Resolves #1252

jgebal · jgebal · commit 962f6fd4446d · 2023-05-21T12:54:53.000+03:00
Replaced all regexp ranges with posix representations to avoid NLS_SORT impact on regex behavior.
diff --git a/source/core/annotations/ut_annotation_manager.pkb b/source/core/annotations/ut_annotation_manager.pkb
@@ -246,7 +246,7 @@ create or replace package body ut_annotation_manager as
         l_sql_clob := regexp_replace(l_sql_clob, '^(.*?\s*create(\s+or\s+replace)?(\s+(editionable|noneditionable))?\s+?)((package|type).*)', '\5', 1, 1, 'ni');
         -- remove "OWNER." from create or replace statement.
         -- Owner is not supported along with AUTHID - see issue https://github.com/utPLSQL/utPLSQL/issues/1088
-        l_sql_clob := regexp_replace(l_sql_clob, '^(package|type)\s+("?[a-zA-Z][a-zA-Z0-9#_$]*"?\.)(.*)', '\1 \3', 1, 1, 'ni');
+        l_sql_clob := regexp_replace(l_sql_clob, '^(package|type)\s+("?[[:alpha:]][[:alnum:]$#_]*"?\.)(.*)', '\1 \3', 1, 1, 'ni');
         l_sql_lines := ut_utils.convert_collection( ut_utils.clob_to_table(l_sql_clob) );
       end if;
       open l_result for
diff --git a/source/core/annotations/ut_annotation_parser.pkb b/source/core/annotations/ut_annotation_parser.pkb
@@ -25,7 +25,7 @@ create or replace package body ut_annotation_parser as
   gc_annot_comment_pattern       constant varchar2(30) := '^( |'||chr(09)||')*-- *('||gc_annotation_qualifier||'.*?)$'; -- chr(09) is a tab character
   gc_comment_replacer_patter     constant varchar2(50) := '{COMMENT#%N%}';
   gc_comment_replacer_regex_ptrn constant varchar2(25) := '{COMMENT#(\d+)}';
-  gc_regexp_identifier           constant varchar2(50) := '[a-zA-Z][a-zA-Z0-9#_$]*';
+  gc_regexp_identifier           constant varchar2(50) := '[[:alpha:]][[:alnum:]$#_]*';
   gc_annotation_block_pattern    constant varchar2(200) := '(({COMMENT#.+}'||chr(10)||')+)( |'||chr(09)||')*(procedure|function)\s+(' ||
                                                            gc_regexp_identifier || ')';
   gc_annotation_pattern          constant varchar2(50) := gc_annotation_qualifier || gc_regexp_identifier || '[ '||chr(9)||']*(\(.*?\)\s*?$)?';
diff --git a/source/core/types/ut_executable_test.tpb b/source/core/types/ut_executable_test.tpb
@@ -159,7 +159,7 @@ create or replace type body ut_executable_test as
       if self.error_stack is null then
         l_fail_message := 'Expected one of exceptions ('||l_expected_error_codes||') but nothing was raised.';
       else
-        l_actual_error_no := regexp_substr(self.error_stack, '^[a-zA-Z]{3}(-[0-9]+)', subexpression=>1);
+        l_actual_error_no := regexp_substr(self.error_stack, '^[[:alpha:]]{3}(-[0-9]+)', subexpression=>1);
         if not l_actual_error_no member of a_expected_error_codes or l_actual_error_no is null then
           l_fail_message := 'Actual: '||l_actual_error_no||' was expected to ';
           if cardinality(a_expected_error_codes) > 1 then
diff --git a/source/core/ut_expectation_processor.pkb b/source/core/ut_expectation_processor.pkb
@@ -159,17 +159,17 @@ create or replace package body ut_expectation_processor as
     -- when 11g and 12c reports only package name
     function cut_header_and_expectations( a_stack varchar2 ) return varchar2 is
     begin
-      return regexp_substr( a_stack, '(.*\.(UT_EQUAL|UT_BE_WITHIN[A-Z0-9#_$]*|UT_EXPECTATION[A-Z0-9#_$]*|UT|UTASSERT2?)(\.[A-Z0-9#_$]+)?\s+)+((.|\s)*)', 1, 1, 'm', 4);
+      return regexp_substr( a_stack, '(.*\.(UT_EQUAL|UT_BE_WITHIN[[:alnum:]$#_]*|UT_EXPECTATION[[:alnum:]$#_]*|UT|UTASSERT2?)(\.[[:alnum:]$#_]+)?\s+)+((.|\s)*)', 1, 1, 'm', 4);
     end;
     function cut_address_columns( a_stack varchar2 ) return varchar2 is
     begin
-      return regexp_replace( a_stack, '^(0x)?[0-9a-f]+\s+', '', 1, 0, 'mi' );
+      return regexp_replace( a_stack, '^(0x)?[[:digit:]abcdef]+\s+', '', 1, 0, 'mi' );
     end;
     function cut_framework_stack( a_stack varchar2 ) return varchar2 is
     begin
       return regexp_replace(
         a_stack,
-        '[0-9]+\s+anonymous\s+block\s+[0-9]+\s+package\s+body\s+sys\.dbms_sql(\.execute)?\s+[0-9]+\s+[0-9_$#a-z ]+\.ut_executable.*',
+        '[0-9]+\s+anonymous\s+block\s+[0-9]+\s+package\s+body\s+sys\.dbms_sql(\.execute)?\s+[0-9]+\s+[[:alnum:]_$# ]+\.ut_executable.*',
         '',
         1, 1, 'mni'
         );
@@ -178,7 +178,7 @@ create or replace package body ut_expectation_processor as
     begin
       return regexp_replace(
         a_stack,
-        '([0-9]+)\s+(.* )?((anonymous block)|(([0-9_$#a-z]+\.[0-9_$#a-z]+(\.([0-9_$#a-z])+)?)))',
+        '([0-9]+)\s+(.* )?((anonymous block)|(([[:alnum:]$#_]+\.[[:alnum:]$#_]+(\.([[:alnum:]$#_])+)?)))',
         'at "\3", line \1', 1, 0, 'i'
         );
     end;
@@ -190,8 +190,8 @@ create or replace package body ut_expectation_processor as
     l_caller_stack_line    := regexp_substr(l_call_stack,'^(.*)');
     if l_caller_stack_line like '%.%' then
       l_line_no          := to_number( regexp_substr( l_caller_stack_line, ', line (\d+)', subexpression => 1 ) );
-      l_owner            := regexp_substr( l_caller_stack_line, 'at "([A-Za-z0-9$#_]+)\.(([A-Za-z0-9$#_]+)(\.([A-Za-z0-9$#_]+))?)", line (\d+)', subexpression => 1 );
-      l_object_name      := regexp_substr( l_caller_stack_line, 'at "([A-Za-z0-9$#_]+)\.(([A-Za-z0-9$#_]+)(\.([A-Za-z0-9$#_]+))?)", line (\d+)', subexpression => 3 );
+      l_owner            := regexp_substr( l_caller_stack_line, 'at "([[:alnum:]$#_]+)\.(([[:alnum:]$#_]+)(\.([[:alnum:]$#_]+))?)", line (\d+)', subexpression => 1 );
+      l_object_name      := regexp_substr( l_caller_stack_line, 'at "([[:alnum:]$#_]+)\.(([[:alnum:]$#_]+)(\.([[:alnum:]$#_]+))?)", line (\d+)', subexpression => 3 );
       l_result :=
         l_caller_stack_line || ' ' || rtrim(ut_metadata.get_source_definition_line(l_owner, l_object_name, l_line_no),chr(10))
         || replace( l_call_stack, l_caller_stack_line );
diff --git a/source/core/ut_metadata.pkb b/source/core/ut_metadata.pkb
@@ -304,7 +304,7 @@ create or replace package body ut_metadata as
   begin
     l_result := regexp_substr(
       a_full_object_name,
-      '^([A-Za-z0-9$#_]+|".*?")\.([A-Za-z0-9$#_]+|".*?")', subexpression => 2
+      '^([[:alnum:]$#_]+|".*?")\.([[:alnum:]$#_]+|".*?")', subexpression => 2
       );
     if not l_result like '"%"' then
       l_result := upper(l_result);
diff --git a/source/core/ut_suite_cache_manager.pkb b/source/core/ut_suite_cache_manager.pkb
@@ -88,7 +88,7 @@ create or replace package body ut_suite_cache_manager is
   end;
 
