Test around escaping \

Sage-Bionetworks · Dec 21, 2023 · a93ed4d · a93ed4d
1 parent 1602452
commit a93ed4d
Show file tree

Hide file tree

Showing 2 changed files with 23 additions and 3 deletions.
diff --git a/synapseutils/sync.py b/synapseutils/sync.py
@@ -60,7 +60,7 @@
 # This is looking for a comma that is not preceded by a backslash
 COMMA_PATTERN = re.compile(r"(?<!\\),")
 # This is looking for a single non-escaped backslash. \ is selected, \\ is not
-BACKSLASH_PATTERN = re.compile(r"(?<!\\)\\")
+BACKSLASH_PATTERN = re.compile(r"(?<!\\\\)(\\)(?!\\)")
 
 tracer = trace.get_tracer("synapseclient")
 

diff --git a/tests/integration/synapseutils/test_synapseutils_sync.py b/tests/integration/synapseutils/test_synapseutils_sync.py
@@ -35,9 +35,9 @@ def __init__(self):
             self.f2 = utils.make_bogus_data_file(n=10)
             self.f3 = "https://www.synapse.org"
 
-            self.header = "path	parent	used	executed	activityName	synapseStore	foo	date_1	datetime_1	datetime_2	datetime_3	multiple_strings	multiple_dates	multiple_bools	multiple_ints	multiple_floats\n"
+            self.header = "path	parent	used	executed	activityName	synapseStore	foo	date_1	datetime_1	datetime_2	datetime_3	multiple_strings	multiple_dates	multiple_bools	multiple_ints	multiple_floats	annotation_with_escaped_commas	annotation_with_escaped_backslash\n"
             self.row1 = (
-                '%s	%s	%s	"%s;https://www.example.com"	provName		bar	2020-01-01	2023-12-04T07:00:00Z	2023-12-05 23:37:02+00:00	2023-12-05 07:00:00+00:00	a,b,c,d	2020-01-01,2023-12-04T07:00:00.111Z,2023-12-05 23:37:02.333+00:00,2023-12-05 07:00:00+00:00	fAlSe,False,tRuE,True	1,2,3,4	1.2,3.4,5.6,7.8\n'
+                '%s	%s	%s	"%s;https://www.example.com"	provName		bar	2020-01-01	2023-12-04T07:00:00Z	2023-12-05 23:37:02+00:00	2023-12-05 07:00:00+00:00	a,b,c,d	2020-01-01,2023-12-04T07:00:00.111Z,2023-12-05 23:37:02.333+00:00,2023-12-05 07:00:00+00:00	fAlSe,False,tRuE,True	1,2,3,4	1.2,3.4,5.6,7.8	my\, string with a comma,another\, string with a comma	my\\\\ string with an escaped backslash,another\\\\ string with an escaped backslash\n'
                 % (
                     self.f1,
                     self.project.id,
@@ -212,6 +212,18 @@ def test_syncToSynapse(test_state):
         np.nan,
     ]
 
+    assert new_anots.loc[:]["annotation_with_escaped_commas"].tolist() == [
+        "my\, string with a comma,another\, string with a comma",
+        np.nan,
+        np.nan,
+    ]
+
+    assert new_anots.loc[:]["annotation_with_escaped_backslash"].tolist() == [
+        "my\\ string with an escaped backslash,another\\ string with an escaped backslash",
+        np.nan,
+        np.nan,
+    ]
+
     # Validate that the Annotations uploaded to Synapse are correct when retrieving
     # them through syn.get. Also verify that they are parsed into the correct Python type
     syn_id_first_file = new_df.loc[:]["id"][0]
@@ -239,6 +251,14 @@ def test_syncToSynapse(test_state):
     assert synapse_file_instance.multiple_bools == [False, False, True, True]
     assert synapse_file_instance.multiple_ints == [1, 2, 3, 4]
     assert synapse_file_instance.multiple_floats == [1.2, 3.4, 5.6, 7.8]
+    assert synapse_file_instance.annotation_with_escaped_commas == [
+        "my, string with a comma",
+        "another, string with a comma",
+    ]
+    assert synapse_file_instance.annotation_with_escaped_backslash == [
+        "my\ string with an escaped backslash",
+        "another\ string with an escaped backslash",
+    ]
 
     # Validate that provenance is correct
     for provenanceType in ["executed", "used"]: