From 4e8d5cf6bfde5f36740a8d1ce819b6bc63999644 Mon Sep 17 00:00:00 2001
From: Ursify <35331253+Ursify@users.noreply.github.com>
Date: Sun, 6 Dec 2020 00:45:15 +1100
Subject: [PATCH] Base 32 Pattern Check (#37)

* Base 32 Pattern Check

BASE 32 works in groups of 8 Characters.
Valid Final 8 Characters(with / without padding):
CC======
CC
CCCC====
CCCC
CCCCC===
CCCCC
CCCCCCC=
CCCCCCC
CCCCCCCC
Invalid Final 8 Characters(with / without padding):
C=======
C
CCCCCC==
CCCCCC

Invalid 8 characters group should throw an error, since it won't be possible to uncoded them back.

* Allow empty string & update tests

* Add tests for incorrect padding

* Fix on Empty String

Empty string was not raising TealInputError.

* Allowing Empty strings

Sorry about that, first I understood that the issue was that empty strings were allowed. I rolled it back.

Co-authored-by: Jason Paulos <jasonpaulos@users.noreply.github.com>
---
 examples/signature/atomic_swap.py   |  2 +-
 examples/signature/atomic_swap.teal |  2 +-
 pyteal/ast/bytes_test.py            | 21 +++++++++++++++++++++
 pyteal/types.py                     |  5 ++---
 4 files changed, 25 insertions(+), 5 deletions(-)

diff --git a/examples/signature/atomic_swap.py b/examples/signature/atomic_swap.py
index 14b7f2140..5b9bc6c57 100644
--- a/examples/signature/atomic_swap.py
+++ b/examples/signature/atomic_swap.py
@@ -6,7 +6,7 @@
 
 alice = Addr("6ZHGHH5Z5CTPCF5WCESXMGRSVK7QJETR63M3NY5FJCUYDHO57VTCMJOBGY")
 bob = Addr("7Z5PWO2C6LFNQFGHWKSK5H47IQP5OJW2M3HA2QPXTY3WTNP5NU2MHBW27M")
-secret = Bytes("base32", "23232323232323")
+secret = Bytes("base32", "2323232323232323")
 timeout = 3000
 
 def htlc(tmpl_seller=alice,
diff --git a/examples/signature/atomic_swap.teal b/examples/signature/atomic_swap.teal
index 3cb7377fe..762e0250d 100644
--- a/examples/signature/atomic_swap.teal
+++ b/examples/signature/atomic_swap.teal
@@ -19,7 +19,7 @@ addr 6ZHGHH5Z5CTPCF5WCESXMGRSVK7QJETR63M3NY5FJCUYDHO57VTCMJOBGY
 ==
 arg 0
 sha256
-byte base32(23232323232323)
+byte base32(2323232323232323)
 ==
 &&
 txn Receiver
diff --git a/pyteal/ast/bytes_test.py b/pyteal/ast/bytes_test.py
index 3a3c19cec..6126eb3d3 100644
--- a/pyteal/ast/bytes_test.py
+++ b/pyteal/ast/bytes_test.py
@@ -86,6 +86,27 @@ def test_bytes_invalid():
     with pytest.raises(TealInputError):
         Bytes("base32", "Zm9vYmE=")
 
+    with pytest.raises(TealInputError):
+        Bytes("base32", "MFRGG====")
+
+    with pytest.raises(TealInputError):
+        Bytes("base32", "MFRGG==")
+
+    with pytest.raises(TealInputError):
+        Bytes("base32", "CCCCCC==")
+
+    with pytest.raises(TealInputError):
+        Bytes("base32", "CCCCCC")
+
+    with pytest.raises(TealInputError):
+        Bytes("base32", "C=======")
+            
+    with pytest.raises(TealInputError):
+        Bytes("base32", "C")
+    
+    with pytest.raises(TealInputError):
+        Bytes("base32", "=")
+
     with pytest.raises(TealInputError):
         Bytes("base64", "?????")
 
diff --git a/pyteal/types.py b/pyteal/types.py
index 08bf1ec66..c9f95cbc6 100644
--- a/pyteal/types.py
+++ b/pyteal/types.py
@@ -48,10 +48,9 @@ def valid_address(address:str):
 def valid_base32(s:str):
     """ check if s is a valid base32 encoding string
     """
-    noPaddingPattern = re.compile(r'[A-Z2-7]*')
-    paddingPattern = re.compile(r'^(?:[A-Z2-7]{8})*(?:([A-Z2-7]{8})|([A-Z2-7]{2}[=]{6})|([A-Z2-7]{4}[=]{4})|([A-Z2-7]{5}[=]{3})|([A-Z2-7]{7}[=]{1}))')
+    pattern = re.compile(r'^(?:[A-Z2-7]{8})*(?:([A-Z2-7]{2}([=]{6})?)|([A-Z2-7]{4}([=]{4})?)|([A-Z2-7]{5}([=]{3})?)|([A-Z2-7]{7}([=]{1})?))?')
 
-    if noPaddingPattern.fullmatch(s) is None and paddingPattern.fullmatch(s) is None:
+    if pattern.fullmatch(s) is None:
         raise TealInputError("{} is not a valid RFC 4648 base 32 string".format(s))
 
 def valid_base64(s:str):