@jurislm/judicial-mcp 1.3.0 → 1.6.2

package/src/public-judgment-source.ts

@@ -0,0 +1,487 @@
+import axios, { type AxiosResponse } from 'axios'
+import { load } from 'cheerio'
+import { JudicialSourceError } from './source-error.js'
+
+const BASE_URL = 'https://judgment.judicial.gov.tw/FJUD/'
+const SEARCH_URL = new URL('Default_AD.aspx', BASE_URL)
+const DETAIL_URL = new URL('data.aspx', BASE_URL)
+const PAGE_SIZE = 10
+const SOURCE_PAGE_SIZE = 20
+const MAX_PAGE = 50
+const MAX_TEXT_LENGTH = 40_000
+const MAX_SUMMARY_LENGTH = 500
+const REQUEST_LIMIT = 2_000_000
+const NO_RESULTS = /查無.{0,12}(?:資料|裁判書)|沒有符合條件的資料/
+
+interface SearchCriteria {
+  keyword?: string
+  court?: string
+  year?: number
+  caseNumber?: string
+  limit: number
+  page: number
+  offset: number
+}
+
+interface CaseNumberParts {
+  year?: number
+  caseWord?: string
+  number: string
+}
+
+interface JudgmentSearchResult {
+  case_id: string
+  case_number: string
+  date: string | null
+  court: string
+  cause: string
+  summary?: string
+  source_url: string
+  search_source_url: string
+  retrieved_at: string
+  cache: { status: 'not_cached' }
+  truncated: boolean
+  partial: false
+  source_version: null
+}
+
+function optionalString(args: Record<string, unknown>, name: string, maxLength: number): string | undefined {
+  const value = args[name]
+  if (value === undefined) return undefined
+  if (typeof value !== 'string' || value.trim().length === 0 || value.length > maxLength) {
+    throw new Error(`參數 ${name} 必須是 1 至 ${maxLength} 個字元的文字`)
+  }
+  return value.trim()
+}
+
+function optionalInteger(
+  args: Record<string, unknown>,
+  name: string,
+  minimum: number,
+  maximum: number,
+  fallback: number,
+): number {
+  const value = args[name]
+  if (value === undefined) return fallback
+  if (typeof value !== 'number' || !Number.isInteger(value) || value < minimum || value > maximum) {
+    throw new Error(`參數 ${name} 必須是 ${minimum} 至 ${maximum} 的整數`)
+  }
+  return value
+}
+
+function parseCriteria(args: Record<string, unknown>): SearchCriteria {
+  const keyword = optionalString(args, 'keyword', 128)
+  const court = optionalString(args, 'court', 80)
+  const year = optionalInteger(args, 'year', 1, 9999, 0)
+  const caseNumber = optionalString(args, 'caseNumber', 80)
+  const limit = optionalInteger(args, 'limit', 1, PAGE_SIZE, 5)
+  const page = optionalInteger(args, 'page', 1, MAX_PAGE, 1)
+  const offset = optionalInteger(args, 'offset', 0, SOURCE_PAGE_SIZE - 1, 0)
+  if (!keyword && !court && !caseNumber && year === 0) {
+    throw new Error('至少需要提供 keyword、court、year 或 caseNumber 其中一項')
+  }
+  return {
+    ...(keyword ? { keyword } : {}),
+    ...(court ? { court } : {}),
+    ...(year > 0 ? { year } : {}),
+    ...(caseNumber ? { caseNumber } : {}),
+    limit,
+    page,
+    offset,
+  }
+}
+
+function parseCaseNumber(value: string): CaseNumberParts {
+  const withYear = value.match(/^(\d{1,4})年度(.+?)字第([A-Za-z0-9-]+)號$/)
+  if (withYear) {
+    const year = Number(withYear[1])
+    if (year < 1 || year > 9999) throw new Error('案號年度不在有效範圍')
+    return { year, caseWord: withYear[2], number: withYear[3] }
+  }
+  const withCaseWord = value.match(/^(.+?)字第([A-Za-z0-9-]+)號$/)
+  if (withCaseWord) return { caseWord: withCaseWord[1], number: withCaseWord[2] }
+  if (/^[A-Za-z0-9-]+$/.test(value)) return { number: value }
+  throw new Error('caseNumber 請使用案號數字或「年度案類字第號」格式')
+}
+
+function toTaiwanYear(year: number): number {
+  return year > 1911 ? year - 1911 : year
+}
+
+function normalizeCourt(value: string): string {
+  return value.replaceAll('台', '臺').replaceAll(/\s+/g, '').toLocaleLowerCase()
+}
+
+function readCourtCode(html: string, court: string): string {
+  const $ = load(html)
+  let code: string | undefined
+  const target = normalizeCourt(court)
+  $('select[name="jud_court"] option').each((_, option) => {
+    const candidate = $(option)
+    const value = candidate.attr('value')?.trim()
+    const label = candidate.text().trim()
+    if (value && (value === court || normalizeCourt(label) === target)) code = value
+  })
+  if (!code) throw new Error(`官方法院清單中找不到「${court}」，請使用司法院顯示的法院名稱`)
+  return code
+}
+
+function readCookieHeader(cookies: string[] | undefined): string | undefined {
+  if (!cookies) return undefined
+  const values = cookies
+    .map(cookie => cookie.split(';', 1)[0]?.trim())
+    .filter((cookie): cookie is string => Boolean(cookie && cookie.includes('=')))
+  return values.length > 0 ? values.join('; ') : undefined
+}
+
+function formFields(html: string, criteria: SearchCriteria): URLSearchParams {
+  const $ = load(html)
+  const form = $('form').first()
+  const state = new URLSearchParams()
+  if (!form.length || !form.find('input[name="__VIEWSTATE"]').length || !form.find('input[name="__EVENTVALIDATION"]').length) {
+    throw new JudicialSourceError('source_changed', SEARCH_URL.href)
+  }
+  form.find('input[type="hidden"][name]').each((_, input) => {
+    const name = $(input).attr('name')
+    const value = $(input).attr('value') ?? ''
+    if (name) state.set(name, value)
+  })
+  const submit = form.find('input[type="submit"][name], button[type="submit"][name]').first()
+  const submitName = submit.attr('name')
+  const submitValue = submit.attr('value') ?? submit.text().trim()
+  if (!submitName) throw new JudicialSourceError('source_changed', SEARCH_URL.href)
+  state.set('judtype', 'JUDBOOK')
+  state.set('whosub', '0')
+  state.set(submitName, submitValue)
+  if (criteria.keyword) state.set('jud_kw', criteria.keyword)
+  if (criteria.court) state.set('jud_court', readCourtCode(html, criteria.court))
+  if (criteria.year) state.set('jud_year', String(toTaiwanYear(criteria.year)))
+  if (criteria.caseNumber) {
+    const parsed = parseCaseNumber(criteria.caseNumber)
+    if (parsed.year && criteria.year && toTaiwanYear(criteria.year) !== toTaiwanYear(parsed.year)) {
+      throw new Error('year 與 caseNumber 的年度不一致')
+    }
+    const queryYear = criteria.year ?? parsed.year
+    if (queryYear) state.set('jud_year', String(toTaiwanYear(queryYear)))
+    if (parsed.caseWord) state.set('jud_case', parsed.caseWord)
+    state.set('jud_no', parsed.number)
+  }
+  return state
+}
+
+function allowedOfficialUrl(value: string, base: URL): URL {
+  const url = new URL(value, base)
+  if (
+    url.origin !== base.origin ||
+    url.username.length > 0 ||
+    url.password.length > 0 ||
+    url.hash.length > 0 ||
+    !url.pathname.startsWith('/FJUD/')
+  ) {
+    throw new JudicialSourceError('source_changed', base.href)
+  }
+  return url
+}
+
+function requestOptions(cookie?: string) {
+  return {
+    timeout: 15_000,
+    maxContentLength: REQUEST_LIMIT,
+    maxBodyLength: REQUEST_LIMIT,
+    maxRedirects: 0,
+    validateStatus: (status: number) => status >= 200 && status < 300,
+    responseType: 'text' as const,
+    headers: {
+      'User-Agent': 'judicial-mcp/1.5 (https://github.com/jurislm/judicial-mcp)',
+      ...(cookie ? { Cookie: cookie } : {}),
+    },
+  }
+}
+
+function upstreamFailure(error: unknown, url: URL): never {
+  let status: number | undefined
+  if (error !== null && typeof error === 'object' && 'response' in error) {
+    const response = error.response
+    if (response !== null && typeof response === 'object' && 'status' in response && typeof response.status === 'number') {
+      status = response.status
+    }
+  }
+  const code = status === 404 && url.pathname.endsWith('/data.aspx') ? 'source_not_found' : 'source_unavailable'
+  throw new JudicialSourceError(code, url.href)
+}
+
+function sourceText(value: unknown, url: URL): string {
+  if (typeof value !== 'string') throw new JudicialSourceError('source_changed', url.href)
+  return value
+}
+
+function visibleBodyText(html: string): string {
+  const $ = load(html)
+  $('script, style, noscript').remove()
+  return $('body').text()
+}
+
+function resultDate(text: string, caseId: string): string | null {
+  const dateMatch = text.match(/(\d{2,4})[./年-](\d{1,2})[./月-](\d{1,2})/)
+  const idMatch = caseId.split(',')[4]?.match(/^(\d{4})(\d{2})(\d{2})$/)
+  const match = dateMatch ?? idMatch
+  if (!match) return null
+  const rawYear = Number(match[1])
+  const year = rawYear < 1912 ? rawYear + 1911 : rawYear
+  const month = Number(match[2])
+  const day = Number(match[3])
+  const parsed = new Date(Date.UTC(year, month - 1, day))
+  if (
+    month < 1 ||
+    month > 12 ||
+    day < 1 ||
+    parsed.getUTCFullYear() !== year ||
+    parsed.getUTCMonth() !== month - 1 ||
+    parsed.getUTCDate() !== day
+  ) return null
+  return `${String(year).padStart(4, '0')}-${String(month).padStart(2, '0')}-${String(day).padStart(2, '0')}`
+}
+
+function caseIdFromLink(href: string, base: URL): string | undefined {
+  const url = allowedOfficialUrl(href, base)
+  if (url.pathname !== '/FJUD/data.aspx' || url.searchParams.get('ty') !== 'JD') return undefined
+  const caseId = url.searchParams.get('id')
+  if (!caseId || !/^[A-Za-z0-9]{2,8},\d{1,4},[^,<>/?#]+,[A-Za-z0-9-]+,\d{8}(?:,\d+)?$/.test(caseId)) return undefined
+  return caseId
+}
+
+function parseResults(html: string, sourceUrl: URL, retrievedAt: string): JudgmentSearchResult[] {
+  const $ = load(html)
+  const noResults = NO_RESULTS.test(visibleBodyText(html))
+  const table = $('table#jud, table.jub-table').first()
+  if (!table.length) {
+    if (noResults) return []
+    throw new JudicialSourceError('source_changed', sourceUrl.href)
+  }
+  const results: JudgmentSearchResult[] = []
+  table.find('tr').each((_, row) => {
+    const currentRow = $(row)
+    if (currentRow.hasClass('summary')) return
+    const link = currentRow.find('a[href*="data.aspx"]').first()
+    const href = link.attr('href')
+    if (!href) return
+    const caseId = caseIdFromLink(href, sourceUrl)
+    if (!caseId) throw new JudicialSourceError('source_changed', sourceUrl.href)
+    const cells = currentRow.find('td')
+    if (cells.length < 4) throw new JudicialSourceError('source_changed', sourceUrl.href)
+    const summaryRow = currentRow.next('tr.summary')
+    const rawSummary = summaryRow.find('span.tdCut').text().trim()
+    const summary = rawSummary.slice(0, MAX_SUMMARY_LENGTH)
+    const date = resultDate(cells.eq(2).text(), caseId)
+    const candidate: JudgmentSearchResult = {
+      case_id: caseId,
+      case_number: link.text().trim(),
+      date,
+      court: caseId.split(',')[0] ?? '',
+      cause: cells.eq(3).text().trim(),
+      source_url: judgmentUrl(caseId).href,
+      search_source_url: sourceUrl.href,
+      retrieved_at: retrievedAt,
+      cache: { status: 'not_cached' },
+      truncated: rawSummary.length > MAX_SUMMARY_LENGTH,
+      partial: false,
+      source_version: null,
+    }
+    if (summary) candidate.summary = summary
+    results.push(candidate)
+  })
+  if (results.length === 0 && !noResults) throw new JudicialSourceError('source_changed', sourceUrl.href)
+  return results
+}
+
+function hasNextPage(html: string, page: number): boolean {
+  const $ = load(html)
+  let found = false
+  $('a[href]').each((_, anchor) => {
+    const link = $(anchor)
+    const label = link.text().trim()
+    const href = link.attr('href') ?? ''
+    if (/^(?:下一頁|下頁|next)$/i.test(label) || new URLSearchParams(href.split('?')[1] ?? '').get('page') === String(page + 1)) {
+      found = true
+    }
+  })
+  return found
+}
+
+function makeSearchResultUrl(iframeSource: string, page: number): URL {
+  const url = allowedOfficialUrl(iframeSource, SEARCH_URL)
+  if (url.pathname !== '/FJUD/qryresultlst.aspx') throw new JudicialSourceError('source_changed', SEARCH_URL.href)
+  if (page > 1) url.searchParams.set('page', String(page))
+  return url
+}
+
+function makeSearchErrorPageUrl(iframeSource: string): URL | undefined {
+  const url = new URL(iframeSource, SEARCH_URL)
+  if (
+    url.origin !== SEARCH_URL.origin ||
+    url.pathname !== '/ErrorPage.aspx' ||
+    url.username.length > 0 ||
+    url.password.length > 0 ||
+    url.hash ||
+    url.href.length > 2_048
+  ) return undefined
+  const parameters = [...url.searchParams.entries()]
+  if (
+    parameters.length !== 2 ||
+    url.searchParams.getAll('frm').length !== 1 ||
+    url.searchParams.getAll('err').length !== 1 ||
+    !['FJUD', 'iframe'].includes(url.searchParams.get('frm') ?? '') ||
+    !/^[A-Za-z0-9_-]{1,64}$/.test(url.searchParams.get('err') ?? '')
+  ) return undefined
+  return url
+}
+
+function emptySearchResult(sourceUrl: URL, retrievedAt: string) {
+  return {
+    count: 0,
+    ambiguous: false,
+    results: [],
+    source_url: sourceUrl.href,
+    retrieved_at: retrievedAt,
+    cache: { status: 'not_cached' as const },
+    truncated: false,
+    partial: false,
+    next_page: null,
+    next_offset: null,
+    source_version: null,
+  }
+}
+
+function validCaseId(value: string): boolean {
+  return /^[A-Za-z0-9]{2,8},\d{1,4},[^,<>/?#]+,[A-Za-z0-9-]+,\d{8}(?:,\d+)?$/.test(value)
+}
+
+function judgmentUrl(caseId: string): URL {
+  const url = new URL(DETAIL_URL.href)
+  url.searchParams.set('ty', 'JD')
+  url.searchParams.set('id', caseId)
+  return url
+}
+
+export async function searchJudgments(args: Record<string, unknown>): Promise<unknown> {
+  const criteria = parseCriteria(args)
+  const retrievedAt = new Date().toISOString()
+  let formResponse: AxiosResponse<string>
+  try {
+    formResponse = await axios.get<string>(SEARCH_URL.href, requestOptions())
+  } catch (error: unknown) {
+    upstreamFailure(error, SEARCH_URL)
+  }
+  const formHtml = sourceText(formResponse.data, SEARCH_URL)
+  const cookie = readCookieHeader(formResponse.headers['set-cookie'])
+  const fields = formFields(formHtml, criteria)
+  let queryResponse: AxiosResponse<string>
+  try {
+    queryResponse = await axios.post<string>(SEARCH_URL.href, fields, {
+      ...requestOptions(cookie),
+      headers: {
+        ...requestOptions(cookie).headers,
+        'Content-Type': 'application/x-www-form-urlencoded',
+        Referer: SEARCH_URL.href,
+      },
+    })
+  } catch (error: unknown) {
+    upstreamFailure(error, SEARCH_URL)
+  }
+  const queryHtml = sourceText(queryResponse.data, SEARCH_URL)
+  const $ = load(queryHtml)
+  const iframeSource = $('iframe').first().attr('src')
+  if (!iframeSource) {
+    if (NO_RESULTS.test(visibleBodyText(queryHtml))) {
+      return emptySearchResult(SEARCH_URL, retrievedAt)
+    }
+    throw new JudicialSourceError('source_changed', SEARCH_URL.href)
+  }
+  const errorPageUrl = makeSearchErrorPageUrl(iframeSource)
+  if (errorPageUrl) {
+    let errorPageResponse: AxiosResponse<string>
+    try {
+      errorPageResponse = await axios.get<string>(errorPageUrl.href, requestOptions())
+    } catch (error: unknown) {
+      upstreamFailure(error, errorPageUrl)
+    }
+    const errorPageHtml = sourceText(errorPageResponse.data, errorPageUrl)
+    if (!NO_RESULTS.test(visibleBodyText(errorPageHtml))) {
+      throw new JudicialSourceError('source_changed', errorPageUrl.href)
+    }
+    return emptySearchResult(errorPageUrl, retrievedAt)
+  }
+  const resultUrl = makeSearchResultUrl(iframeSource, criteria.page)
+  let resultResponse: AxiosResponse<string>
+  try {
+    resultResponse = await axios.get<string>(resultUrl.href, requestOptions(cookie))
+  } catch (error: unknown) {
+    upstreamFailure(error, resultUrl)
+  }
+  const resultsHtml = sourceText(resultResponse.data, resultUrl)
+  const allResults = parseResults(resultsHtml, resultUrl, retrievedAt)
+  const sourceHasNextPage = hasNextPage(resultsHtml, criteria.page)
+  const pageHasNext = sourceHasNextPage && criteria.page < MAX_PAGE
+  const pageResults = allResults.slice(criteria.offset, criteria.offset + criteria.limit)
+  const hasMoreOnPage = allResults.length > criteria.offset + criteria.limit
+  const truncated = hasMoreOnPage || sourceHasNextPage || pageResults.some(result => result.truncated)
+  const ambiguous = Boolean(
+    criteria.caseNumber && (allResults.length > 1 || criteria.page > 1 || criteria.offset > 0 || sourceHasNextPage),
+  )
+  const hasMoreResults = hasMoreOnPage || sourceHasNextPage
+  const hasNextPageResult = hasMoreOnPage || pageHasNext
+  return {
+    count: pageResults.length,
+    ambiguous,
+    results: pageResults,
+    source_url: resultUrl.href,
+    retrieved_at: retrievedAt,
+    cache: { status: 'not_cached' },
+    truncated,
+    partial: hasMoreResults && !hasNextPageResult,
+    next_page: hasMoreOnPage ? criteria.page : pageHasNext ? criteria.page + 1 : null,
+    next_offset: hasMoreOnPage ? criteria.offset + criteria.limit : pageHasNext ? 0 : null,
+    source_version: null,
+  }
+}
+
+export async function fetchJudgmentText(args: Record<string, unknown>): Promise<unknown> {
+  const caseId = optionalString(args, 'caseId', 180)
+  if (!caseId || !validCaseId(caseId)) throw new Error('caseId 必須是司法院裁判書來源提供的裁判 ID')
+  const url = judgmentUrl(caseId)
+  let response: AxiosResponse<string>
+  try {
+    response = await axios.get<string>(url.href, requestOptions())
+  } catch (error: unknown) {
+    upstreamFailure(error, url)
+  }
+  const html = sourceText(response.data, url)
+  const $ = load(html)
+  const noResults = NO_RESULTS.test(visibleBodyText(html))
+  const judgment = $('#jud').first()
+  if (!judgment.length) {
+    throw new JudicialSourceError(noResults ? 'source_not_found' : 'source_changed', url.href)
+  }
+  judgment.find('br').replaceWith('\n')
+  judgment.find('div, p, tr').each((_, element) => {
+    $(element).append('\n')
+  })
+  const fullText = judgment.text()
+    .replace(/\r/g, '')
+    .replace(/[ \t]+\n/g, '\n')
+    .replace(/\n{3,}/g, '\n\n')
+    .trim()
+  if (!fullText) throw new JudicialSourceError(noResults ? 'source_not_found' : 'source_changed', url.href)
+  const truncated = fullText.length > MAX_TEXT_LENGTH
+  return {
+    case_id: caseId,
+    date: resultDate('', caseId),
+    full_text: fullText.slice(0, MAX_TEXT_LENGTH),
+    source_url: url.href,
+    retrieved_at: new Date().toISOString(),
+    cache: { status: 'not_cached' },
+    truncated,
+    partial: truncated,
+    source_version: null,
+  }
+}

package/src/regulation-cache.ts

@@ -0,0 +1,38 @@
+export interface RegulationCacheEntry {
+  value: unknown
+  expiresAt: number
+}
+
+const CACHE_TTL_MS = 15 * 60 * 1000
+const MAX_CACHE_ENTRIES = 100
+const entries = new Map<string, RegulationCacheEntry>()
+
+export function getRegulationCacheEntry(key: string): RegulationCacheEntry | undefined {
+  const entry = entries.get(key)
+  if (!entry) return undefined
+  entries.delete(key)
+  entries.set(key, entry)
+  return entry
+}
+
+export function setRegulationCacheEntry(
+  key: string,
+  entry: Omit<RegulationCacheEntry, 'expiresAt'>,
+  now = Date.now(),
+): void {
+  entries.delete(key)
+  entries.set(key, { ...entry, expiresAt: now + CACHE_TTL_MS })
+  while (entries.size > MAX_CACHE_ENTRIES) {
+    const oldestKey = entries.keys().next().value
+    if (oldestKey === undefined) return
+    entries.delete(oldestKey)
+  }
+}
+
+export function deleteRegulationCacheEntry(key: string): void {
+  entries.delete(key)
+}
+
+export function clearRegulationCache(): void {
+  entries.clear()
+}