@goodandready/dsh-kanban 0.2.36 → 0.2.38

package/lib/server/task-handler.js

@@ -30,12 +30,13 @@ import {
   handleToggleChecklistItem,
 } from '../routes.js'
 import { verifyAndRecordGate } from '../verification.js'
-import { handleTaskSnapshot, handleTaskRollbackSnapshot } from '../routes.js'
+import { handleTaskSnapshot, handleTaskRollbackSnapshot, handleTaskGeneratePlan, handleTaskPrDescription } from '../routes.js'
 import { createInceptionSnapshot } from '../time-machine.js'
 import { runTask, queueTask, unqueueTask, resolveModel, buildStartMessage, PERMISSION_RANKS } from '../launcher.js'
 import { dispatchMove, stopWork, MOVE_DETAIL, STOP_DETAIL } from '../commands.js'
 import { checkWorktreeDirty } from '../worktree.js'
 import { refreshTask } from '../import.js'
+import { findDuplicates } from '../similarity.js'
 import { parseProgressDump, formatProgressDump } from '../progress-dump.js'
 import { rootOf } from '../config.js'
 import { SessionId } from '@deepseek-ai/dsh-session'
@@ -67,6 +68,20 @@ export function createTaskRouteHandler(deps) {
   return async (req, res) => {
       const pathname = urlOf(req).pathname
 
+      if (pathname === '/dsh-kanban/task/find-duplicates') {
+        if (req.method !== 'POST') { writeJson(res, 405, { error: 'method-not-allowed' }); return }
+        if (rejectUntrusted(req, res)) return
+        try {
+          const input = await readBody(req)
+          const candidates = store.listTasks(input?.board || 'main') || []
+          const duplicates = findDuplicates(input, candidates, { threshold: input?.threshold, limit: input?.limit })
+          writeJson(res, 200, { ok: true, duplicates })
+        } catch {
+          writeJson(res, 400, { error: 'bad-request' })
+        }
+        return
+      }
+
       // Path without ID handles local task creation.
       if (pathname === '/dsh-kanban/task' || pathname === '/dsh-kanban/task/') {
         if (req.method !== 'POST') { writeJson(res, 405, { error: 'method-not-allowed' }); return }
@@ -439,6 +454,18 @@ export function createTaskRouteHandler(deps) {
           writeJson(res, resObj.status, resObj.body)
           return
         }
+        if (req.method === 'POST' && action === 'generate-plan') {
+          const resObj = await handleTaskGeneratePlan({ store, id, body: input, eventHub })
+          writeJson(res, resObj.status, resObj.body)
+          return
+        }
+        if ((req.method === 'GET' || req.method === 'POST') && action === 'pr-description') {
+          const cfg = getConfig()
+          const gitRunner = cfg.remoteMode ? createRemoteGitRunner(cfg) : undefined
+          const resObj = await handleTaskPrDescription({ store, id, method: req.method, body: input, gitRunner })
+          writeJson(res, resObj.status, resObj.body)
+          return
+        }
         if (req.method === 'POST' && action === 'resume') {
           watchdog?.resumeTask?.({ store, taskId: id, eventHub })
           const task = store.getTask(id)

