@json-to-office/jto-ops 7.0.1 → 7.2.0

package/dist/index.d.ts

@@ -475,7 +475,7 @@ declare function extractPdfFonts(pdfPath: string, options?: {
  * pass over it, memoised per fact, so eight rules cost one search.
  */
 
-type RenderedRuleId = 'rendered/clip' | 'rendered/spill' | 'rendered/overlap' | 'rendered/text-missing' | 'rendered/font-substituted' | 'rendered/empty-page' | 'rendered/heading-stranded' | 'rendered/paragraph-split' | 'rendered/table-split' | 'rendered/page-underfilled';
+type RenderedRuleId = 'rendered/clip' | 'rendered/spill' | 'rendered/overlap' | 'rendered/text-missing' | 'rendered/placeholder' | 'rendered/font-substituted' | 'rendered/empty-page' | 'rendered/heading-stranded' | 'rendered/paragraph-split' | 'rendered/table-split' | 'rendered/page-underfilled';
 declare const RENDERED_QUALITY_RULES: QualityRulePack;
 
 /**
@@ -544,8 +544,11 @@ interface TextOccurrence {
  * row line by line across every cell, poppler 26 lists it cell by cell, and
  * an authored string in a wrapped cell only survives the second.
  *
- * Rows are grouped by vertical overlap and split into fragments at gaps
- * wider than half the line is tall. A row of two or more fragments opens a
+ * Rows are grouped by vertical overlap and split into fragments, a cell's
+ * line each: at gaps wider than half the line is tall, at gaps wider than
+ * the row's own spaces account for, and before a word that starts on the
+ * edge of a column the rows around it mark, where two cells ran together
+ * at a gap no wider than a space. A row of two or more fragments opens a
  * column block; the rows under it belong to the block while each of their
  * fragments sits inside one column (a right-aligned cell's second line
  * starts further right; a wrapped label's later lines are the only
@@ -644,7 +647,8 @@ declare function assignInventory<T extends InventoryEntry>(pages: readonly PdfTe
  * the preview PDF, the fonts the PDF embeds, and the document's authored
  * text inventory, and reports the defects only ink can show: text past the
  * page or its frame, words drawn over each other, text that never rendered
- * (fully clipped, or set in a place poppler cannot read), faces substituted
+ * (fully clipped, or set in a place poppler cannot read), a placeholder the
+ * renderer painted where it could not draw a component, faces substituted
  * on the way, pages with nothing on them, headings stranded at a page foot
  * and paragraphs split into a lone line.
  *
@@ -673,6 +677,12 @@ interface RenderedTextEntry extends InventoryEntry {
         widthPt?: number;
         heightPt?: number;
     };
+    /**
+     * Set when the entry is the placeholder a renderer paints for a component
+     * it could not draw there: the component's name. Authored text that quotes
+     * the placeholder's wording carries none.
+     */
+    unrendered?: string;
 }
 interface RequestedFont {
     family: string;

package/dist/index.js

@@ -2113,7 +2113,9 @@ async function extractPdfFonts(pdfPath, options = {}) {
 
 // src/rendered-analysis.ts
 import {
-  QualityEngine
+  QualityEngine,
+  unrenderedComponentName,
+  unrenderedComponentText
 } from "@json-to-office/quality";
 
 // src/rendered-rules.ts
@@ -2167,6 +2169,14 @@ var RENDERED_QUALITY_RULES = {
       category: "integrity",
       defaultSeverity: "warning"
     }),
+    renderedRule({
+      id: "rendered/placeholder",
+      code: QUALITY_CODES.RENDERED_PLACEHOLDER,
+      description: `DOCX: the renderer painted "[Unsupported component type: \u2026]" where it could not draw a component \u2014 a table cell holding one it cannot render. Mapped to the cell when the document puts it there; reported on its page, unmapped, when it came from somewhere the document does not show, such as a plugin's output.`,
+      category: "integrity",
+      defaultSeverity: "warning",
+      formats: ["docx"]
+    }),
     renderedRule({
       id: "rendered/font-substituted",
       code: QUALITY_CODES.RENDERED_FONT_SUBSTITUTED,
@@ -2283,10 +2293,26 @@ function linedOrder(words2, lines = []) {
   };
 }
 function rotatedWords(words2, lines) {
+  const upright = words2.filter(
+    (w) => w.text.trim() !== "" && w.xMax - w.xMin >= w.yMax - w.yMin
+  );
+  const standsInALine = (w) => {
+    const height = w.yMax - w.yMin;
+    return upright.some((other) => {
+      if (other === w) return false;
+      const overlap = Math.min(w.yMax, other.yMax) - Math.max(w.yMin, other.yMin);
+      const shorter = Math.min(height, other.yMax - other.yMin);
+      if (shorter <= 0 || overlap <= shorter / 2) return false;
+      const ratio = height / (other.yMax - other.yMin);
+      return ratio > 0.8 && ratio < 1.25;
+    });
+  };
   const rotated = new Set(
     words2.map((_, i) => i).filter((i) => {
       const w = words2[i];
-      return w.text.length > 2 && w.yMax - w.yMin > 1.5 * (w.xMax - w.xMin);
+      if (w.text.length <= 2) return false;
+      if (w.yMax - w.yMin <= 1.5 * (w.xMax - w.xMin)) return false;
+      return !standsInALine(w);
     })
   );
   for (const line of lines) {
@@ -2330,10 +2356,12 @@ function groupRows(words2, indices) {
   }
   return rows;
 }
+var COLUMN_EDGE_TOLERANCE_PT = 0.05;
+var WORD_SPACE_TOLERANCE_PT = 0.1;
 function uprightOrder(words2, indices) {
   const grouped = groupRows(words2, indices);
   const rows = grouped.map((row) => row.words);
-  const fragmentsOf = (row) => {
+  const fragmentsOf = (row, onEdge) => {
     const sorted = [...row].sort((a, b) => words2[a].xMin - words2[b].xMin);
     const gaps = sorted.slice(1).map((i, n) => {
       const height = words2[i].yMax - words2[i].yMin;
@@ -2344,7 +2372,7 @@ function uprightOrder(words2, indices) {
     const typical = spaces.length > 1 ? median(spaces) : void 0;
     const isCellBoundary = (n) => {
       const g = gaps[n - 1];
-      return g.wide || typical !== void 0 && g.gap > g.height / 3 && g.gap >= typical * 1.5;
+      return g.wide || typical !== void 0 && g.gap > g.height / 3 && g.gap >= typical * 1.5 || onEdge !== void 0 && typical !== void 0 && spaces.length >= 3 && Math.abs(g.gap - typical) > WORD_SPACE_TOLERANCE_PT && onEdge(words2[sorted[n - 1]], words2[sorted[n]]);
     };
     const out = [];
     sorted.forEach((i, n) => {
@@ -2357,7 +2385,26 @@ function uprightOrder(words2, indices) {
     });
     return out;
   };
-  const fragments = rows.map(fragmentsOf);
+  const byGaps = rows.map((row) => fragmentsOf(row));
+  const onColumnEdge = (r, before, word) => {
+    const x = word.xMin;
+    for (const step of [-1, 1]) {
+      for (let t = r + step; t >= 0 && t < rows.length; t += step) {
+        const edge = byGaps[t].slice(1).some(
+          (f) => Math.abs(f.xMin - x) <= COLUMN_EDGE_TOLERANCE_PT && before.xMax <= f.xMin
+        );
+        if (edge) return true;
+        const across = rows[t].some(
+          (i) => words2[i].xMin < x - COLUMN_EDGE_TOLERANCE_PT && words2[i].xMax > x + COLUMN_EDGE_TOLERANCE_PT
+        );
+        if (across) break;
+      }
+    }
+    return false;
+  };
+  const fragments = rows.map(
+    (row, r) => fragmentsOf(row, (before, word) => onColumnEdge(r, before, word))
+  );
   const boxes = grouped;
   const order = [];
   const read = /* @__PURE__ */ new Set();
@@ -3160,6 +3207,53 @@ function draftRenderedFindings(input) {
       })
     );
   }
+  const placeholderFinding = (name, page, path9, mapping) => {
+    const text = name ? unrenderedComponentText(name) : "[Unsupported component type: \u2026]";
+    return finding({
+      ruleId: "rendered/placeholder",
+      mapping,
+      page,
+      path: path9,
+      message: `Page ${page} shows "${text}": the renderer could not draw ${name ? `a ${name}` : "a component"} there and painted this instead.`,
+      suggestion: "Put only what the position renders there \u2014 a table cell takes a string, a paragraph, an image, a visual or a highcharts chart \u2014 or move the component out of the table; statistics side by side are a columns component.",
+      evidence: {
+        summary: "Renderer placeholder text in the PDF",
+        actual: text
+      },
+      context: {
+        kind: "unrendered-component",
+        ...name && { component: name }
+      }
+    });
+  };
+  for (const match of matches) {
+    const name = match.entry.unrendered;
+    if (match.status !== "mapped" || name === void 0) continue;
+    for (const o of match.occurrences) {
+      findings.push(
+        placeholderFinding(name, o.pageIndex + 1, match.entry.path, "mapped")
+      );
+    }
+  }
+  pages.forEach((page, pageIndex) => {
+    page.words.forEach((word, index) => {
+      if (!word.text.startsWith("[Unsupported")) return;
+      if (ownerOf(pageIndex, index)) return;
+      const phrase = [];
+      for (const next of page.words.slice(index, index + 6)) {
+        phrase.push(next.text);
+        if (next.text.endsWith("]")) break;
+      }
+      findings.push(
+        placeholderFinding(
+          unrenderedComponentName(phrase.join(" ")),
+          pageIndex + 1,
+          "",
+          "unmapped"
+        )
+      );
+    });
+  });
   let substituted = 0;
   if (input.fonts && input.fonts.length > 0 && input.requestedFonts) {
     const seen = /* @__PURE__ */ new Set();
@@ -3500,6 +3594,9 @@ function renderedInventoryFromFacts(format, facts) {
         ...typeof fact.level === "number" && { level: fact.level },
         ...fact.repeats === true && { repeats: true },
         ...fact.optional === true && { optional: true },
+        ...typeof fact.unrendered === "string" && {
+          unrendered: fact.unrendered
+        },
         ...frame && {
           box: {
             ...frame.widthPt !== void 0 && { widthPt: frame.widthPt },