From 11b79c71f1c121c6556a556721ac658989b2ad1c Mon Sep 17 00:00:00 2001
From: Scott Christley <scott.christley@utsouthwestern.edu>
Date: Fri, 18 Oct 2024 12:12:08 -0500
Subject: [PATCH] fix read gzip, add test case

---
 lang/python/airr/interface.py                 |   2 +-
 .../tests/data/good_rearrangement.tsv.gz      | Bin 0 -> 1106 bytes
 lang/python/tests/test_interface.py           |  19 ++++++++++++++++++
 3 files changed, 20 insertions(+), 1 deletion(-)
 create mode 100644 lang/python/tests/data/good_rearrangement.tsv.gz

diff --git a/lang/python/airr/interface.py b/lang/python/airr/interface.py
index a8f60298f..b58811dc7 100644
--- a/lang/python/airr/interface.py
+++ b/lang/python/airr/interface.py
@@ -41,7 +41,7 @@ def read_rearrangement(filename, validate=False, debug=False):
       airr.io.RearrangementReader: iterable reader class.
     """
     if filename.endswith(".gz"):
-        handle = gzip.open(filename, 'r')
+        handle = gzip.open(filename, 'rt')
     else:
         handle = open(filename, 'r')
         
diff --git a/lang/python/tests/data/good_rearrangement.tsv.gz b/lang/python/tests/data/good_rearrangement.tsv.gz
new file mode 100644
index 0000000000000000000000000000000000000000..c838c584d5e10ba715dd1ebb7fbe1925d847ae23
GIT binary patch
literal 1106
zcmV-Y1g-lYiwFpPH)>`817~k<WM6V+VRCX|Zf9j}Wo~pXbaQq9?U-9{+cp%(pV^OM
z%!YUoMLqaIQxeWu;WaVgq<sklOYkf>il^8C_U(5LsjD3)P1|fK78v5?Paa;Z9}W*y
zW=UBl#WK6iitWuRC1p>z&#ZW#{k6-AMb<4t&xf1E-R*{K%DZ&8*sdP3wqvqhEejli
zEVJ@<y(+pA*d^;VNnMt^Y*E#A1q^jpG=-aWRxG!Fw)G@|0qz%fWd<X-Jg@RamBR$9
zWm4h{HqWbkQRNS?5m?A-`!^J`hh)9Ou^KI{wY=5xT7$XkP2O*la@!TMB86(R<GQ^O
zusz+kPy6=izJ1!a&-d-~zJ0!LpZD#H4XyvuV#DfUy4$Q*i)5R@E8P{_50m-t(~Hw{
zkN!H>7m`r2CukkVainz=8?B9wBB&b^N1#Q}Vzj{&>QIa$TQ{z&hC&=c6X@D0#y}4{
z5{Q^W4;|YcZWbILXS6+R2S_xSKu;5)xs|oEaHB<NTQ4I*yBHH_0V_Md!L>%%$6yC+
zfcDs$*c}<W9$dWLCoW1m6D)&lr^aR+8q{ikxL)`-j5V|S(z@P<9bB8{Kf%=z13yA0
z`nS0laysxhmQUy~7<lv-EXpAbp^Wmm7<i13ubGu_N#bax=diD{lh0Q~aF0+q5+v|P
ziV()8bjAXIOemc)<;~QX$SDiijI%Ml4#xyuJUty!#-PAP0r5s*AQ_RMDdG{Zg_P3(
zN+Dq?AQUE0goTuF&WTS%Km`0K6G6*VbTPl0b(wO)TV^V1nTijXir)-V@qfS+tC&)}
zVT$?^_@N#q0VFCsL`*Qi3wTn%ibBS`iZMt}hfsR~UZTU+AD_>@^w_e(TW0GG_$%{b
zxXvDd>+6U*n2YeBMrwLC`1tYK`YGo1nd<r}Aj?w39Yq}{<&Ef51cgWtv;crWkf;gb
zBxEf@rxSlBx(GSp9igAnJ@G^L^yrbdPtrHKA`p0s(uuyjoODrg!V#44Z62be9ZHb{
z2b+wkI64YZ+=a;eKOh1~!dHTbqvP0zh>zK|^6`kX2|OGgR+Uk+hagpswLny?DcPKT
zk|`vNQg`;rE^snzfa3rht$|ZM;VkHcR-Lo&)6}2#&OR%=JFXWn7d;=VKX&weCv(w0
zbA#4#$zT*d<+yB^#puz1-DkpT{)+<k&JEBrhV^cMR(PlC`y;AnvHI~5)sdTD8&N%W
z^|MjcGk3qEp%wQ!9%?cRdZCpQG@f!5OXi&=9Qt1k&!3}70Zo^eT3-Q6&;WNnaS)&5
z;iu3ONmfxs8>&Qe;)QT@g@_e>GFVs0k=GCf5OLoT1*c!{x_2(bya$vOUOmA2nNa#c
zAEtbON4{EsB@aGK3EzfM%!X0C3aJ=+CsD;%_L2T!eAW0c8)Nx4#)9C`#aNXe^J_mQ
Y152I%RG=m8H>?l;0K5(Ux-J(00KYL1`v3p{

literal 0
HcmV?d00001

diff --git a/lang/python/tests/test_interface.py b/lang/python/tests/test_interface.py
index 5fc54060c..0585cac0d 100644
--- a/lang/python/tests/test_interface.py
+++ b/lang/python/tests/test_interface.py
@@ -23,6 +23,7 @@ def setUp(self):
 
         # Test data
         self.rearrangement_good = os.path.join(data_path, 'good_rearrangement.tsv')
+        self.rearrangement_good_gz = os.path.join(data_path, 'good_rearrangement.tsv.gz')
         self.rearrangement_bad = os.path.join(data_path, 'bad_rearrangement.tsv')
         self.rep_good = os.path.join(data_path, 'good_repertoire.yaml')
         self.rep_bad = os.path.join(data_path, 'bad_repertoire.yaml')
@@ -59,6 +60,24 @@ def test_load_rearrangement(self):
         result = airr.load_rearrangement(self.rearrangement_bad)
         self.assertTupleEqual(result.shape, self.shape_bad, 'load(): bad data failed')
 
+    # @unittest.skip('-> read_rearrangement(): skipped\n')
+    def test_read_rearrangement(self):
+        # Good data
+        result = []
+        reader = airr.read_rearrangement(self.rearrangement_good)
+        for row in reader:
+            result.append(row)
+        reader.close()
+        self.assertTrue(len(result) == self.shape_good[0], 'read_rearrangement(): good data failed')
+
+        # Good data, gzip
+        result = []
+        reader = airr.read_rearrangement(self.rearrangement_good_gz)
+        for row in reader:
+            result.append(row)
+        reader.close()
+        self.assertTrue(len(result) == self.shape_good[0], 'read_rearrangement(): good data (gzip) failed')
+
     # @unittest.skip('-> repertoire_template(): skipped\n')
     def test_repertoire_template(self):
         try: