character === opener[0]);
-}
-function pushFencedCode(blocks, lines, start, opener, info) {
-    let end = start + 1;
-    while (end < lines.length && !closesFence(lines[end], opener))
-        end++;
-    const body = lines.slice(start + 1, end);
-    const language = info.trim().split(/\s+/)[0] ?? '';
-    const attribute = language === '' ? '' : ` class="language-${escapeHtmlAttribute(language)}"`;
-    blocks.push(`<pre><code${attribute}>${escapeHtmlText(body.join('\n'))}${body.length === 0 ? '' : '\n'}</code></pre>`);
-    return end < lines.length ? end + 1 : end;
-}
-function pushBlockquote(blocks, lines, start) {
-    const quoted = [];
-    let index = start;
-    while (index < lines.length && /^ {0,3}>/.test(lines[index])) {
-        quoted.push(lines[index].replace(/^ {0,3}> ?/, ''));
-        index++;
-        // A lazy continuation line belongs to the quoted paragraph.
-        while (index < lines.length && lines[index].trim() !== '' && !/^ {0,3}>/.test(lines[index]) && !startsNewBlock(lines[index])) {
-            quoted.push(lines[index]);
-            index++;
-        }
-    }
-    blocks.push(`<blockquote>\n${renderBlocks(quoted, false)}\n</blockquote>`);
-    return index;
-}
-function startsNewBlock(line) {
-    return /^ {0,3}(`{3,}|~{3,})/.test(line)
-        || /^ {0,3}#{1,6}(\s|$)/.test(line)
-        || THEMATIC_BREAK.test(line)
-        || /^ {0,3}>/.test(line)
-        || LIST_ITEM.test(line)
-        || isHtmlBlockStart(line);
-}
-function isHtmlBlockStart(line) {
-    return /^ {0,3}<(?:[A-Za-z][A-Za-z0-9-]*|\/[A-Za-z][A-Za-z0-9-]*|!--)/.test(line);
-}
-function pushHtmlBlock(blocks, lines, start) {
-    let end = start;
-    while (end < lines.length && lines[end].trim() !== '')
-        end++;
-    blocks.push(lines.slice(start, end).join('\n'));
-    return end;
-}
-// ---------------------------------------------------------------------------
-// Lists
-// ---------------------------------------------------------------------------
-function markerType(marker) {
-    return /^\d/.test(marker) ? 'ordered' : 'bullet';
-}
-function pushList(blocks, lines, start) {
-    const first = LIST_ITEM.exec(lines[start]);
-    const type = markerType(first[2]);
-    const items = [];
-    let current = null;
-    let contentIndent = 0;
-    let loose = false;
-    let pendingBlank = false;
-    let index = start;
-    while (index < lines.length) {
-        const line = lines[index];
-        if (line.trim() === '') {
-            pendingBlank = true;
-            index++;
-            continue;
-        }
-        const item = LIST_ITEM.exec(line);
-        const itemIndent = item === null ? 0 : item[1].length;
-        if (item !== null && !THEMATIC_BREAK.test(line) && markerType(item[2]) === type && (current === null || itemIndent < contentIndent)) {
-            if (current !== null && pendingBlank)
-                loose = true;
-            contentIndent = itemIndent + item[2].length + (item[3] === '' ? 1 : item[3].length);
-            current = [item[4]];
-            items.push(current);
-            pendingBlank = false;
-            index++;
-            continue;
-        }
-        if (current === null)
-            break;
-        const indented = /^ +/.exec(line)?.[0]?.length ?? 0;
-        if (indented >= contentIndent) {
-            if (pendingBlank) {
-                current.push('');
-                loose = true;
-            }
-            current.push(line.slice(contentIndent));
-            pendingBlank = false;
-            index++;
-            continue;
-        }
-        // A lazy continuation of the item's paragraph — but only before a blank line.
