fix(anatomy): V1-2.2 review LOWs, owner export procedure, evidence

Composer check counts url( openings apart from the reader; tab and
newline character references tested; RUST.md sizes and counts of the
final build; owner procedure for Illustrator, Figma and Affinity.
This commit is contained in:
bchanot committed 2026-10-29 17:15:02 +01:00
1 parent a678faaa23
commit f1ef6a466f
9 files changed
+300 -55

No files matched your search

@@ -191,3 +191,193 @@ quoted CSS string (URL Standard: tab and newline are removed from the input).
status` there shows only this file and the owner's `exports/owner/README.md`.
- The verdict runs in the core, client or worker side; the SQL side cannot check the stored
form. That trust boundary is outside this diff.
## 5. Delta review (5281337..HEAD, commit a678faa)
Fresh scan of the delta, mode audit, 2026-10-29. The report above was read first, as the brief
required; nothing else from the earlier round was used.
```
SECURITY — VERDICT: PASS
TOOL: semgrep 1.168.0 present but NOT RUN (p/security-audit, p/secrets, p/owasp-top-ten are registry packs fetched at run time and the brief forbids network) — effectively DEGRADED: checklist + guardrails scanner + manual attack review + the two named cargo suites + a model fuzz (section 5.7)
SCOPE: 13 files (5 Rust: compose.rs, sanitize.rs, ops.rs, svg_clean.rs, svg_scope.rs; 8 docs, fixtures and audit files)
BLOCKING:
(none)
REPORTED (non-blocking):
- [MEDIUM] scope-key-is-view-id-only (carried from 5281337; now a written requirement, still not enforced by code or test) — crates/core/src/anatomy/compose.rs:99 — two documents inlined in one page share `fv.main.`, first in tree order wins; recorded in contract §10 and RUST.md §12.7; no screen inlines two documents today (consumers: apps/app/src/core/sheetOps.ts, services/sheetIndex.ts, anatomy/drawing.ts)
- [LOW] composer-check-shares-the-reader-parser — crates/core/src/anatomy/compose.rs:105 — `references_scoped` calls `urls_are_local` and `url_targets`, the same `url_openings` / `local_url_target` as the sanitiser: it catches a `scope_ids` miss, it cannot see a future reader/browser divergence in those two functions; none today (5.7)
- [LOW] tests-deliver-a-space-not-a-tab — crates/core/tests/svg_clean.rs:356, :357, :381; crates/core/tests/svg_scope.rs:338 — a raw TAB or LF inside an XML attribute is normalised to a space by the XML parser (roxmltree parse.rs:1364), so Chromium's `	` spelling of the first HIGH reaches `local_url_target` in no test; hint: rows written `	` and `
`, or a value-level unit test next to sanitize.rs:710
- [LOW] ts-oracle-plain-spelling-only (carried) — apps/app/tests/helpers/svgTree.ts:112 — still `replaceAll('url(#'`; it mirrors the core's rewrite for the product drawing, and the sanitiser now refuses every other spelling, so the blind spot has no reachable input
- [LOW] mutation-script-rewrites-sources (carried) — .claude/audits/2026-10-29-v1-2-2-mutations-arbitrations.py — five mutants added (A13 to A17); it still edits sources in place and restores in `finally`; not run here, so their kill results are not verified by this gate
PROOF: semgrep 0 rules on 0 files (not run, no network) → 0 findings; checklist 9 checks (secrets, SQL concatenation, unsanitised render, authz, error surfaces, logging, tracked .env, guardrails scanner, new suppressions) → 0 findings in the delta; cargo test svg_clean + svg_scope 21 passed, lib anatomy::svg 24 passed; model fuzz 8.4M strings in fill and style contexts + 2,400-case slot sweep + control run → 0 dangerous mismatches; manual attack review of 5 code files → 0 HIGH, 1 MEDIUM, 4 LOW
```
REPORT_WRITTEN: .claude/audits/2026-10-29-v1-2-2-arbitrations-security.md
### 5.1 Status of the previous findings
| Previous finding | Now |
|---|---|
| HIGH url-reference-spelling-mismatch | Closed. One reading of `url(` (`local_url_target`, sanitize.rs:487) feeds `urls_are_local`, `url_targets`, `scope_url_targets` and `calls_only_local_url`, so the set the sanitiser admits, the set the dangling check and the budget count, and the set the composer rewrites are the same by construction. Fix directions 1, 2, 3 and 4 of section 1 are all done. |
| MEDIUM scope key | Recorded as a requirement (contract §10, RUST.md §12.7). Open as a rule, see the reported list. |
| LOW colour function not whole value | Closed: `colour_function_end` now requires end of text, CSS space, `;`, `,` or `!` after the `)`. `rgb(1,2,3)rgb(4,5,6)` is removed, `url(#a rgb(1,2,3))` is refused at the `url(`. |
| LOW test oracle shares the blind spot | Closed in the core: `browser_references` (svg_scope.rs) reads `href="…"` and every `url(…)` on its own, trims quotes and CSS space, strips tab and newline, percent-decodes. The TS helper is carried, see above. |
| LOW lint allow | Nothing new: no added `allow`, `nosemgrep`, `nosec`, `eslint-disable` or `ts-ignore` line in the delta (grep on the added lines). |
| LOW mutation script | Carried. |
### 5.2 What the delta reads like
`local_url_target` accepts only: CSS space, an optional quote, `#`, a run of `[A-Za-z0-9_.:-]`
that satisfies `is_reference_name` (first character `[A-Za-z_]`), the same quote, CSS space,
`)`. `url_openings` skips a `url(` glued to a longer name (`continues_identifier`: ASCII
alphanumeric, `_`, `-`, `\`, anything above U+007F). Every `urls_are_local` false is an active
removal, so the file is refused, never cleaned. After a strict `url(...)` is cut out and the
strict colour functions are removed, `calls_only_local_url` keeps a value only if no `(` is
left. In the composer, `references_scoped` runs after `scope_ids` and before the layer is
added, and `ops.rs:480` maps its failure to `svg_rejected` / `unscoped_reference`.
