docmost-sync/test/transforms.test.ts

import { describe, expect, it } from 'vitest';
import {
  walk,
  getList,
  insertMarkerAfter,
  setCalloutRange,
  noteItem,
  mdToInlineNodes,
  commentsToFootnotes,
} from '../packages/docmost-client/src/lib/transforms.js';

// ---------------------------------------------------------------------------
// Small inline fixture builders. A ProseMirror node is a plain JSON object of
// shape { type, attrs?, content?, text?, marks? }.
// ---------------------------------------------------------------------------

/** A plain text run, optionally with marks. */
function text(t: string, marks?: any[]): any {
  return marks ? { type: 'text', text: t, marks } : { type: 'text', text: t };
}

/** A paragraph holding the given inline runs. */
function para(...runs: any[]): any {
  return { type: 'paragraph', attrs: { id: 'p' }, content: runs };
}

/** A callout holding the given child blocks. */
function callout(...children: any[]): any {
  return { type: 'callout', content: children };
}

/** A document with the given top-level blocks. */
function doc(...blocks: any[]): any {
  return { type: 'doc', content: blocks };
}

/**
 * Recursively strip every `attrs.id` so docs containing freshId()-generated ids
 * can be deep-compared structurally. Mutates a clone, returns it.
 */
function stripIds<T>(value: T): T {
  const v: any = structuredClone(value);
  const recur = (n: any): void => {
    if (Array.isArray(n)) {
      n.forEach(recur);
      return;
    }
    if (n && typeof n === 'object') {
      if (n.attrs && typeof n.attrs === 'object' && 'id' in n.attrs) {
        delete n.attrs.id;
      }
      for (const k of Object.keys(n)) recur(n[k]);
    }
  };
  recur(v);
  return v;
}

// ===========================================================================
describe('walk', () => {
  it('is a no-op for a nullish or non-object root', () => {
    const seen: any[] = [];
    walk(null, (n) => seen.push(n));
    walk(undefined, (n) => seen.push(n));
    walk('string', (n) => seen.push(n));
    walk(42, (n) => seen.push(n));
    walk([1, 2, 3], (n) => seen.push(n)); // array is not an object root
    expect(seen).toEqual([]);
  });

  it('visits the root itself and all nested nodes (callout/table/list)', () => {
    const tree = {
      type: 'doc',
      content: [
        callout(para(text('a'))),
        {
          type: 'table',
          content: [
            { type: 'tableRow', content: [{ type: 'tableCell', content: [para(text('b'))] }] },
          ],
        },
        {
          type: 'orderedList',
          content: [{ type: 'listItem', content: [para(text('c'))] }],
        },
      ],
    };
    const types: string[] = [];
    walk(tree, (n) => types.push(n.type));
    // Root first, then DFS into every nested container.
    expect(types[0]).toBe('doc');
    expect(types).toContain('callout');
    expect(types).toContain('table');
    expect(types).toContain('tableRow');
    expect(types).toContain('tableCell');
    expect(types).toContain('orderedList');
    expect(types).toContain('listItem');
    expect(types).toContain('paragraph');
    expect(types).toContain('text');
  });

  it('ignores a non-array content field', () => {
    const node = { type: 'weird', content: { not: 'an array' } };
    const seen: any[] = [];
    walk(node, (n) => seen.push(n));
    // Only the root is visited; the object content is never recursed into.
    expect(seen).toEqual([node]);
  });
});

// ===========================================================================
describe('getList', () => {
  it('returns the FIRST match in depth-first order', () => {
    const first = { type: 'orderedList', attrs: { id: 'L1' }, content: [] };
    const second = { type: 'orderedList', attrs: { id: 'L2' }, content: [] };
    const tree = doc(callout(first), second);
    const found = getList(tree, (n) => n.type === 'orderedList');
    expect(found).toBe(first); // DFS reaches the callout's child before the sibling
    expect(found.attrs.id).toBe('L1');
  });

  it('returns null when nothing matches', () => {
    const tree = doc(para(text('x')));
    expect(getList(tree, (n) => n.type === 'orderedList')).toBeNull();
  });

  it('returns a LIVE reference, not a clone', () => {
    const list = { type: 'orderedList', content: [] };
    const tree = doc(list);
    const found = getList(tree, (n) => n.type === 'orderedList');
    expect(found).toBe(list); // same object identity
    found.marker = 'mutated';
    expect(list.marker).toBe('mutated'); // mutation visible on the original
  });

  it('matches a node lacking attrs.id', () => {
    const noId = { type: 'orderedList', content: [] }; // no attrs at all
    const tree = doc(para(text('x')), noId);
    const found = getList(tree, (n) => n.type === 'orderedList');
    expect(found).toBe(noId);
  });
});

// ===========================================================================
describe('insertMarkerAfter', () => {
  it('returns inserted:false when the anchor is not found', () => {
    const d = doc(para(text('hello world')));
    const r = insertMarkerAfter(d, 'absent text', '[1]');
    expect(r.inserted).toBe(false);
    // Returned doc is a clone of the unchanged input.
    expect(r.doc).toEqual(d);
    expect(r.doc).not.toBe(d);
  });

  it('inserts a plain marker run after the anchor in a single text run', () => {
    const d = doc(para(text('see here for details')));
    const r = insertMarkerAfter(d, 'see here', '[1]');
    expect(r.inserted).toBe(true);
    expect(r.doc.content[0].content).toEqual([
      { type: 'text', text: 'see here', marks: [] },
      { type: 'text', text: ' [1]' },
      { type: 'text', text: ' for details', marks: [] },
    ]);
  });

  it('preserves marks across runs and emits a PLAIN marker, no empty runs', () => {
    // Anchor "foo bar" spans a plain run "foo " and a bold run "bar baz".
    const d = doc(
      para(
        text('foo '),
        text('bar baz', [{ type: 'bold' }]),
      ),
    );
    const r = insertMarkerAfter(d, 'foo bar', '[1]');
    expect(r.inserted).toBe(true);
    // The bold run "bar baz" is split at the anchor end (after "bar"); the
    // leading "foo " run is untouched, the marker is plain, surrounding marks
    // are preserved verbatim, and no empty text run is emitted.
    expect(r.doc.content[0].content).toEqual([
      { type: 'text', text: 'foo ' },
      { type: 'text', text: 'bar', marks: [{ type: 'bold' }] },
      { type: 'text', text: ' [1]' },
      { type: 'text', text: ' baz', marks: [{ type: 'bold' }] },
    ]);
  });

  it('splits exactly at a run boundary without emitting an empty run', () => {
    // Anchor ends exactly at the end of the first run "alpha".
    const d = doc(para(text('alpha'), text('beta')));
    const r = insertMarkerAfter(d, 'alpha', '[1]');
    expect(r.inserted).toBe(true);
    // "before" == whole first run, "after" is empty -> no empty run pushed.
    expect(r.doc.content[0].content).toEqual([
      { type: 'text', text: 'alpha', marks: [] },
      { type: 'text', text: ' [1]' },
      { type: 'text', text: 'beta' },
    ]);
  });

  it('beforeBlock scope excludes blocks at/after the boundary', () => {
    const d = doc(
      para(text('body anchor')),     // index 0 (in scope when beforeBlock=1)
      para(text('notes anchor')),    // index 1 (out of scope)
    );
    // Anchor only exists in the out-of-scope block -> not inserted.
    const r = insertMarkerAfter(d, 'notes anchor', '[1]', { beforeBlock: 1 });
    expect(r.inserted).toBe(false);
    // The in-scope anchor still inserts when limited.
    const r2 = insertMarkerAfter(d, 'body anchor', '[1]', { beforeBlock: 1 });
    expect(r2.inserted).toBe(true);
  });

  it('does not mutate the input document', () => {
    const d = doc(para(text('keep me intact please')));
    const snapshot = structuredClone(d);
    insertMarkerAfter(d, 'keep me', '[1]');
    expect(d).toEqual(snapshot);
  });

  it('returns inserted:false for an empty anchor', () => {
    const d = doc(para(text('anything')));
    const r = insertMarkerAfter(d, '', '[1]');
    expect(r.inserted).toBe(false);
    expect(r.doc).toEqual(d);
  });
});

// ===========================================================================
describe('setCalloutRange', () => {
  it('rewrites a Unicode-ellipsis [1]…[K] range inside a callout', () => {
    const d = doc(callout(para(text('Footnotes [1]…[5] follow'))));
    const r = setCalloutRange(d, 7);
    expect(r.changed).toBe(1);
    expect(r.doc.content[0].content[0].content[0].text).toBe('Footnotes [1]…[7] follow');
  });

  it('rewrites an ASCII-ellipsis [1]...[K] range inside a callout', () => {
    const d = doc(callout(para(text('range [1]...[3] here'))));
    const r = setCalloutRange(d, 9);
    expect(r.changed).toBe(1);
    expect(r.doc.content[0].content[0].content[0].text).toBe('range [1]...[9] here');
  });

  it('leaves a paragraph [1]…[K] (outside any callout) untouched', () => {
    const d = doc(para(text('not a callout [1]…[5]')));
    const r = setCalloutRange(d, 9);
    expect(r.changed).toBe(0);
    expect(r.doc.content[0].content[0].text).toBe('not a callout [1]…[5]');
  });

  it('rewrites across multiple callouts and reports the changed count', () => {
    const d = doc(
      callout(para(text('a [1]…[2] b'))),
      para(text('skip [1]…[2]')),
      callout(para(text('c [1]...[4] d'))),
    );
    const r = setCalloutRange(d, 10);
    expect(r.changed).toBe(2);
    expect(r.doc.content[0].content[0].content[0].text).toBe('a [1]…[10] b');
    expect(r.doc.content[2].content[0].content[0].text).toBe('c [1]...[10] d');
  });

  it('reports changed:0 when no range matches', () => {
    const d = doc(callout(para(text('no range here'))));
    const r = setCalloutRange(d, 4);
    expect(r.changed).toBe(0);
  });

  it('does not mutate the input document', () => {
    const d = doc(callout(para(text('x [1]…[5] y'))));
    const snapshot = structuredClone(d);
    setCalloutRange(d, 99);
    expect(d).toEqual(snapshot);
  });

  it('handles TWO matching text nodes in one callout (regex lastIndex reset)', () => {
    // Two separate text nodes, each carrying a range, inside one callout.
    const d = doc(
      callout(
        para(text('first [1]…[2]')),
        para(text('second [1]…[3]')),
      ),
    );
    const r = setCalloutRange(d, 6);
    expect(r.changed).toBe(2);
    expect(r.doc.content[0].content[0].content[0].text).toBe('first [1]…[6]');
    expect(r.doc.content[0].content[1].content[0].text).toBe('second [1]…[6]');
  });
});

// ===========================================================================
describe('noteItem', () => {
  it('wraps inline nodes in listItem > paragraph', () => {
    const inline = [text('hello')];
    const item = noteItem(inline);
    expect(item.type).toBe('listItem');
    expect(item.content).toHaveLength(1);
    const p = item.content[0];
    expect(p.type).toBe('paragraph');
    expect(p.content).toEqual([text('hello')]);
    // The paragraph carries a string id from Math.random()-based freshId().
    expect(typeof p.attrs.id).toBe('string');
    expect(p.attrs.id.length).toBeGreaterThan(0);
  });

  it('produces empty content for non-array input', () => {
    expect(noteItem(undefined as any).content[0].content).toEqual([]);
    expect(noteItem(null as any).content[0].content).toEqual([]);
    expect(noteItem('nope' as any).content[0].content).toEqual([]);
  });

  it('clones the input so the result shares no references', () => {
    const inline = [text('mutable')];
    const item = noteItem(inline);
    inline[0].text = 'changed';
    expect(item.content[0].content[0].text).toBe('mutable'); // unaffected
    expect(item.content[0].content[0]).not.toBe(inline[0]);
  });

  it('matches the expected structure (ignoring the random id)', () => {
    const item = noteItem([text('body', [{ type: 'bold' }])]);
    expect(stripIds(item)).toEqual({
      type: 'listItem',
      content: [
        {
          type: 'paragraph',
          attrs: {},
          content: [{ type: 'text', text: 'body', marks: [{ type: 'bold' }] }],
        },
      ],
    });
  });
});

// ===========================================================================
describe('mdToInlineNodes', () => {
  it('returns [] for empty or non-string input', () => {
    expect(mdToInlineNodes('')).toEqual([]);
    expect(mdToInlineNodes('   ')).toEqual([]);
    expect(mdToInlineNodes(undefined as any)).toEqual([]);
    expect(mdToInlineNodes(null as any)).toEqual([]);
    expect(mdToInlineNodes(123 as any)).toEqual([]);
  });

  it('strips a case-insensitive "комментарий:" prefix', () => {
    expect(mdToInlineNodes('Комментарий: hello')).toEqual([{ type: 'text', text: 'hello' }]);
    expect(mdToInlineNodes('комментарий : hi')).toEqual([{ type: 'text', text: 'hi' }]);
  });

  it('strips a leading "N. " numeric prefix', () => {
    expect(mdToInlineNodes('3. some note')).toEqual([{ type: 'text', text: 'some note' }]);
  });

