v2 chunk 2c: numbered-note format support + real-draft test

Extend the v2 parser and CLI to handle [N] text footnote definitions
in addition to pandoc-style [^marker]: text. The numbered format is
what docx-to-text conversion of footnoted Word documents produces.
The user's "Anti-Communist Formations of LIS" draft uses this format
for all 107 of its notes; without this support, v2 was structurally
incapable of running on real-world docx-derived inputs.

src/cmos/parser.py:
- Add NoteDefinition.original_prefix field so reassembly can
  round-trip the source's marker syntax (pandoc input → pandoc
  output, numbered input → numbered output) without consumers
  needing to know which format was matched.
- Update find_notes() to populate original_prefix as "[^N]: ".
- Add find_numbered_notes() targeting "[N] text" definitions.
  Marker must be all digits (rejects [Smith 2020], [foo], etc.);
  caret prefix is rejected (rejects pandoc-style cleanly).

src/cmos/cli.py:
- reformat_notes now auto-detects format: tries find_notes first,
  falls back to find_numbered_notes if no pandoc definitions found.
  Uses definition.original_prefix for reassembly so both formats
  round-trip correctly.

tests/test_parser.py:
- 11 new tests for find_numbered_notes covering: single/multiple
  definitions, multi-digit markers, line number recording, trailing
  whitespace stripping, ignoring pandoc/non-numeric markers, original
  prefix recording, and the actual Anti-Communist draft format.
- 1 new test for find_notes original_prefix population.

tests/test_cli.py:
- 2 new tests for reformat_notes auto-detect: numbered input round-
  trips as numbered output, pandoc input still round-trips as pandoc.

Total suite: 136/136 (was 123, +13 net new). v1 untouched, 96/96
v1 tests still passing.

Real-draft validation: ran cmos format-notes against the 107-note
Anti-Communist Formations of LIS draft (108 calls in parallel via
the existing concurrency=8 thread pool, completed cleanly). Output
saved to /tmp (not committed). All 107 notes preserved through the
pipeline; ~30-40 first-occurrence full notes produced clean CMOS 18
note form; ~30 shortened-form refs correctly left unchanged;
2 real bugs surfaced for the next iteration (empty input → conver-
sational reply, Ibid → empty string), plus several lower-priority
issues (substantive note truncation, retry waste on shortened
forms, lossy month dropping). Not addressed in this chunk per the
"collect signal, don't fix" plan.

src/cmos/cli.py

@@ -23,7 +23,7 @@ from typing import Callable
 
 from cmos.formatter import format_bibliography_entry
 from cmos.note_formatter import format_note_entry
-from cmos.parser import find_notes, split_bibliography
+from cmos.parser import find_notes, find_numbered_notes, split_bibliography
 
 FormatterFn = Callable[[str], str]
 
@@ -75,31 +75,46 @@ def reformat_notes(
     formatter: FormatterFn | None = None,
     concurrency: int = DEFAULT_CONCURRENCY,
 ) -> str:
-    """Rewrite pandoc-style markdown footnote definitions in ``text``.
+    """Rewrite footnote definitions in ``text`` to CMOS 18 note form.
 
-    Finds every ``[^marker]: definition text`` line via
-    ``cmos.parser.find_notes``, formats each definition's text via
-    ``formatter`` (the v2 note formatter by default), and substitutes
-    the reformatted text back into the same line position. All
-    non-definition lines are preserved byte-for-byte.
+    Auto-detects the source format by trying ``cmos.parser.find_notes``
+    (pandoc-style ``[^marker]: text``) first and falling back to
+    ``cmos.parser.find_numbered_notes`` (``[N] text``) if no pandoc
+    definitions were found. Both formats are common in real drafts:
+    pandoc is used by writers who author directly in markdown, while
+    the numbered form is what docx-to-text conversion typically
+    produces from footnoted Word documents.
+
+    Each definition's body is formatted via ``formatter`` (the v2
+    note formatter by default) and substituted back into the same
+    line position using the definition's ``original_prefix`` — so
+    pandoc input round-trips as pandoc output and numbered input
+    round-trips as numbered output. All non-definition lines are
+    preserved byte-for-byte.
 
     Definitions are reformatted concurrently using a thread pool of
     size ``concurrency``, mirroring v1's ``reformat_draft``. The output
     order is deterministic regardless of which API call finishes first.
 
-    If the document contains no footnote definitions, returns ``text``
-    unchanged — a draft with no notes is a valid (if uninteresting)
-    input rather than an error.
+    If the document contains no footnote definitions in either format,
+    returns ``text`` unchanged — a draft with no notes is a valid (if
+    uninteresting) input rather than an error.
 
-    Scope of v2 chunk 2b (this version): single-line definitions only.
-    Multi-line definitions (where the body continues on indented
-    subsequent lines) are out of scope — only the first line is
-    reformatted, leaving any continuation lines untouched. Real drafts
-    that use multi-line definitions will need a future parser
-    extension before this CLI can handle them safely.
+    Scope: single-line definitions only. Multi-line definitions (where
+    the body continues on indented subsequent lines for pandoc, or on
+    unprefixed continuation lines for numbered) are out of scope —
+    only the first line is reformatted, leaving any continuation lines
+    untouched.
     """
     fmt = formatter or format_note_entry
+
+    # Try pandoc format first; fall back to numbered format if nothing
+    # was found. The two parsers are mutually exclusive on well-formed
+    # input (the regex anchors prevent overlap), so this fallback is
+    # unambiguous in practice.
     parsed = find_notes(text)
+    if not parsed.definitions:
+        parsed = find_numbered_notes(text)
 
     if not parsed.definitions:
         return text
@@ -113,7 +128,7 @@ def reformat_notes(
 
     lines = text.splitlines()
     for definition, formatted in zip(parsed.definitions, rewritten):
-        lines[definition.line_number] = f"[^{definition.marker}]: {formatted}"
+        lines[definition.line_number] = f"{definition.original_prefix}{formatted}"
 
     # Preserve trailing newline if the original had one — splitlines drops it.
     trailing = "\n" if text.endswith("\n") else ""

src/cmos/parser.py

@@ -5,18 +5,32 @@ section as running until the next level-2 (``##``) heading or end of file,
 and return each non-blank line as one entry. Bullet markers (``- ``, ``* ``)
 at the start of a line are stripped.
 
-v2 scope (notes): ``find_notes`` locates pandoc-style markdown footnote
-definitions ``[^marker]: text`` anywhere in the document. Single-line
-definitions only — multi-line continuation (indented continuation lines)
-is deferred. The reference markers in prose (the ``[^marker]`` references
-themselves, without the trailing colon) are NOT collected; only the
-definitions need reformatting.
+v2 scope (notes): two complementary functions for finding footnote
+definitions in two distinct source formats. Both return the same
+``NotesParseResult`` shape; the consumer (typically the CLI) can call
+one and fall back to the other to auto-detect format.
+
+- ``find_notes`` finds pandoc-style markdown footnote definitions
+  ``[^marker]: text``. The marker can be numeric or named.
+- ``find_numbered_notes`` finds plain ``[N] text`` definitions where
+  the marker is digit-only and there is no caret or colon — the format
+  produced by docx-to-text conversion of footnoted Word docs.
+
+Each ``NoteDefinition`` carries the marker, the body text (with
+surrounding whitespace stripped), the line number where it appeared
+(0-indexed, for future reassembly), and the literal ``original_prefix``
+string (e.g., ``"[^1]: "`` or ``"[1] "``) so that reassembly can
+round-trip the source's marker syntax without the consumer needing to
+know which format was matched.
+
+Single-line definitions only in both formats — multi-line continuation
+(indented continuation lines) is deferred. The reference markers in
+prose are NOT collected; only the definitions need reformatting.
 
 Out of scope in v1: in-prose citation rewriting, multi-line entries,
 nested sections, alternative heading names (e.g., "Works Cited",
 "References"). Out of scope in v2 (so far): multi-line note definitions,
-shortened-form generation, reassembly of formatted notes back into the
-document.
+shortened-form generation.
 """
 
 from __future__ import annotations
@@ -38,18 +52,24 @@ class ParseResult:
 
 @dataclass
 class NoteDefinition:
-    """A single pandoc-style markdown footnote definition.
+    """A single footnote definition extracted from a draft.
 
-    ``marker`` is the identifier between ``[^`` and ``]:`` (e.g., ``"1"``
-    or ``"smith2020"``). ``text`` is the note body with surrounding
-    whitespace stripped. ``line_number`` is the 0-indexed line in the
-    source document where the definition appeared — currently unused but
-    captured for future reassembly support.
+    ``marker`` is the identifier between the brackets — e.g., ``"1"`` or
+    ``"smith2020"`` for pandoc-style ``[^1]:`` / ``[^smith2020]:``, or
+    ``"107"`` for the numbered ``[107]`` style. ``text`` is the note
+    body with surrounding whitespace stripped. ``line_number`` is the
+    0-indexed line in the source document where the definition appeared
+    (used for in-place reassembly). ``original_prefix`` is the literal
+    text that preceded the body on the source line — typically
+    ``"[^1]: "`` for pandoc or ``"[1] "`` for numbered. Reassembly
+    concatenates ``original_prefix`` with the reformatted body, so
+    consumers do not need to know which format was matched.
     """
 
     marker: str
     text: str
     line_number: int
+    original_prefix: str = ""
 
 
 @dataclass
@@ -61,6 +81,7 @@ _HEADING_RE = re.compile(r"^##\s+bibliography\s*$", re.IGNORECASE)
 _LEVEL_TWO_RE = re.compile(r"^##\s+\S")
 _BULLET_RE = re.compile(r"^[-*]\s+")
 _NOTE_DEF_RE = re.compile(r"^\[\^([^\]]+)\]:\s*(.*?)\s*$")
+_NUMBERED_NOTE_RE = re.compile(r"^\[(\d+)\]\s+(.*?)\s*$")
 
 
 def split_bibliography(text: str) -> ParseResult:
@@ -111,6 +132,10 @@ def find_notes(text: str) -> NotesParseResult:
     appear in the document. If the document contains no footnote
     definitions, returns an empty list rather than raising — a draft
     with no notes is a valid (if uninteresting) input.
+
+    Each definition's ``original_prefix`` is set to ``"[^MARKER]: "`` so
+    that downstream reassembly can write back the formatted body using
+    the same marker syntax.
     """
     definitions: list[NoteDefinition] = []
     for line_number, line in enumerate(text.splitlines()):
@@ -120,6 +145,52 @@ def find_notes(text: str) -> NotesParseResult:
         marker = match.group(1)
         body = match.group(2)
         definitions.append(
-            NoteDefinition(marker=marker, text=body, line_number=line_number)
+            NoteDefinition(
+                marker=marker,
+                text=body,
+                line_number=line_number,
+                original_prefix=f"[^{marker}]: ",
+            )
+        )
+    return NotesParseResult(definitions=definitions)
+
+
+def find_numbered_notes(text: str) -> NotesParseResult:
+    """Find ``[N] text`` style footnote definitions in ``text``.
+
+    Targets the format produced by docx-to-text conversion of footnoted
+    Word documents: a square-bracket numeric marker followed by a
+    space and the note body, one definition per line. Distinct from
+    pandoc footnote definitions, which require ``[^marker]:`` syntax
+    (caret and colon).
+
+    The marker must be ALL DIGITS — markers like ``[Smith 2020]`` or
+    ``[foo]`` are deliberately not matched, since they could plausibly
+    be many other things (citation keys, link labels, in-line
+    references). Pandoc-style ``[^1]:`` is also not matched because of
+    the leading ``^``.
+
+    Multi-line definitions (where the body wraps onto a continuation
+    line) are out of scope; only the first line is captured.
+
+    Returns a ``NotesParseResult`` with definitions in source order.
+    Each definition's ``original_prefix`` is set to ``"[N] "`` so that
+    downstream reassembly can write back the formatted body using the
+    same marker syntax.
+    """
+    definitions: list[NoteDefinition] = []
+    for line_number, line in enumerate(text.splitlines()):
+        match = _NUMBERED_NOTE_RE.match(line)
+        if match is None:
+            continue
+        marker = match.group(1)
+        body = match.group(2)
+        definitions.append(
+            NoteDefinition(
+                marker=marker,
+                text=body,
+                line_number=line_number,
+                original_prefix=f"[{marker}] ",
+            )
         )
     return NotesParseResult(definitions=definitions)

tests/test_cli.py

@@ -195,6 +195,37 @@ def test_reformat_notes_preserves_trailing_newline():
     assert not reformat_notes(without_trailing, formatter=_fake_note_formatter).endswith("\n")
 
 
+def test_reformat_notes_auto_detects_numbered_format():
+    """Real drafts converted from docx use [N] text format, not pandoc.
+    reformat_notes must auto-detect: try pandoc first, fall back to
+    numbered. Reassembly must use the original prefix (e.g. "[1] ") so
+    the output round-trips in the source's marker syntax."""
+    draft = """\
+## Notes
+
+[1] yu, charles. interior chinatown. 2020. p 45.
+[2] kwon, hyeyoung. inclusion work. 2022, p 1830.
+"""
+    output = reformat_notes(draft, formatter=_fake_note_formatter)
+    # Numbered prefix preserved, NOT rewritten as pandoc.
+    assert "[1] NOTE_FORMATTED(yu, charles. interior chinatown. 2020. p 45.)" in output
+    assert "[2] NOTE_FORMATTED(kwon, hyeyoung. inclusion work. 2022, p 1830.)" in output
+    # No pandoc-style markers should appear in the output (we did not
+    # auto-convert numbered to pandoc).
+    assert "[^1]" not in output
+    assert "[^2]" not in output
+
+
+def test_reformat_notes_pandoc_input_still_round_trips_as_pandoc():
+    """Regression: after the auto-detect change, pandoc-format input
+    must still produce pandoc-format output. The original_prefix path
+    must work for both formats."""
+    draft = "[^1]: messy.\n"
+    output = reformat_notes(draft, formatter=_fake_note_formatter)
+    assert "[^1]: NOTE_FORMATTED(messy.)" in output
+    assert "[1] " not in output  # numbered prefix must NOT appear
+
+
 def test_reformat_notes_preserves_order_under_concurrency():
     """With concurrent execution the formatter is called on all definitions
     in parallel; the output must reassemble them in input order regardless

tests/test_parser.py

@@ -5,14 +5,31 @@ the prose before it, the list of entries inside it, and the prose after it.
 The section ends at the next ``##`` (level-2) heading or end of file.
 Blank lines and bullet markers (``- `` or ``* ``) are stripped from entries.
 
-v2 (notes) parser scope: find pandoc-style markdown footnote definitions
-``[^marker]: text`` anywhere in the document. Single-line definitions only;
-multi-line continuation is deferred to a later chunk.
+v2 (notes) parser scope: two complementary functions for finding footnote
+definitions in two distinct source formats:
+
+- ``find_notes`` finds pandoc-style ``[^marker]: text`` definitions
+  (caret-prefixed marker, colon after the closing bracket).
+- ``find_numbered_notes`` finds ``[N] text`` definitions (square bracket
+  numeric marker, no caret, no colon, just a space before the body) —
+  the format produced by docx-to-text conversion of footnoted Word docs.
+
+Both functions return ``NotesParseResult`` with a list of ``NoteDefinition``
+records carrying the marker, body text, line number, and the original
+prefix string (so reassembly can round-trip the source's marker syntax).
+
+Single-line definitions only in both formats; multi-line continuation is
+deferred.
 """
 
 import pytest
 
-from cmos.parser import NoBibliographyError, find_notes, split_bibliography
+from cmos.parser import (
+    NoBibliographyError,
+    find_notes,
+    find_numbered_notes,
+    split_bibliography,
+)
 
 
 def test_three_entries_under_bibliography_heading():
@@ -159,3 +176,119 @@ Some prose that cites it.[^1]
     result = find_notes(draft)
     assert len(result.definitions) == 1
     assert result.definitions[0].text == "A definition before the reference."
+
+
+def test_find_notes_records_original_prefix_for_pandoc_format():
+    """Reassembly needs the literal prefix string so the round-trip
+    preserves the source's marker syntax. For pandoc this is `[^N]: `."""
+    draft = "[^1]: text.\n[^smith2020]: text.\n"
+    result = find_notes(draft)
+    assert result.definitions[0].original_prefix == "[^1]: "
+    assert result.definitions[1].original_prefix == "[^smith2020]: "
+
+
+# --- find_numbered_notes ([N] text format, no caret, no colon) -------------
+
+
+def test_find_numbered_notes_single_definition():
+    draft = """\
+Some prose.
+
+[1] Charles Yu, Interior Chinatown. Pantheon, 2020. p. 45.
+"""
+    result = find_numbered_notes(draft)
+    assert len(result.definitions) == 1
+    assert result.definitions[0].marker == "1"
+    assert result.definitions[0].text == "Charles Yu, Interior Chinatown. Pantheon, 2020. p. 45."
+
+
+def test_find_numbered_notes_multiple_sequential():
+    draft = """\
+[1] First.
+[2] Second.
+[3] Third.
+"""
+    result = find_numbered_notes(draft)
+    assert len(result.definitions) == 3
+    assert [d.marker for d in result.definitions] == ["1", "2", "3"]
+    assert [d.text for d in result.definitions] == ["First.", "Second.", "Third."]
+
+
+def test_find_numbered_notes_multi_digit_markers():
+    """The Anti-Communist draft has 107 notes, including [107]. Multi-digit
+    markers must work."""
+    draft = "[1] one.\n[42] forty-two.\n[107] one-oh-seven.\n"
+    result = find_numbered_notes(draft)
+    assert [d.marker for d in result.definitions] == ["1", "42", "107"]
+
+
+def test_find_numbered_notes_returns_empty_when_no_definitions():
+    draft = "Just prose with no numbered notes.\n"
+    result = find_numbered_notes(draft)
+    assert result.definitions == []
+
+
+def test_find_numbered_notes_records_line_number():
+    draft = """\
+Line zero.
+Line one.
+
+[1] Note on line three.
+"""
+    result = find_numbered_notes(draft)
+    assert result.definitions[0].line_number == 3
+
+
+def test_find_numbered_notes_strips_trailing_whitespace_from_text():
+    draft = "[1] Note text with trailing spaces.   \n"
+    result = find_numbered_notes(draft)
+    assert result.definitions[0].text == "Note text with trailing spaces."
+
+
+def test_find_numbered_notes_ignores_pandoc_format():
+    """`[^1]: text` is pandoc format, not numbered. find_numbered_notes
+    must NOT match it (or the two parsers would step on each other when
+    used in fallback fashion). The `^` after `[` disqualifies it."""
+    draft = "[^1]: pandoc-style note.\n"
+    result = find_numbered_notes(draft)
+    assert result.definitions == []
+
+
+def test_find_numbered_notes_ignores_non_numeric_markers():
+    """A marker like [Smith 2020] or [foo] is NOT a numbered footnote
+    definition — it could be many other things (citation reference, link
+    label, etc.). Only digit markers count."""
+    draft = """\
+[Smith 2020] Some text — looks like an author-date reference.
+[foo] some kind of label.
+[1] Actual numbered note.
+"""
+    result = find_numbered_notes(draft)
+    assert len(result.definitions) == 1
+    assert result.definitions[0].marker == "1"
+
+
+def test_find_numbered_notes_records_original_prefix():
+    """For numbered format the prefix is `[N] ` — square bracket, number,
+    closing bracket, single space. Reassembly will use this verbatim."""
+    draft = "[1] text.\n[107] text.\n"
+    result = find_numbered_notes(draft)
+    assert result.definitions[0].original_prefix == "[1] "
+    assert result.definitions[1].original_prefix == "[107] "
+
+
+def test_find_numbered_notes_handles_anti_communist_draft_format():
+    """Smoke test mirroring the actual Anti-Communist Formations draft:
+    notes follow a `## Notes` heading, separated by `________________` rule,
+    one definition per line, sequentially numbered."""
+    draft = """\
+## Notes
+________________
+[1] First reference. https://example.org/path
+[2] Second reference, abbreviated.
+[3] Rosen, 7.
+"""
+    result = find_numbered_notes(draft)
+    assert len(result.definitions) == 3
+    assert result.definitions[0].text == "First reference. https://example.org/path"
+    assert result.definitions[2].text == "Rosen, 7."