v0.15.0: sync pushes user attachments verbatim

CHANGELOG.md

@@ -1,5 +1,31 @@
 # Changelog
 
+## v0.15.0 — feat: sync pushes user attachments verbatim
+
+### Added
+
+* **`millet/sync.py`** — files a user drops into `<session>/attachments/` are
+  now pushed into an `attachments/` subdirectory of the meeting folder, with
+  their **names intact**. `_collect_files()` gained `_collect_attachments()`,
+  which deliberately bypasses both `PUSH_SUFFIXES` and the descriptive-rename
+  map: the map keys off the suffix alone, so an attached `slides.pdf` would
+  land as `transcript.pdf` (or force the real transcript to keep its raw
+  name), and the allowlist dropped every image, office document and video —
+  most of what people attach to a meeting. The copy loop in `sync_session()`
+  now creates `dest.parent` so the subdirectory prefix works; `git add`
+  needed no change.
+
+  Guards: symlinks are skipped (the collected pairs are copied into a git
+  clone, and a link could point anywhere on the host), as are dotfiles and
+  nested directories. `MAX_ATTACHMENTS` (50) and `MAX_ATTACHMENTS_BYTES`
+  (100 MB) cap what a mis-aimed folder can push into the archive repo.
+  Sessions without an `attachments/` directory collect exactly what they did
+  before.
+
+  This is the upstream half of vezir's attachment-folder workflow: vezir
+  stores attachments in the session directory it hands to `millet sync`, so
+  nothing further is needed on that side to get them into the team repo.
+
 ## v0.14.1 — fix: speaker relabel swap collapsed both speakers to one name
 
 ### Fixed — correctness

millet/__init__.py

@@ -51,4 +51,4 @@
         # may still be the one installed.  Honor it as a fallback.
         __version__ = _pkg_version("meetscribe-offline")
 except Exception:
-    __version__ = "0.14.1"
+    __version__ = "0.15.0"

millet/sync.py

@@ -154,6 +154,12 @@ def _resolve_sync_config_path(
 # ─── Files to push (by suffix) ───────────────────────────────────────────────
 
 PUSH_SUFFIXES = {".md", ".txt", ".pdf", ".srt", ".json"}
+# User-supplied meeting attachments live in this subdirectory of the session
+# dir and are pushed verbatim into the same subdirectory of the meeting folder.
+# See _collect_attachments.
+ATTACHMENTS_SUBDIR = "attachments"
+MAX_ATTACHMENTS = 50
+MAX_ATTACHMENTS_BYTES = 100 * 1024 * 1024
 # Exclude session metadata and large raw files
 EXCLUDE_PATTERNS = {
     ".session.json",
@@ -696,6 +702,49 @@ def _collect_files(session_dir: Path) -> list[tuple[Path, str]]:
         used_names.add(dest_name)
 
         result.append((f, dest_name))
+
+    result.extend(_collect_attachments(session_dir))
+    return result
+
+
+def _collect_attachments(session_dir: Path) -> list[tuple[Path, str]]:
+    """Return (source, ``attachments/<name>``) pairs for user-supplied files.
+
+    Files a user dropped into ``<session>/attachments/`` — slides, agendas,
+    screenshots — are pushed verbatim: neither ``PUSH_SUFFIXES`` nor the
+    descriptive-rename map above applies to them.  Both would be wrong here.
+    The rename map keys off the suffix alone, so an attached ``slides.pdf``
+    would land as ``transcript.pdf``, and the suffix allowlist would drop
+    every image, office document and video.
+
+    Symlinks are skipped: the pairs returned here are copied into a git
+    clone, and a link could point anywhere on the host.  Count and total
+    size are capped so a mis-aimed folder can't bloat the archive repo.
+    """
+    adir = session_dir / ATTACHMENTS_SUBDIR
+    if not adir.is_dir():
+        return []
+
+    result: list[tuple[Path, str]] = []
+    total = 0
+    for f in sorted(adir.iterdir()):
+        if f.is_symlink() or not f.is_file() or f.name.startswith("."):
+            continue
+        if len(result) >= MAX_ATTACHMENTS:
+            log.warning(
+                "sync: more than %d attachments in %s — pushing the first %d only",
+                MAX_ATTACHMENTS, adir, MAX_ATTACHMENTS,
+            )
+            break
+        size = f.stat().st_size
+        if total + size > MAX_ATTACHMENTS_BYTES:
+            log.warning(
+                "sync: attachments in %s exceed %d bytes — skipping %s and the rest",
+                adir, MAX_ATTACHMENTS_BYTES, f.name,
+            )
+            break
+        total += size
+        result.append((f, f"{ATTACHMENTS_SUBDIR}/{f.name}"))
     return result
 
 
@@ -819,6 +868,8 @@ def _log(msg: str) -> None:
     copied: list[Path] = []
     for src, dest_name in source_files:
         dest = target_dir / dest_name
+        # dest_name carries an "attachments/" prefix for attached files.
+        dest.parent.mkdir(parents=True, exist_ok=True)
         shutil.copy2(src, dest)
         copied.append(dest)
         _log(f"  Staged: {dest.relative_to(repo)}")

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "millet-pipeline"
-version = "0.14.1"
+version = "0.15.0"
 description = "Meeting transcription pipeline (diarization, AI summaries, PDF output) under the vezir ecosystem.  Builds on millet-record for capture.  Named after the Ottoman millet system.  Successor to meetscribe-offline."
 readme = "README.md"
 requires-python = ">=3.10"

tests/test_sync_core.py

@@ -356,6 +356,108 @@ def test_sync_session_disambiguates_different_session(
     assert suffixed, "second session must land in a disambiguated folder"
 
 
+# ─── attachments passthrough ─────────────────────────────────────────────────
+
+
+def _attach(sdir: Path, name: str, content: bytes = b"data") -> Path:
+    adir = sdir / sync.ATTACHMENTS_SUBDIR
+    adir.mkdir(exist_ok=True)
+    p = adir / name
+    p.write_bytes(content)
+    return p
+
+
+def test_collect_attachments_bypasses_allowlist_and_rename_map(tmp_path):
+    """Attachment names are the user's and must survive verbatim — the
+    suffix-keyed rename map would turn slides.pdf into transcript.pdf and
+    PUSH_SUFFIXES would drop images and office documents entirely."""
+    sdir = _make_session(tmp_path, "meeting-20260706-100000", "01ATTACH")
+    _attach(sdir, "slides.pdf")
+    _attach(sdir, "diagram.png")
+    _attach(sdir, "agenda.pptx")
+
+    dests = dict((d, s) for s, d in sync._collect_files(sdir))
+    # The session's own transcript still gets its descriptive name.
+    assert "transcript.txt" in dests
+    # Attachments keep theirs, under the subdir, whatever the suffix.
+    assert "attachments/slides.pdf" in dests
+    assert "attachments/diagram.png" in dests
+    assert "attachments/agenda.pptx" in dests
+    # Nothing was renamed into the pipeline's namespace.
+    assert "transcript.pdf" not in dests
+
+
+def test_collect_attachments_skips_symlinks_dotfiles_and_subdirs(tmp_path):
+    """These pairs get copied into a git clone: a symlink could point
+    anywhere on the host."""
+    sdir = _make_session(tmp_path, "meeting-20260706-100000", "01ATTACH")
+    _attach(sdir, "keep.pdf")
+    _attach(sdir, ".hidden")
+    outside = tmp_path / "secret.txt"
+    outside.write_text("ssh key")
+    (sdir / sync.ATTACHMENTS_SUBDIR / "link.txt").symlink_to(outside)
+    (sdir / sync.ATTACHMENTS_SUBDIR / "nested").mkdir()
+
+    dests = [d for _, d in sync._collect_files(sdir)]
+    assert "attachments/keep.pdf" in dests
+    assert not any(
+        d.endswith(("link.txt", ".hidden", "nested")) for d in dests
+    ), dests
+
+
+def test_collect_files_unchanged_without_attachments_dir(tmp_path):
+    sdir = _make_session(tmp_path, "meeting-20260706-100000", "01ATTACH")
+    dests = [d for _, d in sync._collect_files(sdir)]
+    assert dests == ["summary.md", "transcript.txt"]
+
+
+def test_collect_attachments_caps_count(monkeypatch, tmp_path):
+    monkeypatch.setattr(sync, "MAX_ATTACHMENTS", 2)
+    sdir = _make_session(tmp_path, "meeting-20260706-100000", "01ATTACH")
+    for i in range(5):
+        _attach(sdir, f"f{i}.png")
+
+    attached = [d for _, d in sync._collect_files(sdir) if d.startswith("attachments/")]
+    assert attached == ["attachments/f0.png", "attachments/f1.png"]
+
+
+def test_collect_attachments_caps_total_bytes(monkeypatch, tmp_path):
+    monkeypatch.setattr(sync, "MAX_ATTACHMENTS_BYTES", 20)
+    sdir = _make_session(tmp_path, "meeting-20260706-100000", "01ATTACH")
+    _attach(sdir, "a.png", b"x" * 15)
+    _attach(sdir, "b.png", b"x" * 15)
+
+    attached = [d for _, d in sync._collect_files(sdir) if d.startswith("attachments/")]
+    assert attached == ["attachments/a.png"]
+
+
+def test_sync_session_pushes_attachments_subdir(monkeypatch, tmp_path, local_remote):
+    monkeypatch.setattr(sync, "CLONE_BASE_DIR", tmp_path / "clones")
+    monkeypatch.setattr(
+        sync, "load_sync_config",
+        lambda team=None, config_path=None: {
+            "repo_url": str(local_remote), "meetings": [],
+        },
+    )
+
+    sdir = _make_session(tmp_path, "meeting-20260706-100000", "01TESTULID")
+    _attach(sdir, "slides.pdf", b"%PDF-1.4 slides")
+    _attach(sdir, "photo of board.png", b"\x89PNG board")
+    match = sync.MeetingMatch(name="Weekly", folder="weekly")
+    sync.sync_session(sdir, match, progress_callback=lambda m: None)
+
+    check = tmp_path / "check"
+    _git("clone", str(local_remote), str(check))
+    meeting_dir = check / "meetings" / "2026-07-06_weekly"
+    assert (meeting_dir / "attachments" / "slides.pdf").read_bytes() == b"%PDF-1.4 slides"
+    assert (
+        meeting_dir / "attachments" / "photo of board.png"
+    ).read_bytes() == b"\x89PNG board"
+    # The pipeline's own artifacts are untouched by the attachment pass.
+    assert (meeting_dir / "transcript.txt").read_text() == "transcript\n"
+    assert not (meeting_dir / "transcript.pdf").exists()
+
+
 # ─── sanity: dates used above ────────────────────────────────────────────────