fix(teach): fix regexp matching for certain questions, fix #249

packages/plugin-eval/tests/index.spec.ts

@@ -82,6 +82,8 @@ describe('Eval Plugin', () => {
     await ses.shouldReply('demo', '123')
     await ses.shouldReply('# ^repeat:(.+) ${"$1".repeat(3)} -x', '问答已添加，编号为 2。')
     await ses.shouldReply('repeat:123', '123123123')
+    await ses.shouldReply('# ^我.+ 对，${"$0".replace(/我/, "你")} -x', '问答已添加，编号为 3。')
+    await ses.shouldReply('我是伞兵', '对，你是伞兵')
   })
 
   it('global', async () => {

packages/plugin-teach/src/database/mongo.ts

@@ -70,29 +70,24 @@ export default function apply(ctx: Context) {
   ctx.on('dialogue/mongo', ({ regexp, answer, question, original }, conditionals) => {
     if (regexp) {
       if (answer) conditionals.push({ answer: { $regex: new RegExp(answer, 'i') } })
-      if (question) conditionals.push({ question: { $regex: new RegExp(original, 'i') } })
+      if (original) conditionals.push({ original: { $regex: new RegExp(original, 'i') } })
       return
     }
     if (answer) conditionals.push({ answer })
-    if (question) {
-      if (regexp === false) {
-        conditionals.push({ question })
-      } else {
-        const $expr = {
-          body(field: string, question: string, original: string) {
-            const regex = new RegExp(field, 'i')
-            return regex.test(question) || regex.test(original)
-          },
-          args: ['$name', question, original],
-          lang: 'js',
-        }
-        conditionals.push({
-          $or: [
-            { flag: { $bitsAllClear: Dialogue.Flag.regexp }, question },
-            { flag: { $bitsAllSet: Dialogue.Flag.regexp }, $expr },
-          ],
-        })
+    if (regexp === false) {
+      if (question) conditionals.push({ question })
+    } else if (original) {
+      const $expr = {
+        body(field: string, original: string) {
+          const regex = new RegExp(field, 'i')
+          return regex.test(original)
+        },
+        args: ['$name', original],
+        lang: 'js',
       }
+      const conds = [{ flag: { $bitsAllSet: Dialogue.Flag.regexp }, $expr } as FilterQuery<Dialogue>]
+      if (question) conds.push({ flag: { $bitsAllClear: Dialogue.Flag.regexp }, question })
+      conditionals.push({ $or: conds })
     }
   })
 

packages/plugin-teach/src/database/mysql.ts

@@ -84,22 +84,17 @@ export default function apply(ctx: Context) {
 
     if (regexp) {
       if (answer) conditionals.push('`answer` REGEXP ' + escape(answer))
-      if (question) conditionals.push('`question` REGEXP ' + escape(original))
+      if (original) conditionals.push('`original` REGEXP ' + escape(original))
       return
     }
 
     if (answer) conditionals.push('`answer` = ' + escape(answer))
-    if (question) {
-      if (regexp === false) {
-        conditionals.push('`question` = ' + escape(question))
-      } else {
-        conditionals.push(`(\
-          !(\`flag\` & ${Dialogue.Flag.regexp}) && \`question\` = ${escape(question)} ||\
-          \`flag\` & ${Dialogue.Flag.regexp} && (\
-            ${escape(question)} REGEXP \`question\` || ${escape(original)} REGEXP \`question\`\
-          )\
-        )`)
-      }
+    if (regexp === false) {
+      if (question) conditionals.push('`question` = ' + escape(question))
+    } else if (original) {
+      const conds = [`\`flag\` & ${Dialogue.Flag.regexp} && ${escape(original)} REGEXP \`original\``]
+      if (question) conds.push(`!(\`flag\` & ${Dialogue.Flag.regexp}) && \`question\` = ${escape(question)}`)
+      conditionals.push(`(${conds.join(' || ')})`)
     }
   })
 

packages/plugin-teach/src/internal.ts

@@ -45,11 +45,11 @@ export default function apply(ctx: Context, config: Dialogue.Config) {
       } else if (/\[CQ:(?!face)/.test(question)) {
         return template('teach.prohibited-cq-code')
       }
-      const { unprefixed, prefixed, appellative } = options.regexp
-        ? { unprefixed: question, prefixed: question, appellative: false }
+      const { unprefixed, appellative } = options.regexp
+        ? { unprefixed: question, appellative: false }
         : config._stripQuestion(question)
       defineProperty(options, 'appellative', appellative)
-      defineProperty(options, '_original', prefixed)
+      defineProperty(options, '_original', question)
       defineProperty(options, 'original', question)
       args[0] = unprefixed
       args[1] = answer

packages/plugin-teach/src/receiver.ts

@@ -163,6 +163,10 @@ tokenizer.interpolate('$n', '', (rest) => {
   return { rest, tokens: [], source: '' }
 })
 
+const halfWidth = ',,.~?!()[]'
+const fullWidth = '，、。～？！（）【】'
+const fullWidthRegExp = new RegExp(`[${fullWidth}]`)
+
 export async function triggerDialogue(ctx: Context, session: Session, next: NextFunction = noop) {
   const state = ctx.getSessionState(session)
   state.next = next
@@ -203,8 +207,7 @@ export async function triggerDialogue(ctx: Context, session: Session, next: Next
     .replace(/\$0/g, escapeAnswer(session.content))
 
   if (dialogue.flag & Dialogue.Flag.regexp) {
-    const capture = dialogue._capture || new RegExp(dialogue.question, 'i').exec(state.test.question)
-    if (!capture) console.log(dialogue.question, state.test.question)
+    const capture = dialogue._capture || new RegExp(dialogue.original, 'i').exec(state.test.original)
     capture.map((segment, index) => {
       if (index && index <= 9) {
         state.answer = state.answer.replace(new RegExp(`\\$${index}`, 'g'), escapeAnswer(segment || ''))
@@ -245,7 +248,17 @@ export default function apply(ctx: Context, config: Dialogue.Config) {
   ctx.app._dialogueStates = {}
 
   config._stripQuestion = (source) => {
-    source = prepareSource(source)
+    source = segment.transform(source, {
+      text: ({ content }, index, chain) => {
+        let message = simplify(segment.unescape('' + content))
+          .toLowerCase()
+          .replace(/\s+/g, '')
+          .replace(fullWidthRegExp, $0 => halfWidth[fullWidth.indexOf($0)])
+        if (index === 0) message = message.replace(/^[()\[\]]*/, '')
+        if (index === chain.length - 1) message = message.replace(/[\.,?!()\[\]~]*$/, '')
+        return message
+      },
+    })
     const original = source
     const capture = nicknameRE.exec(source)
     if (capture) source = source.slice(capture[0].length)
@@ -297,12 +310,11 @@ export default function apply(ctx: Context, config: Dialogue.Config) {
 
   ctx.on('dialogue/receive', ({ session, test }) => {
     if (session.content.includes('[CQ:image,')) return true
-    const { unprefixed, prefixed, appellative, activated } = config._stripQuestion(session.content)
+    const { unprefixed, appellative, activated } = config._stripQuestion(session.content)
     test.question = unprefixed
-    test.original = prefixed
+    test.original = session.content
     test.activated = activated
     test.appellative = appellative
-    if (!test.question) return true
   })
 
   // 预判要获取的用户字段
@@ -322,27 +334,3 @@ export default function apply(ctx: Context, config: Dialogue.Config) {
       return triggerDialogue(ctx, session, next)
     })
 }
-
-function prepareSource(source: string) {
-  return segment.transform(source, {
-    text: ({ content }, index, chain) => {
-      let message = simplify(segment.unescape('' + content))
-        .toLowerCase()
-        .replace(/\s+/g, '')
-        .replace(/，/g, ',')
-        .replace(/、/g, ',')
-        .replace(/。/g, '.')
-        .replace(/？/g, '?')
-        .replace(/！/g, '!')
-        .replace(/（/g, '(')
-        .replace(/）/g, ')')
-        .replace(/【/g, '[')
-        .replace(/】/g, ']')
-        .replace(/～/g, '~')
-        .replace(/…/g, '...')
-      if (index === 0) message = message.replace(/^[()\[\]]*/, '')
-      if (index === chain.length - 1) message = message.replace(/[\.,?!()\[\]~]*$/, '')
-      return message
-    },
-  })
-}

packages/plugin-teach/src/search.ts

@@ -165,7 +165,7 @@ async function showSearch(argv: Dialogue.Argv) {
   const { regexp, page = 1, original, pipe, recursive, autoMerge } = options
   const { itemsPerPage = 30, mergeThreshold = 5 } = argv.config
 
-  const test: DialogueTest = { question, answer, regexp, original: options._original }
+  const test: DialogueTest = { question, answer, regexp, original }
   if (app.bail('dialogue/before-search', argv, test)) return
   const dialogues = await app.database.getDialoguesByTest(test)
 
@@ -182,14 +182,14 @@ async function showSearch(argv: Dialogue.Argv) {
     await argv.app.parallel('dialogue/search', argv, test, dialogues)
   }
 
-  if (!question && !answer) {
+  if (!original && !answer) {
     if (!dialogues.length) return '没有搜索到任何回答，尝试切换到其他环境。'
     return sendResult('全部问答如下', formatQuestionAnswers(argv, dialogues))
   }
 
   if (!options.regexp) {
     const suffix = options.regexp !== false ? '，请尝试使用正则表达式匹配' : ''
-    if (!question) {
+    if (!original) {
       if (!dialogues.length) return session.send(`没有搜索到回答“${answer}”${suffix}。`)
       const output = dialogues.map(d => `${formatPrefix(argv, d)}${d.original}`)
       return sendResult(`回答“${answer}”的问题如下`, output)
@@ -227,7 +227,7 @@ async function showSearch(argv: Dialogue.Argv) {
     })
   }
 
-  if (!question) {
+  if (!original) {
     if (!dialogues.length) return `没有搜索到含有正则表达式“${answer}”的回答。`
     return sendResult(`回答正则表达式“${answer}”的搜索结果如下`, output)
   } else if (!answer) {