@json-to-office/jto-ops 5.0.0 → 6.0.0

package/dist/index.js

@@ -2208,40 +2208,61 @@ function readingOrder(words2) {
   }
   return order;
 }
-function uprightOrder(words2, indices) {
+function median(values) {
+  const sorted = [...values].sort((a, b) => a - b);
+  const mid = sorted.length >> 1;
+  return sorted.length % 2 === 1 ? sorted[mid] : (sorted[mid - 1] + sorted[mid]) / 2;
+}
+function groupRows(words2, indices) {
   const rows = [];
   const byY = [...indices].sort(
     (a, b) => words2[a].yMin - words2[b].yMin || words2[a].xMin - words2[b].xMin
   );
   for (const i of byY) {
+    const word = words2[i];
     const row = rows[rows.length - 1];
-    if (row && sameRow(words2[row[0]], words2[i])) row.push(i);
-    else rows.push([i]);
+    if (row && sameRow(words2[row.words[0]], word)) {
+      row.words.push(i);
+      row.yMin = Math.min(row.yMin, word.yMin);
+      row.yMax = Math.max(row.yMax, word.yMax);
+    } else rows.push({ words: [i], yMin: word.yMin, yMax: word.yMax });
   }
+  return rows;
+}
+function uprightOrder(words2, indices) {
+  const grouped = groupRows(words2, indices);
+  const rows = grouped.map((row) => row.words);
   const fragmentsOf = (row) => {
     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;
+      const gap = words2[i].xMin - words2[sorted[n]].xMax;
+      return { gap, height, wide: gap > height / 2 };
+    });
+    const spaces = gaps.filter((g) => !g.wide).map((g) => g.gap);
+    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;
+    };
     const out = [];
-    for (const i of sorted) {
+    sorted.forEach((i, n) => {
       const last = out[out.length - 1];
       const w = words2[i];
-      const height = w.yMax - w.yMin;
-      if (last && w.xMin - last.xMax <= height / 2) {
+      if (last && !isCellBoundary(n)) {
         last.words.push(i);
         last.xMax = Math.max(last.xMax, w.xMax);
       } else out.push({ xMin: w.xMin, xMax: w.xMax, words: [i] });
-    }
+    });
     return out;
   };
   const fragments = rows.map(fragmentsOf);
-  const boxes = rows.map((row) => ({
-    yMin: Math.min(...row.map((i) => words2[i].yMin)),
-    yMax: Math.max(...row.map((i) => words2[i].yMax))
-  }));
+  const boxes = grouped;
   const order = [];
   let cluster = null;
   const flush = () => {
     if (!cluster) return;
-    if (!cluster.multi) {
+    if (cluster.columns.length < 2) {
       for (const r of cluster.rows)
         for (const f of fragments[r]) order.push(...f.words);
     } else {
@@ -2263,9 +2284,8 @@ function uprightOrder(words2, indices) {
     const adjacent = cluster !== null && boxes[r].yMin - boxes[r - 1].yMax <= lineHeight * 0.9;
     const wide = cluster !== null && fragments[r].some((f) => overlapping(cluster.columns, f).length > 1);
     if (!adjacent || wide) flush();
-    if (!cluster) cluster = { rows: [], columns: [], multi: false };
+    if (!cluster) cluster = { rows: [], columns: [] };
     cluster.rows.push(r);
-    if (fragments[r].length > 1) cluster.multi = true;
     for (const f of fragments[r]) {
       const hit = overlapping(cluster.columns, f)[0];
       if (hit) {
@@ -2414,6 +2434,34 @@ function chromeOccurrences(occurrences, pages) {
     )
   );
 }
+function pageNumberRows(pages, claimed) {
+  const hits = [];
+  pages.forEach((page, pageIndex) => {
+    const rows = groupRows(
+      page.words,
+      page.words.map((_, i) => i)
+    );
+    const free = rows.map(
+      (row) => row.words.some((i) => !claimed.has(`${pageIndex}:${i}`))
+    );
+    rows.forEach((row, r) => {
+      if (!free[r] || !inChromeBand(page, row)) return;
+      if (!row.words.every((i) => PAGE_NUMBER_ROW.test(page.words[i].text)))
+        return;
+      const line = row.yMax - row.yMin;
+      const attached = rows.some(
+        (other, o) => o !== r && free[o] && Math.max(other.yMin - row.yMax, row.yMin - other.yMax) < line
