@bendyline/squisq 2.13.5 → 2.13.6

package/dist/{chunk-F5WTRHWL.js → chunk-4O26IMVA.js}

@@ -4,15 +4,21 @@ import {
   flattenRenderableBlocks,
   getPinnedBlockMeta,
   markdownToDoc
-} from "./chunk-ZF4NWRZU.js";
+} from "./chunk-VQXOX3YN.js";
 import {
   deriveTemplateInputs,
+  extractEmbeddedVideos,
   extractImages,
   isAsciiDiagramFence,
   isAsciiTimelineFence,
+  isExplicitDiagramLang,
+  isExplicitTimelineLang,
+  isExplicitTreeLang,
   isPictureFence,
+  isPictureParagraph,
+  isStandaloneHtmlVideo,
   isTreeFence
-} from "./chunk-CZHWFQ7C.js";
+} from "./chunk-CSR6VCSJ.js";
 import {
   extractContent,
   mapElementToBlock,
@@ -20,6 +26,7 @@ import {
 } from "./chunk-SVN7T6BK.js";
 import {
   cloneAndFreezeData,
+  cloneData,
   isTemplateBlock
 } from "./chunk-WNEH3BZP.js";
 import {
@@ -30,7 +37,9 @@ import {
 } from "./chunk-DOOUVVOM.js";
 import {
   extractPlainText,
+  findBlockVideoEmbed,
   getChildren,
+  parseTimeSeconds,
   stringifyHtmlNodes
 } from "./chunk-KVEOD57Z.js";
 import {
@@ -57,7 +66,8 @@ var PRESENTATION_LAYOUTS = [
 var PRESENTATION_LIMITS = {
   sourceCharacters: 4e4,
   beats: 80,
-  serializedCharacters: 16e4
+  serializedCharacters: 16e4,
+  mediaPerBeat: 4
 };
 function record(value) {
   return !!value && typeof value === "object" && !Array.isArray(value);
@@ -90,13 +100,15 @@ function parsePresentationPlan(input) {
     if (!record(beat) || !keys(
       beat,
       ["id", "sourceStart", "sourceEnd", "layout", "headline", "points"],
-      ["imageSrc"]
+      ["imageSrc", "media"]
     ) || !text(beat.id, 80) || ids.has(beat.id) || !Number.isInteger(beat.sourceStart) || !Number.isInteger(beat.sourceEnd) || typeof beat.sourceStart !== "number" || typeof beat.sourceEnd !== "number" || beat.sourceStart < end || beat.sourceEnd <= beat.sourceStart || beat.sourceEnd > value.sourceText.length || value.sourceText.slice(end, beat.sourceStart).trim() || !PRESENTATION_LAYOUTS.includes(beat.layout) || !text(beat.headline, 110) || !Array.isArray(beat.points) || beat.points.length > 4 || !beat.points.every((point) => text(point, 180)))
       return null;
     if (["comparison", "steps", "list"].includes(String(beat.layout)) && beat.points.length < 2)
       return null;
     if (beat.layout === "comparison" && beat.points.length !== 2) return null;
     if (beat.layout === "image" ? !text(beat.imageSrc, 2048) : "imageSrc" in beat) return null;
+    if ("media" in beat && (!Array.isArray(beat.media) || !beat.media.length || beat.media.length > PRESENTATION_LIMITS.mediaPerBeat || new Set(beat.media).size !== beat.media.length || !beat.media.every((ref) => text(ref, 1100))))
+      return null;
     ids.add(beat.id);
     end = beat.sourceEnd;
     beats.push({
@@ -106,7 +118,8 @@ function parsePresentationPlan(input) {
       layout: beat.layout,
       headline: beat.headline,
       points: [...beat.points],
-      ...beat.layout === "image" ? { imageSrc: beat.imageSrc } : {}
+      ...beat.layout === "image" ? { imageSrc: beat.imageSrc } : {},
+      ..."media" in beat ? { media: [...beat.media] } : {}
     });
   }
   if (value.sourceText.slice(end).trim()) return null;
@@ -1638,6 +1651,55 @@ function videoBlock(video, index, segmentId, duration) {
   };
 }
 
+// src/transform/presentationMedia.ts
+function mediaKind(node) {
+  if (node.type === "code")
+    return isPictureFence(node) || isExplicitDiagramLang(node.lang) || isExplicitTreeLang(node.lang) || isExplicitTimelineLang(node.lang) ? { kind: "diagram", continuous: true } : null;
+  if (isStandaloneHtmlVideo(node)) return { kind: "video", continuous: true };
+  if (findBlockVideoEmbed(node) || isPictureParagraph(node) && extractEmbeddedVideos([node]).length > 0)
+    return { kind: "video", continuous: false };
+  return null;
+}
+function collectPresentationMedia(doc, sectionIds) {
+  const sections = new Set(sectionIds);
+  const result = /* @__PURE__ */ new Map();
+  let owner = sectionIds[0];
+  for (const block of flattenRenderableBlocks(doc.blocks)) {
+    if (sections.has(block.id)) owner = block.id;
+    if (!owner) continue;
+    for (const node of block.contents ?? []) {
+      const kind = mediaKind(node);
+      if (!kind) continue;
+      const items = result.get(owner) ?? [];
+      items.push({ ...kind, ref: `${owner}#${items.length}`, node });
+      result.set(owner, items);
+    }
+  }
+  return result;
+}
+function videoElements(node, found = []) {
+  if (!node || typeof node !== "object") return found;
+  const value = node;
+  if (value.type === "htmlElement" && value.tagName === "video") found.push(value);
+  for (const key of ["children", "htmlChildren"])
+    if (Array.isArray(value[key])) for (const child of value[key]) videoElements(child, found);
+  return found;
+}
+function presentationMediaNode(item, elapsed) {
+  const node = cloneData(item.node);
+  if (item.kind !== "video" || elapsed <= 0) return node;
+  for (const element of videoElements(node)) {
+    const attributes = element.attributes ?? {};
+    const seconds = (name) => attributes[name] ? parseTimeSeconds(attributes[name]) : null;
+    const from = (seconds("data-squisq-video-clip-start") ?? 0) + Math.max(0, elapsed - (seconds("data-squisq-video-start-at") ?? 0));
+    const until = seconds("data-squisq-video-clip-end");
+    if (until !== null && from >= until) return null;
+    delete attributes["data-squisq-video-start-at"];
+    element.attributes = { ...attributes, "data-squisq-video-clip-start": String(from) };
+  }
+  return node;
+}
+
 // src/transform/presentationPlanner.ts
 var SENTENCE_END = /[.!?][”"')\]]?$/u;
 var SENTENCE_START = /^[\p{Lu}\p{N}"“‘'([]/u;
@@ -1847,6 +1909,27 @@ function createPresentationPlan(doc, hints = defaultPresentationHints()) {
     throw new Error(
       "This document has too many sections for one presentation. Split it into shorter documents."
     );
+  const sectionMedia = collectPresentationMedia(
+    doc,
+    script.blocks.map((block) => block.blockId)
+  );
+  for (const block of script.blocks) {
+    const items = sectionMedia.get(block.blockId);
+    if (!items?.length) continue;
+    const own = beats.filter(
+      (beat) => beat.sourceStart >= block.charStart && beat.sourceStart < block.charEnd
+    );
+    const open = own.filter((beat) => beat !== beats[0] && beat.layout !== "image");
+    const slots = open.length ? open : own;
+    const shown = /* @__PURE__ */ new Set();
+    slots.forEach((beat, slot) => {
+      const from = Math.floor(slot * items.length / slots.length);
+      const to = Math.max(from + 1, Math.floor((slot + 1) * items.length / slots.length));
+      const refs = items.slice(from, to).filter((item) => item.continuous || !shown.has(item.ref)).slice(0, PRESENTATION_LIMITS.mediaPerBeat).map((item) => item.ref);
+      for (const ref of refs) shown.add(ref);
+      if (refs.length) beat.media = refs;
+    });
+  }
   const closing = beats[beats.length - 1];
   if (beats.length > 2 && closing?.layout === "statement" && !hints.blocks.some(
     (hint) => hint.layout && script.blocks.some(
@@ -1875,6 +1958,7 @@ function createPresentationPlan(doc, hints = defaultPresentationHints()) {
       beat.headline = hint.wording.headline ?? beat.headline;
       beat.points = hint.wording.points ?? beat.points;
     }
+    if (beat.media && beat.layout === "steps" && hint?.layout !== "steps") beat.layout = "list";
     if (["steps", "list", "comparison"].includes(beat.layout) && (beat.points.length < 2 || beat.layout === "comparison" && beat.points.length !== 2))
       beat.layout = "statement";
     if (beat.layout !== "image") delete beat.imageSrc;
@@ -1913,6 +1997,18 @@ function compilePresentationPlan(doc, plan) {
     throw new Error(
       "A presentation image is no longer in the document. Make the presentation again."
     );
+  const media = new Map(
+    [
+      ...collectPresentationMedia(
+        doc,
+        script.blocks.map((block) => block.blockId)
+      ).values()
+    ].flat().map((item) => [item.ref, item])
+  );
+  if (checked.beats.some((beat) => beat.media?.some((ref) => !media.has(ref))))
+    throw new Error(
+      "A presentation video or diagram is no longer in the document. Make the presentation again."
+    );
   const duration = narration ? narration.startTime + narration.duration : Math.max(4, script.tokens.length / 2.5);
   const bookmarks = (narration?.bookmarks ?? []).filter(
     (word) => Number.isFinite(word.time) && word.time >= 0 && word.time <= duration && Number.isInteger(word.charOffset) && word.charOffset >= 0 && word.charOffset < script.sourceText.length
@@ -1932,17 +2028,53 @@ function compilePresentationPlan(doc, plan) {
   });
   const authoredBlocks = flattenRenderableBlocks(doc.blocks);
   const buildHints = parsePresentationHints(doc.frontmatter?.[PRESENTATION_HINTS_KEY]);
+  const sourceOf = (beat) => script.blocks.find(
+    (block) => beat.sourceStart >= block.charStart && beat.sourceStart < block.charEnd
+  );
+  const sectionEnds = /* @__PURE__ */ new Map();
+  checked.beats.forEach((beat, index) => {
+    const section = sourceOf(beat)?.blockId;
+    if (section) sectionEnds.set(section, starts[index + 1] ?? duration);
+  });
+  const clipsPlaced = /* @__PURE__ */ new Set();
+  const mediaStarts = /* @__PURE__ */ new Map();
   const blocks = [];
   checked.beats.forEach((beat, index) => {
     const start = starts[index];
     const end = starts[index + 1] ?? duration;
     if (end <= start) return;
-    const source = script.blocks.find(
-      (block2) => beat.sourceStart >= block2.charStart && beat.sourceStart < block2.charEnd
-    );
+    const source = sourceOf(beat);
     const build = buildHints?.blocks.find((hint) => hint.blockId === source?.blockId)?.build ?? authoredBlocks.find((block2) => block2.id === source?.blockId)?.build;
+    const contents = (beat.media ?? []).flatMap((ref) => {
+      const first = mediaStarts.get(ref) ?? start;
+      mediaStarts.set(ref, first);
+      const node = presentationMediaNode(media.get(ref), start - first);
+      return node ? [node] : [];
+    });
+    let clips = [];
+    if (source && !clipsPlaced.has(source.blockId)) {
+      clipsPlaced.add(source.blockId);
+      const sectionEnd = sectionEnds.get(source.blockId) ?? end;
+      clips = (authoredBlocks.find((block2) => block2.id === source.blockId)?.media ?? []).flatMap(
+        (clip2) => {
+          if (clip2.spillover) return [clip2];
+          const room = sectionEnd - start - clip2.startAt;
+          if (room <= 0) return [];
+          const clipStart = clip2.clipStart ?? 0;
+          return [
+            {
+              ...clip2,
+              spillover: true,
+              clipEnd: Math.min(clip2.clipEnd ?? Infinity, clipStart + room)
+            }
+          ];
+        }
+      );
+    }
     const base = {
       ...build ? { build } : {},
+      ...contents.length ? { contents } : {},
+      ...clips.length ? { media: clips } : {},
       id: beat.id,
       duration: end - start,
       startTime: start,
@@ -1980,7 +2112,14 @@ function visualBlock(beat, base, index) {
         explanation: beat.points.join("\n\n")
       };
     case "list":
-      return { ...base, template: "list", title: beat.headline, items: beat.points, colorScheme };
+      return {
+        ...base,
+        template: "list",
+        title: beat.headline,
+        items: beat.points,
+        ordered: false,
+        colorScheme
+      };
     case "comparison":
       return {
         ...base,
@@ -2069,6 +2208,7 @@ export {
   allocateTiming,
   applyTransform,
   transformNarratedSegment,
+  collectPresentationMedia,
   createPresentationPlan,
   compilePresentationPlan,
   applySavedPresentation

package/dist/{chunk-FSZ6HB7S.js → chunk-B2MOGSWY.js}

@@ -5,10 +5,11 @@ import {
 } from "./chunk-NEWNNLWN.js";
 import {
   buildNarrationScript,
+  estimateSyllables,
   expectedSyllablesAt,
   flattenRenderableBlocks,
   wordPosAtExpectedSyllables
-} from "./chunk-ZF4NWRZU.js";
+} from "./chunk-VQXOX3YN.js";
 
 // src/narration/types.ts
 var DEFAULT_FEATURE_CONFIG = Object.freeze({
@@ -839,6 +840,333 @@ function proportionalAlignment(script, effectiveDuration, takeDuration, c, detec
   return { words, blocks, detectedSyllables, cost: 0 };
 }
 
