@adobe/helix-docx2md 1.0.21 → 1.1.0

package/CHANGELOG.md

@@ -1,3 +1,24 @@
+# [1.1.0](https://github.com/adobe/helix-docx2md/compare/v1.0.23...v1.1.0) (2022-09-09)
+
+
+### Features
+
+* add support for gridtables  ([aff5f95](https://github.com/adobe/helix-docx2md/commit/aff5f951e1172f7f9ca4e14c8935c6e4b0890dd9))
+
+## [1.0.23](https://github.com/adobe/helix-docx2md/compare/v1.0.22...v1.0.23) (2022-08-29)
+
+
+### Bug Fixes
+
+* **deps:** update dependency @adobe/helix-mediahandler to v1.0.42 ([#108](https://github.com/adobe/helix-docx2md/issues/108)) ([ae7726c](https://github.com/adobe/helix-docx2md/commit/ae7726ccce4b26802c42b515288930ce2c328218))
+
+## [1.0.22](https://github.com/adobe/helix-docx2md/compare/v1.0.21...v1.0.22) (2022-07-28)
+
+
+### Bug Fixes
+
+* **deps:** update dependency @adobe/helix-markdown-support to v3.1.8 ([0884578](https://github.com/adobe/helix-docx2md/commit/0884578e7fb5e190e4791972fd307242485a55d2))
+
 ## [1.0.21](https://github.com/adobe/helix-docx2md/compare/v1.0.20...v1.0.21) (2022-07-26)
 
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/helix-docx2md",
-  "version": "1.0.21",
+  "version": "1.1.0",
   "description": "Helix library that converts word documents to markdown",
   "type": "module",
   "main": "./src/index.js",
@@ -13,7 +13,6 @@
   },
   "scripts": {
     "test": "c8 mocha",
-    "test-ci": "c8 mocha && codecov",
     "lint": "eslint .",
     "semantic-release": "semantic-release"
   },
@@ -34,9 +33,9 @@
   },
   "homepage": "https://github.com/adobe/helix-docx2md#readme",
   "dependencies": {
-    "@adobe/helix-markdown-support": "3.1.7",
+    "@adobe/helix-markdown-support": "5.0.7",
     "@adobe/helix-shared-process-queue": "1.1.5",
-    "@adobe/mammoth": "1.4.15-bleeding.1",
+    "@adobe/mammoth": "1.5.1-bleeding.1",
     "dirname-filename-esm": "1.1.1",
     "mdast-util-to-markdown": "1.3.0",
     "mdast-util-to-string": "3.1.0",
@@ -45,26 +44,27 @@
     "remark-stringify": "10.0.2",
     "unified": "10.1.2",
     "unist-util-find": "1.0.2",
-    "unist-util-visit": "4.1.0",
+    "unist-util-visit": "4.1.1",
     "yauzl": "2.10.0"
   },
   "devDependencies": {
     "@adobe/eslint-config-helix": "1.3.2",
-    "@adobe/helix-mediahandler": "1.0.36",
+    "@adobe/helix-mediahandler": "1.2.0",
     "@semantic-release/changelog": "6.0.1",
     "@semantic-release/exec": "6.0.3",
     "@semantic-release/git": "10.0.1",
     "c8": "7.12.0",
-    "dotenv": "16.0.1",
-    "eslint": "8.20.0",
+    "dotenv": "16.0.2",
+    "eslint": "8.23.0",
+    "eslint-import-resolver-exports": "1.0.0-beta.2",
     "eslint-plugin-header": "3.1.1",
     "eslint-plugin-import": "2.26.0",
     "husky": "8.0.1",
-    "junit-report-builder": "3.0.0",
+    "junit-report-builder": "3.0.1",
     "lint-staged": "13.0.3",
     "mocha": "10.0.0",
     "mocha-multi-reporters": "1.5.1",
-    "semantic-release": "19.0.3",
+    "semantic-release": "19.0.5",
     "unist-util-inspect": "7.0.1"
   },
   "lint-staged": {

package/src/dast2mdast/dast2mdast.d.ts

@@ -13,6 +13,7 @@
 /**
  * Converts the docx AST to markdown ast.
  * @param {object} tree the docx ast
+ * @param {boolean} [opts.gridtables = false] generate gridtables
  * @return {object} the markdown ast
  */
-export default function dast2mdast(tree: object): object;
+export default function dast2mdast(tree: object, opts?: object): object;

package/src/dast2mdast/dast2mdast.js

@@ -30,9 +30,11 @@ import handlers from './handlers/index.js';
  *
  * Converts the docx AST to markdown ast.
  * @param {object} tree the docx ast
+ * @param {object} opts options
+ * @param {boolean} [opts.gridtables = false] generate gridtables
  * @return {object} the markdown ast
  */
-export default async function dast2mdast(tree) {
+export default async function dast2mdast(tree, opts = {}) {
   const byId = {};
 
   const h = (type, props, children) => {
@@ -59,6 +61,7 @@ export default async function dast2mdast(tree) {
   h.frozenBaseUrl = null;
   h.handlers = handlers;
   h.numbering = {};
+  h.gridtables = opts.gridtables;
 
   /** @type {ListContainers} */
   h.listContainers = [[]];

package/src/dast2mdast/handlers/table-cell.js

@@ -9,9 +9,45 @@
  * OF ANY KIND, either express or implied. See the License for the specific language
  * governing permissions and limitations under the License.
  */
+import { TYPE_CELL } from '@adobe/helix-markdown-support/gridtable';
 import all from '../all.js';
 
+const V_ALIGN = {
+  top: '',
+  center: 'middle',
+  bottom: 'bottom',
+};
+
+const H_ALIGN = {
+  left: '',
+  right: 'right',
+  center: 'center',
+  distribute: 'justify',
+};
+
+function gridTableCell(h, node) {
+  const props = {};
+  if (node.children.length > 0) {
+    props.align = H_ALIGN[node.children[0].alignment];
+  }
+  props.valign = V_ALIGN[node.verticalAlignment];
+
+  if (node.rowSpan > 1) {
+    props.rowSpan = node.rowSpan;
+  }
+  if (node.colSpan > 1) {
+    props.colSpan = node.colSpan;
+  }
+  h.listContainers.unshift([]);
+  const c = h(TYPE_CELL, props, all(h, node));
+  h.listContainers.shift();
+  return c;
+}
+
 export default function cell(h, node) {
+  if (h.gridtables) {
+    return gridTableCell(h, node);
+  }
   // remember alignment for table
   const props = {};
   if (node.children.length > 0) {

package/src/dast2mdast/handlers/table-row.js

@@ -9,8 +9,20 @@
  * OF ANY KIND, either express or implied. See the License for the specific language
  * governing permissions and limitations under the License.
  */
+import { TYPE_ROW } from '@adobe/helix-markdown-support/gridtable';
 import all from '../all.js';
 
+function gridTableRow(h, node) {
+  const props = {};
+  if (node.isHeader) {
+    props.isHeader = true;
+  }
+  return h(TYPE_ROW, props, all(h, node));
+}
+
 export default function row(h, node) {
+  if (h.gridtables) {
+    return gridTableRow(h, node);
+  }
   return h('tableRow', all(h, node));
 }

package/src/dast2mdast/handlers/table.js

@@ -9,9 +9,32 @@
  * OF ANY KIND, either express or implied. See the License for the specific language
  * governing permissions and limitations under the License.
  */
+import { TYPE_TABLE, TYPE_HEADER, TYPE_BODY } from '@adobe/helix-markdown-support/gridtable';
 import all from '../all.js';
 
+function gridTable(h, node) {
+  let rows = all(h, node);
+  if (rows[0]?.isHeader) {
+    const header = [];
+    while (rows[0]?.isHeader) {
+      const row = rows.shift();
+      delete row.isHeader;
+      header.push(row);
+    }
+    rows = [
+      h(TYPE_HEADER, header),
+      h(TYPE_BODY, rows),
+    ];
+  }
+
+  return h(TYPE_TABLE, rows);
+}
+
 export default function table(h, node) {
+  if (h.gridtables) {
+    return gridTable(h, node);
+  }
+
   const nodes = all(h, node);
 
   // get alignment of first row, if any

package/src/docx2md.d.ts

@@ -66,6 +66,8 @@ export declare interface Docx2mdOptions {
 
   parserOptions?: object;
 
