feat: sql parser (#182)

ydb-platform · Sep 10, 2024 · 9b69b30 · 9b69b30
1 parent e09a738
commit 9b69b30
Show file tree

Hide file tree

Showing 2 changed files with 395 additions and 0 deletions.
diff --git a/src/Ydb.Sdk/src/Ado/Internal/SqlParser.cs b/src/Ydb.Sdk/src/Ado/Internal/SqlParser.cs
@@ -0,0 +1,150 @@
+using System.Collections.Concurrent;
+using System.Text;
+
+namespace Ydb.Sdk.Ado.Internal;
+
+internal static class SqlParser
+{
+    private static readonly IDictionary<string, ParsedResult> CacheQueries =
+        new ConcurrentDictionary<string, ParsedResult>();
+
+    internal static ParsedResult Parse(string sql)
+    {
+        if (CacheQueries.TryGetValue(sql, out var preparedYql))
+        {
+            return preparedYql;
+        }
+
+        var newYql = new StringBuilder();
+        var paramNames = new List<string>();
+
+        var prevToken = 0;
+
+        for (var curToken = 0; curToken < sql.Length; curToken++)
+        {
+            switch (sql[curToken])
+            {
+                case '`':
+                    curToken = SkipTerminals(sql, '`', curToken);
+                    break;
+                case '"':
+                    curToken = SkipTerminals(sql, '"', curToken);
+                    break;
+                case '\'':
+                    curToken = SkipTerminals(sql, '\'', curToken);
+                    break;
+                case '-':
+                    if (curToken + 1 < sql.Length && sql[curToken + 1] == '-')
+                    {
+                        while (curToken + 1 < sql.Length)
+                        {
+                            curToken++;
+                            if (sql[curToken] == '\r' || sql[curToken] == '\n')
+                            {
+                                break;
+                            }
+                        }
+                    }
+
+                    break;
+                case '/':
+                    if (curToken + 1 < sql.Length && sql[curToken + 1] == '*')
+                    {
+                        // /* /* */ */ nest, according to SQL spec
+                        var level = 1;
+                        for (curToken += 2; curToken < sql.Length; curToken++)
+                        {
+                            switch (sql[curToken - 1])
+                            {
+                                case '*':
+                                    if (sql[curToken] == '/')
+                                    {
+                                        --level;
+                                        ++curToken; // don't parse / in */* twice
+                                    }
+
+                                    break;
+                                case '/':
+                                    if (sql[curToken] == '*')
+                                    {
+                                        ++level;
+                                        ++curToken; // don't parse * in /*/ twice
+                                    }
+
+                                    break;
+                            }
+
+                            if (level == 0)
+                            {
+                                break;
+                            }
+                        }
+                    }
+
+                    break;
+                case '@':
+                    if (curToken + 1 < sql.Length && sql[curToken + 1] == '@') // $text = @@ a b c @ @@ 
+                    {
+                        for (curToken += 2; curToken + 1 < sql.Length; curToken++)
+                        {
+                            if (sql[curToken] == '@' && sql[curToken + 1] == '@')
+                            {
+                                curToken++;
+                                break;
+                            }
+                        }
+
+                        break;
+                    }
+
+                    // Parse params
