dravoice 0.1.0 → 0.1.1

package/src/v2/brief.js

@@ -1,146 +1,146 @@
-import fs from "node:fs";
-import path from "node:path";
-import { evidenceTypes } from "./analyzers/evidence.js";
-import { parseDocument } from "./document-model.js";
-import { loadVoicePackV2 } from "./profile.js";
-
-export function voiceArticleBriefV2({ voice, topic, evidence, cwd = process.cwd() }) {
-  const profile = typeof voice === "string" ? loadVoicePackV2(resolvePath(cwd, voice)) : voice;
-  const evidenceResult = evidence ? evidenceAnchorsFromFile({ evidence, cwd }) : {
-    source: null,
-    anchors: [],
-  };
-
-  return {
-    schemaVersion: 2,
-    generatedBy: "dravoice-v2-brief",
-    topic,
-    voice: {
-      corpusConfidence: profile.source.confidence,
-      sourceFileCount: profile.source.documentCount,
-      sourceWordCount: profile.source.wordCount,
-      primaryRegister: profile.families.register.features.primary.value,
-      evidenceSentenceRate: profile.families.evidence.features.evidenceSentenceRate,
-      featureFamilies: Object.keys(profile.families),
-      draftingRules: profile.guidance.draftingRules.slice(0, 5),
-    },
-    workingThesis: `Draft a grounded article about ${topic}. Let the supplied evidence set the size of each claim before broadening the lesson.`,
-    evidence: evidenceResult,
-    missingEvidence: missingEvidenceFor({ topic, evidenceAnchors: evidenceResult.anchors }),
-    outline: outlineFor(profile),
-    voiceCautions: [
-      ...profile.guidance.avoid,
-      "Mark unsupported claims as [specific evidence needed] instead of inventing proof.",
-    ].slice(0, 5),
-  };
-}
-
-export function renderVoiceBriefV2(brief) {
-  const lines = [
-    `# Article Brief: ${brief.topic}`,
-    "",
-    "## Voice Source",
-    "",
-    `- Corpus confidence: ${capitalize(brief.voice.corpusConfidence.band)} - ${brief.voice.corpusConfidence.message}`,
-    `- Source files: ${brief.voice.sourceFileCount}`,
-    `- Primary register: ${brief.voice.primaryRegister}`,
-    `- Evidence sentence rate: ${brief.voice.evidenceSentenceRate}`,
-    "",
-    "## Working Thesis",
-    "",
-    `- ${brief.workingThesis}`,
-    "",
-    "## Evidence Anchors",
-    "",
-  ];
-
-  if (brief.evidence.anchors.length) {
-    for (const item of brief.evidence.anchors) {
-      const typeList = item.types.length ? ` (${item.types.join(", ")})` : "";
-      lines.push(`- ${brief.evidence.source}:${item.line}${typeList} - ${item.text}`);
-    }
-  } else {
-    lines.push("- [specific evidence needed] Add notes, dates, quotes, examples, or source links before drafting broad claims.");
-  }
-
-  lines.push("", "## Missing Evidence", "");
-  lines.push(...brief.missingEvidence.map((item) => `- ${item}`));
-
-  lines.push("", "## Outline", "");
-  lines.push(...brief.outline.map((item, index) => `${index + 1}. ${item}`));
-
-  lines.push("", "## Voice Cautions", "");
-  lines.push(...brief.voiceCautions.map((item) => `- ${item}`));
-
-  lines.push(
-    "",
-    "## Drafting Prompt",
-    "",
-    `Write the article about ${brief.topic} using the evidence anchors above. Keep claims close to concrete support, follow the voice cautions, and write [specific evidence needed] anywhere the brief does not supply enough ground.`,
-    "",
-  );
-
-  return lines.join("\n");
-}
-
-function evidenceAnchorsFromFile({ evidence, cwd }) {
-  const evidencePath = resolvePath(cwd, evidence);
-  const contents = fs.readFileSync(evidencePath, "utf8");
-  const document = parseDocument({
-    filePath: evidencePath,
-    rootDir: cwd,
-    contents,
-  });
-
-  return {
-    source: displayPath(evidencePath, cwd),
-    anchors: document.sentences
-      .map((sentence) => ({
-        line: sentence.line,
