Merge pull request #11 from snguyenthanh/development

Fix compatibility with Python 3.5 (and earlier) and update wordlist
snguyenthanh · Apr 8, 2020 · e352465 · e352465
2 parents 4b2454c + f59238f
commit e352465
Show file tree

Hide file tree

Showing 7 changed files with 97 additions and 15 deletions.
diff --git a/.travis.yml b/.travis.yml
@@ -1,6 +1,30 @@
+dist: xenial
+
 language: python
-python:
-  - "3.6"
+
+services:
+  - redis-server
+
+matrix:
+  include:
+    - python: 3.8
+      env: TOXENV=black
+
+    - python: 3.4
+      env: TOXENV=py34
+    - python: 3.5
+      env: TOXENV=py35
+    - python: 3.6
+      env: TOXENV=py36
+    - python: 3.7
+      env: TOXENV=py37
+    - python: 3.8
+      env: TOXENV=py38
+    - python: pypy3
+      env: TOXENV=pypy3
+
+install:
+  - pip install tox
 
 script:
-  - python tests.py
+  - tox
diff --git a/README.md b/README.md
@@ -30,7 +30,7 @@ Not all languages are supported yet, such as *Chinese*.
 ## Wordlist
 Most of the words in the default [wordlist](./better_profanity/profanity_wordlist.txt) are referred from [Full List of Bad Words and Top Swear Words Banned by Google](https://github.com/RobertJGabriel/Google-profanity-words).
 
-The wordlist contains a total of __106,992 words__, including 317 words from the default [profanity_wordlist.txt](./better_profanity/profanity_wordlist.txt) and their variants by modified spellings.
+The wordlist contains a total of __181,590 words__, including 320 words from the default [profanity_wordlist.txt](./better_profanity/profanity_wordlist.txt) and their variants by modified spellings.
 
 ## Usage
 
@@ -145,7 +145,7 @@ It is best used when there are only a few words that you would like to ignore in
 
 ```
 # Use the default wordlist
-profanity.load_censor_words(whitelist_words=['gay', 'lesbian'])
+profanity.load_censor_words(whitelist_words=['happy', 'merry'])
 
 # or with your custom words as a List
 custom_badwords = ['happy', 'jolly', 'merry']

diff --git a/better_profanity/__init__.py b/better_profanity/__init__.py
@@ -5,6 +5,6 @@
 __all__ = ["name", "__version__", "profanity"]
 
 name = "better_profanity"
-__version__ = "0.6.0"
+__version__ = "0.6.1"
 
 profanity = Profanity()
diff --git a/better_profanity/better_profanity.py b/better_profanity/better_profanity.py
@@ -24,6 +24,7 @@ def __init__(self):
             "l": ("l", "1"),
             "e": ("e", "*", "3"),
             "s": ("s", "$", "5"),
+            "t": ("t", "7",),
         }
         self.MAX_NUMBER_COMBINATIONS = 1
         self.ALLOWED_CHARACTERS = ALLOWED_CHARACTERS
@@ -83,8 +84,8 @@ def _populate_words_to_wordset(self, words, *, whitelist_words=None):
         for index, word in enumerate(whitelist_words):
             if not isinstance(word, str):
                 raise ValueError(
-                    f"Each word in 'whitelist_words' must be 'str' type, "
-                    f"but '{type(word)}' found."
+                    "Each word in 'whitelist_words' must be 'str' type, "
+                    "but '{word}' found.".format(word=type(word))
                 )
             whitelist_words[index] = word.lower()
 

diff --git a/better_profanity/profanity_wordlist.txt b/better_profanity/profanity_wordlist.txt
@@ -1,9 +1,13 @@
 2g1c
 2 girls 1 cup
+4r5e
 anal
 anus
 arse
 ass
+asses
+assfucker
+assfukka
 asshole
 arsehole
 asswhole
@@ -13,8 +17,10 @@ autoerotic
 ballsack
 bastard
 beastial
+bestial
 bellend
 bdsm
+beastiality
 bestiality
 bitch
 bitches
@@ -173,8 +179,6 @@ gang bang
 gokkun
 golden shower
 goldenshower
-gay
-gaylord
 gaysex
 goatse
 handjob
@@ -193,7 +197,6 @@ jizz
 knob
 kinbaku
 labia
-lesbian
 masturbate
 masochist
 mofo

diff --git a/tests.py b/tests.py
@@ -172,12 +172,12 @@ def setUp(self):
         profanity.load_censor_words()
 
     def test_whitelist_words(self):
-        bad_text = "I am gay"
-        censored_text = "I am ****"
+        bad_text = "I have boobs"
+        censored_text = "I have ****"
         self.assertEqual(profanity.censor(bad_text), censored_text)
 
-        # Whitelist the word `gay`
-        profanity.load_censor_words(whitelist_words=["gay"])
+        # Whitelist the word `boobs`
+        profanity.load_censor_words(whitelist_words=["boobs"])
         self.assertEqual(profanity.censor(bad_text), bad_text)
 
 

diff --git a/tox.ini b/tox.ini
@@ -0,0 +1,54 @@
+[tox]
+envlist =
+    black
+    py34, py35, py36, py37, py38, pypy3
+
+skipsdist = True
+skip_missing_interpreters = True
+
+[default]
+basepython = python3.8
+
+setenv =
+    PY_MODULE=better_profanity
+    PYTHONPYCACHEPREFIX={envtmpdir}/pycache
+
+[testenv]
+parallel_show_output = True
+whitelist_externals = make
+                      /bin/bash
+
+basepython =
+		py34: python3.4
+		py35: python3.5
+		py36: python3.6
+    py37: python3.7
+    py38: python3.8
+
+    pypy3: pypy3
+
+
+# run the tests
+# ... or run any other command line tool you need to run here
+commands = python tests.py
+
+
+[testenv:black]
+description = run Black (linter)
+basepython = {[default]basepython}
+skip_install = True
+deps =
+    black==19.10b0
+setenv =
+    BLACK_LINT_ARGS=--check
+commands =
+    black {env:BLACK_LINT_ARGS:} better_profanity
+
+
+[testenv:black-reformat]
+
+description  = {[testenv:black]description} and reformat
+basepython   = {[testenv:black]basepython}
+skip_install = {[testenv:black]skip_install}
+deps         = {[testenv:black]deps}
+commands     = {[testenv:black]commands}