typedframes 0.14.1__tar.gz → 0.14.3__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (71) hide show
  1. {typedframes-0.14.1 → typedframes-0.14.3}/PKG-INFO +14 -13
  2. {typedframes-0.14.1 → typedframes-0.14.3}/README.md +13 -12
  3. {typedframes-0.14.1 → typedframes-0.14.3}/pyproject.toml +6 -4
  4. {typedframes-0.14.1 → typedframes-0.14.3}/rust/Cargo.lock +2 -3
  5. {typedframes-0.14.1 → typedframes-0.14.3}/rust/Cargo.toml +9 -3
  6. {typedframes-0.14.1 → typedframes-0.14.3}/rust/python/Cargo.toml +1 -1
  7. typedframes-0.14.3/rust/python-data/scripts/typedframes-py +4 -0
  8. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/cli/config.rs +128 -33
  9. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/cli/coverage.rs +4 -5
  10. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/cli/report.rs +4 -7
  11. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/sql.rs +96 -20
  12. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/__init__.py +1 -1
  13. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/_binary.py +5 -0
  14. {typedframes-0.14.1 → typedframes-0.14.3}/.cargo/config.toml +0 -0
  15. {typedframes-0.14.1 → typedframes-0.14.3}/LICENSE +0 -0
  16. {typedframes-0.14.1 → typedframes-0.14.3}/rust/README.md +0 -0
  17. {typedframes-0.14.1 → typedframes-0.14.3}/rust/benches/parser_bench.rs +0 -0
  18. {typedframes-0.14.1 → typedframes-0.14.3}/rust/build.rs +0 -0
  19. {typedframes-0.14.1 → typedframes-0.14.3}/rust/python/build.rs +0 -0
  20. {typedframes-0.14.1 → typedframes-0.14.3}/rust/python/src/lib.rs +0 -0
  21. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/apply.rs +0 -0
  22. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/ast_extract.rs +0 -0
  23. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/call_usage.rs +0 -0
  24. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/cli/args.rs +0 -0
  25. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/cli/cache.rs +0 -0
  26. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/cli/check.rs +0 -0
  27. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/cli/collect.rs +0 -0
  28. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/cli/mod.rs +0 -0
  29. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/cli/pathutil.rs +0 -0
  30. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/cli/pyfmt.rs +0 -0
  31. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/cli/style.rs +0 -0
  32. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/cli/suggest.rs +0 -0
  33. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/column_consuming_methods.rs +0 -0
  34. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/column_usage.rs +0 -0
  35. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/config.rs +0 -0
  36. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/constants.rs +0 -0
  37. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/contract.rs +0 -0
  38. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/engine.rs +0 -0
  39. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/errors.rs +0 -0
  40. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/fast_hash.rs +0 -0
  41. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/frame_ops.rs +0 -0
  42. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/helpers.rs +0 -0
  43. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/index.rs +0 -0
  44. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/lib.rs +0 -0
  45. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/library_names.txt +0 -0
  46. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/linter.rs +0 -0
  47. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/main.rs +0 -0
  48. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/notebook.rs +0 -0
  49. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/parallel.rs +0 -0
  50. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/phrase.rs +0 -0
  51. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/report.rs +0 -0
  52. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/report_minimize.rs +0 -0
  53. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/report_names.rs +0 -0
  54. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/static_eval.rs +0 -0
  55. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/suggest.rs +0 -0
  56. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/typo.rs +0 -0
  57. {typedframes-0.14.1 → typedframes-0.14.3}/rust/src/words.txt +0 -0
  58. {typedframes-0.14.1 → typedframes-0.14.3}/rust/tests/cli.rs +0 -0
  59. {typedframes-0.14.1 → typedframes-0.14.3}/rust/tests/integration_test.rs +0 -0
  60. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/__main__.py +0 -0
  61. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/_rust.pyi +0 -0
  62. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/base_schema.py +0 -0
  63. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/column.py +0 -0
  64. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/column_group.py +0 -0
  65. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/column_group_error.py +0 -0
  66. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/column_set.py +0 -0
  67. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/missing_dependency_error.py +0 -0
  68. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/mypy.py +0 -0
  69. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/pandera.py +0 -0
  70. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/py.typed +0 -0
  71. {typedframes-0.14.1 → typedframes-0.14.3}/src/typedframes/schema_algebra.py +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: typedframes
3
- Version: 0.14.1
3
+ Version: 0.14.3
4
4
  Classifier: Development Status :: 3 - Alpha
5
5
  Classifier: Intended Audience :: Developers
6
6
  Classifier: License :: OSI Approved :: MIT License
@@ -44,7 +44,7 @@ Project-URL: Repository, https://github.com/w-martin/typedframes
44
44
 