-        text: sentence.text,
-        types: evidenceTypes(sentence.text),
-      }))
-      .filter((sentence) => sentence.types.length > 0)
-      .slice(0, 8),
-  };
-}
-
-function missingEvidenceFor({ topic, evidenceAnchors }) {
-  const items = [
-    `Add [specific evidence needed] for the central claim about ${topic}.`,
-    "Add [specific evidence needed] for any number, date, quote, source, or example the article depends on.",
-  ];
-  if (evidenceAnchors.length === 0) {
-    items.unshift("No evidence anchors were detected; collect concrete notes before asking for a full draft.");
-  }
-  return items;
-}
-
-function outlineFor(profile) {
-  const opening = profile.families.rhetoricalShape.features.openingMoves.slice(0, 3).join(" -> ");
-  const sentenceMedian = profile.families.rhythm.features.sentenceWords.median;
-  return [
-    opening
-      ? `Start from a concrete artifact or observation, keeping the opening shape compatible with: ${opening}.`
-      : "Start from a concrete artifact or observation before making the larger claim.",
-    "Name the pressure, question, or practical stakes that make the evidence matter.",
-    `Develop the article in the learned register with sentence pacing near the ${sentenceMedian}-word median where it fits.`,
-    "Close by returning to the evidence and leaving the reader with a practical handle, not a generic conclusion.",
-  ];
-}
-
-function resolvePath(cwd, value) {
-  return path.isAbsolute(value) ? value : path.join(cwd, value);
-}
-
-function displayPath(filePath, rootDir) {
-  const relative = path.relative(rootDir, filePath);
-  if (relative && !relative.startsWith("..") && !path.isAbsolute(relative)) {
-    return relative.split(path.sep).join("/");
-  }
-  return filePath.split(path.sep).join("/");
-}
-
-function capitalize(value) {
-  return String(value ?? "").charAt(0).toUpperCase() + String(value ?? "").slice(1);
-}
+import fs from "node:fs";
+import path from "node:path";
+import { evidenceTypes } from "./analyzers/evidence.js";
+import { parseDocument } from "./document-model.js";
+import { loadVoicePackV2 } from "./profile.js";
+
+export function voiceArticleBriefV2({ voice, topic, evidence, cwd = process.cwd() }) {
+  const profile = typeof voice === "string" ? loadVoicePackV2(resolvePath(cwd, voice)) : voice;
+  const evidenceResult = evidence ? evidenceAnchorsFromFile({ evidence, cwd }) : {
+    source: null,
+    anchors: [],
+  };
+
+  return {
+    schemaVersion: 2,
+    generatedBy: "dravoice-v2-brief",
+    topic,
+    voice: {
+      corpusConfidence: profile.source.confidence,
+      sourceFileCount: profile.source.documentCount,
+      sourceWordCount: profile.source.wordCount,
+      primaryRegister: profile.families.register.features.primary.value,
+      evidenceSentenceRate: profile.families.evidence.features.evidenceSentenceRate,
+      featureFamilies: Object.keys(profile.families),
+      draftingRules: profile.guidance.draftingRules.slice(0, 5),
+    },
+    workingThesis: `Draft a grounded article about ${topic}. Let the supplied evidence set the size of each claim before broadening the lesson.`,
+    evidence: evidenceResult,
+    missingEvidence: missingEvidenceFor({ topic, evidenceAnchors: evidenceResult.anchors }),
+    outline: outlineFor(profile),
+    voiceCautions: [
+      ...profile.guidance.avoid,
+      "Mark unsupported claims as [specific evidence needed] instead of inventing proof.",
+    ].slice(0, 5),
+  };
+}
+
+export function renderVoiceBriefV2(brief) {
+  const lines = [
+    `# Article Brief: ${brief.topic}`,
+    "",
+    "## Voice Source",
+    "",
+    `- Corpus confidence: ${capitalize(brief.voice.corpusConfidence.band)} - ${brief.voice.corpusConfidence.message}`,
