feat(python): value rules for every producible type

.github/workflows/fuzz.yml

@@ -1,9 +1,11 @@
 name: Fuzz (crates)
 
 # libFuzzer fuzzing for the stack crates' untrusted-input parsers (the
-# detached `packages/*/fuzz/` crates and the `fuzz:*` mise tasks). Ported from
-# cipherstash-suite's `fuzz.yml`, for the six targets in the imported set; the
-# suite keeps its three cts-common targets. Two jobs with different roles:
+# detached `packages/*/fuzz/` crates and the `fuzz:*` mise tasks), and the
+# Python SDK's value conversion (`languages/python/crates/core/fuzz/`). Ported from
+# cipherstash-suite's `fuzz.yml`, which keeps its three cts-common targets.
+# The two matrices below list every target here. Two jobs with different
+# roles:
 #
 #   * fuzz-regression (pull_request and manual, blocking on PRs): builds every harness, which
 #     catches harness and API drift, and replays the committed seed corpus with
@@ -24,6 +26,18 @@ on:
       - packages/stack-auth/**
       - packages/stack-kms/**
       - packages/stack-encrypt/**
+      # Every other crate a target compiles, from each fuzz crate's
+      # Cargo.lock: scripts/__tests__/fuzz-workflow-paths.test.mjs derives
+      # this list and fails when a path dependency is missing from it.
+      - packages/stack-profile/**
+      - packages/stack-encrypt-derive/**
+      - packages/eql/crates/eql-bindings/**
+      # The Python SDK's core crate, the manifest its dependencies come
+      # from, and the Go guest's EQL resolver it compiles by #[path]: the
+      # fuzz:python-values target.
+      - languages/python/crates/core/**
+      - languages/python/Cargo.toml
+      - languages/golang/encrypt/guest/src/targets.rs
       - Cargo.toml
       - Cargo.lock
       - mise.toml
@@ -80,6 +94,7 @@ jobs:
           - { task: "fuzz:plan-build", slug: plan-build }
           - { task: "fuzz:json-wire", slug: json-wire }
           - { task: "fuzz:json-text", slug: json-text }
+          - { task: "fuzz:python-values", slug: python-values }
     steps:
       - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6
         with:
@@ -104,6 +119,7 @@ jobs:
             packages/stack-auth/fuzz
             packages/stack-kms/fuzz
             packages/stack-encrypt/fuzz
+            languages/python/crates/core/fuzz
           key: ${{ matrix.slug }}
 
       # `-runs=0` replays the committed seed corpus once and exits without
@@ -130,6 +146,7 @@ jobs:
           - { task: "fuzz:plan-build",   slug: plan-build,   dir: packages/stack-encrypt, target: plan_build }
           - { task: "fuzz:json-wire",    slug: json-wire,    dir: packages/stack-encrypt, target: json_wire_decode }
           - { task: "fuzz:json-text",    slug: json-text,    dir: packages/stack-encrypt, target: json_text }
+          - { task: "fuzz:python-values", slug: python-values, dir: languages/python/crates/core, target: value_frame }
     steps:
       - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6
         with:
@@ -169,13 +186,16 @@ jobs:
         run: mise run ${{ matrix.task }} -- "-max_total_time=$MAX_TOTAL_TIME"
 
       # Drops inputs that add no coverage, so the saved corpus stays small.
-      # Skipped when the fuzz step found a crash.
+      # Skipped when the fuzz step found a crash. Run from the root, naming
+      # the fuzz crate: from languages/python, mise would load that folder's
+      # mise.toml and install its Python and uv, which this job does not
+      # cache.
       - name: Minimise the corpus (${{ matrix.slug }})
-        working-directory: ${{ matrix.dir }}
         env:
           TARGET: ${{ matrix.target }}
+          FUZZ_DIR: ${{ matrix.dir }}/fuzz
         run: |
-          cargo +nightly fuzz cmin "$TARGET" --sanitizer none --target "$(rustc -vV | sed -n 's/^host: //p')"
+          cargo +nightly fuzz cmin "$TARGET" --fuzz-dir "$FUZZ_DIR" --sanitizer none --target "$(rustc -vV | sed -n 's/^host: //p')"
 
       # Saved even on a crash: the grown corpus is still worth keeping, and the
       # crash input lives in artifacts/, not corpus/.

.gitignore

@@ -107,6 +107,10 @@ packages/*/fuzz/artifacts/
 packages/*/fuzz/coverage/
 packages/*/fuzz/corpus/*/*
 !packages/*/fuzz/corpus/*/valid-*
+languages/python/crates/core/fuzz/artifacts/
+languages/python/crates/core/fuzz/coverage/
+languages/python/crates/core/fuzz/corpus/*/*
+!languages/python/crates/core/fuzz/corpus/*/valid-*
 # cargo-mutants output (`mise run mutants:<crate>` writes it to the cwd).
 mutants.out/
 

AGENTS.md

@@ -93,7 +93,7 @@ Every npm package except EQL lives under `languages/typescript/`: packages in `l
 - `packages/stack-auth`, `packages/stack-profile`, `packages/stack-kms`, `packages/stack-encrypt`, `packages/stack-encrypt-derive`, `packages/stack-guest-abi`: The Rust crates imported from `cipherstash/cipherstash-suite` with their history — `stack-auth` and `stack-profile` (published to crates.io, one version group), `stack-kms` (published to crates.io from 0.1.0, its own version group, re-exported by `stack-encrypt` as `stack_encrypt::kms`), `stack-encrypt` and `stack-encrypt-derive` (published to crates.io from 0.1.0, one version group; `eql-bindings`' `stack-encrypt` feature depends on them from the registry), and `stack-guest-abi` (`publish = false`). They are the members of the **root Cargo workspace**, with the three node binding crates below. See "Working on the Rust crates".
 - `languages/typescript/packages/auth`, `languages/typescript/packages/profile`, `languages/typescript/packages/stack-auth-wasm`: The node bindings of those crates. `@cipherstash/auth` (napi-rs v2) and its six `platforms/*` packages are published to npm from this repository by `release.yml` (`auth-artifacts`, `publish-auth`); a change to what it ships, the `stack-auth` crate included, needs an `@cipherstash/auth` changeset (`require-auth-npm-changeset.yml`). `@cipherstash/profile` and its platforms are private and never published; `@cipherstash/stack-auth-wasm` is private and builds the wasm that `@cipherstash/auth` ships. Their `build` and `test` scripts never invoke cargo; `build:native`, `build:debug` and `test:cargo` do.
 - `languages/golang`: The Go module — the SDK `encrypt` with its generated-code support `encrypt/gensupport` and the policy packages `encrypt/policy` and `encrypt/policy/protosource`; the credential package `auth`; the generator `stashgen` and its command `cmd/stashgen`; and `internal` (the shared guest plumbing and the `record` wire model). A wazero host with no cgo. Its two WASI guests (`encrypt/guest`, `auth/guest`) are detached Cargo workspaces built by `mise run wasm:guest:build` and `mise run wasm:auth-guest:build`; `mise run wasm:guest:build:deterministic` builds the seeded TEST build the hermetic Go tests use; the `.wasm` files they embed are gitignored. The stack-encrypt guest has a second build with the EQL types (`mise run wasm:guest:build:eql`, cargo feature `eql`, linking `eql-bindings` by path; `wasm:guest:build:eql:deterministic` is its test build), embedded by `encrypt/eql` and registered on import, so a program whose generated code names an EQL type runs it (ADR-0007, amended). `encrypt/eql`'s types and `Types` table are generated from the EQL catalog by `eql-codegen` (`eql_gen.go`, written by `mise run types:generate` in `packages/eql` and drift-gated by `types:check` and `cargo test -p eql-codegen`). Generated `*_stash.go` files are committed and CI fails when `go generate ./...` changes one. Releases are cut by hand with `.github/workflows/release-golang.yml` (`gh workflow run release-golang.yml -f version=X.Y.Z[-pre] --ref <branch>`): because the guests are gitignored on main and Go fetches a version as the files committed at its tag, the workflow builds them, commits them on a release-only commit no branch points at, tags it `languages/golang/vX.Y.Z` and pushes the tag alone, then attaches the EQL v4 SQL bundle to the GitHub release. The ref must carry the EQL v4 bundle, or the build stops at "Build the EQL SQL bundles". `scripts/go-release-version.mjs` refuses, before anything is built, every version Go would refuse (non-canonical semver, `+build` metadata, a v2+ major on a module path with no `/vN` suffix, a version shaped like a pseudo-version), a final version dispatched from anywhere but `main`, and a final version while `languages/golang/go.mod` pins a `github.com/cipherstash/vitaminc/…` module at a pseudo-version (vitaminc's Go bindings have no tags, and a TEMPORARY pin on an unreleased commit must be gone before the module is tagged); a pre-release may come from a branch and may carry such a pin. The build job also refuses a version the Go proxy already lists in `@v/list`, which catches a tag deleted after the proxy fetched it; it never asks the proxy for the