@uurtech/jdf-cli 0.2.2 → 0.2.3

package/dist/index.js

@@ -712,7 +712,7 @@ function columnBands(rows) {
   return bands.map(({ x0, x1 }) => ({ x0, x1 }));
 }
 var numeric = (s) => /^[\s$€£¥+\-−–]*[\d.,]+\s*(%|ms|s|k|m|b|M|K|B|x|×)?\s*(\/\w+)?$/i.test(s.trim()) || /^[+\-−]?\d/.test(s.trim()) && /\d$/.test(s.trim().replace(/[%)]$/, ""));
-function detectTables(runs, shapes, pageWidthMm) {
+function detectTables(runs, shapes, pageWidthMm, gutters = []) {
   const out = [];
   const used = /* @__PURE__ */ new Set();
   const rows = groupRows(runs, () => false);
@@ -761,6 +761,14 @@ function detectTables(runs, shapes, pageWidthMm) {
       i++;
       continue;
     }
+    if (gutters.length && !hasLattice) {
+      const b = best.bands;
+      const straddles = b.some((band, k) => k < b.length - 1 && gutters.some((g) => band.x1 <= g.x1 + 1 && b[k + 1].x0 >= g.x0 - 1) && band.x1 - band.x0 > 30 && b[k + 1].x1 - b[k + 1].x0 > 30);
+      if (straddles) {
+        i++;
+        continue;
+      }
+    }
     const bands = best.bands;
     const cellText = (row, b) => row.cells.filter((c) => c.x0 < bands[b].x1 - 0.2 && c.x1 > bands[b].x0 + 0.2).map((c) => c.run.text.trim()).join(" ").trim();
     const grid = [];
@@ -848,8 +856,165 @@ function detectTables(runs, shapes, pageWidthMm) {
   return out;
 }
 
-// ../../packages/jdf-pdf-import/src/core.ts
+// ../../packages/jdf-pdf-import/src/columns.ts
+function detectGutters(lines, bodyFontSize, pageWmm) {
+  const body = lines.filter((l) => l.text.trim().length > 0 && Math.abs(l.fontSize - bodyFontSize) <= 1.5 && l.width < pageWmm * 0.6 && l.width > 0);
+  if (body.length < 12) return [];
+  const minX = Math.min(...body.map((l) => l.x));
+  const maxX = Math.max(...body.map((l) => l.x + l.width));
+  const span = maxX - minX;
+  if (span < pageWmm * 0.4) return [];
+  const BIN = 0.5;
+  const n = Math.ceil(span / BIN) + 1;
+  const cov = new Array(n).fill(0);
+  for (const l of body) {
+    const a = Math.max(0, Math.floor((l.x - minX) / BIN)), b = Math.min(n, Math.ceil((l.x + l.width - minX) / BIN));
+    for (let i2 = a; i2 < b; i2++) cov[i2]++;
+  }
+  const noise = Math.max(2, Math.round(body.length * 0.05));
+  const gutters = [];
+  let i = 0;
+  while (i < n) {
+    if (cov[i] > noise) {
+      i++;
+      continue;
+    }
+    let j = i;
+    while (j < n && cov[j] <= noise) j++;
+    const g0 = minX + i * BIN, g1 = minX + j * BIN;
+    if (g1 - g0 >= 3 && g0 > minX + span * 0.15 && g1 < maxX - span * 0.15) gutters.push({ x0: g0, x1: g1 });
+    i = j;
+  }
+  const solid = gutters.every((g) => body.filter((l) => l.x + l.width <= g.x0 + 0.5).length >= 4 && body.filter((l) => l.x >= g.x1 - 0.5).length >= 4);
+  return solid ? gutters : [];
+}
+function orderByColumns(els, gutters, pageWmm) {
+  if (!gutters.length || els.length < 2) return els;
+  const bands = [];
+  let left = 0;
+  for (const g of gutters) {
+    bands.push({ x0: left, x1: g.x0 });
+    left = g.x1;
+  }
+  bands.push({ x0: left, x1: pageWmm });
+  const colOf = (e) => {
+    const x = e.position?.x ?? 0;
+    const k = bands.findIndex((b) => x >= b.x0 && x < b.x1);
+    if (k >= 0) return k;
+    let best = 0, d = Infinity;
+    bands.forEach((b, i) => {
+      const dd = Math.min(Math.abs(x - b.x0), Math.abs(x - b.x1));
+      if (dd < d) {
+        d = dd;
+        best = i;
+      }
+    });
+    return best;
+  };
+  const spans = (e) => {
+    const c = colOf(e);
+    if (c >= bands.length - 1) return false;
+    const right = (e.position?.x ?? 0) + (e.width ?? 0);
+    const next = bands[c + 1];
+    return right > (next.x0 + next.x1) / 2;
+  };
+  const sorted = els.slice().sort((a, b) => (a.position?.y ?? 0) - (b.position?.y ?? 0) || (a.position?.x ?? 0) - (b.position?.x ?? 0));
+  const out = [];
+  let band = [];
+  const flush = () => {
+    for (let c = 0; c < bands.length; c++) for (const e of band) if (colOf(e) === c) out.push(e);
+    band = [];
+  };
+  for (const e of sorted) {
+    if (spans(e)) {
+      flush();
+      out.push(e);
+    } else band.push(e);
+  }
+  flush();
+  return out;
+}
+
+// ../../packages/jdf-pdf-import/src/paragraphs.ts
 var PT_TO_MM2 = 0.352778;
+var LIST_START = /^\s*(?:[•◦▪●■\-–—*]\s|\(?\d{1,3}[.)]\s|\(?[a-zA-Z][.)]\s|[ivx]{1,4}[.)]\s)/;
+var TERMINAL = /[.!?:;]["'”’)\]]*$/;
+function textOf(e) {
+  if (e.type === "text") return String(e.content ?? "");
+  return (e.runs ?? []).map((r) => String(r.text ?? "")).join("");
+}
+function foldParagraphs(elements, meta, pageWmm) {
+  const foldable = (e) => !!meta.get(e) && !!e.position && (e.type === "text" && !e.heading || e.type === "richtext");
+  const out = [];
+  let i = 0;
+  while (i < elements.length) {
+    const first = elements[i];
+    if (!foldable(first)) {
+      out.push(first);
+      i++;
+      continue;
+    }
+    const para = [first];
+    let j = i + 1;
+    while (j < elements.length) {
+      const prev = para[para.length - 1], next = elements[j];
+      if (!foldable(next) || !continues(para, prev, next, meta)) break;
+      para.push(next);
+      j++;
+    }
+    out.push(para.length > 1 ? merge(para, meta, pageWmm) : first);
+    i = j;
+  }
+  return out;
+}
+function continues(para, prev, next, meta) {
+  const mp = meta.get(prev), mn = meta.get(next); meta.get(para[0]);
+  if (mp.face !== mn.face || Math.abs(mp.size - mn.size) >= 0.5) return false;
+  const lineH = mp.size * PT_TO_MM2;
+  const pitch = next.position.y - prev.position.y;
+  if (pitch < lineH * 0.9 || pitch > lineH * 1.75) return false;
+  const dx = next.position.x - (para.length === 1 ? prev.position.x : para[0].position.x);
+  if (para.length === 1 ? Math.abs(dx) > 6 : Math.abs(dx) > 1.2) return false;
+  if (para.length >= 2 && Math.abs(next.position.x - para[1].position.x) > 1.2) return false;
+  const nText = textOf(next), pText = textOf(prev).trimEnd();
+  if (LIST_START.test(nText) || !nText.trim()) return false;
+  const maxW = Math.max(mp.w, mn.w, ...para.map((e) => meta.get(e).w));
+  const full = mp.w >= maxW * 0.85;
+  const endsHyphen = /[-‐‑]$/.test(pText);
+  const softEnd = !TERMINAL.test(pText);
+  const startsLower = /^[a-z(\[]/.test(nText.trimStart());
+  return full || endsHyphen || softEnd && (startsLower || mp.w >= maxW * 0.6);
+}
+function merge(para, meta, pageWmm) {
+  const first = para[0], last = para[para.length - 1];
+  const m0 = meta.get(first);
+  const lineH = m0.size * PT_TO_MM2;
+  const x = Math.min(...para.map((e) => e.position.x));
+  const indent = first.position.x - x;
+  const right = Math.max(...para.map((e) => e.position.x + meta.get(e).w));
+  const pitch = (last.position.y - first.position.y) / (para.length - 1);
+  const width = Math.min(pageWmm - x, (right - x) * 1.2 + lineH * 0.4);
