@bendyline/squisq 1.3.0 → 1.4.1

package/dist/transform/index.d.ts

@@ -1,6 +1,175 @@
-export { applyTransform } from './applyTransform.js';
-export { resolveTransformStyle, getTransformStyleIds, getTransformStyleSummaries, DEFAULT_TRANSFORM_STYLE_ID, } from './registry.js';
-export type { TransformStyleId, TransformStyleSummary, TransformStyleConfig, TransformImage, TransformOptions, TransformResult, } from './types.js';
-export type { AnalyzedBlock, BlockImage } from './blockAnalyzer.js';
-export { analyzeBlocks, extractDocImages } from './blockAnalyzer.js';
-//# sourceMappingURL=index.d.ts.map
+import { j as ColorScheme, s as Doc, B as Block } from '../Doc-BeSeH2iX.js';
+import { d as ExtractionType, E as ExtractedElement, b as ExtractionOptions } from '../contentExtractor-BNfVJV2U.js';
+import '../types-BdZkdJ3z.js';
+
+/**
+ * Transform Types
+ *
+ * Defines the type system for doc block transforms — algorithms that take
+ * an existing Doc and produce a new Doc with blocks optimized for visual
+ * presentation. Transforms are read-only on the input and deterministic.
+ */
+
+/** Identifier for a registered transform style. */
+type TransformStyleId = string;
+/** Summary info for UI dropdowns (mirrors getThemeSummaries pattern). */
+interface TransformStyleSummary {
+    id: TransformStyleId;
+    name: string;
+    description: string;
+}
+/** Declarative configuration controlling how a style transforms content. */
+interface TransformStyleConfig {
+    /** Unique style identifier. */
+    id: TransformStyleId;
+    /** Human-readable name for UI display. */
+    name: string;
+    /** Short description for UI tooltips. */
+    description: string;
+    /**
+     * Minimum confidence for extracted elements to become template blocks (0–1).
+     * Higher values = fewer but higher-quality transformations.
+     */
+    minConfidence: number;
+    /**
+     * Maximum fraction of blocks that get transformed (0–1).
+     * 0.2 = light touch, 0.8 = aggressive restructuring.
+     */
+    transformRatio: number;
+    /** Which extraction types this style prefers, in priority order. */
+    preferredTypes: ExtractionType[];
+    /** Color schemes to rotate through on template blocks. */
+    colorSchemes: ColorScheme[];
+    /** Whether to insert sectionHeader blocks between groups. */
+    insertSectionHeaders: boolean;
+    /** Whether to interleave imageWithCaption blocks when images are available. */
+    interleaveImages: boolean;
+    /** Target density of template blocks per section. */
+    blocksPerSection: {
+        min: number;
+        max: number;
+    };
+    /** Transition style preference for transformed blocks. */
+    transitionStyle: 'cut' | 'fade' | 'dissolve' | 'mixed';
+}
+/** Image metadata for image interleaving. */
+interface TransformImage {
+    /** Image source path or URL. */
+    src: string;
+    /** Alt text for accessibility. */
+    alt?: string;
+    /** Photo credit. */
+    credit?: string;
+    /** License identifier. */
+    license?: string;
+}
+/** Options passed to `applyTransform()`. */
+interface TransformOptions {
+    /** Deterministic seed. Default: hash of doc.articleId. */
+    seed?: number;
+    /** Available images for image interleaving / accent images. */
+    images?: TransformImage[];
+    /** Theme ID to embed in the output Doc. */
+    themeId?: string;
+    /** Override specific style config values. */
+    overrides?: Partial<TransformStyleConfig>;
+}
+/** Result of applying a transform. */
+interface TransformResult {
+    /** The transformed Doc. Input doc is NOT mutated. */
+    doc: Doc;
+    /** Statistics about what changed. */
+    stats: {
+        totalInputBlocks: number;
+        transformedBlocks: number;
+        insertedBlocks: number;
+    };
+}
+
+/**
+ * Apply Transform
+ *
+ * Main entry point for the transform module. Takes a Doc and a style,
+ * analyzes block content, selects extractions to promote, and returns
+ * a new Doc with blocks optimized for visual presentation.
+ *
+ * The input Doc is never mutated — a fresh Doc is always returned.
+ */
+
+/**
+ * Apply a transform style to a Doc, producing a new Doc with blocks
+ * optimized for visual presentation.
+ *
+ * @param doc - The source Doc (not mutated).
+ * @param styleId - Registered transform style id (e.g. 'documentary', 'magazine').
+ * @param options - Optional seed, images, theme override.
+ * @returns TransformResult with the new Doc and stats.
+ */
+declare function applyTransform(doc: Doc, styleId: TransformStyleId, options?: TransformOptions): TransformResult;
+
+/**
+ * Transform Style Registry
+ *
+ * Central registry of built-in transform styles. Follows the same pattern
+ * as themeLibrary.ts — a record of configs with lookup/summary helpers.
+ */
+
+/** Default style used when no id is provided. */
+declare const DEFAULT_TRANSFORM_STYLE_ID = "documentary";
+/**
+ * Resolve a transform style by id.
+ * Returns the default style if the id is not found.
+ */
+declare function resolveTransformStyle(id: string): TransformStyleConfig;
+/** Get all registered style ids. */
+declare function getTransformStyleIds(): string[];
+/** Get summary info for all styles (for UI dropdowns). */
+declare function getTransformStyleSummaries(): TransformStyleSummary[];
+
+/**
+ * Block Analyzer
+ *
+ * Walks Doc blocks, extracts plain text from their markdown contents,
+ * and runs the content extractor on each block to identify stats, dates,
+ * quotes, etc. that can be promoted to visual template blocks.
+ */
+
+/** An image found in a block's markdown contents. */
+interface BlockImage {
+    /** Image URL (relative or absolute). */
+    src: string;
+    /** Alt text. */
+    alt: string;
+    /** Index of the source block in the analyzed array. */
+    sourceBlockIndex: number;
+}
+/** A block enriched with extraction analysis. */
+interface AnalyzedBlock {
+    /** The original block. */
+    block: Block;
+    /** Plain text extracted from the block's contents. */
+    plainText: string;
+    /** Content extractions found in the block's text. */
+    extractions: ExtractedElement[];
+    /** Whether this block has child blocks. */
+    hasChildren: boolean;
+    /** Word count of the block's body text. */
+    bodyWordCount: number;
+    /** Images found in this block's markdown contents. */
+    images: BlockImage[];
+}
+/**
+ * Collect all unique images from all blocks in a Doc.
+ * Returns them as TransformImages for use in the transform pipeline.
+ */
+declare function extractDocImages(blocks: Block[]): TransformImage[];
+/**
+ * Analyze all blocks in a Doc, extracting text and running content extraction.
+ *
+ * @param blocks - The Doc's block array (may be nested via children).
+ * @param options - Extraction options (minConfidence, types filter).
+ * @returns Analyzed blocks with extractions, in flattened order.
+ */
+declare function analyzeBlocks(blocks: Block[], options?: ExtractionOptions): AnalyzedBlock[];
+
+export { type AnalyzedBlock, type BlockImage, DEFAULT_TRANSFORM_STYLE_ID, type TransformImage, type TransformOptions, type TransformResult, type TransformStyleConfig, type TransformStyleId, type TransformStyleSummary, analyzeBlocks, applyTransform, extractDocImages, getTransformStyleIds, getTransformStyleSummaries, resolveTransformStyle };

package/dist/transform/index.js

@@ -1,4 +1,491 @@
-export { applyTransform } from './applyTransform.js';
-export { resolveTransformStyle, getTransformStyleIds, getTransformStyleSummaries, DEFAULT_TRANSFORM_STYLE_ID, } from './registry.js';
-export { analyzeBlocks, extractDocImages } from './blockAnalyzer.js';
+import {
+  extractContent,
+  mapElementToBlock,
+  stripMarkdown
+} from "../chunk-FR2RBTKO.js";
+import {
+  SeededRandom,
+  hashString
+} from "../chunk-FVTQSNR7.js";
+import {
+  isTemplateBlock
+} from "../chunk-5WFRKURX.js";
+import {
+  extractPlainText,
+  getChildren
+} from "../chunk-H3AXU4MJ.js";