   function group_paths_by_schema(a_paths ut_varchar2_list) return ut_path_items is
-    c_package_path_regex constant varchar2(100) := '^([A-Za-z0-9$#_]+)(\.([A-Za-z0-9$#_\*]+))?(\.([A-Za-z0-9$#_\*]+))?$';
+    c_package_path_regex constant varchar2(100) := '^([[:alnum:]$#_]+)(\.([[:alnum:]$#_\*]+))?(\.([[:alnum:]$#_\*]+))?$';
     l_results            ut_path_items := ut_path_items();
     l_path_item          ut_path_item;
     i                    pls_integer;
@@ -157,7 +157,7 @@ create or replace package body ut_suite_cache_manager is
         from schema_paths sp left outer join ut_suite_cache c on
         ( c.object_owner = upper(sp.schema_name) 
         --and c.path like replace(sp.suite_path,'*','%'))
-        and regexp_like(c.path,'^'||replace(sp.suite_path,'*','[A-Za-z0-9$#_]*')))
+        and regexp_like(c.path,'^'||replace(sp.suite_path,'*','[[:alnum:]$#_]*')))
         where sp.suite_path is not null and instr(sp.suite_path,'*') > 0
       ),
       straigth_suite_paths as (
diff --git a/source/core/ut_suite_manager.pkb b/source/core/ut_suite_manager.pkb
@@ -32,7 +32,7 @@ create or replace package body ut_suite_manager is
       for i in 1 .. a_paths.count loop
         l_path := a_paths(i);
         if l_path is null or not (
-          regexp_like(l_path, '^[A-Za-z0-9$#_\*]+(\.[A-Za-z0-9$#_\*]+){0,2}$') or regexp_like(l_path, '^([A-Za-z0-9$#_]+)?:[A-Za-z0-9$#_\*]+(\.[A-Za-z0-9$#_\*]+)*$')) then
+          regexp_like(l_path, '^[[:alnum:]$#_\*]+(\.[[:alnum:]$#_\*]+){0,2}$') or regexp_like(l_path, '^([[:alnum:]$#_]+)?:[[:alnum:]$#_\*]+(\.[[:alnum:]$#_\*]+)*$')) then
           raise_application_error(ut_utils.gc_invalid_path_format, 'Invalid path format: ' || nvl(l_path, 'NULL'));
         end if;
       end loop;
@@ -61,9 +61,9 @@ create or replace package body ut_suite_manager is
 
