Issue #357: Fixed error with splitting on batch separator inside quot…

…es (#360)
erikbra · Aug 13, 2023 · 250cdc5 · 250cdc5
1 parent 2e3ddd4
commit 250cdc5
Show file tree

Hide file tree

Showing 2 changed files with 47 additions and 27 deletions.
diff --git a/grate.unittests/Basic/Infrastructure/PostgreSQL/Statement_Splitting/StatementSplitter_.cs b/grate.unittests/Basic/Infrastructure/PostgreSQL/Statement_Splitting/StatementSplitter_.cs
@@ -1,4 +1,5 @@
-using FluentAssertions;
+using System.Linq;
+using FluentAssertions;
 using grate.Infrastructure;
 using grate.Migration;
 using Microsoft.Extensions.Logging.Abstractions;
@@ -36,20 +37,20 @@ USING btree
 
         batches.Should().HaveCount(4);
     }
-
+    
     [Test]
-    public void Ignores_semicolons_in_strings()
+    public void Ignores_semicolons_in_backslash_escaped_strings()
     {
         var original = @"
-DO '
+DO E'
 BEGIN
     IF NOT EXISTS(
         SELECT schema_name
         FROM information_schema.schemata
-        WHERE schema_name = ''random''
+        WHERE schema_name = \'random\'
     )
     THEN
-        EXECUTE ''CREATE SCHEMA random'';
+        EXECUTE \'CREATE SCHEMA random\';
     END IF;
 END
 ';
@@ -59,48 +60,67 @@ FROM information_schema.schemata
         batches.Should().HaveCount(1);
     }
 
-    [Test]
-    public void Ignores_semicolons_in_backslash_escaped_strings()
+    [TestCase("$$")]
+    [TestCase("$sometag$")]
+    public void Ignores_semicolons_in_dollar_quoted_strings(string tag)
     {
-        var original = @"
-DO E'
+        var original = @$"
+DO 
+{tag}
 BEGIN
     IF NOT EXISTS(
         SELECT schema_name
         FROM information_schema.schemata
-        WHERE schema_name = \'random\'
+        WHERE schema_name = 'random'
     )
     THEN
-        EXECUTE \'CREATE SCHEMA random\';
+        EXECUTE 'CREATE SCHEMA random';
     END IF;
 END
-';
+{tag};
 ";
         var batches = Splitter.Split(original);
+        batches.Should().HaveCount(1);
+    }
+
+    [Test]
+    public void Splits_on_semicolon_after_single_quotes_when_there_is_another_semicolon_in_the_quote()
+    {
+        var original = @"SELECT 1 WHERE whatnot = '; ' ; MOO";
+        var batches = Splitter.Split(original).ToList();
+        batches.Should().HaveCount(2);
 
+        batches.First().Should().Be("SELECT 1 WHERE whatnot = '; ' ");
+        batches.Last().Should().Be(" MOO");
+    }
+
+    [Test]
+    public void Ignores_semicolon_in_single_quotes_when_there_is_no_other_semicolon()
+    {
+        var original = @"SELECT 1 WHERE whatnot = '; '";
+        var batches = Splitter.Split(original);
         batches.Should().HaveCount(1);
     }
 
-    [TestCase("$$")]
-    [TestCase("$sometag$")]
-    public void Ignores_semicolons_in_dollar_quoted_strings(string tag)
+    [Test]
+    public void Ignores_semicolons_in_strings()
     {
-        var original = @$"
-DO 
-{tag}
+        var original = @"
+DO '
 BEGIN
     IF NOT EXISTS(
         SELECT schema_name
         FROM information_schema.schemata
-        WHERE schema_name = 'random'
+        WHERE schema_name = ''random''
     )
     THEN
-        EXECUTE 'CREATE SCHEMA random';
+        EXECUTE ''CREATE SCHEMA random'';
     END IF;
 END
-{tag};
+';
 ";
-    var batches = Splitter.Split(original);
+        var batches = Splitter.Split(original);
+
         batches.Should().HaveCount(1);
     }
 }
diff --git a/grate/Infrastructure/PostgreSqlSyntax.cs b/grate/Infrastructure/PostgreSqlSyntax.cs
@@ -7,12 +7,12 @@ public string StatementSeparatorRegex
         get
         {
             const string strings = @"(?<KEEP1>'([^']|\'\')*')";
-            const string backslashEscapedSrings = @"(?<KEEP1>E(?<!\\)('[\S\s]*?(?<!\\)'))";
+            const string backslashEscapedStrings = @"(?<KEEP1>E(?<!\\)('[\S\s]*?(?<!\\)'))";
             const string dollarQuotedStrings = @"(?<KEEP1>\$(?'tag'\w*)\$[\S\s]*?\$\k'tag'\$)";
-            const string dashComments = @"(?<KEEP1>--.*$)";
+            const string dashComments = "(?<KEEP1>--.*$)";
             const string starComments = @"(?<KEEP1>/\*[\S\s]*?\*/)";
-            const string separator = @"(?<KEEP1>.*)(?<BATCHSPLITTER>;)(?<KEEP2>.*)";
-            return strings + "|" + backslashEscapedSrings + "|" + dollarQuotedStrings + "|" + dashComments + "|" + starComments + "|" + separator;
+            const string separator = "(?<KEEP1>.*)(?<BATCHSPLITTER>(;)(?=(?:[^']|'[^']*')*$))(?<KEEP2>.*)";
+            return strings + "|" + backslashEscapedStrings + "|" + dollarQuotedStrings + "|" + dashComments + "|" + starComments + "|" + separator;
         }
     }