+
+// src/transform/styles/documentary.ts
+var documentaryStyle = {
+  id: "documentary",
+  name: "Documentary",
+  description: "Conservative \u2014 highlights only the strongest stats, dates, and quotes",
+  minConfidence: 0.5,
+  transformRatio: 0.3,
+  preferredTypes: ["stat", "date", "quote", "fact"],
+  colorSchemes: ["blue", "green", "purple"],
+  insertSectionHeaders: false,
+  interleaveImages: true,
+  blocksPerSection: { min: 1, max: 3 },
+  transitionStyle: "fade"
+};
+
+// src/transform/styles/magazine.ts
+var magazineStyle = {
+  id: "magazine",
+  name: "Magazine",
+  description: "Visual variety with pull quotes, facts, and rich image layouts",
+  minConfidence: 0.4,
+  transformRatio: 0.6,
+  preferredTypes: ["quote", "fact", "list", "definition", "stat", "impactLine"],
+  colorSchemes: ["blue", "green", "purple", "orange", "red"],
+  insertSectionHeaders: true,
+  interleaveImages: true,
+  blocksPerSection: { min: 2, max: 5 },
+  transitionStyle: "dissolve"
+};
+
+// src/transform/styles/dataDriven.ts
+var dataDrivenStyle = {
+  id: "data-driven",
+  name: "Data-Driven",
+  description: "Aggressively highlights statistics, comparisons, and timelines",
+  minConfidence: 0.3,
+  transformRatio: 0.8,
+  preferredTypes: ["stat", "comparison", "date", "list", "fact"],
+  colorSchemes: ["blue", "orange", "green", "purple", "red"],
+  insertSectionHeaders: true,
+  interleaveImages: false,
+  blocksPerSection: { min: 2, max: 6 },
+  transitionStyle: "cut"
+};
+
+// src/transform/styles/narrative.ts
+var narrativeStyle = {
+  id: "narrative",
+  name: "Narrative",
+  description: "Story-driven with timeline events, quotes, and impact moments",
+  minConfidence: 0.4,
+  transformRatio: 0.5,
+  preferredTypes: ["date", "quote", "impactLine", "fact", "stat"],
+  colorSchemes: ["blue", "purple", "green"],
+  insertSectionHeaders: true,
+  interleaveImages: true,
+  blocksPerSection: { min: 1, max: 4 },
+  transitionStyle: "fade"
+};
+
+// src/transform/styles/minimal.ts
+var minimalStyle = {
+  id: "minimal",
+  name: "Minimal",
+  description: "Light touch \u2014 only the strongest highlights, mostly text",
+  minConfidence: 0.6,
+  transformRatio: 0.2,
+  preferredTypes: ["stat", "quote"],
+  colorSchemes: ["blue", "green"],
+  insertSectionHeaders: false,
+  interleaveImages: false,
+  blocksPerSection: { min: 1, max: 2 },
+  transitionStyle: "fade"
+};
+
+// src/transform/registry.ts
+var TRANSFORM_STYLES = {
+  [documentaryStyle.id]: documentaryStyle,
+  [magazineStyle.id]: magazineStyle,
+  [dataDrivenStyle.id]: dataDrivenStyle,
+  [narrativeStyle.id]: narrativeStyle,
+  [minimalStyle.id]: minimalStyle
+};
+var DEFAULT_TRANSFORM_STYLE_ID = "documentary";
+function resolveTransformStyle(id) {
+  return TRANSFORM_STYLES[id] ?? TRANSFORM_STYLES[DEFAULT_TRANSFORM_STYLE_ID];
+}
+function getTransformStyleIds() {
+  return Object.keys(TRANSFORM_STYLES);
+}
+function getTransformStyleSummaries() {
+  return Object.values(TRANSFORM_STYLES).map(({ id, name, description }) => ({
+    id,
+    name,
+    description
+  }));
+}
+
+// src/transform/blockAnalyzer.ts
+function contentsToPlainText(contents) {
+  return contents.map((node) => extractPlainText(node)).join("\n");
+}
+function collectImages(node) {
+  const results = [];
+  if ("type" in node && node.type === "image" && "url" in node) {
+    const img = node;
+    if (img.url) {
+      results.push({ src: img.url, alt: img.alt ?? "" });
+    }
+  }
+  for (const child of getChildren(node)) {
+    results.push(...collectImages(child));
+  }
+  return results;
+}
+function extractBlockImages(block, blockIndex) {
+  if (!block.contents || block.contents.length === 0) return [];
+  const images = [];
+  for (const node of block.contents) {
+    for (const img of collectImages(node)) {
