aboutsummaryrefslogtreecommitdiffstats
path: root/yql/essentials/tests/postgresql/cases/jsonpath_encoding.sql
diff options
context:
space:
mode:
authorAlexander Smirnov <alex@ydb.tech>2024-11-20 11:14:58 +0000
committerAlexander Smirnov <alex@ydb.tech>2024-11-20 11:14:58 +0000
commit31773f157bf8164364649b5f470f52dece0a4317 (patch)
tree33d0f7eef45303ab68cf08ab381ce5e5e36c5240 /yql/essentials/tests/postgresql/cases/jsonpath_encoding.sql
parent2c7938962d8689e175574fc1e817c05049f27905 (diff)
parenteff600952d5dfe17942f38f510a8ac2b203bb3a5 (diff)
downloadydb-31773f157bf8164364649b5f470f52dece0a4317.tar.gz
Merge branch 'rightlib' into mergelibs-241120-1113
Diffstat (limited to 'yql/essentials/tests/postgresql/cases/jsonpath_encoding.sql')
-rw-r--r--yql/essentials/tests/postgresql/cases/jsonpath_encoding.sql42
1 files changed, 42 insertions, 0 deletions
diff --git a/yql/essentials/tests/postgresql/cases/jsonpath_encoding.sql b/yql/essentials/tests/postgresql/cases/jsonpath_encoding.sql
new file mode 100644
index 0000000000..597c153f5c
--- /dev/null
+++ b/yql/essentials/tests/postgresql/cases/jsonpath_encoding.sql
@@ -0,0 +1,42 @@
+--
+-- encoding-sensitive tests for jsonpath
+--
+SELECT getdatabaseencoding(); -- just to label the results files
+-- checks for double-quoted values
+-- basic unicode input
+SELECT '"\u"'::jsonpath; -- ERROR, incomplete escape
+SELECT '"\u00"'::jsonpath; -- ERROR, incomplete escape
+SELECT '"\u000g"'::jsonpath; -- ERROR, g is not a hex digit
+SELECT '"\u0000"'::jsonpath; -- OK, legal escape
+SELECT '"\uaBcD"'::jsonpath; -- OK, uppercase and lower case both OK
+-- handling of unicode surrogate pairs
+select '"\ud83d\ude04\ud83d\udc36"'::jsonpath as correct_in_utf8;
+select '"\ud83d\ud83d"'::jsonpath; -- 2 high surrogates in a row
+select '"\ude04\ud83d"'::jsonpath; -- surrogates in wrong order
+select '"\ud83dX"'::jsonpath; -- orphan high surrogate
+select '"\ude04X"'::jsonpath; -- orphan low surrogate
+--handling of simple unicode escapes
+select '"the Copyright \u00a9 sign"'::jsonpath as correct_in_utf8;
+select '"dollar \u0024 character"'::jsonpath as correct_everywhere;
+select '"dollar \\u0024 character"'::jsonpath as not_an_escape;
+select '"null \u0000 escape"'::jsonpath as not_unescaped;
+select '"null \\u0000 escape"'::jsonpath as not_an_escape;
+-- checks for quoted key names
+-- basic unicode input
+SELECT '$."\u"'::jsonpath; -- ERROR, incomplete escape
+SELECT '$."\u00"'::jsonpath; -- ERROR, incomplete escape
+SELECT '$."\u000g"'::jsonpath; -- ERROR, g is not a hex digit
+SELECT '$."\u0000"'::jsonpath; -- OK, legal escape
+SELECT '$."\uaBcD"'::jsonpath; -- OK, uppercase and lower case both OK
+-- handling of unicode surrogate pairs
+select '$."\ud83d\ude04\ud83d\udc36"'::jsonpath as correct_in_utf8;
+select '$."\ud83d\ud83d"'::jsonpath; -- 2 high surrogates in a row
+select '$."\ude04\ud83d"'::jsonpath; -- surrogates in wrong order
+select '$."\ud83dX"'::jsonpath; -- orphan high surrogate
+select '$."\ude04X"'::jsonpath; -- orphan low surrogate
+--handling of simple unicode escapes
+select '$."the Copyright \u00a9 sign"'::jsonpath as correct_in_utf8;
+select '$."dollar \u0024 character"'::jsonpath as correct_everywhere;
+select '$."dollar \\u0024 character"'::jsonpath as not_an_escape;
+select '$."null \u0000 escape"'::jsonpath as not_unescaped;
+select '$."null \\u0000 escape"'::jsonpath as not_an_escape;