-        if (!pendingBlank && !startsNewBlock(line)) {
-            current.push(line);
-            index++;
-            continue;
-        }
-        break;
-    }
-    const rendered = items.map((item) => `<li>${wrapListItem(renderBlocks(item, !loose))}</li>`);
-    const tag = type === 'ordered' ? 'ol' : 'ul';
-    const startNumber = type === 'ordered' ? Number.parseInt(first[2], 10) : 1;
-    const startAttribute = type === 'ordered' && startNumber !== 1 ? ` start="${startNumber}"` : '';
-    blocks.push(`<${tag}${startAttribute}>\n${rendered.join('\n')}\n</${tag}>`);
-    return index;
-}
-function wrapListItem(html) {
-    return html.includes('\n') ? `\n${html}\n` : html;
-}
-// ---------------------------------------------------------------------------
-// GFM tables
-// ---------------------------------------------------------------------------
-function splitRow(line) {
-    const trimmed = line.trim().replace(/^\|/, '').replace(/\|$/, '');
-    const cells = [];
-    let cell = '';
-    for (let index = 0; index < trimmed.length; index++) {
-        const character = trimmed[index];
-        if (character === '\\' && trimmed[index + 1] === '|') {
-            cell += '|';
-            index++;
-            continue;
-        }
-        if (character === '|') {
-            cells.push(cell.trim());
-            cell = '';
-            continue;
-        }
-        cell += character;
-    }
-    cells.push(cell.trim());
-    return cells;
-}
-/** Render a GFM pipe table at `start`, or return null when there is none. */
-function pushTable(blocks, lines, start) {
-    const header = lines[start];
-    const delimiter = lines[start + 1];
-    if (!header.includes('|') || delimiter === undefined)
-        return null;
-    if (!/^ {0,3}\|?(?: *:?-+:? *\|)+ *:?-*:? *\|? *$/.test(delimiter) && !/^ {0,3}:?-+:? *$/.test(delimiter))
-        return null;
-    const alignments = splitRow(delimiter).map((cell) => {
-        if (!/^:?-+:?$/.test(cell))
-            return null;
-        if (cell.startsWith(':') && cell.endsWith(':'))
-            return 'center';
-        if (cell.endsWith(':'))
-            return 'right';
-        if (cell.startsWith(':'))
-            return 'left';
-        return '';
-    });
-    if (alignments.some((alignment) => alignment === null))
-        return null;
-    const headerCells = splitRow(header);
-    if (headerCells.length !== alignments.length)
-        return null;
-    const align = (column) => {
-        const alignment = alignments[column];
-        return alignment === undefined || alignment === '' ? '' : ` style="text-align:${alignment}"`;
-    };
-    const rows = [];
-    let index = start + 2;
-    while (index < lines.length && lines[index].trim() !== '' && lines[index].includes('|') && !startsNewBlock(lines[index])) {
-        const cells = splitRow(lines[index]);
-        const padded = alignments.map((_, column) => cells[column] ?? '');
-        rows.push(`<tr>${padded.map((cell, column) => `<td${align(column)}>${renderInline(cell)}</td>`).join('')}</tr>`);
-        index++;
-    }
-    const head = `<thead>\n<tr>${headerCells.map((cell, column) => `<th${align(column)}>${renderInline(cell)}</th>`).join('')}</tr>\n</thead>`;
-    const body = rows.length === 0 ? '' : `\n<tbody>\n${rows.join('\n')}\n</tbody>`;
-    blocks.push(`<table>\n${head}${body}\n</table>`);
-    return index;
-}
-// ---------------------------------------------------------------------------
-// Paragraphs
-// ---------------------------------------------------------------------------
-function pushParagraph(blocks, lines, start, unwrap) {
-    const collected = [lines[start].trim()];
-    let index = start + 1;
-    while (index < lines.length) {
-        const line = lines[index];
-        if (line.trim() === '')
-            break;
-        if (/^ {0,3}=+ *$/.test(line)) {
-            blocks.push(`<h1>${renderInline(collected.join('\n'))}</h1>`);
-            return index + 1;
-        }
-        if (/^ {0,3}-+ *$/.test(line)) {
-            blocks.push(`<h2>${renderInline(collected.join('\n'))}</h2>`);
-            return index + 1;
-        }
-        if (startsNewBlock(line))
-            break;
-        collected.push(line.replace(/^ {0,3}/, ''));
-        index++;
-    }
-    const inline = renderInline(collected.join('\n'));
-    blocks.push(unwrap ? inline : `<p>${inline}</p>`);
-    return index;
-}
-// ---------------------------------------------------------------------------
-// Inline
-// ---------------------------------------------------------------------------
-/** Render inline markdown. Exported for callers rendering a single line. */
-export function renderInline(source) {
-    let out = '';
