@x15ventures/x15-ui-library 3.18.0 → 3.20.0

package/dist/esm/widgets/shared/documents/engines/markdownRenderer/plugins/remarkPlugins.js

@@ -0,0 +1,182 @@
+import { isLikelyLayoutTable, LAYOUT_TABLE_MARKER } from './tableHelpers';
+const getText = node => node.value ?? node.children?.map(getText).join(' ') ?? '';
+
+// Converted documents use `<br>` for line breaks; without raw HTML support they would vanish.
+export const remarkLineBreaks = () => {
+  const transform = node => {
+    if (node.type === 'html' && /^<br\s*\/?>$/i.test(node.value?.trim() ?? '')) {
+      node.type = 'break';
+      delete node.value;
+    }
+    node.children?.forEach(transform);
+  };
+  return transform;
+};
+const splitInlineContent = nodes => {
+  const paragraphs = [[]];
+  nodes.forEach(node => {
+    let segments;
+    if (node.type === 'text') {
+      segments = (node.value ?? '').split(/\r?\n/).map(value => [{
+        ...node,
+        value
+      }]);
+    } else if (node.children) {
+      segments = splitInlineContent(node.children).map(children => [{
+        ...node,
+        children
+      }]);
+    } else {
+      segments = [[node]];
+    }
+    segments.forEach((segment, index) => {
+      if (index > 0) {
+        paragraphs.push([]);
+      }
+      paragraphs[paragraphs.length - 1].push(...segment);
+    });
+  });
+  return paragraphs;
+};
+
+// Source documents treat each line as its own paragraph, whereas markdown joins adjacent lines.
+export const remarkParagraphBreaks = () => {
+  const transform = parent => {
+    if (!parent.children || ['list', 'listItem', 'table', 'tableCell'].includes(parent.type)) {
+      return;
+    }
+    parent.children = parent.children.flatMap(node => {
+      if (node.type === 'paragraph') {
+        return splitInlineContent(node.children ?? []).map(children => ({
+          ...node,
+          children
+        }));
+      }
+      transform(node);
+      return [node];
+    });
+  };
+  return transform;
+};
+const markLayoutTable = node => {
+  const properties = node.data?.hProperties ?? {};
+  const className = [...new Set([...(properties.className ?? []), 'layoutTable'])];
+  node.data = {
+    ...node.data,
+    hProperties: {
+      ...properties,
+      className
+    }
+  };
+};
+export const remarkLayoutTables = () => {
+  const transform = parent => {
+    if (!parent.children) {
+      return;
+    }
+    parent.children = parent.children.filter((node, index, siblings) => {
+      const isMarker = node.type === 'html' && node.value?.trim() === LAYOUT_TABLE_MARKER;
+      if (isMarker && siblings[index + 1]?.type === 'table') {
+        markLayoutTable(siblings[index + 1]);
+        return false;
+      }
+      return true;
+    });
+    parent.children.forEach(node => {
+      if (node.type === 'table' && node.children) {
+        const rows = node.children.map((row, rowIndex) => row.children?.map(cell => ({
+          text: getText(cell),
+          isHeader: rowIndex === 0
+        })) ?? []);
+        if (isLikelyLayoutTable(rows)) {
+          markLayoutTable(node);
+        }
+      }
+      transform(node);
+    });
+  };
+  return transform;
+};
+const getInlineList = (text, allowSingleLeadingBullet = false) => {
+  const leadingBullets = /^-\s+(.+)\.$/.exec(text);
+  if (leadingBullets) {
+    const items = leadingBullets[1].split(/\.\s+-\s+/).map(item => item.trim());
+    if ((items.length >= 2 || allowSingleLeadingBullet) && items.every(Boolean)) {
+      return {
+        prefix: '',
+        items: items.map(item => `${item}.`)
+      };
+    }
+  }
+  const prefixedBullets = /^(.*?:)\s+-\s+(.+)$/.exec(text);
+  if (!prefixedBullets) {
+    return null;
+  }
+  const items = prefixedBullets[2].split(/;\s+(?:and\s+)?-\s+/).map(item => item.trim());
+  if (items.length < 2 || items.some(item => !item)) {
+    return null;
+  }
+  return {
+    prefix: `${prefixedBullets[1]} `,
+    items: items.map((item, index) => index < items.length - 1 ? `${item};` : item)
+  };
+};
+const getPlainText = cell => cell.children?.length === 1 && cell.children[0].type === 'text' ? cell.children[0].value ?? '' : null;
+const setCellList = (cell, list) => {
+  cell.children = [...(list.prefix ? [{
+    type: 'text',
+    value: list.prefix
+  }] : []), {
+    type: 'list',
+    children: list.items.map(value => ({
+      type: 'listItem',
+      children: [{
+        type: 'paragraph',
+        children: [{
+          type: 'text',
+          value
+        }]
+      }]
+    }))
+  }];
+};
+const getCells = row => row?.children?.filter(cell => cell.type === 'tableCell') ?? [];
+const containsList = cell => cell?.children?.some(child => child.type === 'list') ?? false;
+
+// A lone "- item." is only a list when the same column holds a list in a neighbouring row.
+const convertRelatedSingleBullets = table => {
+  const rows = table.children?.filter(row => row.type === 'tableRow') ?? [];
+  rows.forEach((row, rowIndex) => {
+    getCells(row).forEach((cell, cellIndex) => {
+      const text = getPlainText(cell);
+      const list = text === null ? null : getInlineList(text, true);
+      if (list?.items.length !== 1) {
+        return;
+      }
+      if (containsList(getCells(rows[rowIndex - 1])[cellIndex]) || containsList(getCells(rows[rowIndex + 1])[cellIndex])) {
+        setCellList(cell, list);
+      }
+    });
+  });
+};
+
+// Table cells cannot hold markdown lists, so converters flatten them to "- a. - b." text.
+export const remarkTableCellLists = () => {
+  const transform = node => {
+    if (node.type === 'table') {
+      node.children?.forEach(transform);
+      convertRelatedSingleBullets(node);
+      return;
+    }
+    if (node.type === 'tableCell') {
+      const text = getPlainText(node);
+      const list = text === null ? null : getInlineList(text);
+      if (list) {
+        setCellList(node, list);
+      }
+    }
+    node.children?.forEach(transform);
+  };
+  return transform;
+};
+//# sourceMappingURL=remarkPlugins.js.map

package/dist/esm/widgets/shared/documents/engines/markdownRenderer/plugins/remarkPlugins.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"remarkPlugins.js","names":["isLikelyLayoutTable","LAYOUT_TABLE_MARKER","getText","node","value","children","map","join","remarkLineBreaks","transform","type","test","trim","forEach","splitInlineContent","nodes","paragraphs","segments","split","segment","index","push","length","remarkParagraphBreaks","parent","includes","flatMap","markLayoutTable","properties","data","hProperties","className","Set","remarkLayoutTables","filter","siblings","isMarker","rows","row","rowIndex","cell","text","isHeader","getInlineList","allowSingleLeadingBullet","leadingBullets","exec","items","item","every","Boolean","prefix","prefixedBullets","some","getPlainText","setCellList","list","getCells","containsList","child","convertRelatedSingleBullets","table","cellIndex","remarkTableCellLists"],"sources":["../../../../../../../../src/widgets/shared/documents/engines/markdownRenderer/plugins/remarkPlugins.ts"],"sourcesContent":["import { isLikelyLayoutTable, LAYOUT_TABLE_MARKER } from './tableHelpers';\nimport { MarkdownNode, MarkdownTransformer } from './markdownNode.interfaces';\n\nconst getText = (node: MarkdownNode): string => node.value ?? node.children?.map(getText).join(' ') ?? '';\n\n// Converted documents use `<br>` for line breaks; without raw HTML support they would vanish.\nexport const remarkLineBreaks = (): MarkdownTransformer => {\n  const transform = (node: MarkdownNode): void => {\n    if (node.type === 'html' && /^<br\\s*\\/?>$/i.test(node.value?.trim() ?? '')) {\n      node.type = 'break';\n      delete node.value;\n    }\n    node.children?.forEach(transform);\n  };\n  return transform;\n};\n\nconst splitInlineContent = (nodes: MarkdownNode[]): MarkdownNode[][] => {\n  const paragraphs: MarkdownNode[][] = [[]];\n  nodes.forEach((node) => {\n    let segments: MarkdownNode[][];\n    if (node.type === 'text') {\n      segments = (node.value ?? '').split(/\\r?\\n/).map((value) => [{ ...node, value }]);\n    } else if (node.children) {\n      segments = splitInlineContent(node.children).map((children) => [{ ...node, children }]);\n    } else {\n      segments = [[node]];\n    }\n    segments.forEach((segment, index) => {\n      if (index > 0) {\n        paragraphs.push([]);\n      }\n      paragraphs[paragraphs.length - 1].push(...segment);\n    });\n  });\n  return paragraphs;\n};\n\n// Source documents treat each line as its own paragraph, whereas markdown joins adjacent lines.\nexport const remarkParagraphBreaks = (): MarkdownTransformer => {\n  const transform = (parent: MarkdownNode): void => {\n    if (!parent.children || ['list', 'listItem', 'table', 'tableCell'].includes(parent.type)) {\n      return;\n    }\n    parent.children = parent.children.flatMap((node) => {\n      if (node.type === 'paragraph') {\n        return splitInlineContent(node.children ?? []).map((children) => ({ ...node, children }));\n      }\n      transform(node);\n      return [node];\n    });\n  };\n  return transform;\n};\n\nconst markLayoutTable = (node: MarkdownNode): void => {\n  const properties = node.data?.hProperties ?? {};\n  const className = [...new Set([...(properties.className ?? []), 'layoutTable'])];\n  node.data = { ...node.data, hProperties: { ...properties, className } };\n};\n\nexport const remarkLayoutTables = (): MarkdownTransformer => {\n  const transform = (parent: MarkdownNode): void => {\n    if (!parent.children) {\n      return;\n    }\n    parent.children = parent.children.filter((node, index, siblings) => {\n      const isMarker = node.type === 'html' && node.value?.trim() === LAYOUT_TABLE_MARKER;\n      if (isMarker && siblings[index + 1]?.type === 'table') {\n        markLayoutTable(siblings[index + 1]);\n        return false;\n      }\n      return true;\n    });\n    parent.children.forEach((node) => {\n      if (node.type === 'table' && node.children) {\n        const rows = node.children.map((row, rowIndex) =>\n          row.children?.map((cell) => ({ text: getText(cell), isHeader: rowIndex === 0 })) ?? []\n        );\n        if (isLikelyLayoutTable(rows)) {\n          markLayoutTable(node);\n        }\n      }\n      transform(node);\n    });\n  };\n  return transform;\n};\n\ntype InlineList = {\n  prefix: string;\n  items: string[];\n};\n\nconst getInlineList = (text: string, allowSingleLeadingBullet = false): InlineList | null => {\n  const leadingBullets = /^-\\s+(.+)\\.$/.exec(text);\n  if (leadingBullets) {\n    const items = leadingBullets[1].split(/\\.\\s+-\\s+/).map((item) => item.trim());\n    if ((items.length >= 2 || allowSingleLeadingBullet) && items.every(Boolean)) {\n      return { prefix: '', items: items.map((item) => `${ item }.`) };\n    }\n  }\n\n  const prefixedBullets = /^(.*?:)\\s+-\\s+(.+)$/.exec(text);\n  if (!prefixedBullets) {\n    return null;\n  }\n  const items = prefixedBullets[2].split(/;\\s+(?:and\\s+)?-\\s+/).map((item) => item.trim());\n  if (items.length < 2 || items.some((item) => !item)) {\n    return null;\n  }\n  return {\n    prefix: `${ prefixedBullets[1] } `,\n    items: items.map((item, index) => (index < items.length - 1 ? `${ item };` : item))\n  };\n};\n\nconst getPlainText = (cell: MarkdownNode): string | null =>\n  cell.children?.length === 1 && cell.children[0].type === 'text' ? cell.children[0].value ?? '' : null;\n\nconst setCellList = (cell: MarkdownNode, list: InlineList): void => {\n  cell.children = [\n    ...(list.prefix ? [{ type: 'text', value: list.prefix }] : []),\n    {\n      type: 'list',\n      children: list.items.map((value) => ({\n        type: 'listItem',\n        children: [{ type: 'paragraph', children: [{ type: 'text', value }] }]\n      }))\n    }\n  ];\n};\n\nconst getCells = (row: MarkdownNode | undefined): MarkdownNode[] => row?.children?.filter((cell) => cell.type === 'tableCell') ?? [];\n\nconst containsList = (cell: MarkdownNode | undefined): boolean => cell?.children?.some((child) => child.type === 'list') ?? false;\n\n// A lone \"- item.\" is only a list when the same column holds a list in a neighbouring row.\nconst convertRelatedSingleBullets = (table: MarkdownNode): void => {\n  const rows = table.children?.filter((row) => row.type === 'tableRow') ?? [];\n  rows.forEach((row, rowIndex) => {\n    getCells(row).forEach((cell, cellIndex) => {\n      const text = getPlainText(cell);\n      const list = text === null ? null : getInlineList(text, true);\n      if (list?.items.length !== 1) {\n        return;\n      }\n      if (containsList(getCells(rows[rowIndex - 1])[cellIndex]) || containsList(getCells(rows[rowIndex + 1])[cellIndex])) {\n        setCellList(cell, list);\n      }\n    });\n  });\n};\n\n// Table cells cannot hold markdown lists, so converters flatten them to \"- a. - b.\" text.\nexport const remarkTableCellLists = (): MarkdownTransformer => {\n  const transform = (node: MarkdownNode): void => {\n    if (node.type === 'table') {\n      node.children?.forEach(transform);\n      convertRelatedSingleBullets(node);\n      return;\n    }\n    if (node.type === 'tableCell') {\n      const text = getPlainText(node);\n      const list = text === null ? null : getInlineList(text);\n      if (list) {\n        setCellList(node, list);\n      }\n    }\n    node.children?.forEach(transform);\n  };\n  return transform;\n};\n"],"mappings":"AAAA,SAASA,mBAAmB,EAAEC,mBAAmB,QAAQ,gBAAgB;AAGzE,MAAMC,OAAO,GAAIC,IAAkB,IAAaA,IAAI,CAACC,KAAK,IAAID,IAAI,CAACE,QAAQ,EAAEC,GAAG,CAACJ,OAAO,CAAC,CAACK,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE;;AAEzG;AACA,OAAO,MAAMC,gBAAgB,GAAGA,CAAA,KAA2B;EACzD,MAAMC,SAAS,GAAIN,IAAkB,IAAW;IAC9C,IAAIA,IAAI,CAACO,IAAI,KAAK,MAAM,IAAI,eAAe,CAACC,IAAI,CAACR,IAAI,CAACC,KAAK,EAAEQ,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE;MAC1ET,IAAI,CAACO,IAAI,GAAG,OAAO;MACnB,OAAOP,IAAI,CAACC,KAAK;IACnB;IACAD,IAAI,CAACE,QAAQ,EAAEQ,OAAO,CAACJ,SAAS,CAAC;EACnC,CAAC;EACD,OAAOA,SAAS;AAClB,CAAC;AAED,MAAMK,kBAAkB,GAAIC,KAAqB,IAAuB;EACtE,MAAMC,UAA4B,GAAG,CAAC,EAAE,CAAC;EACzCD,KAAK,CAACF,OAAO,CAAEV,IAAI,IAAK;IACtB,IAAIc,QAA0B;IAC9B,IAAId,IAAI,CAACO,IAAI,KAAK,MAAM,EAAE;MACxBO,QAAQ,GAAG,CAACd,IAAI,CAACC,KAAK,IAAI,EAAE,EAAEc,KAAK,CAAC,OAAO,CAAC,CAACZ,GAAG,CAAEF,KAAK,IAAK,CAAC;QAAE,GAAGD,IAAI;QAAEC;MAAM,CAAC,CAAC,CAAC;IACnF,CAAC,MAAM,IAAID,IAAI,CAACE,QAAQ,EAAE;MACxBY,QAAQ,GAAGH,kBAAkB,CAACX,IAAI,CAACE,QAAQ,CAAC,CAACC,GAAG,CAAED,QAAQ,IAAK,CAAC;QAAE,GAAGF,IAAI;QAAEE;MAAS,CAAC,CAAC,CAAC;IACzF,CAAC,MAAM;MACLY,QAAQ,GAAG,CAAC,CAACd,IAAI,CAAC,CAAC;IACrB;IACAc,QAAQ,CAACJ,OAAO,CAAC,CAACM,OAAO,EAAEC,KAAK,KAAK;MACnC,IAAIA,KAAK,GAAG,CAAC,EAAE;QACbJ,UAAU,CAACK,IAAI,CAAC,EAAE,CAAC;MACrB;MACAL,UAAU,CAACA,UAAU,CAACM,MAAM,GAAG,CAAC,CAAC,CAACD,IAAI,CAAC,GAAGF,OAAO,CAAC;IACpD,CAAC,CAAC;EACJ,CAAC,CAAC;EACF,OAAOH,UAAU;AACnB,CAAC;;AAED;AACA,OAAO,MAAMO,qBAAqB,GAAGA,CAAA,KAA2B;EAC9D,MAAMd,SAAS,GAAIe,MAAoB,IAAW;IAChD,IAAI,CAACA,MAAM,CAACnB,QAAQ,IAAI,CAAC,MAAM,EAAE,UAAU,EAAE,OAAO,EAAE,WAAW,CAAC,CAACoB,QAAQ,CAACD,MAAM,CAACd,IAAI,CAAC,EAAE;MACxF;IACF;IACAc,MAAM,CAACnB,QAAQ,GAAGmB,MAAM,CAACnB,QAAQ,CAACqB,OAAO,CAAEvB,IAAI,IAAK;MAClD,IAAIA,IAAI,CAACO,IAAI,KAAK,WAAW,EAAE;QAC7B,OAAOI,kBAAkB,CAACX,IAAI,CAACE,QAAQ,IAAI,EAAE,CAAC,CAACC,GAAG,CAAED,QAAQ,KAAM;UAAE,GAAGF,IAAI;UAAEE;QAAS,CAAC,CAAC,CAAC;MAC3F;MACAI,SAAS,CAACN,IAAI,CAAC;MACf,OAAO,CAACA,IAAI,CAAC;IACf,CAAC,CAAC;EACJ,CAAC;EACD,OAAOM,SAAS;AAClB,CAAC;AAED,MAAMkB,eAAe,GAAIxB,IAAkB,IAAW;EACpD,MAAMyB,UAAU,GAAGzB,IAAI,CAAC0B,IAAI,EAAEC,WAAW,IAAI,CAAC,CAAC;EAC/C,MAAMC,SAAS,GAAG,CAAC,GAAG,IAAIC,GAAG,CAAC,CAAC,IAAIJ,UAAU,CAACG,SAAS,IAAI,EAAE,CAAC,EAAE,aAAa,CAAC,CAAC,CAAC;EAChF5B,IAAI,CAAC0B,IAAI,GAAG;IAAE,GAAG1B,IAAI,CAAC0B,IAAI;IAAEC,WAAW,EAAE;MAAE,GAAGF,UAAU;MAAEG;IAAU;EAAE,CAAC;AACzE,CAAC;AAED,OAAO,MAAME,kBAAkB,GAAGA,CAAA,KAA2B;EAC3D,MAAMxB,SAAS,GAAIe,MAAoB,IAAW;IAChD,IAAI,CAACA,MAAM,CAACnB,QAAQ,EAAE;MACpB;IACF;IACAmB,MAAM,CAACnB,QAAQ,GAAGmB,MAAM,CAACnB,QAAQ,CAAC6B,MAAM,CAAC,CAAC/B,IAAI,EAAEiB,KAAK,EAAEe,QAAQ,KAAK;MAClE,MAAMC,QAAQ,GAAGjC,IAAI,CAACO,IAAI,KAAK,MAAM,IAAIP,IAAI,CAACC,KAAK,EAAEQ,IAAI,CAAC,CAAC,KAAKX,mBAAmB;MACnF,IAAImC,QAAQ,IAAID,QAAQ,CAACf,KAAK,GAAG,CAAC,CAAC,EAAEV,IAAI,KAAK,OAAO,EAAE;QACrDiB,eAAe,CAACQ,QAAQ,CAACf,KAAK,GAAG,CAAC,CAAC,CAAC;QACpC,OAAO,KAAK;MACd;MACA,OAAO,IAAI;IACb,CAAC,CAAC;IACFI,MAAM,CAACnB,QAAQ,CAACQ,OAAO,CAAEV,IAAI,IAAK;MAChC,IAAIA,IAAI,CAACO,IAAI,KAAK,OAAO,IAAIP,IAAI,CAACE,QAAQ,EAAE;QAC1C,MAAMgC,IAAI,GAAGlC,IAAI,CAACE,QAAQ,CAACC,GAAG,CAAC,CAACgC,GAAG,EAAEC,QAAQ,KAC3CD,GAAG,CAACjC,QAAQ,EAAEC,GAAG,CAAEkC,IAAI,KAAM;UAAEC,IAAI,EAAEvC,OAAO,CAACsC,IAAI,CAAC;UAAEE,QAAQ,EAAEH,QAAQ,KAAK;QAAE,CAAC,CAAC,CAAC,IAAI,EACtF,CAAC;QACD,IAAIvC,mBAAmB,CAACqC,IAAI,CAAC,EAAE;UAC7BV,eAAe,CAACxB,IAAI,CAAC;QACvB;MACF;MACAM,SAAS,CAACN,IAAI,CAAC;IACjB,CAAC,CAAC;EACJ,CAAC;EACD,OAAOM,SAAS;AAClB,CAAC;AAOD,MAAMkC,aAAa,GAAGA,CAACF,IAAY,EAAEG,wBAAwB,GAAG,KAAK,KAAwB;EAC3F,MAAMC,cAAc,GAAG,cAAc,CAACC,IAAI,CAACL,IAAI,CAAC;EAChD,IAAII,cAAc,EAAE;IAClB,MAAME