findCargoFiles: take source filters into account when searching (#987)

The fact that we search the original source (e.g. avoid IFD and source
renaming quirks) means we effectively ignore filtering (including git
ignore filters) which means we might end up finding (and later parsing)
totally invalid files. This change remedies that.

#985

Author: ipetkov

CHANGELOG.md

@@ -14,6 +14,12 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
   `--show-progress=...` will take precedence). This avoids progress bar
   (re)painting from resulting in garbled build logs
 
+### Fixed
+* `findCargoFiles` (and by extension `mkDummySrc` and `vendorCargoDeps`) will
+  now always respect source filters, meaning any ignored `Cargo.toml` or
+  `.cargo/config.toml` files will no longer be processed and included during
+  processing.
+
 ## [0.23.1] - 2026-02-14
 
 ### Added

checks/default.nix

@@ -438,6 +438,14 @@ onlyDrvs (
         };
       };
 
+      # https://github.com/ipetkov/crane/issues/985
+      filesetIgnoredInvalidCargotoml = myLib.buildPackage {
+        src = lib.fileset.toSource {
+          root = ./simple-with-extra-invalid-cargo-toml;
+          fileset = lib.fileset.difference ./simple-with-extra-invalid-cargo-toml ./simple-with-extra-invalid-cargo-toml/invalid;
+        };
+      };
+
       filesetWorkspace = myLib.buildPackage {
         src = lib.fileset.toSource {
           root = ./workspace;
@@ -599,6 +607,14 @@ onlyDrvs (
         '';
       };
 
+      # https://github.com/ipetkov/crane/issues/985
+      simpleIgnoredInvalidCargotoml = myLib.buildPackage {
+        src = lib.cleanSourceWith {
+          src = ./simple-with-extra-invalid-cargo-toml;
+          filter = path: type: !(type == "directory" && lib.hasSuffix "invalid" path);
+        };
+      };
+
       simpleNoDeps = myLib.buildPackage {
         src = myLib.cleanCargoSource ./simple-no-deps;
         cargoVendorDir = "/dev/null";

checks/simple-with-extra-invalid-cargo-toml/Cargo.lock

@@ -0,0 +1,4 @@
+[package]
+name = "simple"
+version = "0.1.0"
+edition = "2021"

checks/simple-with-extra-invalid-cargo-toml/Cargo.toml

@@ -0,0 +1,3 @@
+{
+    "something": "invalid"
+}

checks/simple-with-extra-invalid-cargo-toml/invalid/Cargo.toml

@@ -0,0 +1 @@
+fn main() {}

checks/simple-with-extra-invalid-cargo-toml/src/main.rs

@@ -1177,6 +1177,12 @@ The default filter applied by `cleanCargoToml`. Currently corresponds to
 Given a path, recursively search it for any `Cargo.toml`, `.cargo/config` or
 `.cargo/config.toml` files.
 