-    let index = 0;
-    while (index < source.length) {
-        const character = source[index];
-        if (character === '\\') {
-            const next = source[index + 1];
-            if (next === '\n') {
-                out += '<br />\n';
-                index += 2;
-                continue;
-            }
-            if (next !== undefined && ASCII_PUNCTUATION.test(next)) {
-                out += escapeHtmlText(next);
-                index += 2;
-                continue;
-            }
-            out += '\\';
-            index++;
-            continue;
-        }
-        if (character === ' ') {
-            const run = /^( +)(\n|$)/.exec(source.slice(index));
-            if (run !== null) {
-                out += run[1].length >= 2 && run[2] === '\n' ? '<br />\n' : run[2] === '\n' ? '\n' : '';
-                index += run[1].length + run[2].length;
-                continue;
-            }
-            out += ' ';
-            index++;
-            continue;
-        }
-        if (character === '`') {
-            const span = codeSpanAt(source, index);
-            if (span !== null) {
-                out += span.html;
-                index = span.end;
-                continue;
-            }
-            out += '`';
-            index++;
-            continue;
-        }
-        if (character === '<') {
-            const raw = rawInlineAt(source, index);
-            if (raw !== null) {
-                out += raw.html;
-                index = raw.end;
-                continue;
-            }
-            out += '&lt;';
-            index++;
-            continue;
-        }
-        if (character === '&') {
-            const entity = ENTITY.exec(source.slice(index));
-            if (entity !== null) {
-                out += entity[0];
-                index += entity[0].length;
-                continue;
-            }
-            out += '&amp;';
-            index++;
-            continue;
-        }
-        if (character === '>') {
-            out += '&gt;';
-            index++;
-            continue;
-        }
-        if (character === '!' && source[index + 1] === '[') {
-            const image = linkAt(source, index + 1);
-            if (image !== null) {
-                out += `<img src="${escapeHtmlAttribute(image.destination)}" alt="${escapeHtmlAttribute(stripInlineMarkup(image.text))}"${image.title === undefined ? '' : ` title="${escapeHtmlAttribute(image.title)}"`} />`;
-                index = image.end;
-                continue;
-            }
-        }
-        if (character === '[') {
-            const link = linkAt(source, index);
-            if (link !== null) {
-                out += `<a href="${escapeHtmlAttribute(link.destination)}"${link.title === undefined ? '' : ` title="${escapeHtmlAttribute(link.title)}"`}>${renderInline(link.text)}</a>`;
-                index = link.end;
-                continue;
-            }
-        }
-        if (character === '*' || character === '_') {
-            const emphasis = emphasisAt(source, index);
-            if (emphasis !== null) {
-                out += emphasis.html;
-                index = emphasis.end;
-                continue;
-            }
-            out += character;
-            index++;
-            continue;
-        }
-        out += character;
-        index++;
-    }
-    return out;
-}
-/** Plain text for an image `alt`, which carries no markup. */
-function stripInlineMarkup(source) {
-    return source.replace(/[*_`]/g, '');
-}
-function backtickRun(source, at) {
-    let run = 0;
-    while (source[at + run] === '`')
-        run++;
-    return run;
-}
-function codeSpanAt(source, start) {
-    const run = backtickRun(source, start);
-    let index = start + run;
-    while (index < source.length) {
-        if (source[index] !== '`') {
-            index++;
-            continue;
-        }
-        const closing = backtickRun(source, index);
-        if (closing === run) {
-            let content = source.slice(start + run, index).replace(/\n/g, ' ');
-            if (content.length > 2 && content.startsWith(' ') && content.endsWith(' ') && content.trim() !== '')
-                content = content.slice(1, -1);
-            return { html: `<code>${escapeHtmlText(content)}</code>`, end: index + closing };
-        }
-        index += closing;
-    }
-    return null;
-}
-function rawInlineAt(source, start) {
-    const rest = source.slice(start);
-    const uri = AUTOLINK_URI.exec(rest);
-    if (uri !== null)
-        return { html: `<a href="${escapeHtmlAttribute(uri[1])}">${escapeHtmlText(uri[1])}</a>`, end: start + uri[0].length };
-    const email = AUTOLINK_EMAIL.exec(rest);
-    if (email !== null)
