Fix anchor links to sections of documents

Resolves: #472
mieweb · Jun 23, 2024 · cb08db6 · cb08db6
1 parent 0d13ac8
commit cb08db6
Show file tree

Hide file tree

Showing 9 changed files with 51 additions and 44 deletions.
diff --git a/src/containers/transform/LocalLinks.ts b/src/containers/transform/LocalLinks.ts
@@ -1,5 +1,5 @@
 import {FileContentService} from '../../utils/FileContentService.ts';
-import {FileId} from '../../model/model.js';
+import {FileId} from '../../model/model.ts';
 
 interface Link {
   fileId: string;
@@ -78,7 +78,7 @@ export class LocalLinks {
       if (link.fileId === fileId) {
         const links = link.links
           .filter(link => link.startsWith('gdoc:'))
-          .map(link => link.substring('gdoc:'.length));
+          .map(link => link.substring('gdoc:'.length).replace(/#.*/, ''));
 
         return links.map(fileId => ({
           fileId,

diff --git a/src/containers/transform/TransformContainer.ts b/src/containers/transform/TransformContainer.ts
@@ -24,6 +24,7 @@ import {MarkdownTreeProcessor} from './MarkdownTreeProcessor.ts';
 import {LunrIndexer} from '../search/LunrIndexer.ts';
 import {JobManagerContainer} from '../job/JobManagerContainer.ts';
 import {UserConfigService} from '../google_folder/UserConfigService.ts';
+import {getUrlHash} from '../../utils/idParsers.ts';
 
 const __filename = fileURLToPath(import.meta.url);
 
@@ -342,10 +343,10 @@ export class TransformContainer extends Container {
           processed.add(fileId);
           const backLinks = this.localLinks.getBackLinks(fileId);
           for (const backLink of backLinks) {
-            if (processed.has(backLink)) {
+            if (processed.has(backLink.fileId)) {
               continue;
             }
-            filterFilesIds.add(backLink);
+            filterFilesIds.add(backLink.fileId);
           }
         }
         if (filterFilesIds.size > 0) {
@@ -426,7 +427,8 @@ export class TransformContainer extends Container {
       if (fileName.endsWith('.md') || fileName.endsWith('.svg')) {
         const content = await destinationDirectory.readFile(fileName);
         const newContent = content.replace(/(gdoc:[A-Z0-9_-]+)/ig, (str: string) => {
-          const fileId = str.substring('gdoc:'.length);
+          const fileId = str.substring('gdoc:'.length).replace(/#.*/, '');
+          const hash = getUrlHash(str);
           const lastLog = this.localLog.findLastFile(fileId);
           if (lastLog && lastLog.event !== 'removed') {
             if (fileName.endsWith('.svg')) {
@@ -435,7 +437,7 @@ export class TransformContainer extends Container {
               return convertToRelativeMarkDownPath(lastLog.filePath, destinationDirectory.getVirtualPath() + fileName);
             }
           } else {
-            return 'https://drive.google.com/open?id=' + fileId;
+            return 'https://drive.google.com/open?id=' + fileId + hash.replace('#_', '#heading=h.');
           }
         });
         if (content !== newContent) {

diff --git a/src/containers/transform/frontmatters/generateDocumentFrontMatter.ts b/src/containers/transform/frontmatters/generateDocumentFrontMatter.ts
@@ -1,7 +1,7 @@
 import yaml from 'js-yaml';
 
-import {MdFile} from '../../../model/LocalFile';
-import {FRONTMATTER_DUMP_OPTS} from './frontmatter';
+import {MdFile} from '../../../model/LocalFile.ts';
+import {FRONTMATTER_DUMP_OPTS} from './frontmatter.ts';
 
 export function generateDocumentFrontMatter(localFile: MdFile, links: string[],
                                             fm_without_version = false) {

diff --git a/src/odt/LibreOffice.ts b/src/odt/LibreOffice.ts
@@ -76,7 +76,8 @@ export class TextLink implements ParagraphSection {
 