     for i in 1 .. a_paths.count loop
       --if path is suite-path
-      if regexp_like(a_paths(i), '^([A-Za-z0-9$#_]+)?:') then
+      if regexp_like(a_paths(i), '^([[:alnum:]$#_]+)?:') then
       --get schema name / path
-        l_schema := regexp_substr(a_paths(i), '^([A-Za-z0-9$#_]+)?:',subexpression => 1);
+        l_schema := regexp_substr(a_paths(i), '^([[:alnum:]$#_]+)?:',subexpression => 1);
         -- transform ":path1[.path2]" to "schema:path1[.path2]"
         if l_schema is not null then
           l_schema := sys.dbms_assert.schema_name(upper(l_schema));
@@ -78,7 +78,7 @@ create or replace package body ut_suite_manager is
         -- Object name or procedure is allowed to have filter char
         -- However this is not allowed on schema
         begin
-          l_object := regexp_substr(a_paths(i), '^[A-Za-z0-9$#_\*]+');
+          l_object := regexp_substr(a_paths(i), '^[[:alnum:]$#_\*]+');
           l_schema := sys.dbms_assert.schema_name(upper(l_object));
         exception
           when sys.dbms_assert.invalid_schema_name then
diff --git a/source/core/ut_utils.pkb b/source/core/ut_utils.pkb
@@ -19,9 +19,9 @@ create or replace package body ut_utils is
   /**
   * Constants regex used to validate XML name
   */
-  gc_invalid_first_xml_char  constant varchar2(50)  := '[^_a-zA-Z]';
-  gc_invalid_xml_char        constant varchar2(50)  := '[^_a-zA-Z0-9\.-]';
-  gc_full_valid_xml_name     constant varchar2(50)  := '^([_a-zA-Z])([_a-zA-Z0-9\.-])*$';
+  gc_invalid_first_xml_char  constant varchar2(50)  := '[^_[:alpha:]]';
+  gc_invalid_xml_char        constant varchar2(50)  := '[^_[:alnum:]\.-]';
+  gc_full_valid_xml_name     constant varchar2(50)  := '^([[:alpha:]])([_[:alnum:]\.-])*$';
   gc_owner_hash              constant integer(11)   := dbms_utility.get_hash_value( ut_owner(), 0, power(2,31)-1);
 
   
diff --git a/test/ut3_tester/core/test_suite_manager.pkb b/test/ut3_tester/core/test_suite_manager.pkb
@@ -2217,5 +2217,22 @@ end;]';
     end loop;
   end;
 
+  --%test(Path validation does not fail on Estonian NLS_SORT - fix #1252)
+  procedure path_validate_nls_sort is
+    l_schema_names   ut3_develop.ut_varchar2_rows;
+  begin
+    --Arrange
+    execute immediate q'[alter session set nls_sort='estonian']';
+    --Act
+    l_schema_names := ut3_develop.ut_suite_manager.get_schema_names(ut3_develop.ut_varchar2_list('ut3'));
+    --Asseert
+    ut.expect(sqlcode).to_equal(0);
+  end;
+
+
+  procedure reset_nls_sort is
+  begin
+    execute immediate q'[alter session set nls_sort='binary']';
+  end;
 end test_suite_manager;
 /
diff --git a/test/ut3_tester/core/test_suite_manager.pks b/test/ut3_tester/core/test_suite_manager.pks
@@ -237,5 +237,15 @@ create or replace package test_suite_manager is
 
   --%endcontext
 
+  --%context(paths validation)
+
+  --%test(Path validation does not fail on Estonian NLS_SORT - fix #1252)
+  --%aftertest(reset_nls_sort)
+  procedure path_validate_nls_sort;
+
+  procedure reset_nls_sort;
+
+  --%endcontext
+
 end test_suite_manager;
 /
