feat(anonymize): add NULL, NAME, PHONE_NUMBER, DATE_OF_BIRTH, TEXT, WORD types

README.md

@@ -102,7 +102,50 @@ tables:
     sample: 10
 ```
 
-## 3. Migration utils
+## 3. Anonymization
+
+You can scrub PII from selected columns with `padmy anonymize`. Field types map
+to Faker generators (or `NULL` to blank the column). Available types:
+
+| Type | Behavior |
+| --- | --- |
+| `EMAIL` | `faker.email()` — supports `domain:` extra arg |
+| `NULL` | Sets the column to `NULL` (useful for hashes, tokens, anything you don't want to fake) |
+| `FIRST_NAME` | `faker.first_name()` |
+| `LAST_NAME` | `faker.last_name()` |
+| `NAME` | `faker.name()` (full name) |
+| `PHONE_NUMBER` | `faker.phone_number()` |
+| `DATE_OF_BIRTH` | `faker.date_of_birth()` — supports `minimum_age:` / `maximum_age:` |
+| `TEXT` | `faker.text()` — supports `max_nb_chars:` |
+| `WORD` | `faker.word()` |
+
+Example config:
+
+```yaml
+tables:
+  - schema: public
+    table: users
+    fields:
+      - column: email
+        type: EMAIL
+        domain: example.com         # extra arg forwarded to faker.email
+      - column: password_hash
+        type: "NULL"                # quoted: YAML's bare NULL parses as null
+      - column: first_name
+        type: FIRST_NAME
+      - column: birthdate
+        type: DATE_OF_BIRTH
+        minimum_age: 18
+        maximum_age: 80
+```
+
+Run with:
+
+```bash
+uvx padmy anonymize --db test -f config.yml
+```
+
+## 4. Migration utils
 
 **Setting up**
 
@@ -227,7 +270,7 @@ We are all good !
 uvx padmy -vv migrate verify-files --sql-dir /tmp/migrations --no-raise
 ```
 
-## 4. Comparing databases schemas
+## 5. Comparing databases schemas
 
 You can compare two databases by running:
 

padmy/anonymize/anonymize.py

@@ -41,6 +41,25 @@ def _get_fake_value(faker: Faker, field: FieldType, extra_fields: dict | None =
     match field:
         case "EMAIL":
             return faker.email(**_extra_fields)
+        case "NULL":
+            # Drop the column to NULL. Useful when no realistic placeholder is
+            # wanted (password hashes, tokens, secrets) — the column shape is
+            # preserved but the value carries no information.
+            return None
+        case "FIRST_NAME":
+            return faker.first_name(**_extra_fields)
+        case "LAST_NAME":
+            return faker.last_name(**_extra_fields)
+        case "NAME":
+            return faker.name(**_extra_fields)
+        case "PHONE_NUMBER":
+            return faker.phone_number(**_extra_fields)
+        case "DATE_OF_BIRTH":
+            return faker.date_of_birth(**_extra_fields)
+        case "TEXT":
+            return faker.text(**_extra_fields)
+        case "WORD":
+            return faker.word(**_extra_fields)
         case _:
             raise ValueError(f"Got unimplemented field type {field!r}")
 

padmy/config.py

@@ -4,7 +4,17 @@
 
 import yaml
 
-FieldType = Literal["EMAIL"]
+FieldType = Literal[
+    "EMAIL",
+    "NULL",
+    "FIRST_NAME",
+    "LAST_NAME",
+    "NAME",
+    "PHONE_NUMBER",
+    "DATE_OF_BIRTH",
+    "TEXT",
+    "WORD",
+]
 
 SampleType = float | int
 

tests/test_anonymize.py

@@ -78,6 +78,33 @@ async def test():
     ]
 
 
+@pytest.mark.parametrize(
+    "field_type, extra, predicate",
+    [
+        pytest.param("EMAIL", None, lambda v: v and "@" in v, id="EMAIL"),
+        pytest.param("NULL", None, lambda v: v is None, id="NULL"),
+        pytest.param("FIRST_NAME", None, lambda v: isinstance(v, str) and v, id="FIRST_NAME"),
+        pytest.param("LAST_NAME", None, lambda v: isinstance(v, str) and v, id="LAST_NAME"),
+        pytest.param("NAME", None, lambda v: isinstance(v, str) and " " in v, id="NAME"),
+        pytest.param("PHONE_NUMBER", None, lambda v: isinstance(v, str) and v, id="PHONE_NUMBER"),
+        pytest.param("WORD", None, lambda v: isinstance(v, str) and v, id="WORD"),
+    ],
+)
+def test_get_fake_value(faker, field_type, extra, predicate):
+    """Each supported field type returns something matching its shape."""
+    from padmy.anonymize.anonymize import _get_fake_value
+
+    value = _get_fake_value(faker, field_type, extra)
+    assert predicate(value), f"unexpected value for {field_type}: {value!r}"
+
+
+def test_get_fake_value_unknown_type_raises(faker):
+    from padmy.anonymize.anonymize import _get_fake_value
+
+    with pytest.raises(ValueError, match="unimplemented field type"):
+        _get_fake_value(faker, "DOES_NOT_EXIST")  # type: ignore[arg-type]
+
+
 @pytest.mark.usefixtures("add_table_1_data")
 def test_anonymize_db(apool, engine, loop, faker):
     from padmy.anonymize import anonymize_db