Add codespell support (config, workflow to detect/not fix) and make it fix a typo

.github/workflows/codespell.yml

@@ -0,0 +1,25 @@
+# Codespell configuration is within pyproject.toml
+---
+name: Codespell
+
+on:
+  push:
+    branches: [master]
+  pull_request:
+    branches: [master]
+
+permissions:
+  contents: read
+
+jobs:
+  codespell:
+    name: Check for spelling errors
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Annotate locations with typos
+        uses: codespell-project/codespell-problem-matcher@v1
+      - name: Codespell
+        uses: codespell-project/actions-codespell@v2

bazelisk.py

@@ -177,7 +177,7 @@ def get_version_history(bazelisk_directory):
             if not release["prerelease"]
         ),
         # This only handles versions with numeric components, but that is fine
-        # since prerelease verisons have been excluded.
+        # since prerelease versions have been excluded.
         key=lambda version: tuple(int(component)
                                   for component in version.split('.')),
         reverse=True,

docs/conf.py

@@ -363,7 +363,7 @@ def _python_type_to_xref(
         'desc': 'C++ class',
     },
     'absl::StatusOr': {
-        'url': 'https://abseil.io/docs/cpp/guides/statuss#returning-a-status-or-a-value',
+        'url': 'https://abseil.io/docs/cpp/guides/status#returning-a-status-or-a-value',
         'object_type': 'class',
         'desc': 'C++ class',
     },

docs/doctest_test.py

@@ -16,7 +16,7 @@
 This allows doctest examples to be conveniently updated in case of changes to
 the output format.
 
-After commiting or staging changes, you can run this with the `--in-place`
+After committing or staging changes, you can run this with the `--in-place`
 option and then inspect the diff.
 
 This supports top-level `await` in tests, since that provides a convenient way
@@ -327,7 +327,7 @@ def _ast_asyncify(code: str, wrapper_name: str) -> ast.Module:
   This is derived from a similar workaround in IPython.
 
   Args:
-    code: Python source code to pase.
+    code: Python source code to pass.
     wrapper_name: Name to use for function.
 
   Returns:

docs/environment.rst

@@ -96,7 +96,7 @@ Debugging
    Enables debug logging for tensorstore internal subsystems.  Set to comma
    separated list of values, where each value is one of ``name=int`` or just
    ``name``. When ``all`` is, present, then verbose logging will be enabled for
-   each subsytem, otherwise logging is set only for those subsystems present in
+   each subsystem, otherwise logging is set only for those subsystems present in
    the list.
 
    Verbose flag values include: ``curl``, ``distributed``, ``file``,

docs/installation.rst

@@ -264,7 +264,7 @@ The following CMake generators are supported:
 - Ninja and Ninja Multi-Config
 - Makefile generators
 - Visual Studio generators
-- Xcode (targetting arm64 only)
+- Xcode (targeting arm64 only)
 
 The Ninja generator is recommended because it provides the fastest builds.
 

docs/python/indexing.rst

@@ -423,7 +423,7 @@ Ellipsis
 
 Specifying the special `Ellipsis` value (:python:`...`) is equivalent
 to specifying as many full slices :python:`:` as needed to consume the
-remaining dimensions of the original domin not consumed by other
+remaining dimensions of the original domain not consumed by other
 indexing terms:
 
 .. doctest::

docs/schema_schema.yml

@@ -80,7 +80,7 @@ definitions:
       - A numerical :literal:`multiplier`, represented as a `double-precision
         floating-point number
         <https://en.wikipedia.org/wiki/Double-precision_floating-point_format>`_.
-        A multiplier of ``1`` may be used to indicate a quanity equal to a
+        A multiplier of ``1`` may be used to indicate a quantity equal to a
         single base unit.
 
