tools/build-lab.mjs - dev - deka/buffer

tools/build-lab.mjs
189 строк11 KB
Maksim Ratnikov
feat: доска в Markdown — таблицы объектов
07 авг 2026, 12:01
07 авг 2026, 12:013f61bda
import { readFileSync, writeFileSync } from "node:fs";
import { fileURLToPath } from "node:url";
import { dirname, resolve, relative } from "node:path";
import { transformModule } from "./strip-module.mjs";

const __filename = fileURLToPath(import.meta.url);
const __dirname = dirname(__filename);
const ROOT = resolve(__dirname, "..");

const MODULE_PATHS = [
  "src/util/format.js",
  "src/converters/wiki-table.js",
  "src/converters/layout-renderer.js",
  "src/converters/goal-to-layout.js",
  "src/converters/goal-diff.js",
  "src/converters/markdown-to-wiki.js",
  "src/converters/html-to-markdown.js",
  "src/converters/board-text.js",
  "src/converters/board-payload.js",
  "src/converters/board-layout.js",
  "src/converters/board-to-layout.js",
  "src/converters/board-to-markdown.js"
];

const PROMPT_PATH = "prompts/wiki-layout-full.md";
// Промпт доски склеивается из двух частей прямо в инжектируемую константу.
// Третьего файла на диске нет намеренно: сборщик запускается из freshness-теста
// во временный каталог, и побочная запись в репозиторий сделала бы проверку
// мутирующей. Заодно не появляется третий расходящийся контракт.
// Порядок важен: сначала задача и инварианты, потом контракт. Иначе модель
// доходит до собственно задания к 60% текста.
const BOARD_PROMPT_PATHS = ["prompts/text-to-board-layout.md", "prompts/board-layout-reference.md"];
const MEET_PROTOCOL_PROMPT_PATH = "prompts/meet-protocol.md";
const MEETING_NARRATIVE_PROMPT_PATH = "prompts/meeting-narrative.md";
const KB_UPDATE_PROMPT_PATH = "prompts/knowledge-base-update.md";
const GOAL_EXTRACTION_PROMPT_PATH = "prompts/goal-extraction.system.md";
const KNOWLEDGE_BASE_LAYOUT_PATH = "inputs/knowledge-base.layout.json";
const PROTOCOL_EXAMPLE_MARKDOWN_PATH = "Протокол пример.txt";

const args = process.argv.slice(2);
const outIdx = args.indexOf("--out");
const OUT_PATH = outIdx >= 0
  ? resolve(args[outIdx + 1])
  : resolve(ROOT, "lab/wiki-clipboard-lab.html");
const TEMPLATE_PATH = resolve(ROOT, "lab/wiki-clipboard-lab.template.html");
const CONVERTERS_MARKER_RE = /^([ \t]*)\/\/\s*__INJECT_CONVERTERS__[ \t]*\r?\n/m;
const PROMPT_MARKER_RE = /^([ \t]*)\/\/\s*__INJECT_LLM_PROMPT__[ \t]*\r?\n/m;
const BOARD_PROMPT_MARKER_RE = /^([ \t]*)\/\/\s*__INJECT_BOARD_PROMPT__[ \t]*\r?\n/m;
const MEET_PROTOCOL_PROMPT_MARKER_RE = /^([ \t]*)\/\/\s*__INJECT_MEET_PROTOCOL_PROMPT__[ \t]*\r?\n/m;
const MEETING_NARRATIVE_PROMPT_MARKER_RE = /^([ \t]*)\/\/\s*__INJECT_MEETING_NARRATIVE_PROMPT__[ \t]*\r?\n/m;
const KB_UPDATE_PROMPT_MARKER_RE = /^([ \t]*)\/\/\s*__INJECT_KB_UPDATE_PROMPT__[ \t]*\r?\n/m;
const GOAL_EXTRACTION_PROMPT_MARKER_RE = /^([ \t]*)\/\/\s*__INJECT_GOAL_EXTRACTION_PROMPT__[ \t]*\r?\n/m;
const KNOWLEDGE_BASE_LAYOUT_MARKER_RE = /^([ \t]*)\/\/\s*__INJECT_KNOWLEDGE_BASE_LAYOUT__[ \t]*\r?\n/m;
const PROTOCOL_EXAMPLE_MARKDOWN_MARKER_RE = /^([ \t]*)\/\/\s*__INJECT_PROTOCOL_EXAMPLE_MARKDOWN__[ \t]*\r?\n/m;
const FONTS_MARKER_RE = /^([ \t]*)\/\*\s*__INJECT_FONTS__\s*\*\/[ \t]*\r?\n/m;
const FONTS_DIR = resolve(ROOT, "lab/fonts");

function readInlineFontFaces() {
  // lab/fonts/manifest.json (written by tools/fetch-fonts.mjs) describes every
  // WOFF2 file: family, weight (single value or a variable-font range like
  // "100 700") and the unicode-range of its subset. Per-subset unicode-range is
  // what keeps Cyrillic and Latin rendering with the same family instead of one
  // script silently falling back to a system font.
  let manifest;
  try {
    manifest = JSON.parse(readFileSync(resolve(FONTS_DIR, "manifest.json"), "utf8"));
  } catch (err) {
    if (err.code !== "ENOENT") throw err;
    return "";
  }
  const rules = [];
  for (const entry of manifest) {
    const { file, family, weight, range } = entry;
    if (!file || !family || !weight || !range) {
      console.error(`build-lab: lab/fonts/manifest.json entry is missing file/family/weight/range: ${JSON.stringify(entry)}`);
      process.exit(1);
    }
    const b64 = readFileSync(resolve(FONTS_DIR, file)).toString("base64");
    rules.push(
      `@font-face{font-family:'${family}';font-style:normal;font-weight:${weight};font-display:swap;src:url(data:font/woff2;base64,${b64}) format('woff2');unicode-range:${range};}`
    );
  }
  return rules.join("\n");
}

function assertNoLeftoverEsm(relPath, transformed) {
  const lines = transformed.split("\n");
  for (let i = 0; i < lines.length; i += 1) {
    const line = lines[i];
    if (/^\s*(import|export)\s/.test(line)) {
      console.error(`build-lab: ${relPath}:${i + 1} has unsupported ESM syntax that wasn't stripped:`);
      console.error(`  ${line}`);
      console.error(`build-lab supports only:`);
      console.error(`  - single-line: import { x } from "./y.js";`);
      console.error(`  - prefix:      export function|const|let|class NAME ...`);
      process.exit(1);
    }
  }
}

function assertSingleMarker(template, regex, label) {
  const matches = [...template.matchAll(new RegExp(regex.source, "gm"))];
  if (matches.length !== 1) {
    console.error(`build-lab: expected exactly one ${label} marker, found ${matches.length}`);
    process.exit(1);
  }
  return matches[0];
}

function replaceLiteral(source, regex, value) {
  return source.replace(regex, () => value);
}

function main() {
  const template = readFileSync(TEMPLATE_PATH, "utf8");

  const fontsMatch = assertSingleMarker(template, FONTS_MARKER_RE, "__INJECT_FONTS__");
  const promptMatch = assertSingleMarker(template, PROMPT_MARKER_RE, "__INJECT_LLM_PROMPT__");
  const boardPromptMatch = assertSingleMarker(template, BOARD_PROMPT_MARKER_RE, "__INJECT_BOARD_PROMPT__");
  const meetProtocolPromptMatch = assertSingleMarker(template, MEET_PROTOCOL_PROMPT_MARKER_RE, "__INJECT_MEET_PROTOCOL_PROMPT__");
  const meetingNarrativePromptMatch = assertSingleMarker(template, MEETING_NARRATIVE_PROMPT_MARKER_RE, "__INJECT_MEETING_NARRATIVE_PROMPT__");
  const kbUpdatePromptMatch = assertSingleMarker(template, KB_UPDATE_PROMPT_MARKER_RE, "__INJECT_KB_UPDATE_PROMPT__");
  const goalExtractionPromptMatch = assertSingleMarker(template, GOAL_EXTRACTION_PROMPT_MARKER_RE, "__INJECT_GOAL_EXTRACTION_PROMPT__");
  const knowledgeBaseLayoutMatch = assertSingleMarker(template, KNOWLEDGE_BASE_LAYOUT_MARKER_RE, "__INJECT_KNOWLEDGE_BASE_LAYOUT__");
  const protocolExampleMarkdownMatch = assertSingleMarker(template, PROTOCOL_EXAMPLE_MARKDOWN_MARKER_RE, "__INJECT_PROTOCOL_EXAMPLE_MARKDOWN__");
  const convertersMatch = assertSingleMarker(template, CONVERTERS_MARKER_RE, "__INJECT_CONVERTERS__");

  const fontsIndent = fontsMatch[1];
  const fontsRules = readInlineFontFaces();
  const fontsLine = fontsRules
    ? fontsRules.split("\n").map((rule) => `${fontsIndent}${rule}`).join("\n") + "\n"
    : `${fontsIndent}/* no fonts inlined */\n`;
  const promptIndent = promptMatch[1];
  const boardPromptIndent = boardPromptMatch[1];
  const meetProtocolPromptIndent = meetProtocolPromptMatch[1];
  const meetingNarrativePromptIndent = meetingNarrativePromptMatch[1];
  const kbUpdatePromptIndent = kbUpdatePromptMatch[1];
  const goalExtractionPromptIndent = goalExtractionPromptMatch[1];
  const knowledgeBaseLayoutIndent = knowledgeBaseLayoutMatch[1];
  const protocolExampleMarkdownIndent = protocolExampleMarkdownMatch[1];
  const convertersIndent = convertersMatch[1];

  const promptText = readFileSync(resolve(ROOT, PROMPT_PATH), "utf8");
  const meetProtocolPromptText = readFileSync(resolve(ROOT, MEET_PROTOCOL_PROMPT_PATH), "utf8");
  const meetingNarrativePromptText = readFileSync(resolve(ROOT, MEETING_NARRATIVE_PROMPT_PATH), "utf8");
  const kbUpdatePromptText = readFileSync(resolve(ROOT, KB_UPDATE_PROMPT_PATH), "utf8");
  const goalExtractionPromptText = readFileSync(resolve(ROOT, GOAL_EXTRACTION_PROMPT_PATH), "utf8");
  const knowledgeBaseLayoutText = readFileSync(resolve(ROOT, KNOWLEDGE_BASE_LAYOUT_PATH), "utf8");
  const protocolExampleMarkdownText = readFileSync(resolve(ROOT, PROTOCOL_EXAMPLE_MARKDOWN_PATH), "utf8");
  const boardPromptText = BOARD_PROMPT_PATHS
    .map((relPath) => readFileSync(resolve(ROOT, relPath), "utf8").replace(/\s+$/, ""))
    .join("\n\n---\n\n") + "\n";
  const promptLine = `${promptIndent}const LLM_PROMPT_TEXT = ${JSON.stringify(promptText)};\n`;
  const boardPromptLine = `${boardPromptIndent}const BOARD_PROMPT_TEXT = ${JSON.stringify(boardPromptText)};\n`;
  const meetProtocolPromptLine = `${meetProtocolPromptIndent}const MEET_PROTOCOL_PROMPT_TEXT = ${JSON.stringify(meetProtocolPromptText)};\n`;
  const meetingNarrativePromptLine = `${meetingNarrativePromptIndent}const MEETING_NARRATIVE_PROMPT_TEXT = ${JSON.stringify(meetingNarrativePromptText)};\n`;
  const kbUpdatePromptLine = `${kbUpdatePromptIndent}const KB_UPDATE_PROMPT_TEXT = ${JSON.stringify(kbUpdatePromptText)};\n`;
  const goalExtractionPromptLine = `${goalExtractionPromptIndent}const GOAL_EXTRACTION_PROMPT_TEXT = ${JSON.stringify(goalExtractionPromptText)};\n`;
  const knowledgeBaseLayoutLine = `${knowledgeBaseLayoutIndent}const KNOWLEDGE_BASE_LAYOUT_TEXT = ${JSON.stringify(knowledgeBaseLayoutText)};\n`;
  const protocolExampleMarkdownLine = `${protocolExampleMarkdownIndent}const PROTOCOL_EXAMPLE_MARKDOWN_TEXT = ${JSON.stringify(protocolExampleMarkdownText)};\n`;

  const blocks = MODULE_PATHS.map((relPath) => {
    const source = readFileSync(resolve(ROOT, relPath), "utf8");
    const transformed = transformModule(source).replace(/\s+$/, "");
    assertNoLeftoverEsm(relPath, transformed);
    return `// ===== FILE: ${relPath} =====\n\n${transformed}`;
  });
  const joined = blocks.join("\n\n");
  const indentedConverters = joined
    .split("\n")
    .map((line) => (line.length ? `${convertersIndent}${line}` : ""))
    .join("\n");

  let output = replaceLiteral(template, FONTS_MARKER_RE, fontsLine);
  output = replaceLiteral(output, PROMPT_MARKER_RE, promptLine);
  output = replaceLiteral(output, BOARD_PROMPT_MARKER_RE, boardPromptLine);
  output = replaceLiteral(output, MEET_PROTOCOL_PROMPT_MARKER_RE, meetProtocolPromptLine);
  output = replaceLiteral(output, MEETING_NARRATIVE_PROMPT_MARKER_RE, meetingNarrativePromptLine);
  output = replaceLiteral(output, KB_UPDATE_PROMPT_MARKER_RE, kbUpdatePromptLine);
  output = replaceLiteral(output, GOAL_EXTRACTION_PROMPT_MARKER_RE, goalExtractionPromptLine);
  output = replaceLiteral(output, KNOWLEDGE_BASE_LAYOUT_MARKER_RE, knowledgeBaseLayoutLine);
  output = replaceLiteral(output, PROTOCOL_EXAMPLE_MARKDOWN_MARKER_RE, protocolExampleMarkdownLine);
  output = replaceLiteral(output, CONVERTERS_MARKER_RE, `${indentedConverters}\n`);

  writeFileSync(OUT_PATH, output);
  console.log(`build-lab: wrote ${relative(ROOT, OUT_PATH)} (${output.length} bytes)`);
}

main();