task-summary-extractor 9.0.0 → 9.1.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "task-summary-extractor",
-  "version": "9.0.0",
+  "version": "9.1.0",
   "description": "AI-powered meeting analysis & document generation CLI — video + document processing, deep dive docs, dynamic mode, interactive CLI with model selection, confidence scoring, learning loop, git progress tracking",
   "main": "process_and_upload.js",
   "bin": {

package/prompt.json

@@ -225,7 +225,7 @@
                 "what": "string (precise description of the change)",
                 "how": "string (implementation/execution approach discussed or specified)",
                 "why": "string (business, technical, legal, or strategic reason)",
-                "type": "bug_fix|feature|refactor|optimization|cleanup|upgrade|integration|configuration|process_change|policy_update|contract_revision|design_change",
+                "type": "bug_fix|feature|refactor|optimization|cleanup|upgrade|integration|configuration|process_change|policy_update|contract_revision|design_change|content_update|documentation",
                 "priority": "low|medium|high|critical",
                 "status": "actionable|pending_decision|blocked|completed",
                 "dependencies": ["string (IDs of other change requests this depends on)"],
@@ -316,7 +316,7 @@
         "blockers": [
             {
                 "id": "string (BLK-1, BLK-2, or matching ID from tracking docs like DB-1, LEGAL-1, INFRA-1)",
-                "type": "database_prerequisite|pending_decision|dba_work|external_dependency|testing|deployment|legal_review|budget_approval|vendor_dependency|regulatory|access_permission|resource_constraint",
+                "type": "database_prerequisite|pending_decision|dba_work|external_dependency|testing|deployment|legal_review|budget_approval|vendor_dependency|regulatory|access_permission|resource_constraint|backend_work",
                 "description": "string",
                 "owner": "string (who needs to resolve this — person, team, department, external party, etc.)",
                 "blocks": ["string (work-item IDs or action items blocked by this)"],

package/src/logger.js

@@ -13,6 +13,7 @@
 
 const fs = require('fs');
 const path = require('path');
+const { strip: stripAnsi } = require('./utils/colors');
 
 const LOG_LEVELS = { debug: 0, info: 1, warn: 2, error: 3 };
 
@@ -85,12 +86,12 @@ class Logger {
 
   _writeDetailed(line) {
     if (this.closed) return;
-    this._detailedBuffer.push(line + '\n');
+    this._detailedBuffer.push(stripAnsi(line) + '\n');
   }
 
   _writeMinimal(line) {
     if (this.closed) return;
-    this._minimalBuffer.push(line + '\n');
+    this._minimalBuffer.push(stripAnsi(line) + '\n');
   }
 
   _writeBoth(line) {
@@ -104,6 +105,8 @@ class Logger {
    */
   _writeStructured(entry) {
     if (this.closed) return;
+    // Strip ANSI from message fields so structured logs are clean
+    if (entry.message) entry.message = stripAnsi(entry.message);
     const enriched = {
       ...entry,
       elapsedMs: this._elapsedMs(),
@@ -117,8 +120,8 @@ class Logger {
 
   _flush(sync = false) {
     const writeFn = sync
-      ? (p, d) => fs.appendFileSync(p, d)
-      : (p, d) => fs.appendFile(p, d, () => {});
+      ? (p, d) => fs.appendFileSync(p, d, 'utf8')
+      : (p, d) => fs.appendFile(p, d, 'utf8', () => {});
 
     if (this._detailedBuffer.length > 0) {
       const data = this._detailedBuffer.join('');

package/src/phases/init.js

@@ -23,6 +23,13 @@ const { loadHistory, analyzeHistory, printLearningInsights } = require('../utils
 // --- Shared state ---
 const { PKG_ROOT, PROJECT_ROOT, setLog, isShuttingDown, setShuttingDown } = require('./_shared');
 
+/** Parse an integer flag, falling back to `defaultVal` only when the input is absent or NaN. */
+function safeInt(raw, defaultVal) {
+  if (raw === undefined || raw === null || raw === true) return defaultVal;
+  const n = parseInt(raw, 10);
+  return Number.isNaN(n) ? defaultVal : n;
+}
+
 // ======================== PHASE: INIT ========================
 
 /**
@@ -49,12 +56,12 @@ async function phaseInit() {
     reanalyze: !!flags.reanalyze,
     dryRun: !!flags['dry-run'],
     userName: flags.name || null,
-    parallel: parseInt(flags.parallel, 10) || MAX_PARALLEL_UPLOADS,
+    parallel: safeInt(flags.parallel, MAX_PARALLEL_UPLOADS),
     logLevel: flags['log-level'] || LOG_LEVEL,
     outputDir: flags.output || null,
-    thinkingBudget: parseInt(flags['thinking-budget'], 10) || THINKING_BUDGET,
-    compilationThinkingBudget: parseInt(flags['compilation-thinking-budget'], 10) || COMPILATION_THINKING_BUDGET,
-    parallelAnalysis: parseInt(flags['parallel-analysis'], 10) || 2, // concurrent segment analysis
+    thinkingBudget: safeInt(flags['thinking-budget'], THINKING_BUDGET),
+    compilationThinkingBudget: safeInt(flags['compilation-thinking-budget'], COMPILATION_THINKING_BUDGET),
+    parallelAnalysis: safeInt(flags['parallel-analysis'], 2), // concurrent segment analysis
     disableFocusedPass: !!flags['no-focused-pass'],
     disableLearning: !!flags['no-learning'],
     disableDiff: !!flags['no-diff'],
@@ -79,11 +86,19 @@ async function phaseInit() {
     opts.minConfidence = check.normalised.toLowerCase();
   }
 
-  // --- Validate --format flag ---
+  // --- Validate --format flag (supports comma-separated: md,html,pdf) ---
   const VALID_FORMATS = new Set(['md', 'html', 'json', 'pdf', 'docx', 'all']);
-  if (!VALID_FORMATS.has(opts.format)) {
-    throw new Error(`Invalid --format "${opts.format}". Must be: md, html, json, pdf, docx, or all`);
+  const requestedFormats = opts.format.split(',').map(f => f.trim()).filter(Boolean);
+  const invalidFormats = requestedFormats.filter(f => !VALID_FORMATS.has(f));
+  if (invalidFormats.length > 0) {
+    throw new Error(`Invalid --format "${invalidFormats.join(', ')}". Valid: md, html, json, pdf, docx, all`);
   }
+  // Normalise: "all" or set of specific formats
+  opts.formats = requestedFormats.includes('all')
+    ? new Set(['md', 'html', 'json', 'pdf', 'docx'])
+    : new Set(requestedFormats);
+  // Keep opts.format as the original string for backwards compatibility
+  opts.format = requestedFormats.includes('all') ? 'all' : requestedFormats.join(',');
 
   // --- Resolve folder: positional arg or interactive selection ---
   let folderArg = positional[0];

package/src/phases/output.js

@@ -25,6 +25,8 @@ const { getLog, phaseTimer, PROJECT_ROOT } = require('./_shared');
 
 /** Check whether a given output type should be rendered. */
 function shouldRender(opts, type) {
+  if (opts.formats) return opts.formats.has(type);
+  // Fallback for legacy callers
   if (opts.format === 'all') return true;
   return opts.format === type;
 }
@@ -59,14 +61,11 @@ async function phaseOutput(ctx, results, compiledAnalysis, compilationRun, compi
   // Attach cost summary to results
   results.costSummary = costTracker.getSummary();
 
