@bendyline/squisq-formats 2.4.6 → 2.5.1

package/dist/{chunk-N2ZN3MQN.js → chunk-T4PX33AG.js}

@@ -3,6 +3,10 @@ import {
   extractFilename,
   inferMimeType
 } from "./chunk-6RQOV3B3.js";
+import {
+  FootnoteIndex,
+  footnoteIds
+} from "./chunk-BXWNU4T5.js";
 
 // src/html/index.ts
 import JSZip3 from "jszip";
@@ -290,8 +294,9 @@ function markdownDocToPlainHtml(doc, options = {}) {
   } = options;
   const resolveId = themeId ?? readFrontmatterThemeId2(doc.frontmatter);
   const theme = options.theme ?? (resolveId ? resolveThemeForDoc(doc, resolveId, options.themeRegistry) : void 0);
-  const ctx = { images, links, htmlPolicy };
-  const body = renderTopLevel(doc.children, ctx);
+  const footnotes = new FootnoteIndex(doc);
+  const ctx = { images, links, htmlPolicy, footnotes };
+  const body = renderTopLevel(doc.children, ctx) + renderFootnotesSection(footnotes, ctx);
   const fontsLink = theme && externalResources === "allow" ? renderFontsLink(theme) : "";
   const usesIcons = docUsesIcons(doc);
   let iconsLink = "";
@@ -355,6 +360,28 @@ function renderTopLevel(children, ctx) {
   }
   return out.join("\n");
 }
+function renderFootnotesSection(footnotes, ctx) {
+  if (footnotes.isEmpty) return "";
+  const items = footnotes.ordered().map((fn) => {
+    const { def } = footnoteIds(fn.identifier);
+    const body = fn.definition ? childrenToHtml(fn.definition, ctx) : "";
+    const backlinks = Array.from({ length: Math.max(fn.citations, 1) }, (_unused, i) => {
+      const { ref } = footnoteIds(fn.identifier, i + 1);
+      const label = fn.citations > 1 ? `\u21A9${String.fromCharCode(65038)}${i + 1}` : "\u21A9";
+      return `<a href="#${escapeAttr(ref)}" class="squisq-footnote-backref" aria-label="Back to reference ${fn.number}">${label}</a>`;
+    }).join(" ");
+    return `<li id="${escapeAttr(def)}">${body}${backlinks}</li>`;
+  });
+  const lines = [
+    "",
+    '<section class="squisq-footnotes" data-footnotes><hr>',
+    "<ol>",
+    ...items,
+    "</ol>",
+    "</section>"
+  ];
+  return lines.join("\n");
+}
 function findSectionEnd(nodes, from) {
   for (let i = from + 1; i < nodes.length; i++) {
     const n = nodes[i];
@@ -645,6 +672,20 @@ function nodeToHtml(node, ctx) {
       return `<em>${childrenToHtml(node, ctx)}</em>`;
     case "delete":
       return `<del>${childrenToHtml(node, ctx)}</del>`;
+    case "superscript":
+      return `<sup>${childrenToHtml(node, ctx)}</sup>`;
+    case "footnoteReference": {
+      if (!ctx?.footnotes) {
+        return `<sup class="squisq-footnote-ref">${escapeHtml2(node.label ?? node.identifier)}</sup>`;
+      }
+      const { number, occurrence } = ctx.footnotes.cite(node.identifier);
+      const { ref, def } = footnoteIds(node.identifier, occurrence);
+      return `<sup class="squisq-footnote-ref"><a href="#${escapeAttr(def)}" id="${escapeAttr(ref)}">${number}</a></sup>`;
+    }
+    case "footnoteDefinition":
+      return "";
+    case "subscript":
+      return `<sub>${childrenToHtml(node, ctx)}</sub>`;
     case "inlineCode":
       return `<code>${escapeHtml2(node.value ?? "")}</code>`;
     case "code": {
@@ -1238,6 +1279,199 @@ import {
   stringifyMarkdown
 } from "@bendyline/squisq/markdown";
 import { docToMarkdown } from "@bendyline/squisq/doc";
+
+// src/html/footnoteImport.ts
+var TRAILING_WHITESPACE = /[ \t\r\n]+$/;
+var WHITESPACE = /[ \t\r\n]+/;
+var VARIATION_SELECTORS = /[\ufe0e\ufe0f]/gu;
+var RETURN_ARROW = /^[ \t\r\n\u2191\u21a9\u21b5^]+$/u;
+var MARKER_LABEL = /^[ \t[(]*(?:[0-9]{1,4}|[a-z]|[*†‡])[ \t\])]*$/i;
+function isElement(node) {
+  return node.type === "htmlElement";
+}
+function attr(el, name) {
+  return el.attributes[name];
+}
+function classList(el) {
+  return (attr(el, "class") ?? "").split(WHITESPACE).filter(Boolean);
+}
+function isFootnotesContainer(el) {
+  if (attr(el, "data-footnotes") !== void 0) return true;
+  if (attr(el, "epub:type")?.includes("footnotes")) return true;
+  const tag = el.tagName.toLowerCase();
+  if (tag !== "section" && tag !== "aside" && tag !== "div" && tag !== "ol") return false;
+  return classList(el).some(
+    (c) => c === "footnotes" || c === "footnote-list" || c === "squisq-footnotes"
+  );
+}
+function textOf(node) {
+  if (node.type === "htmlText") return node.value;
+  if (!isElement(node)) return "";
+  return node.children.map(textOf).join("");
+}
+function isBacklink(el) {
+  if (el.tagName.toLowerCase() !== "a") return false;
+  if (attr(el, "data-footnote-backref") !== void 0) return true;
+  if (classList(el).some((c) => c.includes("backref") || c === "footnote-back")) return true;
+  return RETURN_ARROW.test(textOf(el).replace(VARIATION_SELECTORS, ""));
+}
+function findElement(nodes, match) {
+  for (const node of nodes) {
+    if (!isElement(node)) continue;
+    if (match(node)) return node;
+    const nested = findElement(node.children, match);
+    if (nested) return nested;
+  }
+  return null;
+}
