@ai-matrx/print 0.8.6 → 0.8.8

package/dist/markdown.js

@@ -23,6 +23,103 @@ function openPrintWindow(htmlDocument, filename = "print", windowFeatures = "wid
 // src/markdown.ts
 import { escapeHtml as escapeHtml2 } from "@ai-matrx/kit/html-escape";
 
+// src/safe-html.ts
+import { fromHtml } from "hast-util-from-html";
+import { defaultSchema, sanitize } from "hast-util-sanitize";
+import { toHtml } from "hast-util-to-html";
+var STRIP_WITH_CONTENT = [
+  "script",
+  "style",
+  "noscript",
+  "template",
+  "iframe",
+  "object",
+  "embed",
+  "textarea",
+  "select",
+  "head",
+  "title"
+];
+var FORBIDDEN_TAGS = [
+  ...STRIP_WITH_CONTENT,
+  "form",
+  "meta",
+  "base",
+  "link",
+  "frame",
+  "frameset",
+  "applet",
+  "button",
+  "svg",
+  "math"
+];
+var baseAttributes = defaultSchema.attributes ?? {};
+var SAFE_HTML_SCHEMA = {
+  ...defaultSchema,
+  tagNames: (defaultSchema.tagNames ?? []).filter((t) => !FORBIDDEN_TAGS.includes(t)),
+  strip: STRIP_WITH_CONTENT,
+  // Authored anchors keep their names; the GitHub `user-content-` rewrite
+  // would also break `<label for>` → `<input id>` pairs.
+  clobberPrefix: "",
+  clobber: []
+};
+var PRINT_CLASS = /^(?:(?:matrx|print|code|doc|language)-[\w-]*|inline-code|is-checked)$/;
+function withPrintClasses(defs = []) {
+  let found = false;
+  const out = defs.map((d) => {
+    if (Array.isArray(d) && d[0] === "className") {
+      found = true;
+      return [...d, PRINT_CLASS];
+    }
+    return d;
+  });
+  return found ? out : [...out, ["className", PRINT_CLASS]];
+}
+var printAttributes = Object.fromEntries(
+  Object.entries(baseAttributes).map(([tag, defs]) => [tag, withPrintClasses(defs)])
+);
+var cellAlign = ["style", /^text-align:\s*(?:left|center|right);?$/];
+var PRINT_HTML_SCHEMA = {
+  ...SAFE_HTML_SCHEMA,
+  tagNames: [
+    .../* @__PURE__ */ new Set([...SAFE_HTML_SCHEMA.tagNames ?? [], "section", "figure", "figcaption", "label"])
+  ],
+  attributes: {
+    ...printAttributes,
+    "*": [
+      ...baseAttributes["*"] ?? [],
+      ["className", PRINT_CLASS],
+      "role",
+      "ariaLabel",
+      "ariaHidden",
+      "dataLabel",
+      "dataLanguage",
+      "dataArtifactType"
+    ],
+    // Table-cell alignment is the one inline style the converter writes.
+    td: withPrintClasses([...baseAttributes["td"] ?? [], [...cellAlign]]),
+    th: withPrintClasses([...baseAttributes["th"] ?? [], [...cellAlign]]),
+    // The flashcard flip is a checkbox + label pair (GitHub forces `disabled`).
+    input: withPrintClasses([["type", "checkbox"], "checked"]),
+    label: withPrintClasses(["htmlFor"])
+  },
+  required: { input: { type: "checkbox" } },
+  protocols: {
+    ...defaultSchema.protocols,
+    href: [...defaultSchema.protocols?.href ?? [], "tel"],
+    // Embedded images (`data:image/png;base64,…`) — an <img> never runs script.
+    src: [...defaultSchema.protocols?.src ?? [], "data"]
+  }
+};
+function sanitizeHtmlFragment(html, schema = SAFE_HTML_SCHEMA) {
+  if (!html) return "";
+  const tree = sanitize(fromHtml(html, { fragment: true }), schema);
+  return toHtml(tree, { characterReferences: { useNamedReferences: true } });
+}
+function sanitizePrintHtml(html) {
+  return sanitizeHtmlFragment(html, PRINT_HTML_SCHEMA);
+}
+
 // src/markdown-css.ts
 var MARKDOWN_DOCUMENT_TOKENS_CSS = `:root {
     --matrx-print-accent: #667eea;
@@ -1053,7 +1150,8 @@ function splitAtPageBreaks(markdown) {
 // src/markdown.ts
 function removeThinkingContent(content) {
   if (!content) return "";
-  let result = content.replace(/\n*<thinking>[\s\S]*?<\/thinking>\n*/gi, "\n").replace(/\n*<think>[\s\S]*?<\/think>\n*/gi, "\n").replace(/\n*<reasoning>[\s\S]*?<\/reasoning>\n*/gi, "\n");
