Updated codebase to include more logging and also accessing large email directories better.

tests/test_cli.py

@@ -1,8 +1,10 @@
+from contextlib import redirect_stderr, redirect_stdout
+from io import StringIO
 from pathlib import Path
 import tempfile
 import unittest
 
-from rspamd_mail_trainer.cli import SearchOptions, iter_candidates, maybe_compile, parse_user_date
+from rspamd_mail_trainer.cli import ScanStats, SearchOptions, iter_candidates, main, maybe_compile, parse_user_date
 
 
 def write_mail(path: Path, sender: str, subject: str, body: str = "hello") -> None:
@@ -41,6 +43,89 @@ class CliTests(unittest.TestCase):
             self.assertEqual(len(matches), 1)
             self.assertEqual(matches[0][0].sender, "no-reply@example.test")
 
+    def test_iter_candidates_populates_scan_stats(self) -> None:
+        with tempfile.TemporaryDirectory() as td:
+            root = Path(td) / "Maildir"
+            cur = root / "cur"
+            cur.mkdir(parents=True)
+            write_mail(cur / "1:2,S", "no-reply@example.test", "Monthly report")
+            opts = SearchOptions(
+                from_re=maybe_compile("example.test", True, True),
+                to_re=None,
+                subject_re=None,
+                any_re=None,
+                after=None,
+                before=None,
+                max_size=1024 * 1024,
+                include_seen_duplicates=False,
+            )
+            stats = ScanStats()
+            matches = list(iter_candidates([root], opts, stats))
+            self.assertEqual(len(matches), 1)
+            self.assertEqual(stats.roots_seen, 1)
+            self.assertEqual(stats.files_seen, 1)
+            self.assertEqual(stats.messages_read, 1)
+            self.assertEqual(stats.matches_found, 1)
+
+    def test_main_logs_scan_progress_to_stderr(self) -> None:
+        with tempfile.TemporaryDirectory() as td:
+            root = Path(td) / "Maildir"
+            cur = root / "cur"
+            cur.mkdir(parents=True)
+            write_mail(cur / "1:2,S", "no-reply@example.test", "Monthly report")
+            stdout = StringIO()
+            stderr = StringIO()
+
+            with redirect_stdout(stdout), redirect_stderr(stderr):
+                rc = main(["search", "--root", str(root), "--from", "example.test", "--literal"])
+
+            self.assertEqual(rc, 0)
+            self.assertIn("[1] maildir-file", stdout.getvalue())
+            self.assertIn("INFO: Using 1 root(s)", stderr.getvalue())
+            self.assertIn("INFO: Scanning mail root:", stderr.getvalue())
+            self.assertIn("INFO: Scan finished:", stderr.getvalue())
+
+    def test_quiet_suppresses_info_logging(self) -> None:
+        with tempfile.TemporaryDirectory() as td:
+            root = Path(td) / "Maildir"
+            cur = root / "cur"
+            cur.mkdir(parents=True)
+            write_mail(cur / "1:2,S", "no-reply@example.test", "Monthly report")
+            stdout = StringIO()
+            stderr = StringIO()
+
+            with redirect_stdout(stdout), redirect_stderr(stderr):
+                rc = main(["search", "--quiet", "--root", str(root), "--from", "example.test", "--literal"])
+
+            self.assertEqual(rc, 0)
+            self.assertIn("[1] maildir-file", stdout.getvalue())
+            self.assertNotIn("INFO:", stderr.getvalue())
+
+    def test_check_reports_missing_rspamc_without_traceback(self) -> None:
+        with tempfile.TemporaryDirectory() as td:
+            root = Path(td) / "Maildir"
+            cur = root / "cur"
+            cur.mkdir(parents=True)
+            write_mail(cur / "1:2,S", "no-reply@example.test", "Monthly report")
+            stdout = StringIO()
+            stderr = StringIO()
+
+            with redirect_stdout(stdout), redirect_stderr(stderr):
+                rc = main([
+                    "check",
+                    "--root",
+                    str(root),
+                    "--from",
+                    "example.test",
+                    "--literal",
+                    "--rspamc",
+                    "/no/such/rspamc",
+                ])
+
+            self.assertEqual(rc, 3)
+            self.assertIn("returncode=127", stdout.getvalue())
+            self.assertIn("Cannot run /no/such/rspamc", stderr.getvalue())
+
     def test_subject_match(self) -> None:
         with tempfile.TemporaryDirectory() as td:
             root = Path(td) / "Maildir"