From 6cc0f86f7472639a07f353f4581b71f78f44fad7 Mon Sep 17 00:00:00 2001
From: Lukas Winkler <git@lw1.at>
Date: Tue, 11 Aug 2020 12:49:49 +0200
Subject: [PATCH] add a lot of typos and a way to quickly find them

---
 typo.py      | 16 ++++++++++------
 typo_find.py | 22 ++++++++++++++++++++++
 2 files changed, 32 insertions(+), 6 deletions(-)
 create mode 100644 typo_find.py

diff --git a/typo.py b/typo.py
index c83fd84..3278401 100644
--- a/typo.py
+++ b/typo.py
@@ -3,19 +3,23 @@ replace common typos of names to unify them in the database
 """
 
 typos = {
-    "Matt": {"Mat", "Mattt", "\"Matt", "Matr"},
-    "Sam": {"San", "Nott", "Sma", "Sasm", "Sm"},
-    "Travis": {"Tarvis", "Travs", "Travia", "Traivs"},
+    "Matt": {"Mat", "Mattt", "\"Matt", "Matr", "Mtt"},
+    "Sam": {"San", "Nott", "Sma", "Sasm", "Sm", "Ssam"},
+    "Travis": {"Tarvis", "Travs", "Travia", "Traivs", "Tavis", "Trvis"},
     "Taliesin": {"Taiesin", "Talisin", "Talisen", "Taleisn", "Talisein", "Talieisin", "Talesin", "Talisan", "Taleisin",
-                 "Talieisn", "Talisien"},
+                 "Talieisn", "Talisien", "Tailesin"},
     "Marisha": {"Beau", "Mariasha", "Maisha", "Marisa", "Marish", "Marihsa", "Marsha", "Marsisha", "Marishaa",
-                "Marihsha", "\\Marisha", "Marisah", "Marissa"},
+                "Marihsha", "\\Marisha", "Marisah", "Marissa", "Marirsha", "Marisaha", "Mairsha", "Marshia", "Marsiha",
+                "Marishia", "Marsiah", "Matisha", "Mraisha", "Amrisha", "<Arisha"},
     "Laura": {"Lauda", "Lauren", "Larua", "Laur", "Lauar", "Vex", "Laira"},
     "Liam": {"Caleb", "Laim", "Vax"},
-    "Ashley": {"Ashly", "Ashely", "Ashey"},
+    "Ashley": {"Ashly", "Ashely", "Ashey", "Aslhey", "Ahsley"},
     "All": {"Everyone", "Everybody"},
     "Mark": {"Marik"},
     "Brian": {"Brain"},
+    "Joe": {"Jroe"},
+    "Man Off-Camera": {"Man Off Camera"},
+    "Off-Screen": {"Offscreen"}
 }
 replacements = {}
 for correct, typoset in typos.items():
diff --git a/typo_find.py b/typo_find.py
new file mode 100644
index 0000000..127e9e6
--- /dev/null
+++ b/typo_find.py
@@ -0,0 +1,22 @@
+from difflib import SequenceMatcher
+from itertools import combinations
+
+from models import Person
+
+people = set()
+for p in Person.select():
+    if "\n" not in p.name:
+        if "," in p.name:
+            names = [n.strip() for n in p.name.split(",")]
+            people.update(names)
+        else:
+            people.add(p.name)
+print(people)
+
+for a, b in combinations(people, r=2):
+    s = SequenceMatcher(None, a, b)
+    ratio = s.ratio()
+    if ratio < 0.8 or ratio == 1:
+        continue
+    print(a, "|", b)
+    print(ratio)