Added/Updated tests\functional\intfunc\unlist\unlist_output_value_has_separator_test.py: Checked on 6.0.0.744 (19-apr-2025) - all fine.

Author: pavel-zotov

tests/functional/intfunc/unlist/unlist_output_value_has_separator_test.py

@@ -0,0 +1,84 @@
+#coding:utf-8
+
+"""
+ID:          issue-8418
+ISSUE:       https://github.com/FirebirdSQL/firebird/issues/8418
+TITLE:  Character <S> must NOT occur in any value generated by UNLIST which gets this <S> as separator.
+DESCRIPTION:
+    Test restores two databases (clones of employee) that contain blob field that must be passed as argument to UNLIST().
+    In both DB blobs were generated as random sequences of unicode characters from any avaliable ranges.
+    None of values generated by UNLIST() must contain separator.
+    But under some circumstatences this rule could be violated, see:
+        https://github.com/FirebirdSQL/firebird/pull/8418#issuecomment-2781057324
+        https://github.com/FirebirdSQL/firebird/pull/8418#issuecomment-2789227832
+NOTES:
+    Confirmed bug on 6.0.0.742 (17-apr-2025): separator presented in some of generated values.
+    Checked on 6.0.0.744 (19-apr-2025) - all fine.
+"""
+
+from pathlib import Path
+import zipfile
+import locale
+import pytest
+from firebird.qa import *
+from firebird.driver import connect
+
+db = db_factory(charset = 'utf8')
+act = python_act('db', substitutions=[('[ \t]+', ' ')])
+
+tmp_fbk = temp_file('unlist_unexpected.tmp.fbk')
+tmp_fdb = temp_file('unlist_unexpected.tmp.fdb')
+
+@pytest.mark.version('>=6.0.0')
+def test_1(act: Action, tmp_fbk: Path, tmp_fdb: Path, capsys):
+
+    test_map = {'unlist-unexpected-6_0_0_716-a5b25c1.fbk' : '0x227b', 'unlist-unexpected-6_0_0_717-fbb6b0c.fbk' : '0x2114'}
+    expected_out = []
+    for i_fbk, i_chr in test_map.items():
+
+        # 716: select u.* from unlist((select blob_fld from t_longblob), unicode_char(0x227b) returning blob character set utf8) as u(x) where x containing unicode_char(0x227b)
+        # 717: select u.x, octet_length(u.x), position(unicode_char(0x2114) in u.x), unicode_char(0x2114) from unlist((select blob_fld from t_longblob), unicode_char(0x2114) returning blob character set utf8) as u(x) where x containing unicode_char(0x2114)
+
+        zipped_fbk_file = zipfile.Path(act.files_dir / 'unlist-unexpected.zip', at = i_fbk)
+        tmp_fbk.write_bytes(zipped_fbk_file.read_bytes())
+
+        act.gbak(switches = ['-rep', str(tmp_fbk), str(tmp_fdb)], combine_output = True, io_enc = locale.getpreferredencoding())
+        assert '' == act.stdout
+        act.reset()
+
+        test_sql = f"""
+            with
+            d as (
+                select
+                     blob_fld
+                    ,unicode_char({i_chr}) as separator
+                from t_longblob
+            )
+            , e as (
+                select * from d, unlist(d.blob_fld, d.separator returning blob character set utf8) as u (unlist_token)
+            )
+            select e.unlist_token, position(e.separator in e.unlist_token) as separator_pos
+            from e
+            where e.unlist_token containing e.separator
+            ;
+        """
+
+        with connect(str(tmp_fdb), user = act.db.user, password = act.db.password, charset = 'utf8') as con:
+            print(i_fbk)
+            expected_out.append(i_fbk)
+            cur = con.cursor()
+            cur.execute(test_sql)
+            data = cur.fetchall()
+            if data:
+                print(f'### ERROR ### separator {i_chr} occurs at least in one record issued by UNLIST():')
+
+                col = cur.description
+                for r in data:
+                    for i in range(len(col)):
+                        print(' '.join((col[i][0], ':', str(r[i]))))
+
+
+    act.expected_stdout = '\n'.join(expected_out)
+
+    act.stdout = capsys.readouterr().out
+    assert act.clean_stdout == act.clean_expected_stdout