+    `- Source files: ${brief.voice.sourceFileCount}`,
+    `- Primary register: ${brief.voice.primaryRegister}`,
+    `- Evidence sentence rate: ${brief.voice.evidenceSentenceRate}`,
+    "",
+    "## Working Thesis",
+    "",
+    `- ${brief.workingThesis}`,
+    "",
+    "## Evidence Anchors",
+    "",
+  ];
+
+  if (brief.evidence.anchors.length) {
+    for (const item of brief.evidence.anchors) {
+      const typeList = item.types.length ? ` (${item.types.join(", ")})` : "";
+      lines.push(`- ${brief.evidence.source}:${item.line}${typeList} - ${item.text}`);
+    }
+  } else {
+    lines.push("- [specific evidence needed] Add notes, dates, quotes, examples, or source links before drafting broad claims.");
+  }
+
+  lines.push("", "## Missing Evidence", "");
+  lines.push(...brief.missingEvidence.map((item) => `- ${item}`));
+
+  lines.push("", "## Outline", "");
+  lines.push(...brief.outline.map((item, index) => `${index + 1}. ${item}`));
+
+  lines.push("", "## Voice Cautions", "");
+  lines.push(...brief.voiceCautions.map((item) => `- ${item}`));
+
+  lines.push(
+    "",
+    "## Drafting Prompt",
+    "",
+    `Write the article about ${brief.topic} using the evidence anchors above. Keep claims close to concrete support, follow the voice cautions, and write [specific evidence needed] anywhere the brief does not supply enough ground.`,
+    "",
+  );
+
+  return lines.join("\n");
+}
+
+function evidenceAnchorsFromFile({ evidence, cwd }) {
+  const evidencePath = resolvePath(cwd, evidence);
+  const contents = fs.readFileSync(evidencePath, "utf8");
+  const document = parseDocument({
+    filePath: evidencePath,
+    rootDir: cwd,
+    contents,
+  });
+
+  return {
+    source: displayPath(evidencePath, cwd),
+    anchors: document.sentences
+      .map((sentence) => ({
+        line: sentence.line,
+        text: sentence.text,
+        types: evidenceTypes(sentence.text),
+      }))
+      .filter((sentence) => sentence.types.length > 0)
+      .slice(0, 8),
+  };
+}
+
+function missingEvidenceFor({ topic, evidenceAnchors }) {
+  const items = [
+    `Add [specific evidence needed] for the central claim about ${topic}.`,
+    "Add [specific evidence needed] for any number, date, quote, source, or example the article depends on.",
+  ];
+  if (evidenceAnchors.length === 0) {
+    items.unshift("No evidence anchors were detected; collect concrete notes before asking for a full draft.");
+  }
+  return items;
+}
+
+function outlineFor(profile) {
+  const opening = profile.families.rhetoricalShape.features.openingMoves.slice(0, 3).join(" -> ");
+  const sentenceMedian = profile.families.rhythm.features.sentenceWords.median;
+  return [
+    opening
+      ? `Start from a concrete artifact or observation, keeping the opening shape compatible with: ${opening}.`
+      : "Start from a concrete artifact or observation before making the larger claim.",
+    "Name the pressure, question, or practical stakes that make the evidence matter.",
+    `Develop the article in the learned register with sentence pacing near the ${sentenceMedian}-word median where it fits.`,
+    "Close by returning to the evidence and leaving the reader with a practical handle, not a generic conclusion.",
+  ];
+}
+
+function resolvePath(cwd, value) {
+  return path.isAbsolute(value) ? value : path.join(cwd, value);
+}
+
+function displayPath(filePath, rootDir) {
+  const relative = path.relative(rootDir, filePath);
+  if (relative && !relative.startsWith("..") && !path.isAbsolute(relative)) {
+    return relative.split(path.sep).join("/");
+  }
+  return filePath.split(path.sep).join("/");
+}
+
+function capitalize(value) {
+  return String(value ?? "").charAt(0).toUpperCase() + String(value ?? "").slice(1);
+}

package/src/v2/profile.js

@@ -6,8 +6,14 @@ import { analyzeLexical } from "./analyzers/lexical.js";
 import { analyzeRegister } from "./analyzers/register.js";
 import { analyzeRhetoricalShape } from "./analyzers/rhetorical-shape.js";
 import { analyzeRhythm } from "./analyzers/rhythm.js";
-import { analyzeStructure } from "./analyzers/structure.js";
-import { loadDocuments } from "./document-model.js";
+import { analyzeStructure } from "./analyzers/structure.js";
+import { loadDocuments } from "./document-model.js";
+import {
+  STYLOMETRIC_REFERENCES,
+  defaultStyleThresholds,
+  distanceByFamily,
+  percentile,
+} from "./stylometry.js";
 
 export function learnVoicePackV2({ examplesDir, outDir }) {
   const documents = loadDocuments({ examplesDir });
@@ -30,38 +36,80 @@ export function loadVoicePackV2(voiceDir) {
   return profile;
 }
 
-export function buildVoiceProfileV2({ documents }) {
-  const source = sourceSummary(documents);
-  const families = {
-    rhythm: analyzeRhythm(documents),
-    lexical: analyzeLexical(documents),
-    register: analyzeRegister(documents),
-    discourse: analyzeDiscourse(documents),
-    rhetoricalShape: analyzeRhetoricalShape(documents),
-    evidence: analyzeEvidence(documents),
-    structure: analyzeStructure(documents),
-  };
-
-  return {
+export function buildVoiceProfileV2({ documents }) {
+  const source = sourceSummary(documents);
+  const families = analyzeFeatureFamilies(documents);
+
+  return {
     schemaVersion: 2,
     generatedBy: "dravoice-v2",
-    tool: { name: "Dravoice", cli: "drav" },
+    tool: { name: "Dravoice", cli: "drav" },
     source,
     families,
     guidance: guidanceFor({ source, families }),
     calibration: {
       featureStability: Object.fromEntries(Object.entries(families).map(([name, family]) => [name, family.confidence])),
-      tolerances: {
-        rhythmMedianWords: toleranceFor(source.confidence.band, 5, 8, 12),
-        evidenceRate: toleranceFor(source.confidence.band, 0.12, 0.18, 0.25),
-      },
-      minimumDraftSize: {
-        words: source.confidence.band === "weak" ? 25 : 35,
-        sentences: source.confidence.band === "weak" ? 3 : 4,
+      tolerances: {
+        rhythmMedianWords: toleranceFor(source.confidence.band, 5, 8, 12),
+        evidenceRate: toleranceFor(source.confidence.band, 0.12, 0.18, 0.25),
+      },
+      styleThresholds: styleThresholdsFor(documents, families),
+      minimumDraftSize: {
+        words: source.confidence.band === "weak" ? 25 : 35,
+        sentences: source.confidence.band === "weak" ? 3 : 4,
       },
     },
-  };
-}
+  };
+}
+
+function analyzeFeatureFamilies(documents) {
+  return {
+    rhythm: analyzeRhythm(documents),
+    lexical: analyzeLexical(documents),
+    register: analyzeRegister(documents),
+    discourse: analyzeDiscourse(documents),
+    rhetoricalShape: analyzeRhetoricalShape(documents),
+    evidence: analyzeEvidence(documents),
+    structure: analyzeStructure(documents),
+  };
+}
+
+function styleThresholdsFor(documents, fallbackFamilies) {
+  const fallbackThresholds = defaultStyleThresholds();
+  const distancesByFamily = Object.fromEntries(Object.keys(fallbackFamilies).map((family) => [family, []]));
+
+  if (documents.length >= 2) {
+    for (let index = 0; index < documents.length; index += 1) {
+      const referenceDocuments = documents.filter((_, candidateIndex) => candidateIndex !== index);
+      const referenceFamilies = analyzeFeatureFamilies(referenceDocuments);
+      const heldoutFamilies = analyzeFeatureFamilies([documents[index]]);
+      for (const family of Object.keys(fallbackFamilies)) {
+        distancesByFamily[family].push(distanceByFamily(
+          family,
+          referenceFamilies[family].features,
+          heldoutFamilies[family].features,
+        ));
+      }
+    }
+  }
+
+  const families = {};
+  for (const family of Object.keys(fallbackFamilies)) {
+    const observations = distancesByFamily[family];
+    const observedThreshold = observations.length > 0 ? percentile(observations, 0.9) : 0;
+    const fallback = fallbackThresholds[family].threshold;
+    families[family] = {
+      threshold: Math.max(0.01, Math.min(0.95, observedThreshold || fallback)),
+      observations: observations.length,
+    };
+  }
+
+  return {
+    method: "leave-one-out-cosine-delta",
+    references: STYLOMETRIC_REFERENCES,
+    families,
+  };
+}
 
 function writeVoicePackV2(outDir, profile) {
   fs.mkdirSync(outDir, { recursive: true });