Moderation

antiabuse/normalize/__init__.py

@@ -5,75 +5,77 @@
 # Used to convert slang in an input string to a more standard form so that it's
 # easier to detect coarse language later on
 _normalization_map = {
+    "[l1]0[l1]i": "loli",
+    "[l1]0[l1]icon": "lolicon",
     "a[s5$]{2}": "ass",
-    "b[a4]ck ?sh[o0]t[s$z]": "backshots",
     "b": "be",
+    "b[a4]ck ?sh[o0]t[s$z]": "backshots",
     "b[i1y]?tch": "bitch",
     "b[o0][o0]b[i1][e3][s$z]": "boobies",
     "btch": "bitch",
+    "c+[uv]+m+": "cum",
+    "c+[uv]+m+s+": "cums",
     "c[a4][s$z]h[a4]pp": "cashapp",
     "c[o0]ck": "cock",
     "c[o0]ck[s$z][uv]ck[e3]r": "cocksucker",
-    "cok": "cock",
-    "c+[uv]+m+": "cum",
-    "c[uv]mming": "cumming",
     "c[uv]m[s$z]h[o0]t": "cumshot",
-    "c+[uv]+m+s+": "cums",
+    "c[uv]mming": "cumming",
+    "cok": "cock",
     "d[iy1]k[e3]": "dyke",
     "dyck": "dick",
     "ejaku[l1]ate": "ejaculate",
-    "fcked": "fucked",
-    "fck": "fuck",
-    "fcking": "fucking",
-    "fked": "fucked",
-    "fk": "fuck",
-    "fking": "fucking",
+    "f[a4]*g+[o0]*t": "faggot",
     "f[o0]{2}tj[o0]b": "footjob",
     "f[o0]{2}tj[o0]b[s$z]": "footjobs",
-    "f[uv]ck[e3]d": "fucked",
     "f[uv]ck": "fuck",
+    "f[uv]ck[e3]d": "fucked",
     "f[uv]ckin": "fucking",
     "f[uv]cking": "fucking",
-    "f[uv]ked": "fucked",
     "f[uv]k": "fuck",
+    "f[uv]ked": "fucked",
     "f[uv]king": "fucking",
-    "gr[o0][o0]mer": "groomer",
+    "fck": "fuck",
+    "fcked": "fucked",
+    "fcking": "fucking",
+    "fk": "fuck",
+    "fked": "fucked",
+    "fking": "fucking",
     "gr[o0][o0]m": "groom",
+    "gr[o0][o0]mer": "groomer",
     "h[@a4]rm": "harm",
     "k[iy1][l1][l1]": "kill",
-    "[l1]0[l1]icon": "lolicon",
-    "[l1]0[l1]i": "loli",
-    "n[i1ye3]*g+(a|uh|e)": "nigga",
     "n[i1ye3]*g+([e3]*r)?": "nigger",
+    "n[i1ye3]*g+(a|uh|e)": "nigga",
     "n[i1ye3]*g+[e3]*r[s$z]": "niggers",
-    "n[i1ye3]gg[l1]et": "niglet",
     "n[i1ye3]g+uh*": "nigga",
+    "n[i1ye3]gg[l1]et": "niglet",
     "n[ie]g{1,2}re{1,2}s+": "negress",
     "p[e3]d[o0]": "pedo",
-    "pissin": "pissing",
     "p[iy1][s$][s$]": "piss",
     "p[o0]rn": "porn",
+    "pissin": "pissing",
     "pr[o0]n": "porn",
-    "r[@a4]p[e3]-?ab[l1]e": "rapeable",
-    "r[a4]p[e3]d": "raped",
-    "r[@a4]p[e3]": "rape",
     "r": "are",
+    "r[@a4]p[e3]": "rape",
+    "r[@a4]p[e3]-?ab[l1]e": "rapeable",
+    "r[@a4]p[e3]d": "raped",
+    "r[@a4]p[i1]ng": "raping",
     "raype": "rape",
     "s[e3][l1]f": "self",
-    "sht": "shit",
     "s[uv][i1]c[i1]d[e3]": "suicide",
+    "sht": "shit",
     "sxy": "sexy",
     "tr[@a4]nnie[s$z]": "trannies",
     "tr[@a4]nny": "tranny",
-    "tr[o0]{2}n[s$z]": "troons",
     "tr[o0]{2}n": "troon",
-    "un-?r[@a4]p[e3]-?ab[l1]e": "unrapeable",
+    "tr[o0]{2}n[s$z]": "troons",
+    "u": "you",
     "un-?r[@a4]p[e3]": "unrape",
-    "urse[l1]f": "yourself",
+    "un-?r[@a4]p[e3]-?ab[l1]e": "unrapeable",
     "ur": "your",
-    "u": "you",
-    "wh[o0]r[e3][s$z]": "whores",
+    "urse[l1]f": "yourself",
     "wh[o0]r[e3]": "whores",
+    "wh[o0]r[e3][s$z]": "whores",
 }
 
 

antiabuse/normalize/test_init.py

@@ -76,6 +76,10 @@ def test_normalize_string(self):
 
         self.assertEqual(normalize_string("nigg"), "nigger")
 
+        self.assertEqual(normalize_string("fag0t"), "faggot")
+
+        self.assertEqual(normalize_string("r@ping"), "raping")
+
 
 if __name__ == '__main__':
     unittest.main()