+      images.push({ ...img, sourceBlockIndex: blockIndex });
+    }
+  }
+  return images;
+}
+function extractDocImages(blocks) {
+  const seen = /* @__PURE__ */ new Set();
+  const images = [];
+  function walk(blockList) {
+    for (const block of blockList) {
+      if (block.contents) {
+        for (const node of block.contents) {
+          for (const img of collectImages(node)) {
+            if (!seen.has(img.src)) {
+              seen.add(img.src);
+              images.push({ src: img.src, alt: img.alt });
+            }
+          }
+        }
+      }
+      if (block.children) {
+        walk(block.children);
+      }
+    }
+  }
+  walk(blocks);
+  return images;
+}
+function collectLeafBlocks(blocks) {
+  const result = [];
+  for (const block of blocks) {
+    if (block.children && block.children.length > 0) {
+      if (!block.contents || block.contents.length === 0) {
+        result.push(block);
+      }
+      result.push(...collectLeafBlocks(block.children));
+    } else {
+      result.push(block);
+    }
+  }
+  return result;
+}
+function analyzeBlocks(blocks, options) {
+  const flat = collectLeafBlocks(blocks);
+  const results = [];
+  for (const block of flat) {
+    let plainText = "";
+    if (block.contents && block.contents.length > 0) {
+      plainText = contentsToPlainText(block.contents);
+    } else if (block.title) {
+      plainText = block.title;
+    }
+    const stripped = stripMarkdown(plainText);
+    const bodyWordCount = stripped.split(/\s+/).filter((w) => w.length > 0).length;
+    const extraction = extractContent(stripped, options);
+    const images = extractBlockImages(block, results.length);
+    results.push({
+      block,
+      plainText: stripped,
+      extractions: extraction.elements,
+      hasChildren: !!(block.children && block.children.length > 0),
+      bodyWordCount,
+      images
+    });
+  }
+  return results;
+}
+
+// src/transform/templateSelector.ts
+var ACCENT_POSITIONS = [
+  "left-strip",
+  "right-strip",
+  "bottom-strip",
+  "corner-inset"
+];
+var AMBIENT_MOTIONS = [
+  "zoomIn",
+  "zoomOut",
+  "panLeft",
+  "panRight"
+];
+function selectAndBuild(analyzed, config, images, seed) {
+  const rng = new SeededRandom(seed);
+  const candidates = collectCandidates(analyzed, config);
+  const maxTransforms = Math.max(1, Math.floor(analyzed.length * config.transformRatio));
+  const selected = candidates.slice(0, maxTransforms);
+  const result = buildBlockSequence(analyzed, selected, config, images, rng);
+  return result;
+}
+function collectCandidates(analyzed, config) {
+  const candidates = [];
+  for (let i = 0; i < analyzed.length; i++) {
+    const ab = analyzed[i];
+    if (ab.bodyWordCount < 5) continue;
+    for (const element of ab.extractions) {
+      if (element.confidence < config.minConfidence) continue;
+      candidates.push({
+        element,
+        sourceBlock: ab.block,
+        sourceIndex: i
+      });
+    }
+  }
+  const typeRank = new Map(config.preferredTypes.map((t, i) => [t, i]));
+  const maxRank = config.preferredTypes.length;
+  candidates.sort((a, b) => {
+    const rankA = typeRank.get(a.element.type) ?? maxRank;
+    const rankB = typeRank.get(b.element.type) ?? maxRank;
+    if (rankA !== rankB) return rankA - rankB;
+    return b.element.confidence - a.element.confidence;
+  });
+  const seenTexts = /* @__PURE__ */ new Set();
+  const deduplicated = [];
+  for (const candidate of candidates) {
+    const key = candidate.element.text.trim().toLowerCase();
+    if (seenTexts.has(key)) continue;
+    seenTexts.add(key);
+    deduplicated.push(candidate);
+  }
+  return deduplicated;
+}
+function buildBlockSequence(analyzed, selected, config, images, rng) {
+  const blockExtractions = /* @__PURE__ */ new Map();
+  for (const sel of selected) {
+    const list = blockExtractions.get(sel.sourceIndex) ?? [];
