Initial integration of dfreport into dfanalyzer

Author: amarathe84

README.md

@@ -1 +1,20 @@
-# dfanalyzer
+# dfanalyzer
+## Report mode
+
+You can now run all of the existing `dfreport.py` reports directly through the main `dfanalyzer` executable:
+
+```bash
+# per-node summary:
+dfanalyzer --report --node /path/to/COMPACT/
+
+# per-process, highlight the max across processes:
+dfanalyzer --report --process --aggregate /path/to/COMPACT/
+
+#### How it works
+
+- **`cli()`** inspects `sys.argv` before Hydra ever sees it.
+- If `--report` is present, we strip it out (so `argparse` in `dfreport.py` still works unchanged) and call `dfreport.main()`.
+- Otherwise we call the original `main()`, so nothing else in your toolchain is disturbed.
+- Finally, by repointing the `dfanalyzer` console script to `cli()`, any `dfanalyzer ...` invocation will first check for `--report`.
+
+This gives you exactly what you asked for `dfanalyzer --report [dfreport options]` with minimal changes to the rest of the repo.

dfanalyzer/__main__.py

@@ -1,3 +1,6 @@
+# File: dfanalyzer/dfanalyzer/__main__.py
+
+import sys
 import hydra
 from distributed import Client
 from hydra.utils import instantiate
@@ -8,9 +11,19 @@
 from .cluster import ExternalCluster
 from .types import Rule
 
-
 init_hydra_config_store()
 
+def cli():
+    """
+    Dispatch between the new --report mode and the usual Hydrab
powered analysis.
+    """
+    if '--report' in sys.argv:
+        # Remove our flag so dfreport.py sees only its own args
+        sys.argv.remove('--report')
+        from .utils.dfreport import main as report_main
+        report_main()
+    else:
+        main()
 
 @hydra.main(version_base=None, config_name="config")
 def main(cfg: Config) -> None:
@@ -21,6 +34,7 @@ def main(cfg: Config) -> None:
             client.restart()
     else:
         client = Client(cluster)
+
     analyzer: AnalyzerType = instantiate(
         cfg.analyzer,
         debug=cfg.debug,
@@ -36,11 +50,13 @@ def main(cfg: Config) -> None:
         unoverlapped_posix_only=cfg.unoverlapped_posix_only,
         view_types=cfg.view_types,
     )
+
     output: OutputType = instantiate(cfg.output)
     output.handle_result(result=result)
+
     client.close()
     cluster.close()  # type: ignore
 
-
 if __name__ == "__main__":
-    main()
+    cli()
+

pyproject.toml

@@ -48,7 +48,7 @@ dependencies = [
 darshan = ["darshan>=3.4"]
 
 [project.scripts]
-dfanalyzer = "dfanalyzer.__main__:main"
+dfanalyzer = "dfanalyzer.__main__:cli"
 dfanalyzer-cluster = "dfanalyzer.cluster:main"
 dfanalyzer-plot = "dfanalyzer.plots:main"