fix(ci): pkg_inventory parser drops inline comments cleanly

The pyproject dep parser did chained `.strip().strip(",").strip('"')`
which left trailing junk on lines with inline comments. The line:
    "pymupdf>=1.24",     # AGPL-3.0 — flag if shipping outside …
became the spec `pymupdf>=1.24",` — pip rejected it and the
pkg-supply-chain workflow failed on every push (#400-403).

Replace the four duplicated chained-strip blocks with a single
`_extract_dep_specs` helper that pulls the first quoted string from
each non-comment line. Same call from prod/dev/optional/build-system.

Manifest pypi count: 62 → 66 (parser was previously dropping a few
specs entirely, not just mangling pymupdf).

Add tests/dev/test_pkg_inventory.py with 4 regression tests covering
inline comments, blank/comment lines, single-quoted strings, and a
full pyproject excerpt with the bug pattern.

Closes the noise from #400-403.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

data/pkg-manifest.json

@@ -1,6 +1,234 @@
 {
-  "apt": {},
-  "apk": {},
+  "apt": {
+    "infra/images/selkies/Dockerfile.upstream": [
+      "\"./virtualgl32_${VIRTUALGL_VERSION}_amd64.deb\"",
+      "\"./virtualgl_${VIRTUALGL_VERSION}_amd64.deb\"",
+      "./google-chrome-stable.deb",
+      "./heroic_launcher.deb",
+      "./kasmvncserver.deb",
+      "./lutris.deb",
+      "./rustdesk.deb",
+      "./selkies-js-interposer.deb",
+      "./virtualgl_${VIRTUALGL_VERSION}_arm64.deb",
+      "GStreamer",
+      "Operating",
+      "adwaita-icon-theme-full",
+      "aom-tools",
+      "appmenu-gtk3-module",
+      "apt-utils",
+      "ark",
+      "aspell",
+      "aspell-en",
+      "breeze",
+      "breeze-cursor-theme",
+      "breeze-gtk-theme",
+      "breeze-icon-theme",
+      "dbus-user-session",
+      "dbus-x11",
+      "debconf-kde-helper",
+      "dependencies",
+      "desktop-file-utils",
+      "dolphin",
+      "dolphin-plugins",
+      "enchant-2",
+      "fakeroot",
+      "fcitx",
+      "fcitx-frontend-gtk2",
+      "fcitx-frontend-gtk3",
+      "fcitx-frontend-qt5",
+      "fcitx-hangul",
+      "fcitx-libpinyin",
+      "fcitx-m17n",
+      "fcitx-module-dbus",
+      "fcitx-module-kimpanel",
+      "fcitx-module-lua",
+      "fcitx-module-x11",
+      "fcitx-mozc",
+      "fcitx-sayura",
+      "fcitx-tools",
+      "fcitx-unikey",
+      "filelight",
+      "firefox",
+      "frameworkintegration",
+      "fuse",
+      "gir1.2-wp-0.5",
+      "gstreamer1.0-libcamera",
+      "gstreamer1.0-pipewire",
+      "gstreamer1.0-plugins-bad",
+      "gwenview",
+      "haveged",
+      "hunspell",
+      "i965-va-driver-shaders",
+      "i965-va-driver-shaders:i386",
+      "im-config",
+      "intel-gpu-tools",
+      "intel-media-va-driver-non-free",
+      "intel-media-va-driver-non-free:i386",
+      "kcalc",
+      "kcharselect",
+      "kde-baseapps",
+      "kde-config-fcitx",
+      "kde-config-gtk-style",
+      "kde-config-gtk-style-preview",
+      "kde-spectacle",
+      "kdeadmin",
+      "kdeconnect",
+      "kdegraphics-thumbnailers",
+      "kdf",
+      "kdialog",
+      "kfind",
+      "kget",
+      "khotkeys",
+      "kimageformat-plugins",
+      "kinfocenter",
+      "kio",
+      "kio-extras",
+      "kmag",
+      "kmenuedit",
+      "kmix",
+      "kmod",
+      "kmousetool",
+      "kmouth",
+      "ksshaskpass",
+      "ktimer",
+      "kwin-addons",
+      "kwin-x11",
+      "kwrite",
+      "libc6:i386",
+      "libdatetime-perl",
+      "libdbusmenu-glib4",
+      "libdbusmenu-gtk3-4",
+      "libdrm2:i386",
+      "libegl-dev",
+      "libegl1:i386",
+      "libffmpeg-nvenc-dev",
+      "libgail-common",
+      "libgdk-pixbuf2.0-bin",
+      "libgl1:i386",
+      "libgles1:i386",
+      "libgles2:i386",
+      "libglu1:i386",
+      "libglvnd0:i386",
+      "libglx0:i386",
+      "libgstreamer-plugins-bad1.0-dev",
+      "libgtk-3-bin",
+      "libgtk2.0-bin",
+      "libkf5baloowidgets-bin",
+      "libkf5dbusaddons-bin",
+      "libkf5iconthemes-bin",
+      "libkf5kdelibs4support5-bin",
+      "libkf5khtml-bin",
+      "libkf5parts-plugins",
+      "libopengl0:i386",
+      "libopenh264-dev",
+      "libpipewire-0.3-modules",
+      "libpipewire-module-x11-bell",
+      "libqt5multimedia5-plugins",
+      "libreoffice",
+      "libreoffice-kf5",
+      "libreoffice-plasma",
+      "libreoffice-style-breeze",
+      "librsvg2-common",
+      "libsm6:i386",
+      "libspa-0.2-bluetooth",
+      "libspa-0.2-jack",
+      "libspa-0.2-modules",
+      "libva-dev",
+      "libva2:i386",
+      "libvulkan-dev:i386",
+      "libx11-6:i386",
+      "libxau6:i386",
+      "libxcb1:i386",
+      "libxdmcp6:i386",
+      "libxext6:i386",
+      "libxtst6:i386",
+      "libxv1:i386",
+      "locales",
+      "media-player-info",
+      "mesa-utils-extra",
+      "mesa-vulkan-drivers:i386",
+      "meson",
+      "nvtop",
+      "okular",
+      "okular-extra-backends",
+      "packages",
+      "pipewire",
+      "pipewire-alsa",
+      "pipewire-audio-client-libraries",
+      "pipewire-jack",
+      "pipewire-libcamera",
+      "pipewire-locales",
+      "pipewire-v4l2",
+      "pipewire-vulkan",
+      "plasma-browser-integration",
+      "plasma-calendar-addons",
+      "plasma-dataengines-addons",
+      "plasma-desktop",
+      "plasma-discover",
+      "plasma-integration",
+      "plasma-runners-addons",
+      "plasma-widgets-addons",
+      "plasma-workspace",
+      "playonlinux",
+      "print-manager",
+      "q4wine",
+      "qapt-deb-installer",
+      "qml-module-org-kde-qqc2desktopstyle",
+      "qml-module-org-kde-runnermodel",
+      "qml-module-qt-labs-platform",
+      "qml-module-qtgraphicaleffects",
+      "qml-module-qtquick-xmllistmodel",
+      "qt5-gtk-platformtheme",
+      "qt5-image-formats-plugins",
+      "qt5-style-plugins",
+      "qtspeech5-flite-plugin",
+      "qtvirtualkeyboard-plugin",
+      "software-properties-qt",
+      "sonnet-plugins",
+      "ssl-cert",
+      "sudo",
+      "svt-av1",
+      "sweeper",
+      "system",
+      "systemsettings",
+      "transmission-qt",
+      "tzdata",
+      "ubuntu-drivers-common",
+      "udev",
+      "va-driver-all",
+      "va-driver-all:i386",
+      "vdpau-driver-all:i386",
+      "vlc",
+      "vlc-plugin-access-extra",
+      "vlc-plugin-notify",
+      "vlc-plugin-samba",
+      "vlc-plugin-skins2",
+      "vlc-plugin-video-splitter",
+      "vlc-plugin-visualization",
+      "winehq-${WINE_BRANCH}",
+      "wireplumber",
+      "wireplumber-locales",
+      "xcvt",
+      "xdg-user-dirs",
+      "xdg-utils",
+      "xvfb"
+    ]
+  },
+  "apk": {
+    "infra/gitea/Dockerfile": [
+      "bash",
+      "build-base",
+      "ca-certificates",
+      "curl",
+      "dumb-init",
+      "gettext",
+      "git",
+      "gnupg",
+      "nodejs",
+      "npm",
+      "openssh-keygen"
+    ]
+  },
   "pypi": {
     "project_prod": [
       {
@@ -8,6 +236,21 @@
         "version": ">=2.0.0",
         "extras": ""
       },
+      {
+        "name": "pymupdf",
+        "version": ">=1.24",
+        "extras": ""
+      },
+      {
+        "name": "python-docx",
+        "version": ">=1.1",
+        "extras": ""
+      },
+      {
+        "name": "xlrd",
+        "version": ">=2.0.2",