+    list.push(sel);
+    blockExtractions.set(sel.sourceIndex, list);
+  }
+  for (const [idx, list] of blockExtractions) {
+    const diverse = [];
+    const usedTypes = /* @__PURE__ */ new Set();
+    for (const sel of list) {
+      if (usedTypes.has(sel.element.type)) continue;
+      usedTypes.add(sel.element.type);
+      diverse.push(sel);
+      if (diverse.length >= config.blocksPerSection.max) break;
+    }
+    blockExtractions.set(idx, diverse);
+  }
+  const blocks = [];
+  let transformedCount = 0;
+  let insertedCount = 0;
+  let colorIndex = 0;
+  let blockIdCounter = 0;
+  let imageIndex = 0;
+  let accentPositionIndex = 0;
+  let elementsSinceImage = 0;
+  let prevTemplateType = "";
+  const totalTransformed = Array.from(blockExtractions.values()).reduce(
+    (sum, list) => sum + list.length,
+    0
+  );
+  const imageInterval = config.interleaveImages && images.length > 0 && totalTransformed > 0 ? Math.max(2, Math.floor(totalTransformed / (images.length + 1))) : Infinity;
+  for (let i = 0; i < analyzed.length; i++) {
+    const ab = analyzed[i];
+    const extractions = blockExtractions.get(i);
+    if (!extractions || extractions.length === 0) {
+      blocks.push(ab.block);
+      continue;
+    }
+    if (config.insertSectionHeaders && ab.block.title) {
+      const headerBlock = {
+        template: "sectionHeader",
+        id: `transform-header-${blockIdCounter++}`,
+        duration: ab.block.duration > 0 ? Math.min(3, ab.block.duration * 0.2) : 3,
+        audioSegment: ab.block.audioSegment,
+        title: ab.block.title,
+        colorScheme: config.colorSchemes[colorIndex % config.colorSchemes.length]
+      };
+      blocks.push(headerBlock);
+      insertedCount++;
+    }
+    for (const sel of extractions) {
+      const colorScheme = config.colorSchemes[colorIndex % config.colorSchemes.length];
+      let accentImage;
+      if (images.length > 0 && config.interleaveImages) {
+        const img = images[imageIndex % images.length];
+        accentImage = {
+          src: img.src,
+          alt: img.alt ?? "",
+          position: ACCENT_POSITIONS[accentPositionIndex % ACCENT_POSITIONS.length],
+          ambientMotion: rng.pick(AMBIENT_MOTIONS) ?? "zoomIn",
+          credit: img.credit,
+          license: img.license
+        };
+        accentPositionIndex++;
+      }
+      const templateBlock = mapElementToBlock(sel.element, {
+        id: `transform-${blockIdCounter++}`,
+        duration: ab.block.duration > 0 ? ab.block.duration / extractions.length : 6,
+        audioSegment: ab.block.audioSegment,
+        colorScheme,
+        accentImage,
+        sourceStartTime: sel.element.sourcePosition
+      });
+      if (config.transitionStyle !== "cut") {
+        templateBlock.transition = {
+          type: config.transitionStyle === "mixed" ? rng.pick(["fade", "dissolve"]) ?? "fade" : config.transitionStyle,
+          duration: 0.5
+        };
+      }
+      if (templateBlock.template === prevTemplateType) {
+        colorIndex++;
+      }
+      prevTemplateType = templateBlock.template;
+      blocks.push(templateBlock);
+      transformedCount++;
+      colorIndex++;
+      elementsSinceImage++;
+      if (config.interleaveImages && imageIndex < images.length && elementsSinceImage >= imageInterval) {
+        const img = images[imageIndex];
+        const imageBlock = {
+          template: "imageWithCaption",
+          id: `transform-img-${blockIdCounter++}`,
+          duration: 4,
+          audioSegment: ab.block.audioSegment,
+          imageSrc: img.src,
+          imageAlt: img.alt ?? "",
+          ambientMotion: rng.pick(AMBIENT_MOTIONS) ?? "zoomIn"
+        };
+        blocks.push(imageBlock);
+        imageIndex++;
+        insertedCount++;
+        elementsSinceImage = 0;
+      }
+    }
+  }
+  return { blocks, transformedCount, insertedCount };
+}
+
+// src/transform/timingAllocator.ts
+var MIN_BLOCK_DURATION = 3;
+var MAX_BLOCK_DURATION = 20;
+var DEFAULT_BLOCK_DURATION = 6;
+function allocateTiming(blocks, totalDuration) {
+  if (blocks.length === 0) return [];
+  const effectiveDuration = totalDuration > 0 ? totalDuration : estimateDefaultDuration(blocks);
+  const durations = blocks.map((block) => {
+    if (!isTemplateBlock(block) && block.duration > 0) {
+      return block.duration;
+    }
+    if (isTemplateBlock(block) && block.duration > 0) {
+      return block.duration;
+    }
+    return DEFAULT_BLOCK_DURATION;
+  });
+  const rawTotal = durations.reduce((sum, d) => sum + d, 0);
+  if (rawTotal > 0 && effectiveDuration > 0) {
+    const scale = effectiveDuration / rawTotal;
+    for (let i = 0; i < durations.length; i++) {
+      durations[i] = clampDuration(durations[i] * scale);
+    }
+  }
+  let currentTime = 0;
+  const result = [];
+  for (let i = 0; i < blocks.length; i++) {
+    const block = blocks[i];
+    const duration = durations[i];
+    if (isTemplateBlock(block)) {
+      const outputBlock = {
+        ...block,
+        startTime: currentTime,
+        duration,
+        title: getTemplateTitle(block)
+      };
+      result.push(outputBlock);
+    } else {
+      result.push({
+        ...block,
+        startTime: currentTime,
+        duration
+      });
+    }
+    currentTime += duration;
+  }
+  return result;
+}
+function clampDuration(d) {
+  return Math.max(MIN_BLOCK_DURATION, Math.min(MAX_BLOCK_DURATION, d));
+}
+function estimateDefaultDuration(blocks) {
+  let totalSeconds = 0;
+  for (const block of blocks) {
+    if (!isTemplateBlock(block) && block.duration > 0) {
+      totalSeconds += block.duration;
+    } else {
+      totalSeconds += DEFAULT_BLOCK_DURATION;
+    }
+  }
+  return Math.max(totalSeconds, blocks.length * MIN_BLOCK_DURATION);
+}
+function getTemplateTitle(block) {
+  if ("title" in block && typeof block.title === "string") {
+    return block.title;
+  }
+  if ("stat" in block && typeof block.stat === "string") {
+    return block.stat;
+  }
+  if ("fact" in block && typeof block.fact === "string") {
+    return block.fact;
+  }
+  if ("quote" in block && typeof block.quote === "string") {
+    return block.quote.slice(0, 60);
+  }
+  return void 0;
+}
+
+// src/transform/applyTransform.ts
+function applyTransform(doc, styleId, options) {
+  const baseConfig = resolveTransformStyle(styleId);
+  const config = options?.overrides ? { ...baseConfig, ...options.overrides, id: baseConfig.id } : baseConfig;
+  const seed = options?.seed ?? hashString(doc.articleId || "transform");
+  const images = options?.images ?? extractDocImages(doc.blocks);
+  const analyzed = analyzeBlocks(doc.blocks, {
+    minConfidence: config.minConfidence,
+    types: config.preferredTypes
+  });
+  const hasContent = analyzed.some((ab) => ab.bodyWordCount >= 5);
+  if (!hasContent) {
+    return {
+      doc: { ...doc },
+      stats: {
+        totalInputBlocks: doc.blocks.length,
+        transformedBlocks: 0,
+        insertedBlocks: 0
+      }
+    };
+  }
+  const selection = selectAndBuild(analyzed, config, images, seed);
+  const timedBlocks = allocateTiming(selection.blocks, doc.duration);
+  const transformedDoc = {
+    ...doc,
+    blocks: timedBlocks,
+    themeId: options?.themeId ?? doc.themeId
+  };
+  return {
+    doc: transformedDoc,
+    stats: {
+      totalInputBlocks: doc.blocks.length,
+      transformedBlocks: selection.transformedCount,
+      insertedBlocks: selection.insertedCount
+    }
+  };
+}
+export {
+  DEFAULT_TRANSFORM_STYLE_ID,
+  analyzeBlocks,
+  applyTransform,
+  extractDocImages,
+  getTransformStyleIds,
+  getTransformStyleSummaries,
+  resolveTransformStyle
+};
 //# sourceMappingURL=index.js.map