From 73c243d83ddfd5e3a8c818bd224dbc3fbbe7d9e9 Mon Sep 17 00:00:00 2001
From: Greg Gauthier <gmgauthier@protonmail.com>
Date: Wed, 21 Oct 2020 21:01:26 +0100
Subject: [PATCH] initial commit

---
 .gitignore      |  3 +++
 email_pruner.py | 62 +++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 65 insertions(+)
 create mode 100644 .gitignore
 create mode 100644 email_pruner.py

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..864ac8e
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,3 @@
+.idea/
+*.iml
+*.pyc
diff --git a/email_pruner.py b/email_pruner.py
new file mode 100644
index 0000000..6c3b519
--- /dev/null
+++ b/email_pruner.py
@@ -0,0 +1,62 @@
+from string import ascii_letters
+from secrets import choice
+from timeit import default_timer as timer
+from datetime import timedelta
+
+
+def reset_stopwatch():
+    return timer()
+
+
+def get_elapsed(starttime):
+    end = timer()
+    return timedelta(seconds=end-starttime)
+
+
+def randstring(strlen=64):
+    return ''.join(choice(ascii_letters) for _ in range(strlen))
+
+
+def spawn(listlen=100):
+    base_list = [randstring(10)+"."+randstring(10)+"@"+randstring(15)+".com" for _ in range(listlen)]
+    dup_list = [choice(base_list) for _ in range(len(base_list))]
+    final_list = []
+    for i in range(listlen):
+        final_list.append(base_list[i])
+        final_list.append(dup_list[i])
+    return final_list
+
+
+def dups(biglist):
+    seen = set()
+    uneek = []
+    for x in biglist:
+        if x not in seen:
+            uneek.append(x)
+            seen.add(x)
+    return seen
+
+
+if __name__ == "__main__":
+    start = reset_stopwatch()
+    list_with_dups = spawn(50000)
+    print(f"GENERATED COMPLETE LIST WITH DUPLICATES: (count = {len(list_with_dups)})")
+    # [print(i) for i in list_with_dups]
+    t1 = get_elapsed(start)
+    print("Elapsed Time: ", t1)
+
+    start = reset_stopwatch()
+    dup_list = dups(list_with_dups)
+    print(f"IDENTIFIED DUPLICATES IN COMPLETE LIST: (count = {len(dup_list)})")
+    # [print(i) for i in dup_list]
+    t2 = get_elapsed(start)
+    print("Elapsed time: ", t2)
+
+    start = reset_stopwatch()
+    list_with_dups = list(dict.fromkeys(list_with_dups))
+    print(f"GENERATED PRUNED LIST WITHOUT DUPLICATES: (count = {len(list_with_dups)})")
+    # [print(i) for i in list_with_dups]
+    t3 = get_elapsed(start)
+    print("Elapsed Time: ", t3)
+    print(f"TOTAL ELAPSED TIME: {t1+t2+t3}")
+    print(f"ELAPSED TIME WITHOUT GENERATOR: {t2+t3}")