dravoice 0.1.2 → 0.1.4

package/src/v2/analyzers/rhetorical-shape.js

@@ -1,59 +1,65 @@
-import { evidenceTypes } from "./evidence.js";
-import { transitionLabel } from "./discourse.js";
-import { topItems } from "../text-utils.js";
-
-export function analyzeRhetoricalShape(documents) {
-  const documentMoves = documents.map((document) => document.sentences.map((sentence) => moveFor(sentence.text)));
-  const sentenceMoves = documentMoves.flat();
-  const openingMoves = documents.flatMap((document) => document.sentences.slice(0, 3).map((sentence) => moveFor(sentence.text)));
-  const bigrams = [];
-  const trigrams = [];
-  const openingMovePatterns = [];
-  for (const moves of documentMoves) {
-    if (moves.length > 0) {
-      openingMovePatterns.push(moves.slice(0, 3).join(" -> "));
-    }
-    for (let index = 0; index < moves.length - 1; index += 1) {
-      bigrams.push(`${moves[index]} -> ${moves[index + 1]}`);
-    }
-    for (let index = 0; index < moves.length - 2; index += 1) {
-      trigrams.push(`${moves[index]} -> ${moves[index + 1]} -> ${moves[index + 2]}`);
-    }
-  }
-
-  return {
-    family: "rhetoricalShape",
-    confidence: sentenceMoves.length >= 12 ? "medium" : "low",
-    features: {
-      moveRates: topItems(sentenceMoves, 12),
-      openingMoves: openingMoves.slice(0, 9),
-      openingMovePatterns: topItems(openingMovePatterns, 8),
-      moveBigrams: topItems(bigrams, 12),
-      moveTrigrams: topItems(trigrams, 12),
-      commonSequences: topItems(bigrams, 12),
-    },
-    examples: openingMoves.slice(0, 5),
-    warnings: sentenceMoves.length < 12 ? ["Rhetorical-shape confidence is limited because the corpus has fewer than 12 sentences."] : [],
-    revisionHandles: ["Compare the draft's opening and move sequence against observed source patterns."],
-  };
-}
-
-export function moveFor(text) {
-  if (evidenceTypes(text).length > 0) {
-    return "scene-or-evidence";
-  }
-  const transition = transitionLabel(text);
-  if (transition === "contrast") {
-    return "turn";
-  }
-  if (transition === "causal" || transition === "conclusion") {
-    return "implication";
-  }
-  if (/\b(should|must|rule|lesson|start|fix|avoid|keep)\b/i.test(text)) {
-    return "advice";
-  }
-  if (/\b(always|never|everyone|everything|important|best|better|worse|should|must|all|none|every)\b/i.test(text)) {
-    return "abstract-claim";
-  }
-  return "statement";
-}
+import { evidenceTypes } from "./evidence.js";
+import { transitionLabel } from "./discourse.js";
+import { splitSentences, topItems } from "../text-utils.js";
+
+export function analyzeRhetoricalShape(documents) {
+  const documentMoves = documents.map((document) => document.sentences.map((sentence) => moveFor(sentence.text)));
+  const sentenceMoves = documentMoves.flat();
+  const openingMoves = documents.flatMap((document) => document.sentences.slice(0, 3).map((sentence) => moveFor(sentence.text)));
+  const paragraphMovePatterns = documents.flatMap((document) =>
+    document.paragraphs
+      .map((paragraph) => splitSentences(paragraph.text).map((sentence) => moveFor(sentence)).join(" -> "))
+      .filter(Boolean)
+  );
+  const bigrams = [];
+  const trigrams = [];
+  const openingMovePatterns = [];
+  for (const moves of documentMoves) {
+    if (moves.length > 0) {
+      openingMovePatterns.push(moves.slice(0, 3).join(" -> "));
+    }
+    for (let index = 0; index < moves.length - 1; index += 1) {
+      bigrams.push(`${moves[index]} -> ${moves[index + 1]}`);
+    }
+    for (let index = 0; index < moves.length - 2; index += 1) {
+      trigrams.push(`${moves[index]} -> ${moves[index + 1]} -> ${moves[index + 2]}`);
+    }
+  }
+
+  return {
+    family: "rhetoricalShape",
+    confidence: sentenceMoves.length >= 12 ? "medium" : "low",
+    features: {
+      moveRates: topItems(sentenceMoves, 12),
+      openingMoves: openingMoves.slice(0, 9),
+      openingMovePatterns: topItems(openingMovePatterns, 8),
+      paragraphMovePatterns: topItems(paragraphMovePatterns, 12),
+      moveBigrams: topItems(bigrams, 12),
+      moveTrigrams: topItems(trigrams, 12),
+      commonSequences: topItems(bigrams, 12),
+    },
+    examples: openingMoves.slice(0, 5),
+    warnings: sentenceMoves.length < 12 ? ["Rhetorical-shape confidence is limited because the corpus has fewer than 12 sentences."] : [],
+    revisionHandles: ["Compare the draft's opening and move sequence against observed source patterns."],
+  };
+}
+
+export function moveFor(text) {
+  if (evidenceTypes(text).length > 0) {
+    return "scene-or-evidence";
+  }
+  const transition = transitionLabel(text);
+  if (transition === "contrast") {
+    return "turn";
+  }
+  if (transition === "causal" || transition === "conclusion") {
+    return "implication";
+  }
+  if (/\b(should|must|rule|lesson|start|fix|avoid|keep)\b/i.test(text)) {
+    return "advice";
+  }
+  if (/\b(always|never|everyone|everything|important|best|better|worse|should|must|all|none|every)\b/i.test(text)) {
+    return "abstract-claim";
+  }
+  return "statement";
+}