package/lib/similarity.js

@@ -0,0 +1,206 @@
+/**
+ * Text Similarity & Duplicate Task Detection Engine (#380)
+ *
+ * Implements multilingual tokenization and character n-gram Dice/Jaccard
+ * similarity algorithms to identify duplicate or highly similar tasks
+ * without external dependencies.
+ */
+
+const STOP_WORDS = new Set([
+  'the', 'a', 'an', 'and', 'or', 'but', 'in', 'on', 'at', 'to', 'for',
+  'of', 'with', 'by', 'from', 'up', 'about', 'into', 'over', 'after',
+  'is', 'are', 'was', 'were', 'be', 'been', 'being', 'have', 'has', 'had',
+  'do', 'does', 'did', 'can', 'could', 'should', 'would', 'will', 'shall',
+  'this', 'that', 'these', 'those', 'it', 'its', 'as', 'if', 'when',
+  'feature', 'bug', 'task', 'fix', 'add', 'update', 'implement', 'test', 'tests'
+])
+
+/**
+ * Normalizes input text by removing markdown syntax, code fences,
+ * URLs, issue references, and non-letter symbols.
+ *
+ * @param {string} text
+ * @returns {string}
+ */
+function normalizeText(text) {
+  if (typeof text !== 'string') return ''
+  return text
+    .toLowerCase()
+    .replace(/```[\s\S]*?```/g, ' ')
+    .replace(/`/g, ' ')
+    .replace(/https?:\/\/[^\s]+/g, ' ')
+    .replace(/#\d+/g, ' ')
+    .replace(/[^\p{L}\p{N}\s]/gu, ' ')
+    .replace(/\s+/g, ' ')
+    .trim()
+}
+
+/**
+ * Splits text into cleaned words, filtering out short tokens and common stop words.
+ *
+ * @param {string} text
+ * @returns {string[]}
+ */
+function tokenizeWords(text) {
+  const normalized = normalizeText(text)
+  if (!normalized) return []
+  return normalized
+    .split(' ')
+    .filter((w) => w.length > 1 && !STOP_WORDS.has(w))
+}
+
+/**
+ * Extracts character n-grams from normalized text for sub-word multilingual matching.
+ *
+ * @param {string} text
+ * @param {number} [n=3]
+ * @returns {Set<string>}
+ */
+function extractNgrams(text, n = 3) {
+  const normalized = normalizeText(text).replace(/\s+/g, '')
+  const ngrams = new Set()
+  if (normalized.length < n) {
+    if (normalized.length > 0) ngrams.add(normalized)
+    return ngrams
+  }
+  for (let i = 0; i <= normalized.length - n; i++) {
+    ngrams.add(normalized.slice(i, i + n))
+  }
+  return ngrams
+}
+
+/**
+ * Calculates Dice similarity coefficient between two sets: 2 * |A ∩ B| / (|A| + |B|).
+ *
+ * @param {Set<string>} setA
+ * @param {Set<string>} setB
+ * @returns {number} Float between 0.0 and 1.0
+ */
+function diceCoefficient(setA, setB) {
+  if (!setA.size || !setB.size) return 0
+  let intersection = 0
+  for (const item of setA) {
+    if (setB.has(item)) intersection++
+  }
+  return (2 * intersection) / (setA.size + setB.size)
+}
+
+/**
+ * Calculates Jaccard index between two sets: |A ∩ B| / |A ∪ B|.
+ *
+ * @param {Set<string>} setA
+ * @param {Set<string>} setB
+ * @returns {number} Float between 0.0 and 1.0
+ */
+function jaccardIndex(setA, setB) {
+  if (!setA.size || !setB.size) return 0
+  let intersection = 0
+  for (const item of setA) {
+    if (setB.has(item)) intersection++
+  }
+  const union = setA.size + setB.size - intersection
+  return union > 0 ? intersection / union : 0
+}
+
+/**
+ * Calculates hybrid text similarity blending word token overlap and character 3-grams.
+ *
+ * @param {string} a
+ * @param {string} b
+ * @returns {number} Similarity score between 0.0 and 1.0
+ */
+function calculateTextSimilarity(a, b) {
+  if (!a || !b) return 0
+  if (a.trim().toLowerCase() === b.trim().toLowerCase()) return 1.0
+
+  const wordsA = new Set(tokenizeWords(a))
+  const wordsB = new Set(tokenizeWords(b))
+  const wordScore = diceCoefficient(wordsA, wordsB)
+
+  const ngramsA = extractNgrams(a, 3)
+  const ngramsB = extractNgrams(b, 3)
+  const ngramScore = diceCoefficient(ngramsA, ngramsB)
+
+  // Blend word tokens (0.4) and character n-grams (0.6) for morphological tolerance
+  return wordScore * 0.4 + ngramScore * 0.6
+}
+
+/**
+ * Calculates composite similarity between query task parameters and an existing task card.
+ *
+ * @param {{ title?: string, description?: string }} query
+ * @param {{ title?: string, description?: string }} candidate
+ * @returns {number} Score between 0.0 and 1.0
+ */
+function compareTasks(query, candidate) {
+  if (!query || !candidate) return 0
+  const titleA = query.title || ''
+  const titleB = candidate.title || ''
+
+  if (titleA && titleB && titleA.trim().toLowerCase() === titleB.trim().toLowerCase()) {
+    return 1.0
+  }
+
+  const titleSim = calculateTextSimilarity(titleA, titleB)
+  const descA = query.description || ''
+  const descB = candidate.description || ''
+
+  if (!descA || !descB) {
+    return titleSim
+  }
+
+  const descSim = calculateTextSimilarity(descA, descB)
+  // Title carries 70% weight, description carries 30% weight
+  return titleSim * 0.7 + descSim * 0.3
+}
+
+/**
+ * Finds duplicate or highly similar tasks from an array of candidate cards.
+ *
+ * @param {{ title?: string, description?: string, excludeId?: string }} query
+ * @param {Array<Object>} candidates
+ * @param {{ threshold?: number, limit?: number }} [options]
+ * @returns {Array<{ id: string, title: string, issueNumber?: number, score: number, percent: number }>}
+ */
+function findDuplicates(query, candidates, options = {}) {
+  if (!query || !Array.isArray(candidates) || candidates.length === 0) {
+    return []
+  }
+  const threshold = typeof options.threshold === 'number' ? options.threshold : 0.45
+  const limit = typeof options.limit === 'number' ? options.limit : 5
+  const excludeId = query.excludeId || query.id
+
+  const results = []
+  for (const candidate of candidates) {
+    if (!candidate || !candidate.id) continue
+    if (excludeId && candidate.id === excludeId) continue
+
+    const score = compareTasks(query, candidate)
+    if (score >= threshold) {
+      results.push({
+        id: candidate.id,
+        title: candidate.title,
+        issueNumber: candidate.issueNumber,
+        repo: candidate.repo,
+        column: candidate.column,
+        score: Math.round(score * 1000) / 1000,
+        percent: Math.round(score * 100),
+      })
+    }
+  }
+
+  return results
+    .sort((a, b) => b.score - a.score)
+    .slice(0, limit)
+}
+
+export {
+  normalizeText,
+  tokenizeWords,
+  extractNgrams,
+  diceCoefficient,
+  jaccardIndex,
+  calculateTextSimilarity,
+  compareTasks,
+  findDuplicates,
+}

package/lib/store.js

@@ -81,6 +81,8 @@ CREATE TABLE IF NOT EXISTS tasks (
   stalledAt             INTEGER NOT NULL DEFAULT 0,
   blockedBy             TEXT NOT NULL DEFAULT '[]',
   inceptionSnapshotId   TEXT NOT NULL DEFAULT '',
+  spec                  TEXT NOT NULL DEFAULT '',
+  prDescription         TEXT NOT NULL DEFAULT '',
   updatedAt    INTEGER NOT NULL
 );
 CREATE INDEX IF NOT EXISTS tasks_board_col ON tasks(board, col, position);
@@ -116,7 +118,7 @@ const WRITABLE = [
   'verificationCommand', 'verificationStatus',
   'costMetrics', 'costLimitExceeded',
   'stalled', 'stalledReason', 'stalledAt',
-  'blockedBy', 'inceptionSnapshotId',
+  'blockedBy', 'inceptionSnapshotId', 'spec', 'prDescription',
 ]
 
 /** Valid transition sources; unexpected values are rejected to preserve log integrity. */
@@ -124,7 +126,7 @@ const SOURCES = new Set(['manual', 'session', 'gitea', 'github', 'tool', 'workfl
 
 function rowToTask(row) {
   if (row === undefined) return undefined
-  const { col, labels, waiting, labelColors, runs, checklist, report, comments, rejectReason, rejectedAt, isEpic, parentId, stalled, blockedBy, inceptionSnapshotId, ...rest } = row
+  const { col, labels, waiting, labelColors, runs, checklist, report, comments, rejectReason, rejectedAt, isEpic, parentId, stalled, blockedBy, inceptionSnapshotId, spec, prDescription, ...rest } = row
   let parsedChecklist = []
   try {
     const value = JSON.parse(checklist ?? '[]')
@@ -187,6 +189,8 @@ function rowToTask(row) {
       } catch { return [] }
     })(),
     inceptionSnapshotId: typeof inceptionSnapshotId === 'string' ? inceptionSnapshotId : (typeof rest.inceptionSnapshotId === 'string' ? rest.inceptionSnapshotId : ''),
+    spec: typeof spec === 'string' ? spec : (typeof rest.spec === 'string' ? rest.spec : ''),
+    prDescription: typeof prDescription === 'string' ? prDescription : (typeof rest.prDescription === 'string' ? rest.prDescription : ''),
   }
 }
 
@@ -284,6 +288,8 @@ export function openStore({ dir }) {
     ['stalledAt', 'INTEGER NOT NULL DEFAULT 0'],
     ['blockedBy', "TEXT NOT NULL DEFAULT '[]'"],
     ['inceptionSnapshotId', "TEXT NOT NULL DEFAULT ''"],
+    ['spec', "TEXT NOT NULL DEFAULT ''"],
+    ['prDescription', "TEXT NOT NULL DEFAULT ''"],
   ]) {
     try { db.exec(`ALTER TABLE tasks ADD COLUMN ${column} ${definition}`) } catch { /* column already exists */ }
   }
@@ -614,7 +620,8 @@ export function openStore({ dir }) {
       for (const key of WRITABLE) {
         if (!Object.hasOwn(patch, key)) continue
         fields.push(`${key} = ?`)
-        if (key === 'labels') values.push(JSON.stringify(patch[key] ?? []))
+        if (key === 'plan') values.push(typeof patch[key] === 'string' ? patch[key] : JSON.stringify(patch[key] ?? { items: [], current: 0 }))
+        else if (key === 'labels') values.push(JSON.stringify(patch[key] ?? []))
         else if (key === 'labelColors') values.push(JSON.stringify(patch[key] ?? {}))
         else if (key === 'runs') values.push(typeof patch[key] === 'string' ? patch[key] : JSON.stringify(patch[key] ?? []))
         else if (key === 'checklist') values.push(typeof patch[key] === 'string' ? patch[key] : JSON.stringify(patch[key] ?? []))
@@ -628,6 +635,8 @@ export function openStore({ dir }) {
         else if (key === 'parentId') values.push(String(patch[key] ?? ''))
         else if (key === 'blockedBy') values.push(Array.isArray(patch[key]) ? JSON.stringify(patch[key]) : (typeof patch[key] === 'string' ? patch[key] : '[]'))
         else if (key === 'inceptionSnapshotId') values.push(String(patch[key] ?? ''))
+        else if (key === 'spec') values.push(String(patch[key] ?? ''))
+        else if (key === 'prDescription') values.push(String(patch[key] ?? ''))
         else values.push(patch[key] ?? null)
       }
       if (fields.length > 0) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@goodandready/dsh-kanban",
-  "version": "0.2.36",
+  "version": "0.2.38",
   "description": "Visual Kanban board for DeepSeek Harness: Gitea and GitHub-backed tasks, workflow columns, and a dedicated agent session per task.",
   "type": "module",
   "main": "lib/index.js",