From ca223ff3f489f8ce5905c5f4f8d318e4994680b1 Mon Sep 17 00:00:00 2001
From: Ash Garcia <github@garcia.sh>
Date: Mon, 12 Aug 2024 16:59:47 -0700
Subject: [PATCH] make data_dir required & save output to {data_dir}/output by
 default

---
 anonymize_entries.py | 38 ++++++++++++++++++++++++--------------
 1 file changed, 24 insertions(+), 14 deletions(-)

diff --git a/anonymize_entries.py b/anonymize_entries.py
index f6d5107..cfa28e3 100644
--- a/anonymize_entries.py
+++ b/anonymize_entries.py
@@ -31,13 +31,13 @@ from simfile.types import Simfile
 
 
 class AnonymizeEntriesRawArgs:
-    data_dir: str | None
+    data_dir: str
     csv: str | None
     file_uploads: str | None
     deanonymized: bool
     dry_run: bool
     emails: str
-    output: str
+    output: str | None
     regenerate: bool
     seed: str
 
@@ -47,6 +47,7 @@ class AnonymizeEntriesArgs(AnonymizeEntriesRawArgs):
 
     csv: str
     file_uploads: str
+    output: str
 
 
 def argparser():
@@ -74,21 +75,26 @@ def argparser():
     )
     parser.add_argument(
         "data_dir",
-        nargs="?",
         type=str,
-        help="directory containing both the CSV form data and the file responses (uploads)",
+        help="working directory - used to find the form responses CSV, file responses directory, and for output",
     )
     parser.add_argument(
         "-c",
         "--csv",
         type=str,
-        help="override path to the CSV file of form responses",
+        help="override CSV form responses path (defaults to first file matching {data_dir}/*.csv)",
     )
     parser.add_argument(
         "-f",
         "--file-uploads",
         type=str,
-        help="override path to the directory of file responses (uploads)",
+        help="override file responses directory path (defaults to first subdirectory matching {data_dir}/*/*.zip)",
+    )
+    parser.add_argument(
+        "-o",
+        "--output",
+        type=str,
+        help="override output path (defaults to {data_dir}/output)",
     )
     parser.add_argument(
         "-d",
@@ -108,13 +114,6 @@ def argparser():
         type=str,
         help="limit output to files from the specified emails (comma-separated)",
     )
-    parser.add_argument(
-        "-o",
-        "--output",
-        type=str,
-        default="output/",
-        help="output directory",
-    )
     parser.add_argument(
         "-r",
         "--regenerate",
@@ -176,12 +175,22 @@ def process_args(args: AnonymizeEntriesRawArgs) -> AnonymizeEntriesArgs:
         for dir_entry in os.scandir(args.data_dir):
             if not args.csv and dir_entry.is_file() and dir_entry.name.endswith(".csv"):
                 args.csv = dir_entry.path
+                print(
+                    f"Using {repr(dir_entry.name)} for form responses (override with --csv)"
+                )
             if not args.file_uploads and dir_entry.is_dir():
                 if any(
                     subdir_entry.name.endswith(".zip")
                     for subdir_entry in os.scandir(dir_entry.path)
                 ):
                     args.file_uploads = dir_entry.path
+                    print(
+                        f"Using {repr(dir_entry.name)} for file responses (override with --file-uploads)"
+                    )
+
+    if not args.output:
+        args.output = os.path.join(args.data_dir, "output")
+        print(f"Using {args.output} for output (override with --output)")
 
     assert args.csv, "Unable to find a CSV file in the provided directory"
     assert (
@@ -300,7 +309,8 @@ def maybe_mark_resubmitted_entries(
                     row[KnownColumns.IgnoreFile] = "resubmitted" if resubmitted else ""
                     if resubmitted:
                         resubmitted_total += 1
-        print(f"Marked {resubmitted_total} resubmitted files to be ignored")
+        s = "" if resubmitted_total == 1 else "s"
+        print(f"Marked {resubmitted_total} resubmitted file{s} to be ignored")
         return True