+// src/narration/transcriptAlign.ts
+var MAX_DP_CELLS = 1e6;
+var ANCHOR_GRAM = 3;
+var EXACT_SCORE = 2;
+var NEAR_SCORE = 1;
+var MISMATCH_SCORE = -1;
+var GAP_SCORE = -1;
+var NEAR_SIMILARITY = 0.75;
+var ISOLATED_MATCH_MAX_LENGTH = 3;
+var NON_SPEECH = /^\s*[[(].*[\])]\s*$/u;
+function transcriptWordKeys(text) {
+  return text.normalize("NFKD").replace(/\p{M}+/gu, "").toLowerCase().replace(/['’]/gu, "").split(/[^\p{L}\p{N}]+/u).filter(Boolean);
+}
+function clampTime(value, durationSec) {
+  return Number.isFinite(value) ? Math.min(Math.max(0, value), durationSec) : 0;
+}
+function heardFromSegments(segments, durationSec) {
+  const heard = [];
+  const units = [];
+  const add = (text, startSec, endSec) => {
+    if (NON_SPEECH.test(text)) return;
+    const keys = transcriptWordKeys(text);
+    if (!keys.length) return;
+    for (const key of keys) units.push({ key, index: heard.length });
+    heard.push({ startSec, endSec: Math.max(startSec, endSec) });
+  };
+  const ordered = [...segments].sort((a, b) => a.startSec - b.startSec);
+  for (const segment of ordered) {
+    const start = clampTime(segment.startSec, durationSec);
+    const end = Math.max(start, clampTime(segment.endSec, durationSec));
+    if (segment.words?.length) {
+      for (const word of segment.words) {
+        const wordStart = clampTime(word.startSec, durationSec);
+        add(word.text, wordStart, clampTime(word.endSec, durationSec));
+      }
+      continue;
+    }
+    const pieces = segment.text.split(/\s+/u).filter((piece) => piece && !NON_SPEECH.test(piece));
+    const weights = pieces.map((piece) => estimateSyllables(piece));
+    const total = weights.reduce((sum, weight) => sum + weight, 0) || 1;
+    let before = 0;
+    pieces.forEach((piece, i) => {
+      const from = start + before / total * (end - start);
+      before += weights[i];
+      add(piece, from, start + before / total * (end - start));
+    });
+  }
+  let running = 0;
+  for (const word of heard) {
+    word.startSec = Math.max(running, word.startSec);
+    word.endSec = Math.max(word.startSec, word.endSec);
+    running = word.startSec;
+  }
+  return { heard, units };
+}
+function scriptUnits(script) {
+  const units = [];
+  script.tokens.forEach((token, index) => {
+    if (!token.spoken) return;
+    for (const key of transcriptWordKeys(token.text)) units.push({ key, index });
+  });
+  return units;
+}
+function uniqueGrams(units) {
+  const grams = /* @__PURE__ */ new Map();
+  for (let i = 0; i + ANCHOR_GRAM <= units.length; i++) {
+    const gram = units.slice(i, i + ANCHOR_GRAM).map((unit) => unit.key).join(" ");
+    grams.set(gram, grams.has(gram) ? -1 : i);
+  }
+  return grams;
+}
+function anchorChain(script, heard) {
+  const heardGrams = uniqueGrams(heard);
+  const pairs = [];
+  for (const [gram, s] of uniqueGrams(script)) {
+    const t = heardGrams.get(gram);
+    if (s >= 0 && t !== void 0 && t >= 0) pairs.push([s, t]);
+  }
+  pairs.sort((a, b) => a[0] - b[0]);
+  const tails = [];
+  const previous = new Int32Array(pairs.length).fill(-1);
+  for (let i = 0; i < pairs.length; i++) {
+    let lo = 0;
+    let hi = tails.length;
+    while (lo < hi) {
+      const mid = lo + hi >> 1;
+      if (pairs[tails[mid]][1] < pairs[i][1]) lo = mid + 1;
+      else hi = mid;
+    }
+    if (lo > 0) previous[i] = tails[lo - 1];
+    tails[lo] = i;
+  }
+  const chain = [];
+  for (let i = tails.length ? tails[tails.length - 1] : -1; i >= 0; i = previous[i]) {
+    chain.push(pairs[i]);
+  }
+  return chain.reverse();
+}
+function editDistance(a, b) {
+  let row = Array.from({ length: b.length + 1 }, (_, j) => j);
+  for (let i = 1; i <= a.length; i++) {
+    const next = [i];
+    for (let j = 1; j <= b.length; j++) {
+      next[j] = Math.min(
+        row[j] + 1,
+        next[j - 1] + 1,
+        row[j - 1] + (a.charCodeAt(i - 1) === b.charCodeAt(j - 1) ? 0 : 1)
+      );
+    }
+    row = next;
+  }
+  return row[b.length];
+}
+function createScorer() {
+  const near = /* @__PURE__ */ new Map();
+  return (a, b) => {
+    if (a === b) return EXACT_SCORE;
+    if (a.length < 4 || b.length < 4 || a[0] !== b[0] || Math.abs(a.length - b.length) > 3)
+      return MISMATCH_SCORE;
+    const key = a < b ? `${a} ${b}` : `${b} ${a}`;
+    let similar = near.get(key);
+    if (similar === void 0) {
+      similar = 1 - editDistance(a, b) / Math.max(a.length, b.length) >= NEAR_SIMILARITY;
+      near.set(key, similar);
+    }
+    return similar ? NEAR_SCORE : MISMATCH_SCORE;
+  };
+}
+function alignStretch(script, sFrom, sTo, heard, tFrom, tTo, score) {
+  const rows = sTo - sFrom;
+  const cols = tTo - tFrom;
+  if (rows <= 0 || cols <= 0 || rows * cols > MAX_DP_CELLS) return [];
+  const width = cols + 1;
+  const moves = new Uint8Array((rows + 1) * width);
+  let prev = new Float64Array(width);
+  for (let j2 = 1; j2 <= cols; j2++) {
+    prev[j2] = j2 * GAP_SCORE;
+    moves[j2] = 2;
+  }
+  for (let i2 = 1; i2 <= rows; i2++) {
+    const row = new Float64Array(width);
+    row[0] = i2 * GAP_SCORE;
+    moves[i2 * width] = 1;
+    const key = script[sFrom + i2 - 1].key;
+    for (let j2 = 1; j2 <= cols; j2++) {
+      const diagonal = prev[j2 - 1] + score(key, heard[tFrom + j2 - 1].key);
+      const up = prev[j2] + GAP_SCORE;
+      const left = row[j2 - 1] + GAP_SCORE;
+      if (diagonal >= up && diagonal >= left) {
+        row[j2] = diagonal;
+        moves[i2 * width + j2] = 0;
+      } else if (up >= left) {
+        row[j2] = up;
+        moves[i2 * width + j2] = 1;
+      } else {
+        row[j2] = left;
+        moves[i2 * width + j2] = 2;
+      }
+    }
+    prev = row;
+  }
+  const pairs = [];
+  let i = rows;
+  let j = cols;
+  while (i > 0 || j > 0) {
+    const move = moves[i * width + j];
+    if (move === 0) {
+      const s = sFrom + i - 1;
+      const t = tFrom + j - 1;
+      if (score(script[s].key, heard[t].key) > 0) pairs.push([s, t]);
+      i--;
+      j--;
+    } else if (move === 1) i--;
+    else j--;
+  }
+  return pairs.reverse();
+}
+function matchUnits(script, heard) {
+  const anchored = [];
+  let lastS = -1;
+  let lastT = -1;
+  for (const [s, t] of anchorChain(script, heard)) {
+    for (let k = 0; k < ANCHOR_GRAM; k++) {
+      if (s + k > lastS && t + k > lastT) {
+        anchored.push([s + k, t + k]);
+        lastS = s + k;
+        lastT = t + k;
+      }
+    }
+  }
+  const score = createScorer();
+  const pairs = [];
+  let from = [-1, -1];
+  for (const to of [...anchored, [script.length, heard.length]]) {
+    pairs.push(...alignStretch(script, from[0] + 1, to[0], heard, from[1] + 1, to[1], score));
+    if (to[0] < script.length) pairs.push(to);
+    from = to;
+  }
+  return pairs.filter(([s, t], i) => {
+    if (script[s].key.length > ISOLATED_MATCH_MAX_LENGTH) return true;
+    const before = pairs[i - 1];
+    const after = pairs[i + 1];
+    return before !== void 0 && before[0] === s - 1 && before[1] === t - 1 || after !== void 0 && after[0] === s + 1 && after[1] === t + 1;
+  });
+}
+function placeWords(script, matched, heard, speechEnd, durationSec, gapMinSec) {
+  const n = script.tokens.length;
+  const cumulative = script.cumulativeSyllables;
+  if (!heard.length) {
+    const total = Math.max(1, script.totalSyllables);
+    return script.tokens.map((_, k) => ({
+      tokenIndex: k,
+      tSec: cumulative[k] / total * durationSec,
+      interpolated: true
+    }));
+  }
+  const startOf = (h) => h < heard.length ? heard[h].startSec : speechEnd;
+  const words = new Array(n);
+  let leftToken = -1;
+  let leftHeard = -1;
+  const fill = (rightToken, rightHeard) => {
+    const from = leftToken + 1;
+    const between = rightHeard - leftHeard - 1;
+    const leftBlock = leftToken >= 0 ? script.tokens[leftToken].blockIndex : -1;
+    let leftCount = 0;
+    let rightCount = 0;
+    for (let u = from; u < rightToken; u++) {
+      if (!script.tokens[u].spoken) continue;
+      if (script.tokens[u].blockIndex === leftBlock) leftCount++;
+      else rightCount++;
+    }
+    if (between > 0 && between >= leftCount + rightCount) {
+      let rightFloor = rightHeard - rightCount;
+      for (let h = rightHeard; h > rightFloor && h > 0; h--) {
+        if (startOf(h) - heard[h - 1].endSec >= gapMinSec) {
+          rightFloor = h;
+          break;
+        }
+      }
+      let leftRank = 0;
+      let rightRank = 0;
+      for (let u = from; u < rightToken; u++) {
+        const token = script.tokens[u];
+        const left = token.blockIndex === leftBlock;
+        const position = left ? leftHeard + 1 + leftRank : Math.max(rightFloor, rightHeard - rightCount + rightRank);
+        words[u] = { tokenIndex: u, tSec: startOf(position), interpolated: true };
+        if (token.spoken && left) leftRank++;
+        else if (token.spoken) rightRank++;
+      }
+      return;
+    }
+    const base = cumulative[from] ?? 0;
+    const span = (rightToken < n ? cumulative[rightToken] : script.totalSyllables) - base;
+    for (let u = from; u < rightToken; u++) {
+      let tSec = startOf(rightHeard);
+      if (between > 0) {
+        const fraction = span > 0 ? (cumulative[u] - base) / span : (u - from) / (rightToken - from);
+        const position = leftHeard + 1 + fraction * between;
+        const index = Math.floor(position);
+        tSec = startOf(index) + (position - index) * (startOf(index + 1) - startOf(index));
+      }
+      words[u] = { tokenIndex: u, tSec, interpolated: true };
+    }
+  };
+  for (let k = 0; k < n; k++) {
+    const h = matched[k];
+    if (h === void 0) continue;
+    fill(k, h);
+    words[k] = { tokenIndex: k, tSec: heard[h].startSec, interpolated: false };
+    leftToken = k;
+    leftHeard = h;
+  }
+  fill(n, heard.length);
+  let running = 0;
+  for (const word of words) {
+    running = Math.max(running, word.tSec);
+    word.tSec = running;
+  }
+  return words;
+}
+function alignTranscript(input, config) {
+  const c = { ...DEFAULT_ALIGN_CONFIG, ...config };
+  const { script } = input;
+  const durationSec = Number.isFinite(input.durationSec) && input.durationSec > 0 ? input.durationSec : 0;
+  if (script.tokens.length > c.maxScriptTokens) {
+    throw new RangeError(`Narration script exceeds the ${c.maxScriptTokens}-token safety limit`);
+  }
+  const spokenWords = script.tokens.filter((token) => token.spoken).length;
+  const { heard, units: heardUnits } = heardFromSegments(input.segments, durationSec);
+  if (script.tokens.length === 0 || durationSec === 0) {
+    return { words: [], blocks: [], detectedSyllables: 0, cost: 0, matchedWords: 0, spokenWords };
+  }
+  const units = scriptUnits(script);
+  const matched = new Array(script.tokens.length);
+  for (const [s, t] of matchUnits(units, heardUnits)) {
+    const token = units[s].index;
+    matched[token] ?? (matched[token] = heardUnits[t].index);
+  }
+  const matchedWords = matched.filter((heardIndex) => heardIndex !== void 0).length;
+  const lastSpeechEnd = heard.reduce((end, word) => Math.max(end, word.endSec), 0);
+  const words = placeWords(
+    script,
+    matched,
+    heard,
+    lastSpeechEnd || durationSec,
+    durationSec,
+    c.gapMinSec
+  );
+  const gaps = [];
+  for (let i = 1; i < heard.length; i++) {
+    const startSec = heard[i - 1].endSec;
+    const endSec = heard[i].startSec;
+    if (endSec - startSec >= c.gapMinSec) {
+      gaps.push({ kind: "gap", tSec: (startSec + endSec) / 2, startSec, endSec });
+    }
+  }
+  const blocks = deriveBlockRanges(script, words, gaps, durationSec, lastSpeechEnd, c);
+  return {
+    words,
+    blocks,
+    detectedSyllables: heard.length,
+    cost: 0,
+    matchedWords,
+    spokenWords
+  };
+}
+
 // src/narration/advanceTiming.ts
 var EMPTY_ADVANCE_LOG = Object.freeze([]);
 function recordSlideShown(log, blockId, atMs) {
@@ -963,6 +1291,8 @@ export {
   traceWordPosAt,
   downsampleTrace,
   alignNarration,
+  transcriptWordKeys,
+  alignTranscript,
   EMPTY_ADVANCE_LOG,
   recordSlideShown,
   buildAdvanceTimingJson,