45
45
  > ⚠️ **Project Status: Proof of Concept**
46
46
  >
47
- > `typedframes` (v0.14.1) is currently an experimental proof-of-concept. The core static analysis and mypy/Rust
47
+ > `typedframes` (v0.14.3) is currently an experimental proof-of-concept. The core static analysis and mypy/Rust
48
48
  > integrations work, but expect rough edges. The codebase prioritizes demonstrating the viability of static DataFrame
49
49
  > column checking over production-grade stability.
50
50
  >
@@ -145,7 +145,8 @@ Where the Python interpreter may run code but no other executable may be started
145
145
  the wheel, instead of starting the `typedframes` executable. Results and exit status are the same; when the checker is
146
146
  asked to capture its output (as the mypy plugin does), a usage error or a missing path comes back with it, and any
147
147
  warning a command prints still goes to the terminal. Without the extension installed (for example from a source
148
- distribution) `TYPEDFRAMES_INPROC=1` is an error.
148
+ distribution) `TYPEDFRAMES_INPROC=1` is an error. The `typedframes-py` command always runs in the interpreter, without
149
+ the variable.
149
150
 
150
151
  ---
151
152
 
@@ -514,7 +515,7 @@ repos:
514
515
  name: typedframes check
515
516
  entry: typedframes check . --strict
516
517
  language: python
517
- additional_dependencies: ["typedframes==0.14.1"]
518
+ additional_dependencies: ["typedframes==0.14.3"]
518
519
  types_or: [python, jupyter]
519
520
  pass_filenames: false
520
521
  ```
@@ -563,7 +564,7 @@ The action installs the PyPI wheel into a throwaway virtualenv and runs the chec
563
564
  | Input | Default | |
564
565
  |-------|---------|-|
565
566
  | `path` | `.` | File or directory to check |
566
- | `version` | `latest` | PyPI version to install, e.g. `"0.14.1"` |
567
+ | `version` | `latest` | PyPI version to install, e.g. `"0.14.3"` |
567
568
  | `strict` | `true` | Fail the step on errors — `typedframes check` exits 0 without it |
568
569
  | `coverage-fail-under` | *(unset)* | Minimum DataFrame schema coverage, e.g. `"90"` |
569
570
  | `coverage-detail` | `summary` | Or `term-missing` for the per-file breakdown, `explain` to diagnose a lower-than-expected total |
@@ -703,17 +704,17 @@ plus the requested features, so both halves get checked normally. See
703
704
  Fast feedback reduces development time. The typedframes Rust binary provides near-instant column checking.
704
705
 
705
706
  **Benchmark results** (median of 5 runs ± one standard deviation, 3 warmup, caches cleared between runs):
706
- *2026-10-13 · Darwin 27.0.0 · arm · CPython 3.15.0 · 64GiB RAM · Great Expectations pinned @ 1.23.2*
707
+ *2026-10-18 · Darwin 27.0.0 · arm · CPython 3.15.0 · 64GiB RAM · Great Expectations pinned @ 1.23.2*
707
708
 
708
709
  | Tool | Version | What it does | typedframes (12 files) | great_expectations (494 files) |
709
710
  |------|---------|--------------|------------------------|--------------------------------|
710
- | typedframes | 0.14.0 | DataFrame column checker | 18.1 ± 0.2 ms | 48.6 ± 0.1 ms |
711
- | ruff | 0.17.0 | Linter (no type checking) | 25.1 ± 1.7 ms | 214.6 ± 2.0 ms |
712
- | ty | 0.0.84 | Type checker | 51.1 ± 10.3 ms | 754.2 ± 9.0 ms |
713
- | pyrefly | 1.3.2 | Type checker | 99.3 ± 11.0 ms | 299.0 ± 6.4 ms |
714
- | mypy | 2.4.0 | Type checker (no plugin) | 1.79 ± 0.01 s | 2.99 ± 0.03 s |
715
- | mypy + typedframes | 2.4.0 | Type checker + column checker | 1.81 ± 0.01 s | 3.16 ± 0.01 s |
716
- | pyright | 1.1.414 | Type checker | 656.8 ± 5.8 ms | 3.86 ± 0.07 s |
711
+ | typedframes | 0.14.2 | DataFrame column checker | 21.2 ± 0.8 ms | 50.8 ± 0.6 ms |
712
+ | ruff | 0.17.1 | Linter (no type checking) | 25.9 ± 0.4 ms | 224.4 ± 3.4 ms |
713
+ | ty | 0.0.89 | Type checker | 52.5 ± 0.6 ms | 818.7 ± 24.5 ms |
714
+ | pyrefly | 1.4.0 | Type checker | 116.8 ± 5.8 ms | 323.6 ± 10.1 ms |
715
+ | mypy | 2.4.0 | Type checker (no plugin) | 2.01 ± 0.14 s | 3.37 ± 0.03 s |
716
+ | mypy + typedframes | 2.4.0 | Type checker + column checker | 2.04 ± 0.02 s | 3.68 ± 0.10 s |
717
+ | pyright | 1.1.414 | Type checker | 714.9 ± 20.2 ms | 4.59 ± 0.33 s |
717
718
 
718
719
  *Run `uv run python benchmarks/benchmark_checkers.py` to reproduce.*
719
720
 
@@ -8,7 +8,7 @@
8
8
 
9
9
  > ⚠️ **Project Status: Proof of Concept**
10
10
  >
11
- > `typedframes` (v0.14.1) is currently an experimental proof-of-concept. The core static analysis and mypy/Rust
11
+ > `typedframes` (v0.14.3) is currently an experimental proof-of-concept. The core static analysis and mypy/Rust
12
12
  > integrations work, but expect rough edges. The codebase prioritizes demonstrating the viability of static DataFrame
13
13
  > column checking over production-grade stability.
14
14
  >
@@ -109,7 +109,8 @@ Where the Python interpreter may run code but no other executable may be started
109
109
  the wheel, instead of starting the `typedframes` executable. Results and exit status are the same; when the checker is
110
110
  asked to capture its output (as the mypy plugin does), a usage error or a missing path comes back with it, and any
111
111
  warning a command prints still goes to the terminal. Without the extension installed (for example from a source
112
- distribution) `TYPEDFRAMES_INPROC=1` is an error.
112
+ distribution) `TYPEDFRAMES_INPROC=1` is an error. The `typedframes-py` command always runs in the interpreter, without
113
+ the variable.
113
114
 
114
115
  ---
115
116
 
@@ -478,7 +479,7 @@ repos:
478
479
  name: typedframes check
479
480
  entry: typedframes check . --strict
480
481
  language: python
481
- additional_dependencies: ["typedframes==0.14.1"]
482
+ additional_dependencies: ["typedframes==0.14.3"]
482
483
  types_or: [python, jupyter]
483
484
  pass_filenames: false
484
485
  ```
@@ -527,7 +528,7 @@ The action installs the PyPI wheel into a throwaway virtualenv and runs the chec
527
528
  | Input | Default | |
528
529
  |-------|---------|-|
529
530
  | `path` | `.` | File or directory to check |
530
- | `version` | `latest` | PyPI version to install, e.g. `"0.14.1"` |
531
+ | `version` | `latest` | PyPI version to install, e.g. `"0.14.3"` |
531
532
  | `strict` | `true` | Fail the step on errors — `typedframes check` exits 0 without it |
532
533
  | `coverage-fail-under` | *(unset)* | Minimum DataFrame schema coverage, e.g. `"90"` |
533
534
  | `coverage-detail` | `summary` | Or `term-missing` for the per-file breakdown, `explain` to diagnose a lower-than-expected total |
@@ -667,17 +668,17 @@ plus the requested features, so both halves get checked normally. See
667
668
  Fast feedback reduces development time. The typedframes Rust binary provides near-instant column checking.
668
669
 
669
670
  **Benchmark results** (median of 5 runs ± one standard deviation, 3 warmup, caches cleared between runs):
670
- *2026-10-13 · Darwin 27.0.0 · arm · CPython 3.15.0 · 64GiB RAM · Great Expectations pinned @ 1.23.2*
671
+ *2026-10-18 · Darwin 27.0.0 · arm · CPython 3.15.0 · 64GiB RAM · Great Expectations pinned @ 1.23.2*
671
672
 
672
673
  | Tool | Version | What it does | typedframes (12 files) | great_expectations (494 files) |
673
674
  |------|---------|--------------|------------------------|--------------------------------|
674
- | typedframes | 0.14.0 | DataFrame column checker | 18.1 ± 0.2 ms | 48.6 ± 0.1 ms |
675
- | ruff | 0.17.0 | Linter (no type checking) | 25.1 ± 1.7 ms | 214.6 ± 2.0 ms |
676
- | ty | 0.0.84 | Type checker | 51.1 ± 10.3 ms | 754.2 ± 9.0 ms |
677
- | pyrefly | 1.3.2 | Type checker | 99.3 ± 11.0 ms | 299.0 ± 6.4 ms |
678
- | mypy | 2.4.0 | Type checker (no plugin) | 1.79 ± 0.01 s | 2.99 ± 0.03 s |
679
- | mypy + typedframes | 2.4.0 | Type checker + column checker | 1.81 ± 0.01 s | 3.16 ± 0.01 s |
680
- | pyright | 1.1.414 | Type checker | 656.8 ± 5.8 ms | 3.86 ± 0.07 s |
675
+ | typedframes | 0.14.2 | DataFrame column checker | 21.2 ± 0.8 ms | 50.8 ± 0.6 ms |
676
+ | ruff | 0.17.1 | Linter (no type checking) | 25.9 ± 0.4 ms | 224.4 ± 3.4 ms |
677
+ | ty | 0.0.89 | Type checker | 52.5 ± 0.6 ms | 818.7 ± 24.5 ms |
678
+ | pyrefly | 1.4.0 | Type checker | 116.8 ± 5.8 ms | 323.6 ± 10.1 ms |
679
+ | mypy | 2.4.0 | Type checker (no plugin) | 2.01 ± 0.14 s | 3.37 ± 0.03 s |
680
+ | mypy + typedframes | 2.4.0 | Type checker + column checker | 2.04 ± 0.02 s | 3.68 ± 0.10 s |
681
+ | pyright | 1.1.414 | Type checker | 714.9 ± 20.2 ms | 4.59 ± 0.33 s |
681
682
 
682
683
  *Run `uv run python benchmarks/benchmark_checkers.py` to reproduce.*
683
684
 
@@ -4,7 +4,7 @@ build-backend = "maturin"
4
4
 
5
5
  [project]
6
6
  name = "typedframes"
7
- version = "0.14.1"
7
+ version = "0.14.3"
8
8
  description = "Static analysis for pandas and polars DataFrames. Catch column errors at lint-time, not runtime."
9
9
  keywords = ["pandas", "polars", "type-checking", "static-analysis", "dataframe", "linter", "mypy-plugin"]
10
10
  readme = "README.md"
@@ -47,6 +47,8 @@ python-source = "src"
47
47
  # The wheel carries the Rust `typedframes` executable (installed as a script) next to the Python package.
48
48
  bindings = "bin"
49
49
  module-name = "typedframes"
50
+ # `typedframes-py`: a Python script installed beside the executable, which maturin cannot declare as an entry point.
51
+ data = "rust/python-data"
50
52
  # The cargo config carries the allocator tuning, so a build from the sdist is tuned too.
51
53
  include = [
52
54
  { path = "src/typedframes/_rust.*", format = "wheel" },
@@ -140,13 +142,13 @@ dev = [
140
142
  "invoke>=3.0.3",
141
143
  "maturin>=1.15.0",
142
144
  "pre-commit>=4.7.0",
143
- "pyrefly>=1.3.2",
145
+ "pyrefly>=1.4.0",
144
146
  "pyright>=1.1.414",
145
147
  "pytest>=9.1.1",
146
148
  "pytest-cov>=7.1.0",
147
149
  "pytest-xdist>=3.8.0",
148
- "ruff>=0.17.0",
149
- "ty>=0.0.87",
150
+ "ruff>=0.17.1",
151
+ "ty>=0.0.89",
150
152
  "openpyxl>=3.1.5",
151
153
  "pyarrow>=26.0.0",
152
154
  "xlsxwriter>=3.2.9",
@@ -1469,13 +1469,12 @@ checksum = "7d56353a2a665ad0f41a421187180aab746c8c325620617ad883a99a1cbe66d2"
1469
1469
 
1470
1470
  [[package]]
1471
1471
  name = "typedframes_checker"
1472
- version = "0.14.1"
1472
+ version = "0.14.3"
1473
1473
  dependencies = [
1474
1474
  "anyhow",
1475
1475
  "clap",
1476
1476
  "criterion",
1477
1477
  "mimalloc",
1478
- "regex",
1479
1478
  "rmp-serde",
1480
1479
  "ruff_notebook",
1481
1480
  "ruff_python_ast",
@@ -1492,7 +1491,7 @@ dependencies = [
1492
1491
 
1493
1492
  [[package]]
1494
1493
  name = "typedframes_python"
1495
- version = "0.14.1"
1494
+ version = "0.14.3"
1496
1495
  dependencies = [
1497
1496
  "pyo3",
1498
1497
  "pyo3-build-config",
@@ -1,6 +1,6 @@
1
1
  [package]
2
2
  name = "typedframes_checker"
3
- version = "0.14.1"
3
+ version = "0.14.3"
4
4
  edition = "2021"
5
5
  description = "A high-performance Rust-based checker for typedframes DataFrame column schemas"
6
6
  license = "MIT"
@@ -20,6 +20,13 @@ codegen-units = 1
20
20
  # a Python exception and not end the interpreter that loaded it.
21
21
  strip = "symbols"
22
22
 
23
+ # The executable wheel is built with this profile: it never loads into another process, so a panic
24
+ # may end it, and it favours size over the last ~10% of speed. The extension module keeps `release`.
25
+ [profile.cli]
26
+ inherits = "release"
27
+ panic = "abort"
28
+ opt-level = "s"
29
+
23
30
  [lints.rust]
24
31
  unsafe_code = "forbid"
25
32
 
@@ -38,9 +45,8 @@ serde_json = { version = "1.0", features = ["preserve_order"] }
38
45
  rmp-serde = "1"
39
46
  anyhow = "1.0"
40
47
  toml = { version = "1.1", features = ["preserve_order"] }
41
- sqlparser = "0.62"
48
+ sqlparser = { version = "0.62", default-features = false, features = ["std", "recursive-protection"] }
42
49
  clap = { version = "4.6.7", features = ["derive"] }
43
- regex = "1.13.1"
44
50
  # The allocator split ruff and ty use: jemalloc where it builds, mimalloc on Windows (jemalloc
45
51
  # has no MSVC support), the platform allocator elsewhere. Keep the `cfg`s in `src/lib.rs` in
46
52
  # step with these. jemalloc is tuned for a short run on many threads in `../.cargo/config.toml`.
@@ -1,6 +1,6 @@
1
1
  [package]
2
2
  name = "typedframes_python"
3
- version = "0.14.1"
3
+ version = "0.14.3"
4
4
  edition = "2021"
5
5
  description = "The typedframes checker as a Python extension module, for interpreters that may not start new executables"
6
6
  license = "MIT"
@@ -0,0 +1,4 @@
1
+ #!python
2
+ from typedframes._binary import main_in_process
3
+
4
+ main_in_process()
@@ -2,7 +2,6 @@
2
2
  //! collecting files, and the opt-in DataFrame schema coverage gate.
3
3
 
4
4
  use super::pyfmt::{float_str, repr_str};
5
- use regex::Regex;
6
5
  use std::collections::HashSet;
7
6
  use std::fs;
8
7
  use std::path::Path;
@@ -241,42 +240,98 @@ pub(crate) fn load_configured_excludes(path: &Path) -> Option<HashSet<String>> {
241
240
  )
242
241
  }
243
242
 
244
- /// An anchored regex for a project-relative path glob: `**` spans any number of path segments
245
- /// (`**/` consumes its separator too, so `**/x.py` still matches a bare `x.py`), `*` and `?` never
246
- /// cross a `/`, everything else is literal.
247
- pub(crate) fn glob_to_regex(pattern: &str) -> Regex {
248
- let chars: Vec<char> = pattern.chars().collect();
249
- let mut out = String::from("^(?:");
250
- let mut i = 0;
251
- while i < chars.len() {
252
- let ch = chars[i];
253
- if ch == '*' {
254
- let mut star_end = i;
255
- while star_end < chars.len() && chars[star_end] == '*' {
256
- star_end += 1;
243
+ /// One step of a compiled glob.
244
+ enum Step {
245
+ Literal(char),
246
+ /// `?`: one character other than `/`.
247
+ One,
248
+ /// `*`: any run of characters other than `/`.
249
+ Segment,
250
+ /// `**`: any run of characters, line breaks aside.
251
+ Any,
252
+ /// Front of `**/`: the two steps after it may be skipped, so `**/x.py` still matches a bare `x.py`.
253
+ Optional,
254
+ }
255
+
256
+ /// A project-relative path glob, matched against the whole path: `**` spans any number of path
257
+ /// segments (`**/` consumes its separator too), `*` and `?` never cross a `/`, everything else is
258
+ /// literal.
259
+ pub(crate) struct Glob {
260
+ steps: Vec<Step>,
261
+ }
262
+
263
+ impl Glob {
264
+ pub(crate) fn new(pattern: &str) -> Self {
265
+ let chars: Vec<char> = pattern.chars().collect();
266
+ let mut steps = Vec::new();
267
+ let mut i = 0;
268
+ while i < chars.len() {
269
+ match chars[i] {
270
+ '*' => {
271
+ let mut end = i;
272
+ while chars.get(end) == Some(&'*') {
273
+ end += 1;
274
+ }
275
+ if end - i < 2 {
276
+ steps.push(Step::Segment);
277
+ } else if chars.get(end) == Some(&'/') {
278
+ steps.extend([Step::Optional, Step::Any, Step::Literal('/')]);
279
+ end += 1;
280
+ } else {
281
+ steps.push(Step::Any);
282
+ }
283
+ i = end;
284
+ }
285
+ '?' => {
286
+ steps.push(Step::One);
287
+ i += 1;
288
+ }
289
+ ch => {
290
+ steps.push(Step::Literal(ch));
291
+ i += 1;
292
+ }
257
293
  }
258
- if chars[i..].starts_with(&['*', '*']) {
259
- if star_end < chars.len() && chars[star_end] == '/' {
260
- out.push_str("(?:.*/)?");
261
- star_end += 1;
262
- } else {
263
- out.push_str(".*");
294
+ }
295
+ Self { steps }
296
+ }
297
+
298
+ /// Runs every live position through the path at once, so a pattern with several `**` cannot
299
+ /// backtrack exponentially.
300
+ pub(crate) fn is_match(&self, path: &str) -> bool {
301
+ let mut live = vec![false; self.steps.len() + 1];
302
+ self.enter(&mut live, 0);
303
+ for ch in path.chars() {
304
+ let mut next = vec![false; live.len()];
305
+ for (at, step) in self.steps.iter().enumerate() {
306
+ if !live[at] {
307
+ continue;
308
+ }
309
+ match step {
310
+ Step::Literal(expected) if *expected == ch => self.enter(&mut next, at + 1),
311
+ Step::One if ch != '/' => self.enter(&mut next, at + 1),
312
+ Step::Segment if ch != '/' => self.enter(&mut next, at),
313
+ Step::Any if ch != '\n' => self.enter(&mut next, at),
314
+ _ => {}
264
315
  }
265
- } else {
266
- out.push_str("[^/]*");
267
316
  }
268
- i = star_end;
269
- } else if ch == '?' {
270
- out.push_str("[^/]");
271
- i += 1;
272
- } else {
273
- out.push_str(&regex::escape(&ch.to_string()));
274
- i += 1;
317
+ live = next;
318
+ }
319
+ live[self.steps.len()]
320
+ }
321
+
322
+ fn enter(&self, live: &mut [bool], at: usize) {
323
+ if std::mem::replace(&mut live[at], true) {
324
+ return;
325
+ }
326
+ match self.steps.get(at) {
327
+ Some(Step::Segment | Step::Any) => self.enter(live, at + 1),
328
+ Some(Step::Optional) => {
329
+ self.enter(live, at + 1);
330
+ self.enter(live, at + 3);
331
+ }
332
+ _ => {}
275
333
  }
276
334
  }
277
- out.push_str(")$");
278
- // The pattern is built from escaped literals and fixed pieces, so it always compiles.
279
- Regex::new(&out).unwrap_or_else(|_| Regex::new("$^").expect("empty-match regex compiles"))
280
335
  }
281
336
 