 @XmlElement()
 @XmlAttribute('text:name', 'name')
-export class TextBookmark {
+export class TextBookmark implements ParagraphSection {
+  type = 'bookmark';
   name: string;
 }
 
@@ -200,7 +201,7 @@ export class TextChangeEnd {
 @XmlElement()
 @XmlText('list', {isArray: true})
 @XmlAttribute('text:style-name', 'styleName')
-@XmlElementChild('text:bookmark', 'bookmark', 'TextBookmark')
+@XmlElementChild('text:bookmark', 'list', 'TextBookmark', {isArray: true})
 @XmlElementChild('text:a', 'list', 'TextLink', {isArray: true})
 @XmlElementChild('text:span', 'list', 'TextSpan', {isArray: true})
 @XmlElementChild('draw:rect', 'list', 'DrawRect', {isArray: true})
@@ -215,8 +216,7 @@ export class TextChangeEnd {
 @XmlElementChild('text:change-end', 'list', 'TextChangeEnd', {isArray: true})
 export class TextParagraph implements TextSection {
   type = 'paragraph';
-  bookmark: TextBookmark;
-  list: Array<string | TextLink | TextSpan | DrawRect | DrawFrame | TextTab | TextLineBreak | TextSpace | DrawG | TextChangeStart | TextChangeEnd | DrawCustomShape> = [];
+  list: Array<string | TextLink | TextSpan | DrawRect | DrawFrame | TextTab | TextLineBreak | TextSpace | DrawG | TextChangeStart | TextChangeEnd | DrawCustomShape | TextBookmark> = [];
   annotations: OfficeAnnotation[] = [];
   styleName: string;
 }

diff --git a/src/odt/MarkdownNodes.ts b/src/odt/MarkdownNodes.ts
@@ -16,7 +16,7 @@ export type TAG = 'BODY' | 'HR/' | 'B' | 'I' | 'BI' | 'BLANK/' | // | '/B' | '/I
   'EMB_SVG' | 'EMB_SVG_G' | 'EMB_SVG_P/' | 'EMB_SVG_TEXT' | // | '/EMB_SVG' | '/EMB_SVG_G' | '/EMB_SVG_TEXT'
   'EMB_SVG_TSPAN' | // | '/EMB_SVG_TSPAN'
   'MATHML' |
-  'CHANGE_START' | 'CHANGE_END' | 'RAW_MODE/' | 'HTML_MODE/' | 'MD_MODE/' | 'MACRO_MODE/' | 'COMMENT';
+  'CHANGE_START' | 'CHANGE_END' | 'RAW_MODE/' | 'HTML_MODE/' | 'MD_MODE/' | 'MACRO_MODE/' | 'COMMENT' | 'BOOKMARK/';
 
 export interface TagPayload {
   lang?: string;
@@ -34,7 +34,6 @@ export interface TagPayload {
   listStyle?: ListStyle;
   continueNumbering?: boolean;
   listLevel?: number;
-  bookmarkName?: string;
   pathD?: string;
 
   x?: number;

diff --git a/src/odt/OdtToMarkdown.ts b/src/odt/OdtToMarkdown.ts
@@ -10,15 +10,15 @@ import {
   TableCell,
   TableOfContent,
   TableRow,
-  TableTable,
+  TableTable, TextBookmark,
   TextLink,
   TextList,
   TextParagraph,
   TextProperty,
   TextSpace,
   TextSpan
 } from './LibreOffice.ts';
-import {urlToFolderId} from '../utils/idParsers.ts';
+import {getUrlHash, urlToFolderId} from '../utils/idParsers.ts';
 import {MarkdownNodes, MarkdownTagNode} from './MarkdownNodes.ts';
 import {inchesToPixels, inchesToSpaces, spaces} from './utils.ts';
 import {extractPath} from './extractPath.ts';
@@ -230,13 +230,14 @@ export class OdtToMarkdown {
   async linkToText(currentTagNode: MarkdownTagNode, link: TextLink): Promise<void> {
     let href = link.href;
     const id = urlToFolderId(href);
+    const hash = getUrlHash(link.href);
     if (id) {
       href = 'gdoc:' + id;
     }
 
-    this.addLink(href);
+    this.addLink(href + hash);
 
-    const block = this.chunks.createNode('A', { href: href });
+    const block = this.chunks.createNode('A', { href: href + hash });
     this.chunks.append(currentTagNode, block);
     currentTagNode = block;
 
@@ -458,38 +459,36 @@ export class OdtToMarkdown {
     return false;
   }
 
-
   async paragraphToText(currentTagNode: MarkdownTagNode, paragraph: TextParagraph): Promise<void> {
     const style = this.getStyle(paragraph.styleName);
     const listStyle = this.getListStyle(style.listStyleName);
-    const bookmarkName = paragraph.bookmark?.name || null;
 
     if (this.hasStyle(paragraph, 'Heading_20_1')) {
-      const header = this.chunks.createNode('H1', { marginLeft: inchesToSpaces(style.paragraphProperties?.marginLeft), style, listStyle, bookmarkName });
+      const header = this.chunks.createNode('H1', { marginLeft: inchesToSpaces(style.paragraphProperties?.marginLeft), style, listStyle });
       this.chunks.append(currentTagNode, header);
       currentTagNode = header;
     } else
     if (this.hasStyle(paragraph, 'Heading_20_2')) {
-      const header = this.chunks.createNode('H2', { marginLeft: inchesToSpaces(style.paragraphProperties?.marginLeft), style, listStyle, bookmarkName });
+      const header = this.chunks.createNode('H2', { marginLeft: inchesToSpaces(style.paragraphProperties?.marginLeft), style, listStyle });
       this.chunks.append(currentTagNode, header);
       currentTagNode = header;
     } else
     if (this.hasStyle(paragraph, 'Heading_20_3')) {
-      const header = this.chunks.createNode('H3', { marginLeft: inchesToSpaces(style.paragraphProperties?.marginLeft), style, listStyle, bookmarkName });
+      const header = this.chunks.createNode('H3', { marginLeft: inchesToSpaces(style.paragraphProperties?.marginLeft), style, listStyle });
       this.chunks.append(currentTagNode, header);
       currentTagNode = header;
     } else
     if (this.hasStyle(paragraph, 'Heading_20_4')) {
-      const header = this.chunks.createNode('H4', { marginLeft: inchesToSpaces(style.paragraphProperties?.marginLeft), style, listStyle, bookmarkName });
+      const header = this.chunks.createNode('H4', { marginLeft: inchesToSpaces(style.paragraphProperties?.marginLeft), style, listStyle });
       this.chunks.append(currentTagNode, header);
       currentTagNode = header;
     } else
     if (this.isCourier(paragraph.styleName)) {
-      const block = this.chunks.createNode('PRE', { marginLeft: inchesToSpaces(style.paragraphProperties?.marginLeft), style, listStyle, bookmarkName });
+      const block = this.chunks.createNode('PRE', { marginLeft: inchesToSpaces(style.paragraphProperties?.marginLeft), style, listStyle });
       this.chunks.append(currentTagNode, block);
       currentTagNode = block;
     } else {
-      const block = this.chunks.createNode('P', { marginLeft: inchesToSpaces(style.paragraphProperties?.marginLeft), style, listStyle, bookmarkName });
+      const block = this.chunks.createNode('P', { marginLeft: inchesToSpaces(style.paragraphProperties?.marginLeft), style, listStyle });
       this.chunks.append(currentTagNode, block);
       currentTagNode = block;
     }
@@ -596,6 +595,12 @@ export class OdtToMarkdown {
         case 'change_end':
           this.chunks.append(currentTagNode, this.chunks.createNode('CHANGE_END'));
           break;
+        case 'bookmark':
+          {
+            const bookmark = <TextBookmark>child;
+            this.chunks.append(currentTagNode, this.chunks.createNode('BOOKMARK/', { id: bookmark.name }));
+          }
+          break;
       }
     }
   }

diff --git a/src/odt/markdownNodesUtils.ts b/src/odt/markdownNodesUtils.ts
@@ -221,6 +221,8 @@ function chunkToText(chunk: MarkdownNode, ctx: ToTextContext) {
           return addLiNumbers(chunk, ctx, chunksToText(chunk.children, { ...ctx, inListItem: true, parentLevel: chunk.payload.listLevel }));
         case 'TOC':
           return chunksToText(chunk.children, ctx); // TODO
+        case 'BOOKMARK/':
+          return `<a id="${chunk.payload.id}"></a>`;
       }
       return chunksToText(chunk.children, ctx);
     case 'html':
@@ -296,6 +298,8 @@ function chunkToText(chunk: MarkdownNode, ctx: ToTextContext) {
             const fontSize = inchesToPixels(chunk.payload.style?.textProperties.fontSize);
             return `<tspan style="${textStyleToString(chunk.payload.style?.textProperties)}" font-size="${fontSize}">` + chunksToText(chunk.children, ctx) + '</tspan>\n';
           }
+        case 'BOOKMARK/':
+          return `<a id="${chunk.payload.id}"></a>`;
       }
       return chunksToText(chunk.children, ctx);
     default:

diff --git a/src/odt/postprocess/rewriteHeaders.ts b/src/odt/postprocess/rewriteHeaders.ts
@@ -1,27 +1,16 @@
-import slugify from 'slugify';
-import {extractText, walkRecursiveAsync, walkRecursiveSync} from '../markdownNodesUtils.ts';
+import {walkRecursiveAsync} from '../markdownNodesUtils.ts';
 import {MarkdownNodes} from '../MarkdownNodes.ts';
 
 export async function rewriteHeaders(markdownChunks: MarkdownNodes) {
-  const headersMap = {};
-
   await walkRecursiveAsync(markdownChunks.body, async (chunk) => {
-    if (chunk.isTag === true && ['H1', 'H2', 'H3', 'H4'].includes(chunk.tag)) { // && 'md' === this.currentMode) {
-      if (chunk.payload.bookmarkName) {
-        const innerTxt = extractText(chunk);
-        const slug = slugify(innerTxt.trim(), { replacement: '-', lower: true, remove: /[#*+~.()'"!:@]/g });
-        if (slug) {
-          headersMap['#' + chunk.payload.bookmarkName] = '#' + slug;
+    if (chunk.isTag && ['H1', 'H2', 'H3', 'H4'].includes(chunk.tag)) {
+      if (chunk.children.length > 1) {
+        const first = chunk.children[0];
+        if (first.isTag && first.tag === 'BOOKMARK/') {
+          const toMove = chunk.children.splice(0, 1);
+          chunk.children.splice(chunk.children.length, 0, ...toMove);
         }
       }
     }
   });
-
-  walkRecursiveSync(markdownChunks.body, (chunk) => {
-    if (chunk.isTag === true && chunk.payload?.href) {
-      if (headersMap[chunk.payload.href]) {
-        chunk.payload.href = headersMap[chunk.payload.href];
-      }
-    }
-  });
 }
diff --git a/src/utils/idParsers.ts b/src/utils/idParsers.ts
@@ -71,3 +71,11 @@ export function urlToFolderId(url: string): string | null {
 
   return null;
 }
+
+export function getUrlHash(url: string): string {
+  const idx = url.indexOf('#');
+  if (idx >= 0) {
+    return url.substring(idx).replace('#heading=h.', '#_');
+  }
+  return '';
+}