@portone/docx-editor 0.3.0 → 0.5.0

package/dist/docx/headersFooters.js

@@ -1,131 +1,32 @@
 // src/docx/headersFooters.ts
-import { ALIGN_BY_JC, isOnElement } from "../ooxml/units.js";
+import { toParagraphFormat } from "../model/format.js";
+import { NAMESPACES } from "../ooxml/names.js";
 import {
-  attributeByLocalName,
-  decodeUtf8,
-  elementChildren,
-  parseXml,
-  R_NS,
-  W_NS
-} from "../ooxml/xml.js";
+  ensureRootDeclarations
+} from "../ooxml/partSplice.js";
+import { isOnElement } from "../ooxml/units.js";
+import { decodeUtf8, encodeUtf8, parseXml, R_NS, W_NS } from "../ooxml/xml.js";
+import { docxSchema } from "../schema/index.js";
+import { sameSource } from "../schema/sourceEquality.js";
+import { storyKey, storyNodeOf } from "../schema/stories.js";
+import { NO_EXPORT_REFS } from "./exportRefs.js";
+import { fieldSpans, isFieldCharacter } from "./fields.js";
 import { relatedPartPath } from "./packageParts.js";
 import { readRelationships, relsPathOf, resolveTarget } from "./relationships.js";
+import {
+  HEADER_FOOTER_VARIANTS
+} from "./sections.js";
+import { serializeStory } from "./serializeStory.js";
+import {
+  readStory,
+  storyLeafText
+} from "./story.js";
 var EMPTY_VARIANTS = {
   default: null,
   first: null,
   even: null
 };
-var NO_HEADERS_FOOTERS = {
-  headers: EMPTY_VARIANTS,
-  footers: EMPTY_VARIANTS,
-  firstPageDifferent: false,
-  evenAndOdd: false,
-  pageNumberStart: 1
-};
-function firstSectPrIn(root) {
-  const namespaced = root.getElementsByTagNameNS(W_NS, "sectPr").item(0);
-  if (namespaced) return namespaced;
-  for (const el of root.getElementsByTagName("*")) {
-    if (el.localName === "sectPr") return el;
-  }
-  return null;
-}
-function pageField(instruction) {
-  const name = instruction.trim().split(/\s+/, 1)[0]?.toUpperCase();
-  return name === "PAGE" || name === "NUMPAGES" ? name : null;
-}
-function paragraphSegments(paragraph) {
-  const segments = [];
-  const fields = [];
-  const suppressed = () => fields.some((field) => !field.separated || field.field !== null);
-  const text = (value) => {
-    if (!value || suppressed()) return;
-    const previous = segments.at(-1);
-    if (previous?.kind === "text") previous.value += value;
-    else segments.push({ kind: "text", value });
-  };
-  const dynamic = (field) => {
-    if (!suppressed()) segments.push({ kind: "field", field });
-  };
-  const visit = (el) => {
-    if (el.namespaceURI === W_NS && (el.localName === "drawing" || el.localName === "pict" || el.localName === "object" || el.localName === "tbl" || el.localName === "txbxContent")) {
-      return;
-    }
-    if (el.namespaceURI === W_NS && el.localName === "fldSimple") {
-      const field = pageField(attributeByLocalName(el, "instr") ?? "");
-      if (field) dynamic(field);
-      else for (const child of elementChildren(el)) visit(child);
-      return;
-    }
-    if (el.namespaceURI === W_NS && el.localName === "fldChar") {
-      const kind = attributeByLocalName(el, "fldCharType");
-      if (kind === "begin") {
-        fields.push({ instruction: "", separated: false, field: null });
-      } else if (kind === "separate") {
-        const active = fields.at(-1);
-        if (active) {
-          active.field = pageField(active.instruction);
-          active.separated = true;
-          if (active.field) {
-            const outerSuppresses = fields.slice(0, -1).some((field) => !field.separated || field.field !== null);
-            if (!outerSuppresses) {
-              segments.push({ kind: "field", field: active.field });
-            }
-          }
-        }
-      } else if (kind === "end") {
-        const active = fields.pop();
-        if (active && !active.separated) {
-          const field = pageField(active.instruction);
-          if (field) dynamic(field);
-        }
-      }
-      return;
-    }
-    if (el.namespaceURI === W_NS && el.localName === "instrText") {
-      const active = fields.at(-1);
-      if (active && !active.separated)
-        active.instruction += el.textContent ?? "";
-      return;
-    }
-    if (el.namespaceURI === W_NS && el.localName === "t") {
-      text(el.textContent ?? "");
-      return;
-    }
-    if (el.namespaceURI === W_NS && el.localName === "tab") {
-      text("	");
-      return;
-    }
-    if (el.namespaceURI === W_NS && (el.localName === "br" || el.localName === "cr")) {
-      text("\n");
-      return;
-    }
-    for (const child of elementChildren(el)) visit(child);
-  };
-  for (const child of elementChildren(paragraph)) visit(child);
-  return segments;
-}
-function readContent(bytes) {
-  const root = parseXml(decodeUtf8(bytes).text).documentElement;
-  const paragraphs = elementChildren(root).filter(