,KAAK,GAAGF,cAAc,CAAC,CAAC,CAAC,CAAC3B,KAAK,CAAC,WAAW,CAAC,CAACZ,GAAG,CAAE0C,IAAI,IAAKA,IAAI,CAACpC,IAAI,CAAC,CAAC,CAAC;IAC7E,IAAI,CAACmC,KAAK,CAACzB,MAAM,IAAI,CAAC,IAAIsB,wBAAwB,KAAKG,KAAK,CAACE,KAAK,CAACC,OAAO,CAAC,EAAE;MAC3E,OAAO;QAAEC,MAAM,EAAE,EAAE;QAAEJ,KAAK,EAAEA,KAAK,CAACzC,GAAG,CAAE0C,IAAI,IAAK,GAAIA,IAAI,GAAI;MAAE,CAAC;IACjE;EACF;EAEA,MAAMI,eAAe,GAAG,qBAAqB,CAACN,IAAI,CAACL,IAAI,CAAC;EACxD,IAAI,CAACW,eAAe,EAAE;IACpB,OAAO,IAAI;EACb;EACA,MAAML,KAAK,GAAGK,eAAe,CAAC,CAAC,CAAC,CAAClC,KAAK,CAAC,qBAAqB,CAAC,CAACZ,GAAG,CAAE0C,IAAI,IAAKA,IAAI,CAACpC,IAAI,CAAC,CAAC,CAAC;EACxF,IAAImC,KAAK,CAACzB,MAAM,GAAG,CAAC,IAAIyB,KAAK,CAACM,IAAI,CAAEL,IAAI,IAAK,CAACA,IAAI,CAAC,EAAE;IACnD,OAAO,IAAI;EACb;EACA,OAAO;IACLG,MAAM,EAAE,GAAIC,eAAe,CAAC,CAAC,CAAC,GAAI;IAClCL,KAAK,EAAEA,KAAK,CAACzC,GAAG,CAAC,CAAC0C,IAAI,EAAE5B,KAAK,KAAMA,KAAK,GAAG2B,KAAK,CAACzB,MAAM,GAAG,CAAC,GAAG,GAAI0B,IAAI,GAAI,GAAGA,IAAK;EACpF,CAAC;AACH,CAAC;AAED,MAAMM,YAAY,GAAId,IAAkB,IACtCA,IAAI,CAACnC,QAAQ,EAAEiB,MAAM,KAAK,CAAC,IAAIkB,IAAI,CAACnC,QAAQ,CAAC,CAAC,CAAC,CAACK,IAAI,KAAK,MAAM,GAAG8B,IAAI,CAACnC,QAAQ,CAAC,CAAC,CAAC,CAACD,KAAK,IAAI,EAAE,GAAG,IAAI;AAEvG,MAAMmD,WAAW,GAAGA,CAACf,IAAkB,EAAEgB,IAAgB,KAAW;EAClEhB,IAAI,CAACnC,QAAQ,GAAG,CACd,IAAImD,IAAI,CAACL,MAAM,GAAG,CAAC;IAAEzC,IAAI,EAAE,MAAM;IAAEN,KAAK,EAAEoD,IAAI,CAACL;EAAO,CAAC,CAAC,GAAG,EAAE,CAAC,EAC9D;IACEzC,IAAI,EAAE,MAAM;IACZL,QAAQ,EAAEmD,IAAI,CAACT,KAAK,CAACzC,GAAG,CAAEF,KAAK,KAAM;MACnCM,IAAI,EAAE,UAAU;MAChBL,QAAQ,EAAE,CAAC;QAAEK,IAAI,EAAE,WAAW;QAAEL,QAAQ,EAAE,CAAC;UAAEK,IAAI,EAAE,MAAM;UAAEN;QAAM,CAAC;MAAE,CAAC;IACvE,CAAC,CAAC;EACJ,CAAC,CACF;AACH,CAAC;AAED,MAAMqD,QAAQ,GAAInB,GAA6B,IAAqBA,GAAG,EAAEjC,QAAQ,EAAE6B,MAAM,CAAEM,IAAI,IAAKA,IAAI,CAAC9B,IAAI,KAAK,WAAW,CAAC,IAAI,EAAE;AAEpI,MAAMgD,YAAY,GAAIlB,IAA8B,IAAcA,IAAI,EAAEnC,QAAQ,EAAEgD,IAAI,CAAEM,KAAK,IAAKA,KAAK,CAACjD,IAAI,KAAK,MAAM,CAAC,IAAI,KAAK;;AAEjI;AACA,MAAMkD,2BAA2B,GAAIC,KAAmB,IAAW;EACjE,MAAMxB,IAAI,GAAGwB,KAAK,CAACxD,QAAQ,EAAE6B,MAAM,CAAEI,GAAG,IAAKA,GAAG,CAAC5B,IAAI,KAAK,UAAU,CAAC,IAAI,EAAE;EAC3E2B,IAAI,CAACxB,OAAO,CAAC,CAACyB,GAAG,EAAEC,QAAQ,KAAK;IAC9BkB,QAAQ,CAACnB,GAAG,CAAC,CAACzB,OAAO,CAAC,CAAC2B,IAAI,EAAEsB,SAAS,KAAK;MACzC,MAAMrB,IAAI,GAAGa,YAAY,CAACd,IAAI,CAAC;MAC/B,MAAMgB,IAAI,GAAGf,IAAI,KAAK,IAAI,GAAG,IAAI,GAAGE,aAAa,CAACF,IAAI,EAAE,IAAI,CAAC;MAC7D,IAAIe,IAAI,EAAET,KAAK,CAACzB,MAAM,KAAK,CAAC,EAAE;QAC5B;MACF;MACA,IAAIoC,YAAY,CAACD,QAAQ,CAACpB,IAAI,CAACE,QAAQ,GAAG,CAAC,CAAC,CAAC,CAACuB,SAAS,CAAC,CAAC,IAAIJ,YAAY,CAACD,QAAQ,CAACpB,IAAI,CAACE,QAAQ,GAAG,CAAC,CAAC,CAAC,CAACuB,SAAS,CAAC,CAAC,EAAE;QAClHP,WAAW,CAACf,IAAI,EAAEgB,IAAI,CAAC;MACzB;IACF,CAAC,CAAC;EACJ,CAAC,CAAC;AACJ,CAAC;;AAED;AACA,OAAO,MAAMO,oBAAoB,GAAGA,CAAA,KAA2B;EAC7D,MAAMtD,SAAS,GAAIN,IAAkB,IAAW;IAC9C,IAAIA,IAAI,CAACO,IAAI,KAAK,OAAO,EAAE;MACzBP,IAAI,CAACE,QAAQ,EAAEQ,OAAO,CAACJ,SAAS,CAAC;MACjCmD,2BAA2B,CAACzD,IAAI,CAAC;MACjC;IACF;IACA,IAAIA,IAAI,CAACO,IAAI,KAAK,WAAW,EAAE;MAC7B,MAAM+B,IAAI,GAAGa,YAAY,CAACnD,IAAI,CAAC;MAC/B,MAAMqD,IAAI,GAAGf,IAAI,KAAK,IAAI,GAAG,IAAI,GAAGE,aAAa,CAACF,IAAI,CAAC;MACvD,IAAIe,IAAI,EAAE;QACRD,WAAW,CAACpD,IAAI,EAAEqD,IAAI,CAAC;MACzB;IACF;IACArD,IAAI,CAACE,QAAQ,EAAEQ,OAAO,CAACJ,SAAS,CAAC;EACnC,CAAC;EACD,OAAOA,SAAS;AAClB,CAAC","ignoreList":[]}

package/dist/esm/widgets/shared/documents/engines/markdownRenderer/plugins/remarkSafeTables.js

@@ -0,0 +1,78 @@
+import DOMPurify from 'dompurify';
+import { hasMeaningfulTableText } from './tableHelpers';
+const ALLOWED_TABLE_TAGS = ['table', 'thead', 'tbody', 'tfoot', 'tr', 'th', 'td', 'caption', 'colgroup', 'col', 'p', 'br', 'strong', 'em', 'b', 'i', 'u', 'del', 's', 'code', 'pre', 'ul', 'ol', 'li', 'a', 'img', 'span', 'div', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6'];
+const ALLOWED_TABLE_ATTRIBUTES = ['colspan', 'rowspan', 'href', 'src', 'alt', 'title', 'class'];
+const ALLOWED_CLASS_NAMES = ['layoutTable'];
+const toHastProperties = element => {
+  const properties = {};
+  Array.from(element.attributes).forEach(({
+    name,
+    value
+  }) => {
+    if (name === 'class') {
+      const className = value.split(/\s+/).filter(item => ALLOWED_CLASS_NAMES.includes(item));
+      if (className.length) {
+        properties.className = className;
+      }
+    } else if (name === 'colspan' || name === 'rowspan') {
+      const span = Number(value);
+      if (Number.isInteger(span) && span > 0) {
+        properties[name === 'colspan' ? 'colSpan' : 'rowSpan'] = span;
+      }
+    } else {
+      properties[name] = value;
+    }
+  });
+  return properties;
+};
+const toHast = node => {
+  if (node.nodeType === Node.TEXT_NODE) {
+    return {
+      type: 'text',
+      value: node.textContent ?? ''
+    };
+  }
+  if (node.nodeType !== Node.ELEMENT_NODE) {
+    return null;
+  }
+  const element = node;
+  return {
+    type: 'element',
+    tagName: element.tagName.toLowerCase(),
+    properties: toHastProperties(element),
+    children: Array.from(element.childNodes).map(toHast).filter(child => child !== null)
+  };
+};
+const sanitizeTable = html => {
+  const fragment = DOMPurify.sanitize(html, {
+    ALLOWED_TAGS: ALLOWED_TABLE_TAGS,
+    ALLOWED_ATTR: ALLOWED_TABLE_ATTRIBUTES,
+    RETURN_DOM_FRAGMENT: true
+  });
+  const meaningfulNodes = Array.from(fragment.childNodes).filter(child => child.nodeType !== Node.TEXT_NODE || hasMeaningfulTableText(child.textContent ?? ''));
+  const [onlyNode] = meaningfulNodes;
+  return meaningfulNodes.length === 1 && onlyNode instanceof Element && onlyNode.tagName === 'TABLE' ? onlyNode : null;
+};
+
+// Converters fall back to raw `<table>` HTML for merged cells, which markdown cannot express.
+// Only a single, sanitised table is rendered; any other raw HTML stays inert text.
