From b51b6b92d7ac69800f1f1cb9c531676409d2e9b0 Mon Sep 17 00:00:00 2001
From: ahiijny <ahiijny@gmail.com>
Date: Fri, 12 Feb 2021 12:58:03 -0500
Subject: [PATCH 1/2] Fix backslash handling in rows_from_chunks

---
 pydruid/db/api.py                 | 21 ++++++++++++++-------
 tests/db/test_rows_from_chunks.py | 12 ++++++++++++
 2 files changed, 26 insertions(+), 7 deletions(-)

diff --git a/pydruid/db/api.py b/pydruid/db/api.py
index 86b64e5e..3fbe94ae 100644
--- a/pydruid/db/api.py
+++ b/pydruid/db/api.py
@@ -394,20 +394,27 @@ def rows_from_chunks(chunks):
             body = "".join((body, chunk))
 
         # find last complete row
+        # see also: https://www.json.org/
         boundary = 0
         brackets = 0
         in_string = False
+        in_escape = False
         for i, char in enumerate(body):
-            if char == '"':
-                if not in_string:
-                    in_string = True
-                elif body[i - 1] != "\\":
-                    in_string = False
-
             if in_string:
+                if in_escape:
+                    # we're just looking for string boundaries, so we can
+                    # ignore the trailing X in escapes like \uXXXX, since each
+                    # of those X characters must be alphanumeric anyway
+                    in_escape = False
+                elif char == '\\':
+                    in_escape = True
+                elif char == '"':
+                    in_string = False
                 continue
 
-            if char == "{":
+            if char == '"':
+                in_string = True
+            elif char == "{":
                 brackets += 1
             elif char == "}":
                 brackets -= 1
diff --git a/tests/db/test_rows_from_chunks.py b/tests/db/test_rows_from_chunks.py
index 5b811811..e143a1e4 100644
--- a/tests/db/test_rows_from_chunks.py
+++ b/tests/db/test_rows_from_chunks.py
@@ -36,6 +36,18 @@ def test_rows_from_chunks_quote_in_string(self):
         result = list(rows_from_chunks(chunks))
         self.assertEqual(result, expected)
 
+    def test_rows_from_chunks_string_ending_with_backslash(self):
+        chunks = [r'[{"name": "\\"}]']
+        expected = [{"name": "\\"}]
+        result = list(rows_from_chunks(chunks))
+        self.assertEqual(result, expected)
+
+    def test_rows_from_chunks_multiple_rows_ending_with_backslashes(self):
+        chunks = [r'[{"name": "alice"}, {"name": "bob\\"}, {"name": "charlie\\"}]']
+        expected = [{"name": "alice"}, {"name": "bob\\"}, {"name": "charlie\\"}]
+        result = list(rows_from_chunks(chunks))
+        self.assertEqual(result, expected)
+
 
 if __name__ == "__main__":
     unittest.main()

From 4a3f7cce08de5b870ef4575215624756f125a301 Mon Sep 17 00:00:00 2001
From: ahiijny <ahiijny@gmail.com>
Date: Fri, 12 Feb 2021 14:42:58 -0500
Subject: [PATCH 2/2] Satisfy the code formatter

---
 pydruid/db/api.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pydruid/db/api.py b/pydruid/db/api.py
index 3fbe94ae..550fed0d 100644
--- a/pydruid/db/api.py
+++ b/pydruid/db/api.py
@@ -406,7 +406,7 @@ def rows_from_chunks(chunks):
                     # ignore the trailing X in escapes like \uXXXX, since each
                     # of those X characters must be alphanumeric anyway
                     in_escape = False
-                elif char == '\\':
+                elif char == "\\":
                     in_escape = True
                 elif char == '"':
                     in_string = False