package/dist/{chunk-CZHWFQ7C.js → chunk-CSR6VCSJ.js}

@@ -3977,6 +3977,7 @@ export {
   parseAnnotationInner,
   parseStandaloneAnnotation,
   isMediaAnnotationName,
+  isStandaloneHtmlVideo,
   extractMediaFromContents,
   toCells,
   cellWidth,

package/dist/{chunk-K75GTKSN.js → chunk-FGN4OZVK.js}

@@ -3,7 +3,7 @@ import {
 } from "./chunk-FTN6SGW5.js";
 import {
   applySavedPresentation
-} from "./chunk-F5WTRHWL.js";
+} from "./chunk-4O26IMVA.js";
 import {
   SHAPE_NAMES,
   buildChartData,
@@ -31,7 +31,7 @@ import {
   templateRegistry,
   writeCustomTemplatesToFrontmatter,
   writeCustomThemesToFrontmatter
-} from "./chunk-ZF4NWRZU.js";
+} from "./chunk-VQXOX3YN.js";
 import {
   ASCII_TREE_VOCAB,
   ASCII_VOCAB,
@@ -54,7 +54,7 @@ import {
   toCells,
   toGrid,
   traceBoxes
-} from "./chunk-CZHWFQ7C.js";
+} from "./chunk-CSR6VCSJ.js";
 import {
   iconMarker
 } from "./chunk-7ZAAICW4.js";

package/dist/{chunk-ZF4NWRZU.js → chunk-VQXOX3YN.js}

@@ -32,7 +32,7 @@ import {
   profileBlockContents,
   renderMarkdownBlocksHtml,
   treeToTemplateData
-} from "./chunk-CZHWFQ7C.js";
+} from "./chunk-CSR6VCSJ.js";
 import {
   anchorPoint,
   base64ToUtf8,
@@ -4508,6 +4508,17 @@ function dataTable(input, context) {
 