       - A :literal:`base_unit`, represented as a string.  An empty string may be used

examples/image_convolution.cc

@@ -34,7 +34,7 @@ tensorstore::SharedArray<int, 2> ApplyKernel(
     const tensorstore::ArrayView<const int, 2> in,
     const tensorstore::ArrayView<const double, 2> kernel) {
   // Compute bounds for the offset.
-  // FIXME: It's akward that we cannot do this:
+  // FIXME: It's awkward that we cannot do this:
   //   std::array<Index, 2> k(kernel.shape());
   //
   std::array<Index, 2> k;
@@ -80,7 +80,7 @@ tensorstore::SharedArray<int, 2> ApplyKernel(
               }
             });
 
-        // Again, the most intutive way to write an array value
+        // Again, the most intuitive way to write an array value
         // is not permitted:
         //
         //   dest[indices] = (sum / count);  // error: no viable overloaded '='
@@ -154,7 +154,7 @@ void AffineWarpGrid(size_t xmax, size_t ymax,
 }
 
 // AffineWarpInverseGrid computes the inverse mapping from AffineWarpGrid,
-// so it can be used to map from a destination image to a souce image.
+// so it can be used to map from a destination image to a source image.
 void AffineWarpInverseGrid(size_t xmax, size_t ymax,
                            tensorstore::span<const double, 6> M,
                            AffineWarpGridFunction fn) {
@@ -237,7 +237,7 @@ void PrintCSVArray(tensorstore::ArrayView<T, N> data) {
   // reference for every element.
   //
   // There is a streaming operator already, but the output is
-  // this is equvalent to:
+  // this is equivalent to:
   // for (int x = 0; x < data.shape()[0]; x++)
   //  for (int y = 0; y < data.shape()[1]; y++) {
   //     ... body ...

pyproject.toml

@@ -39,3 +39,11 @@ version_scheme = "no-guess-dev"
 # Test PyPI does not support local versions.
 local_scheme = "no-local-version"
 fallback_version = "0.0.0"
+
+[tool.codespell]
+# Ref: https://github.com/codespell-project/codespell#using-a-config-file
+skip = '.git*,third_party'
+check-hidden = true
+# Do not bother with mixed case words -- variables, and test lines
+ignore-regex = '\b[a-zA-Z]+[A-Z][a-z]*\b|\bEXPECT_.*'
+ignore-words-list = 'ehr,ans'

python/tensorstore/array_type_caster.cc

@@ -277,7 +277,7 @@ pybind11::object GetNumpyArrayImpl(SharedArrayView<const void> value,
   auto obj = py::reinterpret_steal<py::array>(PyArray_NewFromDescr(
       /*subtype=*/&PyArray_Type,
       /*descr=*/reinterpret_cast<PyArray_Descr*>(py_dtype.release().ptr()),
-      /*nd=*/static_cast<int>(value.rank()),
+      /*nd=*/static_cast<int>(value.rank()),  // codespell:ignore nd
       /*dims=*/shape,
       /*strides=*/strides,
       /*data=*/const_cast<void*>(value.data()), flags, nullptr));

python/tensorstore/keyword_arguments.h

@@ -23,7 +23,7 @@
 /// This mechanism allows individual keyword arguments to be defined as
 /// `ParamDef` types, that specifies the name, documentation, argument type, and
 /// operation to perform to "apply" the argument.  These `ParamDef` types can
-/// then be re-used by multiple pybind11 functions while avoiding duplication.
+/// then be reused by multiple pybind11 functions while avoiding duplication.
 ///
 /// Each `ParamDef` type should be a struct with the following members, and no
 /// non-static data members (i.e. it should be empty):

python/tensorstore/keyword_arguments_test.cc

@@ -88,7 +88,7 @@ Does something or other with keyword arguments.
     //     Args:
     //       required_arg: This is required
     //
-    //       a: Specifies a.  This documentaiton string is allowed
+    //       a: Specifies a.  This documentation string is allowed
     //         to be more than one line.
     //       b: Specifies b.
     //

python/tensorstore/tests/exit_test.py

@@ -37,7 +37,7 @@ def run_during_finalization():
     async def test_read():
       t = ts.array([1, 2, 3], dtype=ts.int64)
       await asyncio.wait_for(t.read(), timeout=1)
-      # Normally, await won't suceed.  However, await may still succeed, if it
+      # Normally, await won't succeed.  However, await may still succeed, if it
       # happens that the read completed before the call to `await`.
       os._exit(0)
 

python/tensorstore/unit.cc

@@ -54,7 +54,7 @@ The quantity is specified as the combination of:
 
 - A numerical :py:obj:`.multiplier`, represented as a
   `double-precision floating-point number <https://en.wikipedia.org/wiki/Double-precision_floating-point_format>`_.
-  A multiplier of :python:`1` may be used to indicate a quanity equal to a
+  A multiplier of :python:`1` may be used to indicate a quantity equal to a
   single base unit.
 
 - A :py:obj:`.base_unit`, represented as a string.  An empty string may be used

setup.py

@@ -160,7 +160,7 @@ def _get_action_env():
   Unfortunately there is no kosher way to detect the PEP517 build environment,
   so this is a heuristic approach based on inspection of the PATH variable
   passed to the build and review of the PIP sources. The source, as reviewed,
-  creates a temporary directory inclusing pip-{kind}, where kind=build-env.
+  creates a temporary directory including pip-{kind}, where kind=build-env.
 
   See: dist-packages/pip/_internal/utils/temp_dir.py
   Also: https://github.com/bazelbuild/bazel/issues/18809
@@ -178,7 +178,7 @@ def _get_action_env():
   if not build_env:
     return []
 
-  # There may be mutliple path entries added under the build-env directory,
+  # There may be multiple path entries added under the build-env directory,
   # so remove them all.
   while 'pip-build-env' in os.path.dirname(build_env):
     build_env = os.path.dirname(build_env)
@@ -216,7 +216,7 @@ def run(self):
         # from the PATH as bazel is already hermetic to improve cache use.
         action_env = _get_action_env()
 
-        # Ensure python_configure.bzl finds the correct Python verison.
+        # Ensure python_configure.bzl finds the correct Python version.
         os.environ['PYTHON_BIN_PATH'] = sys.executable
 
         # Ensure it is built against the version of `numpy` in the current

tensorstore/context.cc

@@ -392,7 +392,7 @@ class ResourceReference : public ResourceSpecImplBase {
   ResourceSpecImplPtr UnbindContext(
       const internal::ContextSpecBuilder& spec_builder) final {
     auto& builder_impl = *internal_context::Access::impl(spec_builder);
-    // Ensure the referent is not re-used as an identifier for another resource.
+    // Ensure the referent is not reused as an identifier for another resource.
     ++builder_impl.ids_[referent_];
     return ResourceSpecImplPtr(this);
   }

tensorstore/contiguous_layout.h

@@ -148,7 +148,7 @@ bool PermutationMatchesOrder(span<const DimensionIndex> permutation,
 void InvertPermutation(DimensionIndex rank, const DimensionIndex* perm,
                        DimensionIndex* inverse_perm);
 
-/// Normalizes `source` to a permutation if it is not already a permuation.
+/// Normalizes `source` to a permutation if it is not already a permutation.
 ///
 /// \relates ContiguousLayoutPermutation
 template <typename LayoutOrder, DimensionIndex Rank>

tensorstore/driver/downsample/downsample_util.h

@@ -61,8 +61,8 @@ struct PropagatedIndexTransformDownsampling {
 ///
 /// is equivalent to:
 ///
-///   transforming `b` by `propgated.transform` and then downsampling by
-///   `propgated.input_downsample_factors` (and possibly "squeezing" some
+///   transforming `b` by `propagated.transform` and then downsampling by
+///   `propagated.input_downsample_factors` (and possibly "squeezing" some
 ///   singleton dimensions that were added).
 ///
 /// Note that this function assumes downsampling is performed using a method

tensorstore/driver/kvs_backed_chunk_driver.h

@@ -853,7 +853,7 @@ class OpenState : public MetadataOpenState {
 ///    is compatible with the open request by calling
 ///    `open_state->GetComponentIndex`.
 ///
-///    - If it is, either re-uses an existing `DataCache` with a cache key that
+///    - If it is, either reuses an existing `DataCache` with a cache key that
 ///      matches `open_state->GetDataCacheKey`, or obtain a new `DataCache` from
 ///      `open_state->GetDataCache`.
 ///

