Replace uses of code base dictionary with CodeBase

Signed-off-by: John Pennycook <john.pennycook@intel.com>

Author: Pennycook

bin/codebasin

@@ -10,7 +10,7 @@ import logging
 import os
 import sys
 
-from codebasin import config, finder, report, util
+from codebasin import CodeBase, config, finder, report, util
 from codebasin.walkers.platform_mapper import PlatformMapper
 
 version = "1.2.0"
@@ -107,14 +107,7 @@ def main():
     # Determine the root directory based on where codebasin is run.
     rootdir = os.path.realpath(os.getcwd())
 
-    # Set up a default codebase and configuration object.
-    codebase = {
-        "files": [],
-        "platforms": [],
-        "exclude_files": set(),
-        "exclude_patterns": args.excludes,
-        "rootdir": rootdir,
-    }
+    # Set up a default configuration object.
     configuration = {}
 
     # Load the analysis file if it exists.
@@ -132,8 +125,7 @@ def main():
 
         if "codebase" in analysis_toml:
             if "exclude" in analysis_toml["codebase"]:
-                excludes = analysis_toml["codebase"]["exclude"]
-                codebase["exclude_patterns"] += excludes
+                args.excludes += analysis_toml["codebase"]["exclude"]
 
         for name in args.platforms:
             if name not in analysis_toml["platform"].keys():
@@ -142,16 +134,20 @@ def main():
                     + "does not exist in the configuration file.",
                 )
 
+        cmd_platforms = args.platforms.copy()
         for name in analysis_toml["platform"].keys():
-            if args.platforms and name not in args.platforms:
+            if cmd_platforms and name not in cmd_platforms:
                 continue
             if "commands" not in analysis_toml["platform"][name]:
                 raise ValueError(f"Missing 'commands' for platform {name}")
             p = analysis_toml["platform"][name]["commands"]
             db = config.load_database(p, rootdir)
-            codebase["platforms"].append(name)
+            args.platforms.append(name)
             configuration.update({name: db})
 
+    # Construct a codebase object associated with the root directory.
+    codebase = CodeBase(rootdir, exclude_patterns=args.excludes)
+
     # Parse the source tree, and determine source line associations.
     # The trees and associations are housed in state.
     state = finder.find(
@@ -180,8 +176,7 @@ def main():
     if report_enabled("clustering"):
         basename = os.path.basename(args.analysis_file)
         filename = os.path.splitext(basename)[0]
-        platform_names = [p for p in codebase["platforms"]]
-        output_prefix = "-".join([filename] + platform_names)
+        output_prefix = "-".join([filename] + args.platforms)
 
         clustering_output_name = output_prefix + "-dendrogram.png"
         clustering = report.clustering(clustering_output_name, setmap)

codebasin/finder.py

@@ -142,11 +142,11 @@ def find(
 
     # Build a tree for each unique file for all platforms.
     state = ParserState(summarize_only)
-    for f in codebase["files"]:
+    for f in codebase:
         state.insert_file(f)
     for p in configuration:
         for e in configuration[p]:
-            if e["file"] not in codebase["files"]:
+            if e["file"] not in codebase:
                 filename = e["file"]
                 if legacy_warnings:
                     log.warning(

codebasin/walkers/exporter.py

@@ -6,7 +6,6 @@
 
 from codebasin import util
 from codebasin.preprocessor import CodeNode, FileNode
-from codebasin.walkers.platform_mapper import exclude
 from codebasin.walkers.tree_walker import TreeWalker
 
 log = logging.getLogger("codebasin")
@@ -38,10 +37,7 @@ def walk(self, state):
     def _export_node(self, _filename, _node, _map):
         # Do not export files that the user does not consider to be part of
         # the codebase
-        if isinstance(_node, FileNode) and exclude(
-            _node.filename,
-            self.codebase,
-        ):
+        if isinstance(_node, FileNode) and _node.filename not in self.codebase:
             return
 
         if isinstance(_node, CodeNode):

codebasin/walkers/platform_mapper.py

@@ -2,44 +2,13 @@
 # SPDX-License-Identifier: BSD-3-Clause
 
 import logging
-import os
-
-import pathspec
 
 from codebasin.preprocessor import CodeNode, FileNode
 from codebasin.walkers.tree_mapper import TreeMapper
 
 log = logging.getLogger("codebasin")
 
 
-def exclude(filename, cb):
-    # Always exclude files that were explicitly listed as excluded.
-    if filename in cb["exclude_files"]:
-        log.info(f"Excluding {filename}; matches 'exclude_files'.")
-        return True
-
-    # Only exclude files outside of the root directory if they weren't
-    # explicitly listed as part of the codebase.
-    path = os.path.realpath(filename)
-    if not path.startswith(cb["rootdir"]):
-        if filename in cb["files"]:
-            return False
-        log.info(f"Excluding {filename}; outside of root directory.")
-        return True
-
-    # Exclude files matching an exclude pattern.
-    #
-    # Use GitIgnoreSpec to match git behavior in weird corner cases.
-    # Convert relative paths to match .gitignore subdirectory behavior.
-    spec = pathspec.GitIgnoreSpec.from_lines(cb["exclude_patterns"])
-    rel = os.path.relpath(path, cb["rootdir"])
-    if spec.match_file(rel):
-        log.info(f"Excluding {filename}; matches exclude pattern.")
-        return True
-
-    return False
-
-
 class PlatformMapper(TreeMapper):
     """
     Specific TreeMapper that builds a mapping of nodes to platforms.
@@ -57,10 +26,7 @@ def _map_node(self, _node, _map):
         """
         # Do not map files that the user does not consider to be part of
         # the codebase
-        if isinstance(_node, FileNode) and exclude(
-            _node.filename,
-            self.codebase,
-        ):
+        if isinstance(_node, FileNode) and _node.filename not in self.codebase:
             return
 
         if isinstance(_node, CodeNode):