282
337
  /// How specific an override pattern is, for most-specific-match-wins: the length of the literal
@@ -296,7 +351,7 @@ mod tests {
296
351
  use super::*;
297
352
 
298
353
  fn matches(pattern: &str, path: &str) -> bool {
299
- glob_to_regex(pattern).is_match(path)
354
+ Glob::new(pattern).is_match(path)
300
355
  }
301
356
 
302
357
  #[test]
@@ -326,6 +381,46 @@ mod tests {
326
381
  assert!(!matches("a.py", "abpy"));
327
382
  }
328
383
 
384
+ #[test]
385
+ fn test_should_match_a_trailing_double_star_directory_only_with_its_separator() {
386
+ assert!(matches("src/**/", "src/a/"));
387
+ assert!(matches("src/**/", "src/"));
388
+ assert!(!matches("src/**/", "src"));
389
+ }
390
+
391
+ #[test]
392
+ fn test_should_treat_three_stars_as_a_double_star() {
393
+ assert!(matches("a/***", "a/b/c.py"));
394
+ assert!(matches("a/***/c.py", "a/c.py"));
395
+ }
396
+
397
+ #[test]
398
+ fn test_should_match_the_whole_path_not_a_part_of_it() {
399
+ assert!(!matches("a", "ab"));
400
+ assert!(!matches("a", "ba"));
401
+ assert!(matches("", ""));
402
+ assert!(!matches("", "a"));
403
+ }
404
+
405
+ #[test]
406
+ fn test_should_not_let_double_star_cross_a_line_break() {
407
+ assert!(!matches("a/**", "a/b\nc"));
408
+ assert!(matches("a/*", "a/b\nc"));
409
+ }
410
+
411
+ #[test]
412
+ fn test_should_count_characters_not_bytes_for_a_question_mark() {
413
+ assert!(matches("src/?.py", "src/\u{e9}.py"));
414
+ }
415
+
416
+ #[test]
417
+ fn test_should_match_many_double_stars_in_linear_time() {
418
+ let pattern = "**/".repeat(40) + "x.py";
419
+ let path = "a/".repeat(200) + "y.py";
420
+ assert!(!matches(&pattern, &path));
421
+ assert!(matches(&pattern, &("a/".repeat(200) + "x.py")));
422
+ }
423
+
329
424
  #[test]
330
425
  fn test_should_rank_a_longer_literal_prefix_as_more_specific() {
331
426
  assert!(pattern_specificity("src/new_module/**") > pattern_specificity("src/**"));
@@ -3,10 +3,9 @@
3
3
  //! Sites and events are the JSON objects the checker produced for each file, tagged with the
4
4
  //! `file` they came from.
5
5
 
6
- use super::config::{glob_to_regex, pattern_specificity, CoverageConfig};
6
+ use super::config::{pattern_specificity, CoverageConfig, Glob};
7
7
  use super::pathutil::relative_posix;
8
8
  use super::pyfmt::{repr_str, round_half_even};
9
- use regex::Regex;
10
9
  use serde_json::{json, Map, Value};
11
10
  use std::collections::{BTreeMap, BTreeSet, HashMap, HashSet};
12
11
  use std::path::Path;
@@ -133,7 +132,7 @@ pub(crate) fn evaluate_coverage(
133
132
  root: &Path,
134
133
  cli_fail_under: Option<f64>,
135
134
  ) -> Vec<Bucket> {
136
- let mut regexes: HashMap<&str, Regex> = HashMap::new();
135
+ let mut globs: HashMap<&str, Glob> = HashMap::new();
137
136
  let mut tallies: Vec<(Option<String>, f64, Tally)> = Vec::new();
138
137
  for (file, tally) in per_file {
139
138
  let (label, threshold) = match cli_fail_under {
@@ -144,9 +143,9 @@ pub(crate) fn evaluate_coverage(
144
143
  .overrides
145
144
  .iter()
146
145
  .filter(|(pattern, _)| {
147
- regexes
146
+ globs
148
147
  .entry(pattern.as_str())
149
- .or_insert_with(|| glob_to_regex(pattern))
148
+ .or_insert_with(|| Glob::new(pattern))
150
149
  .is_match(&relative)
151
150
  })
152
151
  .max_by(|a, b| pattern_specificity(&a.0).cmp(&pattern_specificity(&b.0)));
@@ -6,13 +6,11 @@ use super::collect::collect_files;
6
6
  use super::config::load_configured_excludes;
7
7
  use super::pyfmt::{dumps_indent2, repr_str};
8
8
  use crate::engine::{report_all_batch, report_batch};
9
- use regex::Regex;
10
9
  use serde_json::{json, Value};
11
10
  use std::collections::HashMap;
12
11
  use std::fs;
13
12
  use std::io::{self, Write};
14
13
  use std::path::{Path, PathBuf};
15
- use std::sync::OnceLock;
16
14
 
17
15
  const REVIEW_NOTICE: &str = "<!-- Review before posting, then delete this line. Names, comments and most strings are replaced, but numbers 0 and 1, library and option names and the shape of the code are kept as written. -->";
18
16
 
@@ -34,11 +32,10 @@ fn text<'a>(value: &'a Value, key: &str) -> &'a str {
34
32
 
35
33
  // How many more uses a reason mentions; sites differing only in that are one problem.
36
34
  fn strip_more_uses(reason: &str) -> String {
37
- static PATTERN: OnceLock<Regex> = OnceLock::new();
38
- PATTERN
39
- .get_or_init(|| Regex::new(", and N more$").expect("a fixed pattern compiles"))
40
- .replace(reason, "")
41
- .into_owned()
35
+ reason
36
+ .strip_suffix(", and N more")
37
+ .unwrap_or(reason)
38
+ .to_string()
42
39
  }
43
40
 
44
41
  fn snippet(report: &Value) -> Option<&str> {
@@ -7,9 +7,10 @@
7
7
  //! `untracked-dataframe` nudge is recoverable but a false `unknown-column` error (from a
8
8
  //! short, wrongly-inferred column set) is not.
9
9
 
10
- use sqlparser::ast::{Query, SelectItem, SetExpr, Statement};
10
+ use sqlparser::ast::{Query, SelectItem, SetExpr};
11
11
  use sqlparser::dialect::Dialect;
12
12
  use sqlparser::parser::Parser;
13
+ use sqlparser::tokenizer::Token;
13
14
 
14
15
  /// The grammar every query is parsed with, regardless of `SqlDialect` — see
15
16
  /// [`columns_from_select`]. Identical to [`sqlparser::dialect::GenericDialect`] (its
@@ -423,26 +424,23 @@ pub(crate) fn columns_from_select(sql: &str, dialect: SqlDialect) -> SqlOutcome
423
424
  // identifiers — both perfectly valid SQL text) would otherwise cost real
424
425
  // inferences for no correctness benefit — we only ever read identifier names out
425
426
  // of a successfully parsed `SELECT` list, never validate engine-specific syntax.
426
- let statements =
427
- match Parser::new(&PermissiveDialect)
428
- .try_with_sql(sql)
429
- .and_then(|mut parser| {
430
- parser = parser.with_recursion_limit(50);
431
- parser.parse_statements()
432
- }) {
433
- Ok(stmts) => stmts,
434
- Err(_) => return SqlOutcome::Unparsed,
435
- };
436
-
437
- let [statement] = statements.as_slice() else {
438
- return SqlOutcome::Unparsed;
439
- };
440
-
441
- let Statement::Query(query) = statement else {
442
- return SqlOutcome::Unparsed;
443
- };
427
+ match parse_single_query(sql) {
428
+ Some(query) => columns_from_query(&query, dialect),
429
+ None => SqlOutcome::Unparsed,
430
+ }
431
+ }
444
432
 
445
- columns_from_query(query, dialect)
433
+ /// The one query `sql` holds, with any `;` around it; `None` for anything else. Parsing a query
434
+ /// directly, not a statement list, leaves the statement grammar out of the binary.
435
+ fn parse_single_query(sql: &str) -> Option<Box<Query>> {
436
+ let mut parser = Parser::new(&PermissiveDialect)
437
+ .try_with_sql(sql)
438
+ .ok()?
439
+ .with_recursion_limit(50);
440
+ while parser.consume_token(&Token::SemiColon) {}
441
+ let query = parser.parse_query().ok()?;
442
+ while parser.consume_token(&Token::SemiColon) {}
443
+ (parser.peek_token().token == Token::EOF).then_some(query)
446
444
  }
447
445
 
448
446
  fn columns_from_query(query: &Query, dialect: SqlDialect) -> SqlOutcome {
@@ -533,6 +531,84 @@ mod tests {
533
531
  columns_from_select(sql, dialect)
534
532
  }
535
533
 
534
+ // The statement-list path `parse_single_query` replaced.
535
+ fn via_statements(sql: &str) -> SqlOutcome {
536
+ let parsed = Parser::new(&PermissiveDialect)
537
+ .try_with_sql(sql)
538
+ .and_then(|parser| parser.with_recursion_limit(50).parse_statements());
539
+ match parsed.as_deref() {
540
+ Ok([sqlparser::ast::Statement::Query(query)]) => {
541
+ columns_from_query(query, SqlDialect::Generic)
542
+ }
543
+ _ => SqlOutcome::Unparsed,
544
+ }
545
+ }
546
+
547
+ #[test]
548
+ fn single_query_parse_agrees_with_the_statement_list_parse() {
549
+ let inputs = [
550
+ "SELECT a, b FROM t",
551
+ "SELECT a FROM t;",
552
+ ";SELECT a FROM t",
553
+ ";;SELECT a FROM t;;",
554
+ "SELECT a FROM t; SELECT b FROM u",
555
+ "SELECT a FROM t;; SELECT b FROM u;",
556
+ "SELECT a FROM t SELECT b FROM u",
557
+ "",
558
+ ";",
559
+ ";;",
560
+ "(SELECT a FROM t)",
561
+ "((SELECT a FROM t))",
562
+ "(SELECT a FROM t) UNION (SELECT a FROM u)",
563
+ "SELECT a FROM t UNION SELECT b FROM u",
564
+ "WITH c AS (SELECT a FROM t) SELECT a FROM c",
565
+ "WITH c AS (SELECT a FROM t) INSERT INTO u SELECT a FROM c",
566
+ "VALUES (1, 2)",
567
+ "FROM t SELECT a",
568
+ "TABLE t",
569
+ "INSERT INTO t VALUES (1)",
570
+ "UPDATE t SET a = 1",
571
+ "DELETE FROM t",
572
+ "DROP TABLE t",
573
+ "EXPLAIN SELECT a FROM t",
574
+ "CREATE TABLE t AS SELECT a FROM u",
575
+ "SELECT",
576
+ "SELECT a FROM",
577
+ "SELECT a, FROM t",
578
+ "SELECT * FROM t",
579
+ "SELECT a + 1 FROM t",
580
+ "SELECT 'x;y' AS a FROM t",
581
+ "SELECT a FROM t -- trailing; comment",
582
+ "/* lead */ SELECT a FROM t",
583
+ "SELECT a FROM t WHERE b = ?",
584
+ "SELECT [a b] FROM t",
585
+ "SELECT `a` FROM t;\n",
586
+ "\n\t SELECT a FROM t \n",
587
+ "SELECT a FROM t END",
588
+ "not sql at all",
589
+ ];
590
+ for sql in inputs {
591
+ if sql == "SELECT a FROM t END" {
592
+ // `parse_statements` stops at a stray `END`; a query followed by anything is `Unparsed` here.
593
+ assert_eq!(SqlOutcome::Unparsed, cols(sql), "{sql:?}");
594
+ } else {
595
+ assert_eq!(via_statements(sql), cols(sql), "{sql:?}");
596
+ }
597
+ }
598
+ }
599
+
600
+ #[test]
601
+ fn deeply_nested_input_is_unparsed_not_a_stack_overflow() {
602
+ let sql = format!("SELECT {}1{} AS a", "(".repeat(5000), ")".repeat(5000));
603
+ let outcome = std::thread::Builder::new()
604
+ .stack_size(512 * 1024)
605
+ .spawn(move || cols(&sql))
606
+ .unwrap()
607
+ .join()
608
+ .unwrap();
609
+ assert_eq!(SqlOutcome::Unparsed, outcome);
610
+ }
611
+
536
612
  #[test]
537
613
  fn plain_select_list() {
538
614
  assert_eq!(
@@ -33,7 +33,7 @@ Usage:
33
33
  df.filter(UserSchema.user_id.col > 10) # ✓ Refactor-safe polars expression
34
34
  """
35
35
 
36
- __version__ = "0.14.1"
36
+ __version__ = "0.14.3"
37
37
 
38
38
  from .base_schema import BaseSchema as BaseSchema
39
39
  from .column import Column as Column
@@ -113,3 +113,8 @@ def main() -> None:
113
113
  sys.exit(run_binary(sys.argv[1:]).returncode)
114
114
  binary = str(find_binary())
115
115
  os.execv(binary, [binary, *sys.argv[1:]])
116
+
117
+
118
+ def main_in_process() -> None:
119
+ """Entry point for `typedframes-py`: always the checker inside this process, never the executable."""
120
+ sys.exit(_extension().main(["typedframes", *sys.argv[1:]]))
File without changes
File without changes