 // src/doc/templates/chart/parse.ts
 var NUMERIC_COLUMN_THRESHOLD = 0.6;
+function cellText(cell) {
+  if (typeof cell === "string") return cell;
+  if (cell === null || cell === void 0) return "";
+  if (typeof cell === "number") return Number.isFinite(cell) ? String(cell) : "";
+  return typeof cell === "object" ? "" : String(cell);
+}
+function normalizeChartTable(table) {
+  const headers = Array.isArray(table.headers) ? table.headers.map(cellText) : [];
+  const rows = Array.isArray(table.rows) ? table.rows.filter(Array.isArray).map((row) => row.map(cellText)) : [];
+  return { headers, rows };
+}
 function parseCellNumber(raw) {
   if (typeof raw !== "string") return null;
   let text = raw.trim();
@@ -4553,8 +4564,7 @@ function resolveColumnRef(ref, headers) {
   return null;
 }
 function buildChartData(table, options = {}) {
-  const headers = Array.isArray(table.headers) ? table.headers : [];
-  const rows = Array.isArray(table.rows) ? table.rows : [];
+  const { headers, rows } = normalizeChartTable(table);
   if (headers.length === 0 || rows.length === 0) return null;
   const warnings = [];
   let labelIndex = 0;
@@ -4972,7 +4982,7 @@ function renderChart(kind, input, context) {
 }
 function resolveTable(input, context) {
   if (Array.isArray(input.headers) && input.headers.length > 0 && Array.isArray(input.rows) && input.rows.length > 0) {
-    return { headers: input.headers, rows: input.rows };
+    return normalizeChartTable(input);
   }
   const contents = input.contents ?? context.block?.contents;
   return extractTableFromContents(contents);
@@ -6276,14 +6286,15 @@ function mapForegroundLayers(layers, contentRect, viewport, fontScale, template)
     if (isFullViewportBackground(layer, viewport)) return layer;
     const position = mapPosition(layer.position, contentRect, viewport);
     if (layer.type === "text") {
+      const fontSize = Math.max(12, Math.round(layer.content.style.fontSize * fontScale));
       return {
         ...layer,
-        position,
+        position: layer.content.flow?.role === "marker" ? markerPosition(layer, position, fontSize / layer.content.style.fontSize, viewport) : position,
         content: {
           ...layer.content,
           style: {
             ...layer.content.style,
-            fontSize: Math.max(12, Math.round(layer.content.style.fontSize * fontScale)),
+            fontSize,
             ...layer.content.style.padding !== void 0 ? { padding: Math.round(layer.content.style.padding * fontScale) } : {}
           }
         }
@@ -6293,6 +6304,20 @@ function mapForegroundLayers(layers, contentRect, viewport, fontScale, template)
   });
   return arrangeTitleFamily(mapped, template, contentRect, viewport);
 }
+function markerPosition(layer, mapped, textScale, viewport) {
+  if (mapped.width === void 0 || layer.position.width === void 0) return mapped;
+  const width = resolvePositionValue(mapped.width, viewport.width);
+  const needed = Math.ceil(resolvePositionValue(layer.position.width, viewport.width) * textScale);
+  if (needed <= width) return mapped;
+  const grow = needed - width;
+  const anchor = mapped.anchor ?? "top-left";
+  const shift = layer.content.style.textAlign !== "right" || anchor.includes("right") ? 0 : anchor === "center" ? grow / 2 : grow;
+  return {
+    ...mapped,
+    x: resolvePositionValue(mapped.x, viewport.width) - shift,
+    width: needed
+  };
+}
 function arrangeTitleFamily(layers, template, contentRect, viewport) {
   const centerX = contentRect.x + contentRect.width / 2;
   const centerY = contentRect.y + contentRect.height / 2;

package/dist/doc/index.js

@@ -75,9 +75,9 @@ import {
   validateMarkdownSource,
   wordRevealDelay,
   writeDashboardLayoutsToFrontmatter
-} from "../chunk-K75GTKSN.js";
+} from "../chunk-FGN4OZVK.js";
 import "../chunk-FTN6SGW5.js";
-import "../chunk-F5WTRHWL.js";
+import "../chunk-4O26IMVA.js";
 import {
   BASE_INPUT_DESCRIPTORS,
   BLOCK_MEDIA_LAYOUT_POLICIES,
@@ -195,7 +195,7 @@ import {
   wrapWithPersistentLayers,
   writeCustomTemplatesToFrontmatter,
   writeCustomThemesToFrontmatter
-} from "../chunk-ZF4NWRZU.js";
+} from "../chunk-VQXOX3YN.js";
 import {
   ASCII_CHAR_H,
   ASCII_CHAR_W,
@@ -248,7 +248,7 @@ import {
   treeFromMarkdownList,
   treeFromTemplateData,
   treeToTemplateData
-} from "../chunk-CZHWFQ7C.js";
+} from "../chunk-CSR6VCSJ.js";
 import "../chunk-SVN7T6BK.js";
 import {
   PATH_SHAPE_KINDS,