+function stripBacklinks(nodes) {
+  const out = [];
+  for (const node of nodes) {
+    if (isElement(node)) {
+      if (isBacklink(node)) continue;
+      out.push({ ...node, children: stripBacklinks(node.children) });
+      continue;
+    }
+    out.push(node);
+  }
+  return out;
+}
+function removeElement(nodes, target) {
+  const out = [];
+  for (const node of nodes) {
+    if (node === target) continue;
+    if (isElement(node)) {
+      out.push({ ...node, children: removeElement(node.children, target) });
+      continue;
+    }
+    out.push(node);
+  }
+  return out;
+}
+function trimTrailingText(nodes) {
+  const out = [...nodes];
+  for (let i = out.length - 1; i >= 0; i--) {
+    const node = out[i];
+    if (node.type === "htmlText") {
+      const trimmed = node.value.replace(TRAILING_WHITESPACE, "");
+      if (trimmed === "") {
+        out.pop();
+        continue;
+      }
+      out[i] = { ...node, value: trimmed };
+      break;
+    }
+    if (isElement(node)) {
+      out[i] = { ...node, children: trimTrailingText(node.children) };
+    }
+    break;
+  }
+  return out;
+}
+function collectDefinitionItems(container) {
+  const items = /* @__PURE__ */ new Map();
+  const walk = (nodes) => {
+    for (const node of nodes) {
+      if (!isElement(node)) continue;
+      const id = attr(node, "id");
+      if (node.tagName.toLowerCase() === "li" && id) {
+        items.set(id, trimTrailingText(stripBacklinks(node.children)));
+        continue;
+      }
+      walk(node.children);
+    }
+  };
+  walk(container.children);
+  return items;
+}
+function readableIdentifier(rawId, taken) {
+  const short = rawId.replace(/^user-content-/, "").replace(/^fn[-_:]?/i, "").replace(/^footnote[-_:]?/i, "");
+  if (short === "" || taken.has(short)) return rawId;
+  return short;
+}
+function extractFootnoteSection(nodes) {
+  const container = findElement(nodes, isFootnotesContainer);
+  if (!container) return { nodes, bodies: /* @__PURE__ */ new Map() };
+  const raw = collectDefinitionItems(container);
+  if (raw.size === 0) return { nodes, bodies: /* @__PURE__ */ new Map() };
+  const bodies = /* @__PURE__ */ new Map();
+  const identifiers = /* @__PURE__ */ new Map();
+  const taken = /* @__PURE__ */ new Set();
+  for (const [rawId, body] of raw) {
+    const identifier = readableIdentifier(rawId, taken);
+    taken.add(identifier);
+    identifiers.set(rawId, identifier);
+    bodies.set(identifier, body);
+  }
+  return { nodes: removeElement(nodes, container), bodies, identifiers };
+}
+function inlineText(nodes) {
+  return nodes.map(
+    (n) => "value" in n && typeof n.value === "string" ? n.value : "children" in n && Array.isArray(n.children) ? inlineText(n.children) : ""
+  ).join("");
+}
+function isBlockish(node) {
+  switch (node.type) {
+    case "paragraph":
+    case "heading":
+    case "blockquote":
+    case "list":
+    case "listItem":
+    case "table":
+    case "tableRow":
+    case "tableCell":
+    case "code":
+    case "footnoteDefinition":
+      return true;
+    default:
+      return false;
+  }
+}
+function linkFootnoteReferences(blocks, known) {
+  const rewriteInlines = (nodes, inSuperscript) => {
+    for (let i = 0; i < nodes.length; i++) {
+      const node = nodes[i];
+      if (node.type === "link") {
+        const id = node.url.startsWith("#") ? node.url.slice(1) : null;
+        const identifier = id ? known.get(id) : void 0;
+        if (identifier && (inSuperscript || MARKER_LABEL.test(inlineText(node.children)))) {
+          nodes[i] = { type: "footnoteReference", identifier };
+        }
+        continue;
+      }
+      if (node.type === "superscript") {
+        rewriteInlines(node.children, true);
+        const only = node.children.length === 1 ? node.children[0] : null;
+        if (only?.type === "footnoteReference") nodes[i] = only;
+        continue;
+      }
+      if ("children" in node && Array.isArray(node.children)) {
+        rewriteInlines(node.children, inSuperscript);
+      }
+    }
+  };
+  const walk = (list) => {
+    for (const block of list) {
+      if (!("children" in block) || !Array.isArray(block.children)) continue;
+      const children = block.children;
+      if (children.length > 0 && children.every(isBlockish)) {
+        walk(children);
+      } else {
+        rewriteInlines(children, false);
+      }
+    }
+  };
+  walk(blocks);
+}
+function buildFootnoteDefinitions(bodies) {
+  return [...bodies.entries()].map(([identifier, children]) => ({
+    type: "footnoteDefinition",
+    identifier,
+    children
+  }));
+}
+
+// src/html/import.ts
 var HEADINGS = {
   h1: 1,
   h2: 2,
@@ -1267,8 +1501,10 @@ var DROP = /* @__PURE__ */ new Set(["script", "style", "head", "title", "noscrip
 var INLINE_STRONG = /* @__PURE__ */ new Set(["strong", "b"]);
 var INLINE_EM = /* @__PURE__ */ new Set(["em", "i"]);
 var INLINE_DEL = /* @__PURE__ */ new Set(["del", "s", "strike"]);