+export const remarkSafeTables = () => {
+  const transform = node => {
+    if (node.type === 'html' && /^<table(?:\s|>)/i.test(node.value?.trim() ?? '')) {
+      const table = sanitizeTable(node.value ?? '');
+      const hast = table ? toHast(table) : null;
+      if (hast?.type === 'element') {
+        node.type = 'safeDocumentTable';
+        node.data = {
+          hName: 'table',
+          hProperties: hast.properties,
+          hChildren: hast.children
+        };
+        delete node.value;
+      }
+    }
+    node.children?.forEach(transform);
+  };
+  return transform;
+};
+//# sourceMappingURL=remarkSafeTables.js.map

package/dist/esm/widgets/shared/documents/engines/markdownRenderer/plugins/remarkSafeTables.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"remarkSafeTables.js","names":["DOMPurify","hasMeaningfulTableText","ALLOWED_TABLE_TAGS","ALLOWED_TABLE_ATTRIBUTES","ALLOWED_CLASS_NAMES","toHastProperties","element","properties","Array","from","attributes","forEach","name","value","className","split","filter","item","includes","length","span","Number","isInteger","toHast","node","nodeType","Node","TEXT_NODE","type","textContent","ELEMENT_NODE","tagName","toLowerCase","children","childNodes","map","child","sanitizeTable","html","fragment","sanitize","ALLOWED_TAGS","ALLOWED_ATTR","RETURN_DOM_FRAGMENT","meaningfulNodes","onlyNode","Element","remarkSafeTables","transform","test","trim","table","hast","data","hName","hProperties","hChildren"],"sources":["../../../../../../../../src/widgets/shared/documents/engines/markdownRenderer/plugins/remarkSafeTables.ts"],"sourcesContent":["import DOMPurify from 'dompurify';\n\nimport { hasMeaningfulTableText } from './tableHelpers';\nimport { MarkdownNode, MarkdownTransformer } from './markdownNode.interfaces';\n\nconst ALLOWED_TABLE_TAGS = [\n  'table', 'thead', 'tbody', 'tfoot', 'tr', 'th', 'td', 'caption', 'colgroup', 'col',\n  'p', 'br', 'strong', 'em', 'b', 'i', 'u', 'del', 's', 'code', 'pre', 'ul', 'ol', 'li',\n  'a', 'img', 'span', 'div', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6'\n];\nconst ALLOWED_TABLE_ATTRIBUTES = ['colspan', 'rowspan', 'href', 'src', 'alt', 'title', 'class'];\nconst ALLOWED_CLASS_NAMES = ['layoutTable'];\n\ntype HastNode =\n  | { type: 'text'; value: string }\n  | { type: 'element'; tagName: string; properties: Record<string, unknown>; children: HastNode[] };\n\nconst toHastProperties = (element: Element): Record<string, unknown> => {\n  const properties: Record<string, unknown> = {};\n  Array.from(element.attributes).forEach(({ name, value }) => {\n    if (name === 'class') {\n      const className = value.split(/\\s+/).filter((item) => ALLOWED_CLASS_NAMES.includes(item));\n      if (className.length) {\n        properties.className = className;\n      }\n    } else if (name === 'colspan' || name === 'rowspan') {\n      const span = Number(value);\n      if (Number.isInteger(span) && span > 0) {\n        properties[name === 'colspan' ? 'colSpan' : 'rowSpan'] = span;\n      }\n    } else {\n      properties[name] = value;\n    }\n  });\n  return properties;\n};\n\nconst toHast = (node: Node): HastNode | null => {\n  if (node.nodeType === Node.TEXT_NODE) {\n    return { type: 'text', value: node.textContent ?? '' };\n  }\n  if (node.nodeType !== Node.ELEMENT_NODE) {\n    return null;\n  }\n  const element = node as Element;\n  return {\n    type: 'element',\n    tagName: element.tagName.toLowerCase(),\n    properties: toHastProperties(element),\n    children: Array.from(element.childNodes).map(toHast).filter((child): child is HastNode => child !== null)\n  };\n};\n\nconst sanitizeTable = (html: string): Element | null => {\n  const fragment = DOMPurify.sanitize(html, {\n    ALLOWED_TAGS: ALLOWED_TABLE_TAGS,\n    ALLOWED_ATTR: ALLOWED_TABLE_ATTRIBUTES,\n    RETURN_DOM_FRAGMENT: true\n  });\n  const meaningfulNodes = Array.from(fragment.childNodes).filter((child) =>\n    child.nodeType !== Node.TEXT_NODE || hasMeaningfulTableText(child.textContent ?? '')\n  );\n  const [onlyNode] = meaningfulNodes;\n  return meaningfulNodes.length === 1 && onlyNode instanceof Element && onlyNode.tagName === 'TABLE' ? onlyNode : null;\n};\n\n// Converters fall back to raw `<table>` HTML for merged cells, which markdown cannot express.\n// Only a single, sanitised table is rendered; any other raw HTML stays inert text.\nexport const remarkSafeTables = (): MarkdownTransformer => {\n  const transform = (node: MarkdownNode): void => {\n    if (node.type === 'html' && /^<table(?:\\s|>)/i.test(node.value?.trim() ?? '')) {\n      const table = sanitizeTable(node.value ?? '');\n      const hast = table ? toHast(table) : null;\n      if (hast?.type === 'element') {\n        node.type = 'safeDocumentTable';\n        node.data = {\n          hName: 'table',\n          hProperties: hast.properties as Record<string, unknown> & { className?: string[] },\n          hChildren: hast.children\n        };\n        delete node.value;\n      }\n    }\n    node.children?.forEach(transform);\n  };\n  return transform;\n};\n"],"mappings":"AAAA,OAAOA,SAAS,MAAM,WAAW;AAEjC,SAASC,sBAAsB,QAAQ,gBAAgB;AAGvD,MAAMC,kBAAkB,GAAG,CACzB,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,UAAU,EAAE,KAAK,EAClF,GAAG,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,KAAK,EAAE,GAAG,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EACrF,GAAG,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAC9D;AACD,MAAMC,wBAAwB,GAAG,CAAC,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,KAAK,EAAE,KAAK,EAAE,OAAO,EAAE,OAAO,CAAC;AAC/F,MAAMC,mBAAmB,GAAG,CAAC,aAAa,CAAC;AAM3C,MAAMC,gBAAgB,GAAIC,OAAgB,IAA8B;EACtE,MAAMC,UAAmC,GAAG,CAAC,CAAC;EAC9CC,KAAK,CAACC,IAAI,CAACH,OAAO,CAACI,UAAU,CAAC,CAACC,OAAO,CAAC,CAAC;IAAEC,IAAI;IAAEC;EAAM,CAAC,KAAK;IAC1D,IAAID,IAAI,KAAK,OAAO,EAAE;MACpB,MAAME,SAAS,GAAGD,KAAK,CAACE,KAAK,CAAC,KAAK,CAAC,CAACC,MAAM,CAAEC,IAAI,IAAKb,mBAAmB,CAACc,QAAQ,CAACD,IAAI,CAAC,CAAC;MACzF,IAAIH,SAAS,CAACK,MAAM,EAAE;QACpBZ,UAAU,CAACO,SAAS,GAAGA,SAAS;MAClC;IACF,CAAC,MAAM,IAAIF,IAAI,KAAK,SAAS,IAAIA,IAAI,KAAK,SAAS,EAAE;MACnD,MAAMQ,IAAI,GAAGC,MAAM,CAACR,KAAK,CAAC;MAC1B,IAAIQ,MAAM,CAACC,SAAS,CAACF,IAAI,CAAC,IAAIA,IAAI,GAAG,CAAC,EAAE;QACtCb,UAAU,CAACK,IAAI,KAAK,SAAS,GAAG,SAAS,GAAG,SAAS,CAAC,GAAGQ,IAAI;MAC/D;IACF,CAAC,MAAM;MACLb,UAAU,CAACK,IAAI,CAAC,GAAGC,KAAK;IAC1B;EACF,CAAC,CAAC;EACF,OAAON,UAAU;AACnB,CAAC;AAED,MAAMgB,MAAM,GAAIC,IAAU,IAAsB;EAC9C,IAAIA,IAAI,CAACC,QAAQ,KAAKC,IAAI,CAACC,SAAS,EAAE;IACpC,OAAO;MAAEC,IAAI,EAAE,MAAM;MAAEf,KAAK,EAAEW,IAAI,CAACK,WAAW,IAAI;IAAG,CAAC;EACxD;EACA,IAAIL,IAAI,CAACC,QAAQ,KAAKC,IAAI,CAACI,YAAY,EAAE;IACvC,OAAO,IAAI;EACb;EACA,MAAMxB,OAAO,GAAGkB,IAAe;EAC/B,OAAO;IACLI,IAAI,EAAE,SAAS;IACfG,OAAO,EAAEzB,OAAO,CAACyB,OAAO,CAACC,WAAW,CAAC,CAAC;IACtCzB,UAAU,EAAEF,gBAAgB,CAACC,OAAO,CAAC;IACrC2B,QAAQ,EAAEzB,KAAK,CAACC,IAAI,CAACH,OAAO,CAAC4B,UAAU,CAAC,CAACC,GAAG,CAACZ,MAAM,CAAC,CAACP,MAAM,CAAEoB,KAAK,IAAwBA,KAAK,KAAK,IAAI;EAC1G,CAAC;AACH,CAAC;AAED,MAAMC,aAAa,GAAIC,IAAY,IAAqB;EACtD,MAAMC,QAAQ,GAAGvC,SAAS,CAACwC,QAAQ,CAACF,IAAI,EAAE;IACxCG,YAAY,EAAEvC,kBAAkB;IAChCwC,YAAY,EAAEvC,wBAAwB;IACtCwC,mBAAmB,EAAE;EACvB,CAAC,CAAC;EACF,MAAMC,eAAe,GAAGpC,KAAK,CAACC,IAAI,CAAC8B,QAAQ,CAACL,UAAU,CAAC,CAAClB,MAAM,CAAEoB,KAAK,IACnEA,KAAK,CAACX,QAAQ,KAAKC,IAAI,CAACC,SAAS,IAAI1B,sBAAsB,CAACmC,KAAK,CAACP,WAAW,IAAI,EAAE,CACrF,CAAC;EACD,MAAM,CAACgB,QAAQ,CAAC,GAAGD,eAAe;EAClC,OAAOA,eAAe,CAACzB,MAAM,KAAK,CAAC,IAAI0B,QAAQ,YAAYC,OAAO,IAAID,QAAQ,CAACd,OAAO,KAAK,OAAO,GAAGc,QAAQ,GAAG,IAAI;AACtH,CAAC;;AAED;AACA;AACA,OAAO,MAAME,gBAAgB,GAAGA,CAAA,KAA2B;EACzD,MAAMC,SAAS,GAAIxB,IAAkB,IAAW;IAC9C,IAAIA,IAAI,CAACI,IAAI,KAAK,MAAM,IAAI,kBAAkB,CAACqB,IAAI,CAACzB,IAAI,CAACX,KAAK,EAAEqC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE;MAC7E,MAAMC,KAAK,GAAGd,aAAa,CAACb,IAAI,CAACX,KAAK,IAAI,EAAE,CAAC;MAC7C,MAAMuC,IAAI,GAAGD,KAAK,GAAG5B,MAAM,CAAC4B,KAAK,CAAC,GAAG,IAAI;MACzC,IAAIC,IAAI,EAAExB,IAAI,KAAK,SAAS,EAAE;QAC5BJ,IAAI,CAACI,IAAI,GAAG,mBAAmB;QAC/BJ,IAAI,CAAC6B,IAAI,GAAG;UACVC,KAAK,EAAE,OAAO;UACdC,WAAW,EAAEH,IAAI,CAAC7C,UAAgE;UAClFiD,SAAS,EAAEJ,IAAI,CAACnB;QAClB,CAAC;QACD,OAAOT,IAAI,CAACX,KAAK;MACnB;IACF;IACAW,IAAI,CAACS,QAAQ,EAAEtB,OAAO,CAACqC,SAAS,CAAC;EACnC,CAAC;EACD,OAAOA,SAAS;AAClB,CAAC","ignoreList":[]}

package/dist/esm/widgets/shared/documents/engines/markdownRenderer/plugins/tableHelpers.js

@@ -0,0 +1,30 @@
+// Converters can emit this marker before a table they know is used for layout rather than data.