package/dist/index.d.ts

@@ -16,7 +16,7 @@ export { BuildCaptionTrackOptions, CaptionSegmentSource, DEFAULT_WORDS_PER_SECON
 export { SeededRandom, hashString } from './random/index.js';
 export { C as ComparisonData, D as DateData, a as DefinitionData, E as ExtractedElement, b as ExtractionOptions, c as ExtractionResult, d as ExtractionType, F as FactData, I as ImpactLineData, L as ListData, Q as QuoteData, S as StatData, e as extractContent, s as stripMarkdown } from './contentExtractor-BNfVJV2U.js';
 export { AffinityAssignment, AffinityMatchResult, AffinityTarget, ClipSelectionOptions, ClipSource, KeywordExtractionOptions, MapOptions, SelectedClip, TranscriptSegmentSource, calculateKeywordAffinity, extractKeywords, mapElementToBlock, matchByKeywordAffinity, selectSafeZoneClips, selectTranscriptClips } from './generate/index.js';
-export { AnalyzedBlock, BlockImage, DEFAULT_TRANSFORM_STYLE_ID, DYNAMIC_PRESENTATION_STYLE, NarratedSegmentInput, NarratedSegmentResult, NarratedSegmentTransformOptions, NarratedSegmentVideo, PRESENTATION_HINTS_KEY, PRESENTATION_KEY, PRESENTATION_LAYOUTS, PRESENTATION_LIMITS, PresentationBeat, PresentationBlockHint, PresentationHints, PresentationLayout, PresentationPlan, TransformImage, TransformOptions, TransformResult, TransformStyleConfig, TransformStyleId, TransformStyleInput, TransformStyleRegistry, TransformStyleSummary, allocateTiming, analyzeBlocks, applySavedPresentation, applyTransform, compilePresentationPlan, createPresentationPlan, createTransformStyleRegistry, defaultPresentationHints, extractDocImages, getTransformStyleIds, getTransformStyleSummaries, parsePresentationHints, parsePresentationPlan, presentationSections, resolveTransformStyle, serializePresentationHints, serializePresentationPlan, transformNarratedSegment, usesDynamicPresentation } from './transform/index.js';
+export { AnalyzedBlock, BlockImage, DEFAULT_TRANSFORM_STYLE_ID, DYNAMIC_PRESENTATION_STYLE, NarratedSegmentInput, NarratedSegmentResult, NarratedSegmentTransformOptions, NarratedSegmentVideo, PRESENTATION_HINTS_KEY, PRESENTATION_KEY, PRESENTATION_LAYOUTS, PRESENTATION_LIMITS, PresentationBeat, PresentationBlockHint, PresentationHints, PresentationLayout, PresentationMediaItem, PresentationPlan, TransformImage, TransformOptions, TransformResult, TransformStyleConfig, TransformStyleId, TransformStyleInput, TransformStyleRegistry, TransformStyleSummary, allocateTiming, analyzeBlocks, applySavedPresentation, applyTransform, collectPresentationMedia, compilePresentationPlan, createPresentationPlan, createTransformStyleRegistry, defaultPresentationHints, extractDocImages, getTransformStyleIds, getTransformStyleSummaries, parsePresentationHints, parsePresentationPlan, presentationSections, resolveTransformStyle, serializePresentationHints, serializePresentationPlan, transformNarratedSegment, usesDynamicPresentation } from './transform/index.js';
 export { C as CoalesceOptions, P as PrunePolicy, R as RevertFailureReason, a as RevertOptions, b as RevertResult, S as SaveVersionOptions, c as SaveVersionResult, V as Version } from './types-8QjefM9J.js';
 export { DATA_BACKUP_PREFIX, DocumentVersionManager, DocumentVersionManagerOptions, VERSIONS_PREFIX, VERSION_TIMESTAMP_PATTERN, buildDataBackupPath, buildVersionPath, coalesceVersions, formatVersionTimestamp, getDocBasename, listVersions, parseVersionPath, parseVersionTimestamp, pruneVersions, readVersion, revertToVersion, saveVersion } from './versions/index.js';
 export { BuildJsonFormTokensOptions, ControlKind, InferSchemaOptions, JsonFormValidationError, JsonFormValidator, SquisqAnnotatedSchema, SquisqHints, SquisqWhen, appendPointer, arrayItemKind, buildJsonFormTokens, chooseControl, escapePointerSegment, evaluateWhen, getByPointer, inferSchema, isSafeConditionRegex, pointerSegments, primaryType, resolveFlag, resolveJsonFormTheme, resolveRef, setByPointer, toPointer } from './jsonForm/index.js';
@@ -24,7 +24,7 @@ export { AnimatedImageFormat, AnimatedImageInfo, IMAGE_EDIT_ASSETS_PREFIX, IMAGE
 export { I as ICONS, a as IconEntry, b as IconFamily, c as IconTextRun, h as hasIconMarker, i as iconClass, d as iconMarker, s as splitIconMarkers, e as stripIconMarkers } from './inlineIconMarker-BGjv-W4j.js';
 export { IconSuggestion, canonicalIconToken, iconGlyph, looksLikeIconToken, resolveIcon, suggestIcons } from './icons/index.js';
 export { BlockContentProfile, RecommendationResult, profileBlockContents, recommendTemplatesForBlock } from './recommend/index.js';
-export { AdvanceCoverage, AlignConfig, AlignInput, BandpassState, BuildAdvanceTimingOptions, BuildNarrationTimingOptions, BuildScriptOptions, DEFAULT_ALIGN_CONFIG, DEFAULT_FEATURE_CONFIG, DEFAULT_NUCLEI_CONFIG, DEFAULT_PACING_CONFIG, DEFAULT_VAD_CONFIG, EMPTY_ADVANCE_LOG, FeatureConfig, FeatureState, FrameFeatures, NarrationAlignment, NarrationBlockRange, NarrationScript, NarrationSessionConfig, NarrationSessionState, NarrationTimingBlock, NarrationTimingGenerator, NarrationTimingJsonV3, NarrationTimingMethod, NarrationTrace, NucleiConfig, NucleiState, PacingConfig, PacingState, PacingTick, ScriptBlockRange, ScriptToken, SlideAdvance, SlideAdvanceLog, TraceSample, VadConfig, VadState, WordTiming, advanceCoverage, alignNarration, bandpassRun, buildAdvanceTimingJson, buildNarrationScript, buildNarrationTimingJson, createBandpass, createFeatureState, createNarrationSession, createNucleiState, createPacingState, createVadState, detectSyllableOnsets, downsampleTrace, estimateSyllables, expectedSyllablesAt, extractFrameFeatures, featureStep, narrationSessionStep, nucleiStep, pacingStep, parseNarrationTimingJson, reanchorPacing, reanchorSession, recordSlideShown, resolveNarrationTimingGenerator, traceWordPosAt, vadStep, wordIndexAtChar, wordIndexAtTime, wordPosAtExpectedSyllables } from './narration/index.js';
+export { AdvanceCoverage, AlignConfig, AlignInput, BandpassState, BuildAdvanceTimingOptions, BuildNarrationTimingOptions, BuildScriptOptions, DEFAULT_ALIGN_CONFIG, DEFAULT_FEATURE_CONFIG, DEFAULT_NUCLEI_CONFIG, DEFAULT_PACING_CONFIG, DEFAULT_VAD_CONFIG, EMPTY_ADVANCE_LOG, FeatureConfig, FeatureState, FrameFeatures, NarrationAlignment, NarrationBlockRange, NarrationScript, NarrationSessionConfig, NarrationSessionState, NarrationTimingBlock, NarrationTimingGenerator, NarrationTimingJsonV3, NarrationTimingMethod, NarrationTrace, NucleiConfig, NucleiState, PacingConfig, PacingState, PacingTick, ScriptBlockRange, ScriptToken, SlideAdvance, SlideAdvanceLog, TraceSample, TranscriptAlignInput, TranscriptAlignment, TranscriptSegment, TranscriptWord, VadConfig, VadState, WordTiming, advanceCoverage, alignNarration, alignTranscript, bandpassRun, buildAdvanceTimingJson, buildNarrationScript, buildNarrationTimingJson, createBandpass, createFeatureState, createNarrationSession, createNucleiState, createPacingState, createVadState, detectSyllableOnsets, downsampleTrace, estimateSyllables, expectedSyllablesAt, extractFrameFeatures, featureStep, narrationSessionStep, nucleiStep, pacingStep, parseNarrationTimingJson, reanchorPacing, reanchorSession, recordSlideShown, resolveNarrationTimingGenerator, traceWordPosAt, transcriptWordKeys, vadStep, wordIndexAtChar, wordIndexAtTime, wordPosAtExpectedSyllables } from './narration/index.js';
 export { FenceRenderContext, FenceRenderer, FenceRendererMap, fenceRendererLangs } from './fence/index.js';
 export { P as PROOF_DIALECTS, a as ProofCategory, b as ProofDialect, c as ProofFinding, d as ProofRange, e as ProofSuggestion, f as ProofSuggestionKind } from './types-BQWu0gzi.js';
 export { BlankProtectedSpansOptions, BlankedText, DEFAULT_PROOF_SETTINGS, JoinedSegments, LINT_KIND_CATEGORIES, PROOF_FRONTMATTER_KEYS, PROOF_JOIN_SEPARATOR, ProofingSettings, SegmentSpan, blankProtectedSpans, buildJoinedText, categorizeLintKind, dropMaskedFindings, formatProofDictionary, mapJoinedSpanToSegment, parseProofDictionary, readProofingSettings, resolveProofDialect } from './proof/index.js';

package/dist/index.js

@@ -7,6 +7,7 @@ import {
   EMPTY_ADVANCE_LOG,
   advanceCoverage,
   alignNarration,
+  alignTranscript,
   bandpassRun,
   buildAdvanceTimingJson,
   createBandpass,
@@ -26,8 +27,9 @@ import {
   reanchorSession,
   recordSlideShown,
   traceWordPosAt,
+  transcriptWordKeys,
   vadStep
-} from "./chunk-FSZ6HB7S.js";
+} from "./chunk-B2MOGSWY.js";
 import {
   fenceRendererLangs
 } from "./chunk-PGI7HSWE.js";
@@ -230,7 +232,7 @@ import {
   validateMarkdownSource,
   wordRevealDelay,
   writeDashboardLayoutsToFrontmatter
-} from "./chunk-K75GTKSN.js";
+} from "./chunk-FGN4OZVK.js";
 import {
   buildNarrationTimingJson,
   parseNarrationTimingJson,
@@ -247,6 +249,7 @@ import {
   analyzeBlocks,
   applySavedPresentation,
   applyTransform,
+  collectPresentationMedia,
   compilePresentationPlan,
   createPresentationPlan,
   createTransformStyleRegistry,
@@ -262,7 +265,7 @@ import {
   serializePresentationPlan,
   transformNarratedSegment,
   usesDynamicPresentation
-} from "./chunk-F5WTRHWL.js";
+} from "./chunk-4O26IMVA.js";
 import {
   BASE_INPUT_DESCRIPTORS,
   BLOCK_MEDIA_LAYOUT_POLICIES,
@@ -386,7 +389,7 @@ import {
   wrapWithPersistentLayers,
   writeCustomTemplatesToFrontmatter,
   writeCustomThemesToFrontmatter
-} from "./chunk-ZF4NWRZU.js";
+} from "./chunk-VQXOX3YN.js";
 import {
   ASCII_CHAR_H,
   ASCII_CHAR_W,
@@ -441,7 +444,7 @@ import {
   treeFromMarkdownList,
   treeFromTemplateData,
   treeToTemplateData
-} from "./chunk-CZHWFQ7C.js";
+} from "./chunk-CSR6VCSJ.js";
 import {
   extractContent,
   mapElementToBlock,
@@ -830,6 +833,7 @@ export {
   adjustY,
   advanceCoverage,
   alignNarration,
+  alignTranscript,
   allocateTiming,
   analyzeBlocks,
   anchorPoint,
@@ -894,6 +898,7 @@ export {
   coalesceImageEditVersions,
   coalesceVersions,
   coerceTemplateParams,
+  collectPresentationMedia,
   comparisonBar,
   compilePresentationPlan,
   compileTheme,
@@ -1298,6 +1303,7 @@ export {
   topFeature,
   touch,
   traceWordPosAt,
+  transcriptWordKeys,
   transformNarratedSegment,
   transformOriginFor,
   transposeCells,

package/dist/narration/index.d.ts

@@ -568,6 +568,60 @@ interface AlignInput {
 }
 declare function alignNarration(input: AlignInput, config?: Partial<AlignConfig>): NarrationAlignment;
 