+  const joiner = () => "\n";
+  const style = { ...first.style ?? {}, lineHeight: Math.round(pitch / lineH * 100) / 100, ...indent > 0.5 ? { textIndent: Math.round(indent * 100) / 100 } : {} };
+  const allText = para.every((e) => e.type === "text" && !e.link);
+  const base = { position: { x, y: first.position.y }, width: Math.round(width * 100) / 100, style };
+  if (allText) {
+    const content = para.map((e) => String(e.content ?? "").trim().replace(/[ \t]+/g, " ")).join(joiner());
+    const { height: _hh2, ...firstRest } = first;
+    return { ...firstRest, ...base, content };
+  }
+  const runs = [];
+  for (const e of para) {
+    const lineRuns = e.type === "text" ? [{ text: String(e.content ?? "").trim(), ...e.style?.fontWeight === "bold" ? { bold: true } : {}, ...e.style?.fontStyle === "italic" ? { italic: true } : {}, ...e.style?.color && e.style.color !== "#000000" ? { color: e.style.color } : {}, ...e.link ? { link: e.link } : {} }] : (e.runs ?? []).map((r) => ({ ...r }));
+    if (runs.length) runs[runs.length - 1].text = `${String(runs[runs.length - 1].text).trimEnd()}${joiner()}`;
+    runs.push(...lineRuns);
+  }
+  const { content: _c, heading: _h, link: _l, height: _hh, ...rest } = first;
+  return { ...rest, ...base, type: "richtext", runs, style: { fontSize: style.fontSize, fontFamily: style.fontFamily, lineHeight: style.lineHeight, ...style.textIndent ? { textIndent: style.textIndent } : {}, ...style.opacity != null ? { opacity: style.opacity } : {} } };
+}
+
+// ../../packages/jdf-pdf-import/src/core.ts
+var PT_TO_MM3 = 0.352778;
 function classifyFont(name) {
   const n = (name || "").toLowerCase();
   const bold = /bold|black|heavy|semibold|demibold|extrabold/.test(n);
@@ -959,10 +1124,10 @@ async function walkOps(page, OPS, viewport) {
     const minY = Math.min(...ys), maxY = Math.max(...ys);
     imagePositions.push({
       name,
-      x: minX * PT_TO_MM2,
-      y: minY * PT_TO_MM2,
-      w: (maxX - minX) * PT_TO_MM2,
-      h: (maxY - minY) * PT_TO_MM2,
+      x: minX * PT_TO_MM3,
+      y: minY * PT_TO_MM3,
+      w: (maxX - minX) * PT_TO_MM3,
+      h: (maxY - minY) * PT_TO_MM3,
       inline,
       maskFill
     });
@@ -978,13 +1143,13 @@ async function walkOps(page, OPS, viewport) {
       const br = toViewport(r.x + r.w, r.y);
       shapes.push({
         kind: "rect",
-        x: Math.min(tl.x, br.x) * PT_TO_MM2,
-        y: Math.min(tl.y, br.y) * PT_TO_MM2,
-        width: Math.abs(br.x - tl.x) * PT_TO_MM2,
-        height: Math.abs(br.y - tl.y) * PT_TO_MM2,
+        x: Math.min(tl.x, br.x) * PT_TO_MM3,
+        y: Math.min(tl.y, br.y) * PT_TO_MM3,
+        width: Math.abs(br.x - tl.x) * PT_TO_MM3,
+        height: Math.abs(br.y - tl.y) * PT_TO_MM3,
         fill: isFill ? gs.fill : void 0,
         stroke: isStroke ? gs.stroke : void 0,
-        strokeWidth: isStroke ? gs.lineWidth * PT_TO_MM2 : void 0,
+        strokeWidth: isStroke ? gs.lineWidth * PT_TO_MM3 : void 0,
         opacity: isFill ? gs.fillAlpha : gs.strokeAlpha
       });
     }
@@ -998,13 +1163,13 @@ async function walkOps(page, OPS, viewport) {
       const h = Math.abs(br.y - tl.y);
       shapes.push({
         kind: "rect",
-        x: x * PT_TO_MM2,
-        y: y * PT_TO_MM2,
-        width: w * PT_TO_MM2,
-        height: h * PT_TO_MM2,
+        x: x * PT_TO_MM3,
+        y: y * PT_TO_MM3,
+        width: w * PT_TO_MM3,
+        height: h * PT_TO_MM3,
         fill: isFill ? gs.fill : void 0,
         stroke: isStroke ? gs.stroke : void 0,
-        strokeWidth: isStroke ? gs.lineWidth * PT_TO_MM2 : void 0,
+        strokeWidth: isStroke ? gs.lineWidth * PT_TO_MM3 : void 0,
         opacity: isFill ? gs.fillAlpha : gs.strokeAlpha
       });
     } else if (pathSegments.length === 2 && pathSegments[0].type === "M" && pathSegments[1].type === "L") {
@@ -1016,35 +1181,35 @@ async function walkOps(page, OPS, viewport) {
       const minY = Math.min(va.y, vb.y);
       const maxX = Math.max(va.x, vb.x);
       const maxY = Math.max(va.y, vb.y);
-      const x1Local = (va.x - minX) * PT_TO_MM2;
-      const y1Local = (va.y - minY) * PT_TO_MM2;
-      const x2Local = (vb.x - minX) * PT_TO_MM2;
-      const y2Local = (vb.y - minY) * PT_TO_MM2;
-      const wLocal = Math.max(0.05, (maxX - minX) * PT_TO_MM2);
-      const hLocal = Math.max(0.05, (maxY - minY) * PT_TO_MM2);
+      const x1Local = (va.x - minX) * PT_TO_MM3;
+      const y1Local = (va.y - minY) * PT_TO_MM3;
+      const x2Local = (vb.x - minX) * PT_TO_MM3;
+      const y2Local = (vb.y - minY) * PT_TO_MM3;
+      const wLocal = Math.max(0.05, (maxX - minX) * PT_TO_MM3);
+      const hLocal = Math.max(0.05, (maxY - minY) * PT_TO_MM3);
       const dx = Math.abs(va.x - vb.x);
       const dy = Math.abs(va.y - vb.y);
       const axisAligned = dx < 0.5 || dy < 0.5;
       if (axisAligned) {
         shapes.push({
           kind: "line",
-          x: minX * PT_TO_MM2,
-          y: minY * PT_TO_MM2,
+          x: minX * PT_TO_MM3,
+          y: minY * PT_TO_MM3,
           width: wLocal,
           height: hLocal,
           stroke: isStroke ? gs.stroke : void 0,
-          strokeWidth: isStroke ? gs.lineWidth * PT_TO_MM2 : void 0,
+          strokeWidth: isStroke ? gs.lineWidth * PT_TO_MM3 : void 0,
           opacity: gs.strokeAlpha
         });
       } else {
         shapes.push({
           kind: "path",
-          x: minX * PT_TO_MM2,
-          y: minY * PT_TO_MM2,
+          x: minX * PT_TO_MM3,
+          y: minY * PT_TO_MM3,
           width: wLocal,
           height: hLocal,
           stroke: isStroke ? gs.stroke : void 0,
-          strokeWidth: isStroke ? gs.lineWidth * PT_TO_MM2 : void 0,
+          strokeWidth: isStroke ? gs.lineWidth * PT_TO_MM3 : void 0,
           opacity: gs.strokeAlpha,
           path: `M ${x1Local.toFixed(2)} ${y1Local.toFixed(2)} L ${x2Local.toFixed(2)} ${y2Local.toFixed(2)}`
         });
@@ -1077,20 +1242,20 @@ async function walkOps(page, OPS, viewport) {
           if (seg.type === "Z") return "Z";
           const p = [];
           for (let i = 0; i < seg.pts.length; i += 2) {
-            p.push(((seg.pts[i] - minX) * PT_TO_MM2).toFixed(2));
-            p.push(((seg.pts[i + 1] - minY) * PT_TO_MM2).toFixed(2));
+            p.push(((seg.pts[i] - minX) * PT_TO_MM3).toFixed(2));
+            p.push(((seg.pts[i + 1] - minY) * PT_TO_MM3).toFixed(2));
           }
           return `${seg.type} ${p.join(" ")}`;
         }).join(" ");
         shapes.push({
           kind: "path",
-          x: minX * PT_TO_MM2,
-          y: minY * PT_TO_MM2,
-          width: bw * PT_TO_MM2,
-          height: bh * PT_TO_MM2,
+          x: minX * PT_TO_MM3,
+          y: minY * PT_TO_MM3,
+          width: bw * PT_TO_MM3,