version itself before the tag exists, because the proxy would cache "not found" for up to 30 minutes and `verify` would fail after the push. The build checks the `Cargo.lock` of the EQL workspace and both guest workspaces with `--locked` first, since none of those builds passes it. It runs in a job with no write token and hands the guests to a `publish` job that holds `contents: write` and runs only git and gh. `publish` fetches the commit with plain unauthenticated git rather than `actions/checkout`, whose `token` input defaults to the job's writable token, and refuses a release commit whose tree is not the one the build tested. `publish` takes the paths it commits from the workflow's fixed `GUESTS` list, never from the artifact, and refuses a `guests.sha256` that is not exactly one SHA-256 line for each of those three guests: the build job runs third-party code after writing that manifest, so a path read from it could put any file on the tag (`scripts/__tests__/go-release-artifact.test.mjs`). A version is immutable once the Go proxy has fetched it, so a tag is never moved or deleted: if `publish` or `verify` fails after the tag is pushed, use "Re-run failed jobs" on the same run, which keeps a tag that run pushed and creates or completes the release; a new dispatch of the same version is refused. Never tag a main commit by hand: `go get` would fetch a module with no guests.
-- `languages/python`: The Python SDK, `stackencrypt`, a pyo3 extension over `stack-encrypt`, built by maturin. **The fourth Cargo workspace**, excluded from the root one. It has two crates. `crates/core` has no pyo3: the engine-side logic, linked and tested without a Python interpreter. `crates/engine` is the `_engine` cdylib, `abi3-py39`, one wheel per platform for Python 3.9 to 3.14. `crates/core` compiles the Go guest's EQL resolver, `languages/golang/encrypt/guest/src/targets.rs`, in place through `#[path]`. So both SDKs run one resolver, and a change to that file is a change to both. The `extension-module` feature is set only by maturin in `pyproject.toml`, never in a crate manifest, so `cargo test` links libpython. Its tasks live in its own `mise.toml` and run from that folder. `mise run build` builds the wheel into `.venv`. `mise run test` runs pytest. `mise run test:cargo` runs plain `cargo test --locked` for both crates, with libpython on the loader path. It does not use nextest, and its CI job sets no `MISE_ENV`. The nextest rule under "Working on the Rust crates" is for the root workspace, whose `stack-kms` tests set environment variables. No test in this workspace does, and nextest comes from `mise.test.toml`, which this workspace's CI does not load. The wheel's version is `crates/engine/Cargo.toml`'s. `stackencrypt.__engine__` is the commit of this checkout, recorded by `crates/core/build.rs`. Its `Cargo.lock` is seeded from the root one, so every crate it shares with the root workspace resolves to the version the root tests. After a root dependency moves, re-seed it the same way: copy the root `Cargo.lock` over it and run `cargo metadata --offline` there. That command prunes the copy to what the workspace uses. Until then, `tests-python.yml` fails on `--locked`. Nothing is published. CI is `tests-python.yml`. It runs the Rust checks and pytest on Python 3.9 and 3.14, where `MISE_PYTHON_VERSION` moves a leg off the local version. A `live` job runs `mise run test:live`: `encrypt`, `decrypt` and `query` against ZeroKMS with the repository's `CS_*` secrets, and the rows inserted into, searched in and read back from a `postgres:17` service with the EQL v4 bundle built from the tree. It skips on a fork's pull request, and the task fails on any skipped test. It then builds one abi3 wheel per FFI platform, uploads them as artifacts, and writes each build time to the job summary. `scripts/__tests__/python-ci.test.mjs` compares its path filter with the workspace's `Cargo.lock` and its wheel matrix with the protect-ffi platform packages. There is no Python lockfile yet. The package has no runtime dependency, so only the `test` extra could be locked: pytest, attrs, Pydantic, SQLAlchemy, and psycopg. The middle three are for the declaration tests, and psycopg is for the live tests. None has an upper bound, so the tests meet each new release when users do. `python/stackencrypt/_eql.py`, the EQL type markers, is generated from the EQL catalog by `eql-codegen py-eql`. To regenerate it, run `mise run types:generate` in `packages/eql`. Never edit it by hand: `tests-python.yml` runs its parity tests and fails on a hand edit. Design: the Python ADRs under `languages/python/docs/adr/` (ADR-0001, the declaration; ADR-0002, the call shape of `query`), the spec in #1341.