+  const hide = (_match, before, after) => before.length >= 2 || after.length >= 2 ? "\n\n" : "\n";
+  let result = content.replace(/(\n*)<thinking>[\s\S]*?<\/thinking>(\n*)/gi, hide).replace(/(\n*)<think>[\s\S]*?<\/think>(\n*)/gi, hide).replace(/(\n*)<reasoning>[\s\S]*?<\/reasoning>(\n*)/gi, hide);
   result = result.replace(/\n{3,}/g, "\n\n");
   return result.trim();
 }
@@ -1361,11 +1459,11 @@ function protectRegions(markdown, g, stash, ctx) {
     text = next;
   }
   text = sanitizeRawTags(text);
-  if (text !== before && !ctx.strippedAnnounced) {
+  if ((text !== before || hasDisallowedTag(text)) && !ctx.strippedAnnounced) {
     ctx.strippedAnnounced = true;
     ctx.options.onNotice?.({
       code: "html-stripped",
-      message: "Unsafe or page-level markup (a <script>, <style>, <head>, an <html>/<body> shell, an on\u2026 event handler, or a javascript: URL) was removed from the rendered copy.",
+      message: "Markup that is not on the safe list (a <script>, <style>, <iframe>, <form>, <object>/<embed>, <meta>, <link>, an on\u2026 event handler, inline style, or a javascript: URL) was removed from the rendered copy.",
       remedy: 'To show a full HTML page, put it in an ```html fence or an <artifact type="html"> block \u2014 it is printed as its source.'
     });
   }
@@ -1377,6 +1475,14 @@ ${stash.block(flashcardsHtml(g, cards, ctx))}
   });
   return text;
 }
+var ALLOWED_TAGS = new Set(PRINT_HTML_SCHEMA.tagNames ?? []);
+function hasDisallowedTag(text) {
+  for (const m of text.matchAll(/<\/?([a-zA-Z][\w:-]*)/g)) {
+    const tag = m[1].toLowerCase();
+    if (tag !== "flashcards" && !ALLOWED_TAGS.has(tag)) return true;
+  }
+  return false;
+}
 var URL_ATTRS = /* @__PURE__ */ new Set(["href", "src", "action", "formaction", "xlink:href", "poster", "background"]);
 function decodeForSchemeCheck(value) {
   return value.replace(/&#x([0-9a-f]+);?/gi, (_m, h) => String.fromCharCode(parseInt(h, 16))).replace(/&#(\d+);?/g, (_m, d) => String.fromCharCode(Number(d))).replace(/&colon;/gi, ":").replace(/[\u0000-\u0020]+/g, "").toLowerCase();
@@ -1431,7 +1537,7 @@ function splitTableRow(row) {
   if (r.endsWith("|") && !r.endsWith("\\|")) r = r.slice(0, -1);
   return r.split(/(?<!\\)\|/).map((c) => c.trim().replace(/\\\|/g, "|"));
 }
-function cellAlign(separator) {
+function cellAlign2(separator) {
   const s = separator.trim();
   const left = s.startsWith(":");
   const right = s.endsWith(":");
@@ -1445,7 +1551,7 @@ function convertTables(html, g) {
     tableRegex,
     (_m, headerRow, separatorRow, bodyRows) => {
       const headers = splitTableRow(headerRow);
-      const aligns = splitTableRow(separatorRow).map(cellAlign);
+      const aligns = splitTableRow(separatorRow).map(cellAlign2);
       const rows = bodyRows.trim().split("\n").filter((r) => r.trim()).map(splitTableRow);
       let t = `<table${cls(g.table)}>
 <thead${cls(g.thead)}>
@@ -1559,7 +1665,9 @@ function markdownToHtml(markdown, options = {}) {
     return "";
   }
   const source = scrubTokens(options.includeThinking ? markdown : removeThinkingContent(markdown));
-  return `<div class="matrx-content-container">${documentBody(source, newContext(options))}</div>`;
+  return sanitizePrintHtml(
+    `<div class="matrx-content-container">${documentBody(source, newContext(options))}</div>`
+  );
 }
 function documentBody(markdown, ctx) {
   const g = DOCUMENT_GRAMMAR;
@@ -1629,7 +1737,7 @@ function markdownToArticleHtml(markdown, options = {}) {
     return "";
   }
   const source = scrubTokens(options.includeThinking ? markdown : removeThinkingContent(markdown));
-  return articleBody(source, newContext(options));
+  return sanitizePrintHtml(articleBody(source, newContext(options)));
 }
 function articleBody(markdown, ctx) {
   const g = ARTICLE_GRAMMAR;