+          height: bh * PT_TO_MM3,
           fill: isFill ? gs.fill : void 0,
           stroke: isStroke ? gs.stroke : void 0,
-          strokeWidth: isStroke ? gs.lineWidth * PT_TO_MM2 : void 0,
+          strokeWidth: isStroke ? gs.lineWidth * PT_TO_MM3 : void 0,
           opacity: isFill ? gs.fillAlpha : gs.strokeAlpha,
           path: d
         });
@@ -1431,10 +1596,10 @@ async function extractLinks(doc, page, viewport) {
     const xMax = Math.max(c1.x, c2.x);
     const yMax = Math.max(c1.y, c2.y);
     const rectMm = {
-      x: xMin * PT_TO_MM2,
-      y: yMin * PT_TO_MM2,
-      w: (xMax - xMin) * PT_TO_MM2,
-      h: (yMax - yMin) * PT_TO_MM2
+      x: xMin * PT_TO_MM3,
+      y: yMin * PT_TO_MM3,
+      w: (xMax - xMin) * PT_TO_MM3,
+      h: (yMax - yMin) * PT_TO_MM3
     };
     const url = a.url || a.unsafeUrl;
     const destPage = url ? void 0 : await resolveDestPage(doc, a.dest);
@@ -1472,10 +1637,10 @@ async function extractFormWidgets(page, viewport) {
     })).filter((o) => o.value !== "") : [];
     out.push({
       rectMm: {
-        x: xMin * PT_TO_MM2,
-        y: yMin * PT_TO_MM2,
-        w: (xMax - xMin) * PT_TO_MM2,
-        h: (yMax - yMin) * PT_TO_MM2
+        x: xMin * PT_TO_MM3,
+        y: yMin * PT_TO_MM3,
+        w: (xMax - xMin) * PT_TO_MM3,
+        h: (yMax - yMin) * PT_TO_MM3
       },
       fieldType: a.fieldType || "",
       fieldName: a.fieldName || `field-${out.length + 1}`,
@@ -1764,12 +1929,12 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
       const w = safeNum(it.width, 0);
       runs.push({
         text: it.str,
-        x: safeNum(vx * PT_TO_MM2, 0),
-        y: safeNum(yTop * PT_TO_MM2, 0),
+        x: safeNum(vx * PT_TO_MM3, 0),
+        y: safeNum(yTop * PT_TO_MM3, 0),
         fontSize: safeNum(fontSize, 10),
         fontName: it.fontName,
-        width: safeNum(w * PT_TO_MM2, 0),
-        height: safeNum((it.height || fontSize) * PT_TO_MM2, fontSize * PT_TO_MM2),
+        width: safeNum(w * PT_TO_MM3, 0),
+        height: safeNum((it.height || fontSize) * PT_TO_MM3, fontSize * PT_TO_MM3),
         color: op?.fill || "#000000",
         opacity: invisible ? 0 : safeNum(op?.alpha, 1)
       });
@@ -1792,10 +1957,10 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
       }
       const sameLine = Math.abs(last.y - r.y) <= Y_TOL;
       const sameStyle = Math.abs(last.fontSize - r.fontSize) < 0.4 && (last.fontName === r.fontName || fontKey(last.fontName) === fontKey(r.fontName)) && last.color === r.color && Math.abs(last.opacity - r.opacity) < 0.05;
-      const emMm = r.fontSize * PT_TO_MM2;
+      const emMm = r.fontSize * PT_TO_MM3;
       const extent = (t) => {
         if (!/\s$/.test(t.text)) return t.width;
-        const em = t.fontSize * PT_TO_MM2;
+        const em = t.fontSize * PT_TO_MM3;
         const est = Math.max(1, t.text.trim().length) * em * kGlyph + em * 0.25;
         if (stretchedSpaces) return Math.min(t.width, est);
         return t.width > est * 1.4 ? est : t.width;
@@ -1814,11 +1979,19 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
       }
     }
     const elements = [];
+    const lineMeta = /* @__PURE__ */ new WeakMap();
     const tRuns = lines.map((l) => {
       const cls = fontMap.get(l.fontName) || classifyFont(l.fontName || "");
       return { text: l.text, x: l.x, y: l.y, width: l.width, height: l.height, fontSize: l.fontSize, fontName: l.fontName, color: l.color, bold: cls.weight === "bold" };
     });
-    const detected = options.detectTables === false ? [] : detectTables(tRuns, ops.shapes, pageW * PT_TO_MM2);
+    const sizeChars = /* @__PURE__ */ new Map();
+    for (const l of lines) {
+      const k = Math.round(l.fontSize * 2) / 2;
+      sizeChars.set(k, (sizeChars.get(k) ?? 0) + l.text.length);
+    }
+    const bodyFontSize = [...sizeChars.entries()].sort((a, b) => b[1] - a[1])[0]?.[0] ?? 0;
+    const gutters = options.readingOrder === false ? [] : detectGutters(lines.map((l) => ({ text: l.text, x: l.x, y: l.y, width: l.width, fontSize: l.fontSize })), bodyFontSize, pageW * PT_TO_MM3);
+    const detected = options.detectTables === false ? [] : detectTables(tRuns, ops.shapes, pageW * PT_TO_MM3, gutters);
     const consumedLines = /* @__PURE__ */ new Set();
     const consumedShapes = /* @__PURE__ */ new Set();
     const tableAtLine = /* @__PURE__ */ new Map();
@@ -1827,7 +2000,7 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
       for (const k of t.shapeIdx) consumedShapes.add(k);
       tableAtLine.set(Math.min(...t.lineIdx), t.element);
     }
-    const pageWmm = pageW * PT_TO_MM2, pageHmm = pageH * PT_TO_MM2;
+    const pageWmm = pageW * PT_TO_MM3, pageHmm = pageH * PT_TO_MM3;
     ops.shapes.forEach((sh, shapeIdx) => {
       if (consumedShapes.has(shapeIdx)) return;
       if (sh.width < 0.3 && sh.height < 0.3) return;
@@ -1871,12 +2044,6 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
         fit: "fill"
       });
     }
-    const sizeChars = /* @__PURE__ */ new Map();
-    for (const l of lines) {
-      const k = Math.round(l.fontSize * 2) / 2;
-      sizeChars.set(k, (sizeChars.get(k) ?? 0) + l.text.length);
-    }
-    const bodyFontSize = [...sizeChars.entries()].sort((a, b) => b[1] - a[1])[0]?.[0] ?? 0;
     const rowOf = /* @__PURE__ */ new Map();
     const rowStartOf = /* @__PURE__ */ new Map();
     const nextOnRow = /* @__PURE__ */ new Map();
@@ -1884,10 +2051,10 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
       const order = lines.map((_, i) => i).filter((i) => !consumedLines.has(i));
       for (let a = 0; a < order.length; a++) {
         const i = order[a], li = lines[i];
-        const tolY = Math.max(0.6, li.fontSize * PT_TO_MM2 * 0.35);
         let bestNext = -1, bestX = Infinity;
         for (let b = 0; b < order.length; b++) {
           const j = order[b], lj = lines[j];
+          const tolY = Math.max(0.6, Math.max(li.fontSize, lj.fontSize) * PT_TO_MM3 * 0.5);
           if (j === i || Math.abs(lj.y - li.y) > tolY || lj.x <= li.x) continue;
           if (lj.x < bestX) {
             bestX = lj.x;
@@ -1904,7 +2071,7 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
         let cur = i;
         while (nextOnRow.has(cur)) {
           const j = nextOnRow.get(cur), lc = lines[cur], lj = lines[j];
-          const em = Math.min(lc.fontSize, lj.fontSize) * PT_TO_MM2;
+          const em = Math.min(lc.fontSize, lj.fontSize) * PT_TO_MM3;
           const gap = lj.x - (lc.x + lc.width);
           if (gap < -em * 0.3 || gap > em * 0.6) break;
           row.push(j);
@@ -1929,9 +2096,9 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
         const first = lines[row[0]], last = lines[row[row.length - 1]];
         const base = runStyle(first);
         const rowEnd = last.x + last.width;
-        const measuredW = Math.max((rowEnd - first.x) * 1.2 + first.fontSize * PT_TO_MM2 * 0.4, first.fontSize * PT_TO_MM2);
+        const measuredW = Math.max((rowEnd - first.x) * 1.2 + first.fontSize * PT_TO_MM3 * 0.4, first.fontSize * PT_TO_MM3);
         const nextIdx2 = nextOnRow.get(row[row.length - 1]);
-        const cap2 = nextIdx2 != null ? lines[nextIdx2].x - first.x - first.fontSize * PT_TO_MM2 * 0.3 : pageWmm - first.x;
+        const cap2 = nextIdx2 != null ? lines[nextIdx2].x - first.x - first.fontSize * PT_TO_MM3 * 0.3 : pageWmm - first.x;
         const runs2 = [];
         row.forEach((idx, k) => {
           const r = lines[idx];
@@ -1940,7 +2107,7 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
           if (k > 0) {
             const prev2 = lines[row[k - 1]];
             const gap = r.x - (prev2.x + prev2.width);
-            if (gap > r.fontSize * PT_TO_MM2 * 0.08 && !/\s$/.test(prev2.text) && !/^\s/.test(text2)) text2 = " " + text2;
+            if (gap > r.fontSize * PT_TO_MM3 * 0.08 && !/\s$/.test(prev2.text) && !/^\s/.test(text2)) text2 = " " + text2;
           }
           const run = { text: text2 };
           if (st.bold) run.bold = true;
@@ -1954,13 +2121,16 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
         });
         const style2 = { fontSize: Math.round(first.fontSize * 10) / 10, fontFamily: base.cls.family };
         if (first.opacity < 0.999) style2.opacity = Math.round(first.opacity * 100) / 100;
-        elements.push({
+        const rt = {
           type: "richtext",
           runs: runs2,
           position: { x: Math.max(0, Math.round(first.x * 100) / 100), y: Math.max(0, Math.round(Math.min(...row.map((i) => lines[i].y)) * 100) / 100) },
-          width: Math.max(2, Math.round(Math.max(first.fontSize * PT_TO_MM2, Math.min(measuredW, cap2)) * 100) / 100),
+          width: Math.max(2, Math.round(Math.max(first.fontSize * PT_TO_MM3, Math.min(measuredW, cap2)) * 100) / 100),
           style: style2
-        });
+        };
+        const dominant = row.map((i) => lines[i]).sort((a, b) => b.text.trim().length - a.text.trim().length)[0];
+        lineMeta.set(rt, { w: Math.max(first.fontSize * PT_TO_MM3, rowEnd - first.x), size: dominant.fontSize, face: fontKey(dominant.fontName) });
+        elements.push(rt);
         return;
       }
       const cls = fontMap.get(l.fontName) || classifyFont(l.fontName || "");
@@ -1973,10 +2143,10 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
       if (l.color !== "#000000") style.color = l.color;
       if (l.opacity < 0.999) style.opacity = Math.round(l.opacity * 100) / 100;
       const link = findLinkForRun2(l);
-      const measured = Math.max(l.width * 1.2 + l.fontSize * PT_TO_MM2 * 0.4, l.fontSize * PT_TO_MM2);
+      const measured = Math.max(l.width * 1.2 + l.fontSize * PT_TO_MM3 * 0.4, l.fontSize * PT_TO_MM3);
       const remaining = Math.max(measured, pageWmm - l.x);
       const nextIdx = nextOnRow.get(lineIdx);
-      const cap = nextIdx != null ? Math.max(l.fontSize * PT_TO_MM2, lines[nextIdx].x - l.x - l.fontSize * PT_TO_MM2 * 0.3) : Infinity;
+      const cap = nextIdx != null ? Math.max(l.fontSize * PT_TO_MM3, lines[nextIdx].x - l.x - l.fontSize * PT_TO_MM3 * 0.3) : Infinity;
       const elWidth = Math.min(measured, remaining, cap);
       const text = {
         type: "text",
@@ -1997,14 +2167,26 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
         else if (link.destPage != null) text.link = { type: "internal", target: `#page-${link.destPage + 1}` };
       }
       const prev = elements[elements.length - 1];
-      if (text.heading && prev && prev.type === "text" && prev.heading === text.heading && !link && !prev.link && Math.abs(prev.style?.fontSize - style.fontSize) < 0.5 && Math.abs(prev.position.x - text.position.x) < 1 && text.position.y - prev.position.y < l.fontSize * PT_TO_MM2 * 2.2 && text.position.y > prev.position.y) {
+      if (text.heading && prev && prev.type === "text" && prev.heading === text.heading && !link && !prev.link && Math.abs(prev.style?.fontSize - style.fontSize) < 0.5 && Math.abs(prev.position.x - text.position.x) < 1 && text.position.y - prev.position.y < l.fontSize * PT_TO_MM3 * 2.2 && text.position.y > prev.position.y) {
         prev.content = `${prev.content} ${text.content}`.replace(/\s+/g, " ");
         prev.tocEntry = prev.content;
         prev.width = Math.max(prev.width ?? 0, text.width ?? 0);
         return;
       }
+      lineMeta.set(text, { w: Math.max(l.fontSize * PT_TO_MM3, l.width), size: l.fontSize, face: fontKey(l.fontName) });
       elements.push(text);
     });
+    if (options.readingOrder !== false) {
+      if (gutters.length) {
+        const isFlow = (e) => e.type === "text" || e.type === "richtext" || e.type === "table";
+        const flow = elements.filter(isFlow), rest = elements.filter((e) => !isFlow(e));
+        elements.splice(0, elements.length, ...rest, ...orderByColumns(flow, gutters, pageWmm));
+      }
+    }
+    if (options.foldParagraphs !== false) {
+      const folded = foldParagraphs(elements, lineMeta, pageWmm);
+      elements.splice(0, elements.length, ...folded);
+    }
     for (const w of formWidgets) {
       if (w.pushButton) continue;
       const baseEl = {
@@ -2039,7 +2221,7 @@ async function importPdfToJdf(source, title, runtime, options = {}) {
     }
     pages.push({
       id: `page-${pi}`,
-      pageSize: { width: Math.round(pageW * PT_TO_MM2 * 100) / 100, height: Math.round(pageH * PT_TO_MM2 * 100) / 100 },
+      pageSize: { width: Math.round(pageW * PT_TO_MM3 * 100) / 100, height: Math.round(pageH * PT_TO_MM3 * 100) / 100 },
       margins: { top: 0, right: 0, bottom: 0, left: 0 },
       elements
     });

package/dist/jdf-schema.json

@@ -103,6 +103,7 @@
         "backgroundColor": { "type": "string" },
         "textAlign": { "type": "string", "enum": ["left","center","right","justify"] },
         "lineHeight": { "type": "number" },
+        "textIndent": { "type": "number", "description": "First-line indent in the document unit." },
         "letterSpacing": { "oneOf": [{ "type": "number" }, { "type": "string" }] },
         "padding": { "oneOf": [{ "type": "number" }, { "type": "string" }, { "$ref": "#/definitions/Margins" }] },
         "margin": { "oneOf": [{ "type": "number" }, { "type": "string" }, { "$ref": "#/definitions/Margins" }] },

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@uurtech/jdf-cli",
-  "version": "0.2.2",
+  "version": "0.2.3",
   "description": "Command-line tool for the JDF (JSON Document Format) — validate and convert documents.",
   "license": "MIT",
   "author": "Ugur Kazdal",