Replace FileLanguage class with a Language enum

- Avoids referencing languages by name (as strings).
- Enables multiple constructors (from_path and from_ext).
- Enables a dedicated sentinel value (UNKNOWN) instead of None.

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

codebasin/file_parser.py

@@ -10,6 +10,7 @@
 
 from codebasin import preprocessor, util
 from codebasin.file_source import get_file_source
+from codebasin.source import Language
 
 log = logging.getLogger("codebasin")
 
@@ -156,7 +157,7 @@ def insert_directive_node(tree, line_group, logical_line):
 
         tree.insert(new_node)
 
-    def parse_file(self, *, summarize_only=True, language=None):
+    def parse_file(self, *, summarize_only=True, language=Language.UNKNOWN):
         """
         Parse the file that this parser points at, build a SourceTree
         representing this file, and return it.

codebasin/file_source.py

@@ -8,7 +8,7 @@
 import itertools as it
 import logging
 
-from codebasin.language import FileLanguage
+from codebasin.source import Language
 
 log = logging.getLogger("codebasin")
 
@@ -698,20 +698,20 @@ def asm_file_source(fp, relaxed=False):
     return (total_sloc, total_physical_lines)
 
 
-def get_file_source(path, assumed_lang=None):
+def get_file_source(path, assumed_lang=Language.UNKNOWN):
     """
     Return a C or Fortran line source for path depending on
     the language we can detect, or fail.
     """
-    lang = FileLanguage(path).get_language()
-    if assumed_lang:
+    lang = Language.from_path(path)
+    if assumed_lang != Language.UNKNOWN:
         lang = assumed_lang
 
-    if lang == "fortran-free":
+    if lang == Language.FORTRAN_FREE:
         return fortran_file_source
-    elif lang in ["c", "c++"]:
+    elif lang in [Language.C, Language.CPLUSPLUS]:
         return c_file_source
-    elif lang in ["asm"]:
+    elif lang in [Language.ASM]:
         return asm_file_source
     else:
         raise RuntimeError(f"Could not determine language of {path}.")

codebasin/finder.py

@@ -10,7 +10,7 @@
 import os
 
 from codebasin import file_parser, platform, preprocessor, util
-from codebasin.language import FileLanguage
+from codebasin.source import Language
 from codebasin.walkers.tree_associator import TreeAssociator
 
 log = logging.getLogger("codebasin")
@@ -84,7 +84,7 @@ def _map_filename(self, fn):
         self.fileinfo[bn].append(FileInfo(fn, size, sha))
         return fn
 
-    def insert_file(self, fn, language=None):
+    def insert_file(self, fn, language=Language.UNKNOWN):
         """
         Build a new tree for a source file, and create an association
         map for it.
@@ -97,10 +97,10 @@ def insert_file(self, fn, language=None):
                 language=language,
             )
             self.maps[fn] = collections.defaultdict(set)
-            if language:
-                self.langs[fn] = language
+            if language == Language.UNKNOWN:
+                self.langs[fn] = Language.from_path(fn)
             else:
-                self.langs[fn] = FileLanguage(fn).get_language()
+                self.langs[fn] = language
 
     def get_filenames(self):
         """

codebasin/language.py

@@ -8,6 +8,7 @@
 
 import logging
 import os
+import warnings
 
 log = logging.getLogger(__name__)
 
@@ -60,6 +61,7 @@ def __init__(self, filename):
             if self._extension in self._language_extensions[lang]:
                 self._language = lang
                 break
+        warnings.warn("FileLanguage is deprecated. Use Language instead.")
 
     def get_language(self):
         return self._language

codebasin/source/__init__.py

@@ -0,0 +1,55 @@
+# Copyright (C) 2019-2024 Intel Corporation
+# SPDX-License-Identifier: BSD-3-Clause
+
+import os
+from enum import Enum, auto
+
+
+class Language(Enum):
+    ASM = auto()
+    C = auto()
+    CPLUSPLUS = auto()
+    FORTRAN_FREE = auto()
+    FORTRAN_MIXED = auto()
+    UNKNOWN = auto()
+
+    @classmethod
+    def from_extension(cls, ext: str) -> "Language":
+        if ext in [".s", ".S", ".asm"]:
+            return cls.ASM
+
+        if ext in [".c", ".h"]:
+            return cls.C
+
+        if ext in [
+            ".c++",
+            ".cxx",
+            ".cpp",
+            ".cc",
+            ".hpp",
+            ".hxx",
+            ".h++",
+            ".hh",
+            ".inc",
+            ".inl",
+            ".tcc",
+            ".icc",
+            ".ipp",
+            ".cu",
+            ".cuh",
+            ".cl",
+        ]:
+            return cls.CPLUSPLUS
+
+        if ext in [".f90", ".F90"]:
+            return cls.FORTRAN_FREE
+
+        if ext in [".f", ".ftn", ".fpp", ".F", ".FOR", ".FTN", ".FPP"]:
+            return cls.FORTRAN_FIXED
+
+        return cls.UNKNOWN
+
+    @classmethod
+    def from_path(cls, path: str) -> "Language":
+        ext = os.path.splitext(path)[1]
+        return cls.from_extension(ext)

setup.py

@@ -11,7 +11,12 @@
     author="John Pennycook",
     author_email="[email protected]",
     url="https://www.github.com/intel/code-base-investigator",
-    packages=["codebasin", "codebasin.schema", "codebasin.walkers"],
+    packages=[
+        "codebasin",
+        "codebasin.source",
+        "codebasin.schema",
+        "codebasin.walkers",
+    ],
     include_package_data=True,
     scripts=["codebasin.py"],
     classifiers=[