+                    newYql.Append(sql[prevToken .. curToken]);
+                    prevToken = ++curToken;
+
+                    for (;
+                         curToken < sql.Length && (char.IsLetterOrDigit(sql[curToken]) || sql[curToken] == '_');
+                         curToken++)
+                    {
+                    }
+
+                    if (curToken - prevToken == 0)
+                    {
+                        throw new YdbException($"Have empty name parameter, invalid SQL [position: {prevToken}]");
+                    }
+
+                    var originalParamName = $"${sql[prevToken .. curToken]}";
+
+                    paramNames.Add(originalParamName);
+                    newYql.Append(originalParamName);
+                    prevToken = curToken;
+
+                    break;
+            }
+        }
+
+        newYql.Append(sql.AsSpan(prevToken, sql.Length - prevToken));
+
+        return CacheQueries[sql] = new ParsedResult(newYql.ToString(), paramNames.ToArray());
+    }
+
+    private static int SkipTerminals(string sql, char stopSymbol, int curToken)
+    {
+        while (++curToken < sql.Length)
+        {
+            if (sql[curToken] == '\\')
+            {
+                ++curToken;
+                continue;
+            }
+
+            if (sql[curToken] == stopSymbol)
+            {
+                return curToken;
+            }
+        }
+
+        return sql.Length;
+    }
+}
+
+internal record ParsedResult(string ParsedSql, IReadOnlyList<string> ParamNames);
diff --git a/src/Ydb.Sdk/tests/Ado/Internal/SqlParserTests.cs b/src/Ydb.Sdk/tests/Ado/Internal/SqlParserTests.cs
@@ -0,0 +1,245 @@
+using Xunit;
+using Ydb.Sdk.Ado.Internal;
+
+namespace Ydb.Sdk.Tests.Ado.Internal;
+
+[Trait("Category", "Unit")]
+public class SqlParserTests
+{
+    [Fact]
+    public void Parse_WhenDigitOrLetterOrUnderliningInParamName_ReturnParsedSqlAndCached()
+    {
+        var (sql, paramNames) = SqlParser.Parse("SELECT @p; SELECT @p2; SELECT @p_3;");
+
+        Assert.Equal("SELECT $p; SELECT $p2; SELECT $p_3;", sql);
+        Assert.Equal(new[] { "$p", "$p2", "$p_3" }, paramNames);
+    }
+
+    [Fact]
+    public void Parse_WhenEmptySql_ReturnEmpty()
+    {
+        var (sql, paramsNames) = SqlParser.Parse("");
+        Assert.Equal("", sql);
+        Assert.True(paramsNames.Count == 0);
+    }
+
+    [Fact]
+    public void Parse_WhenLineComment_ReturnSqlWithComment()
+    {
+        var (sql, paramNames) = SqlParser.Parse(@"-- Comment with params @param, @p2, @p_3
+SELECT @param; SELECT @p2; SELECT @p_3;");
+
+        Assert.Equal(@"-- Comment with params @param, @p2, @p_3
+SELECT $param; SELECT $p2; SELECT $p_3;", sql);
+        Assert.Equal(new[] { "$param", "$p2", "$p_3" }, paramNames);
+    }
+
+    [Fact]
+    public void Parse_WhenBlockComment_ReturnSqlWithComment()
+    {
+        var (sql, paramNames) = SqlParser.Parse(@"/* Comment SQL
+/*
+Comment with params @param, @p2, @p_3
+--
+@
+*/
+SELECT @param; SELECT @p2; SELECT @p_3
+*/
+INSERT INTO Table 
+    (id, bool_column, bigint_column, smallint_column, tinyint_column, float_column, double_column, decimal_column, 
+     uint8_column, uint16_column, uint32_column, uint64_column, text_column, binary_column, json_column,
+     jsondocument_column, date_column, datetime_column, timestamp_column, interval_column) VALUES
+(@name1, @name2, @name3, @name4, @name5, @name6, @name7, @name8, @name9, @name10, @name11, @name12, @name13, @name14,
+ @name15, @name16, @name17, @name18, @name19, @name20);");
+
+        Assert.Equal(@"/* Comment SQL
+/*
+Comment with params @param, @p2, @p_3
+--
+@
+*/
+SELECT @param; SELECT @p2; SELECT @p_3
+*/
+INSERT INTO Table 
+    (id, bool_column, bigint_column, smallint_column, tinyint_column, float_column, double_column, decimal_column, 
+     uint8_column, uint16_column, uint32_column, uint64_column, text_column, binary_column, json_column,
+     jsondocument_column, date_column, datetime_column, timestamp_column, interval_column) VALUES
+($name1, $name2, $name3, $name4, $name5, $name6, $name7, $name8, $name9, $name10, $name11, $name12, $name13, $name14,
+ $name15, $name16, $name17, $name18, $name19, $name20);", sql);
+        Assert.Equal(new[]
+        {
+            "$name1", "$name2", "$name3", "$name4", "$name5", "$name6", "$name7", "$name8", "$name9", "$name10",
+            "$name11", "$name12", "$name13", "$name14", "$name15", "$name16", "$name17", "$name18", "$name19", "$name20"
+        }, paramNames);
+    }
+
+    [Fact]
+    public void Parse_WhenDoubleQuotes_ReturnSql()
+    {
+        var (sql, paramNames) = SqlParser.Parse(@"REPLACE INTO episodes
+(
+    series_id,
+    season_id,
+    episode_id,
+    title,
+    air_date
+)
+VALUES
+(
+    2,
+    5,
+    12,
+    ""@ @ @ @ @ @Test Episode !!! \"" \"" \"" \"" SELECT @param; SELECT @p2; SELECT @p_3"",
+    @air_date
+)
+
+COMMIT;
+
+-- View result:
+SELECT * FROM episodes WHERE series_id = ""123 @ \"" @ @"" AND season_id = @param;
+;");
+        Assert.Equal(@"REPLACE INTO episodes
+(
+    series_id,
+    season_id,
+    episode_id,
+    title,
+    air_date
+)
+VALUES
+(
+    2,
+    5,
+    12,
+    ""@ @ @ @ @ @Test Episode !!! \"" \"" \"" \"" SELECT @param; SELECT @p2; SELECT @p_3"",
+    $air_date
+)
+
+COMMIT;
+
+-- View result:
+SELECT * FROM episodes WHERE series_id = ""123 @ \"" @ @"" AND season_id = $param;
+;", sql);
+        Assert.Equal(new[] { "$air_date", "$param" }, paramNames);
+    }
+
+    [Fact]
+    public void Parse_WhenSingleQuotes_ReturnSql()
+    {
+        var (sql, paramNames) = SqlParser.Parse(@"UPSERT INTO episodes
+(
+    series_id,
+    season_id,
+    episode_id,
+    title,
+    air_date
+)
+VALUES
+(
+    2,
+    5,
+    13,
+    'Test Episode @ \'@ @ \'@ \'@ \'@ @ @ @ @ @ @ @ ',
+    @air_date
+)
+;
+
+COMMIT;
+
+-- View result:
+SELECT * FROM episodes WHERE series_id = '123 @ \' @ @' AND season_id = @param;");
+        Assert.Equal(@"UPSERT INTO episodes
+(
+    series_id,
+    season_id,
+    episode_id,
+    title,
+    air_date
+)
+VALUES
+(
+    2,
+    5,
+    13,
+    'Test Episode @ \'@ @ \'@ \'@ \'@ @ @ @ @ @ @ @ ',
+    $air_date
+)
+;
+
+COMMIT;
+
+-- View result:
+SELECT * FROM episodes WHERE series_id = '123 @ \' @ @' AND season_id = $param;", sql);
+        Assert.Equal(new[] { "$air_date", "$param" }, paramNames);
+    }
+
+    [Fact]
+    public void Parse_WhenBacktickQuotes_ReturnSql()
+    {
+        var (sql, paramNames) = SqlParser.Parse(@"UPSERT INTO `episodes @ @ @ @  "" "" \` @ @ @`
+(
+    series_id,
+    season_id,
+    episode_id,
+    title,
+    air_date
+)
+VALUES
+(
+    2,
+    5,
+    13,
+    'Test Episode @ \'@ @ \'@ \'@ \'@ @ @ @ @ @ @ @ ',
+    @air_date
+)
+;
+
+COMMIT;
+
+-- View result:
+SELECT * FROM episodes WHERE series_id = '123 @ \' @ @' AND season_id = @param;");
+
+        Assert.Equal(@"UPSERT INTO `episodes @ @ @ @  "" "" \` @ @ @`
+(
+    series_id,
+    season_id,
+    episode_id,
+    title,
+    air_date
+)
+VALUES
+(
+    2,
+    5,
+    13,
+    'Test Episode @ \'@ @ \'@ \'@ \'@ @ @ @ @ @ @ @ ',
+    $air_date
+)
+;
+
+COMMIT;
+
+-- View result:
+SELECT * FROM episodes WHERE series_id = '123 @ \' @ @' AND season_id = $param;", sql);
+        Assert.Equal(new[] { "$air_date", "$param" }, paramNames);
+    }
+
+    [Fact]
+    public void Parse_WhenMultilineStringLiterals_ReturnSql()
+    {
+        var (sql, paramNames) = SqlParser.Parse(@"$text = @@some
+multiline with double at: @@@@
+text@@;
+SELECT $text;
+-- Comment with params @param, @p2, @p_3
+SELECT @param; SELECT @p2; SELECT @p_3;");
+
+        Assert.Equal(@"$text = @@some
+multiline with double at: @@@@
+text@@;
+SELECT $text;
+-- Comment with params @param, @p2, @p_3
+SELECT $param; SELECT $p2; SELECT $p_3;", sql);
+        Assert.Equal(new[] { "$param", "$p2", "$p_3" }, paramNames);
+    }
+}