  it('turns a leading **bold lead** into a bold node + plain remainder, space preserved', () => {
    const nodes = mdToInlineNodes('**Lead** rest of text');
    expect(nodes).toEqual([
      { type: 'text', text: 'Lead', marks: [{ type: 'bold' }] },
      { type: 'text', text: ' rest of text' }, // separating space preserved
    ]);
  });

  it('splits an inline **bold** mid-text', () => {
    const nodes = mdToInlineNodes('start **mid** end');
    expect(nodes).toEqual([
      { type: 'text', text: 'start ' },
      { type: 'text', text: 'mid', marks: [{ type: 'bold' }] },
      { type: 'text', text: ' end' },
    ]);
  });

  it('passes plain text through unchanged when there is no bold', () => {
    expect(mdToInlineNodes('just plain text')).toEqual([{ type: 'text', text: 'just plain text' }]);
  });

  it('handles a bold-only string', () => {
    // A bold-only string is treated as a leading bold lead with empty remainder.
    expect(mdToInlineNodes('**only**')).toEqual([
      { type: 'text', text: 'only', marks: [{ type: 'bold' }] },
    ]);
  });
});

// ===========================================================================
describe('commentsToFootnotes', () => {
  const HEADING = 'Примечания переводчика';

  /**
   * Build a realistic doc: body paragraphs, then the notes heading, then the
   * notes orderedList. `notes` is an array of inline-text strings for existing
   * list items.
   */
  function buildDoc(opts: {
    body: any[];
    notes?: string[];
    omitHeading?: boolean;
    omitList?: boolean;
    disclaimer?: any;
  }): any {
    const blocks: any[] = [];
    if (opts.disclaimer) blocks.push(opts.disclaimer);
    blocks.push(...opts.body);
    if (!opts.omitHeading) {
      blocks.push({ type: 'heading', attrs: { id: 'h', level: 2 }, content: [text(HEADING)] });
    }
    if (!opts.omitList) {
      const items = (opts.notes ?? []).map((t, i) => ({
        type: 'listItem',
        attrs: { id: `li${i}` },
        content: [para(text(t))],
      }));
      blocks.push({ type: 'orderedList', attrs: { id: 'ol' }, content: items });
    }
    return doc(...blocks);
  }

  function findNotesList(d: any): any {
    return d.content.find((n: any) => n.type === 'orderedList');
  }

  it('is identity (renumber pass only) when there are zero comments', () => {
    const d = buildDoc({ body: [para(text('plain body'))], notes: [] });
    const r = commentsToFootnotes(d, []);
    expect(r.consumed).toEqual([]);
    expect(r.doc.content[0]).toEqual(para(text('plain body')));
  });

  it('inserts a marker and appends one note for one comment with a selection', () => {
    const d = buildDoc({ body: [para(text('the quick brown fox'))], notes: [] });
    const r = commentsToFootnotes(d, [
      { id: 'c1', content: 'A note', selection: 'quick brown' },
    ]);
    expect(r.consumed).toEqual(['c1']);
    // Body now carries "[1]" right after the selection.
    const bodyText = r.doc.content[0].content.map((n: any) => n.text).join('');
    expect(bodyText).toBe('the quick brown [1] fox');
    // The notes list holds exactly one note built from the comment content.
    const list = findNotesList(r.doc);
    expect(list.content).toHaveLength(1);
    expect(stripIds(list.content[0])).toEqual(
      stripIds(noteItem(mdToInlineNodes('A note'))),
    );
  });

  it('numbers many comments by BODY reading order, not comment-array order', () => {
    // Body order: "alpha" then "omega". Comments are given out of order.
    const d = buildDoc({
      body: [para(text('alpha then omega here'))],
      notes: [],
    });
    const r = commentsToFootnotes(d, [
      { id: 'cOmega', content: 'note for omega', selection: 'omega' },
      { id: 'cAlpha', content: 'note for alpha', selection: 'alpha' },
    ]);
    // Both consumed, in comment-array processing order (NOT reading order, NOT sorted).
    expect(r.consumed).toEqual(['cOmega', 'cAlpha']);
    const bodyText = r.doc.content[0].content.map((n: any) => n.text).join('');
    // "alpha" precedes "omega" in reading order => [1] then [2].
    expect(bodyText).toBe('alpha [1] then omega [2] here');
    const list = findNotesList(r.doc);
    // Note list reordered to reading order: alpha-note first, omega-note second.
    expect(list.content[0].content[0].content[0].text).toBe('note for alpha');
    expect(list.content[1].content[0].content[0].text).toBe('note for omega');
  });

  it('skips a comment with no selection without consuming it', () => {
    const d = buildDoc({ body: [para(text('body text here'))], notes: [] });
    const r = commentsToFootnotes(d, [
      { id: 'c1', content: 'no anchor', selection: null },
      { id: 'c2', content: 'anchored', selection: 'body text' },
    ]);
    expect(r.consumed).toEqual(['c2']);
    const list = findNotesList(r.doc);
    expect(list.content).toHaveLength(1);
  });

  it('skips a comment whose selection is absent (no orphan note)', () => {
    const d = buildDoc({ body: [para(text('present text'))], notes: [] });
    const r = commentsToFootnotes(d, [
      { id: 'c1', content: 'orphan', selection: 'this string is not in the body' },
    ]);
    expect(r.consumed).toEqual([]); // nothing anchored
    const list = findNotesList(r.doc);
    expect(list.content).toHaveLength(0); // no orphan note appended
    const bodyText = r.doc.content[0].content.map((n: any) => n.text).join('');
    expect(bodyText).toBe('present text'); // body unchanged
  });

  it('renumbers existing [N] markers mixed with new placeholders by reading order', () => {
    // Body already has an existing "[1]" marker after "first"; a new comment
    // anchors before it in reading order at "intro".
    const d = buildDoc({
      body: [para(text('intro and first [1] then more'))],
      notes: ['existing note one'],
    });
    const r = commentsToFootnotes(d, [
      { id: 'cNew', content: 'fresh note', selection: 'intro' },
    ]);
    expect(r.consumed).toEqual(['cNew']);
    const bodyText = r.doc.content[0].content.map((n: any) => n.text).join('');
    // Reading order: "intro" placeholder -> [1]; existing "[1]" -> [2].
    expect(bodyText).toBe('intro [1] and first [2] then more');
    const list = findNotesList(r.doc);
    // Notes reordered: the new note (for "intro") first, the existing note second.
    expect(list.content).toHaveLength(2);
    expect(list.content[0].content[0].content[0].text).toBe('fresh note');
    expect(list.content[1].content[0].content[0].text).toBe('existing note one');
  });

  it('throws "document is inconsistent" when a body [N] has no matching note', () => {
    // Body references [9] but the notes list has only 3 items.
    const d = buildDoc({
      body: [para(text('see footnote [9] here'))],
      notes: ['n1', 'n2', 'n3'],
    });
    expect(() => commentsToFootnotes(d, [])).toThrow(/document is inconsistent/);
  });

  it('throws when the notes heading is missing', () => {
    const d = buildDoc({ body: [para(text('x'))], notes: [], omitHeading: true });
    expect(() => commentsToFootnotes(d, [])).toThrow(/not found/);
  });

  it('throws when the notes orderedList is missing', () => {
    const d = buildDoc({ body: [para(text('x'))], omitList: true });
    expect(() => commentsToFootnotes(d, [])).toThrow(/orderedList not found/);
  });

  it('does not mutate the input document', () => {
    const d = buildDoc({ body: [para(text('the quick brown fox'))], notes: [] });
    const snapshot = structuredClone(d);
    commentsToFootnotes(d, [{ id: 'c1', content: 'A note', selection: 'quick brown' }]);
    expect(d).toEqual(snapshot);
  });

  it('does not renumber a top-level disclaimer callout but syncs its range', () => {
    // A disclaimer callout carries "[1]…[K]"; it must be preserved (not consumed
    // as a footnote marker) and its range synced to the final note count.
    const disclaimer = callout(para(text('Notes range [1]…[1] applies')));
    const d = buildDoc({
      body: [para(text('alpha and beta here'))],
      notes: [],
      disclaimer,
    });
    const r = commentsToFootnotes(d, [
      { id: 'c1', content: 'note a', selection: 'alpha' },
      { id: 'c2', content: 'note b', selection: 'beta' },
    ]);
    expect(r.consumed.sort()).toEqual(['c1', 'c2']);
    // Disclaimer callout is at index 0; its body must NOT have been renumbered
    // into [1][2], it remains a "[1]…[n]" range synced to 2 notes.
    const calloutText = r.doc.content[0].content[0].content
      .map((n: any) => n.text)
      .join('');
    expect(calloutText).toBe('Notes range [1]…[2] applies');
    // Body markers (index 1) are the real footnotes.
    const bodyText = r.doc.content[1].content.map((n: any) => n.text).join('');
    expect(bodyText).toBe('alpha [1] and beta [2] here');
    const list = findNotesList(r.doc);
    expect(list.content).toHaveLength(2);
  });
});