SSAUGenerator/docx/src/render/document.js

'use strict';

const fs = require('fs');
const path = require('path');
const { Document, ExternalHyperlink, TextRun } = require('docx');

const {
  pageBreak,
  paraRuns,
  textRun,
  sectionHeading,
  subsectionHeading,
  plainHeading,
  pointParagraph,
  structSection,
  appendixStart,
  tocSection,
  figureBlock,
  tableSto,
  formulaFromText,
  whereBlock,
  noteSingle,
  notesBlock,
  definitionsBlock,
  bibliographyList,
  listParagraphRuns,
  nextListInstance,
  resetListInstances,
  defaultSectionProps,
  pageNumberFooter,
  REFS,
} = require('./primitives');
const { isPlantUmlSource } = require('../plantuml/render');

function createState(md, sections) {
  return {
    md,
    sections,
    labelMap: new Map(),
    biblioCount: 0,
    errors: [],
    warnings: [],
  };
}

function firstInlineText(tokens) {
  for (const token of tokens) {
    if (token.type === 'inline' && String(token.content || '').trim()) {
      return token.content.trim();
    }
  }
  return '';
}

// ── inline ────────────────────────────────────────────────────────

function citationRun(ctx, meta, state) {
  const { bracketOpen, bracketClose, pagePrefix } = ctx.config.citation;
  if (meta.n < 1 || (state.biblioCount > 0 && meta.n > state.biblioCount)) {
    state.errors.push(
      `Ссылка [@${meta.n}] выходит за пределы списка источников (${state.biblioCount})`
    );
    return textRun(ctx, `[@${meta.n}]`);
  }
  const text = meta.page
    ? `${bracketOpen}${meta.n}, ${pagePrefix} ${meta.page}${bracketClose}`
    : `${bracketOpen}${meta.n}${bracketClose}`;
  return textRun(ctx, text);
}

function xrefRun(ctx, meta, state) {
  const key = `${meta.kind}:${meta.label}`;
  const value = state.labelMap.get(key);
  if (value == null) {
    state.errors.push(`Не найдена метка @${key}`);
    return textRun(ctx, `@${key}`);
  }
  return textRun(ctx, String(value));
}

function renderInline(ctx, inlineToken, state) {
  const runs = [];
  const flags = { bold: false, italics: false };
  const linkStack = [];
  if (!inlineToken || !inlineToken.children) return runs;

  const push = (run) => {
    if (linkStack.length) linkStack[linkStack.length - 1].children.push(run);
    else runs.push(run);
  };
  const closeLink = () => {
    const link = linkStack.pop();
    const hyper = new ExternalHyperlink({ children: link.children, link: link.href });
    if (linkStack.length) linkStack[linkStack.length - 1].children.push(hyper);
    else runs.push(hyper);
  };

  for (const child of inlineToken.children) {
    switch (child.type) {
      case 'text':
        push(textRun(ctx, child.content, flags));
        break;
      case 'softbreak':
        push(textRun(ctx, ' '));
        break;
      case 'hardbreak':
        push(
          new TextRun({
            break: 1,
            font: ctx.typography.font,
            size: ctx.typography.sizeBody,
            ...(ctx.typography.language ? { language: { value: ctx.typography.language } } : {}),
          })
        );
        break;
      case 'strong_open':
        flags.bold = true;
        break;
      case 'strong_close':
        flags.bold = false;
        break;
      case 'em_open':
        flags.italics = true;
        break;
      case 'em_close':
        flags.italics = false;
        break;
      case 's_open':
      case 's_close':
        break;
      case 'code_inline':
        push(textRun(ctx, child.content, { size: ctx.typography.sizeSmall }));
        break;
      case 'link_open':
        linkStack.push({ href: child.attrGet('href') || '', children: [] });
        break;
      case 'link_close':
        if (linkStack.length) closeLink();
        break;
      case 'citation':
        push(citationRun(ctx, child.meta, state));
        break;
      case 'xref':
        push(xrefRun(ctx, child.meta, state));
        break;
      case 'image':
        state.warnings.push(
          'Инлайн-изображение пропущено; используйте директиву ::: figure'
        );
        break;
      default:
        break;
    }
  }
  while (linkStack.length) closeLink();
  return runs;
}

// ── списки ────────────────────────────────────────────────────────

function renderList(ctx, tokens, startIndex, depth, listCtx, state) {
  const open = tokens[startIndex];
  const ordered = open.type === 'ordered_list_open';
  const current = depth === 0 ? { ref: ordered ? REFS.number : REFS.dash } : listCtx;
  if (depth === 0) current.instance = nextListInstance(current.ref);

  const elements = [];
  const level = Math.min(depth, 3);
  let i = startIndex + 1;
  const closeType = ordered ? 'ordered_list_close' : 'bullet_list_close';

  while (i < tokens.length && tokens[i].type !== closeType) {
    if (tokens[i].type !== 'list_item_open') {
      i += 1;
      continue;
    }
    i += 1;
    let firstParagraph = true;
    while (i < tokens.length && tokens[i].type !== 'list_item_close') {
      const token = tokens[i];
      if (token.type === 'paragraph_open') {
        const runs = renderInline(ctx, tokens[i + 1], state);
        if (firstParagraph) {
          elements.push(listParagraphRuns(ctx, runs, current.ref, level, current.instance));
          firstParagraph = false;
        } else {
          elements.push(
            paraRuns(ctx, runs, {
              extra: {
                indent: {
                  left: ctx.config.lists.nestedLevelStep * (depth + 1),
                  firstLine: ctx.typography.indentFirst,
                },
              },
            })
          );
        }
        i += 3;
      } else if (token.type === 'bullet_list_open' || token.type === 'ordered_list_open') {
        const nested = renderList(ctx, tokens, i, depth + 1, current, state);
        elements.push(...nested.elements);
        i = nested.next;
      } else if (token.type === 'directive') {
        elements.push(...renderDirective(ctx, token, state));
        i += 1;
      } else if (token.type === 'table_open') {
        const table = readTable(tokens, i);
        elements.push(...tableSto(ctx, { label: '', title: '', rows: table.rows }));
        i = table.next;
      } else {
        i += 1;
      }
    }
    i += 1;
  }
  return { elements, next: i + 1 };
}

// ── таблицы из markdown ───────────────────────────────────────────

function readTable(tokens, startIndex) {
  const rows = [];
  let current = null;
  let i = startIndex + 1;
  for (; i < tokens.length; i++) {
    const token = tokens[i];
    if (token.type === 'tr_open') {
      current = [];
    } else if (token.type === 'tr_close') {
      if (current) rows.push(current);
      current = null;
    } else if (token.type === 'inline') {
      if (current) current.push(token.content);
    } else if (token.type === 'table_close') {
      i += 1;
      break;
    }
  }
  return { rows, next: i };
}

// ── директивы ─────────────────────────────────────────────────────

function parseWhereBody(body) {
  return body
    .split(/\r?\n/)
    .map((line) => line.replace(/^\s*(?:[-*]|\d+[.)])\s*/, '').trim())
    .filter(Boolean)
    .map((line) => {
      const parts = line.split(/\s+[—–-]\s+/);
      if (parts.length < 2) return { symbol: line, desc: '' };
      return { symbol: parts[0].trim(), desc: parts.slice(1).join(' — ').trim() };
    });
}

function splitNoteItems(body) {
  return body
    .split(/\r?\n/)
    .map((line) => line.replace(/^\s*(?:[-*]|\d+[.)])\s*/, '').trim())
    .filter(Boolean);
}

function parseDefinitions(body) {
  return parseWhereBody(body).map(({ symbol, desc }) => ({ term: symbol, definition: desc }));
}

function findTableRows(md, body) {
  const tokens = md.parse(body, {});
  for (let i = 0; i < tokens.length; i++) {
    if (tokens[i].type === 'table_open') return readTable(tokens, i).rows;
  }
  return null;
}

function renderDirective(ctx, token, state) {
  const { name, attrs, body } = token.info;
  switch (name) {
    case 'pagebreak':
      return [pageBreak()];
    case 'figure': {
      const cap = token.__caption || { label: '?' };
      const notes = body
        ? body.split(/\r?\n/).map((s) => s.trim()).filter(Boolean)
        : [];
      let imagePath = attrs.file;
      if (isPlantUmlSource(attrs.file)) {
        imagePath = attrs.__renderedImage || null;
      } else if (attrs.file) {
        const resolved = path.isAbsolute(attrs.file)
          ? attrs.file
          : path.resolve(ctx.assetsRoot, attrs.file);
        if (!fs.existsSync(resolved)) {
          state.warnings.push(`Рисунок "${attrs.title || ''}": файл не найден: ${attrs.file}`);
        }
      }
      return figureBlock(ctx, {
        label: cap.label,
        title: attrs.title || '',
        imagePath,
        notesLines: notes,
      });
    }
    case 'table': {
      const cap = token.__caption || { label: '?' };
      const rows = findTableRows(state.md, body);
      if (!rows) {
        state.errors.push(
          `Директива ::: table "${attrs.title || ''}": не найдена markdown-таблица`
        );
        return [];
      }
      return tableSto(ctx, {
        label: cap.label,
        title: attrs.title || '',
        rows,
        note: attrs.note,
      });
    }
    case 'formula': {
      const cap = token.__caption || { label: '?' };
      return formulaFromText(ctx, body.trim(), cap.label);
    }
    case 'where':
      return whereBlock(ctx, parseWhereBody(body));
    case 'note':
      return [noteSingle(ctx, body.trim().replace(/\s*\n\s*/g, ' '))];
    case 'notes':
      return notesBlock(ctx, splitNoteItems(body));
    case 'definitions':
      return definitionsBlock(ctx, parseDefinitions(body));
    default:
      state.warnings.push(`Неизвестная директива ::: ${name}`);
      return [];
  }
}

// ── общий обход токенов ───────────────────────────────────────────

function renderTokens(ctx, tokens, state, opts = {}) {
  const mode = opts.mode || 'plain';
  const out = [];
  let firstH1 = true;

  for (let i = 0; i < tokens.length; i++) {
    const token = tokens[i];
    switch (token.type) {
      case 'heading_open': {
        const level = Number(token.tag.slice(1));
        const text = tokens[i + 1] ? tokens[i + 1].content : '';
        if (mode === 'numbered') {
          if (level <= 1) {
            if (!(firstH1 && opts.suppressFirstPageBreak)) out.push(pageBreak());
            firstH1 = false;
            out.push(sectionHeading(ctx, text));
          } else if (level === 2) {
            out.push(subsectionHeading(ctx, text));
          } else {
            out.push(pointParagraph(ctx, text));
          }
        } else {
          out.push(plainHeading(ctx, text, level));
        }
        i += 2;
        break;
      }
      case 'paragraph_open':
        out.push(paraRuns(ctx, renderInline(ctx, tokens[i + 1], state)));
        i += 2;
        break;
      case 'directive':
        out.push(...renderDirective(ctx, token, state));
        break;
      case 'bullet_list_open':
      case 'ordered_list_open': {
        const res = renderList(ctx, tokens, i, 0, null, state);
        out.push(...res.elements);
        i = res.next - 1;
        break;
      }
      case 'table_open': {
        const res = readTable(tokens, i);
        out.push(...tableSto(ctx, { label: '', title: '', rows: res.rows }));
        i = res.next - 1;
        break;
      }
      case 'fence':
        state.warnings.push('Блок кода проигнорирован (функция листингов удалена)');
        break;
      case 'hr':
        out.push(pageBreak());
        break;
      default:
        break;
    }
  }
  return out;
}

function renderSections(ctx, state) {
  resetListInstances();
  const out = [];
  let first = true;

  for (const section of state.sections) {
    const entry = section.entry;

    if (entry.type === 'toc') {
      if (!ctx.config.toc.enabled) continue;
      out.push(...tocSection(ctx, [], { isFirstInDocument: first }));
      first = false;
    } else if (entry.type === 'structural') {
      const title = entry.title || section.meta.title || firstInlineText(section.tokens) || '';
      const body = renderTokens(ctx, section.tokens, state, { mode: 'plain' });
      out.push(...structSection(ctx, title, body, { pageBreakBefore: !first }));
      first = false;
    } else if (entry.type === 'numbered') {
      out.push(
        ...renderTokens(ctx, section.tokens, state, {
          mode: 'numbered',
          suppressFirstPageBreak: first,
        })
      );
      first = false;
    } else if (entry.type === 'bibliography') {
      const body = bibliographyList(ctx, section.biblioEntries || []);
      out.push(
        ...structSection(ctx, ctx.config.bibliography.title, body, { pageBreakBefore: !first })
      );
      first = false;
    } else if (entry.type === 'appendix') {
      out.push(...appendixStart(ctx, entry.letter, entry.title || ''));
      out.push(...renderTokens(ctx, section.tokens, state, { mode: 'plain' }));
      first = false;
    }
  }
  return out;
}

function buildDocx(ctx, children) {
  resetListInstances();
  return new Document({
    features: ctx.config.document.features,
    numbering: ctx.numberingConfig,
    styles: ctx.documentStyles,
    sections: [
      {
        properties: defaultSectionProps(ctx),
        footers: { default: pageNumberFooter(ctx) },
        children,
      },
    ],
  });
}

module.exports = {
  createState,
  renderSections,
  renderTokens,
  renderInline,
  readTable,
  parseWhereBody,
  buildDocx,
};