feat(canary): operator-upload instrumenters + tests

Seven instrumenters that mutate operator-supplied artifacts to
embed the callback URL:

- passthrough — bytes unchanged; only DNS-callback tokens trip
  detection, with the slug embedded in the placement path
- plain      — substitutes {{CANARY_URL}}/{{CANARY_HOST}} placeholders;
  falls back to appending a comment line whose prefix adapts to the
  apparent file syntax (#, //, ;)
- html       — injects a 1x1 tracking pixel before </body>, appends
  if the close tag is missing
- docx       — direct zipfile manipulation (no python-docx dep):
  inserts an external-image Relationship into word/_rels/document.xml.rels
  and a matching <w:drawing> element before </w:body>
- xlsx       — sibling of docx; injects an external-image relationship
  into xl/_rels/workbook.xml.rels (orphan rels are still fetched on
  open by most viewers)
- pdf        — uses pikepdf to install /OpenAction /URI on the catalog;
  rejects with a clear message when pikepdf isn't installed
- image      — uses Pillow to embed slug + URL in PNG tEXt / JPEG
  comment; rejects with a clear message when Pillow isn't installed

DOCX and XLSX share the rId allocator + relationship injector via
the docx module; both work on stdlib zipfile only.

Tests synthesise minimal real DOCX/XLSX fixtures inline, round-trip
each instrumenter, and assert the callback URL ends up in the
mutated bytes while the file still parses.
This commit is contained in:
2026-04-27 13:03:42 -04:00
parent c7658ea65e
commit 19ceff4417
10 changed files with 819 additions and 0 deletions

View File

@@ -0,0 +1,95 @@
"""XLSX instrumenter — embed an external-image link.
XLSX is structurally identical to DOCX (Office Open XML zip). The
injection target is the workbook's relationships file
(``xl/_rels/workbook.xml.rels``). We add an external image
relationship there; Excel/LibreOffice fetch external images on
workbook open in the same way Word does.
We don't inject a ``<drawing>`` element into a sheet because that
requires touching ``xl/worksheets/sheetN.xml`` *and* allocating a new
``xl/drawings/drawingN.xml`` part — much higher chance of mangling
the file. An orphan external image relationship is enough: many
Office viewers fetch all relationships at open time regardless of
whether they're referenced from a sheet.
If the operator wants a stronger trigger (image visible in the
sheet, fetched even by viewers that lazy-load external resources)
they should embed the slug as a hyperlink cell content via the
``plain``/``passthrough`` instrumenters.
"""
from __future__ import annotations
import io
import zipfile
from typing import Tuple
from decnet.canary.base import (
CanaryArtifact,
CanaryContext,
CanaryInstrumenter,
InstrumenterRejectedError,
)
from decnet.canary.instrumenters.docx import _inject_relationship, _next_rid
_RELS_PATHS = (
"xl/_rels/workbook.xml.rels",
"xl/_rels/sharedStrings.xml.rels",
)
def _mutate(blob: bytes, url: str) -> Tuple[bytes, str, str]:
try:
with zipfile.ZipFile(io.BytesIO(blob), "r") as zf:
members = [(zi, zf.read(zi.filename)) for zi in zf.infolist()]
except zipfile.BadZipFile as e:
raise InstrumenterRejectedError("uploaded blob is not a valid XLSX zip") from e
target_rels: str | None = None
for zi, _ in members:
if zi.filename in _RELS_PATHS:
target_rels = zi.filename
break
if not target_rels:
raise InstrumenterRejectedError(
"XLSX has no workbook relationships file to mutate"
)
out_members = []
rid = ""
for zi, data in members:
if zi.filename == target_rels:
rid = _next_rid(data)
data = _inject_relationship(data, rid, url)
out_members.append((zi, data))
out = io.BytesIO()
with zipfile.ZipFile(out, "w", zipfile.ZIP_DEFLATED) as zf_out:
for zi, data in out_members:
zf_out.writestr(zi, data)
return out.getvalue(), rid, target_rels
class XlsxInstrumenter(CanaryInstrumenter):
name = "xlsx"
mime_prefixes = (
"application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
)
def instrument(
self, blob: bytes, ctx: CanaryContext, *, target_path: str,
) -> CanaryArtifact:
url = f"{ctx.http_base.rstrip('/')}/c/{ctx.callback_token}"
mutated, rid, target_rels = _mutate(blob, url)
return CanaryArtifact(
path=target_path,
content=mutated,
mode=0o644,
mtime_offset=-86400 * 14,
instrumenter=self.name,
notes=[
f"injected external-image relationship {rid} into "
f"{target_rels} -> {url}",
],
)