feat: custom word list

cool_domain/__main__.py

@@ -1,6 +1,8 @@
 import logging
 import sys
 
+from cool_domain.word_list import WordList
+
 from . import PKG_NAME, PKG_VERSION
 from .cache import Cache
 from .constants import LANGS
@@ -14,10 +16,12 @@ def main() -> int:
     params = parse_parameters(sys.argv[1:])
     setup_logs(params)
     logging.getLogger().info("%s %s", PKG_NAME, PKG_VERSION)
+    langs = LANGS
+    langs["custom"] = WordList("custom", params.custom_words)
     with (
         Cache(filepath=params.cache_file, no_load=params.cache) as cache,
         TLDList(cache=cache, remote_uri=params.tld_list_uri) as tld_list,
-        LANGS[params.lang] as word_list,
+        langs[params.lang] as word_list,
     ):
         factory = Factory(tld_list=tld_list, word_list=word_list, words=params.words)
         logging.getLogger().info(

cool_domain/constants.py

@@ -1,6 +1,6 @@
 from .word_list import WordList
 
 LANGS: dict[str, WordList] = {
-    "english": WordList("english", "english", "words_alpha.txt"),
-    "french": WordList("french", "french", "francais.txt"),
+    "english": WordList("english", "english/words_alpha.txt"),
+    "french": WordList("french", "french/francais.txt"),
 }

cool_domain/params.py

@@ -20,6 +20,7 @@ class Parameters:
     lang: str = "english"
     cache_file: str = str(pathlib.Path(tempfile.gettempdir()) / ".cool-domain-cache")
     tld_list_uri: str = "https://publicsuffix.org/list/public_suffix_list.dat"
+    custom_words: str = "words.txt"
 
     @classmethod
     def from_namespace(cls, args: argparse.Namespace) -> "Parameters":
@@ -98,12 +99,6 @@ def parse_parameters(args: typing.Sequence[str]) -> Parameters:
         default=default_values.count,
         help_txt="number of domains to generate",
     )
-    parser.add_argument(
-        "--lang",
-        default=default_values.lang,
-        help="word list lang (default: english)",
-        choices=LANGS.keys(),
-    )
     __add_arg_int(
         parser,
         "--words",
@@ -111,6 +106,19 @@ def parse_parameters(args: typing.Sequence[str]) -> Parameters:
         default=default_values.words,
         help_txt="number of words per domain",
     )
+    parser.add_argument(
+        "--lang",
+        default=default_values.lang,
+        help="word list lang (default: english)",
+        choices=[*list(LANGS.keys()), "custom"],
+    )
+    __add_arg_str(
+        parser,
+        "--custom-words",
+        env_var="CUSTOM_WORDS",
+        default=default_values.custom_words,
+        help_txt="custom word list location",
+    )
     __add_arg_bool(
         parser,
         "--dig",

cool_domain/word_list.py

@@ -8,10 +8,10 @@ import unidecode
 class WordList:
     __slots__ = ["__filepath", "__name", "logger", "words"]
 
-    def __init__(self, name: str, path: str, filename: str) -> None:
+    def __init__(self, name: str, path: str) -> None:
         self.logger: logging.Logger = logging.getLogger(self.__class__.__name__)
         self.__name = name
-        self.__filepath = pathlib.Path(__file__).parent / ".." / path / filename
+        self.__filepath = pathlib.Path(__file__).parent / ".." / path
         self.words: list[str] = []
 
     def open(self) -> None: