src/converters/markdown-to-wiki.js - dev - deka/buffer

src/converters/markdown-to-wiki.js
318 строк10 KB
Maksim Ratnikov
Extract markdown-to-wiki converter into src/
08 май 2026, 16:24
08 май 2026, 16:24feea842
import { escapeHtml, layoutId, layoutWidths } from "../util/format.js";

export function mdInlineToWiki(text, depth = 0) {
  if (text === "" || text == null) return "";
  if (depth > 64) return escapeHtml(String(text));
  const out = [];
  let i = 0;
  const src = String(text);

  function emitText(value) {
    out.push(escapeHtml(value));
  }

  function readDelim(delim) {
    let close = src.indexOf(delim, i + delim.length);
    if (close < 0) return null;
    const ch = delim[0];
    while (close + delim.length < src.length && src[close + delim.length] === ch) {
      close += 1;
    }
    return { inner: src.slice(i + delim.length, close), end: close + delim.length };
  }

  while (i < src.length) {
    const ch = src[i];

    if (ch === "\\" && i + 1 < src.length) {
      const next = src[i + 1];
      if ("\\*_`[]~".includes(next)) {
        emitText(next);
        i += 2;
        continue;
      }
    }

    if (ch === "`") {
      const close = src.indexOf("`", i + 1);
      if (close > i) {
        out.push(`<code data-set="inline">${escapeHtml(src.slice(i + 1, close))}</code>`);
        i = close + 1;
        continue;
      }
    }

    if (ch === "[") {
      const closeBracket = src.indexOf("]", i + 1);
      if (closeBracket > i && src[closeBracket + 1] === "(") {
        const closeParen = src.indexOf(")", closeBracket + 2);
        if (closeParen > closeBracket + 1) {
          const label = src.slice(i + 1, closeBracket);
          const rawUrl = src.slice(closeBracket + 2, closeParen);
          const safeUrl = /^\s*(?:javascript|data|vbscript):/i.test(rawUrl) ? "#" : rawUrl;
          out.push(`<a href="${escapeHtml(safeUrl)}">${mdInlineToWiki(label, depth + 1)}</a>`);
          i = closeParen + 1;
          continue;
        }
      }
    }

    if (ch === "*" || ch === "_") {
      const doubled = src.startsWith(ch + ch, i);
      if (doubled) {
        const result = readDelim(ch + ch);
        if (result) {
          out.push(`<strong>${mdInlineToWiki(result.inner, depth + 1)}</strong>`);
          i = result.end;
          continue;
        }
      } else {
        const result = readDelim(ch);
        if (result) {
          out.push(`<em>${mdInlineToWiki(result.inner, depth + 1)}</em>`);
          i = result.end;
          continue;
        }
      }
    }

    if (ch === "~" && src.startsWith("~~", i)) {
      const result = readDelim("~~");
      if (result) {
        out.push(`<s>${mdInlineToWiki(result.inner, depth + 1)}</s>`);
        i = result.end;
        continue;
      }
    }

    emitText(ch);
    i += 1;
  }

  return out.join("");
}

export function mdParaTag(content) {
  return `<p data-id="${layoutId()}" style="text-align: justify; margin-left: 0px !important;">${content}</p>`;
}

export function mdListItemMatch(line) {
  const bullet = line.match(/^(\s*)([-+*])\s+(.*)$/);
  if (bullet) {
    const indent = bullet[1].length;
    const taskMatch = bullet[3].match(/^\[( |x|X)\]\s+(.*)$/);
    return {
      kind: "ul",
      indent,
      marker: bullet[2],
      isTask: Boolean(taskMatch),
      checked: taskMatch ? taskMatch[1].toLowerCase() === "x" : false,
      text: taskMatch ? taskMatch[2] : bullet[3]
    };
  }
  const ordered = line.match(/^(\s*)(\d+)\.\s+(.*)$/);
  if (ordered) {
    return {
      kind: "ol",
      indent: ordered[1].length,
      marker: ordered[2],
      isTask: false,
      checked: false,
      text: ordered[3]
    };
  }
  return null;
}

export function mdGroupListItems(items) {
  const root = [];
  const stack = [{ indent: -1, items: root }];

  for (const item of items) {
    while (stack.length > 1 && item.indent <= stack[stack.length - 1].indent) {
      stack.pop();
    }
    const parent = stack[stack.length - 1];
    const node = { ...item, children: [] };
    parent.items.push(node);
    stack.push({ indent: item.indent, items: node.children });
  }

  return root;
}

export function mdEmitTaskList(items) {
  const inner = items.map((item) => {
    const checkedAttr = item.checked ? ' checked=""' : "";
    const dataChecked = item.checked ? "true" : "false";
    const childrenHtml = item.children.length ? mdEmitListNodes(item.children) : "";
    return `<li data-id="${layoutId()}" data-checked="${dataChecked}" data-type="taskItem" style="display: flex; align-items: flex-start; gap: 0.5rem;"><label style="display: flex; align-items: center; margin-top: 0.25rem; pointer-events: none;"><input type="checkbox" disabled=""${checkedAttr}><span></span></label><div>${mdParaTag(mdInlineToWiki(item.text))}${childrenHtml}</div></li>`;
  }).join("");
  return `<ul data-id="${layoutId()}" data-type="taskList" style="text-align: justify; margin-left: 0px !important;">${inner}</ul>`;
}

export function mdEmitListNodes(items) {
  if (!items.length) return "";
  const kind = items[0].kind;
  const allTask = kind === "ul" && items.every((item) => item.isTask);
  if (allTask) {
    return mdEmitTaskList(items);
  }
  const inner = items.map((item) => {
    const visibleText = item.isTask ? `[${item.checked ? "x" : " "}] ${item.text}` : item.text;
    const childrenHtml = item.children.length ? mdEmitListNodes(item.children) : "";
    if (kind === "ol") {
      return `<li class="listItem_14jqc" data-id="${layoutId()}">${mdParaTag(mdInlineToWiki(visibleText))}${childrenHtml}</li>`;
    }
    return `<li data-id="${layoutId()}">${mdParaTag(mdInlineToWiki(visibleText))}${childrenHtml}</li>`;
  }).join("");
  if (kind === "ol") {
    return `<ol class="orderedList_14jqc" data-id="${layoutId()}">${inner}</ol>`;
  }
  return `<ul data-id="${layoutId()}">${inner}</ul>`;
}

export function mdEmitTable(headerCells, bodyRows) {
  const colCount = Math.max(headerCells.length, ...bodyRows.map((row) => row.length));
  const widths = layoutWidths(undefined, colCount);
  const tableId = layoutId();
  const headerCellHtml = Array.from({ length: colCount }, (_, idx) => {
    const text = headerCells[idx] ?? "";
    return `<th data-id="${layoutId()}" colspan="1" rowspan="1" backgroundcolor="var(--editor-highlight-bgc)"><span>${mdParaTag(mdInlineToWiki(text))}</span></th>`;
  }).join("");
  const bodyHtml = bodyRows.map((row, rowIdx) => {
    const cellsHtml = Array.from({ length: colCount }, (_, idx) => {
      const text = row[idx] ?? "";
      return `<td data-id="${layoutId()}" colspan="1" rowspan="1" data-bgc="#ffffff" style="background-color: rgb(255, 255, 255);">${mdParaTag(mdInlineToWiki(text))}</td>`;
    }).join("");
    return `<tr data-id="${layoutId()}" data-original-row-index="${rowIdx}">${cellsHtml}</tr>`;
  }).join("");
  return `<section data-id="${layoutId()}" data-table-id="${tableId}" columnwidths="${widths}" numberedactive="false" data-type="table"><table><tbody><tr data-id="${layoutId()}">${headerCellHtml}</tr>${bodyHtml}</tbody></table></section>`;
}

export function mdSplitTableRow(line) {
  const trimmed = line.replace(/^\s*\|/, "").replace(/\|\s*$/, "");
  const cells = [];
  let current = "";
  let escaped = false;
  for (const ch of trimmed) {
    if (escaped) {
      current += ch;
      escaped = false;
      continue;
    }
    if (ch === "\\") {
      escaped = true;
      continue;
    }
    if (ch === "|") {
      cells.push(current.trim());
      current = "";
      continue;
    }
    current += ch;
  }
  cells.push(current.trim());
  return cells;
}

export function mdParseBlocks(lines) {
  const out = [];
  let i = 0;

  while (i < lines.length) {
    const raw = lines[i];

    if (!raw.trim()) {
      i += 1;
      continue;
    }

    const fence = raw.match(/^(\s*)(`{3,}|~{3,})(.*)$/);
    if (fence) {
      const closer = fence[2];
      const buf = [];
      let j = i + 1;
      while (j < lines.length && !lines[j].trim().startsWith(closer)) {
        buf.push(lines[j]);
        j += 1;
      }
      out.push(`<pre data-set="highlighted-code" data-id="${layoutId()}"><code>${escapeHtml(buf.join("\n"))}</code></pre>`);
      i = j < lines.length ? j + 1 : j;
      continue;
    }

    const heading = raw.match(/^(#{1,6})\s+(.+?)\s*#*\s*$/);
    if (heading) {
      const level = heading[1].length;
      out.push(`<h${level} data-id="${layoutId()}">${mdInlineToWiki(heading[2])}</h${level}>`);
      i += 1;
      continue;
    }

    if (/^\s{0,3}(-{3,}|\*{3,}|_{3,})\s*$/.test(raw)) {
      out.push("<hr>");
      i += 1;
      continue;
    }

    if (/^\s*>/.test(raw)) {
      const buf = [];
      while (i < lines.length && /^\s*>/.test(lines[i])) {
        buf.push(lines[i].replace(/^\s*>\s?/, ""));
        i += 1;
      }
      out.push(`<blockquote data-id="${layoutId()}">${mdParseBlocks(buf)}</blockquote>`);
      continue;
    }

    if (raw.includes("|") && i + 1 < lines.length && /^\s*\|?\s*:?-{1,}/.test(lines[i + 1])) {
      const headerCells = mdSplitTableRow(raw);
      const delim = mdSplitTableRow(lines[i + 1]);
      if (delim.every((cell) => /^:?-{1,}:?$/.test(cell.trim()))) {
        const bodyRows = [];
        let j = i + 2;
        while (j < lines.length && lines[j].trim() && lines[j].includes("|")) {
          bodyRows.push(mdSplitTableRow(lines[j]));
          j += 1;
        }
        out.push(mdEmitTable(headerCells, bodyRows));
        i = j;
        continue;
      }
    }

    const firstListItem = mdListItemMatch(raw);
    if (firstListItem) {
      const items = [];
      while (i < lines.length) {
        const item = mdListItemMatch(lines[i]);
        if (!item) break;
        items.push(item);
        i += 1;
      }
      const grouped = mdGroupListItems(items);
      out.push(mdEmitListNodes(grouped));
      continue;
    }

    const paraBuf = [];
    while (i < lines.length && lines[i].trim() && !/^(`{3,}|~{3,}|#{1,6}\s|\s{0,3}(-{3,}|\*{3,}|_{3,})\s*$|\s*>|\s*[-+*]\s+|\s*\d+\.\s+)/.test(lines[i])) {
      paraBuf.push(lines[i].trim());
      i += 1;
    }
    if (paraBuf.length) {
      out.push(mdParaTag(mdInlineToWiki(paraBuf.join(" "))));
    }
  }

  return out.join("\n");
}

export function markdownToWikiHtml(source) {
  if (source == null) return "";
  const text = String(source);
  if (!text.trim()) return "";
  const lines = text.replace(/\r\n?/g, "\n").replace(/\t/g, "    ").split("\n");
  return mdParseBlocks(lines);
}