+      );
+      if (!attached) hits.push({ pageIndex, box: row });
+    });
+  });
+  if (pages.length < 2) return hits;
+  return hits.filter(
+    (hit) => hits.some(
+      (other) => other.pageIndex !== hit.pageIndex && Math.abs(other.box.yMin - hit.box.yMin) <= CHROME_ROW_TOLERANCE_PT
+    )
+  );
+}
 function assignInventory(pages, inventory) {
   const results = /* @__PURE__ */ new Map();
   const chromeWords = /* @__PURE__ */ new Set();
@@ -2450,16 +2498,8 @@ function assignInventory(pages, inventory) {
       });
     }
   }
-  pages.forEach((page, pageIndex) => {
-    page.words.forEach((word, i) => {
-      if (chromeWords.has(`${pageIndex}:${i}`)) return;
-      if (!inChromeBand(page, word)) return;
-      const row = page.words.filter((w) => sameRow(w, word));
-      if (row.every((w) => PAGE_NUMBER_ROW.test(w.text))) {
-        claimRow(pageIndex, word);
-      }
-    });
-  });
+  for (const row of pageNumberRows(pages, chromeWords))
+    claimRow(row.pageIndex, row.box);
   const body = indexDocument(pages, chromeWords);
   const claimed = /* @__PURE__ */ new Map();
   let cursor = 0;
@@ -2855,7 +2895,8 @@ function draftRenderedFindings(input) {
           }
         }
       }
-      const section = owner ? /^\/children\/\d+/.exec(owner.path)?.[0] : void 0;
+      const owned = owner ? /^\/children\/\d+/.exec(owner.path)?.[0] : void 0;
+      const section = owned && last ? closingSection(input.inventory) ?? owned : owned;
       findings.push(
         finding({
           ruleId: "rendered/page-underfilled",
@@ -2863,7 +2904,7 @@ function draftRenderedFindings(input) {
           page: pageIndex + 1,
           path: section ?? "",
           message: last ? `Page ${pageIndex + 1}, the last, is ${Math.round(fill * 100)}% filled: only the tail of the document reached it.` : `Page ${pageIndex + 1} is ${Math.round(fill * 100)}% filled: its content stops ${Math.round(footerTop - bottom)}pt above the footer and the next page begins anyway.`,
-          suggestion: last ? "Keep the notes and sources with the paragraph before them, or tighten the page before so the document closes on a designed page rather than a stub." : "Merge this section into its neighbour, or give the page the table or chart its argument owes. Do not pad it with prose: a short section on its own page reads as unfinished, a padded one reads as filler.",
+          suggestion: last ? "Set `pageBreak: true` on this section \u2014 or on the one before it, when this section already starts a page \u2014 so the closing argument and its notes share one designed page instead of the last page taking only their tail, then re-preview. Keeping the notes with the paragraph above them moves the tail, not the page." : "Merge this section into its neighbour, or give the page the table or chart its argument owes. Do not pad it with prose: a short section on its own page reads as unfinished, a padded one reads as filler.",
           evidence: {
             summary: "Share of the body area the page paints, header to footer",
             expected: `\u2265 ${Math.round((last ? MINIMUM_LAST_PAGE_FILL : MINIMUM_PAGE_FILL) * 100)}%`,
@@ -2964,6 +3005,17 @@ function renderedEngine() {
   engine ??= new QualityEngine(RENDERED_QUALITY_RULES.rules);
   return engine;
 }
+function closingSection(inventory) {
+  let highest;
+  for (const entry of inventory) {
+    if (entry.repeats) continue;
+    const index = /^\/children\/(\d+)/.exec(entry.path)?.[1];
+    if (index === void 0) continue;
+    const value = Number(index);
+    if (highest === void 0 || value > highest) highest = value;
+  }
+  return highest === void 0 ? void 0 : `/children/${highest}`;
+}
 function analyzeRenderedDocument(input, options = {}) {
   const fact = {
     id: "rendered:geometry",