namespace TextParser.Tests; using Parsing; using Parsing.Schema; using Parsing.Schema.BuildingBlocks; using Parsing.Tokenization; public class TextParserTests { private const string testInput1 = "2 4 6 8"; private const string testInput2 = "2 ab ba 8 cd dc"; private const string testInput3 = @"2 4 6 1 3 5 7 2 4 6 8 3"; private const string testInput4 = @"2 ab ba fd er sd 8 cd dc 7 uh 6 yp rt"; private const string testInput5 = @"asdfnums(2,5,3)ght cv strs(test) jh 4,3,2 34,54,2nums(2,8) strs(aa,ab,ba,bb)aa,bb"; [Fact] public void LineParser_TestSimpleRepetition() { var schemaBuilder = new InputSchemaBuilder(); var schema = schemaBuilder .Repeat(4) .Expect(InputType.Integer) .EndRepetition() .Build(); var parser = new LineParser(schema); var tokens = parser.ParseLine(testInput1); Assert.Equal(4, tokens.Count); Assert.Equal(InputType.Integer, tokens[0].GetInputType()); Assert.Equal(InputType.Integer, tokens[1].GetInputType()); Assert.Equal(InputType.Integer, tokens[2].GetInputType()); Assert.Equal(InputType.Integer, tokens[3].GetInputType()); Assert.Equal(2, (tokens[0] as IntegerToken)?.GetValue()); Assert.Equal(4, (tokens[1] as IntegerToken)?.GetValue()); Assert.Equal(6, (tokens[2] as IntegerToken)?.GetValue()); Assert.Equal(8, (tokens[3] as IntegerToken)?.GetValue()); } [Fact] public void LineParser_TestSimpleInput() { var schemaBuilder = new InputSchemaBuilder(); var schema = schemaBuilder .Expect(InputType.Integer) .Expect(InputType.Integer) .Expect(InputType.Integer) .Expect(InputType.Integer) .Build(); var parser = new LineParser(schema); var tokens = parser.ParseLine(testInput1); Assert.Equal(4, tokens.Count); Assert.Equal(InputType.Integer, tokens[0].GetInputType()); Assert.Equal(InputType.Integer, tokens[1].GetInputType()); Assert.Equal(InputType.Integer, tokens[2].GetInputType()); Assert.Equal(InputType.Integer, tokens[3].GetInputType()); Assert.Equal(2, (tokens[0] as IntegerToken)?.GetValue()); Assert.Equal(4, (tokens[1] as IntegerToken)?.GetValue()); Assert.Equal(6, (tokens[2] as IntegerToken)?.GetValue()); Assert.Equal(8, (tokens[3] as IntegerToken)?.GetValue()); } [Fact] public void LineParser_TestNestedRepetition() { var schemaBuilder = new InputSchemaBuilder(); var schema = schemaBuilder .Repeat(2) .Expect(InputType.Integer) .Repeat(2) .Expect(InputType.String) .EndRepetition() .EndRepetition() .Build(); var parser = new LineParser(schema); var tokens = parser.ParseLine(testInput2); Assert.Equal(6, tokens.Count); Assert.Equal(InputType.Integer, tokens[0].GetInputType()); Assert.Equal(InputType.String, tokens[1].GetInputType()); Assert.Equal(InputType.String, tokens[2].GetInputType()); Assert.Equal(InputType.Integer, tokens[3].GetInputType()); Assert.Equal(InputType.String, tokens[4].GetInputType()); Assert.Equal(InputType.String, tokens[5].GetInputType()); Assert.Equal(2, (tokens[0] as IntegerToken)?.GetValue()); Assert.Equal("ab", (tokens[1] as StringToken)?.GetValue()); Assert.Equal("ba", (tokens[2] as StringToken)?.GetValue()); Assert.Equal(8, (tokens[3] as IntegerToken)?.GetValue()); Assert.Equal("cd", (tokens[4] as StringToken)?.GetValue()); Assert.Equal("dc", (tokens[5] as StringToken)?.GetValue()); } [Fact] public void TextParser_TestRepetitionAsRows() { var schemaBuilder = new InputSchemaBuilder(); var schema = schemaBuilder .Repeat(4) .Expect(InputType.Integer) .EndRepetition() .Build(); var parser = new TextParser(schema); var rows = parser .SetInputText(testInput3) .Parse() .AsRows(); Assert.Equal(3, rows.Count); Assert.Equal(4, rows[0].Length); Assert.Equal(2, rows[0][0]); Assert.Equal(4, rows[0][1]); Assert.Equal(6, rows[0][2]); Assert.Equal(1, rows[0][3]); Assert.Equal(3, rows[1][0]); Assert.Equal(5, rows[1][1]); Assert.Equal(7, rows[1][2]); Assert.Equal(2, rows[1][3]); Assert.Equal(4, rows[2][0]); Assert.Equal(6, rows[2][1]); Assert.Equal(8, rows[2][2]); Assert.Equal(3, rows[2][3]); } [Fact] public void TextParser_TestRepetitionAsColumns() { var schemaBuilder = new InputSchemaBuilder(); var schema = schemaBuilder .Repeat(4) .Expect(InputType.Integer) .EndRepetition() .Build(); var parser = new TextParser(schema); var columns = parser .SetInputText(testInput3) .Parse() .AsColumns(); Assert.Equal(4, columns.Count); Assert.Equal(3, columns[0].Length); Assert.Equal(2, columns[0][0]); Assert.Equal(3, columns[0][1]); Assert.Equal(4, columns[0][2]); Assert.Equal(4, columns[1][0]); Assert.Equal(5, columns[1][1]); Assert.Equal(6, columns[1][2]); Assert.Equal(6, columns[2][0]); Assert.Equal(7, columns[2][1]); Assert.Equal(8, columns[2][2]); Assert.Equal(1, columns[3][0]); Assert.Equal(2, columns[3][1]); Assert.Equal(3, columns[3][2]); } [Fact] public void TextParser_TestGreedyRepetitionAsRows() { var schemaBuilder = new InputSchemaBuilder(); var schema = schemaBuilder .Repeat() .Expect(InputType.Integer) .Repeat() .Expect(InputType.String) .EndRepetition() .EndRepetition() .Build(); var parser = new TextParser(schema); var rows = parser .SetInputText(testInput4) .Parse() .AsRawData(); Assert.Equal(3, rows.Count); Assert.Equal(6, rows[0].Count); Assert.Equal(3, rows[1].Count); Assert.Equal(5, rows[2].Count); Assert.Equal(InputType.Integer, rows[0][0].GetInputType()); Assert.Equal(InputType.String, rows[0][1].GetInputType()); Assert.Equal(InputType.String, rows[0][2].GetInputType()); Assert.Equal(InputType.String, rows[0][3].GetInputType()); Assert.Equal(InputType.String, rows[0][4].GetInputType()); Assert.Equal(InputType.String, rows[0][5].GetInputType()); Assert.Equal(InputType.Integer, rows[1][0].GetInputType()); Assert.Equal(InputType.String, rows[1][1].GetInputType()); Assert.Equal(InputType.String, rows[1][2].GetInputType()); Assert.Equal(InputType.Integer, rows[2][0].GetInputType()); Assert.Equal(InputType.String, rows[2][1].GetInputType()); Assert.Equal(InputType.Integer, rows[2][2].GetInputType()); Assert.Equal(InputType.String, rows[2][3].GetInputType()); Assert.Equal(InputType.String, rows[2][4].GetInputType()); } [Fact] public void FragmentParser_SimpleTest() { var schemaBuilder = new FragmentSchemaBuilder(); var schema = schemaBuilder .StartOptions() .Option() .Expect("nums(") .Expect(InputType.Integer, "num") .Repeat() .Expect(",") .Expect(InputType.Integer, "num") .EndRepetition() .Expect(")") .Option() .Expect("strs(") .Expect(InputType.String, "str") .Repeat() .Expect(",") .Expect(InputType.String, "str") .EndRepetition() .Expect(")") .EndOptions() .Build(); var parser = TextParser.Create(schema); var fragmentData = parser .SetInputText(testInput5) .Parse() .AsFragments(); var convertedData = fragmentData .ConvertAll((Fragment f) => { int numSum = 0; foreach (var numString in f["num"]) { numSum += int.Parse(numString); } return f["num"].Count + f["str"].Count + numSum; }); Assert.Equal(4, fragmentData.Count); Assert.Equal(3, fragmentData[0]["num"].Count); Assert.Single(fragmentData[1]["str"]); Assert.Equal(2, fragmentData[2]["num"].Count); Assert.Equal(4, fragmentData[3]["str"].Count); Assert.Equal(13, convertedData[0]); Assert.Equal(1, convertedData[1]); Assert.Equal(12, convertedData[2]); Assert.Equal(4, convertedData[3]); } }