minimal-agent 0.13.0 → 0.14.0

package/workflows/wrong-question-docx/workflow.mjs

@@ -0,0 +1,485 @@
+export const meta = {
+  apiVersion: 2,
+  description: '调用question-crops分题，只抄明确指定题号，生成可编辑Word及PDF',
+  dependencies: ['question-crops'],
+  inputs: [
+    {
+      name: 'source',
+      type: 'resource',
+      required: true,
+      greedy: true,
+      accepts: ['image/*'],
+      allowFile: true,
+      allowDirectory: false,
+      multiple: false,
+      maxItems: 1,
+      attachmentFallback: true,
+    },
+    {
+      name: 'questions',
+      description: '必须指定题号，如5,9,13-16或章节:题号；全部为all',
+    },
+    {
+      name: 'answer_space',
+      type: 'enum',
+      values: ['compact', 'standard', 'generous'],
+      default: 'standard',
+    },
+    { name: 'vision_timeout_seconds', type: 'number', default: 900 },
+  ],
+  artifacts: [],
+  resumeBackoff: 0,
+};
+
+export default async function run(wf) {
+const { values: args } = wf.input;
+const { sessionId: session, taskId: task, workflowDir } = wf.runtime;
+const { generate: llm, require: requireModel } = wf.model;
+const { call: tool } = wf.tools;
+const { file: execFile } = wf.exec;
+const { readText, readJson } = wf.fs;
+const { ask, child: workflow } = wf.flow;
+const echo = wf.output.message;
+const declareArtifact = wf.output.artifact;
+const log = (message) => wf.output.progress({ phase: 'workflow', status: 'update', message: String(message) });
+
+let ROOT = `out/wrong-question-docx/${session}/${task}`;
+let seq = 0;
+const timeout = Number(args.vision_timeout_seconds ?? 900);
+if (!Number.isFinite(timeout) || timeout < 60 || timeout > 3600)
+  throw new Error('视觉调用时限必须为60–3600秒');
+const fatal = (e) =>
+  /AbortError|aborted|cancel|用户中断|EACCES|EPERM|ENOSPC|EDQUOT|EIO|EROFS|ENOENT|MODULE_NOT_FOUND|Cannot find module|权限|permission/i.test(
+    String(e),
+  );
+const manifest = {
+  version: 2,
+  status: 'running',
+  questions: [],
+  calls: [],
+  answerSpace: args.answer_space ?? 'standard',
+  contentVerified: false,
+};
+async function write(path, value) {
+  declareArtifact(path);
+  await tool('Write', {
+    file_path: path,
+    content: JSON.stringify(value, null, 2),
+  });
+}
+async function command(action, payload) {
+  const job = `${ROOT}/qa/job-${++seq}.json`;
+  await write(job, payload);
+  if (action === 'prepare')
+    for (const p of [
+      `questions/${payload.question.id}-faithful.png`,
+      `qa/${payload.question.id}-preview.png`,
+      `qa/${payload.question.id}-faithful-preview.png`,
+    ])
+      declareArtifact(`${ROOT}/${p}`);
+  if (action === 'crop')
+    for (const [i] of payload.figures.entries())
+      for (const p of [
+        `figures/${payload.id}-${i + 1}-a${payload.attempt}.png`,
+        `qa/${payload.id}-${i + 1}-a${payload.attempt}.png`,
+        `qa/${payload.id}-${i + 1}-a${payload.attempt}-box.png`,
+      ])
+        declareArtifact(`${ROOT}/${p}`);
+  if (action === 'build') declareArtifact(`${ROOT}/wrong-questions.docx`);
+  if (action === 'render') {
+    declareArtifact(`${ROOT}/wrong-questions.pdf`);
+    declareArtifact(`${ROOT}/qa/rendered/page-*.png`);
+  }
+  const r = await execFile('./scripts/workbook.mjs', [action, job]);
+  if (r.exitCode !== 0) throw new Error(r.stderr || r.stdout);
+  const value = JSON.parse(r.stdout.trim());
+  for (const path of value?.artifacts ?? []) declareArtifact(path);
+  return value;
+}
+ROOT = await command('absolute', { path: ROOT });
+async function report() {
+  manifest.success = manifest.questions.filter(
+    (q) => q.status === 'success',
+  ).length;
+  manifest.partial = manifest.questions.filter(
+    (q) => q.status === 'partial',
+  ).length;
+  manifest.missing = manifest.questions.filter(
+    (q) => q.status === 'missing',
+  ).length;
+  manifest.ready = manifest.success + manifest.partial;
+  await write(`${ROOT}/manifest.json`, manifest);
+  declareArtifact(`${ROOT}/report.md`);
+  await tool('Write', {
+    file_path: `${ROOT}/report.md`,
+    content:
+      '# 指定题号错题本\n\n' +
+      `状态：${manifest.status}；请求${manifest.questions.length}题，完整${manifest.success}题，局部待补${manifest.partial}题，无法整理${manifest.missing}题。\n\n` +
+      manifest.questions
+        .map(
+          (q) =>
+            `- 第${q.number}题（${q.section}）：${q.status} ${q.reason || ''}`,
+        )
+        .join('\n') +
+      '\n\n只抄目标题。流程完成不等于人工内容验收通过。\n' +
+      (manifest.error || ''),
+  });
+}
+const str = { type: 'string' };
+const box = {
+  type: 'object',
+  required: ['x', 'y', 'width', 'height'],
+  properties: {
+    x: { type: 'number', minimum: 0, maximum: 1000 },
+    y: { type: 'number', minimum: 0, maximum: 1000 },
+    width: { type: 'number', minimum: 1, maximum: 1000 },
+    height: { type: 'number', minimum: 1, maximum: 1000 },
+  },
+  additionalProperties: false,
+};
+const detailSchema = {
+  type: 'object',
+  required: [
+    'question_id',
+    'observed_number',
+    'status',
+    'reason',
+    'type',
+    'blocks',
+    'figures',
+    'missing_parts',
+  ],
+  properties: {
+    question_id: str,
+    observed_number: str,
+    status: { enum: ['ready', 'partial', 'unclear', 'missing', 'ambiguous'] },
+    reason: str,
+    type: { enum: ['short', 'long'] },
+    blocks: {
+      type: 'array',
+      items: {
+        type: 'object',
+        required: ['kind', 'text'],
+        properties: {
+          kind: { enum: ['text', 'option', 'formula'] },
+          text: str,
+        },
+        additionalProperties: false,
+      },
+    },
+    figures: {
+      type: 'array',
+      items: {
+        type: 'object',
+        required: ['id', 'kind', 'box', 'description'],
+        properties: {
+          id: str,
+          kind: { enum: ['figure', 'formula'] },
+          box,
+          description: str,
+        },
+        additionalProperties: false,
+      },
+    },
+    missing_parts: {
+      type: 'array',
+      items: {
+        type: 'object',
+        required: ['kind', 'reason', 'source_limited'],
+        properties: {
+          kind: { enum: ['text', 'option', 'figure', 'continuation'] },
+          reason: str,
+          source_limited: { type: 'boolean' },
+        },
+        additionalProperties: false,
+      },
+    },
+  },
+  additionalProperties: false,
+};
+const qaSchema = {
+  type: 'object',
+  required: ['groups'],
+  properties: {
+    groups: {
+      type: 'array',
+      items: {
+        type: 'object',
+        required: [
+          'id',
+          'belongs_to_target',
+          'complete',
+          'cut_edges',
+          'reason',
+          'corrected_box',
+        ],
+        properties: {
+          id: str,
+          belongs_to_target: { type: 'boolean' },
+          complete: { type: 'boolean' },
+          cut_edges: { type: 'boolean' },
+          reason: str,
+          corrected_box: box,
+        },
+        additionalProperties: false,
+      },
+    },
+  },
+  additionalProperties: false,
+};
+async function observe(prompt, schema, images, label) {
+  const call = { label, status: 'running' };
+  manifest.calls.push(call);
+  try {
+    const value = await llm(prompt, {
+      schema,
+      images: images.map((source) => ({ source, detail: 'high' })),
+      isolated: true,
+      model: 'vision',
+      timeoutMs: timeout * 1000,
+      maxTokens: 8192,
+      label,
+    });
+    call.status = 'returned';
+    return value;
+  } catch (e) {
+    call.status = 'failed';
+    call.reason = String(e);
+    throw e;
+  }
+}
+function target(q) {
+  return (
+    '\n目标绑定：' +
+    JSON.stringify({
+      question_id: q.sourceId,
+      printed_number: q.number,
+      section: q.section,
+      content_status: q.content_status ?? 'legacy_unknown',
+      missing_parts: q.missing_parts ?? [],
+      warnings: q.warnings ?? [],
+      fragments: q.fragments.map(f => ({ order: f.order, sourceBox: f.sourceBox, canvas: f.canvas })),
+    }) +
+    '\n本次只处理该题，忽略所有邻题及其配图。'
+  );
+}
+
+try {
+  let requested = String(args.questions ?? '').trim();
+  if (!requested && args.requestText) {
+    const intent = await llm(
+      '只从用户要求提取明确要抄录的题号，不能自行判错题。清晰指定全部才all=true；未指定则items=[]。章节不明确留空。用户要求：' +
+        args.requestText,
+      {
+        isolated: true,
+        schema: {
+          type: 'object',
+          required: ['all', 'items'],
+          properties: {
+            all: { type: 'boolean' },
+            items: {
+              type: 'array',
+              items: {
+                type: 'object',
+                required: ['section', 'number'],
+                properties: { section: str, number: str },
+                additionalProperties: false,
+              },
+            },
+          },
+          additionalProperties: false,
+        },
+        label: '提取用户指定题号',
+      },
+    );
+    requested = intent.all
+      ? 'all'
+      : intent.items
+          .map((i) => (i.section ? i.section + ':' : '') + i.number)
+          .join(',');
+  }
+  if (!requested)
+    requested = await ask(
+      '请指定要整理的题号，例如5,9,13-16；或明确输入全部。不会自动判断错题。',
+      [
+        { label: '等待指定题号 (推荐)', description: '停止本次，补充题号后重跑' },
+        { label: '全部', description: '抄录全部题目' },
+      ],
+    );
+  if (!requested || requested.startsWith('等待指定题号')) {
+    manifest.status = 'awaiting_selection';
+    await report();
+    echo(
+      '请指定题号后重跑：/workflow wrong-question-docx <图片> --input questions=5,9,13-16',
+    );
+    return { status: manifest.status, report: `${ROOT}/report.md` };
+  }
+  const selection = await command('selection', { text: requested });
+  if (!selection.all && !selection.items.length)
+    throw new Error('没有指定题号');
+  await requireModel('vision', { vision: true });
+  const child = await workflow(
+    'question-crops',
+    { source: args.source, vision_timeout_seconds: timeout },
+    { label: '试卷逐题切图' },
+  );
+  manifest.child = child;
+  const source = await readJson(child.result?.manifest);
+  if (!source || !Array.isArray(source.questions))
+    throw new Error('子流程没有返回有效题目清单');
+  manifest.questions = await command('select', {
+    manifest: source,
+    selection,
+  });
+  echo(
+    `请求${manifest.questions.length}题：${manifest.questions.map((q) => q.number).join('、')}；找到${manifest.questions.filter((q) => q.status === 'pending').length}题。只抄指定题号。`,
+  );
+  await report();
+  const transcribe = await readText(`${workflowDir}/prompts/transcribe.md`),
+    figureQA = await readText(`${workflowDir}/prompts/figure-qa.md`);
+  if (!transcribe || !figureQA) throw new Error('缺少抄录或配图提示词');
+  for (const q of manifest.questions) {
+    if (q.status !== 'pending') continue;
+    log(`抄录第${q.number}题（${q.sourceId}），仅处理该题及所属配图`);
+    try {
+      const prepared = await command('prepare', { root: ROOT, question: q });
+      const binding = target(q);
+      let detail = await observe(
+        transcribe + binding,
+        detailSchema,
+        [prepared.enhanced.path],
+        `抄录第${q.number}题`,
+      );
+      await write(`${ROOT}/questions/${q.id}-initial.json`, detail);
+      if (detail.status === 'unclear') {
+        detail = await observe(
+          transcribe +
+            binding +
+            '\n本次改用未经增强的原题预览辨认模糊符号；只抄目标题。此前问题：' +
+            detail.reason,
+          detailSchema,
+          [prepared.plain.path],
+          `第${q.number}题原貌重读`,
+        );
+      }
+      const check = await command('detail', { question: q, detail });
+      if (check.reason) throw new Error(check.reason);
+      detail.figures = check.figures;
+      await write(`${ROOT}/questions/${q.id}-detail.json`, detail);
+      let figures = detail.figures,
+        accepted = [],
+        componentIssues = [...check.missingParts];
+      if (figures.length)
+        try {
+          for (let attempt = 1; attempt <= 3; attempt++) {
+          const crops = await command('crop', {
+            root: ROOT,
+            id: q.id,
+            base: prepared.faithful,
+            geometry: prepared.plain,
+            figures,
+            attempt,
+          });
+          const qa = await observe(
+            figureQA +
+              binding +
+              '\n目标题干：' +
+              detail.blocks.map((b) => b.text).join('\n') +
+              '\ngroups：' +
+              JSON.stringify(figures),
+            qaSchema,
+            crops.figures.flatMap((f) => [f.overlay, f.preview]),
+            `第${q.number}题配图归属与完整性${attempt}`,
+          );
+          await write(`${ROOT}/qa/${q.id}-figures-${attempt}.json`, qa);
+          if (
+            qa.groups.length !== figures.length ||
+            qa.groups.some((g, i) => g.id !== figures[i].id)
+          )
+            throw new Error('配图验收返回图组不匹配');
+          if (
+            qa.groups.every(
+              (g) => g.belongs_to_target && g.complete && !g.cut_edges,
+            )
+          ) {
+            accepted = crops.figures;
+            break;
+          }
+            if (attempt === 3)
+            throw new Error(
+              '配图不完整或归属不明：' +
+                qa.groups.map((g) => g.reason).join('；'),
+            );
+          const next = figures.map((f, i) => ({
+            ...f,
+            box: qa.groups[i].corrected_box,
+          }));
+          if (figures.every((f, i) => ['x', 'y', 'width', 'height'].every(k => f.box[k] === next[i].box[k])))
+            throw new Error('没有可靠的新配图框，不重复同图验收');
+            figures = next;
+          }
+        } catch (figureError) {
+          if (fatal(figureError)) throw figureError;
+          accepted = [];
+          componentIssues.push({
+            kind: 'figure',
+            reason: String(figureError.message ?? figureError),
+            source_limited: false,
+          });
+        }
+      Object.assign(q, {
+        status: componentIssues.length ? 'partial' : 'success',
+        type: detail.type,
+        blocks: detail.blocks,
+        figures: accepted,
+        missingParts: componentIssues,
+        reason: componentIssues.map((part) => part.reason).join('；'),
+      });
+    } catch (e) {
+      if (fatal(e)) throw e;
+      q.status = 'missing';
+      q.reason = String(e.message ?? e);
+      q.blocks = [];
+      q.figures = [];
+      log(`第${q.number}题待补全：${q.reason}`);
+    }
+    await report();
+  }
+  manifest.status = manifest.questions.some((q) => q.status !== 'success')
+    ? 'partial'
+    : 'complete';
+  await report();
+  if (!manifest.ready) {
+    echo(`没有完整抄录的题目，仅提供清单与报告：${ROOT}/report.md`);
+    return {
+      status: 'partial',
+      manifest: `${ROOT}/manifest.json`,
+      report: `${ROOT}/report.md`,
+    };
+  }
+  const built = await command('build', { root: ROOT, manifest });
+  manifest.questions = built.questions;
+  const checked = await command('validate', { path: built.path, manifest });
+  await write(`${ROOT}/qa/docx-check.json`, checked);
+  const rendered = await command('render', { path: built.path, root: ROOT });
+  manifest.docx = built.path;
+  manifest.pdf = rendered.pdf;
+  manifest.renderedPages = rendered.pages;
+  await report();
+  echo(
+    `完整整理${manifest.success}题，局部待补${manifest.partial}题，无法整理${manifest.missing}题。\nWord：${built.path}\nPDF：${rendered.pdf}\n报告：${ROOT}/report.md`,
+  );
+  return {
+    status: manifest.status,
+    docx: built.path,
+    pdf: rendered.pdf,
+    manifest: `${ROOT}/manifest.json`,
+    report: `${ROOT}/report.md`,
+  };
+} catch (e) {
+  manifest.status = 'failed';
+  manifest.error = String(e);
+  await report();
+  throw e;
+}
+}