Fix codespell

Author: ludgerpaehler

llvm_ir_dataset_utils/builders/builder.py

@@ -171,7 +171,7 @@ def parse_and_build_from_description(corpus_description,
       logging.warn('Cargo builder detected no targets from git repository, '
                    'retrying with tar archive.')
       shutil.rmtree(source_dir)
-      # The git repositry is always guaranteed to be the first source as long
+      # The git repository is always guaranteed to be the first source as long
       # as parse_crates_database.py was the source
       corpus_description['sources'].pop(0)
       build_future = get_build_future(corpus_description, source_base_dir,

llvm_ir_dataset_utils/builders/cmake_builder.py

@@ -17,7 +17,7 @@ def generate_configure_command(root_path, options_dict):
   # Add some default flags that are needed for bitcode extraction
   command_vector.append("-DCMAKE_C_COMPILER=clang")
   command_vector.append("-DCMAKE_CXX_COMPILER=clang++")
-  # These two flags assume this is a stanard non-LTO build, will need to fix
+  # These two flags assume this is a standard non-LTO build, will need to fix
   # later when we want to support (Thin)LTO builds.
   command_vector.append("-DCMAKE_C_FLAGS='-Xclang -fembed-bitcode=all'")
   command_vector.append("-DCMAKE_CXX_FLAGS='-Xclang -fembed-bitcode=all'")

llvm_ir_dataset_utils/tools/parse_crates_database.py

@@ -54,7 +54,7 @@ def dedeuplicate_repositories(crates_list):
 
 
 def canonicalize_license(license_string):
-  # Some of the licenses include / as a seperator. This is equivalent to OR
+  # Some of the licenses include / as a separator. This is equivalent to OR
   # within the rust crates index, but not standard in the SPDX format.
   license_string = license_string.replace('/', ' OR ')
   return license_string

pyproject.toml

@@ -54,3 +54,6 @@ line-length = 120
 [tool.ruff.lint]
 select = ["E", "F"]
 ignore = ["F401", "E731"]
+
+[tool.codespell]
+ignore-words-list = "crate,"