package/src/v2/analyzers/rhythm.js

@@ -1,47 +1,39 @@
-import { distribution, rate } from "../text-utils.js";
-
-export function analyzeRhythm(documents) {
-  const sentences = documents.flatMap((document) => document.sentences);
-  const paragraphs = documents.flatMap((document) => document.paragraphs);
-  const blocks = documents.flatMap((document) => document.blocks);
-
-  return {
-    family: "rhythm",
-    confidence: confidenceFor(sentences.length),
-    features: {
-      documentCount: documents.length,
-      sentenceCount: sentences.length,
-      paragraphCount: paragraphs.length,
-      sentenceWords: distribution(sentences.map((sentence) => sentence.tokens.length)),
-      paragraphWords: distribution(paragraphs.map((paragraph) => paragraph.text.split(/\s+/).filter(Boolean).length)),
-      paragraphSentences: distribution(paragraphs.map((paragraph) => countParagraphSentences(sentences, paragraph))),
-      headingDensity: rate(documents.reduce((sum, document) => sum + document.headings.length, 0), sentences.length, 2),
-      listDensity: rate(blocks.filter((block) => block.type === "list").length, paragraphs.length, 2),
-      quoteDensity: rate(blocks.filter((block) => block.type === "quote").length, paragraphs.length, 2),
-    },
-    examples: [
-      `sentenceWords.median: ${distribution(sentences.map((sentence) => sentence.tokens.length)).median}`,
-      `paragraphWords.median: ${distribution(paragraphs.map((paragraph) => paragraph.text.split(/\s+/).filter(Boolean).length)).median}`,
-    ],
-    warnings: sentences.length < 12 ? ["Rhythm confidence is limited because the corpus has fewer than 12 usable sentences."] : [],
-    revisionHandles: ["Compare sentence and paragraph pacing against the learned range."],
-  };
-}
-
-function countParagraphSentences(sentences, paragraph) {
-  return Math.max(1, sentences.filter((sentence) =>
-    sentence.line >= paragraph.line &&
-    sentence.line < paragraph.line + 8 &&
-    sentence.headingId === paragraph.headingId
-  ).length);
-}
-
-function confidenceFor(sentenceCount) {
-  if (sentenceCount >= 30) {
-    return "high";
-  }
-  if (sentenceCount >= 8) {
-    return "medium";
-  }
-  return "low";
-}
+import { distribution, rate, splitSentences } from "../text-utils.js";
+
+export function analyzeRhythm(documents) {
+  const sentences = documents.flatMap((document) => document.sentences);
+  const paragraphs = documents.flatMap((document) => document.paragraphs);
+  const blocks = documents.flatMap((document) => document.blocks);
+
+  return {
+    family: "rhythm",
+    confidence: confidenceFor(sentences.length),
+    features: {
+      documentCount: documents.length,
+      sentenceCount: sentences.length,
+      paragraphCount: paragraphs.length,
+      sentenceWords: distribution(sentences.map((sentence) => sentence.tokens.length)),
+      paragraphWords: distribution(paragraphs.map((paragraph) => paragraph.text.split(/\s+/).filter(Boolean).length)),
+      paragraphSentences: distribution(paragraphs.map((paragraph) => Math.max(1, splitSentences(paragraph.text).length))),
+      headingDensity: rate(documents.reduce((sum, document) => sum + document.headings.length, 0), sentences.length, 2),
+      listDensity: rate(blocks.filter((block) => block.type === "list").length, paragraphs.length, 2),
+      quoteDensity: rate(blocks.filter((block) => block.type === "quote").length, paragraphs.length, 2),
+    },
+    examples: [
+      `sentenceWords.median: ${distribution(sentences.map((sentence) => sentence.tokens.length)).median}`,
+      `paragraphWords.median: ${distribution(paragraphs.map((paragraph) => paragraph.text.split(/\s+/).filter(Boolean).length)).median}`,
+    ],
+    warnings: sentences.length < 12 ? ["Rhythm confidence is limited because the corpus has fewer than 12 usable sentences."] : [],
+    revisionHandles: ["Compare sentence and paragraph pacing against the learned range."],
+  };
+}
+
+function confidenceFor(sentenceCount) {
+  if (sentenceCount >= 30) {
+    return "high";
+  }
+  if (sentenceCount >= 8) {
+    return "medium";
+  }
+  return "low";
+}

package/src/v2/analyzers/structure.js

@@ -1,24 +1,41 @@
-import { distribution, rate } from "../text-utils.js";
-import { moveFor } from "./rhetorical-shape.js";
-
-export function analyzeStructure(documents) {
-  const sectionLengths = documents.flatMap((document) =>
-    document.sections.map((section) => section.blocks.reduce((sum, block) => sum + block.lines.join(" ").split(/\s+/).filter(Boolean).length, 0))
-  );
-  const openingMoves = documents.flatMap((document) => document.sentences.slice(0, 2).map((sentence) => moveFor(sentence.text)));
-
-  return {
-    family: "structure",
-    confidence: documents.length >= 3 ? "medium" : "low",
-    features: {
-      sectionWords: distribution(sectionLengths),
-      headingCount: distribution(documents.map((document) => document.headings.length)),
-      openingMoves,
-      listDocumentRate: rate(documents.filter((document) => document.blocks.some((block) => block.type === "list")).length, documents.length, 2),
-      quoteDocumentRate: rate(documents.filter((document) => document.blocks.some((block) => block.type === "quote")).length, documents.length, 2),
-    },
-    examples: openingMoves.slice(0, 5),
-    warnings: documents.length < 3 ? ["Structure confidence is limited because the corpus has fewer than 3 documents."] : [],
-    revisionHandles: ["Compare headings, list/quote use, section size, and opening structure."],
-  };
-}
+import { distribution, rate, topItems } from "../text-utils.js";
+import { moveFor } from "./rhetorical-shape.js";
+
+export function analyzeStructure(documents) {
+  const sectionLengths = documents.flatMap((document) =>
+    document.sections.map((section) => section.blocks.reduce((sum, block) => sum + block.lines.join(" ").split(/\s+/).filter(Boolean).length, 0))
+  );
+  const openingMoves = documents.flatMap((document) => document.sentences.slice(0, 2).map((sentence) => moveFor(sentence.text)));
+  const sectionOrderPatterns = documents
+    .map((document) => document.headings.map((heading) => `h${heading.depth}`).join(" -> "))
+    .filter(Boolean);
+  const listPlacementPatterns = documents.flatMap((document) =>
+    document.blocks
+      .filter((block) => block.type === "list")
+      .map((block) => `h${block.headingDepth || 0}:list`)
+  );
+  const quotePlacementPatterns = documents.flatMap((document) =>
+    document.blocks
+      .filter((block) => block.type === "quote")
+      .map((block) => `h${block.headingDepth || 0}:quote`)
+  );
+
+  return {
+    family: "structure",
+    confidence: documents.length >= 3 ? "medium" : "low",
+    features: {
+      sectionWords: distribution(sectionLengths),
+      headingCount: distribution(documents.map((document) => document.headings.length)),
+      maxHeadingDepth: distribution(documents.map((document) => Math.max(0, ...document.headings.map((heading) => heading.depth)))),
+      sectionOrderPatterns: topItems(sectionOrderPatterns, 12),
+      listPlacementPatterns: topItems(listPlacementPatterns, 12),
+      quotePlacementPatterns: topItems(quotePlacementPatterns, 12),
+      openingMoves,
+      listDocumentRate: rate(documents.filter((document) => document.blocks.some((block) => block.type === "list")).length, documents.length, 2),
+      quoteDocumentRate: rate(documents.filter((document) => document.blocks.some((block) => block.type === "quote")).length, documents.length, 2),
+    },
+    examples: openingMoves.slice(0, 5),
+    warnings: documents.length < 3 ? ["Structure confidence is limited because the corpus has fewer than 3 documents."] : [],
+    revisionHandles: ["Compare headings, list/quote use, section size, and opening structure."],
+  };
+}