-  // Write results JSON (always written unless --format excludes it)
+  // Write results JSON (always written; logged only when JSON format is requested)
   const jsonPath = path.join(runDir, 'results.json');
-  if (shouldRender(opts, 'json') || opts.format === 'all') {
-    fs.writeFileSync(jsonPath, JSON.stringify(results, null, 2), 'utf8');
+  fs.writeFileSync(jsonPath, JSON.stringify(results, null, 2), 'utf8');
+  if (shouldRender(opts, 'json')) {
     log.step(`Results JSON saved → ${jsonPath}`);
-  } else {
-    // Still write JSON internally (needed for uploads/diffs) but don't advertise
-    fs.writeFileSync(jsonPath, JSON.stringify(results, null, 2), 'utf8');
   }
 
   // Generate Markdown

package/src/phases/process-media.js

@@ -440,7 +440,7 @@ async function phaseProcessVideo(ctx, videoPath, videoIndex) {
 
         // === AUTO-RETRY on FAIL ===
         if (qualityReport.shouldRetry && !isShuttingDown()) {
-          console.log(`    ↻ Quality below threshold (${qualityReport.score}/${THRESHOLDS.PASS}) — retrying with enhanced hints...`);
+          console.log(`    ↻ Quality below threshold (${qualityReport.score}/${THRESHOLDS.FAIL_BELOW}) — retrying with enhanced hints...`);
           log.step(`Quality gate FAIL for ${segName} (score: ${qualityReport.score}) — retrying`);
           retried = true;
 

package/src/phases/services.js

@@ -64,7 +64,7 @@ async function phaseServices(ctx) {
       .map(({ absPath, relPath }) => ({
         type: 'inlineText',
         fileName: relPath,
-        content: fs.readFileSync(absPath, 'utf8'),
+        content: fs.readFileSync(absPath, 'utf8').replace(/^\uFEFF/, ''),
       }));
   }
 

package/src/pipeline.js

@@ -370,7 +370,7 @@ async function runDocOnly(ctx) {
   const jsonPath = path.join(runDir, 'results.json');
   fs.writeFileSync(jsonPath, JSON.stringify(results, null, 2), 'utf8');
 
-  const shouldRender = (type) => opts.format === 'all' || opts.format === type;
+  const shouldRender = (type) => opts.formats ? opts.formats.has(type) : (opts.format === 'all' || opts.format === type);
 
   if (compiledAnalysis) {
     const mdMeta = {
@@ -641,7 +641,7 @@ async function runDynamic(initCtx) {
     const ext = path.extname(absPath).toLowerCase();
     try {
       if (INLINE_EXTS.includes(ext)) {
-        let content = fs.readFileSync(absPath, 'utf8');
+        let content = fs.readFileSync(absPath, 'utf8').replace(/^\uFEFF/, '');
         if (content.length > 8000) {
           content = content.slice(0, 8000) + '\n... (truncated)';
         }

package/src/renderers/html.js

@@ -455,10 +455,10 @@ function renderResultsHtml({ compiled, meta }) {
       const comments = t.comments || [];
       if (comments.length > 0) {
         ln('<h4>🗣️ Key Quotes</h4><ul>');
-        for (const c of comments) {
-          const speaker = c.speaker ? resolve(c.speaker, clusterMap) : 'Unknown';
-          const ts = c.timestamp ? `<code>${e(c.timestamp)}</code> ` : '';
-          ln(`<li>${ts}<strong>${e(speaker)}</strong>: "${e(c.text)}"</li>`);
+        for (const cmt of comments) {
+          const speaker = cmt.speaker ? resolve(cmt.speaker, clusterMap) : 'Unknown';
+          const ts = cmt.timestamp ? `<code>${e(cmt.timestamp)}</code> ` : '';
+          ln(`<li>${ts}<strong>${e(speaker)}</strong>: "${e(cmt.text)}"</li>`);
         }
         ln('</ul>');
       }

package/src/renderers/shared.js

@@ -54,7 +54,10 @@ function clusterNames(rawNames) {
 
     let merged = false;
     for (const [existNk, c] of normToCluster) {
-      if (existNk.includes(nk) || nk.includes(existNk)) {
+      // Only merge via substring if the shorter name is at least 5 chars
+      // to prevent false merges (e.g. "Ed" matching "Jeddah Dev")
+      const shorter = nk.length < existNk.length ? nk : existNk;
+      if (shorter.length >= 5 && (existNk.includes(nk) || nk.includes(existNk))) {
         c.variants.add(raw);
         normToCluster.set(nk, c);
         if (stripped.length >= c.canonical.length && stripped[0] === stripped[0].toUpperCase()) {
@@ -92,7 +95,8 @@ function resolve(name, clusterMap) {
       if (normalizeKey(v) === nk) return canonical;
     }
     const cnk = normalizeKey(canonical);
-    if (cnk.includes(nk) || nk.includes(cnk)) return canonical;
+    const shorter = nk.length < cnk.length ? nk : cnk;
+    if (shorter.length >= 5 && (cnk.includes(nk) || nk.includes(cnk))) return canonical;
   }
   return stripParens(name).trim() || name;
 }

package/src/services/doc-parser.js

@@ -234,7 +234,7 @@ function stripHtml(html) {
  */
 async function parseBuiltinText(filePath) {
   try {
-    const content = await fs.promises.readFile(filePath, 'utf8');
+    const content = (await fs.promises.readFile(filePath, 'utf8')).replace(/^\uFEFF/, '');
     return { text: content.trim(), warnings: [] };
   } catch (err) {
     return { text: '', warnings: [`Failed to read file: ${err.message}`] };
@@ -248,7 +248,7 @@ async function parseBuiltinText(filePath) {
  */
 async function parseHtmlFile(filePath) {
   try {
-    const html = await fs.promises.readFile(filePath, 'utf8');
+    const html = (await fs.promises.readFile(filePath, 'utf8')).replace(/^\uFEFF/, '');
     const text = stripHtml(html);
     return { text, warnings: [] };
   } catch (err) {

package/src/services/gemini.js

@@ -61,7 +61,7 @@ async function prepareDocsForGemini(ai, docFileList) {
     try {
       if (INLINE_TEXT_EXTS.includes(ext)) {
         console.log(`    Reading ${name} (inline text)...`);
-        const content = await fs.promises.readFile(docPath, 'utf8');
+        const content = (await fs.promises.readFile(docPath, 'utf8')).replace(/^\uFEFF/, '');
         prepared.push({ type: 'inlineText', fileName: name, content });
         console.log(`    ${c.success(`${name} ready (${(content.length / 1024).toFixed(1)} KB)`)}`);
       } else if (DOC_PARSER_EXTS.includes(ext)) {
@@ -442,6 +442,22 @@ async function processWithGemini(ai, filePath, displayName, contextDocs = [], pr
         console.error(`    ${c.error(`File API fallback also failed: ${fallbackErr.message}`)}`);
         throw fallbackErr;
       }
+    } else if (!usedExternalUrl && errMsg.includes('INVALID_ARGUMENT')) {
+      // File API upload was used but still got INVALID_ARGUMENT — re-upload fresh and retry once
+      console.log(`    ${c.warn('INVALID_ARGUMENT with File API — re-uploading and retrying...')}`);
+      try {
+        file = await uploadViaFileApi();
+        contentParts[0] = { fileData: { mimeType: file.mimeType, fileUri: file.uri } };
+        requestPayload.contents[0].parts = contentParts;
+        response = await withRetry(
+          () => ai.models.generateContent(requestPayload),
+          { label: `Gemini segment analysis — re-upload retry (${displayName})`, maxRetries: 1, baseDelay: 5000 }
+        );
+        console.log(`    ${c.success('Re-upload retry succeeded')}`);
+      } catch (reuploadErr) {
+        console.error(`    ${c.error(`Re-upload retry also failed: ${reuploadErr.message}`)}`);
+        throw reuploadErr;
+      }
     } else {
       // Log request diagnostics for other errors to aid debugging
       const partSummary = contentParts.map((p, i) => {

package/src/utils/cli.js

@@ -246,7 +246,7 @@ async function selectModel(GEMINI_MODELS, currentModel) {
   const indexMap = {}; // index → modelId
   for (const id of modelIds) {
     const m = GEMINI_MODELS[id];
-    const tier = tiers[m.tier] || tiers.fast;
+    const tier = tiers[m.tier] || tiers.economy;
     idx++;
     indexMap[idx] = id;
     tier.models.push({ idx, id, ...m });
@@ -370,7 +370,7 @@ ${f('--deep-dive', 'Generate explanatory docs per topic')}
   ${h('CORE OPTIONS')}
 ${f('--name <name>', 'Your name (skip interactive prompt)')}
 ${f('--model <id>', 'Gemini model (skip interactive selector)')}
-${f('--format <type>', 'Output formats: md, html, json, pdf, docx, all (default: all)')}
+${f('--format <type>', 'Output: md, html, json, pdf, docx, all — comma-separated (default: all)')}
 ${f('--min-confidence <level>', 'Filter: high, medium, low (default: all)')}
 ${f('--output <dir>', 'Custom output directory for results')}
 ${f('--skip-upload', 'Skip Firebase Storage uploads')}
@@ -424,6 +424,7 @@ ${f('--version, -v', 'Show version')}
     ${c.dim('$')} taskex --dynamic --request "Plan API migration" "specs"
     ${c.dim('$')} taskex --min-confidence medium "call 1" ${c.dim('# Filter low-confidence')}
     ${c.dim('$')} taskex --format md "call 1" ${c.dim('# Markdown only')}
+    ${c.dim('$')} taskex --format md,html,pdf "call 1" ${c.dim('# Multiple formats')}
     ${c.dim('$')} taskex --format pdf "call 1" ${c.dim('# PDF report')}
     ${c.dim('$')} taskex --format docx "call 1" ${c.dim('# Word document')}
     ${c.dim('$')} taskex --resume "call 1" ${c.dim('# Resume interrupted run')}

package/src/utils/confidence-filter.js

@@ -103,7 +103,8 @@ function filterByConfidence(compiled, minLevel = 'LOW') {
       tasks_todo: filterArr(compiled.your_tasks.tasks_todo),
       tasks_waiting_on_others: filterArr(compiled.your_tasks.tasks_waiting_on_others),
       decisions_needed: filterArr(compiled.your_tasks.decisions_needed),
-      completed_in_call: filterArr(compiled.your_tasks.completed_in_call),
+      // completed_in_call items are plain strings (no confidence field) — preserve unconditionally
+      completed_in_call: compiled.your_tasks.completed_in_call || [],
     };
   }
 

package/src/utils/quality-gate.js

@@ -19,10 +19,10 @@ const { c } = require('./colors');
 // ======================== QUALITY THRESHOLDS ========================
 
 const THRESHOLDS = {
-  /** Minimum score to PASS without retry (0-100) */
-  PASS: 45,
-  /** Score range for WARNING — will pass but flag issues (45-65 is typical) */
-  WARN: 65,
+  /** Minimum score to avoid FAIL. Below this → FAIL + retry (0-100) */
+  FAIL_BELOW: 45,
+  /** Minimum score for a clean PASS. Between FAIL_BELOW and PASS_ABOVE → WARN (45-65 is typical) */
+  PASS_ABOVE: 65,
   /** Maximum retries per segment */
   MAX_RETRIES: 1,
 };
@@ -301,9 +301,9 @@ function assessQuality(analysis, context = {}) {
   ];
 
   let grade;
-  if (compositeScore >= THRESHOLDS.WARN) {
+  if (compositeScore >= THRESHOLDS.PASS_ABOVE) {
     grade = 'PASS';
-  } else if (compositeScore >= THRESHOLDS.PASS) {
+  } else if (compositeScore >= THRESHOLDS.FAIL_BELOW) {
     grade = 'WARN';
   } else {
     grade = 'FAIL';