fix(task-16): custom properties refuse pre-R2004 headlessly, line breaks, and AutoCAD's invalid keys before any write

Three header-only-write / half-write classes in drawing_properties_set:

1. Measured: ezdxf emits $CUSTOMPROPERTYTAG / $CUSTOMPROPERTY only after
   $LASTSAVEDBY, which does not exist before R2004, so on an R12 (AC1009) or
   R2000 (AC1015) document the write reported ok, drawing_properties_get
   echoed it from memory, and drawing_save_as (version kept) wrote nothing --
   the class f3ff703 removed for CANNOSCALE. The headless engine now refuses
   with a ValueError naming the DXF version and "save as R2004 or newer",
   before custom_vars is touched; the dwgprops capability reason no longer
   claims custom_properties_supported unconditionally.

2. A line break in a custom key or value passed validation, ezdxf wrote it
   unescaped, and the saved DXF failed to reopen (DXFStructureError) -- the
   whole drawing lost, reported as success. Refused by name in the shared
   validator.

3. Measured on AutoCAD 2026: AddCustomInfo rejects keys with leading/trailing
   whitespace and keys containing '=', ';' or a no-break space as 'Invalid
   key' -- mid-write, after the summary and earlier keys were applied, as an
   unnamed COM error -- while the headless engine wrote them. The shared
   validator now mirrors those rules, so both engines refuse before writing
   (internal spaces, tabs, unicode and 300-char keys stay accepted, as
   AutoCAD accepts them).

Pinned by save+reopen tests on R12/R2000 (refused, file reads back empty),
R2004 (round-trips), the newline case (refused, the file still opens), and
the fake-SummaryInfo COM test (no ActiveX call reaches SummaryInfo).

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FKFStjEFzpGyLq5bMpWxTd
This commit is contained in:
Umut CAN
2026-09-21 18:45:02 +03:00
co-authored by Claude Opus 5
parent 10a7372ce7
commit 2b9791fbcd
4 changed files with 219 additions and 4 deletions
+59
View File
@@ -515,6 +515,49 @@ def _decode_setting(key: str, kind: str, raw: Any) -> Any:
SUMMARY_FIELDS = ("title", "subject", "author", "keywords", "comments") SUMMARY_FIELDS = ("title", "subject", "author", "keywords", "comments")
#: Characters AutoCAD's ``SummaryInfo.AddCustomInfo`` rejects inside a custom
#: key with ``Invalid key`` (measured on AutoCAD 2026; internal spaces, tabs
#: and other unicode are accepted). ``\xa0`` is the no-break space.
_CUSTOM_KEY_FORBIDDEN = {"=": "'='", ";": "';'", "\xa0": "a no-break space (U+00A0)"}
def _check_custom_text(key: str, text: str, what: str) -> None:
"""A newline in a custom key or value corrupts the DXF on save.
ezdxf writes custom properties verbatim as ``9 / $CUSTOMPROPERTYTAG / 1 /
{key}`` and ``9 / $CUSTOMPROPERTY / 1 / {value}`` header lines — a line
break inside either text starts a new (invalid) group and the reopened
file fails with ``DXFStructureError``. Refused here so neither engine
writes it.
"""
if "\n" in text or "\r" in text:
raise ValueError(
f"custom[{key!r}]: the {what} contains a line break, which corrupts the "
"DXF on save (the whole drawing would fail to reopen, not just the "
"property). Custom properties are single-line."
)
def _check_custom_key(key: str) -> None:
"""Mirror AutoCAD's own custom-key rules so the live engine cannot refuse
mid-write. Measured: ``AddCustomInfo`` raises ``Invalid key`` for a key
with leading/trailing whitespace and for ``=``, ``;`` or a no-break space
anywhere in it — after the summary fields and earlier keys were already
applied — while the headless engine happily writes the same key."""
if key != key.strip():
raise ValueError(
f"custom: key {key!r} has leading or trailing whitespace — AutoCAD's "
"AddCustomInfo rejects it as 'Invalid key'. Strip the key."
)
for char, name in _CUSTOM_KEY_FORBIDDEN.items():
if char in key:
raise ValueError(
f"custom: key {key!r} contains {name} — AutoCAD's AddCustomInfo rejects "
"it as 'Invalid key'. Use another separator."
)
_check_custom_text(key, key, "key")
def validate_drawing_properties( def validate_drawing_properties(
summary: dict | None, custom: dict | None summary: dict | None, custom: dict | None
) -> tuple[dict[str, str], dict[str, str], list[str]]: ) -> tuple[dict[str, str], dict[str, str], list[str]]:
@@ -526,6 +569,20 @@ def validate_drawing_properties(
string nor ``None`` (``None`` deletes) is a ``TypeError``. Values are never string nor ``None`` (``None`` deletes) is a ``TypeError``. Values are never
coerced with ``str()`` — a number the caller meant as text is theirs to coerced with ``str()`` — a number the caller meant as text is theirs to
format. format.
Custom keys and values are also held to the rules the *other* engine would
enforce or the file format would break on, so the same call cannot succeed
on one engine and half-apply on the other (``ValueError`` naming the key):
* a key with leading or trailing whitespace, or containing ``=``, ``;`` or
a no-break space — AutoCAD's ``AddCustomInfo`` rejects these as
``Invalid key`` mid-write (measured on AutoCAD 2026), after the summary
fields and the earlier keys were already applied; the headless engine
writes them without complaint, and AutoCAD then loads them from the DXF;
* a key or value containing a line break (LF or CR) — ezdxf emits custom
properties unescaped, one header line per text, so the saved DXF is
corrupt and the whole drawing fails to reopen (``DXFStructureError``),
not just the property.
""" """
if summary is not None and not isinstance(summary, dict): if summary is not None and not isinstance(summary, dict):
raise TypeError("summary: must be an object of {field: text}") raise TypeError("summary: must be an object of {field: text}")
@@ -545,9 +602,11 @@ def validate_drawing_properties(
for key, value in (custom or {}).items(): for key, value in (custom or {}).items():
if not isinstance(key, str) or not key.strip(): if not isinstance(key, str) or not key.strip():
raise TypeError(f"custom: key {key!r} must be a non-empty string") raise TypeError(f"custom: key {key!r} must be a non-empty string")
_check_custom_key(key)
if value is None: if value is None:
to_delete.append(key) to_delete.append(key)
elif isinstance(value, str): elif isinstance(value, str):
_check_custom_text(key, value, "value")
to_write[key] = value to_write[key] = value
else: else:
raise TypeError( raise TypeError(
+17 -1
View File
@@ -1117,7 +1117,7 @@ class EzdxfBackend(AutoCADBackend):
), ),
"dwgprops": FeatureCapability( "dwgprops": FeatureCapability(
False, False,
reason="summary_fields_need_live_autocad;custom_properties_supported", reason="summary_fields_need_live_autocad;custom_properties_need_r2004_or_newer",
), ),
}, },
) )
@@ -6125,6 +6125,22 @@ class EzdxfBackend(AutoCADBackend):
def _sync(): def _sync():
doc = self._require_doc() doc = self._require_doc()
if (to_write or to_delete) and doc.dxfversion < "AC1018":
# Measured: ezdxf emits $CUSTOMPROPERTYTAG / $CUSTOMPROPERTY only
# after $LASTSAVEDBY, a header variable that does not exist
# before R2004 (its own CustomVars docstring says so). On an
# R12 or R2000 document the in-memory write "succeeded",
# drawing_properties_get echoed it back, and drawing_save_as
# (version kept) wrote nothing -- the header-only write that
# vanishes on save, the class f3ff703 removed for CANNOSCALE.
# Refused here, before custom_vars is touched.
raise ValueError(
f"custom properties: this document is DXF {doc.dxfversion} "
f"({doc.acad_release}), and $CUSTOMPROPERTYTAG / $CUSTOMPROPERTY "
"header pairs are only written for R2004 (AC1018) or newer, so "
f"{sorted(to_write) + sorted(to_delete)} would be lost on save. "
"Save as R2004 or newer first."
)
custom_vars = doc.header.custom_vars custom_vars = doc.header.custom_vars
for key, value in to_write.items(): for key, value in to_write.items():
if custom_vars.has_tag(key): if custom_vars.has_tag(key):
+8 -2
View File
@@ -6962,8 +6962,14 @@ async def drawing_properties_set(
engine (`capability: dwgprops` — SummaryInfo lives in the DWG, a DXF has engine (`capability: dwgprops` — SummaryInfo lives in the DWG, a DXF has
no slot for it; custom properties still work headlessly), a non-string no slot for it; custom properties still work headlessly), a non-string
value (`TypeError` naming the field or key — nothing is coerced with value (`TypeError` naming the field or key — nothing is coerced with
`str()`), an empty custom key. On the live engine a custom key is added `str()`), an empty custom key, a custom key AutoCAD's AddCustomInfo would
with AddCustomInfo when new and changed with SetCustomByKey when present. reject mid-write (leading/trailing whitespace, `=`, `;`, no-break space —
`ValueError` naming the key, on both engines), a line break in a custom
key or value (it corrupts the DXF on save), and headlessly a document
older than R2004 (`ValueError` — ezdxf only writes the custom-property
header pairs for AC1018+, so they would vanish at save; save as R2004 or
newer first). On the live engine a custom key is added with AddCustomInfo
when new and changed with SetCustomByKey when present.
""" """
summary = { summary = {
"title": title, "title": title,
+135 -1
View File
@@ -4,7 +4,15 @@ Headlessly the five summary fields live in the DWG SummaryInfo stream, which
ezdxf cannot write, so any non-null summary field is refused with ezdxf cannot write, so any non-null summary field is refused with
`capability: dwgprops` before anything is touched; custom properties are `capability: dwgprops` before anything is touched; custom properties are
`$CUSTOMPROPERTYTAG` / `$CUSTOMPROPERTY` header pairs and work on both `$CUSTOMPROPERTYTAG` / `$CUSTOMPROPERTY` header pairs and work on both
engines (measured: they survive save + reopen). engines (measured: they survive save + reopen on R2004 and newer; ezdxf
only emits the pairs for AC1018+, so an older document is refused headlessly
instead of losing the write at save).
Custom keys and values are validated once, by name, against what would break
*either* engine: keys AutoCAD's AddCustomInfo rejects mid-write as
`Invalid key` (measured on AutoCAD 2026: leading/trailing whitespace, `=`,
`;`, no-break space) and line breaks, which ezdxf writes unescaped and
which corrupt the whole DXF.
""" """
from __future__ import annotations from __future__ import annotations
@@ -42,6 +50,19 @@ def test_validation_splits_writes_and_deletes():
(None, {"": "x"}, TypeError, "non-empty"), (None, {"": "x"}, TypeError, "non-empty"),
("Gearbox", None, TypeError, "summary"), ("Gearbox", None, TypeError, "summary"),
(None, ["PROJECT"], TypeError, "custom"), (None, ["PROJECT"], TypeError, "custom"),
# Keys AutoCAD's AddCustomInfo rejects as 'Invalid key' (measured on
# AutoCAD 2026) -- refused here so the live engine never fails
# mid-write after the summary and earlier keys were already applied.
(None, {" BAD ": "x"}, ValueError, "' BAD '"),
(None, {"PROJECT ": "x"}, ValueError, "trailing whitespace"),
(None, {"A=B": "x"}, ValueError, "'A=B' contains '='"),
(None, {"A;B": "x"}, ValueError, "'A;B' contains ';'"),
(None, {"NB\xa0SP": "x"}, ValueError, "no-break space"),
# A line break is written unescaped by ezdxf and corrupts the DXF.
(None, {"NOTE": "line1\nline2"}, ValueError, "custom['NOTE']: the value"),
(None, {"NOTE": "line1\rline2"}, ValueError, "line break"),
(None, {"K\nEY": "x"}, ValueError, "the key contains a line break"),
(None, {"K\nEY": None}, ValueError, "the key contains a line break"),
], ],
) )
def test_validation_refuses_by_name(summary, custom, exc, fragment): def test_validation_refuses_by_name(summary, custom, exc, fragment):
@@ -50,6 +71,13 @@ def test_validation_refuses_by_name(summary, custom, exc, fragment):
assert fragment in str(excinfo.value) assert fragment in str(excinfo.value)
@pytest.mark.parametrize("key", ["IN SP", "T\tAB", "\u00dcn\u00efcode", "X" * 300])
def test_validation_accepts_keys_autocad_accepts(key):
"""Measured: internal spaces, tabs, unicode and 300-char keys pass
AddCustomInfo; the validator must not be stricter than AutoCAD."""
assert validate_drawing_properties(None, {key: "v"}) == ({}, {key: "v"}, [])
# ── headless engine ───────────────────────────────────────────────────────── # ── headless engine ─────────────────────────────────────────────────────────
@@ -85,6 +113,96 @@ async def test_custom_properties_write_replace_delete_and_survive_reopen(backend
assert (await backend.drawing_properties_get())["custom"] == {"PROJECT": "X-2"} assert (await backend.drawing_properties_get())["custom"] == {"PROJECT": "X-2"}
async def test_line_break_is_refused_before_any_write_and_the_file_still_opens(backend, tmp_path):
"""Measured: `custom={'NOTE': 'line1\\nline2', 'AFTER': 'ok'}` reported
`ok: True`, and the saved DXF failed to reopen with
`DXFStructureError: Invalid group code "line2"` -- the whole drawing lost,
reported as success. The request is refused as a whole, so AFTER is not
written either, and the file saved afterwards reopens."""
with pytest.raises(ValueError, match=r"custom\['NOTE'\]: the value contains a line break"):
await backend.drawing_properties_set(custom={"NOTE": "line1\nline2", "AFTER": "ok"})
assert (await backend.drawing_properties_get())["custom"] == {}
path = str(tmp_path / "clean.dxf")
await backend.drawing_save_as(path)
await backend.drawing_open(path)
assert (await backend.drawing_properties_get())["custom"] == {}
async def test_autocad_invalid_keys_are_refused_headlessly_too(backend):
"""The headless engine used to write ' BAD ', 'A=B' and 'A;B' without
complaint (and AutoCAD loaded them from the DXF) while the live engine
half-applied the same call; both now refuse before writing."""
with pytest.raises(ValueError, match="' BAD '"):
await backend.drawing_properties_set(custom={"AAA": "first", " BAD ": "x", "ZZZ": "n"})
assert (await backend.drawing_properties_get())["custom"] == {}
async def _backend_for_version(tmp_path, version: str):
import ezdxf
from backends.ezdxf_backend import EzdxfBackend
path = tmp_path / f"{version}.dxf"
ezdxf.new(version).saveas(str(path))
backend = EzdxfBackend()
await backend.connect()
await backend.drawing_open(str(path))
return backend
@pytest.mark.parametrize("version, dxfversion", [("R12", "AC1009"), ("R2000", "AC1015")])
async def test_custom_properties_before_r2004_are_refused_rather_than_lost(
tmp_path, version, dxfversion
):
"""Measured: on an R12 or R2000 document `drawing_properties_set` reported
`ok: True, custom_written: ['PROJECT']`, `drawing_properties_get` echoed
it from memory, and `drawing_save_as` (version kept) wrote nothing --
ezdxf emits $CUSTOMPROPERTYTAG / $CUSTOMPROPERTY only after $LASTSAVEDBY,
which does not exist before R2004. The header-only write that vanishes on
save, the class f3ff703 removed for CANNOSCALE. Refused up front, before
custom_vars is touched; the reopened file is read back to prove it."""
backend = await _backend_for_version(tmp_path, version)
try:
assert backend._doc.dxfversion == dxfversion
with pytest.raises(ValueError, match=dxfversion) as excinfo:
await backend.drawing_properties_set(custom={"PROJECT": "X-1", "REV": "B"})
assert "R2004 or newer" in str(excinfo.value)
assert "PROJECT" in str(excinfo.value)
assert list(backend._doc.header.custom_vars) == []
assert (await backend.drawing_properties_get())["custom"] == {}
# A delete on the same document is refused for the same reason: it
# would report a key removed that the file could never have held.
with pytest.raises(ValueError, match="R2004 or newer"):
await backend.drawing_properties_set(custom={"PROJECT": None})
# An empty request touches nothing and is not a write to refuse.
res = await backend.drawing_properties_set(custom={})
assert res["ok"] is True and res["custom_written"] == []
out = tmp_path / f"{version}_out.dxf"
await backend.drawing_save_as(str(out))
await backend.drawing_open(str(out))
assert backend._doc.dxfversion == dxfversion
assert (await backend.drawing_properties_get())["custom"] == {}
finally:
await backend.disconnect()
async def test_custom_properties_on_r2004_survive_save_and_reopen(tmp_path):
"""The oldest version that carries the pairs -- the boundary the refusal
sits on -- round-trips."""
backend = await _backend_for_version(tmp_path, "R2004")
try:
res = await backend.drawing_properties_set(custom={"PROJECT": "X-1"})
assert res["custom_written"] == ["PROJECT"]
out = tmp_path / "r2004_out.dxf"
await backend.drawing_save_as(str(out))
await backend.drawing_open(str(out))
assert backend._doc.dxfversion == "AC1018"
assert (await backend.drawing_properties_get())["custom"] == {"PROJECT": "X-1"}
finally:
await backend.disconnect()
async def test_summary_is_refused_headlessly_before_any_write(backend): async def test_summary_is_refused_headlessly_before_any_write(backend):
with pytest.raises(UnsupportedCapabilityError) as excinfo: with pytest.raises(UnsupportedCapabilityError) as excinfo:
await backend.drawing_properties_set(summary={"title": "T"}, custom={"LATE": "no"}) await backend.drawing_properties_set(summary={"title": "T"}, custom={"LATE": "no"})
@@ -231,3 +349,19 @@ async def test_com_validates_before_touching_activex(com_backend):
with pytest.raises(ValueError): with pytest.raises(ValueError):
await backend.drawing_properties_set(summary={"edition": "1"}) await backend.drawing_properties_set(summary={"edition": "1"})
assert info.calls == [] and info.Title == "Old title" assert info.calls == [] and info.Title == "Old title"
async def test_com_refuses_autocad_invalid_keys_before_any_write(com_backend):
"""Measured on AutoCAD 2026: `summary={'subject': 'PARTIAL?'},
custom={'AAA': 'first', ' BAD ': 'x', 'ZZZ': 'never'}` raised an unnamed
COM error from AddCustomInfo(' BAD ') and left subject and AAA written,
ZZZ absent -- a half-applied write reported as an error. The shared
validator now names the key and nothing reaches SummaryInfo."""
backend, info = com_backend
with pytest.raises(ValueError, match="' BAD '"):
await backend.drawing_properties_set(
summary={"subject": "PARTIAL?"},
custom={"AAA": "first", " BAD ": "x", "ZZZ": "never"},
)
assert info.calls == [] and info.Subject == ""
assert info._custom == [("PROJECT", "X-1"), ("OLD", "1")]