tensorstore/driver/n5/schema.yml

@@ -9,7 +9,7 @@ allOf:
     metadata:
       title: N5 array metadata.
       description: |
-        Specifies constraints on the metdata of a dataset exactly as in the
+        Specifies constraints on the metadata of a dataset exactly as in the
         `attributes.json file
         <https://github.com/saalfeldlab/n5#file-system-specification-version-203-snapshot>`_,
         except that all members are optional.  When creating a new array, the

tensorstore/driver/neuroglancer_precomputed/metadata.cc

@@ -1481,7 +1481,7 @@ bool GetShardChunkHierarchy(const ShardingSpec& sharding_spec,
                             ShardChunkHierarchy& hierarchy) {
   if (sharding_spec.hash_function != ShardingSpec::HashFunction::identity) {
     // For non-identity hash functions, the number of chunks per shard is not
-    // predicable and the shard doesn't correspond to a rectangular region
+    // predictable and the shard doesn't correspond to a rectangular region
     // anyway.
     return false;
   }

tensorstore/index_interval_test.cc

@@ -387,7 +387,7 @@ TEST(IndexIntervalTest, IntersectPreferringExplicit) {
           OIII{IndexInterval::UncheckedClosed(0, 10), false, false}),
       ::testing::Eq(OIII{IndexInterval::UncheckedClosed(0, 10), false, false}));
 
-  // These may surprise you!  explicit takes prededence over implicit!
+  // These may surprise you!  explicit takes precedence over implicit!
   EXPECT_THAT(
       IntersectPreferringExplicit(
           OIII{IndexInterval::UncheckedClosed(-kInfIndex, kMaxFiniteIndex),

tensorstore/index_space/internal/numpy_indexing_spec.cc

@@ -281,7 +281,7 @@ void GetIndexedInputDims(const NumpyIndexingSpec& spec,
   }
   input_dims_per_intermediate_dim[intermediate_rank] = input_dim;
 
-  // Compute `indexed_input_dims` by reodering `input_dims_per_intermediate_dim`
+  // Compute `indexed_input_dims` by reordering `input_dims_per_intermediate_dim`
   // by `selected_dims`.
   for (const DimensionIndex intermediate_dim : selected_dims) {
     for (DimensionIndex

tensorstore/index_space/internal/propagate_bounds.cc

@@ -92,7 +92,7 @@ absl::Status PropagateBoundsImpl(BoxView<> b,
   for (DimensionIndex b_dim = 0; b_dim < b.rank(); ++b_dim) {
     auto& map = maps[b_dim];
     const Index output_stride = map.stride();
-    // We dont't validate or propagate bounds for array-based output index maps.
+    // We don't validate or propagate bounds for array-based output index maps.
     if (map.method() == OutputIndexMethod::array) continue;
     OptionallyImplicitIndexInterval b_bounds_oi{b[b_dim],
                                                 b_implicit_lower_bounds[b_dim],

tensorstore/index_space/internal/transform_rep.h

@@ -368,8 +368,8 @@ inline void NormalizeImplicitBounds(TransformRep& rep) {
 // Check that OutputIndexMap and std::string don't have a greater alignment
 // value than Index, as that would require more complicated logic for accessing
 // the variable length fields than is currently implemented.  In practice these
-// constraints should always be satisified.  If this code needs to work on a
-// platform that doesn't satisfy these contraints, the more complicated logic
+// constraints should always be satisfied.  If this code needs to work on a
+// platform that doesn't satisfy these constraints, the more complicated logic
 // could be implemented.
 static_assert(alignof(OutputIndexMap) <= sizeof(Index),
               "Platform has unsupported alignment.");

tensorstore/internal/cache/async_cache.h

@@ -378,7 +378,7 @@ class AsyncCache : public Cache {
     /// \param transaction[in,out] Transaction associated with the entry.  If
     ///     non-null, must specify an explicit transaction, and an associated
     ///     transaction node will be created if one does not already exist.  In
-    ///     this case, the `tranaction` pointer itself will not be modified.  An
+    ///     this case, the `transaction` pointer itself will not be modified.  An
     ///     implicit transaction node associated with a new implicit transaction
     ///     is requested by specifying `transaction` initially equally to
     ///     `nullptr`.  Upon return, `transaction` will hold an open transaction

