dsh-titlecraft 0.1.0-rc.0

package/src/core.ts

@@ -0,0 +1,516 @@
+import iconCatalogData from './icons.json' with { type: 'json' }
+
+export const ALLOWED_TYPES = Object.freeze([
+  '功能',
+  '设计',
+  '修复',
+  '优化',
+  '发布',
+  '探索',
+  '文档',
+  '研究',
+])
+
+export const TITLE_TEMPLATES = Object.freeze({
+  standard: '{icon} {date}｜{type}｜{topic}',
+  concise: '{icon} {type}｜{topic}',
+  'topic-first': '{icon} {topic}｜{type} · {date}',
+  bracketed: '{icon} [{type}] {topic} · {date}',
+  compact: '{icon} {date} · {topic}',
+  timeline: '{icon} {date} — {topic} [{type}]',
+  minimal: '{icon} {topic}',
+})
+export const TITLE_ICON_GROUPS = Object.freeze(
+  iconCatalogData.groups.map((group) =>
+    Object.freeze({
+      id: group.id,
+      zh: group.zh,
+      en: group.en,
+      icons: Object.freeze(group.icons.map((icon) => Object.freeze({ ...icon }))),
+    }),
+  ),
+)
+export const TITLE_ICON_CATALOG = Object.freeze(TITLE_ICON_GROUPS.flatMap(({ icons }) => icons))
+export const TITLE_ICON_LEGACY_GLYPHS = Object.freeze([...iconCatalogData.legacyGlyphs])
+export const TITLE_ICONS = Object.freeze(
+  Object.fromEntries([['none', ''], ...TITLE_ICON_CATALOG.map(({ id, glyph }) => [id, glyph])]),
+)
+const FONT_FAMILIES = new Set(['dsh', 'system-ui', 'ui-monospace', 'ui-serif'])
+
+export const DEFAULTS: any = Object.freeze({
+  enabled: true,
+  dateSource: 'createdAt',
+  types: ALLOWED_TYPES,
+  enabledTypes: ALLOWED_TYPES,
+  templatePreset: 'standard',
+  formatTemplate: TITLE_TEMPLATES.standard,
+  iconMode: 'none',
+  iconId: 'spark',
+  fontFamily: 'dsh',
+  fontSize: 14,
+  sessionTitleLeftOffset: 0,
+  timezone: 'Asia/Shanghai',
+  datePattern: 'MMDD',
+  instantEnabled: true,
+  firstMessageLlmEnabled: false,
+  refineEnabled: true,
+  idleDelayMs: 3000,
+  maxInputBytes: 6000,
+  maxOutputTokens: 96,
+  timeoutMs: 60000,
+  topicMaxChars: 16,
+  minConfidence: 0.72,
+  preferCheap: true,
+  fallbackToSessionModel: false,
+  modelProvider: '',
+  modelName: '',
+  respectManualTitles: true,
+  topicGuidance: '主题应简洁、具体，基于对话实际内容；不要重复项目名称。无法判断时返回空主题。',
+})
+
+const TYPE_RULES: ReadonlyArray<readonly [string, RegExp]> = Object.freeze([
+  ['修复', /(?:修复|修好|排查|解决|报错|错误|故障|崩溃|失败|异常|\bfix(?:ed|ing)?\b|\bbug\b)/iu],
+  [
+    '发布',
+    /(?:发布|部署|上线|发版|提交代码到\s*github|推送到\s*github|\brelease\b|\bdeploy\b|\bpublish\b)/iu,
+  ],
+  [
+    '优化',
+    /(?:优化|改进|提升|加速|精简|重构|性能|体验|\boptimi[sz]e\b|\brefactor\b|\bperformance\b)/iu,
+  ],
+  ['文档', /(?:文档|说明书|使用说明|教程|指南|readme|\bdocs?\b|\bdocument(?:ation)?\b)/iu],
+  ['研究', /(?:研究|调研|分析|对比|评估|验证原理|benchmark|\bresearch\b|\binvestigate\b)/iu],
+  ['设计', /(?:设计|规划|方案|架构|界面|交互|布局|样式|视觉|\bdesign\b|\barchitecture\b)/iu],
+  ['探索', /(?:探索|讨论|尝试|可行性|设想|脑暴|brainstorm|\bexplore\b|\bprototype\b)/iu],
+  [
+    '功能',
+    /(?:新增|添加|实现|开发|创建|整合|集成|接入|支持|完成|功能|\badd\b|\bimplement\b|\bintegrat(?:e|ion)\b|\bfeature\b)/iu,
+  ],
+])
+
+const GENERIC_TOPICS = new Set([
+  '新功能',
+  '新功能讨论',
+  '功能讨论',
+  '讨论',
+  '继续',
+  '继续处理',
+  '帮我看看',
+  '处理一下',
+  '看看这个',
+  '解决问题',
+  '修复问题',
+  '优化一下',
+  '做个设计',
+  'new feature',
+  'discussion',
+  'continue',
+  'help me',
+])
+
+// Control characters are deliberately stripped from untrusted title input.
+// eslint-disable-next-line no-control-regex
+const CONTROL_PATTERN = /[\u0000-\u0008\u000B\u000C\u000E-\u001F\u007F]/g
+const EDGE_PUNCTUATION =
+  /^[\s.,;:!?。，；：！？、·\-—_'"“”‘’()[\]{}<>/\\|~`#*+=&%$@]+|[\s.,;:!?。，；：！？、·\-—_'"“”‘’()[\]{}<>/\\|~`#*+=&%$@]+$/g
+
+export function normalizeConfig(value: any = {}) {
+  const out = { ...DEFAULTS, ...(value && typeof value === 'object' ? value : {}) }
+  out.enabled = Boolean(out.enabled)
+  out.dateSource = ['createdAt', 'lastUserMessage', 'lastCompletedTurn', 'lastActivity'].includes(
+    out.dateSource,
+  )
+    ? out.dateSource
+    : DEFAULTS.dateSource
+  out.types = normalizeTypes(out.types, 'types')
+  out.enabledTypes = normalizeTypes(value?.enabledTypes ?? out.types, 'enabledTypes')
+  if (out.enabledTypes.some((type) => !out.types.includes(type))) {
+    throw new Error('启用类型必须存在于类型列表中')
+  }
+  out.instantEnabled = Boolean(out.instantEnabled)
+  out.firstMessageLlmEnabled = Boolean(out.firstMessageLlmEnabled)
+  out.refineEnabled = Boolean(out.refineEnabled)
+  out.preferCheap = Boolean(out.preferCheap)
+  out.fallbackToSessionModel = Boolean(out.fallbackToSessionModel)
+  out.respectManualTitles = Boolean(out.respectManualTitles)
+  out.templatePreset = Object.hasOwn(TITLE_TEMPLATES, out.templatePreset)
+    ? out.templatePreset
+    : 'custom'
+  out.formatTemplate = String(
+    out.formatTemplate || TITLE_TEMPLATES[out.templatePreset] || DEFAULTS.formatTemplate,
+  ).trim()
+  const requestedIconMode = out.iconMode === 'type' ? 'smart' : out.iconMode
+  out.iconMode = ['none', 'selected', 'smart'].includes(requestedIconMode)
+    ? requestedIconMode
+    : DEFAULTS.iconMode
+  out.iconId = Object.hasOwn(TITLE_ICONS, out.iconId) ? out.iconId : DEFAULTS.iconId
+  out.fontFamily = FONT_FAMILIES.has(out.fontFamily) ? out.fontFamily : DEFAULTS.fontFamily
+  out.fontSize = integerIn(out.fontSize, 12, 20, DEFAULTS.fontSize)
+  out.sessionTitleLeftOffset = integerIn(
+    out.sessionTitleLeftOffset,
+    0,
+    8,
+    DEFAULTS.sessionTitleLeftOffset,
+  )
+  out.timezone = String(out.timezone || DEFAULTS.timezone).trim()
+  out.datePattern = String(out.datePattern || DEFAULTS.datePattern).trim()
+  out.modelProvider = String(out.modelProvider || '').trim()
+  out.modelName = String(out.modelName || '').trim()
+  out.topicGuidance = String(out.topicGuidance || DEFAULTS.topicGuidance).trim()
+  out.idleDelayMs = integerIn(out.idleDelayMs, 250, 600000, DEFAULTS.idleDelayMs)
+  out.maxInputBytes = integerIn(out.maxInputBytes, 512, 65536, DEFAULTS.maxInputBytes)
+  out.maxOutputTokens = integerIn(out.maxOutputTokens, 32, 512, DEFAULTS.maxOutputTokens)
+  out.timeoutMs = integerIn(out.timeoutMs, 1000, 2 ** 31 - 1, DEFAULTS.timeoutMs)
+  out.topicMaxChars = integerIn(out.topicMaxChars, 4, 40, DEFAULTS.topicMaxChars)
+  out.minConfidence = numberIn(out.minConfidence, 0, 1, DEFAULTS.minConfidence)
+
+  const tokens = [...out.formatTemplate.matchAll(/\{([^{}]+)\}/g)].map((match) => match[1])
+  if (
+    tokens.some((token) => !['date', 'type', 'topic', 'icon'].includes(token)) ||
+    /[{}]/.test(out.formatTemplate.replace(/\{(?:date|type|topic|icon)\}/g, ''))
+  )
+    throw new Error('formatTemplate 包含未知或不完整的占位符')
+  if (!out.formatTemplate.includes('{topic}')) {
+    throw new Error('formatTemplate 必须包含 {topic}')
+  }
+  if (out.modelProvider.length > 0 !== out.modelName.length > 0) {
+    throw new Error('modelProvider 与 modelName 必须同时设置或同时留空')
+  }
+  try {
+    new Intl.DateTimeFormat('en-US', { timeZone: out.timezone }).format(0)
+  } catch {
+    throw new Error(`无效时区：${out.timezone}`)
+  }
+  return Object.freeze(out)
+}
+
+export function normalizeTypes(value, field = 'types') {
+  if (!Array.isArray(value) || value.length < 1 || value.length > 32) {
+    throw new Error(`${field} 必须包含 1–32 个类型`)
+  }
+  const types = value.map((type) => {
+    if (typeof type !== 'string') throw new Error('类型名称必须是文字')
+    const name = type.trim()
+    if (!/^[\p{L}\p{N}][\p{L}\p{N} _-]*$/u.test(name) || [...name].length > 12) {
+      throw new Error('类型名称须为 1–12 个字符，仅允许文字、数字、空格、下划线和短横线')
+    }
+    return name
+  })
+  if (new Set(types.map((type) => type.toLowerCase())).size !== types.length) {
+    throw new Error('类型名称不能重复')
+  }
+  return Object.freeze(types)
+}
+
+export function formatCreatedDate(
+  createdAt,
+  timezone = DEFAULTS.timezone,
+  pattern = DEFAULTS.datePattern,
+) {
+  if (!Number.isFinite(createdAt) || createdAt < 0)
+    throw new Error('createdAt 必须是非负 Unix 毫秒时间戳')
+  const parts = new Intl.DateTimeFormat('en-US', {
+    timeZone: timezone,
+    year: 'numeric',
+    month: '2-digit',
+    day: '2-digit',
+  }).formatToParts(new Date(createdAt))
+  const table = Object.fromEntries(parts.map((part) => [part.type, part.value]))
+  return pattern
+    .replaceAll('YYYY', table.year)
+    .replaceAll('MM', table.month)
+    .replaceAll('DD', table.day)
+}
+
+export function formatSessionTitle({
+  createdAt,
+  dateTimestamp,
+  type,
+  topic,
+  iconId: semanticIconId = '',
+  config = DEFAULTS,
+  maxBytes = 80,
+}) {
+  const policy = normalizeConfig(config)
+  if (!policy.enabledTypes.includes(type)) throw new Error(`未启用的类型：${String(type)}`)
+  const cleanTopic = normalizeTopic(topic, policy.topicMaxChars)
+  if (cleanTopic.length === 0) return null
+  const timestamp = Number.isFinite(dateTimestamp) ? dateTimestamp : createdAt
+  const date = formatCreatedDate(timestamp, policy.timezone, policy.datePattern)
+  const resolvedIconId =
+    policy.iconMode === 'smart' &&
+    Object.hasOwn(TITLE_ICONS, semanticIconId) &&
+    semanticIconId !== 'none'
+      ? semanticIconId
+      : policy.iconId
+  const icon = policy.iconMode === 'none' ? '' : TITLE_ICONS[resolvedIconId] || ''
+  const render = () =>
+    policy.formatTemplate
+      .replaceAll('{date}', date)
+      .replaceAll('{type}', type)
+      .replaceAll('{topic}', cleanTopic)
+      .replaceAll('{icon}', icon ? `${icon} ` : '')
+      .replace(/\s+/g, ' ')
+      .trim()
+  const title = render()
+  return Buffer.byteLength(title, 'utf8') <= maxBytes ? title : null
+}
+
+export function inferInstant(text, options: any = {}) {
+  const policy = normalizeConfig(options.config ?? options)
+  if (
+    /(?:[\n\r`“”「」]|https?:\/\/|^\s*>|^(?:继续|已重启|不要|无需|别|还是|现在|好的))/u.test(
+      String(text),
+    )
+  )
+    return null
+  const cleaned = cleanSourceText(text)
+  if (cleaned.length === 0) return null
+  // Extract an explicit capability instead of truncating a provider/model prefix.
+  if (
+    /^(?:请\s*)?(?:修改|调整|配置)[^\n]{0,100}(?:模型|参数)[^\n]{0,40}(?:使之|使其|让其|以便)[^\n]{0,20}(?:调整|选择|设置)(?:思考|推理)强度[。！!]?$/u.test(
+      cleaned,
+    )
+  ) {
+    if (!policy.enabledTypes.includes('功能')) return null
+    const topic = normalizeTopic('思考强度选择', policy.topicMaxChars)
+    return topic ? { type: '功能', topic, confidence: 0.9 } : null
+  }
+  const type = classifyType(cleaned)
+  if (type === null || !policy.enabledTypes.includes(type)) return null
+  const topic = extractHeuristicTopic(cleaned, type, options.projectName, policy.topicMaxChars)
+  if (topic === null) return null
+  return { type, topic, confidence: 0.78 }
+}
+
+export function classifyType(text) {
+  for (const [type, pattern] of TYPE_RULES) {
+    if (pattern.test(text)) return type
+  }
+  return null
+}
+
+export function extractHeuristicTopic(text, type, projectName, maxChars = DEFAULTS.topicMaxChars) {
+  if (/[，,；;。！？!?：:]/u.test(text.replace(/[。！!]$/u, ''))) return null
+  let topic = cleanSourceText(text)
+  if (projectName) topic = removeProjectName(topic, projectName)
+  topic = topic
+    .replace(
+      /^(?:麻烦你?|请你?|请|帮我(?:一下|看看|把)?|你能(?:不能)?|可以帮我|我想(?:要)?|我需要|please|could you|can you)\s*/iu,
+      '',
+    )
+    .replace(/^(?:关于|针对)\s*/u, '')
+
+  if (type === '优化') topic = topic.replace(/^(?:优化|改进|提升|精简|重构)\s*/u, '')
+  if (type === '修复') topic = topic.replace(/^(?:修复|修好|解决|排查)\s*/u, '')
+  if (type === '设计') topic = topic.replace(/^(?:设计|规划)\s*/u, '')
+  if (type === '文档') topic = topic.replace(/^(?:编写|补充|更新)\s*(?:文档)?\s*/u, '')
+  if (type === '研究') topic = topic.replace(/^(?:研究|调研|分析|评估)\s*/u, '')
+
+  if (/(?:这个|那个|一下|使之|使其|以便|然后|同时|怎么|如何)/u.test(topic)) return null
+  topic = normalizeTopic(topic, maxChars)
+  if (topic.length < 2 || isGenericTopic(topic)) return null
+  return topic
+}
+
+export function inspectModelDecision(raw, options: any = {}) {
+  const policy = normalizeConfig(options.config ?? options)
+  let parsed
+  try {
+    parsed = JSON.parse(String(raw ?? '').trim())
+  } catch {
+    return { kind: 'invalid', reason: 'invalid-json' }
+  }
+  if (
+    !parsed ||
+    Array.isArray(parsed) ||
+    typeof parsed !== 'object' ||
+    Object.keys(parsed).some(
+      (key) =>
+        ![
+          'type',
+          'topic',
+          'confidence',
+          ...(policy.iconMode === 'smart' ? ['iconId'] : []),
+        ].includes(key),
+    ) ||
+    typeof parsed.type !== 'string' ||
+    typeof parsed.topic !== 'string' ||
+    typeof parsed.confidence !== 'number' ||
+    !Number.isFinite(parsed.confidence) ||
+    parsed.confidence < 0 ||
+    parsed.confidence > 1
+  ) {
+    return { kind: 'invalid', reason: 'invalid-schema' }
+  }
+  if (parsed.confidence < policy.minConfidence || !parsed.topic.trim())
+    return { kind: 'uncertain', reason: 'uncertain-topic' }
+  if (!policy.enabledTypes.includes(parsed.type)) return { kind: 'invalid', reason: 'invalid-type' }
+  if (
+    policy.iconMode === 'smart' &&
+    (typeof parsed.iconId !== 'string' ||
+      parsed.iconId === 'none' ||
+      !Object.hasOwn(TITLE_ICONS, parsed.iconId))
+  )
+    return { kind: 'invalid', reason: 'invalid-icon' }
+  let topic = parsed.topic
+  if (options.projectName) topic = removeProjectName(topic, options.projectName)
+  topic = normalizeTopic(topic, policy.topicMaxChars)
+  if (!topic) return { kind: 'invalid', reason: 'invalid-topic-length' }
+  if ([...topic].length < 2 || isGenericTopic(topic))
+    return { kind: 'uncertain', reason: 'generic-topic' }
+  if (/[\r\n｜|{}]/u.test(parsed.topic)) return { kind: 'invalid', reason: 'invalid-topic-format' }
+  return {
+    kind: 'valid',
+    decision: {
+      type: parsed.type,
+      topic,
+      confidence: parsed.confidence,
+      ...(policy.iconMode === 'smart' ? { iconId: parsed.iconId } : {}),
+    },
+  }
+}
+
+export function parseModelDecision(raw, options: any = {}) {
+  const result = inspectModelDecision(raw, options)
+  return result.kind === 'valid' ? result.decision : null
+}
+
+export function normalizeTopic(text, maxChars = DEFAULTS.topicMaxChars) {
+  let topic = String(text ?? '')
+    .replace(CONTROL_PATTERN, '')
+    .replace(/^["'“”‘’]+|["'“”‘’]+$/g, '')
+    .replace(EDGE_PUNCTUATION, '')
+    .replace(/\s+/g, ' ')
+    .trim()
+  if (/\p{Script=Han}/u.test(topic)) topic = topic.replace(/\s+/g, '')
+  return [...topic].length <= maxChars ? topic : ''
+}
+
+export function sessionDateTimestamp(session, source = DEFAULTS.dateSource) {
+  const createdAt = session?.header?.createdAt
+  if (source === 'createdAt' || !session || typeof session.snapshotEvents !== 'function')
+    return createdAt
+  const events = session.snapshotEvents()
+  for (let i = events.length - 1; i >= 0; i--) {
+    const event = events[i]
+    if (!Number.isFinite(event?.time)) continue
+    if (source === 'lastActivity') return event.time
+    if (
+      source === 'lastUserMessage' &&
+      event.type === 'user/message' &&
+      event.data?.source?.kind === 'user'
+    )
+      return event.time
+    if (
+      source === 'lastCompletedTurn' &&
+      event.type === 'turn/end' &&
+      event.data?.reason?.kind === 'completed'
+    )
+      return event.time
+  }
+  return createdAt
+}
+
+export function projectNameOf(session) {
+  const cwd = session?.header?.cwd
+  if (typeof cwd !== 'string' || cwd.trim() === '') return ''
+  return (
+    cwd
+      .replace(/[\\/]+$/u, '')
+      .split(/[\\/]/u)
+      .pop() ?? ''
+  )
+}
+
+export function isSameTitle(a, b) {
+  const comparable = (value) =>
+    String(value ?? '')
+      .toLowerCase()
+      .replace(/[^\p{L}\p{N}]+/gu, '')
+  const left = comparable(a)
+  return left.length > 0 && left === comparable(b)
+}
+
+export function selectWithinBudget(messages, maxBytes) {
+  const selected = []
+  const bytes = (items) => Buffer.byteLength(JSON.stringify({ messages: items }), 'utf8')
+  // Keep the initiating task plus recent context; measure the actual JSON payload.
+  const ordered = messages.length > 1 ? [messages[0], ...messages.slice(1).reverse()] : messages
+  for (const message of ordered) {
+    if (!message || typeof message.text !== 'string') continue
+    const item = { seq: message.seq, text: message.text }
+    const candidate = [...selected, item].sort((a, b) => a.seq - b.seq)
+    if (bytes(candidate) <= maxBytes) {
+      selected.push(item)
+      continue
+    }
+    if (selected.length) continue
+    const chars = [...item.text]
+    let low = 0,
+      high = chars.length
+    while (low < high) {
+      const mid = Math.ceil((low + high) / 2)
+      if (bytes([{ ...item, text: chars.slice(0, mid).join('') }]) <= maxBytes) low = mid
+      else high = mid - 1
+    }
+    if (low) selected.push({ ...item, text: chars.slice(0, low).join('') })
+  }
+  return selected.sort((a, b) => a.seq - b.seq)
+}
+
+export function truncateUtf8(text, maxBytes) {
+  const value = String(text ?? '')
+  if (Buffer.byteLength(value, 'utf8') <= maxBytes) return value
+  let out = ''
+  let bytes = 0
+  for (const char of value) {
+    const width = Buffer.byteLength(char, 'utf8')
+    if (bytes + width > maxBytes) break
+    out += char
+    bytes += width
+  }
+  return out
+}
+
+function cleanSourceText(text) {
+  return String(text ?? '')
+    .replace(/```[\s\S]*?```/g, ' ')
+    .replace(/`[^`]*`/g, ' ')
+    .replace(/https?:\/\/\S+/giu, ' ')
+    .replace(/^#{1,6}\s+/gm, '')
+    .replace(/^\s*[-*+]\s+/gm, '')
+    .replace(/^\s*>\s?/gm, '')
+    .replace(/[*_~]{1,3}([^*_~]+)[*_~]{1,3}/g, '$1')
+    .replace(CONTROL_PATTERN, '')
+    .replace(/\s+/g, ' ')
+    .trim()
+}
+
+function removeProjectName(text, projectName) {
+  const name = String(projectName ?? '').trim()
+  if (name.length < 2) return text
+  return text
+    .replace(new RegExp(escapeRegExp(name), 'giu'), ' ')
+    .replace(/\s+/g, ' ')
+    .trim()
+}
+
+function isGenericTopic(topic) {
+  const normalized = topic.toLowerCase().replace(/[。.!！?？\s]/g, '')
+  return GENERIC_TOPICS.has(topic.toLowerCase()) || GENERIC_TOPICS.has(normalized)
+}
+
+function escapeRegExp(text) {
+  return text.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')
+}
+
+function integerIn(value, min, max, fallback) {
+  const number = Number(value)
+  return Number.isSafeInteger(number) && number >= min && number <= max ? number : fallback
+}
+
+function numberIn(value, min, max, fallback) {
+  const number = Number(value)
+  return Number.isFinite(number) && number >= min && number <= max ? number : fallback
+}