-var INLINE_TRANSPARENT = /* @__PURE__ */ new Set(["span", "font", "abbr", "mark", "small", "sub", "sup", "u"]);
-var isElement = (n) => n.type === "htmlElement";
+var INLINE_SUP = /* @__PURE__ */ new Set(["sup"]);
+var INLINE_SUB = /* @__PURE__ */ new Set(["sub"]);
+var INLINE_TRANSPARENT = /* @__PURE__ */ new Set(["span", "font", "abbr", "mark", "small", "u"]);
+var isElement2 = (n) => n.type === "htmlElement";
 var isText = (n) => n.type === "htmlText";
 var collapseWs = (s) => s.replace(/\s+/g, " ");
 function inlinesFromNodes(nodes) {
@@ -1279,7 +1515,7 @@ function inlinesFromNodes(nodes) {
       if (value) out.push({ type: "text", value });
       continue;
     }
-    if (!isElement(node)) continue;
+    if (!isElement2(node)) continue;
     const tag = node.tagName.toLowerCase();
     if (DROP.has(tag)) continue;
     if (tag === "br") {
@@ -1290,6 +1526,10 @@ function inlinesFromNodes(nodes) {
       out.push({ type: "emphasis", children: inlinesFromNodes(node.children) });
     } else if (INLINE_DEL.has(tag)) {
       out.push({ type: "delete", children: inlinesFromNodes(node.children) });
+    } else if (INLINE_SUP.has(tag)) {
+      out.push({ type: "superscript", children: inlinesFromNodes(node.children) });
+    } else if (INLINE_SUB.has(tag)) {
+      out.push({ type: "subscript", children: inlinesFromNodes(node.children) });
     } else if (tag === "code" || tag === "kbd" || tag === "samp" || tag === "tt") {
       out.push({ type: "inlineCode", value: textContent(node) });
     } else if (tag === "a") {
@@ -1312,7 +1552,7 @@ function inlinesFromNodes(nodes) {
 }
 function textContent(node) {
   if (isText(node)) return node.value;
-  if (isElement(node)) return node.children.map(textContent).join("");
+  if (isElement2(node)) return node.children.map(textContent).join("");
   return "";
 }
 var onlyWhitespace = (inlines) => inlines.every((n) => n.type === "text" && n.value.trim() === "");
@@ -1332,7 +1572,7 @@ function blocksFromNodes(nodes) {
       inlineBuffer.push(node);
       continue;
     }
-    if (!isElement(node)) continue;
+    if (!isElement2(node)) continue;
     const tag = node.tagName.toLowerCase();
     if (DROP.has(tag)) continue;
     const block = blockForElement(node, tag);
@@ -1381,7 +1621,7 @@ function blockForElement(node, tag) {
 function listFromElement(node, ordered) {
   const items = [];
   for (const child of node.children) {
-    if (isElement(child) && child.tagName.toLowerCase() === "li") {
+    if (isElement2(child) && child.tagName.toLowerCase() === "li") {
       const blocks = blocksFromNodes(child.children);
       items.push({ type: "listItem", children: blocks });
     }
@@ -1399,12 +1639,12 @@ function tableFromElement(node) {
   const rows = [];
   const collectRows = (n) => {
     for (const child of n.children) {
-      if (!isElement(child)) continue;
+      if (!isElement2(child)) continue;
       const t = child.tagName.toLowerCase();
       if (t === "tr") {
         const cells = [];
         for (const cell of child.children) {
-          if (isElement(cell) && (cell.tagName.toLowerCase() === "td" || cell.tagName.toLowerCase() === "th")) {
+          if (isElement2(cell) && (cell.tagName.toLowerCase() === "td" || cell.tagName.toLowerCase() === "th")) {
             cells.push({ type: "tableCell", children: inlinesFromNodes(cell.children) });
           }
         }
@@ -1426,7 +1666,7 @@ function toHtmlString(data) {
 }
 function findEmbeddedSquisqDoc(nodes) {
   for (const node of nodes) {
-    if (!isElement(node)) continue;
+    if (!isElement2(node)) continue;
     if (node.tagName.toLowerCase() === "script" && "data-squisq-doc" in node.attributes) {
       return node;
     }
@@ -1460,11 +1700,19 @@ function htmlToMarkdownDocSync(html, options = {}) {
   if (html.length > maxInputChars) {
     throw new RangeError(`HTML exceeds the ${maxInputChars}-character safety limit`);
   }
-  let nodes = parseHtmlToNodes(html);
+  const nodes = parseHtmlToNodes(html);
   const embedded = findEmbeddedSquisqDoc(nodes);
   if (embedded) return docToMarkdown(parseEmbeddedSquisqDoc(embedded));
-  if (options.sanitize !== false) nodes = sanitizeHtmlNodes2(nodes);
-  return { type: "document", children: blocksFromNodes(nodes) };
+  const { nodes: bodyNodes, bodies, identifiers } = extractFootnoteSection(nodes);
+  const clean = (input) => options.sanitize === false ? input : sanitizeHtmlNodes2(input);
+  const children = blocksFromNodes(clean(bodyNodes));
+  if (bodies.size > 0) {
+    linkFootnoteReferences(children, identifiers ?? /* @__PURE__ */ new Map());
+    const definitions = /* @__PURE__ */ new Map();
+    for (const [id, body] of bodies) definitions.set(id, blocksFromNodes(clean(body)));
+    children.push(...buildFootnoteDefinitions(definitions));
+  }
+  return { type: "document", children };
 }
 async function htmlToMarkdownDoc(data, options = {}) {
   return htmlToMarkdownDocSync(toHtmlString(data), options);

package/dist/{chunk-BVGJ54NV.js → chunk-TVUX3RUC.js}

@@ -1,7 +1,7 @@
 import {
   convert,
   defaultRegistry
-} from "./chunk-DNGNODJP.js";
+} from "./chunk-5JQ5NDRC.js";
 import {
   ConversionError
 } from "./chunk-KXOZMWBS.js";

package/dist/docx/index.js

@@ -4,9 +4,9 @@ import {
   docxToDoc,
   docxToMarkdownDoc,
   markdownDocToDocx
-} from "../chunk-BGDN7BXK.js";
-import "../chunk-DWNNYO5H.js";
-import "../chunk-6X5XN3CZ.js";
+} from "../chunk-FIOSE4BO.js";
+import "../chunk-A6N6IN3I.js";
+import "../chunk-AVOZAKGP.js";
 import "../chunk-ILCJ3WFD.js";
 import "../chunk-JU2RHXUB.js";
 import "../chunk-S5PCVMKU.js";
@@ -14,6 +14,7 @@ import "../chunk-7AWFHP5U.js";
 import "../chunk-IIQYS2YH.js";
 import "../chunk-USU6HTKB.js";
 import "../chunk-AONELFLA.js";
+import "../chunk-BXWNU4T5.js";
 export {
   docToDocx,
   docxToContainer,

package/dist/epub/index.js

@@ -1,12 +1,13 @@
 import {
   docToEpub,
   markdownDocToEpub
-} from "../chunk-KN62QJYK.js";
-import "../chunk-6X5XN3CZ.js";
+} from "../chunk-2LT3JL7U.js";
+import "../chunk-AVOZAKGP.js";
 import "../chunk-JU2RHXUB.js";
 import "../chunk-USU6HTKB.js";
 import "../chunk-6RQOV3B3.js";
 import "../chunk-AONELFLA.js";
+import "../chunk-BXWNU4T5.js";
 export {
   docToEpub,
   markdownDocToEpub