tensorstore/internal/cache/cache_test.cc

@@ -70,7 +70,7 @@ class TestCache : public Cache {
     std::deque<std::string> entry_allocate_log;
     // Log of calls to DoDeleteEntry.  Contains the cache key and entry key.
     std::deque<std::pair<std::string, std::string>> entry_destroy_log;
-    // Log of calls to GetTestCache (defined below).  ontains the cache key.
+    // Log of calls to GetTestCache (defined below).  contains the cache key.
     std::deque<std::string> cache_allocate_log;
     // Log of calls to TestCache destructor.  Contains the cache key.
     std::deque<std::string> cache_destroy_log;

tensorstore/internal/container/compressed_tuple_test.cc

@@ -308,7 +308,7 @@ TEST(CompressedTupleTest, Nested) {
   std::set<Empty<0>*> empties{&y.get<0>(), &y.get<1>(), &y.get<2>().get<0>(),
                               &y.get<2>().get<1>().get<0>()};
 #ifdef _MSC_VER
-  // MSVC has a bug where many instances of the same base class are layed out in
+  // MSVC has a bug where many instances of the same base class are laid out in
   // the same address when using __declspec(empty_bases).
   // This will be fixed in a future version of MSVC.
   int expected = 1;

tensorstore/internal/curl/curl_factory.h

@@ -23,7 +23,7 @@ namespace internal_http {
 /// CurlHandleFactory creates and cleans up CURL* (CurlPtr) handles
 /// and CURLM* (CurlMulti) handles.
 ///
-/// NOTE: These methods are virtual so that a curl factory can re-use
+/// NOTE: These methods are virtual so that a curl factory can reuse
 /// curl handles.
 class CurlHandleFactory {
  public:

tensorstore/internal/file_io_concurrency_resource.cc

@@ -27,7 +27,7 @@ namespace {
 struct FileIoConcurrencyResourceTraits
     : public ConcurrencyResourceTraits,
       public ContextResourceTraits<FileIoConcurrencyResource> {
-  // TODO(jbms): use beter method of picking concurrency limit
+  // TODO(jbms): use better method of picking concurrency limit
   FileIoConcurrencyResourceTraits()
       : ConcurrencyResourceTraits(
             std::max(size_t(4), size_t(std::thread::hardware_concurrency()))) {}

tensorstore/internal/image/image_reader_test.cc

@@ -207,7 +207,7 @@ TEST_P(ReaderTest, ReadImageTruncated) {
 }
 
 // Most images generated via tiffcp <source> <options> <dest>.
-// Query image paramters using tiffinfo <image>
+// Query image parameters using tiffinfo <image>
 std ::vector<V> GetD75_08_Values() {
   return {
       // upper-left corner: hw=0,0   => 151,75,83

tensorstore/internal/image/image_writer_test.cc

@@ -234,7 +234,7 @@ TEST_P(WriterTest, RoundTrip) {
 
   double rmse = ComputeRMSE(decoded.data(), source.data(), source.size());
 
-  /// When RMSE is not 0, verify that the actual value is witin 5%.
+  /// When RMSE is not 0, verify that the actual value is within 5%.
   if (GetParam().rmse_error_limit == 0) {
     EXPECT_EQ(0, rmse) << "\nA: " << source_info << " "
                        << "\nB: " << decoded_info;

tensorstore/internal/json_binding/json_binding.h

@@ -484,7 +484,7 @@ constexpr auto Projection(Proj projection, Binder binder = DefaultBinder<>) {
   };
 }
 
-/// Binder adapter that projects the parsed representation using gettter/setter
+/// Binder adapter that projects the parsed representation using getter/setter
 /// functions.
 ///
 /// Commonly this is used with `Member`, in order to bind a

tensorstore/internal/riegeli/find.cc

@@ -34,7 +34,7 @@ bool StartsWith(riegeli::Reader &reader, std::string_view needle) {
          memcmp(reader.cursor(), needle.data(), needle.size()) == 0;
 }
 
-/// Seeks for the first occurence of data string starting from the current pos.
+/// Seeks for the first occurrence of data string starting from the current pos.
 /// This works well enough for ZIP archives, since the tags do not have
 /// internal repetition.
 bool FindFirst(riegeli::Reader &reader, std::string_view needle) {