From 65138c5cd99a4f55a092a74014e48ce76f81ff27 Mon Sep 17 00:00:00 2001
From: ABW <andrzejl@syncad.com>
Date: Fri, 4 Sep 2020 19:54:19 +0200
Subject: [PATCH] [ABW]: some characters apparently need long 8-byte encoding
(problem actually caused by use of ujson)
---
hive/utils/normalize.py | 26 +++++++++++++++-----------
tests/tests_api | 2 +-
2 files changed, 16 insertions(+), 12 deletions(-)
diff --git a/hive/utils/normalize.py b/hive/utils/normalize.py
index 0894cd1df..0fe4f5d13 100644
--- a/hive/utils/normalize.py
+++ b/hive/utils/normalize.py
@@ -78,19 +78,23 @@ def escape_characters(text):
if ch in SPECIAL_CHARS:
dw = SPECIAL_CHARS[ch]
ret = ret + dw
- elif ch.isprintable():
- ret = ret + ch
else:
- # escaped_value = ch.encode('unicode-escape').decode('utf-8')
ordinal = ord(ch)
- hexstr = hex(ordinal)[2:]
- escaped_value = '\\u'
- i = len(hexstr)
- while i < 4:
- escaped_value += '0'
- i += 1
- escaped_value += hexstr
- ret = ret + escaped_value
+ if ordinal <= 0x80 and ch.isprintable():
+ ret = ret + ch
+ else:
+ hexstr = hex(ordinal)[2:]
+ i = len(hexstr)
+ max = 4
+ escaped_value = '\\u'
+ if i > max:
+ max = 8
+ escaped_value = '\\U'
+ while i < max:
+ escaped_value += '0'
+ i += 1
+ escaped_value += hexstr
+ ret = ret + escaped_value
ret = ret + "'"
return ret
diff --git a/tests/tests_api b/tests/tests_api
index 4ee51004b..fa660ef0e 160000
--- a/tests/tests_api
+++ b/tests/tests_api
@@ -1 +1 @@
-Subproject commit 4ee51004b4d83d2c12ca8f6e10faab762cc0262f
+Subproject commit fa660ef0ee019ba9c2da91e7a9140423593d944e
--
GitLab