fix(extract): preserve empty postgres DSNs

Author: oleksii-tumanov

graphify/cli.py

@@ -2406,35 +2406,36 @@ def _parse_percentile(name: str, raw: str) -> float:
             "force": os.environ.get("GRAPHIFY_FORCE", "").lower()
             in ("1", "true", "yes"),
         }
-        # option -> (destination, action, converter, allow_equals)
+        # option -> (destination, action, converter, allow_equals, allow_empty)
         option_specs = {
-            "--backend": ("backend", "store", None, True),
-            "--model": ("model", "store", None, True),
-            "--mode": ("extract_mode", "store", None, True),
-            "--out": ("out_dir", "store", lambda _n, raw: Path(raw), True),
-            "--no-cluster": ("no_cluster", "flag", None, False),
-            "--dedup-llm": ("dedup_llm", "flag", None, False),
-            "--code-only": ("code_only", "flag", None, False),
-            "--google-workspace": ("google_workspace", "flag", None, False),
-            "--no-gitignore": ("no_gitignore", "flag", None, False),
-            "--global": ("global_merge", "flag", None, False),
-            "--as": ("global_repo_tag", "store", None, False),
-            "--max-workers": ("cli_max_workers", "store", _parse_int, True),
-            "--token-budget": ("cli_token_budget", "store", _parse_int, True),
+            "--backend": ("backend", "store", None, True, False),
+            "--model": ("model", "store", None, True, False),
+            "--mode": ("extract_mode", "store", None, True, False),
+            "--out": ("out_dir", "store", lambda _n, raw: Path(raw), True, False),
+            "--no-cluster": ("no_cluster", "flag", None, False, False),
+            "--dedup-llm": ("dedup_llm", "flag", None, False, False),
+            "--code-only": ("code_only", "flag", None, False, False),
+            "--google-workspace": ("google_workspace", "flag", None, False, False),
+            "--no-gitignore": ("no_gitignore", "flag", None, False, False),
+            "--global": ("global_merge", "flag", None, False, False),
+            "--as": ("global_repo_tag", "store", None, False, False),
+            "--max-workers": ("cli_max_workers", "store", _parse_int, True, False),
+            "--token-budget": ("cli_token_budget", "store", _parse_int, True, False),
             "--max-concurrency": (
-                "cli_max_concurrency", "store", _parse_int, True
+                "cli_max_concurrency", "store", _parse_int, True, False
             ),
-            "--api-timeout": ("cli_api_timeout", "store", _parse_float, True),
-            "--resolution": ("cli_resolution", "store", _parse_float, True),
+            "--api-timeout": ("cli_api_timeout", "store", _parse_float, True, False),
+            "--resolution": ("cli_resolution", "store", _parse_float, True, False),
             "--exclude-hubs": (
-                "cli_exclude_hubs", "store", _parse_percentile, True
+                "cli_exclude_hubs", "store", _parse_percentile, True, False
             ),
-            "--exclude": ("cli_excludes", "append", None, True),
-            "--postgres": ("cli_postgres_dsn", "store", None, True),
-            "--cargo": ("cli_cargo", "flag", None, False),
-            "--force": ("force", "flag", None, False),
-            "--allow-partial": ("cli_allow_partial", "flag", None, False),
-            "--timing": ("cli_timing", "flag", None, False),
+            "--exclude": ("cli_excludes", "append", None, True, False),
+            # Empty DSN intentionally lets psycopg use PG* environment variables.
+            "--postgres": ("cli_postgres_dsn", "store", None, True, True),
+            "--cargo": ("cli_cargo", "flag", None, False, False),
+            "--force": ("force", "flag", None, False, False),
+            "--allow-partial": ("cli_allow_partial", "flag", None, False, False),
+            "--timing": ("cli_timing", "flag", None, False, False),
         }
 
         i = 0
@@ -2449,7 +2450,7 @@ def _parse_percentile(name: str, raw: str) -> float:
                     print("error: unexpected extract positional argument", file=sys.stderr)
                 sys.exit(2)
 
-            destination, action, converter, allow_equals = spec
+            destination, action, converter, allow_equals, allow_empty = spec
             if action == "flag":
                 if equals:
                     print(f"error: {name} does not take a value", file=sys.stderr)
@@ -2466,15 +2467,15 @@ def _parse_percentile(name: str, raw: str) -> float:
             else:
                 if (
                     i + 1 >= len(args)
-                    or not args[i + 1]
+                    or (not args[i + 1] and not allow_empty)
                     or args[i + 1].startswith("--")
                 ):
                     print(f"error: {name} requires a value", file=sys.stderr)
                     sys.exit(2)
                 raw_value = args[i + 1]
                 i += 1
 
-            if not raw_value:
+            if not raw_value and not allow_empty:
                 print(f"error: {name} requires a value", file=sys.stderr)
                 sys.exit(2)
             value = converter(name, raw_value) if converter else raw_value

tests/test_extract_cli_args.py

@@ -42,6 +42,8 @@
     ["--exclude=vendor"],
     ["--postgres", "postgresql://localhost/db"],
     ["--postgres=postgresql://localhost/db"],
+    ["--postgres", ""],
+    ["--postgres="],
     ["--cargo"],
     ["--force"],
     ["--allow-partial"],
@@ -175,7 +177,6 @@ def test_extract_value_options_require_a_value(
         "--resolution=",
         "--exclude-hubs=",
         "--exclude=",
-        "--postgres=",
     ],
 )
 def test_extract_inline_value_options_reject_empty_values(
@@ -201,7 +202,6 @@ def test_extract_inline_value_options_reject_empty_values(
         "--resolution",
         "--exclude-hubs",
         "--exclude",
-        "--postgres",
     ],
 )
 def test_extract_value_options_reject_empty_separate_values(