### 5.3 Re-attack, item by item
Everything below was thrown at the strict reader and compared with a browser model (5.7).
- Percent escapes, quoted and unquoted: `url(#%61)`, `url('#%61')`, `url(#a%61)`, `url(#a.%2e)`,
`%23`, `%00`, `%20`. The id run stops at `%`, which is not followed by a quote, a space or
`)`, so the value is not local and the file is refused. Same for `href="#%61"` and
`xlink:href` (strict `#` + name).
- Every C0 control and Unicode space, inside and around the quotes: an exhaustive sweep of
8 positions (before `url(`, after `url(`, after the quote, between `#` and the id, inside the
id, at the id end, after the closing quote, before `)`) x 3 quote modes x 100 characters
(U+0001 to U+0020, U+007F, U+0080 to U+00A0, U+1680, U+180E, U+2000 to U+200F, U+2028,
U+2029, U+202F, U+205F, U+3000, U+FEFF, U+FFFD, `/**/`, `/*x*/`, `%`-forms, fullwidth `#`
and `'`) = 2,400 values. 163 are accepted; every one has only CSS space (space, tab, LF,
CR, FF) at the four legal slots, or a character before `url(` that is not part of the
reference; 0 differ from the browser model. A NBSP or U+3000 after the quote or before the
closing quote is refused, which is the Chromium/Firefox divergence of the first round.
Controls other than tab, LF and CR cannot be carried by XML 1.0 at all.
- CSS comments inside `url(`: `url(/**/#a)` is, in a browser, an unquoted URL whose text is
`/**/#a` (a relative fetch); the reader refuses it (`/` is neither a quote nor `#`). `url(#a/**/)`,
`url('#a'/**/)`, `url(/**/'#a')` are refused. `/**/url(#%61)` has a non-name character in
front and is read as an opening, then refused on the `%`.
- Escapes: a backslash still refuses the value for every attribute, before anything else runs
(`value_acceptable`, `value_is_active`); unchanged.
- `URL(`, `Url(`: ASCII-lowercase scan with identical byte offsets; no non-ASCII character
folds to `url`.
- Nested or repeated: `url(url(#a))`, `url(#a)url(#a)`, `url(#a) url(x)`: every opening must
be strict. The 25,000-reference `%61` flood is refused as active content
(`svg_clean.rs:346` test); the plain flood is `too_complex` as before.
- `url(` inside a colour: `rgb(1,url(#a),3)` and `url(#a rgb(1,2,3))` leave a `(` or fail the
strict window and are refused.
- `)` inside quotes: `url(')')`, `url('#a)')` are refused (id then wrong closer).
- `xlink:href` against `href`: both need `#` + a reference name after an ASCII trim (space, tab,
LF, CR, FF); NBSP does not trim; `#a\tb`, `#a%61`, `# a` are refused. The composer demands
`#fv.<view>.` on both. A prefixed `x:href` is not on the allowlist; `xlink:href` must sit in
the xlink namespace.
- Ids with `.` or `:`: `url(#a.b:c-d_e)` is accepted, scoped as `fv.main.a.b:c-d_e`, and a browser
reads the same string. `svgView(` and `xpointer(` need `(`, which the id grammar excludes.
`url(#-a)`, `url(#1a)`, `url(#é)` are refused.