+Note: if `src` is the result of `lib.cleanSource`, `lib.cleanSourceWith`, or
+`lib.fileset.toSource` the embedded filtering logic will be respected (i.e. if
+the result of the cleaned source would ignore certain `Cargo.toml` or
+`.cargo/config.toml` files, they will also be omitted in the results here). The
+resulting files will use paths from the original (unfiltered) source.
+
 ```nix
 craneLib.findCargoFiles ./src
 # { cargoTomls = [ "..." ]; cargoConfigs = [ "..." ]; }

docs/API.md

@@ -2,56 +2,70 @@
   lib,
 }:
 
-src:
 let
   inherit (lib)
     flatten
+    flip
     groupBy
     mapAttrs
     mapAttrsToList
+    optionals
     ;
+  defaultFilter = _: _: true;
 
   # A specialized form of lib.listFilesRecursive except it will only look
   # for Cargo.toml and config.toml files to keep the intermediate results lean
   listFilesRecursive =
-    parentIsDotCargo: dir:
+    filter: parentIsDotCargo: dir:
     flatten (
-      mapAttrsToList (
+      flip mapAttrsToList (builtins.readDir dir) (
         name: type:
         let
-          cur = dir + "/${name}";
+          cur = builtins.unsafeDiscardStringContext ((toString dir) + "/${name}");
+
           isConfig = parentIsDotCargo && (name == "config.toml" || name == "config");
           isCargoToml = name == "Cargo.toml";
         in
-        if type == "directory" then
-          listFilesRecursive (name == ".cargo") cur
-        else if isCargoToml then
-          [
-            {
-              path = cur;
-              type = "cargoTomls";
-            }
-          ]
-        else if isConfig then
-          [
-            {
-              path = cur;
-              type = "cargoConfigs";
-            }
-          ]
-        else
-          [ ]
-      ) (builtins.readDir dir)
+        # NB: manually apply any cleanSourceWith filtering here to avoid surprises
+        # in accidentally finding files that were meant to be excluded
+        # https://github.com/ipetkov/crane/issues/985
+        optionals (filter cur type) (
+          if type == "directory" then
+            listFilesRecursive filter (name == ".cargo") cur
+          else if isCargoToml then
+            [
+              {
+                path = cur;
+                type = "cargoTomls";
+              }
+            ]
+          else if isConfig then
+            [
+              {
+                path = cur;
+                type = "cargoConfigs";
+              }
+            ]
+          else
+            [ ]
+        )
+      )
     );
 
-  foundFiles = listFilesRecursive false src;
-  grouped = groupBy (x: x.type) foundFiles;
-  cleaned = mapAttrs (_: map (y: y.path)) grouped;
-
   # Ensure we have a well typed result
   default = {
     cargoTomls = [ ];
     cargoConfigs = [ ];
   };
 in
+src:
+let
+  isCleanSourceWith = src._isLibCleanSourceWith or false;
+  origSrc = if isCleanSourceWith then src.origSrc or src else src;
+  filter = if isCleanSourceWith then src.filter or defaultFilter else defaultFilter;
+
+  foundFiles = listFilesRecursive filter false origSrc;
+  grouped = groupBy (x: x.type) foundFiles;
+  cleaned = mapAttrs (_: map (y: y.path)) grouped;
+in
 default // cleaned

lib/findCargoFiles.nix

@@ -96,10 +96,12 @@ let
   # whose prefix won't match the paths we observe when we try to clean the source a bit further down
   # (Nix optimizes multiple filters by running them all once against the original source).
   # https://github.com/ipetkov/crane/issues/46
-  origSrc = if src ? _isLibCleanSourceWith then src.origSrc else src;
-
+  origSrc = if src._isLibCleanSourceWith or false then src.origSrc else src;
   uncleanSrcBasePath = builtins.unsafeDiscardStringContext ((toString origSrc) + "/");
-  uncleanFiles = findCargoFiles origSrc;
+  # NB: findCargoFiles will apply any cleanSourceWith filters so we pass in the original
+  # src (not origSrc) here which will give us the best of both worlds
+  # https://github.com/ipetkov/crane/issues/985
+  uncleanFiles = findCargoFiles src;
 
   cargoTomlsBase = uncleanSrcBasePath;
   inherit (uncleanFiles) cargoTomls;

lib/mkDummySrc.nix

@@ -13,10 +13,10 @@ let
 
   inherit (lib.attrsets) optionalAttrs;
 
-  origSrc = src: if src ? _isLibCleanSourceWith then src.origSrc else src;
-
-  cargoConfigs = if args ? src then (findCargoFiles (origSrc args.src)).cargoConfigs else [ ];
+  # NB: findCargoFiles will apply any filtering for us here so we don't need to check origSrc
+  cargoConfigs = if args ? src then (findCargoFiles args.src).cargoConfigs else [ ];
 
+  origSrc = src: if src._isLibCleanSourceWith or false then src.origSrc else src;
   src = origSrc (
     args.src or (throw ''
       unable to find `src` attribute. consider one of the following: