Add country types

outlines/fsm/types.py

@@ -32,10 +32,12 @@ def custom_format_fn(sequence: str) -> Any:
 
     if isinstance(python_type, EnumMeta):
         values = python_type.__members__.keys()
-        regex_str = "(" + "|".join(values) + ")"
-        format_fn = lambda x: str(x)
+        enum_regex_str: str = "(" + "|".join(values) + ")"
 
-        return regex_str, format_fn
+        def enum_format_fn(sequence: str) -> str:
+            return str(sequence)
+
+        return enum_regex_str, enum_format_fn
 
     if python_type == float:
 

outlines/types/__init__.py

@@ -1,4 +1,4 @@
+from . import airports, countries
 from .isbn import ISBN
 from .phone_numbers import PhoneNumber
 from .zip_codes import ZipCode
-from . import airports

outlines/types/countries.py

@@ -0,0 +1,25 @@
+"""Generate valid country codes and names."""
+from typing import List
+from enum import Enum
+
+try:
+    import pycountry
+except ImportError:
+    raise ImportError(
+        'The `countries` module requires "pycountry" to be installed. You can install it with "pip install pycountry"'
+    )
+
+ALPHA_2_CODE = [(country.alpha_2, country.alpha_2) for country in pycountry.countries]
+Alpha2 = Enum("Alpha_2", ALPHA_2_CODE)  # type:ignore
+
+ALPHA_3_CODE = [(country.alpha_3, country.alpha_3) for country in pycountry.countries]
+Alpha3 = Enum("Alpha_2", ALPHA_3_CODE)  # type:ignore
+
+NUMERIC_CODE = [(country.numeric, country.numeric) for country in pycountry.countries]
+Numeric = Enum("Numeric_code", NUMERIC_CODE)  # type:ignore
+
+NAME = [(country.name, country.name) for country in pycountry.countries]
+Name = Enum("Name", NAME)  # type:ignore
+
+FLAG = [(country.flag, country.flag) for country in pycountry.countries]
+Flag = Enum("Flag", FLAG)  # type:ignore

pyproject.toml

@@ -126,6 +126,7 @@ module = [
     "vllm.*",
     "uvicorn.*",
     "fastapi.*",
+    "pycountry.*",
     "pyairports.*",
 ]
 ignore_missing_imports = true

tests/test_types.py

@@ -47,10 +47,19 @@ class Model(BaseModel):
     [
         (types.airports.IATA, "CDG", True),
         (types.airports.IATA, "XXX", False),
+        (types.countries.Alpha2, "FR", True),
+        (types.countries.Alpha2, "XX", False),
+        (types.countries.Alpha3, "UKR", True),
+        (types.countries.Alpha3, "XXX", False),
+        (types.countries.Numeric, "004", True),
+        (types.countries.Numeric, "900", False),
+        (types.countries.Name, "Ukraine", True),
+        (types.countries.Name, "Wonderland", False),
+        (types.countries.Flag, "🇿🇼", True),
+        (types.countries.Flag, "🤗", False),
     ],
 )
 def test_type_enum(custom_type, test_string, should_match):
-
     type_name = custom_type.__name__
 
     class Model(BaseModel):