+/**
+ * Transcript alignment: word/block timestamps for a recorded take from a
+ * speech recognizer's timed transcript.
+ *
+ * `alignNarration` infers timing from the take's acoustics against the
+ * script's expected syllables, so it needs a reader who follows the script
+ * closely. This matches the WORDS a recognizer heard to the script's words
+ * instead, so it tolerates paraphrase, ad-libs, skipped passages and long
+ * silent demos.
+ *
+ * Pipeline: both sides become normalized word units → word trigrams that
+ * occur exactly once on each side anchor a monotone chain (longest increasing
+ * subsequence) → the short stretches between anchors are aligned with a small
+ * global DP that also accepts near-miss spellings → script words left without
+ * a match are placed on the unmatched words heard between their matched
+ * neighbors, so a passage the speaker skipped collapses onto the next thing
+ * they said, and a long pause or an off-script aside stays with the slide
+ * already showing. Block
+ * ranges come from the same contiguous-range builder the acoustic aligner
+ * uses, with the silences between heard words as its breath gaps.
+ *
+ * Pure and deterministic, like the rest of the narration engine.
+ */
+
+/** One recognized word, in seconds on the take's timeline. */
+interface TranscriptWord {
+    text: string;
+    startSec: number;
+    endSec: number;
+}
+/** One recognized phrase. Without `words`, its time is shared out by syllables. */
+interface TranscriptSegment {
+    text: string;
+    startSec: number;
+    endSec: number;
+    words?: readonly TranscriptWord[];
+}
+interface TranscriptAlignInput {
+    script: NarrationScript;
+    /** Recognized segments on the take's timeline (seconds from its start). */
+    segments: readonly TranscriptSegment[];
+    /** Take length in seconds (decoded-PCM length). */
+    durationSec: number;
+}
+interface TranscriptAlignment extends NarrationAlignment {
+    /** Spoken script words whose time came from a matched recognized word. */
+    matchedWords: number;
+    /** Spoken script words in total. */
+    spokenWords: number;
+}
+/** Lowercased letter/digit runs; apostrophes join ("you're" → "youre"). */
+declare function transcriptWordKeys(text: string): string[];
+declare function alignTranscript(input: TranscriptAlignInput, config?: Partial<AlignConfig>): TranscriptAlignment;
+
 /** Narration time range of one doc block at record time. */
 interface NarrationTimingBlock {
     /** Doc block id at record time (primary match key; ids can drift on heading edits). */
@@ -614,8 +668,11 @@ interface NarrationTimingJsonV3 {
  * - `'tts'` — a host synthesized the narration (text-to-speech). Block and
  *   sentence boundaries are exact because each was synthesized separately;
  *   word times are interpolated.
+ * - `'transcript-align'` — a speech recognizer transcribed the take and its
+ *   timed words were matched to the script's words. Paraphrase and skipped
+ *   passages are tolerated; unmatched words are interpolated.
  */
-type NarrationTimingMethod = 'dsp-align' | 'presenter-advance' | 'tts';
+type NarrationTimingMethod = 'dsp-align' | 'presenter-advance' | 'tts' | 'transcript-align';
 /** Provenance recorded in a sidecar's `generator` field. */
 interface NarrationTimingGenerator {
     /** Producer id, e.g. `'squisq-teleprompter'` or a host's TTS engine name. */
@@ -747,4 +804,4 @@ interface AdvanceCoverage {
 /** Summarize an advance log against a doc. Pure; drives the review-panel warning. */
 declare function advanceCoverage(doc: Doc, advances: SlideAdvanceLog): AdvanceCoverage;
 
-export { type AdvanceCoverage, type AlignConfig, type AlignInput, type BandpassState, type BuildAdvanceTimingOptions, type BuildNarrationTimingOptions, type BuildScriptOptions, DEFAULT_ALIGN_CONFIG, DEFAULT_FEATURE_CONFIG, DEFAULT_NUCLEI_CONFIG, DEFAULT_PACING_CONFIG, DEFAULT_VAD_CONFIG, EMPTY_ADVANCE_LOG, type FeatureConfig, type FeatureState, type FrameFeatures, type NarrationAlignment, type NarrationBlockRange, type NarrationScript, type NarrationSessionConfig, type NarrationSessionState, type NarrationTimingBlock, type NarrationTimingGenerator, type NarrationTimingJsonV3, type NarrationTimingMethod, type NarrationTrace, type NucleiConfig, type NucleiState, type PacingConfig, type PacingState, type PacingTick, type ScriptBlockRange, type ScriptToken, type SlideAdvance, type SlideAdvanceLog, type TraceSample, type VadConfig, type VadState, type WordTiming, advanceCoverage, alignNarration, bandpassRun, buildAdvanceTimingJson, buildNarrationScript, buildNarrationTimingJson, createBandpass, createFeatureState, createNarrationSession, createNucleiState, createPacingState, createVadState, detectSyllableOnsets, downsampleTrace, estimateSyllables, expectedSyllablesAt, extractFrameFeatures, featureStep, narrationSessionStep, nucleiStep, pacingStep, parseNarrationTimingJson, reanchorPacing, reanchorSession, recordSlideShown, resolveNarrationTimingGenerator, traceWordPosAt, vadStep, wordIndexAtChar, wordIndexAtTime, wordPosAtExpectedSyllables };
+export { type AdvanceCoverage, type AlignConfig, type AlignInput, type BandpassState, type BuildAdvanceTimingOptions, type BuildNarrationTimingOptions, type BuildScriptOptions, DEFAULT_ALIGN_CONFIG, DEFAULT_FEATURE_CONFIG, DEFAULT_NUCLEI_CONFIG, DEFAULT_PACING_CONFIG, DEFAULT_VAD_CONFIG, EMPTY_ADVANCE_LOG, type FeatureConfig, type FeatureState, type FrameFeatures, type NarrationAlignment, type NarrationBlockRange, type NarrationScript, type NarrationSessionConfig, type NarrationSessionState, type NarrationTimingBlock, type NarrationTimingGenerator, type NarrationTimingJsonV3, type NarrationTimingMethod, type NarrationTrace, type NucleiConfig, type NucleiState, type PacingConfig, type PacingState, type PacingTick, type ScriptBlockRange, type ScriptToken, type SlideAdvance, type SlideAdvanceLog, type TraceSample, type TranscriptAlignInput, type TranscriptAlignment, type TranscriptSegment, type TranscriptWord, type VadConfig, type VadState, type WordTiming, advanceCoverage, alignNarration, alignTranscript, bandpassRun, buildAdvanceTimingJson, buildNarrationScript, buildNarrationTimingJson, createBandpass, createFeatureState, createNarrationSession, createNucleiState, createPacingState, createVadState, detectSyllableOnsets, downsampleTrace, estimateSyllables, expectedSyllablesAt, extractFrameFeatures, featureStep, narrationSessionStep, nucleiStep, pacingStep, parseNarrationTimingJson, reanchorPacing, reanchorSession, recordSlideShown, resolveNarrationTimingGenerator, traceWordPosAt, transcriptWordKeys, vadStep, wordIndexAtChar, wordIndexAtTime, wordPosAtExpectedSyllables };

