Improve compile command handling

codebasin/__init__.py

@@ -1,7 +1,9 @@
 # Copyright (C) 2019-2024 Intel Corporation
 # SPDX-License-Identifier: BSD-3-Clause
+import shlex
 import warnings
 
+import codebasin.source
 import codebasin.walkers
 
 warnings.warn(
@@ -11,3 +13,113 @@
     + "a future release of Code Base Investigator.",
     DeprecationWarning,
 )
+
+
+class CompileCommand:
+    """
+    A single compile command from a compilation database.
+
+    Attributes
+    ----------
+    filename: string
+        The name of the source file compiled by this command.
+
+    directory: string, optional
+        The working directory for this command.
+
+    arguments: list[string], optional
+        The `argv` for this command, including the executable as `argv[0]`.
+
+    output: string, optional
+        The name of the file produced by this command, or None if not
+        specified.
+    """
+
+    def __init__(
+        self,
+        filename,
+        directory=None,
+        arguments=None,
+        command=None,
+        output=None,
+    ):
+        """
+        Raises
+        ------
+        ValueError
+            If both arguments and command are None.
+        """
+        self._filename = filename
+        self._directory = directory
+        if arguments is None and command is None:
+            raise ValueError("CompileCommand requires arguments or command.")
+        self._arguments = arguments
+        self._command = command
+        self._output = output
+
+    @property
+    def directory(self):
+        return self._directory
+
+    @property
+    def filename(self):
+        return self._filename
+
+    @property
+    def arguments(self):
+        if self._arguments is None:
+            return shlex.split(self._command)
+        else:
+            return self._arguments
+
+    @property
+    def output(self):
+        return self._output
+
+    def __str__(self):
+        if self._command is None:
+            return " ".join(self._arguments)
+        else:
+            return self._command
+
+    def is_supported(self):
+        """
+        Returns
+        -------
+        bool
+            True if the command can be emulated and False otherwise.
+            Commands that are not supported will not impact analysis.
+        """
+        # Commands must be non-empty in order to do something.
+        # Commands must operate on source files.
+        if len(self.arguments) > 0 and codebasin.source.is_source_file(
+            self.filename,
+        ):
+            return True
+
+        return False
+
+    @classmethod
+    def from_json(cls, instance: dict):
+        """
+        Parameters
+        ----------
+        instance: dict
+            A JSON object representing a single compile command.
+
+        Returns
+        -------
+        CompileCommand
+            A CompileCommand corresponding to the JSON object.
+        """
+        directory = instance.get("directory", None)
+        arguments = instance.get("arguments", None)
+        command = instance.get("command", None)
+        output = instance.get("output", None)
+        return cls(
+            instance["file"],
+            directory=directory,
+            arguments=arguments,
+            command=command,
+            output=output,
+        )

codebasin/config.py

@@ -11,12 +11,11 @@
 import logging
 import os
 import re
-import shlex
 import warnings
 
 import yaml
 
-from codebasin import util
+from codebasin import CompileCommand, util
 
 log = logging.getLogger("codebasin")
 
@@ -418,11 +417,10 @@ def load_database(dbpath, rootdir):
 
     configuration = []
     for e in db:
-        # Database may not have tokenized arguments
-        if "command" in e:
-            argv = shlex.split(e["command"])
-        elif "arguments" in e:
-            argv = e["arguments"]
+        command = CompileCommand.from_json(e)
+        if not command.is_supported():
+            continue
+        argv = command.arguments
 
         # Extract defines, include paths and include files
         # from command-line arguments
@@ -444,19 +442,19 @@ def load_database(dbpath, rootdir):
         # - relative to a directory
         # - as an absolute path
         filedir = rootdir
-        if "directory" in e:
-            if os.path.isabs(e["directory"]):
-                filedir = e["directory"]
+        if command.directory is not None:
+            if os.path.isabs(command.directory):
+                filedir = command.directory
             else:
                 filedir = os.path.realpath(
                     rootdir,
-                    os.path.join(e["directory"]),
+                    os.path.join(command.directory),
                 )
 
-        if os.path.isabs(e["file"]):
-            path = os.path.realpath(e["file"])
+        if os.path.isabs(command.filename):
+            path = os.path.realpath(command.filename)
         else:
-            path = os.path.realpath(os.path.join(filedir, e["file"]))
+            path = os.path.realpath(os.path.join(filedir, command.filename))
 
         # Compilation database may contain files that don't
         # exist without running make

codebasin/source.py

@@ -0,0 +1,63 @@
+# Copyright (C) 2019-2024 Intel Corporation
+# SPDX-License-Identifier: BSD-3-Clause
+
+import os
+from pathlib import Path
+from typing import Union
+
+
+def is_source_file(filename: Union[str, os.PathLike]) -> bool:
+    """
+    Parameters
+    ----------
+    filename: Union[str, os.Pathlike]
+        The filename of a potential source file.
+
+    Returns
+    -------
+    bool
+        True if the file ends in a recognized extension and False otherwise.
+        Only files that can be parsed correctly have recognized extensions.
+
+    Raises
+    ------
+    TypeError
+        If filename is not a string or Path.
+    """
+    if not (isinstance(filename, str) or isinstance(filename, Path)):
+        raise TypeError("filename must be a string or Path")
+
+    extension = Path(filename).suffix
+    supported_extensions = [
+        ".f90",
+        ".F90",
+        ".f",
+        ".ftn",
+        ".fpp",
+        ".F",
+        ".FOR",
+        ".FTN",
+        ".FPP",
+        ".c",
+        ".h",
+        ".c++",
+        ".cxx",
+        ".cpp",
+        ".cc",
+        ".hpp",
+        ".hxx",
+        ".h++",
+        ".hh",
+        ".inc",
+        ".inl",
+        ".tcc",
+        ".icc",
+        ".ipp",
+        ".cu",
+        ".cuh",
+        ".cl",
+        ".s",
+        ".S",
+        ".asm",
+    ]
+    return extension in supported_extensions

tests/compile-command/__init__.py

@@ -0,0 +1,2 @@
+# Copyright (C) 2019-2024 Intel Corporation
+# SPDX-License-Identifier: BSD-3-Clause

tests/compile-command/test_compile_command.py

@@ -0,0 +1,70 @@
+# Copyright (C) 2019-2024 Intel Corporation
+# SPDX-License-Identifier: BSD-3-Clause
+
+import unittest
+
+from codebasin import CompileCommand
+
+
+class TestCompileCommand(unittest.TestCase):
+    """
+    Test CompileCommand class.
+    """
+
+    def test_commands_and_arguments(self):
+        """Check commands and arguments are not both None"""
+
+        with self.assertRaises(ValueError):
+            CompileCommand("file.cpp", command=None, arguments=None)
+
+        with self.assertRaises(ValueError):
+            instance = {
+                "file": "file.cpp",
+            }
+            CompileCommand.from_json(instance)
+
+    def test_command_to_arguments(self):
+        """Check commands convert to arguments"""
+        command = CompileCommand("file.cpp", command="c++ file.cpp")
+        self.assertEqual(command.arguments, ["c++", "file.cpp"])
+
+        instance = {
+            "file": "file.cpp",
+            "command": "c++ file.cpp",
+        }
+        command = CompileCommand.from_json(instance)
+        self.assertEqual(command.arguments, ["c++", "file.cpp"])
+
+    def test_arguments_to_command(self):
+        """Check arguments convert to command"""
+        command = CompileCommand("file.cpp", arguments=["c++", "file.cpp"])
+        self.assertEqual(str(command), "c++ file.cpp")
+
+        instance = {
+            "file": "file.cpp",
+            "arguments": [
+                "c++",
+                "file.cpp",
+            ],
+        }
+        command = CompileCommand.from_json(instance)
+        self.assertEqual(str(command), "c++ file.cpp")
+
+    def test_empty_command(self):
+        """Check empty commands are not supported"""
+        command = CompileCommand("file.cpp", command="")
+        self.assertFalse(command.is_supported())
+
+    def test_link_command(self):
+        """Check link commands are not supported"""
+        command = CompileCommand("file.o", command="c++ -o a.out file.o")
+        self.assertFalse(command.is_supported())
+
+    def test_valid_command(self):
+        """Check valid commands are supported"""
+        command = CompileCommand("file.cpp", command="c++ file.cpp")
+        self.assertTrue(command.is_supported())
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/source/__init__.py

@@ -0,0 +1,2 @@
+# Copyright (C) 2019-2024 Intel Corporation
+# SPDX-License-Identifier: BSD-3-Clause

tests/source/test_source.py

@@ -0,0 +1,36 @@
+# Copyright (C) 2019-2024 Intel Corporation
+# SPDX-License-Identifier: BSD-3-Clause
+
+import unittest
+from pathlib import Path
+
+import codebasin.source as source
+
+
+class TestSource(unittest.TestCase):
+    """
+    Test functionality in the source module.
+    """
+
+    def test_is_source_file_string(self):
+        """Check source file identification for string filenames"""
+        self.assertTrue(source.is_source_file("file.cpp"))
+        self.assertTrue(source.is_source_file("/path/to/file.cpp"))
+        self.assertFalse(source.is_source_file("file.o"))
+        self.assertFalse(source.is_source_file("/path/to/file.o"))
+
+    def test_is_source_file_path(self):
+        """Check source file identification for Path filenames"""
+        self.assertTrue(source.is_source_file(Path("file.cpp")))
+        self.assertTrue(source.is_source_file(Path("/path/to/file.cpp")))
+        self.assertFalse(source.is_source_file(Path("file.o")))
+        self.assertFalse(source.is_source_file(Path("/path/to/file.o")))
+
+    def test_is_source_types(self):
+        """Check type validation for is_source"""
+        with self.assertRaises(TypeError):
+            source.is_source_file(1)
+
+
+if __name__ == "__main__":
+    unittest.main()