+  gridtables?: boolean;
+
 }
 
 /**

package/src/docx2md.js

@@ -32,7 +32,7 @@ export default async function docx2md(doc, opts) {
     listener('beforeToMdast', dast);
   }
 
-  const mdast = await dast2mdast(dast);
+  const mdast = await dast2mdast(dast, opts);
 
   if (listener) {
     listener('afterToMdast', mdast);

package/src/mdast2md/mdast2md.d.ts

@@ -23,6 +23,8 @@ export declare interface MDast2mdOptions {
   mediaHandler?: MediaHandler,
 
   source?: string,
+
+  gridtables?: boolean,
 }
 
 /**

package/src/mdast2md/mdast2md.js

@@ -21,12 +21,15 @@ import {
   robustTables,
   suppressSpaceCode,
   sanitizeHeading,
-  remarkMatter,
   breaksAsSpaces,
   sanitizeFormats,
   sanitizeText,
+  sanitizeTextAndFormats,
   sanitizeLinks,
+  imageReferences,
 } from '@adobe/helix-markdown-support';
+import { remarkMatter } from '@adobe/helix-markdown-support/matter';
+import { remarkGridTable } from '@adobe/helix-markdown-support/gridtable';
 
 import processImages from './mdast-process-images.js';
 import sanitizeAutoEmbeds from './mdast-sanitize-autoembeds.js';
@@ -45,18 +48,46 @@ export default async function mdast2md(mdast, opts = {}) {
     listener('beforePostProcess', mdast);
   }
 
+  const processor = unified()
+    .use(stringify, {
+      strong: '*',
+      emphasis: '_',
+      bullet: '-',
+      fence: '`',
+      fences: true,
+      incrementListMarker: true,
+      rule: '-',
+      ruleRepetition: 3,
+      ruleSpaces: false,
+    })
+    .use(gfm)
+    .use(remarkMatter)
+    .use(orderedListPlugin);
+
   // process.stdout.write(inspect(mdast));
   // process.stdout.write('\n');
 
-  await sanitizeHeading(mdast);
-  await sanitizeFormats(mdast); // collapse formats once
-  await sanitizeLinks(mdast);
-  await sanitizeFormats(mdast); // and again for sanitized links
-  await sanitizeText(mdast);
-  await suppressSpaceCode(mdast);
-  await sanitizeAutoEmbeds(mdast);
-  await processImages(log, mdast, opts.mediaHandler, opts.source);
-  await robustTables(mdast);
+  if (opts.gridtables) {
+    await sanitizeHeading(mdast);
+    await sanitizeLinks(mdast);
+    await sanitizeTextAndFormats(mdast);
+    await suppressSpaceCode(mdast);
+    await sanitizeAutoEmbeds(mdast);
+    await processImages(log, mdast, opts.mediaHandler, opts.source);
+    await imageReferences(mdast);
+    processor.use(remarkGridTable);
+  } else {
+    await sanitizeHeading(mdast);
+    await sanitizeFormats(mdast); // collapse formats once
+    await sanitizeLinks(mdast);
+    await sanitizeFormats(mdast); // and again for sanitized links
+    await sanitizeText(mdast);
+    await suppressSpaceCode(mdast);
+    await sanitizeAutoEmbeds(mdast);
+    await processImages(log, mdast, opts.mediaHandler, opts.source);
+    await robustTables(mdast);
+    processor.use(breaksAsSpaces);
+  }
 
   // process.stdout.write(inspect(mdast));
   // process.stdout.write('\n');
@@ -67,26 +98,7 @@ export default async function mdast2md(mdast, opts = {}) {
   }
 
   // noinspection JSVoidFunctionReturnValueUsed
-  const md = unified()
-    .use(stringify, {
-      strong: '*',
-      emphasis: '_',
-      bullet: '-',
-      fence: '`',
-      fences: true,
-      incrementListMarker: true,
-      rule: '-',
-      ruleRepetition: 3,
-      ruleSpaces: false,
-    })
-    .use(gfm, {
-      // tableCellPadding: false,
-      // tablePipeAlign: false,
-    })
-    .use(breaksAsSpaces)
-    .use(remarkMatter)
-    .use(orderedListPlugin)
-    .stringify(mdast);
+  const md = processor.stringify(mdast);
 
   if (listener) {
     listener('afterToMarkdown', md);