-    (child) => child.namespaceURI === W_NS && child.localName === "p"
-  );
-  const segments = [];
-  paragraphs.forEach((paragraph, index) => {
-    if (index > 0) segments.push({ kind: "text", value: "\n" });
-    segments.push(...paragraphSegments(paragraph));
-  });
-  const pPr = paragraphs[0] ? elementChildren(paragraphs[0]).find(
-    (child) => child.namespaceURI === W_NS && child.localName === "pPr"
-  ) : void 0;
-  const jc = pPr ? elementChildren(pPr).find(
-    (child) => child.namespaceURI === W_NS && child.localName === "jc"
-  ) : void 0;
-  return {
-    segments,
-    align: jc ? ALIGN_BY_JC[attributeByLocalName(jc, "val") ?? ""] ?? null : null
-  };
-}
+var PART_ROOT = { header: "hdr", footer: "ftr" };
 function settingsEvenAndOdd(parts, mainPartPath) {
   const path = relatedPartPath(parts, mainPartPath, `${R_NS}/settings`);
   const bytes = path === null ? void 0 : parts.get(path);
@@ -134,76 +35,166 @@ function settingsEvenAndOdd(parts, mainPartPath) {
   const setting = root.getElementsByTagNameNS(W_NS, "evenAndOddHeaders").item(0);
   return isOnElement(setting);
 }
-function readVariants(parts, mainPartPath, sectPr, kind) {
-  const relationships = new Map(
-    readRelationships(parts, relsPathOf(mainPartPath)).map((entry) => [
-      entry.id,
-      entry
-    ])
-  );
-  const variants = { ...EMPTY_VARIANTS };
-  for (const reference of elementChildren(sectPr)) {
-    if (reference.localName !== `${kind}Reference`) continue;
-    const type = attributeByLocalName(reference, "type") ?? "default";
-    if (type !== "default" && type !== "first" && type !== "even") continue;
-    const id = reference.getAttributeNS(R_NS, "id") ?? attributeByLocalName(reference, "id");
-    if (!id) continue;
-    const relationship = relationships.get(id);
-    if (!relationship || relationship.external || relationship.type !== `${R_NS}/${kind}`) {
+function kindOf(relationshipType) {
+  if (relationshipType === `${R_NS}/header`) return "header";
+  if (relationshipType === `${R_NS}/footer`) return "footer";
+  return null;
+}
+function readHeaderFooterStories(parts, mainPartPath, referenced, depsFor) {
+  const keyByRelId = /* @__PURE__ */ new Map();
+  const read = /* @__PURE__ */ new Map();
+  for (const relationship of readRelationships(
+    parts,
+    relsPathOf(mainPartPath)
+  )) {
+    if (!referenced.has(relationship.id)) continue;
+    const kind = relationship.external ? null : kindOf(relationship.type);
+    if (kind === null) continue;
+    const partPath = resolveTarget(mainPartPath, relationship.target);
+    const key = storyKey(kind, partPath);
+    if (read.has(key)) {
+      keyByRelId.set(relationship.id, key);
+      continue;
+    }
+    const bytes = parts.get(partPath);
+    if (!bytes) continue;
+    const xml = decodeUtf8(bytes).text;
+    const root = parseXml(xml).documentElement;
+    if (root.namespaceURI !== W_NS || root.localName !== PART_ROOT[kind]) {
       continue;
     }
-    const bytes = parts.get(resolveTarget(mainPartPath, relationship.target));
-    if (bytes) variants[type] = readContent(bytes);
+    read.set(
+      key,
+      readStory(
+        { kind, id: partPath, partPath },
+        { el: root, xml },
+        depsFor(partPath)
+      )
+    );
+    keyByRelId.set(relationship.id, key);
   }
-  return variants;
+  return {
+    stories: Array.from(read.values()),
+    refs: { keyByRelId, evenAndOdd: settingsEvenAndOdd(parts, mainPartPath) }
+  };
 }
-function pageNumberStart(sectPr) {
-  const pgNumType = elementChildren(sectPr).find(
-    (child) => child.localName === "pgNumType"
-  );
-  const declared = pgNumType ? attributeByLocalName(pgNumType, "start") : null;
-  if (declared === null) return 1;
-  const start = Number(declared);
-  return Number.isSafeInteger(start) && start >= 0 ? start : 1;
+function firstAlign(story) {
+  for (let at = 0; at < story.childCount; at += 1) {
+    const block = story.child(at);
+    if (block.type !== docxSchema.nodes.paragraph) continue;
+    return toParagraphFormat(block.attrs.format)?.align ?? null;
+  }
+  return null;
 }
-function readHeadersFooters(parts, mainPartPath, body) {
-  const sectPr = firstSectPrIn(body);
-  if (!sectPr) return NO_HEADERS_FOOTERS;
+function variantContent(ids, kind, refs, storyOf) {
+  const variants = { ...EMPTY_VARIANTS };
+  for (const variant of HEADER_FOOTER_VARIANTS) {
+    const id = ids[variant];
+    const key = id === null ? void 0 : refs.keyByRelId.get(id);
+    if (key === void 0 || !key.startsWith(`${kind}:`)) continue;
+    const story = storyOf(key);
+    if (story !== null) variants[variant] = { story, align: firstAlign(story) };
+  }
+  return variants;
+}
+function variantsFor(section, refs, storyOf) {
   return {
-    headers: readVariants(parts, mainPartPath, sectPr, "header"),
-    footers: readVariants(parts, mainPartPath, sectPr, "footer"),
-    firstPageDifferent: isOnElement(
-      elementChildren(sectPr).find((child) => child.localName === "titlePg") ?? null
-    ),
-    evenAndOdd: settingsEvenAndOdd(parts, mainPartPath),
-    pageNumberStart: pageNumberStart(sectPr)
+    headers: variantContent(section.props.headerRefs, "header", refs, storyOf),
+    footers: variantContent(section.props.footerRefs, "footer", refs, storyOf),
+    firstPageDifferent: section.props.titlePg,
+    evenAndOdd: refs.evenAndOdd,
+    pageNumberStart: section.props.pageNumberStart ?? 1
   };
 }
 function displayPageNumber(headersFooters, page) {
   return headersFooters.pageNumberStart + page - 1;
 }
-function contentForPage(variants, headersFooters, page) {
-  if (page === 1 && headersFooters.firstPageDifferent) return variants.first;
-  if (headersFooters.evenAndOdd && displayPageNumber(headersFooters, page) % 2 === 0) {
-    return variants.even;
+function headerFooterOn(variants, headersFooters, pageInSection) {
+  if (pageInSection === 1 && headersFooters.firstPageDifferent) {
+    return variants.first;
   }
+  const number = headersFooters.pageNumberStart + pageInSection - 1;
+  if (headersFooters.evenAndOdd && number % 2 === 0) return variants.even;
   return variants.default;
 }
-function headerFooterText(variants, headersFooters, page, totalPages) {
-  const content = contentForPage(variants, headersFooters, page);
-  if (!content) return null;
-  return content.segments.map((segment) => {
-    if (segment.kind === "text") return segment.value;
-    return segment.field === "PAGE" ? `${displayPageNumber(headersFooters, page)}` : `${totalPages}`;
-  }).join("");
+function pageField(instruction) {
+  const name = instruction.split(/\s+/, 1)[0]?.toUpperCase();
+  return name === "PAGE" || name === "NUMPAGES" ? name : null;
+}
+function hiddenBy(field) {
+  if (pageField(field.instr) !== null) {
+    return { field, from: field.begin, to: field.end };
+  }
+  if (field.begin === field.end) return null;
+  return { field, from: field.begin, to: field.separate ?? field.end };
+}
+function nested(hidden, field) {
+  return hidden.some(
+    (other) => other.field !== field && field.begin >= other.from && field.begin <= other.to
+  );
+}
+function paragraphText(paragraph, page, totalPages) {
+  const fields = fieldSpans(paragraph, 0);
+  const hidden = fields.flatMap((field) => {
+    const range = hiddenBy(field);
+    return range === null ? [] : [range];
+  });
+  const numbers = new Map(
+    fields.flatMap((field) => {
+      const name = pageField(field.instr);
+      if (name === null || nested(hidden, field)) return [];
+      return [[field.begin, name === "PAGE" ? `${page}` : `${totalPages}`]];
+    })
+  );
+  const pieces = [];
+  paragraph.forEach((child, at) => {
+    const number = numbers.get(at);
+    if (number !== void 0) pieces.push(number);
+    if (isFieldCharacter(child)) return;
+    if (hidden.some(({ from, to }) => at >= from && at <= to)) return;
+    pieces.push(storyLeafText(child));
+  });
+  return pieces.join("");
+}
+function headerFooterText(story, page, totalPages) {
+  const lines = [];
+  story.forEach((block) => {
+    if (block.type === docxSchema.nodes.paragraph) {
+      lines.push(paragraphText(block, page, totalPages));
+    }
+  });
+  return lines.join("\n");
 }
-function headerFooterAlign(variants, headersFooters, page) {
-  return contentForPage(variants, headersFooters, page)?.align ?? null;
+var HEADER_MARKUP = { namespaces: { w: NAMESPACES.w } };
+function rewrittenParts(doc, session) {
+  const parts = /* @__PURE__ */ new Map();
+  for (const [key, imported] of session.stories) {
+    if (imported.kind !== "header" && imported.kind !== "footer") continue;
+    const current = storyNodeOf(doc, key);
+    if (current === null || sameSource(current, imported.doc)) continue;
+    const written = serializeStory(current, imported, imported, {
+      ...NO_EXPORT_REFS,
+      session
+    });
+    const hadBom = decodeUtf8(
+      session.parts.get(imported.partPath) ?? new Uint8Array()
+    ).hadBom;
+    parts.set(
+      imported.partPath,
+      encodeUtf8(ensureRootDeclarations(written, HEADER_MARKUP), hadBom)
+    );
+  }
+  return parts.size === 0 ? null : parts;
 }
+var headerFooterPlanner = {
+  name: "headers and footers",
+  plan: (doc, session) => rewrittenParts(doc, session)
+};
 export {
-  NO_HEADERS_FOOTERS,
   displayPageNumber,
-  headerFooterAlign,
+  headerFooterOn,
+  headerFooterPlanner,
   headerFooterText,
-  readHeadersFooters
+  readHeaderFooterStories,
+  variantsFor
 };

package/dist/docx/hyperlink.d.ts

@@ -45,9 +45,8 @@ export declare function relIdIn(prefix: string): string | null;
  *
  * A tag that already names one has only that value rewritten, so a link nobody retargeted comes
  * out as the very string it went in as. A tag that named none is given the attribute, which the
- * specification has supersede any `w:anchor` beside it, along with the namespace it lives in: the
- * body we write into is not always one that declares it, and declaring it again where it is
- * already bound to the same namespace changes nothing.
+ * specification has supersede any `w:anchor` beside it. The namespace it lives in is the part
+ * root's to bind, which `hyperlinkRefs` reports and `docx/exportDocx` sees to.
  */
 export declare function withRelId(prefix: string, relId: string): string;
 /** Which relationship each address goes out on while the body is being written */
@@ -62,6 +61,17 @@ export interface LinkRefs {
 }
 /** What a serializer called on its own, outside an export, has to work with */
 export declare const NO_LINK_REFS: LinkRefs;
+/**
+ * The refs a body is written through, and what the part it went into has to declare afterwards.
+ *
+ * `withRelId` puts the attribute on exactly the tag that arrived without one, and that is the tag
+ * `relIdOf` is handed a null `was` for, so what is recorded here and what is written cannot come
+ * apart.
+ */
+export interface LinkWriting extends LinkRefs {
+    /** Whether a link went out under an `r:id` its own opening tag did not arrive carrying */
+    addedRelId(): boolean;
+}
 /**
  * Hands out the relationship for every address the body points at, adding one where the package
  * has none.
@@ -72,4 +82,4 @@ export declare const NO_LINK_REFS: LinkRefs;
  * relationship a link arrived on is kept wherever it still leads where the link now goes, so
  * retargeting one link in a document leaves every other link's XML alone.
  */
-export declare function hyperlinkRefs(relationships: RelationshipWriter): LinkRefs;
+export declare function hyperlinkRefs(relationships: RelationshipWriter): LinkWriting;

package/dist/docx/hyperlink.js

@@ -34,7 +34,7 @@ function withRelId(prefix, relId) {
     return prefix.replace(REL_ID_VALUE, ` ${REL_ID_ATTR}="${relId}"`);
   }
   const name = "<w:hyperlink";
-  return `${name} xmlns:r="${R_NS}" ${REL_ID_ATTR}="${relId}"${prefix.slice(name.length)}`;
+  return `${name} ${REL_ID_ATTR}="${relId}"${prefix.slice(name.length)}`;
 }
 var NO_LINK_REFS = { relIdOf: () => void 0 };
 function hyperlinkRefs(relationships) {
@@ -42,8 +42,11 @@ function hyperlinkRefs(relationships) {
     (rel) => rel.type === HYPERLINK_REL_TYPE && rel.external
   );
   const added = /* @__PURE__ */ new Map();
+  let attributeAdded = false;
   return {
+    addedRelId: () => attributeAdded,
     relIdOf: (href, was) => {
+      if (was === null) attributeAdded = true;
       const kept = external.find((rel) => rel.id === was);
       if (kept?.target === href) return was ?? void 0;
       const found = external.find((rel) => rel.target === href);

package/dist/docx/importDocx.d.ts

@@ -3,7 +3,7 @@
  *
  * This module knows nothing about the screen. All it knows is the file and the document model.
  */
-import { type Node as PMNode } from "prosemirror-model";
+import type { Node as PMNode } from "prosemirror-model";
 import { type XmlParser } from "../ooxml/xml";
 import { type FidelityNote } from "./fidelity";
 import { SessionStore } from "./session";