-        return { html: `<a href="mailto:${escapeHtmlAttribute(email[1])}">${escapeHtmlText(email[1])}</a>`, end: start + email[0].length };
-    const comment = HTML_COMMENT.exec(rest);
-    if (comment !== null)
-        return { html: comment[0], end: start + comment[0].length };
-    const tag = RAW_INLINE_TAG.exec(rest);
-    if (tag !== null)
-        return { html: tag[0], end: start + tag[0].length };
-    return null;
-}
-function linkAt(source, start) {
-    let depth = 0;
-    let index = start;
-    let textEnd = -1;
-    while (index < source.length) {
-        const character = source[index];
-        if (character === '\\') {
-            index += 2;
-            continue;
-        }
-        if (character === '`') {
-            const span = codeSpanAt(source, index);
-            index = span === null ? index + 1 : span.end;
-            continue;
-        }
-        if (character === '[')
-            depth++;
-        else if (character === ']') {
-            depth--;
-            if (depth === 0) {
-                textEnd = index;
-                break;
-            }
-        }
-        index++;
-    }
-    if (textEnd === -1 || source[textEnd + 1] !== '(')
-        return null;
-    let cursor = textEnd + 2;
-    while (source[cursor] === ' ' || source[cursor] === '\n')
-        cursor++;
-    let destination = '';
-    if (source[cursor] === '<') {
-        const close = source.indexOf('>', cursor + 1);
-        if (close === -1)
-            return null;
-        destination = source.slice(cursor + 1, close);
-        cursor = close + 1;
-    }
-    else {
-        let parens = 0;
-        while (cursor < source.length) {
-            const character = source[cursor];
-            if (character === '\\' && source[cursor + 1] !== undefined) {
-                destination += source[cursor + 1];
-                cursor += 2;
-                continue;
-            }
-            if (/\s/.test(character))
-                break;
-            if (character === '(')
-                parens++;
-            if (character === ')') {
-                if (parens === 0)
-                    break;
-                parens--;
-            }
-            destination += character;
-            cursor++;
-        }
-    }
-    while (source[cursor] === ' ' || source[cursor] === '\n')
-        cursor++;
-    let title;
-    const quote = source[cursor];
-    if (quote === '"' || quote === "'") {
-        const close = source.indexOf(quote, cursor + 1);
-        if (close === -1)
-            return null;
-        title = source.slice(cursor + 1, close);
-        cursor = close + 1;
-        while (source[cursor] === ' ' || source[cursor] === '\n')
-            cursor++;
-    }
-    if (source[cursor] !== ')')
-        return null;
-    return { text: source.slice(start + 1, textEnd), destination, ...(title === undefined ? {} : { title }), end: cursor + 1 };
-}
-function emphasisAt(source, start) {
-    const marker = source[start];
-    let run = 0;
-    while (source[start + run] === marker)
-        run++;
-    const need = Math.min(run, 3);
-    const opened = source[start + need];
-    if (opened === undefined || /\s/.test(opened))
-        return null;
-    if (marker === '_') {
-        const before = source[start - 1];
-        if (before !== undefined && /[A-Za-z0-9]/.test(before))
-            return null;
-    }
-    const closer = findEmphasisCloser(source, start + need, marker, need);
-    if (closer === null)
-        return null;
-    const inner = renderInline(source.slice(start + need, closer));
-    const html = need === 1 ? `<em>${inner}</em>` : need === 2 ? `<strong>${inner}</strong>` : `<strong><em>${inner}</em></strong>`;
-    return { html, end: closer + need };
-}
-function findEmphasisCloser(source, from, marker, need) {
-    let index = from;
-    while (index < source.length) {
-        const character = source[index];
-        if (character === '\\') {
-            index += 2;
-            continue;
-        }
-        if (character === '`') {
-            const span = codeSpanAt(source, index);
-            index = span === null ? index + 1 : span.end;
-            continue;
-        }
-        if (character !== marker) {
-            index++;
-            continue;
-        }
-        let run = 0;
-        while (source[index + run] === marker)
-            run++;
-        const before = source[index - 1];
-        const after = source[index + run];
-        const closes = run >= need && before !== undefined && !/\s/.test(before)
-            && !(marker === '_' && after !== undefined && /[A-Za-z0-9]/.test(after));
-        if (closes && index > from)
-            return index;
-        index += run;
-    }
-    return null;
-}