package/dist/narration/index.js

@@ -7,6 +7,7 @@ import {
   EMPTY_ADVANCE_LOG,
   advanceCoverage,
   alignNarration,
+  alignTranscript,
   bandpassRun,
   buildAdvanceTimingJson,
   createBandpass,
@@ -26,8 +27,9 @@ import {
   reanchorSession,
   recordSlideShown,
   traceWordPosAt,
+  transcriptWordKeys,
   vadStep
-} from "../chunk-FSZ6HB7S.js";
+} from "../chunk-B2MOGSWY.js";
 import "../chunk-NEWNNLWN.js";
 import {
   buildNarrationTimingJson,
@@ -41,8 +43,8 @@ import {
   wordIndexAtChar,
   wordIndexAtTime,
   wordPosAtExpectedSyllables
-} from "../chunk-ZF4NWRZU.js";
-import "../chunk-CZHWFQ7C.js";
+} from "../chunk-VQXOX3YN.js";
+import "../chunk-CSR6VCSJ.js";
 import "../chunk-FUBMT6SJ.js";
 import "../chunk-WNEH3BZP.js";
 import "../chunk-L7QZORLB.js";
@@ -60,6 +62,7 @@ export {
   EMPTY_ADVANCE_LOG,
   advanceCoverage,
   alignNarration,
+  alignTranscript,
   bandpassRun,
   buildAdvanceTimingJson,
   buildNarrationScript,
@@ -85,6 +88,7 @@ export {
   recordSlideShown,
   resolveNarrationTimingGenerator,
   traceWordPosAt,
+  transcriptWordKeys,
   vadStep,
   wordIndexAtChar,
   wordIndexAtTime,

package/dist/recommend/index.js

@@ -1,7 +1,7 @@
 import {
   profileBlockContents,
   recommendTemplatesForBlock
-} from "../chunk-CZHWFQ7C.js";
+} from "../chunk-CSR6VCSJ.js";
 import "../chunk-L7QZORLB.js";
 import "../chunk-KVEOD57Z.js";
 import "../chunk-P53Z4RYC.js";

package/dist/transform/index.d.ts

@@ -1,7 +1,7 @@
 import { R as Doc, s as ColorScheme, k as Block, bt as TemplateBlock } from '../Doc-BQUVDooL.js';
 import { i as PageTransformHints } from '../materializePageSection-_ftXlwvB.js';
 import { d as ExtractionType, E as ExtractedElement, b as ExtractionOptions } from '../contentExtractor-BNfVJV2U.js';
-import { f as BuildSpec } from '../types-Zk8qYuaT.js';
+import { f as BuildSpec, M as MarkdownBlockNode } from '../types-Zk8qYuaT.js';
 
 /** Identifier for a built-in or caller-registered transform style. */
 type TransformStyleId = string;
@@ -336,6 +336,8 @@ interface PresentationBeat {
     headline: string;
     points: string[];
     imageSrc?: string;
+    /** The section's videos and diagrams shown beside this slide's text (`presentationMedia`). */
+    media?: string[];
 }
 interface PresentationPlan {
     version: 1;
@@ -347,6 +349,7 @@ declare const PRESENTATION_LIMITS: {
     readonly sourceCharacters: 40000;
     readonly beats: 80;
     readonly serializedCharacters: 160000;
+    readonly mediaPerBeat: 4;
 };
 declare function parsePresentationPlan(input: unknown): PresentationPlan | null;
 declare function serializePresentationPlan(plan: PresentationPlan): string;
@@ -400,4 +403,22 @@ declare function createPresentationPlan(doc: Doc, hints?: PresentationHints): Pr
 declare function compilePresentationPlan(doc: Doc, plan: PresentationPlan): Doc;
 declare function applySavedPresentation(doc: Doc): Doc;
 
-export { type AnalyzedBlock, type BlockImage, DEFAULT_TRANSFORM_STYLE_ID, DYNAMIC_PRESENTATION_STYLE, type NarratedSegmentInput, type NarratedSegmentResult, type NarratedSegmentTransformOptions, type NarratedSegmentVideo, PRESENTATION_HINTS_KEY, PRESENTATION_KEY, PRESENTATION_LAYOUTS, PRESENTATION_LIMITS, type PresentationBeat, type PresentationBlockHint, type PresentationHints, type PresentationLayout, type PresentationPlan, type TransformImage, type TransformOptions, type TransformResult, type TransformStyleConfig, type TransformStyleId, type TransformStyleInput, type TransformStyleRegistry, type TransformStyleSummary, allocateTiming, analyzeBlocks, applySavedPresentation, applyTransform, compilePresentationPlan, createPresentationPlan, createTransformStyleRegistry, defaultPresentationHints, extractDocImages, getTransformStyleIds, getTransformStyleSummaries, parsePresentationHints, parsePresentationPlan, presentationSections, resolveTransformStyle, serializePresentationHints, serializePresentationPlan, transformNarratedSegment, usesDynamicPresentation };
+/** The videos and diagrams a narrated section embeds, which dynamic slides show
+ * beside its summaries. Images keep their own full-bleed slides; these are the
+ * visuals rich-media layout places next to slide text (`appendRichContentLayers`).
+ */
+
+interface PresentationMediaItem {
+    /** `<section block id>#<index>`, stable while the section's visuals are unchanged. */
+    ref: string;
+    kind: 'video' | 'diagram';
+    /** The body node exactly as authored; slides render it through rich-media layout. */
+    node: MarkdownBlockNode;
+    /** A diagram, or an HTML video that can resume mid-clip on a following slide. */
+    continuous: boolean;
+}
+/** Each section's visuals in source order, claimed like images: a nested
+ * block's visuals belong to the nearest narrated section at or before it. */
+declare function collectPresentationMedia(doc: Doc, sectionIds: string[]): Map<string, PresentationMediaItem[]>;
+
+export { type AnalyzedBlock, type BlockImage, DEFAULT_TRANSFORM_STYLE_ID, DYNAMIC_PRESENTATION_STYLE, type NarratedSegmentInput, type NarratedSegmentResult, type NarratedSegmentTransformOptions, type NarratedSegmentVideo, PRESENTATION_HINTS_KEY, PRESENTATION_KEY, PRESENTATION_LAYOUTS, PRESENTATION_LIMITS, type PresentationBeat, type PresentationBlockHint, type PresentationHints, type PresentationLayout, type PresentationMediaItem, type PresentationPlan, type TransformImage, type TransformOptions, type TransformResult, type TransformStyleConfig, type TransformStyleId, type TransformStyleInput, type TransformStyleRegistry, type TransformStyleSummary, allocateTiming, analyzeBlocks, applySavedPresentation, applyTransform, collectPresentationMedia, compilePresentationPlan, createPresentationPlan, createTransformStyleRegistry, defaultPresentationHints, extractDocImages, getTransformStyleIds, getTransformStyleSummaries, parsePresentationHints, parsePresentationPlan, presentationSections, resolveTransformStyle, serializePresentationHints, serializePresentationPlan, transformNarratedSegment, usesDynamicPresentation };

package/dist/transform/index.js

@@ -9,6 +9,7 @@ import {
   analyzeBlocks,
   applySavedPresentation,
   applyTransform,
+  collectPresentationMedia,
   compilePresentationPlan,
   createPresentationPlan,
   createTransformStyleRegistry,
@@ -24,9 +25,9 @@ import {
   serializePresentationPlan,
   transformNarratedSegment,
   usesDynamicPresentation
-} from "../chunk-F5WTRHWL.js";
-import "../chunk-ZF4NWRZU.js";
-import "../chunk-CZHWFQ7C.js";
+} from "../chunk-4O26IMVA.js";
+import "../chunk-VQXOX3YN.js";
+import "../chunk-CSR6VCSJ.js";
 import "../chunk-SVN7T6BK.js";
 import "../chunk-FUBMT6SJ.js";
 import "../chunk-WNEH3BZP.js";
@@ -49,6 +50,7 @@ export {
   analyzeBlocks,
   applySavedPresentation,
   applyTransform,
+  collectPresentationMedia,
   compilePresentationPlan,
   createPresentationPlan,
   createTransformStyleRegistry,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bendyline/squisq",
-  "version": "2.13.5",
+  "version": "2.13.6",
   "description": "Headless utilities for doc/block rendering, spatial math, Markdown, and storage",
   "license": "MIT",
   "author": "Bendyline",