Better statement splitting

src/grate.core/Infrastructure/IStatementSplitter.cs

@@ -0,0 +1,9 @@
+namespace grate.Infrastructure;
+
+/// <summary>
+/// Splits a batch of SQL text into individual statements to be run separately.
+/// </summary>
+public interface IStatementSplitter
+{
+    IEnumerable<string> Split(string statement);
+}

src/grate.core/Infrastructure/ISyntax.cs

@@ -5,7 +5,7 @@
 /// </summary>
 public interface ISyntax
 {
-    string StatementSeparatorRegex { get; }
+    IStatementSplitter StatementSplitter { get; }
     string CurrentDatabase { get; }
     string ListDatabases { get; }
     string CreateDatabase(string databaseName, string? password);

src/grate.core/Infrastructure/RegexStatementSplitter.cs

@@ -0,0 +1,59 @@
+using System.Text.RegularExpressions;
+using static System.Text.RegularExpressions.RegexOptions;
+
+namespace grate.Infrastructure;
+
+/// <summary>
+/// Base class for splitting SQL batches into statements using a dialect-specific regex.
+/// The regex must tag the parts to keep as KEEP1/KEEP2 groups, and the separator to remove as BATCHSPLITTER.
+/// </summary>
+public abstract class RegexStatementSplitter : IStatementSplitter
+{
+    private const string BatchTerminatorReplacementString = @" |{[_REMOVE_]}| ";
+
+    protected abstract string StringsRegex { get; }
+    protected abstract string DashCommentsRegex { get; }
+    protected abstract string StarCommentsRegex { get; }
+    protected abstract string SeparatorRegex { get; }
+
+    /// <summary>
+    /// Additional regex fragments to be inserted between the strings and comments patterns (e.g. for extra string forms).
+    /// </summary>
+    protected virtual IEnumerable<string> AdditionalRegexes => Enumerable.Empty<string>();
+
+    private Regex? _regex;
+    private Regex Regex => _regex ??= new Regex(Pattern, IgnoreCase | Multiline);
+
+    private string Pattern =>
+        string.Join("|", new[] { StringsRegex }
+            .Concat(AdditionalRegexes)
+            .Append(DashCommentsRegex)
+            .Append(StarCommentsRegex)
+            .Append(SeparatorRegex));
+
+    public IEnumerable<string> Split(string statement)
+    {
+        var replaced = Replace(statement);
+
+        var statements = replaced.Split(BatchTerminatorReplacementString);
+        return statements.Where(HasScriptsToRun);
+    }
+
+    private string Replace(string text) => Regex.Replace(text, ReplaceBatchSeparator);
+
+    private static string ReplaceBatchSeparator(Match match)
+    {
+        var groups = match.Groups;
+        var replacement = groups["BATCHSPLITTER"].Success ? BatchTerminatorReplacementString : string.Empty;
+        return groups["KEEP1"].Value + replacement + groups["KEEP2"].Value;
+    }
+
+    private static bool HasScriptsToRun(string sqlStatement)
+    {
+        var trimmedStatement = sqlStatement.Replace(BatchTerminatorReplacementString, string.Empty, StringComparison.InvariantCultureIgnoreCase);
+        return !string.IsNullOrEmpty(trimmedStatement.ToLower()
+            .Replace(Environment.NewLine, string.Empty)
+            .Replace("\n", string.Empty) // This is necessary to make script with unix-style line endings work on grate on Windows
+            .Replace(" ", string.Empty));
+    }
+}

src/grate.core/Migration/AnsiSqlDatabase.cs

@@ -36,7 +36,6 @@ protected AnsiSqlDatabase(ILogger logger, ISyntax syntax)
     {
         Logger = logger;
         _syntax = syntax;
-        StatementSplitter = new StatementSplitter(syntax);
     }
 
     public string ServerName => Connection.DataSource;
@@ -54,15 +53,13 @@ protected AnsiSqlDatabase(ILogger logger, ISyntax syntax)
     public abstract bool SupportsSchemas { get; }
 
     public virtual bool SplitBatchStatements => true;
-    private StatementSplitter StatementSplitter { get; }
+    private IStatementSplitter StatementSplitter => _syntax.StatementSplitter;
 
     public virtual IEnumerable<string> GetStatements(string sql)
         => SplitBatchStatements ? this.StatementSplitter.Split(sql) : new[] { sql };
 
     public abstract void ThrowScriptFailed(MigrationsFolder folder, string file, string? scriptText, Exception exception);
 
-    public string StatementSeparatorRegex => _syntax.StatementSeparatorRegex;
-
     public string ScriptsRunTable => _syntax.TableWithSchema(SchemaName, ScriptsRunTableName);
     public string ScriptsRunErrorsTable => _syntax.TableWithSchema(SchemaName, ScriptsRunErrorsTableName);
     public string VersionTable => _syntax.TableWithSchema(SchemaName, VersionTableName);

src/grate.mariadb/Infrastructure/MariaDbStatementSplitter.cs

@@ -0,0 +1,11 @@
+using grate.Infrastructure;
+
+namespace grate.MariaDb.Infrastructure;
+
+public class MariaDbStatementSplitter : RegexStatementSplitter
+{
+    protected override string StringsRegex => @"(?<KEEP1>'[^']*')";
+    protected override string DashCommentsRegex => @"(?<KEEP1>--.*$)";
+    protected override string StarCommentsRegex => @"(?<KEEP1>/\*[\S\s]*?\*/)";
+    protected override string SeparatorRegex => @"(?<KEEP1>^|\s)(?<BATCHSPLITTER>GO)(?<KEEP2>\s|;|$)";
+}

src/grate.mariadb/Infrastructure/MariaDbSyntax.cs

@@ -3,17 +3,8 @@ namespace grate.MariaDb.Infrastructure;
 
 public readonly struct MariaDbSyntax : ISyntax
 {
-    public string StatementSeparatorRegex
-    {
-        get
-        {
-            const string strings = @"(?<KEEP1>'[^']*')";
-            const string dashComments = @"(?<KEEP1>--.*$)";
-            const string starComments = @"(?<KEEP1>/\*[\S\s]*?\*/)";
-            const string separator = @"(?<KEEP1>^|\s)(?<BATCHSPLITTER>GO)(?<KEEP2>\s|;|$)";
-            return strings + "|" + dashComments + "|" + starComments + "|" + separator;
-        }
-    }
+    private static readonly IStatementSplitter _statementSplitter = new MariaDbStatementSplitter();
+    public IStatementSplitter StatementSplitter => _statementSplitter;
 
     public string CurrentDatabase => "SELECT DATABASE()";
     public string ListDatabases => "SHOW DATABASES";

src/grate.oracle/Infrastructure/OracleStatementSplitter.cs

@@ -0,0 +1,11 @@
+using grate.Infrastructure;
+
+namespace grate.Oracle.Infrastructure;
+
+public class OracleStatementSplitter : RegexStatementSplitter
+{
+    protected override string StringsRegex => @"(?<KEEP1>'[^']*')";
+    protected override string DashCommentsRegex => @"(?<KEEP1>--.*$)";
+    protected override string StarCommentsRegex => @"(?<KEEP1>/\*[\S\s]*?\*/)";
+    protected override string SeparatorRegex => @"(?<KEEP1>^|\s)(?<BATCHSPLITTER>/)(?<KEEP2>\s|;|$)";
+}

src/grate.oracle/Infrastructure/OracleSyntax.cs

@@ -4,17 +4,8 @@ namespace grate.Oracle.Infrastructure;
 
 public readonly struct OracleSyntax : ISyntax
 {
-    public string StatementSeparatorRegex
-    {
-        get
-        {
-            const string strings = @"(?<KEEP1>'[^']*')";
-            const string dashComments = @"(?<KEEP1>--.*$)";
-            const string starComments = @"(?<KEEP1>/\*[\S\s]*?\*/)";
-            const string separator = @"(?<KEEP1>^|\s)(?<BATCHSPLITTER>/)(?<KEEP2>\s|;|$)";
-            return strings + "|" + dashComments + "|" + starComments + "|" + separator;
-        }
-    }
+    private static readonly IStatementSplitter _statementSplitter = new OracleStatementSplitter();
+    public IStatementSplitter StatementSplitter => _statementSplitter;
 
     public string CurrentDatabase => "select user from dual";
     public string ListDatabases => "SELECT * FROM all_users";

src/grate.postgresql/Infrastructure/PostgreSqlStatementSplitter.cs

@@ -0,0 +1,22 @@
+using grate.Infrastructure;
+
+namespace grate.PostgreSql.Infrastructure;
+
+public class PostgreSqlStatementSplitter : RegexStatementSplitter
+{
+    protected override string StringsRegex => @"(?<KEEP1>'([^']|\'\')*')";
+    protected override string DashCommentsRegex => "(?<KEEP1>--.*$)";
+    protected override string StarCommentsRegex => @"(?<KEEP1>/\*[\S\s]*?\*/)";
+    protected override string SeparatorRegex => "(?<KEEP1>.*)(?<BATCHSPLITTER>(;)(?=(?:[^']|'[^']*')*$))(?<KEEP2>.*)";
+
+    protected override IEnumerable<string> AdditionalRegexes
+    {
+        get
+        {
+            const string backslashEscapedStrings = @"(?<KEEP1>E(?<!\\)('[\S\s]*?(?<!\\)'))";
+            const string dollarQuotedStrings = @"(?<KEEP1>\$(?'tag'\w*)\$[\S\s]*?\$\k'tag'\$)";
+            yield return backslashEscapedStrings;
+            yield return dollarQuotedStrings;
+        }
+    }
+}

src/grate.postgresql/Infrastructure/PostgreSqlSyntax.cs

@@ -3,19 +3,8 @@ namespace grate.PostgreSql.Infrastructure;
 
 public readonly struct PostgreSqlSyntax : ISyntax
 {
-    public string StatementSeparatorRegex
-    {
-        get
-        {
-            const string strings = @"(?<KEEP1>'([^']|\'\')*')";
-            const string backslashEscapedStrings = @"(?<KEEP1>E(?<!\\)('[\S\s]*?(?<!\\)'))";
-            const string dollarQuotedStrings = @"(?<KEEP1>\$(?'tag'\w*)\$[\S\s]*?\$\k'tag'\$)";
-            const string dashComments = "(?<KEEP1>--.*$)";
-            const string starComments = @"(?<KEEP1>/\*[\S\s]*?\*/)";
-            const string separator = "(?<KEEP1>.*)(?<BATCHSPLITTER>(;)(?=(?:[^']|'[^']*')*$))(?<KEEP2>.*)";
-            return strings + "|" + backslashEscapedStrings + "|" + dollarQuotedStrings + "|" + dashComments + "|" + starComments + "|" + separator;
-        }
-    }
+    private static readonly IStatementSplitter _statementSplitter = new PostgreSqlStatementSplitter();
+    public IStatementSplitter StatementSplitter => _statementSplitter;
 
     public string CurrentDatabase => "SELECT current_database()";
     public string ListDatabases => "SELECT datname FROM pg_database";

src/grate.sqlite/Infrastructure/SqliteStatementSplitter.cs

@@ -0,0 +1,11 @@
+using grate.Infrastructure;
+
+namespace grate.Sqlite.Infrastructure;
+
+public class SqliteStatementSplitter : RegexStatementSplitter
+{
+    protected override string StringsRegex => @"(?<KEEP1>'[^']*')";
+    protected override string DashCommentsRegex => @"(?<KEEP1>--.*$)";
+    protected override string StarCommentsRegex => @"(?<KEEP1>/\*[\S\s]*?\*/)";
+    protected override string SeparatorRegex => @"(?<KEEP1>^|\s)(?<BATCHSPLITTER>GO)(?<KEEP2>\s|;|$)";
+}

src/grate.sqlite/Infrastructure/SqliteSyntax.cs

@@ -3,17 +3,8 @@ namespace grate.Sqlite.Infrastructure;
 
 public readonly struct SqliteSyntax : ISyntax
 {
-    public string StatementSeparatorRegex
-    {
-        get
-        {
-            const string strings = @"(?<KEEP1>'[^']*')";
-            const string dashComments = @"(?<KEEP1>--.*$)";
-            const string starComments = @"(?<KEEP1>/\*[\S\s]*?\*/)";
-            const string separator = @"(?<KEEP1>^|\s)(?<BATCHSPLITTER>GO)(?<KEEP2>\s|;|$)";
-            return strings + "|" + dashComments + "|" + starComments + "|" + separator;
-        }
-    }
+    private static readonly IStatementSplitter _statementSplitter = new SqliteStatementSplitter();
+    public IStatementSplitter StatementSplitter => _statementSplitter;
 
     public string CurrentDatabase => "SELECT name FROM pragma_database_list ORDER BY seq DESC LIMIT 1";
     public string ListDatabases => "select name from pragma_database_list";