+export const LAYOUT_TABLE_MARKER = '<!-- layoutTable -->';
+
+// Zero-width spaces and byte-order marks are common in converted cells and carry no content.
+const INVISIBLE_CHARACTERS = [0x200b, 0xfeff].map(code => String.fromCharCode(code));
+export const hasMeaningfulTableText = text => Array.from(text).some(character => !/\s/.test(character) && !INVISIBLE_CHARACTERS.includes(character));
+const isProse = text => text.trim().length > 40 && text.trim().split(/\s+/).length >= 8;
+const isLabelAndBodyRow = row => {
+  const label = row[0]?.text.trim() ?? '';
+  const body = row[1]?.text.trim() ?? '';
+  return label.length > 0 && label.length <= 90 && label.split(/\s+/).length <= 12 && !/[.!?]$/.test(label) && isProse(body) && body.length > label.length * 2;
+};
+
+// Documents converted from Word often use one- or two-column tables to lay out prose
+// (a short label beside a paragraph). Those read better styled as text than as a data grid.
+export const isLikelyLayoutTable = rows => {
+  const meaningfulRows = rows.filter(row => row.some(cell => cell.text.trim()));
+  if (!meaningfulRows.length || Math.max(...meaningfulRows.map(row => row.length)) > 2) {
+    return false;
+  }
+  if (meaningfulRows[0].every(cell => cell.isHeader && cell.text.trim())) {
+    return false;
+  }
+  const layoutRows = meaningfulRows.filter(row => {
+    const populated = row.filter(cell => cell.text.trim());
+    return populated.length === 1 ? isProse(populated[0].text) : isLabelAndBodyRow(row);
+  });
+  return layoutRows.length >= Math.max(1, Math.ceil(meaningfulRows.length * 0.6));
+};
+//# sourceMappingURL=tableHelpers.js.map

package/dist/esm/widgets/shared/documents/engines/markdownRenderer/plugins/tableHelpers.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"tableHelpers.js","names":["LAYOUT_TABLE_MARKER","INVISIBLE_CHARACTERS","map","code","String","fromCharCode","hasMeaningfulTableText","text","Array","from","some","character","test","includes","isProse","trim","length","split","isLabelAndBodyRow","row","label","body","isLikelyLayoutTable","rows","meaningfulRows","filter","cell","Math","max","every","isHeader","layoutRows","populated","ceil"],"sources":["../../../../../../../../src/widgets/shared/documents/engines/markdownRenderer/plugins/tableHelpers.ts"],"sourcesContent":["export type LayoutTableCell = {\n  text: string;\n  isHeader?: boolean;\n};\n\n// Converters can emit this marker before a table they know is used for layout rather than data.\nexport const LAYOUT_TABLE_MARKER = '<!-- layoutTable -->';\n\n// Zero-width spaces and byte-order marks are common in converted cells and carry no content.\nconst INVISIBLE_CHARACTERS = [0x200b, 0xfeff].map((code) => String.fromCharCode(code));\n\nexport const hasMeaningfulTableText = (text: string): boolean =>\n  Array.from(text).some((character) => !/\\s/.test(character) && !INVISIBLE_CHARACTERS.includes(character));\n\nconst isProse = (text: string): boolean => text.trim().length > 40 && text.trim().split(/\\s+/).length >= 8;\n\nconst isLabelAndBodyRow = (row: LayoutTableCell[]): boolean => {\n  const label = row[0]?.text.trim() ?? '';\n  const body = row[1]?.text.trim() ?? '';\n  return label.length > 0\n    && label.length <= 90\n    && label.split(/\\s+/).length <= 12\n    && !/[.!?]$/.test(label)\n    && isProse(body)\n    && body.length > label.length * 2;\n};\n\n// Documents converted from Word often use one- or two-column tables to lay out prose\n// (a short label beside a paragraph). Those read better styled as text than as a data grid.\nexport const isLikelyLayoutTable = (rows: LayoutTableCell[][]): boolean => {\n  const meaningfulRows = rows.filter((row) => row.some((cell) => cell.text.trim()));\n  if (!meaningfulRows.length || Math.max(...meaningfulRows.map((row) => row.length)) > 2) {\n    return false;\n  }\n  if (meaningfulRows[0].every((cell) => cell.isHeader && cell.text.trim())) {\n    return false;\n  }\n  const layoutRows = meaningfulRows.filter((row) => {\n    const populated = row.filter((cell) => cell.text.trim());\n    return populated.length === 1 ? isProse(populated[0].text) : isLabelAndBodyRow(row);\n  });\n  return layoutRows.length >= Math.max(1, Math.ceil(meaningfulRows.length * 0.6));\n};\n"],"mappings":"AAKA;AACA,OAAO,MAAMA,mBAAmB,GAAG,sBAAsB;;AAEzD;AACA,MAAMC,oBAAoB,GAAG,CAAC,MAAM,EAAE,MAAM,CAAC,CAACC,GAAG,CAAEC,IAAI,IAAKC,MAAM,CAACC,YAAY,CAACF,IAAI,CAAC,CAAC;AAEtF,OAAO,MAAMG,sBAAsB,GAAIC,IAAY,IACjDC,KAAK,CAACC,IAAI,CAACF,IAAI,CAAC,CAACG,IAAI,CAAEC,SAAS,IAAK,CAAC,IAAI,CAACC,IAAI,CAACD,SAAS,CAAC,IAAI,CAACV,oBAAoB,CAACY,QAAQ,CAACF,SAAS,CAAC,CAAC;AAE1G,MAAMG,OAAO,GAAIP,IAAY,IAAcA,IAAI,CAACQ,IAAI,CAAC,CAAC,CAACC,MAAM,GAAG,EAAE,IAAIT,IAAI,CAACQ,IAAI,CAAC,CAAC,CAACE,KAAK,CAAC,KAAK,CAAC,CAACD,MAAM,IAAI,CAAC;AAE1G,MAAME,iBAAiB,GAAIC,GAAsB,IAAc;EAC7D,MAAMC,KAAK,GAAGD,GAAG,CAAC,CAAC,CAAC,EAAEZ,IAAI,CAACQ,IAAI,CAAC,CAAC,IAAI,EAAE;EACvC,MAAMM,IAAI,GAAGF,GAAG,CAAC,CAAC,CAAC,EAAEZ,IAAI,CAACQ,IAAI,CAAC,CAAC,IAAI,EAAE;EACtC,OAAOK,KAAK,CAACJ,MAAM,GAAG,CAAC,IAClBI,KAAK,CAACJ,MAAM,IAAI,EAAE,IAClBI,KAAK,CAACH,KAAK,CAAC,KAAK,CAAC,CAACD,MAAM,IAAI,EAAE,IAC/B,CAAC,QAAQ,CAACJ,IAAI,CAACQ,KAAK,CAAC,IACrBN,OAAO,CAACO,IAAI,CAAC,IACbA,IAAI,CAACL,MAAM,GAAGI,KAAK,CAACJ,MAAM,GAAG,CAAC;AACrC,CAAC;;AAED;AACA;AACA,OAAO,MAAMM,mBAAmB,GAAIC,IAAyB,IAAc;EACzE,MAAMC,cAAc,GAAGD,IAAI,CAACE,MAAM,CAAEN,GAAG,IAAKA,GAAG,CAACT,IAAI,CAAEgB,IAAI,IAAKA,IAAI,CAACnB,IAAI,CAACQ,IAAI,CAAC,CAAC,CAAC,CAAC;EACjF,IAAI,CAACS,cAAc,CAACR,MAAM,IAAIW,IAAI,CAACC,GAAG,CAAC,GAAGJ,cAAc,CAACtB,GAAG,CAAEiB,GAAG,IAAKA,GAAG,CAACH,MAAM,CAAC,CAAC,GAAG,CAAC,EAAE;IACtF,OAAO,KAAK;EACd;EACA,IAAIQ,cAAc,CAAC,CAAC,CAAC,CAACK,KAAK,CAAEH,IAAI,IAAKA,IAAI,CAACI,QAAQ,IAAIJ,IAAI,CAACnB,IAAI,CAACQ,IAAI,CAAC,CAAC,CAAC,EAAE;IACxE,OAAO,KAAK;EACd;EACA,MAAMgB,UAAU,GAAGP,cAAc,CAACC,MAAM,CAAEN,GAAG,IAAK;IAChD,MAAMa,SAAS,GAAGb,GAAG,CAACM,MAAM,CAAEC,IAAI,IAAKA,IAAI,CAACnB,IAAI,CAACQ,IAAI,CAAC,CAAC,CAAC;IACxD,OAAOiB,SAAS,CAAChB,MAAM,KAAK,CAAC,GAAGF,OAAO,CAACkB,SAAS,CAAC,CAAC,CAAC,CAACzB,IAAI,CAAC,GAAGW,iBAAiB,CAACC,GAAG,CAAC;EACrF,CAAC,CAAC;EACF,OAAOY,UAAU,CAACf,MAAM,IAAIW,IAAI,CAACC,GAAG,CAAC,CAAC,EAAED,IAAI,CAACM,IAAI,CAACT,cAAc,CAACR,MAAM,GAAG,GAAG,CAAC,CAAC;AACjF,CAAC","ignoreList":[]}

package/dist/esm/widgets/shared/documents/engines/wordRenderer/WordRenderer.js

@@ -0,0 +1,41 @@
+import React, { useEffect, useState } from 'react';
+import { renderWordDocument } from './renderWordDocument';
+export const WordRenderer = ({
+  arrayBuffer,
+  title,
+  cssClass = '',
+  loadingIndicator = null,
+  onError
+}) => {
+  const [previewDocument, setPreviewDocument] = useState(null);
+  useEffect(() => {
+    let isCurrent = true;
+    setPreviewDocument(null);
+    renderWordDocument(arrayBuffer).then(html => {
+      if (isCurrent) {
+        setPreviewDocument(html);
+      }
+    }).catch(error => {
+      if (isCurrent) {
+        onError?.(error);
+      }
+    });
+    return () => {
+      isCurrent = false;
+    };
+  }, [arrayBuffer, onError]);
+  if (previewDocument === null) {
+    return /*#__PURE__*/React.createElement(React.Fragment, null, loadingIndicator);
+  }
+
+  // An empty sandbox blocks scripts, forms and same-origin access; the CSP inside the document
+  // additionally blocks any network requests from the converted content.
+  return /*#__PURE__*/React.createElement("iframe", {
+    title: title,
+    className: `wordRenderer ${cssClass}`.trim(),
+    sandbox: "",
+    srcDoc: previewDocument
+  });
+};
+export default WordRenderer;
+//# sourceMappingURL=WordRenderer.js.map

package/dist/esm/widgets/shared/documents/engines/wordRenderer/WordRenderer.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"WordRenderer.js","names":["React","useEffect","useState","renderWordDocument","WordRenderer","arrayBuffer","title","cssClass","loadingIndicator","onError","previewDocument","setPreviewDocument","isCurrent","then","html","catch","error","createElement","Fragment","className","trim","sandbox","srcDoc"],"sources":["../../../../../../../src/widgets/shared/documents/engines/wordRenderer/WordRenderer.tsx"],"sourcesContent":["import React, { FC, ReactNode, useEffect, useState } from 'react';\n\nimport { renderWordDocument } from './renderWordDocument';\n\nexport type WordRendererProps = {\n  arrayBuffer: ArrayBuffer;\n  title: string;\n  cssClass?: string;\n  // Shown while the document is being converted, after the engine itself has loaded.\n  loadingIndicator?: ReactNode;\n  /* eslint-disable-next-line no-unused-vars */\n  onError?: (error: unknown) => void;\n};\n\nexport const WordRenderer: FC<WordRendererProps> = (\n  {\n    arrayBuffer,\n    title,\n    cssClass = '',\n    loadingIndicator = null,\n    onError\n  }\n) => {\n  const [previewDocument, setPreviewDocument] = useState<string | null>(null);\n\n  useEffect(() => {\n    let isCurrent = true;\n    setPreviewDocument(null);\n    renderWordDocument(arrayBuffer)\n      .then((html) => {\n        if (isCurrent) {\n          setPreviewDocument(html);\n        }\n      })\n      .catch((error: unknown) => {\n        if (isCurrent) {\n          onError?.(error);\n        }\n      });\n    return () => {\n      isCurrent = false;\n    };\n  }, [arrayBuffer, onError]);\n\n  if (previewDocument === null) {\n    return <>{ loadingIndicator }</>;\n  }\n\n  // An empty sandbox blocks scripts, forms and same-origin access; the CSP inside the document\n  // additionally blocks any network requests from the converted content.\n  return <iframe title={ title } className={ `wordRenderer ${ cssClass }`.trim() } sandbox=\"\" srcDoc={ previewDocument } />;\n};\n\nexport default WordRenderer;\n"],"mappings":"AAAA,OAAOA,KAAK,IAAmBC,SAAS,EAAEC,QAAQ,QAAQ,OAAO;AAEjE,SAASC,kBAAkB,QAAQ,sBAAsB;AAYzD,OAAO,MAAMC,YAAmC,GAAGA,CACjD;EACEC,WAAW;EACXC,KAAK;EACLC,QAAQ,GAAG,EAAE;EACbC,gBAAgB,GAAG,IAAI;EACvBC;AACF,CAAC,KACE;EACH,MAAM,CAACC,eAAe,EAAEC,kBAAkB,CAAC,GAAGT,QAAQ,CAAgB,IAAI,CAAC;EAE3ED,SAAS,CAAC,MAAM;IACd,IAAIW,SAAS,GAAG,IAAI;IACpBD,kBAAkB,CAAC,IAAI,CAAC;IACxBR,kBAAkB,CAACE,WAAW,CAAC,CAC5BQ,IAAI,CAAEC,IAAI,IAAK;MACd,IAAIF,SAAS,EAAE;QACbD,kBAAkB,CAACG,IAAI,CAAC;MAC1B;IACF,CAAC,CAAC,CACDC,KAAK,CAAEC,KAAc,IAAK;MACzB,IAAIJ,SAAS,EAAE;QACbH,OAAO,GAAGO,KAAK,CAAC;MAClB;IACF,CAAC,CAAC;IACJ,OAAO,MAAM;MACXJ,SAAS,GAAG,KAAK;IACnB,CAAC;EACH,CAAC,EAAE,CAACP,WAAW,EAAEI,OAAO,CAAC,CAAC;EAE1B,IAAIC,eAAe,KAAK,IAAI,EAAE;IAC5B,oBAAOV,KAAA,CAAAiB,aAAA,CAAAjB,KAAA,CAAAkB,QAAA,QAAIV,gBAAoB,CAAC;EAClC;;EAEA;EACA;EACA,oBAAOR,KAAA,CAAAiB,aAAA;IAAQX,KAAK,EAAGA,KAAO;IAACa,SAAS,EAAG,gBAAiBZ,QAAQ,EAAG,CAACa,IAAI,CAAC,CAAG;IAACC,OAAO,EAAC,EAAE;IAACC,MAAM,EAAGZ;EAAiB,CAAE,CAAC;AAC3H,CAAC;AAED,eAAeN,YAAY","ignoreList":[]}

package/dist/esm/widgets/shared/documents/engines/wordRenderer/renderWordDocument.js

@@ -0,0 +1,207 @@
+import { renderAsync } from 'docx-preview';
+import DOMPurify from 'dompurify';
+import JSZip from 'jszip';
+import { prepareWordPreviewNumbering } from './wordPreviewNumbering';
+import { createWordNumberingResolver, getWordStyles, wordAttribute, wordChild, wordNamespace, wordValue } from './wordXml';
+
+// The preview runs in a sandboxed iframe that cannot read the host page's CSS variables.
+const PAGE_GUTTER = '15px';
+const CONTENT_SECURITY_POLICY = 'default-src \'none\'; img-src data:; font-src data:; style-src \'unsafe-inline\';';
+
+/*
+  Limits against "zip bombs": a small .docx can declare, or simply contain, gigabytes of compressed
+  content. The sizes a zip declares cannot be trusted, so every entry is decompressed through a
+  stream that stops once the running total passes the limit, into a new zip that docx-preview reads.
+  Its own decompression then only ever sees content whose size has been measured.
+ */
+export const WORD_DOCUMENT_LIMITS = {
+  maxFileBytes: 50 * 1024 * 1024,
+  maxUncompressedBytes: 200 * 1024 * 1024,
+  maxEntries: 5000
+};
+
+// Documented by JSZip but missing from its typings.
+
+const TOO_LARGE = 'This Word document is too large to preview.';
+const readZipEntry = (file, budget) => new Promise((resolve, reject) => {
+  const chunks = [];
+  let length = 0;
+  const stream = file.internalStream('uint8array');
+  stream.on('data', chunk => {
+    budget.remaining -= chunk.length;
+    if (budget.remaining < 0) {
+      stream.pause();
+      reject(new Error(TOO_LARGE));
+      return;
+    }
+    chunks.push(chunk);
+    length += chunk.length;
+  }).on('error', reject).on('end', () => {
+    const content = new Uint8Array(length);
+    let offset = 0;
+    chunks.forEach(chunk => {
+      content.set(chunk, offset);
+      offset += chunk.length;
+    });
+    resolve(content);
+  }).resume();
+});
+const loadBoundedZip = async arrayBuffer => {
+  if (arrayBuffer.byteLength > WORD_DOCUMENT_LIMITS.maxFileBytes) {
+    throw new Error(TOO_LARGE);
+  }
+  const zip = await JSZip.loadAsync(arrayBuffer);
+  const entries = Object.values(zip.files).filter(file => !file.dir);
+  if (entries.length > WORD_DOCUMENT_LIMITS.maxEntries) {
+    throw new Error(TOO_LARGE);
+  }
+  const bounded = new JSZip();
+  const budget = {
+    remaining: WORD_DOCUMENT_LIMITS.maxUncompressedBytes
+  };
+  // Sequential, so the budget is spent in order and the first entry over it stops the rest.
+  for (const file of entries) {
+    bounded.file(file.name, await readZipEntry(file, budget));
+  }
+  return bounded;
+};
+const parseWordPart = async (zip, name) => {
+  const file = zip.file(`word/${name}.xml`);
+  if (!file) {
+    return undefined;
+  }
+  const xml = new DOMParser().parseFromString(await file.async('string'), 'application/xml');
+  if (xml.querySelector('parsererror')) {
+    throw new Error(`Invalid Word XML: ${name}`);
+  }
+  return xml;
+};
+const hasLatinFont = fonts => Boolean(wordAttribute(fonts, 'ascii') || wordAttribute(fonts, 'asciiTheme'));
+
+// docx-preview replaces the whole CSS font-family when a style only names an East Asian font,
+// so copy the Latin font each style would otherwise inherit.
+const inheritLatinFonts = (styleDefinitions, styles) => {
+  const defaultFonts = styleDefinitions?.getElementsByTagNameNS(wordNamespace, 'rPrDefault')[0]?.getElementsByTagNameNS(wordNamespace, 'rFonts')[0];
+  const resolveLatinFont = (styleId, visited = new Set()) => {
+    if (!styleId) {
+      return defaultFonts;
+    }
+    if (visited.has(styleId)) {
+      throw new Error(`Circular Word style inheritance: ${styleId}`);
+    }
+    visited.add(styleId);
+    const style = styles.byId.get(styleId);
+    const fonts = wordChild(wordChild(style, 'rPr'), 'rFonts');
+    return hasLatinFont(fonts) ? fonts : resolveLatinFont(wordValue(style, 'basedOn'), visited);
+  };
+  styles.byId.forEach((style, id) => {
+    const fonts = wordChild(wordChild(style, 'rPr'), 'rFonts');
+    if (!fonts || hasLatinFont(fonts)) {
+      return;
+    }
+    const inheritedFonts = resolveLatinFont(id);
+    ['ascii', 'asciiTheme'].forEach(name => {
+      const font = wordAttribute(inheritedFonts, name);
+      if (font) {
+        fonts.setAttributeNS(wordNamespace, `w:${name}`, font);
+      }
+    });
+  });
+};
+const ensureChild = (source, parent, name, prepend = false) => {
+  let child = wordChild(parent, name);
+  if (!child) {
+    child = source.createElementNS(wordNamespace, `w:${name}`);
+    if (prepend) {
+      parent.prepend(child);
+    } else {
+      parent.append(child);
+    }
+  }
+  return child;
+};
+const isPageBreak = element => element.localName === 'lastRenderedPageBreak' || element.localName === 'br' && wordAttribute(element, 'type') === 'page';
+
+// A leading inline page break otherwise produces an empty numbered fragment before the paragraph.
+const convertLeadingPageBreaks = (source, paragraph) => {
+  const runContent = Array.from(paragraph.getElementsByTagNameNS(wordNamespace, 'r')).flatMap(run => Array.from(run.children).filter(child => child.localName !== 'rPr'));
+  for (const child of runContent) {
+    if (!isPageBreak(child)) {
+      break;
+    }
+    const properties = ensureChild(source, paragraph, 'pPr', true);
+    ensureChild(source, properties, 'pageBreakBefore').setAttributeNS(wordNamespace, 'w:val', '1');
+    child.remove();
+  }
+};
+const applyResolvedNumbering = (source, paragraph, resolveNumbering) => {
+  const {
+    id,
+    level
+  } = resolveNumbering(wordChild(paragraph, 'pPr'));
+  if (!id || id === '0') {
+    return;
+  }
+  const numbering = ensureChild(source, ensureChild(source, paragraph, 'pPr', true), 'numPr');
+  ensureChild(source, numbering, 'numId').setAttributeNS(wordNamespace, 'w:val', id);
+  ensureChild(source, numbering, 'ilvl').setAttributeNS(wordNamespace, 'w:val', level);
+};
+const wrapPreviewDocument = html => `<!doctype html><html><head><meta charset="utf-8">
+<meta http-equiv="Content-Security-Policy" content="${CONTENT_SECURITY_POLICY}">
+</head><body>${html}
+<style>body { margin: 0; } .docx-wrapper { padding: ${PAGE_GUTTER}; align-items: flex-start; }
+.docx-wrapper > section.docx { flex-shrink: 0; margin: 0 auto ${PAGE_GUTTER}; }
+.docx-wrapper p { text-indent: 0 !important; margin-inline-start: 0 !important; }</style>
+</body></html>`;
+
+/*
+  Renders a .docx file to a self-contained, sanitised HTML document for a sandboxed iframe.
+  The Word XML is patched first to fix fonts, page breaks and list numbering that docx-preview
+  would otherwise render incorrectly.
+ */
+export const renderWordDocument = async arrayBuffer => {
+  const zip = await loadBoundedZip(arrayBuffer);
+  const source = await parseWordPart(zip, 'document');
+  if (!source) {
+    throw new Error('DOCX is missing word/document.xml.');
+  }
+  const styleDefinitions = await parseWordPart(zip, 'styles');
+  const styles = getWordStyles(styleDefinitions);
+  inheritLatinFonts(styleDefinitions, styles);
+  if (styleDefinitions) {
+    zip.file('word/styles.xml', new XMLSerializer().serializeToString(styleDefinitions));
+  }
+  const numberingDefinitions = await parseWordPart(zip, 'numbering');
+  const resolveNumbering = createWordNumberingResolver(numberingDefinitions, styles);
+  Array.from(source.getElementsByTagNameNS(wordNamespace, 'p')).forEach(paragraph => {
+    convertLeadingPageBreaks(source, paragraph);
+    applyResolvedNumbering(source, paragraph, resolveNumbering);
+  });
+  const numberingCss = prepareWordPreviewNumbering(numberingDefinitions, styles);
+  if (numberingDefinitions) {
+    zip.file('word/numbering.xml', new XMLSerializer().serializeToString(numberingDefinitions));
+  }
+  zip.file('word/document.xml', new XMLSerializer().serializeToString(source));
+  const container = document.createElement('div');
+  await renderAsync(await zip.generateAsync({
+    type: 'arraybuffer'
+  }), container, undefined, {
+    useBase64URL: true,
+    ignoreLastRenderedPageBreak: false,
+    renderAltChunks: false
+  });
+  const numberingStyle = document.createElement('style');
+  numberingStyle.textContent = numberingCss;
+  container.append(numberingStyle);
+  container.querySelectorAll('section.docx').forEach(page => {
+    if (page.style.width) {
+      // Scale fixed-width pages down to fit narrow viewers instead of scrolling sideways.
+      page.style.setProperty('zoom', `min(1, calc((100vw - 2 * ${PAGE_GUTTER}) / ${page.style.width}))`);
+    }
+  });
+  return wrapPreviewDocument(DOMPurify.sanitize(container.innerHTML, {
+    ADD_TAGS: ['style'],
+    FORCE_BODY: true
+  }));
+};
+//# sourceMappingURL=renderWordDocument.js.map