+- `languages/python`: The Python SDK, `stackencrypt`, a pyo3 extension over `stack-encrypt`, built by maturin. **The fourth Cargo workspace**, excluded from the root one. It has two crates. `crates/core` has no pyo3: the engine-side logic, linked and tested without a Python interpreter. `crates/engine` is the `_engine` cdylib, `abi3-py39`, one wheel per platform for Python 3.9 to 3.14. `crates/core/fuzz` is a detached cargo-fuzz crate over the value conversion. The root `mise.toml`'s `fuzz:python-values` task runs it, from `fuzz.yml`, so it is the one Python task outside `languages/python/mise.toml`. `crates/core/testdata/value-conformance.json` is a copy of vitaminc's shared value corpus; its `README.md` records the commit it came from. `crates/core` compiles the Go guest's EQL resolver, `languages/golang/encrypt/guest/src/targets.rs`, in place through `#[path]`. So both SDKs run one resolver, and a change to that file is a change to both. The `extension-module` feature is set only by maturin in `pyproject.toml`, never in a crate manifest, so `cargo test` links libpython. Its tasks live in its own `mise.toml` and run from that folder. `mise run build` builds the wheel into `.venv`. `mise run test` runs pytest. `mise run test:cargo` runs plain `cargo test --locked` for both crates, with libpython on the loader path. It does not use nextest, and its CI job sets no `MISE_ENV`. The nextest rule under "Working on the Rust crates" is for the root workspace, whose `stack-kms` tests set environment variables. No test in this workspace does, and nextest comes from `mise.test.toml`, which this workspace's CI does not load. The wheel's version is `crates/engine/Cargo.toml`'s. `stackencrypt.__engine__` is the commit of this checkout, recorded by `crates/core/build.rs`. Its `Cargo.lock` is seeded from the root one, so every crate it shares with the root workspace resolves to the version the root tests. After a root dependency moves, re-seed it the same way: copy the root `Cargo.lock` over it and run `cargo metadata --offline` there. That command prunes the copy to what the workspace uses. Until then, `tests-python.yml` fails on `--locked`. Nothing is published. CI is `tests-python.yml`. It runs the Rust checks and pytest on Python 3.9 and 3.14, where `MISE_PYTHON_VERSION` moves a leg off the local version. A `live` job runs `mise run test:live`: `encrypt`, `decrypt` and `query` against ZeroKMS with the repository's `CS_*` secrets, and the rows inserted into, searched in and read back from a `postgres:17` service with the EQL v4 bundle built from the tree. It skips on a fork's pull request, and the task fails on any skipped test. It then builds one abi3 wheel per FFI platform, uploads them as artifacts, and writes each build time to the job summary. `scripts/__tests__/python-ci.test.mjs` compares its path filter with the workspace's `Cargo.lock` and its wheel matrix with the protect-ffi platform packages. There is no Python lockfile yet. The package has no runtime dependency, so only the `test` extra could be locked: pytest, attrs, Pydantic, SQLAlchemy, and psycopg. The middle three are for the declaration tests, and psycopg is for the live tests. None has an upper bound, so the tests meet each new release when users do. `python/stackencrypt/_eql.py`, the EQL type markers, is generated from the EQL catalog by `eql-codegen py-eql`. To regenerate it, run `mise run types:generate` in `packages/eql`. Never edit it by hand: `tests-python.yml` runs its parity tests and fails on a hand edit. Design: the Python ADRs under `languages/python/docs/adr/` (ADR-0001, the declaration; ADR-0002, the call shape of `query`), the spec in #1341.
 - `e2e/*`: Cross-package end-to-end tests (package managers, supply chain, Prisma example README)
 - `languages/typescript/examples/*`: Working apps (basic, prisma, supabase-worker)
 - `docs/plans/*`: Internal design plans. User-facing documentation lives at https://cipherstash.com/docs (not in this repo).
@@ -720,7 +720,8 @@ exact pin would stop the suite sharing one `cts-common` with it.
   crates and the three node binding crates), protect-ffi's, EQL's and the
   Python SDK's (`languages/python`). The root `Cargo.toml` excludes the other
   three; protect-ffi and EQL set their own vitaminc version. The
-  three fuzz crates, the two Go guests and the EQL v4 block ORE timing
+  four fuzz crates (three under `packages/*/fuzz`, one under
+  `languages/python/crates/core/fuzz`), the two Go guests and the EQL v4 block ORE timing
   harness's ciphertext generator
   (`packages/eql/tests/timing/ore_block_v2/ctgen`) are detached workspaces
   too, each with its own `Cargo.lock`. None of them has a Dependabot entry.