- Glued names with non-ASCII letters: `érgb(1,2,3)`, `éurl(#%61)`, ` url(`. A browser reads
every code point above U+007F as part of the name, so these are single longer function names;
the reader agrees and leaves the `(` in place, which refuses the value on a presentation
attribute. A digit in front (`1url(`) is a dimension plus a block in a browser, never a fetch;
the reader treats it as glued as well. A non-name character in front (`%`, `+`, `,`, `"`, `#`,
`@`) keeps the `url(` an opening, so it must be strict.
- A `(` surviving in a kept value: none. After the strict `url(...)` cut and the strict
`rgb(`/`rgba(` removal, any remaining `(` refuses the value. In 2.5 million mutated accepted
values (fill and style), the browser model saw only `url` tokens equal to a reader target
and no function other than `rgb` or `rgba`.
### 5.4 The composer check
`references_scoped` (compose.rs:105) walks the whole tree once, root included, and demands
`#fv.<view>.` on every `href` and `xlink:href`, and for every other attribute whose
lowercased value contains `url(`: `urls_are_local` and every target starting with the scope.
Anything a sanitised drawing can still carry as a reference is one of those: the presentation
attributes and `style` are the only CSS-parsed attributes on the allowlist that take a `<url>`
(`fill`, `stroke`, `filter`, `mask`, `clip-path`), `d`, `transform` and the rest are not read
as CSS fetches, and no other attribute names an element. The layer is added afterwards and
carries no `id`, `href` or `url(`: its colours are validated as hex at deserialisation
(schema.rs:89, `js::is_hex_color`). It cannot be bypassed by a value the reader accepts. Its
limit is that it shares the parser: it is a guard on `scope_ids`, not a second opinion on the
reader (reported, LOW).
### 5.5 Cost
No new quadratic walk. `references_scoped` is one pass with one lowercase copy per attribute
value. `url_openings` searches forward from the last match; each opening's space and id runs
end before the next opening's `(`, so runs do not overlap; `colour_function_end` adds one byte
test. Slice indexes sit on ASCII boundaries (`#`, `)`, `u`, `r`), so no panic path appeared.
Measured on a transliteration (5.7): 11 adversarial families (`url(#a) ` x N, `xurl(` x N,
`url(` plus N spaces, `url(#aaa…`, `url(#a` x N overlapping, `rgb(1,2,3) ` x N, `rgb(1,url(#a),3)`
x N, and others) doubled from 50k to 200k units: every ratio between 1.85 and 2.13. A first run
showed a 4.5x ratio on the `rgb(1,2,3) ` family; that was the Python slice copy `text[at:]`, which
the Rust `&text[at..]` does not make, and an offset-based port is linear.
### 5.6 Reported findings, detail
- MEDIUM, scope key (carried). The requirement is written in two places and nothing enforces it.
Today it has no victim: the only users of `view.canvas` in the app are the editor canvas and
its detail modal, which draw views of one document, with different view ids and therefore
different scopes. It becomes a defect on the first page that inlines two documents or two
versions at once. Keep it as an entry criterion of that page (a per-handle token in
`id_scope`, or a test that mounts two canvases with the same view id and fails).
- LOW, composer shares the parser. A coarse second opinion would not share the code: after
scoping, count the case-insensitive `url(` substrings of each value and require the count to
equal `url_targets(...).len()`. It would also fail closed on a glued `xurl(` kept on a
non-presentation attribute, which is harmless but unwanted in a drawing.
- LOW, tests deliver a space. roxmltree turns a raw TAB, LF or CR in attribute text into a space
(parse.rs:1364), so the rows `url('#\ta')`, `url('#a\tb')`, `url(\n#a)` and the hostile loop's
`url('#\tg')` test a space. A character reference (`&#9;`) stays a tab per XML 1.0 §3.3.3; I
could not read that branch of the parser (see 5.8). The reader is right on a tab (the model
and the slot sweep refuse it), but a regression that skipped only tabs would pass the suite.
- LOW, carried: TS oracle and mutation script, see the block.
- Not a finding, product note: a reference to an id the grammar rejects (`url(#1a)`, `url(#é)`,
for example an Illustrator export that keeps accented layer names as ids) now refuses the whole
file with `svg_active_content`, where it used to pass unread. That is the owner's strict
rule (§9.2), but the label tells the admin "active content" for what is only an unsupported id.
### 5.7 Method and what the probes are
- Run for real: `cargo test -p faunosteo-core --test svg_clean --test svg_scope` (21 passed,
0.73 s) and `cargo test -p faunosteo-core --lib anatomy::svg` (24 passed), in the worktree.
- The built wasm module cannot serve this round: `packages/core-wasm/pkg` is dated 16:28 and the
delta's sources 16:46 and 16:48, so it still holds the first-round code. No build was started.
- The fuzz and the sweeps ran on a Python transliteration of `url_openings`,
`local_url_target`, `url_targets`, `scope_url_targets`, `urls_are_local`,
`calls_only_local_url`, `colour_function_end`, `without_colour_functions` and the value rules
of `value_acceptable`, typed in from sanitize.rs at a678faa, fed on stdin (no file written).
Before use it was checked against the repo's own assertions (the 14 refused and 4 accepted
`url` rows of svg_clean, the 5 glued or loose colour rows, the colour accepted and refused
lists, the scoping unit test): all passed. It is not the compiled code.
- The oracle is a second, independent piece: a CSS Syntax Level 3 tokenizer (comments, strings,
numbers and dimensions, name code points above U+007F, `url(` unquoted token versus function,
bad-url) plus the URL Standard reading of a reference (strip C0 and space at the ends, remove
tab and newline, percent-decode the fragment). The property checked, per accepted value in a
`fill` and in a `style` context: every URL the browser model resolves is exactly `#id` with
the id one of the reader's targets, nothing external, no function other than `rgb` or `rgba`.
After scoping, the same holds with the `fv.main.` prefix and `references_scoped` passes.
- Volume: 400,000 grammar-generated strings, 3,000,000 and 2 x 2,500,000 mutants of 10 to 11
valid seeds (inserting, deleting or replacing from a 60-fragment vocabulary that includes
`%61`, tabs, controls, Unicode spaces, comments, quotes, `url(`, `rgb(`, backslash), each in
both contexts: 0 dangerous mismatches. About 543,000 accepted values carried a browser URL.
10,582 accepted values had a reference inside a comment or a string that the reader counts and
the browser ignores: the safe direction (over-counting for the dangling check and the budget).
- Control: the same harness on the pre-delta rules (`skip_url_prefix` plus longest id run)
reported 52,752 divergent values out of 177,587 accepted (`url(#éa)`, `url(#a%'6)`, `url(#g'-h%2e)`),
so the harness does detect the class the first round found.
- Not run: any browser. The first round's Chromium and Firefox results are the only browser
evidence; this round's browser behaviour comes from the specifications as modelled. WebKit was
never probed. semgrep did not run (registry packs, no network). The guardrails scanner ran:
0 FAIL, 28 WARN, none in the delta's files.
### 5.8 Commands refused
- A scratch cargo crate (path dependency on the worktree's core, own target directory, to run
the real functions on the new spellings) was refused by the permission rules: "Permission to
use Bash with command … has been denied", no rule named. My contract also forbids writing a
code file outside this report, so I did not retry it by another tool or route; the Python
transliteration above is the substitute and is weaker than running the Rust.
- A `sed` on `parse.rs` of roxmltree in the cargo registry after a `cd` with a glob was refused
with the deny rule `**/secrets/**`; not retried. The raw-whitespace normalisation at
parse.rs:1364 had been read earlier through a `grep`; the character-reference branch was not.
- Nothing else was written: `git status` in the worktree showed a clean tree before this
section, and this file is the only one touched.
@@ -79,3 +79,38 @@ Containers after the gates (`docker inspect`, read-only): identical to the snaps
`supabase_*_faunosteo` and the V1-2.1 sandbox: same start times and restart counts).
Known: commit `64a4384` alone does not build (`facade::clean_svg` names the module of `cf37aa2`); the branch head does.
## 7. Owner arbitrations at the checkpoint (brief « V1-2.2 : arbitrages du propriétaire et finalisation », on 17d9199)
Owner answers (2026-10-29, contract §17.4 last six rows): references refused when dangling and canvas ids scoped per
view, implemented in V1-2.2 (not deferred); one stored form for every upload, verdict and preview refusing another;
strict numeric colours kept; `remove` refused atomically while named; Inkscape fetched for real exports; the wasm debug
`name` section dropped. Commits: `e823655` (SVG), `bb5efa0` (removal), `661212a` (build flag), `93e4e81` (docs),
`5281337` (contract criteria B1-B8, mutation script), `a678faa` (strict `url(` after the security review), then the
review LOWs and this evidence.
| Gate | Result |
|---|---|
| `pnpm check:all` on `5281337` | exit 0; cargo 1 949 passed, 0 failed, 3 ignored; core-wasm 65, supabase 128, app 566; e2e:ui 26 passed; app image, worker test stage, worker runtime image |
| `pnpm check:all` on `a678faa` | exit 0; cargo 1 959 passed, 0 failed, 3 ignored; same suites green; module 1 512 071 B raw, 558 793 B gzip -9, 568 625 B by Node's gzip (budget 600 000, margin 31 375 B) |
| first run on the working tree | stopped at the committed-types guard only (`DrawingResult.ts`, `SvgCleaning.ts` regenerated, not yet committed); every stage before it green |
| mutations (`2026-10-29-v1-2-2-mutations-arbitrations.py`) | A1-A12 on 5281337: 12 killed (A11 first written so it did not compile, rewritten, killed); A13-A17 after a678faa: 4 killed, A17 (`%` taken into the id run) survives as an equivalent mutant, `is_reference_name` refusing `%` either way (A14 kills the grammar check itself) |
| protected readers, N-1 copies, frozen oracles, goldens, `supabase/` | no file changed (`git diff --name-only 17d9199`) |
| DEV and sandbox containers | `docker inspect` identical before and after every gate (start times, restart counts) |
Reviews (fresh; the pinned Fable 5.1 had no credits left and the Opus alias answered `model_not_found`, so both ran on
Sonnet):
| Round | Verifier | Security |
|---|---|---|
| `17d9199..5281337` | ECARTS(2): B4 LOW (`érgb(`, `rgurl(#b(` passed the colour exception), A6 open; doc drift | BLOCK(1) HIGH: `url(#%61)` (and Chromium's `url('#<TAB>a')`) passed the dangling check, the scoping and the expansion budget, resolving in Chromium 1248 and Firefox 1555 (budget hole pre-existing since R3); MEDIUM scope unique per document only; 3 LOW |
| `5281337..a678faa` | ECARTS(2): A6 open; B10 LOW (stale size figures, one wording in ANNOTATION-RENDERING); owner procedure not precise enough | PASS; MEDIUM carried as a written requirement; LOWs: composer check shares the reader's parser, no test delivers a real tab, TS helper oracle, mutation script |
| after | the LOWs fixed in the last commit: sizes, wording and counts in RUST.md; a coarse `url(` count in the composer check that shares no parser; `&#9;`, `&#10;`, `&#13;` spellings in `svg_clean`; the owner procedure rewritten (same source file, live-text variants, `--nocapture`, date in the name, Affinity) | — |
Inkscape provenance: not installed; inkscape.org 504 then timeout; `apt-get download` of `inkscape_1.4.3-0ubuntu2_amd64.deb`
(sha256 `24a2e9225c2cdcaad8f3e36a8b7a0f5a68db58ed67fc6f67455395fdd918326e`, equal to `apt-cache show`) and 22 library
packages into the session scratch folder, `dpkg-deb -x`, run with `LD_LIBRARY_PATH` (no install, no sudo).
Open at the checkpoint: A6 (Illustrator, Figma, Affinity exports from the owner); the MEDIUM (a page inlining two
documents needs a per-document prefix); the probe folder of the first security review, whose deletion the permission
rules refused (`scratchpad/probe-audit` of session 0cee2841), to be removed by the owner.
+1 -1
View File
@@ -579,7 +579,7 @@ Order (BDR-071): P0 docs + decisions → V1-1 clients and animals → V1-2 sheet
- [x] I4 edit commands, owner go 2026-10-29 (option « refus atomique »): `remove` refused `has_dependents` when any dependent is not a descendant (view: drawn annotations, child views, links opening it; annotation category; region; finding zone: linked annotations; reference: citing findings); finding descendants still go with their parent; contract §8.1 amended
- [x] I5 A6 inventory: Inkscape / Illustrator / Figma (version, drawing, steps, file name, folder, tests, status); Inkscape not installed → owner go 2026-10-29 to download it; inkscape.org answered 504 / timed out, so the Ubuntu archive package 1.4.3 (sha256 = signed index) + 22 libs fetched with `apt-get download`, unpacked in the scratchpad (no install, no sudo); 3 real exports + generator committed; Illustrator / Figma owner procedure; A6 stays open
- [x] I6 wasm size: contributors measured, optimisation options for later steps; cap and test kept; owner go 2026-10-29: `--remove-name-section` in scripts/wasm-build.sh (debug `name` section, 279 KB raw, −30 691 B gzip measured)
- [ ] I7 gates: rust:check, SVG security suites, property + mutation, wasm + ts-rs, size, check:all; fresh verifier + security; evidence, memory, commit + push
- [x] I7 gates: check:all green on 5281337 and a678faa (final rerun on the last commit); mutations 16 killed + 1 equivalent; fresh reviews (Sonnet: Fable out of credits, Opus alias 404): security BLOCK(1) HIGH url(#%61) → fixed a678faa → PASS; verifier ECARTS → LOWs fixed, A6 open; evidence §7; commit + push. Open: A6 owner exports, MEDIUM per-document prefix before any two-document page
- [ ] V1-2.3 drawings and worker verification (M6) · V1-2.4 wireframes (editor, review, admin entry) · V1-2.5 catalogue, admin space, rights, species, MFA · V1-2.6 editor level 1 · V1-2.7 preview, review, publication UI · V1-2.8 equine working copy and review preparation · V1-2.9 FR/EN acceptance and closure (owner checkpoints V1-2.3, V1-2.4, V1-2.6, V1-2.7, V1-2.8, V1-2.9 + each DEV operation of contract §16.11)
## Release readiness track (owner 2026-10-25, distinct from the functional roadmap; nothing here authorises a release or exposure)
+5 -3
View File
@@ -108,10 +108,12 @@ fn references_scoped(element: &Element, scope: &str) -> bool {
.strip_prefix('#')
.is_some_and(|id| id.starts_with(scope)),
_ if attr.value.to_ascii_lowercase().contains("url(") => {
let targets = url_targets(&attr.value);
// A coarse count that shares no parser with the reader: every `url(` is a target.
let openings = attr.value.to_ascii_lowercase().matches("url(").count();
urls_are_local(&attr.value)
&& url_targets(&attr.value)
.iter()
.all(|id| id.starts_with(scope))
&& openings == targets.len()
&& targets.iter().all(|id| id.starts_with(scope))
}
_ => true,
});
+6 -6
View File
@@ -26,9 +26,9 @@ sudo); inkscape.org, where the AppImage lives, answered 504 and timed out that d
## Editor exports, supplied by the owner (`owner/`)
Illustrator, Figma and Affinity are not available on the development machine (Inkscape is
covered above by a real 1.4.3 export); the owner exports a small drawing from each and puts
the files in `owner/`, named `<tool>-<version>-<what>.svg` (for example
`illustrator-29.3-export-as.svg`, `figma-125.8-export.svg`). The suite reads every file
there: each must be refused (with a closed reason) or cleaned into a drawing that reads back
with no removal, and a file whose name contains `-active-` must be refused for active content.
Until files arrive, that test runs on an empty folder and the report of V1-2.2 says so.
covered above by a real 1.4.3 export). The owner opens `inkscape-source.svg` in each tool,
exports it unchanged and puts the files in `owner/`, named `<tool>-<version>-<date>-<variant>.svg`;
`owner/README.md` gives the exact steps. The suite reads every file there: each must be refused
(with a closed reason) or cleaned into a stored form that reads back with no removal, and a file
whose name contains `-active-` must be refused for active content. Until files arrive, that test
runs on an empty folder and A6 stays open.
+45 -34
View File
@@ -2,46 +2,57 @@ Owner-supplied editor exports land here (see ../README.md).
# What to export (A6, contract §9.2 and §15)
One small drawing per tool, made in that tool, never retouched by hand afterwards. Same
reference content as `../inkscape-source.svg`, so the results compare:
Every tool exports the same drawing, so the results compare: open
`../inkscape-source.svg` (200 × 150 px: a dashed curved outline filled with a
two-stop gradient, an ellipse, a rounded rectangle with `fill-opacity` 0.5, the
group clipped by a rectangle, the live text « Garrot ») in the tool, change
nothing, and export it. Never retouch an exported file by hand.
- an outline with a curve, filled with a two-stop linear gradient, a 2 px dashed stroke;
- an ellipse with a flat colour;
- a rounded rectangle at 50 % opacity;
- the whole group clipped by a rectangle (Illustrator: clipping mask; Figma: a frame with
« Clip content »);
- the word « Garrot » as live text.
Name each file `<tool>-<version>-<date>-<variant>.svg`: the version from the
tool's « About » box (Figma in the browser shows none: write `web`), the date
as `2026-10-30`. Example: `illustrator-29.3-2026-10-30-presentation.svg`.
Artboard or frame 200 × 150 px. Note the exact version (Illustrator: Aide › À propos
d'Illustrator; Figma: menu › Aide › À propos de Figma, or the desktop app's version).
The menu labels below are the English ones; the French interface words them
differently, and I could not check them here.
## Illustrator
## Illustrator (Fichier › Ouvrir… `inkscape-source.svg`)
1. Fichier › Exporter › Exporter sous… › format SVG (`svg`), « Utiliser les plans de travail »
ticked.
2. Options SVG, two files:
- `illustrator-<version>-export-as.svg`: Style « Attributs de présentation », Police
« Vectoriser », Images « Incorporer », ID d'objet « Noms de calques », Décimales 2, Minifier
off, Responsive on;
- `illustrator-<version>-internal-css.svg`: the same with Style « CSS interne » (the export
most users keep by default; expected cleanable, its `<style>` named as removed).
3. If a third is easy: Fichier › Enregistrer sous… › SVG (`illustrator-<version>-save-as.svg`),
the older dialog, default settings.
1. File › Export › Export As… › format SVG, « Use Artboards » ticked.
2. SVG Options, two exports of the same drawing:
- `…-presentation.svg`: Styling « Presentation Attributes », Font « SVG »
(text kept as text), Images « Embed », Object IDs « Layer Names »,
Decimal 2, Minify off, Responsive on;
- `…-internal-css.svg`: the same with Styling « Internal CSS » and Font
« Convert to Outlines » (the export most users keep).
3. Optional: File › Save As… › SVG with the dialog's defaults
(`…-save-as.svg`). This older path may keep Illustrator data that the
core refuses as active content; a refusal is a valid result, tell me.
## Figma
## Figma (import `inkscape-source.svg` into a file, or drag it onto the canvas)
1. Select the frame, right panel › Exporter › « + » › format SVG.
2. Options, two files:
- `figma-<version>-export.svg`: defaults (« Contours du texte » ticked, « Inclure
l'attribut id » unticked);
- `figma-<version>-ids.svg`: « Inclure l'attribut id » ticked (the layer names become ids).
3. Exporter.
1. Select the imported frame, right panel › Export › « + » › SVG.
2. Two exports:
- `…-default.svg`: the defaults;
- `…-ids.svg`: « Include "id" attribute » ticked, « Outline text » unticked
(text kept as text).
## Affinity (Designer 2 or Affinity, if you use it; otherwise say so and A6 drops it)
File › Open `inkscape-source.svg`, then File › Export › SVG, preset « SVG (for
web) » (`…-web.svg`); if offered, a second with « SVG (digital – high quality) »
(`…-hq.svg`).
## Then
Copy the files here, unchanged, and tell Claude. The suite
`cargo test -p faunosteo-core --test svg_clean the_owner_editor_exports` reads every `.svg` in
this folder: each must be refused with a closed reason or cleaned into a stored form that reads
back with no removal. A file whose name contains `-active-` must be refused for active content
(useful if you want to try an export with a linked image or an animation). Nothing is uploaded
anywhere: these are test fixtures, committed with the tool and version in their names.
Copy the files here unchanged, then run, from the repository root:
cargo test -p faunosteo-core --test svg_clean the_owner_editor_exports -- --nocapture
It prints one line per file: the outcome (clean, cleanable or rejected), the
reason and the removed names. Send me that output with the files. A green run
alone proves only that every file was cleaned into a readable stored form or
refused with a closed reason; the printed lines say which, and I turn them into
fixed expectations in the suite. A file whose name contains `-active-` must be
refused for active content (useful if you also export one with a linked image
or an animation). Nothing is uploaded anywhere: these are test fixtures,
committed with the tool, version and date in their names.
+4
View File
@@ -365,6 +365,10 @@ fn every_url_spelling_but_the_strict_one_is_refused() {
"url(#a rgb(1,2,3))",
"url(#/a)",
"url(#a",
"url(\'#&#9;a\')",
"url(\'#a&#9;\')",
"url(\'#a&#10;\')",
"url(&quot;&#13;#a&quot;)",
] {
let cleaning = clean_svg(svg(value).as_bytes());
assert_eq!(
+5 -3
View File
@@ -79,9 +79,11 @@ whole document. Since V1-2.2 (owner decision 2026-10-29) the core writes
reference to it (`href`, `xlink:href`, `url(#…)`), so each view resolves its
gradients, clip paths and `<use>` targets in its own drawing only, never in
another view's or in the page, and the composer refuses a canvas that would
keep a reference outside its scope; a drawing whose reference names an id it
does not define, or spells a `url(` any other way than `url(#id)` with optional
quotes and spaces, is refused by the reader. The annotation
keep a reference outside its scope. A drawing whose reference names an id it
does not define is refused by the reader (`dangling_reference`); a `url(` spelt
any other way than `url(#id)` with optional quotes and spaces is removed by the
reader, so the cleaning refuses the file as active content and the verdict as a
removal. The annotation
layer carries no `id`. The PDF view keeps the drawing's ids: it is rasterised
as a document of its own. The R3 canvas goldens stay frozen; the tests apply
this declared change to them (`withIdsScopedPerView`,
+9 -8
View File
@@ -3885,7 +3885,7 @@ trapping test stub (`apps/app/tests/helpers/core.ts`) lists the two new glue mem
### 12.3 Behaviour (declared)
1. The sanitiser keeps, removes and refuses exactly as before (R3 oracles O4/O5 green); its walk also lists, in
1. The sanitiser keeps, removes and refuses exactly as before (R3 oracles O4/O5 green; the declared changes of §12.7 apart); its walk also lists, in
`active`, each removal that carried a script, an embedded document or image (`image`, `feImage`, `foreignObject`,
`iframe`, `object`, `embed`, media, and the HTML fetchers `img`, `input`, `source`, `frame`, `form`, `meta`, `link`,
`base`…), any element of the XHTML namespace, animation, a link (`a`), an event handler (`on*`), a fetching attribute
@@ -3950,7 +3950,7 @@ trapping test stub (`apps/app/tests/helpers/core.ts`) lists the two new glue mem
| Measure | Value |
|---|---|
| shipped module | 1 511 410 B raw, 558 651 B gzip -9 (script), 568 416 B by Node's gzip in `size.test.ts` (final, after §12.7: the debug `name` section dropped); budget 600 000 (D-R13); 596 671 B at ca31b78, 531 715 B before V1-2.2 (RUST.md §11.22) |
| shipped module | 1 512 071 B raw, 558 793 B gzip -9 (script), 568 625 B by Node's gzip in `size.test.ts` (final, a678faa and after: the debug `name` section dropped, §12.7); budget 600 000 (D-R13); 596 671 B at ca31b78, 531 715 B before V1-2.2 (RUST.md §11.22) |
| verdict at every cap, native | 107 ms release, 493 ms debug (document 1 165 850 bytes) |
| `clean_svg`, shipped module | 45 hostile inputs and floods, slowest 0.4 ms |
@@ -3969,7 +3969,7 @@ with a PI shown `rejected` by the preview as the verdict refuses it.
### 12.6 Residual risks and points for V1-2.3
- The module's margin under its budget is about 31.6 KB by the test's measure since §12.7 (3.3 KB before): the V1-2.5 and V1-2.6 pages must not grow the
- The module's margin under its budget is about 31.4 KB by the test's measure since §12.7 (3.3 KB before): the V1-2.5 and V1-2.6 pages must not grow the
core's browser surface without measuring.
- cairo colours: closed at the checkpoint (§12.7), strict numeric `rgb()` / `rgba()` kept.
- Editor exports: Inkscape 1.4.3 covered at the checkpoint (§12.7); Illustrator, Figma and Affinity pending from the
@@ -4004,9 +4004,9 @@ canvas goldens (`canvas.{main,head}.json`) get their ids scoped (`with_ids_scope
`withIdsScopedPerView` in the app's `canvas.trees.test.tsx`). The equine drawing is already in the stored form: its
sha256 `0ea5202f…`, the fingerprint and every R4 digest are unchanged.
Tests: `svg_scope` (6: valid reference, absent reference refused by reader, cleaning, preview and verdict, two views with
Tests: `svg_scope` (7: valid reference, absent reference refused by reader, cleaning, preview and verdict, two views with
the same ids resolving each in its own view when both canvases are inlined, a cross-view attempt, three references to
one id, deterministic and repeated composition); `svg_clean` (13: cairo export clean with its five colours, shapes and
one id, deterministic and repeated composition, a browser model over composed canvases); `svg_clean` (14: cairo export clean with its five colours, shapes and
references; seven dangerous constructs next to a colour refused as active, five other functions removed; the three
Inkscape exports; clean files uploaded in the stored form; generated drawings now also refused for dangling references);
`sheet_verdict` and `sheet_preview` (a drawing outside the stored form fails / shows rejected); `sheet_edit` (removal
@@ -4026,15 +4026,16 @@ the expansion budget was walked around the same way (pre-existing since R3). Now
reading of a `url(`: CSS whitespace, an optional quote, `#`, an id in the reference grammar, the same quote, CSS
whitespace, `)`; any other spelling is active content and refuses the file. A `url(` or `rgb(` glued to a longer name
(`xurl(`, `érgb(`, a non-ASCII letter included) is another function; a colour must end its value or list item. The
composer checks, fail-closed, that every reference of a canvas carries its view's scope (`unscoped_reference`). Tests
composer checks, fail-closed, that every reference of a canvas carries its view's scope and that each value holds
no more `url(` than local references (`unscoped_reference`), a coarse count that shares no parser with the reader. Tests
written apart from the reader: 14 refused spellings, 5 removed glued names, a 25 000-reference flood in `%61`, and a
browser model (percent-decoding, tab and newline stripping) over composed canvases (`svg_scope`). The R3 sanitiser
oracles needed no new declared change.
Residual (MEDIUM, security review): the scope is the view id, unique in one document only. A page that inlines canvases
of two documents at once (two versions side by side, two sheets) would share `fv.main.` and resolve in the first one in
tree order; such a page (V1-2.7 or later; the side-by-side diff is level 2, contract §8.3) must add a per-document
prefix before it ships. The detail modal of the report editor draws another view of the same document, so its scope
tree order; any such page (a V1-2.5 catalogue grid, a V1-2.6 preview beside the editor, a V1-2.7 comparison; the
side-by-side diff itself is level 2, contract §8.3) must add a per-document prefix before it ships. The detail modal of the report editor draws another view of the same document, so its scope
differs.
Size (V1-2.2 final, by Node's gzip in `size.test.ts`): see §12.5. Contributors measured on a build keeping the name