Follow symlinks when reading files

Now that we no longer use realpath() when locating files, there is a
possibility that paths will still contain symlinks when we read them.

Signed-off-by: John Pennycook <[email protected]>

codebasin/__init__.py

@@ -191,7 +191,7 @@ def from_file(cls, filename: str | os.PathLike[str]):
         -------
             A CompilationDatbase corresponding to the provided JSON file.
         """
-        with codebasin.util.safe_open_read_nofollow(filename, "r") as f:
+        with open(filename) as f:
             db = codebasin.util._load_json(f, schema_name="compiledb")
         return CompilationDatabase.from_json(db)
 

codebasin/__main__.py

@@ -283,7 +283,7 @@ def main():
             if not os.path.splitext(path)[1] == ".toml":
                 raise RuntimeError(f"Analysis file {path} must end in .toml.")
 
-        with util.safe_open_read_nofollow(path, "rb") as f:
+        with open(path, "rb") as f:
             try:
                 analysis_toml = util._load_toml(f, "analysis")
             except BaseException:

codebasin/config.py

@@ -82,7 +82,7 @@ def load_importcfg():
     path = ".cbi/config"
     if os.path.exists(path):
         log.info(f"Found configuration file at {path}")
-        with util.safe_open_read_nofollow(path, "rb") as f:
+        with open(path, "rb") as f:
             try:
                 _importcfg_toml = util._load_toml(f, "cbiconfig")
                 for name, compiler in _importcfg_toml["compiler"].items():

codebasin/file_parser.py

@@ -8,7 +8,7 @@
 import logging
 import os
 
-from codebasin import preprocessor, util
+from codebasin import preprocessor
 from codebasin.file_source import get_file_source
 
 log = logging.getLogger(__name__)
@@ -170,11 +170,7 @@ def parse_file(self, *, summarize_only=True, language=None):
                 f"{filename} doesn't appear "
                 + "to be a language this tool can process",
             )
-        with util.safe_open_read_nofollow(
-            filename,
-            mode="r",
-            errors="replace",
-        ) as source_file:
+        with open(filename, errors="replace") as source_file:
             groups = {
                 "code": LineGroup(),
                 "directive": LineGroup(),

codebasin/preprocessor.py

@@ -616,7 +616,7 @@ def __init__(self, _filename):
     def __compute_file_hash(self):
         chunk_size = 4096
         hasher = hashlib.sha512()
-        with util.safe_open_read_nofollow(self.filename, "rb") as in_file:
+        with open(self.filename, "rb") as in_file:
             for chunk in iter(lambda: in_file.read(chunk_size), b""):
                 hasher.update(chunk)
 

codebasin/util.py

@@ -75,12 +75,6 @@ def safe_open_write_binary(fname):
     return os.fdopen(fpid, "wb")
 
 
-def safe_open_read_nofollow(fname, *args, **kwargs):
-    """Open fname for reading, but don't follow links."""
-    fpid = os.open(fname, os.O_RDONLY | os.O_NOFOLLOW)
-    return os.fdopen(fpid, *args, **kwargs)
-
-
 def valid_path(path):
     """Return true if the path passed in is valid"""
     valid = True