chore: 1.3.0

.gitignore

@@ -9,3 +9,4 @@ coverage.xml
 build
 dist
 .pytest_cache
+words.txt

README.md

@@ -23,15 +23,18 @@ uvx --from git+https://git.klemek.fr/klemek/cool-domain cool-domain
 ## Usage
 
 ```txt
-usage: cool-domain [-h] [--count DOMAIN_COUNT] [--lang {english,french}] [--words WORD_COUNT] [--dig | --no-dig] [--debug | --no-debug]
-                   [--quiet | --no-quiet] [--cache | --no-cache] [--cache-file CACHE_FILE] [--tld-list-uri TLD_LIST_URI]
+usage: cool-domain [-h] [--count DOMAIN_COUNT] [--words WORD_COUNT] [--lang {english,french,custom}] [--custom-words CUSTOM_WORDS]
+                   [--dig | --no-dig] [--debug | --no-debug] [--quiet | --no-quiet] [--cache | --no-cache] [--cache-file CACHE_FILE]
+                   [--tld-list-uri TLD_LIST_URI]
 
 options:
   -h, --help            show this help message and exit
   --count DOMAIN_COUNT  number of domains to generate (default: 1)
-  --lang {english,french}
-                        word list lang (default: english)
   --words WORD_COUNT    number of words per domain (default: 1)
+  --lang {english,french,custom}
+                        word list lang (default: english)
+  --custom-words CUSTOM_WORDS
+                        custom word list location (default: words.txt)
   --dig, --no-dig       check availability with dig (default: true)
   --debug, --no-debug
   --quiet, --no-quiet

cool_domain/__main__.py

@@ -1,23 +1,26 @@
 import logging
 import sys
 
-from . import PKG_NAME, PKG_VERSION
-from .cache import Cache
-from .constants import LANGS
-from .factory import Factory
-from .logs import setup_logs
-from .params import parse_parameters
-from .tld_list import TLDList
+from cool_domain import PKG_NAME, PKG_VERSION
+from cool_domain.cache import Cache
+from cool_domain.constants import LANGS
+from cool_domain.factory import Factory
+from cool_domain.logs import setup_logs
+from cool_domain.params import parse_parameters
+from cool_domain.tld_list import TLDList
+from cool_domain.word_list import WordList
 
 
 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(
@@ -29,3 +32,7 @@ def main() -> int:
             else:
                 logging.getLogger().info(factory.random_domain())
     return 0
+
+
+if __name__ == "__main__":
+    main()

cool_domain/factory.py

@@ -4,8 +4,8 @@ import shutil
 import subprocess
 import typing
 
-from .tld_list import TLDList
-from .word_list import WordList
+from cool_domain.tld_list import TLDList
+from cool_domain.word_list import WordList
 
 
 class FactoryError(Exception):

cool_domain/logs.py

@@ -2,7 +2,7 @@ import enum
 import logging
 import typing
 
-from .params import Parameters
+from cool_domain.params import Parameters
 
 
 class TermColor(enum.StrEnum):

cool_domain/params.py

@@ -5,8 +5,8 @@ import pathlib
 import tempfile
 import typing
 
-from . import PKG_NAME
-from .constants import LANGS
+from cool_domain import PKG_NAME
+from cool_domain.constants import LANGS
 
 
 @dataclasses.dataclass(frozen=True, slots=True)
@@ -19,7 +19,8 @@ class Parameters:
     dig: bool = True
     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"
+    tld_list_uri: str = "https://data.iana.org/TLD/tlds-alpha-by-domain.txt"
+    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/tld_list.py

@@ -5,7 +5,8 @@ import typing
 import requests
 import unidecode
 
-from .cache import Cache
+from cool_domain.cache import Cache
+from cool_domain.constants import BLACKLIST
 
 
 class TLDError(Exception):
@@ -29,20 +30,17 @@ class TLDList:
         raw_lines = response.content.decode().split("\n")
         out_lines = []
         for line in raw_lines:
-            stripped_line = line.strip()
+            stripped_line = line.strip().lower()
             if (
                 len(stripped_line)
                 and not stripped_line.startswith("//")
                 and not stripped_line.startswith("*")
+                and not stripped_line.startswith("#")
+                and not stripped_line.startswith("xn--")
+                and "." not in stripped_line
                 and unidecode.unidecode(stripped_line) == stripped_line
+                and stripped_line not in BLACKLIST
             ):
-                if "." in stripped_line:
-                    for other in out_lines:
-                        if stripped_line.endswith(other):
-                            break
-                    else:
-                        out_lines += [stripped_line]
-                else:
                 out_lines += [stripped_line]
         return out_lines
 

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:

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "cool-domain"
-version = "1.1.2"
+version = "1.3.0"
 description = "Add your description here"
 readme = "README.md"
 requires-python = ">=3.12"

uv.lock

@@ -86,7 +86,7 @@ wheels = [
 
 [[package]]
 name = "cool-domain"
-version = "1.1.2"
+version = "1.3.0"
 source = { editable = "." }
 dependencies = [
     { name = "requests" },