Add support for glob patterns in `src` (#1225)

Cargo.lock

@@ -796,6 +796,12 @@ dependencies = [
  "wasm-bindgen",
 ]
 
+[[package]]
+name = "glob"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9b919933a397b79c37e33b77bb2aa3dc8eb6e165ad809e58ff75bc7db2e34574"
+
 [[package]]
 name = "globset"
 version = "0.4.9"
@@ -1841,6 +1847,7 @@ dependencies = [
  "fern",
  "filetime",
  "getrandom 0.2.8",
+ "glob",
  "globset",
  "insta",
  "itertools",

Cargo.toml

@@ -28,6 +28,7 @@ common-path = { version = "1.0.0" }
 dirs = { version = "4.0.0" }
 fern = { version = "0.6.1" }
 filetime = { version = "0.2.17" }
+glob = { version = "0.3.0" }
 globset = { version = "0.4.9" }
 itertools = { version = "0.10.5" }
 libcst = { git = "https://github.com/charliermarsh/LibCST", rev = "f2f0b7a487a8725d161fe8b3ed73a6758b21e177" }

README.md

@@ -1744,6 +1744,25 @@ show-source = true
 
 The source code paths to consider, e.g., when resolving first- vs. third-party imports.
 
+As an example: given a Python package structure like:
+
+```text
+my_package/
+  pyproject.toml
+  src/
+    my_package/
+      __init__.py
+      foo.py
+      bar.py
+```
+
+The `src` directory should be included in `source` (e.g., `source = ["src"]`), such that
+when resolving imports, `my_package.foo` is considered a first-party import.
+
+This field supports globs. For example, if you have a series of Python packages in
+a `python_modules` directory, `src = ["python_modules/*"]` would expand to incorporate
+all of the packages in that directory.
+
 **Default value**: `["."]`
 
 **Type**: `Vec<PathBuf>`

resources/test/project/README.md

@@ -9,24 +9,26 @@ Running from the repo root should pick up and enforce the appropriate settings f
 
 ```
 ∴ cargo run resources/test/project/
-Found 4 error(s).
+Found 5 error(s).
 resources/test/project/examples/docs/docs/file.py:1:1: I001 Import block is un-sorted or un-formatted
 resources/test/project/examples/docs/docs/file.py:8:5: F841 Local variable `x` is assigned to but never used
 resources/test/project/src/file.py:1:8: F401 `os` imported but unused
 resources/test/project/src/file.py:5:5: F841 Local variable `x` is assigned to but never used
-2 potentially fixable with the --fix option.
+resources/test/project/src/import_file.py:1:1: I001 Import block is un-sorted or un-formatted
+3 potentially fixable with the --fix option.
 ```
 
 Running from the project directory itself should exhibit the same behavior:
 
 ```
 ∴ cd resources/test/project/ && cargo run .
-Found 4 error(s).
+Found 5 error(s).
 examples/docs/docs/file.py:1:1: I001 Import block is un-sorted or un-formatted
 examples/docs/docs/file.py:8:5: F841 Local variable `x` is assigned to but never used
 src/file.py:1:8: F401 `os` imported but unused
 src/file.py:5:5: F841 Local variable `x` is assigned to but never used
-2 potentially fixable with the --fix option.
+src/import_file.py:1:1: I001 Import block is un-sorted or un-formatted
+3 potentially fixable with the --fix option.
 ```
 
 Running from the sub-package directory should exhibit the same behavior, but omit the top-level
@@ -45,7 +47,7 @@ file paths from the current working directory:
 
 ```
 ∴ cargo run -- --config=resources/test/project/pyproject.toml resources/test/project/
-Found 8 error(s).
+Found 9 error(s).
 resources/test/project/examples/docs/docs/concepts/file.py:1:8: F401 `os` imported but unused
 resources/test/project/examples/docs/docs/concepts/file.py:5:5: F841 Local variable `x` is assigned to but never used
 resources/test/project/examples/docs/docs/file.py:1:8: F401 `os` imported but unused
@@ -54,7 +56,8 @@ resources/test/project/examples/docs/docs/file.py:4:27: F401 `docs.concepts.file
 resources/test/project/examples/docs/docs/file.py:8:5: F841 Local variable `x` is assigned to but never used
 resources/test/project/src/file.py:1:8: F401 `os` imported but unused
 resources/test/project/src/file.py:5:5: F841 Local variable `x` is assigned to but never used
-5 potentially fixable with the --fix option.
+resources/test/project/src/import_file.py:1:1: I001 Import block is un-sorted or un-formatted
+6 potentially fixable with the --fix option.
 ```
 
 Running from a parent directory should this "ignore" the `exclude` (hence, `concepts/file.py` gets

resources/test/project/pyproject.toml

@@ -1,2 +1,3 @@
 [tool.ruff]
-src = ["."]
+src = [".", "python_modules/*"]
+extend-select = ["I001"]

resources/test/project/src/import_file.py

@@ -0,0 +1,7 @@
+import numpy as np
+from app import app_file
+from core import core_file
+
+np.array([1, 2, 3])
+app_file()
+core_file()

src/settings/configuration.rs

@@ -5,6 +5,7 @@
 use std::path::{Path, PathBuf};
 
 use anyhow::{anyhow, Result};
+use glob::{glob, GlobError, Paths, PatternError};
 use regex::Regex;
 
 use crate::checks_gen::CheckCodePrefix;
@@ -65,11 +66,10 @@ impl Configuration {
                 .map(|pattern| Regex::new(&pattern))
                 .transpose()
                 .map_err(|e| anyhow!("Invalid `dummy-variable-rgx` value: {e}"))?,
-            src: options.src.map(|src| {
-                src.iter()
-                    .map(|path| fs::normalize_path_to(Path::new(path), project_root))
-                    .collect()
-            }),
+            src: options
+                .src
+                .map(|src| resolve_src(&src, project_root))
+                .transpose()?,
             target_version: options.target_version,
             exclude: options.exclude.map(|paths| {
                 paths
@@ -216,3 +216,19 @@ impl Configuration {
         }
     }
 }
+
+/// Given a list of source paths, which could include glob patterns, resolve the
+/// matching paths.
+pub fn resolve_src(src: &[String], project_root: &Path) -> Result<Vec<PathBuf>> {
+    let globs = src
+        .iter()
+        .map(Path::new)
+        .map(|path| fs::normalize_path_to(path, project_root))
+        .map(|path| glob(&path.to_string_lossy()))
+        .collect::<Result<Vec<Paths>, PatternError>>()?;
+    let paths: Vec<PathBuf> = globs
+        .into_iter()
+        .flatten()
+        .collect::<Result<Vec<PathBuf>, GlobError>>()?;
+    Ok(paths)
+}

src/settings/mod.rs

@@ -19,7 +19,7 @@ use crate::settings::configuration::Configuration;
 use crate::settings::types::{FilePattern, PerFileIgnore, PythonVersion, SerializationFormat};
 use crate::{
     flake8_annotations, flake8_bugbear, flake8_import_conventions, flake8_quotes,
-    flake8_tidy_imports, fs, isort, mccabe, pep8_naming, pyupgrade,
+    flake8_tidy_imports, isort, mccabe, pep8_naming, pyupgrade,
 };
 
 pub mod configuration;
@@ -296,12 +296,6 @@ pub fn resolve_per_file_ignores(
         .collect()
 }
 
-pub fn resolve_src(src: Vec<PathBuf>, project_root: &Path) -> Vec<PathBuf> {
-    src.into_iter()
-        .map(|path| fs::normalize_path_to(&path, project_root))
-        .collect()
-}
-
 /// Given a set of selected and ignored prefixes, resolve the set of enabled
 /// error codes.
 fn resolve_codes(select: &[CheckCodePrefix], ignore: &[CheckCodePrefix]) -> FxHashSet<CheckCode> {

src/settings/options.rs

@@ -235,8 +235,28 @@ pub struct Options {
     )]
     pub show_source: Option<bool>,
     #[option(
-        doc = "The source code paths to consider, e.g., when resolving first- vs. third-party \
-               imports.",
+        doc = r#"
+            The source code paths to consider, e.g., when resolving first- vs. third-party imports.
+
+            As an example: given a Python package structure like:
+
+            ```text
+            my_package/
+              pyproject.toml
+              src/
+                my_package/
+                  __init__.py
+                  foo.py
+                  bar.py
+            ```
+
+            The `src` directory should be included in `source` (e.g., `source = ["src"]`), such that
+            when resolving imports, `my_package.foo` is considered a first-party import.
+
+            This field supports globs. For example, if you have a series of Python packages in
+            a `python_modules` directory, `src = ["python_modules/*"]` would expand to incorporate
+            all of the packages in that directory.
+            "#,
         default = r#"["."]"#,
         value_type = "Vec<PathBuf>",
         example = r#"