+        "extras": ""
+      },
       {
         "name": "pydantic",
         "version": ">=2.0.0",
@@ -65,7 +308,7 @@
       },
       {
         "name": "cryptography",
-        "version": ">=46.0.5",
+        "version": ">=46.0.7",
         "extras": ""
       },
       {
@@ -108,6 +351,21 @@
         "version": ">=0.41.0",
         "extras": ""
       },
+      {
+        "name": "httpx",
+        "version": ">=0.28.1",
+        "extras": ""
+      },
+      {
+        "name": "pydo",
+        "version": ">=0.29.0",
+        "extras": ""
+      },
+      {
+        "name": "resend",
+        "version": ">=2.0.0",
+        "extras": ""
+      },
       {
         "name": "narwhals",
         "version": ">=2.17.0",
@@ -128,6 +386,31 @@
         "version": ">=2.17.0",
         "extras": ""
       },
+      {
+        "name": "duckdb",
+        "version": ">=1.0.0",
+        "extras": ""
+      },
+      {
+        "name": "narwhals",
+        "version": ">=2.17.0",
+        "extras": ""
+      },
+      {
+        "name": "duckdb",
+        "version": ">=1.0.0",
+        "extras": ""
+      },
+      {
+        "name": "narwhals",
+        "version": ">=2.17.0",
+        "extras": ""
+      },
+      {
+        "name": "typer",
+        "version": ">=0.24.1",
+        "extras": ""
+      },
       {
         "name": "narwhals",
         "version": ">=2.17.0",
@@ -143,6 +426,36 @@
         "version": ">=2024.1.0",
         "extras": ""
       },
+      {
+        "name": "anthropic",
+        "version": ">=0.40.0",
+        "extras": ""
+      },
+      {
+        "name": "httpx",
+        "version": ">=0.28.1",
+        "extras": "socks"
+      },
+      {
+        "name": "pyyaml",
+        "version": ">=6.0.0",
+        "extras": ""
+      },
+      {
+        "name": "pydo",
+        "version": ">=0.29.0",
+        "extras": ""
+      },
+      {
+        "name": "pdfminer",
+        "version": ".six>=20221105",
+        "extras": ""
+      },
+      {
+        "name": "resend",
+        "version": ">=2.0.0",
+        "extras": ""
+      },
       {
         "name": "opentelemetry-api",
         "version": ">=1.25.0",
@@ -179,8 +492,8 @@
         "extras": ""
       },
       {
-        "name": "databricks-cli",
-        "version": ">=0.18.0",
+        "name": "coverage",
+        "version": ">=7.13.4",
         "extras": ""
       },
       {
@@ -188,6 +501,11 @@
         "version": ">=0.85.0",
         "extras": ""
       },
+      {
+        "name": "databricks-bundles",
+        "version": ">=0.295.0",
+        "extras": ""
+      },
       {
         "name": "boto3",
         "version": ">=1.35.0",
@@ -226,7 +544,28 @@
     ],
     "project_dev": [],
     "notebook": [],
-    "dockerfile_adhoc": []
+    "dockerfile_adhoc": [
+      {
+        "name": "altair",
+        "version": "",
+        "extras": ""
+      },
+      {
+        "name": "marimo",
+        "version": "==${marimo_version}",
+        "extras": ""
+      },
+      {
+        "name": "numpy",
+        "version": "",
+        "extras": ""
+      },
+      {
+        "name": "pandas",
+        "version": "",
+        "extras": ""
+      }
+    ]
   },
   "npm": [
     {
@@ -288,5 +627,18 @@
       "version": "latest"
     }
   ],
-  "base_images": {}
+  "base_images": {
+    "infra/gitea/Dockerfile": [
+      "docker.io/library/golang:1.25-alpine3.22",
+      "docker.io/library/alpine:3.22"
+    ],
+    "infra/images/selkies/Dockerfile.upstream": [
+      "${DISTRIB_IMAGE}:${DISTRIB_RELEASE}"
+    ],
+    "infra/marimo/src/docker/Dockerfile": [
+      "python:3.13-slim",
+      "base",
+      "data"
+    ]
+  }
 }

dev/scripts/pkg_inventory.py

@@ -164,6 +164,25 @@ def _is_self_ref(spec: str) -> bool:
     return name == "stack"
 
 
+def _extract_dep_specs(block: str) -> list[str]:
+    """Extract dependency spec strings from a TOML array body.
+
+    Pulls the first quoted string from each non-comment line. This
+    correctly skips inline comments (e.g. ``"pymupdf>=1.24",  # AGPL``),
+    trailing commas, and surrounding whitespace — the previous chained
+    ``.strip()`` approach broke on inline comments.
+    """
+    specs: list[str] = []
+    for line in block.splitlines():
+        s = line.strip()
+        if not s or s.startswith("#"):
+            continue
+        m = re.match(r'["\']([^"\']+)["\']', s)
+        if m:
+            specs.append(m.group(1).strip())
+    return specs
+
+
 def _parse_pyproject_deps(text: str) -> tuple[list[dict], list[dict]]:
     """Parse dependencies from pyproject.toml without a TOML library.
 
@@ -181,10 +200,9 @@ def _parse_pyproject_deps(text: str) -> tuple[list[dict], list[dict]]:
     # prod deps — [project] dependencies
     m = re.search(r"^dependencies\s*=\s*\[(.*?)\]", text, re.MULTILINE | re.DOTALL)
     if m:
-        for line in m.group(1).splitlines():
-            line = line.strip().strip(",").strip('"').strip("'")
-            if line and not line.startswith("#") and not _is_self_ref(line):
-                prod.append(_parse_pypi_spec(line))
+        for spec in _extract_dep_specs(m.group(1)):
+            if not _is_self_ref(spec):
+                prod.append(_parse_pypi_spec(spec))
 
     # optional deps — [project.optional-dependencies] all sections
     # Extract the section content, then parse each "name = [...]" block.
@@ -214,18 +232,16 @@ def _parse_pyproject_deps(text: str) -> tuple[list[dict], list[dict]]:
                     depth -= 1
                 i += 1
             block = rest[: i - 1] if depth == 0 else rest
-            for line in block.splitlines():
-                line = line.strip().strip(",").strip('"').strip("'")
-                if line and not line.startswith("#") and not _is_self_ref(line):
-                    prod.append(_parse_pypi_spec(line))
+            for spec in _extract_dep_specs(block):
+                if not _is_self_ref(spec):
+                    prod.append(_parse_pypi_spec(spec))
 
     # dev deps — [dependency-groups] dev
     m = re.search(r"dev\s*=\s*\[(.*?)\]", text, re.MULTILINE | re.DOTALL)
     if m:
-        for line in m.group(1).splitlines():
-            line = line.strip().strip(",").strip('"').strip("'")
-            if line and not line.startswith("#") and not _is_self_ref(line):
-                dev.append(_parse_pypi_spec(line))
+        for spec in _extract_dep_specs(m.group(1)):
+            if not _is_self_ref(spec):
+                dev.append(_parse_pypi_spec(spec))
 
     # build-system requires
     m = re.search(
@@ -234,10 +250,9 @@ def _parse_pyproject_deps(text: str) -> tuple[list[dict], list[dict]]:
         re.MULTILINE | re.DOTALL,
     )
     if m:
-        for line in m.group(1).splitlines():
-            line = line.strip().strip(",").strip('"').strip("'")
-            if line and not line.startswith("#") and not _is_self_ref(line):
-                prod.append(_parse_pypi_spec(line))
+        for spec in _extract_dep_specs(m.group(1)):
+            if not _is_self_ref(spec):
+                prod.append(_parse_pypi_spec(spec))
 
     return prod, dev
 

tests/dev/test_pkg_inventory.py

@@ -0,0 +1,75 @@
+"""Regression tests for dev/scripts/pkg_inventory.py.
+
+Specifically guards the inline-comment bug that caused pkg-supply-chain
+to fail on every push (issue #400-403): the previous parser did
+``line.strip().strip(",").strip('"')`` which left trailing junk when a
+dep line had an inline ``# comment``, producing invalid pip specs like
+``pymupdf>=1.24",`` that broke the downstream PyPI mirror sync.
+"""
+
+from __future__ import annotations
+
+import importlib.util
+import sys
+from pathlib import Path
+
+# Load the script as a module — it lives outside the src/ tree.
+_SCRIPT = Path(__file__).resolve().parents[2] / "dev" / "scripts" / "pkg_inventory.py"
+_spec = importlib.util.spec_from_file_location("_pkg_inventory", _SCRIPT)
+assert _spec and _spec.loader
+_pkg_inventory = importlib.util.module_from_spec(_spec)
+sys.modules["_pkg_inventory"] = _pkg_inventory
+_spec.loader.exec_module(_pkg_inventory)
+
+
+def test_extract_dep_specs_strips_inline_comments():
+    block = """
+    "pymupdf>=1.24",     # AGPL-3.0 — flag if shipping outside internal use
+    "python-docx>=1.1",
+    "ruff>=0.11.0",  # linter
+    "no-comment>=2.0",
+    """
+    specs = _pkg_inventory._extract_dep_specs(block)
+    assert specs == [
+        "pymupdf>=1.24",
+        "python-docx>=1.1",
+        "ruff>=0.11.0",
+        "no-comment>=2.0",
+    ]
+
+
+def test_extract_dep_specs_skips_blank_and_comment_lines():
+    block = """
+
+    # this whole line is a comment
+    "pkg>=1.0",
+       # indented comment
+    """
+    assert _pkg_inventory._extract_dep_specs(block) == ["pkg>=1.0"]
+
+
+def test_extract_dep_specs_handles_single_quotes():
+    block = """
+    'singlequoted>=1.0',
+    "doublequoted>=2.0",
+    """
+    assert _pkg_inventory._extract_dep_specs(block) == [
+        "singlequoted>=1.0",
+        "doublequoted>=2.0",
+    ]
+
+
+def test_parse_pyproject_deps_real_pymupdf_line():
+    """Full integration — parse a small pyproject excerpt with the bug pattern."""
+    text = """[project]
+name = "test"
+dependencies = [
+    "pymupdf>=1.24",     # AGPL-3.0 — flag if shipping outside internal use
+    "python-docx>=1.1",
+]
+"""
+    prod, _dev = _pkg_inventory._parse_pyproject_deps(text)
+    names = sorted(p["name"] for p in prod)
+    assert names == ["pymupdf", "python-docx"]
+    versions = {p["name"]: p["version"] for p in prod}
+    assert versions["pymupdf"] == ">=1.24"  # NOT '>=1.24",'