Improve module_analyzer (#3079)

- Fixed a bug that `direct_deps` wasn't actually returned.
- Added `--exclude-dev-deps` to support calculating without dev
dependencies
- Added `--name-only` flag to only print out module names without
scores.
- Filtered out potential modules parsed that actually don't exist in
BCR. (bazel_dep with override)

tools/README.md

@@ -105,14 +105,16 @@ This script calculates the importance of each module in the BCR based on their P
 The graph is constructed based on dependencies of the latest version of each module.
 
 ```
-usage: module_analyzer.py [-h] [--registry REGISTRY] [--top_n TOP_N]
+usage: module_analyzer.py [-h] [--registry REGISTRY] [--top_n TOP_N] [--exclude-dev-deps] [--name-only]
 
 Select module versions matching given patterns.
 
 optional arguments:
   -h, --help           show this help message and exit
   --registry REGISTRY  Specify the root path of the registry (default: the current working directory or the workspace root if running with Bazel).
   --top_n TOP_N        Specify the top N important modules to print out (default: 50).
+  --exclude-dev-deps   Exclude dev dependencies when constructing the dependency graph (default: False).
+  --name-only          Only print the module names without the scores (default: False).
 ```
 
 You can also run with Bazel, for example:

tools/module_analyzer.py

@@ -35,7 +35,7 @@ def get_buildozer_path():
     return "buildozer"
 
 
-def get_direct_dependencies(module_name, version, registry_dir, buildozer):
+def get_direct_dependencies(module_name, version, registry_dir, buildozer, exclude_dev_deps):
     deps = (
         subprocess.check_output(
             [buildozer, "print name", f"//modules/{module_name}/{version}/MODULE.bazel:%bazel_dep"],
@@ -45,6 +45,9 @@ def get_direct_dependencies(module_name, version, registry_dir, buildozer):
         .split()
     )
 
+    if not exclude_dev_deps:
+        return deps
+
     dev_deps_stat = (
         subprocess.check_output(
             [buildozer, "print dev_dependency", f"//modules/{module_name}/{version}/MODULE.bazel:%bazel_dep"],
@@ -60,7 +63,7 @@ def get_direct_dependencies(module_name, version, registry_dir, buildozer):
         if dev_deps_stat[i] != "True":
             direct_deps.append(dep)
 
-    return deps
+    return direct_deps
 
 
 def main():
@@ -77,6 +80,16 @@ def main():
         default=50,
         help="Specify the top N important modules to print out (default: 50).",
     )
+    parser.add_argument(
+        "--exclude-dev-deps",
+        action="store_true",
+        help="Exclude dev dependencies when constructing the dependency graph (default: False).",
+    )
+    parser.add_argument(
+        "--name-only",
+        action="store_true",
+        help="Only print the module names without the scores (default: False).",
+    )
 
     args = parser.parse_args()
 
@@ -91,14 +104,23 @@ def main():
     G = nx.DiGraph()
     for module in modules:
         module_name, version = module.split("@")
-        for dep in get_direct_dependencies(module_name, version, args.registry, buildozer):
+        for dep in get_direct_dependencies(module_name, version, args.registry, buildozer, args.exclude_dev_deps):
+            # It is possible for a MODULE.bazel to contain a bazel_dep with an override that is not in the registry
+            if not registry.contains(dep):
+                continue
             G.add_edge(module_name, dep)
 
     pagerank = nx.pagerank(G)
 
     sorted_modules = sorted(pagerank.items(), key=lambda x: x[1], reverse=True)
 
     N = min(args.top_n, len(sorted_modules))
+
+    if args.name_only:
+        for module, _ in sorted_modules[:N]:
+            print(module)
+        return
+
     print(f"Top {N} Modules by PageRank:")
     for module, score in sorted_modules[:N]:
         print(f"{module}: {score:.6f}")