10 Commits
0.6.0 ... 0.9.0

Author SHA1 Message Date
4c67e8efb0 release: version 0.9.0 🚀
Some checks failed
Upload Python Package / Create Release (push) Successful in 17s
Upload Python Package / deploy (push) Failing after 4m23s
2024-12-13 13:13:50 +01:00
e83e99758a feat: add filter option to TokenConverter, ref: NOISSUE 2024-12-13 13:13:43 +01:00
81ac797b4c release: version 0.8.0 🚀
Some checks failed
Upload Python Package / Create Release (push) Successful in 17s
Upload Python Package / deploy (push) Failing after 4m6s
2024-12-12 20:03:14 +01:00
e9aa60524c feat: adding sensible index constructors refs: NOISSUE 2024-12-12 20:03:07 +01:00
7e5ab9f799 release: version 0.7.2 🚀
Some checks failed
Upload Python Package / Create Release (push) Successful in 17s
Upload Python Package / deploy (push) Failing after 4m24s
2024-12-06 00:33:43 +01:00
fc137ebd03 fix: add some missing API methods, ref: NOISSUE 2024-12-06 00:33:37 +01:00
1b2e9ad1ee release: version 0.7.1 🚀
Some checks failed
Upload Python Package / Create Release (push) Successful in 17s
Upload Python Package / deploy (push) Failing after 4m25s
2024-12-06 00:03:23 +01:00
550c8280a6 fix: allow for parsing single chars as input, ref: NOISSUE 2024-12-05 23:58:11 +01:00
Simon Diesenreiter
b261773b13 release: version 0.7.0 🚀
Some checks failed
Upload Python Package / Create Release (push) Successful in 13s
Upload Python Package / deploy (push) Failing after 4m26s
2024-12-05 23:36:29 +01:00
Simon Diesenreiter
85d94eca2f feat: implement generic data set manipulator, ref: NOISSUE 2024-12-05 23:36:23 +01:00
27 changed files with 913 additions and 23 deletions

View File

@@ -4,6 +4,47 @@ Changelog
(unreleased)
------------
- Feat: add filter option to TokenConverter, ref: NOISSUE. [Simon
Diesenreiter]
0.8.0 (2024-12-12)
------------------
- Feat: adding sensible index constructors refs: NOISSUE. [Simon
Diesenreiter]
0.7.2 (2024-12-05)
------------------
Fix
~~~
- Add some missing API methods, ref: NOISSUE. [Simon Diesenreiter]
Other
~~~~~
0.7.1 (2024-12-05)
------------------
Fix
~~~
- Allow for parsing single chars as input, ref: NOISSUE. [Simon
Diesenreiter]
Other
~~~~~
0.7.0 (2024-12-05)
------------------
- Feat: implement generic data set manipulator, ref: NOISSUE. [Simon
Diesenreiter]
0.6.0 (2024-12-03)
------------------
- Feat: enable named literals, ref: NOISSUE. [Simon Diesenreiter]

View File

@@ -1,8 +1,8 @@
namespace TextParser.Tests;
using Parsing;
using Parsing.Data;
using Parsing.Schema;
using Parsing.Schema.BuildingBlocks;
using Parsing.Tokenization;
public class TextParserTests
@@ -26,6 +26,13 @@ public class TextParserTests
private const string testInput7 = @"adfdf1()324ddf3()()()svsdvs
davnsldkvjs2()()m23423()()()
mcsodkcn owdjnfj 1() asdfnad 23234 2()() sdvsdv";
private const string testInput8 = @"2 4 6 4 1 3 5 4 7 2 4 6 8 3";
private const string testInput9 = @"2 4 6 4 1
3 5 4 7 6
4 6 8 3 9";
private const string testInput10 = @"abc
bca
cab";
[Fact]
public void LineParser_TestSimpleRepetition()
@@ -295,11 +302,11 @@ public class TextParserTests
if (saw1)
{
indicator += 1;
}
}
if (saw2)
{
indicator += 2;
}
}
if (saw3)
{
indicator += 4;
@@ -315,4 +322,97 @@ public class TextParserTests
Assert.Equal(1, convertedData[4]);
Assert.Equal(2, convertedData[5]);
}
[Fact]
public void DataManipulator_SimpleOneDimensionalTest()
{
var schemaBuilder = new InputSchemaBuilder();
var schema = schemaBuilder
.Repeat()
.Expect(InputType.Integer)
.EndRepetition()
.Build();
var parser = new TextParser<InputSchemaContext>(schema);
var row = parser
.SetInputText(testInput8)
.Parse()
.AsSingleStream<int>();
var searchSequence = new List<int> { 4, 6 };
var manipulator = DefaultOneDimensionalManipulator.Create(row);
var searchResults = manipulator.FindInSet(searchSequence);
Assert.Equal(3, searchResults.Count);
Assert.Equal(1, searchResults[0].DataIndex.GetIndices()[0]);
Assert.Equal(3, searchResults[1].DataIndex.GetIndices()[0]);
Assert.Equal(10, searchResults[2].DataIndex.GetIndices()[0]);
Assert.Equal(Direction.Forward, searchResults[0].Direction);
Assert.Equal(Direction.Backward, searchResults[1].Direction);
Assert.Equal(Direction.Forward, searchResults[2].Direction);
}
[Fact]
public void DataManipulator_SimpleTwoDimensionalTest()
{
var schemaBuilder = new InputSchemaBuilder();
var schema = schemaBuilder
.Repeat()
.Expect(InputType.Integer)
.EndRepetition()
.Build();
var parser = new TextParser<InputSchemaContext>(schema);
var row = parser
.SetInputText(testInput9)
.Parse()
.AsListRows<int>();
var searchSequence = new List<int> { 4, 6 };
var manipulator = DefaultTwoDimensionalManipulator.Create(row);
var searchResults = manipulator.FindInSet(searchSequence);
Assert.Equal(6, searchResults.Count);
Assert.Equal(0, searchResults[0].DataIndex.GetIndices()[0]);
Assert.Equal(0, searchResults[0].DataIndex.GetIndices()[1]);
Assert.Equal(2, searchResults[1].DataIndex.GetIndices()[0]);
Assert.Equal(1, searchResults[1].DataIndex.GetIndices()[1]);
Assert.Equal(2, searchResults[2].DataIndex.GetIndices()[0]);
Assert.Equal(1, searchResults[2].DataIndex.GetIndices()[1]);
Assert.Equal(1, searchResults[3].DataIndex.GetIndices()[0]);
Assert.Equal(2, searchResults[3].DataIndex.GetIndices()[1]);
Assert.Equal(3, searchResults[4].DataIndex.GetIndices()[0]);
Assert.Equal(2, searchResults[4].DataIndex.GetIndices()[1]);
Assert.Equal(3, searchResults[5].DataIndex.GetIndices()[0]);
Assert.Equal(2, searchResults[5].DataIndex.GetIndices()[1]);
Assert.Equal(Direction.E, searchResults[0].Direction);
Assert.Equal(Direction.N, searchResults[1].Direction);
Assert.Equal(Direction.SW, searchResults[2].Direction);
Assert.Equal(Direction.E, searchResults[3].Direction);
Assert.Equal(Direction.SE, searchResults[4].Direction);
Assert.Equal(Direction.W, searchResults[5].Direction);
}
[Fact]
public void TextPArser_TestReadingChars()
{
var schemaBuilder = new InputSchemaBuilder();
var schema = schemaBuilder
.Repeat()
.Expect(InputType.Char)
.EndRepetition()
.Build();
var parser = new TextParser<InputSchemaContext>(schema);
var row = parser
.SetInputText(testInput10)
.Parse()
.AsListRows<string>();
Assert.Equal(3, row.Count);
Assert.Equal("a", row[0][0]);
Assert.Equal(3, row[0].Count);
Assert.Equal(3, row[1].Count);
Assert.Equal(3, row[2].Count);
}
}

View File

@@ -0,0 +1,49 @@
namespace Parsing.Data;
using Parsing;
using Parsing.Tokenization;
public static class DataConversionHelpers
{
public static List<TNewType> ConvertData<TTokenType, TNewType, TOldType>(this List<IToken> tokenList, Func<TOldType, TNewType> converter) where TTokenType : IValueToken<TOldType>
{
var newList = new List<TNewType>();
foreach (var token in tokenList)
{
var typedToken = token as IValueToken<TOldType>;
if (typedToken == null)
{
throw new Exception("Invalid Token type encountered during value conversion");
}
newList.Add(converter(typedToken.GetValue()));
}
return newList;
}
public static List<TNewType> ConvertData<TTokenType, TNewType, TOldType>(this List<IToken> tokenList, Func<TOldType, List<TNewType>> converter) where TTokenType : IValueToken<TOldType>
{
var newList = new List<TNewType>();
foreach (var token in tokenList)
{
var typedToken = token as IValueToken<TOldType>;
if (typedToken == null)
{
throw new Exception("Invalid Token type encountered during value conversion");
}
newList.AddRange(converter(typedToken.GetValue()));
}
return newList;
}
public static List<List<TNewType>> ConvertData<TTokenType, TNewType, TOldType>(this List<List<IToken>> tokenListList, Func<TOldType, TNewType> converter) where TTokenType : IValueToken<TOldType>
{
var newListList = new List<List<TNewType>>();
foreach (var tokenList in tokenListList)
{
newListList.Add(tokenList.ConvertData<TTokenType, TNewType, TOldType>(converter));
}
return newListList;
}
}

View File

@@ -0,0 +1,43 @@
namespace Parsing.Data;
public static class DataManipulationHelpers
{
public static TType ReduceData<TType>(this List<TType> data, Func<TType, TType, TType> reducer)
{
if (data.Count < 2)
{
return data[0];
}
TType result = data[0];
for (int i = 1; i < data.Count; i++)
{
result = reducer(result, data[i]);
}
return result;
}
public static TType ReduceData<TType>(this List<TType> data, Func<List<TType>, TType> reducer)
{
return reducer(data);
}
public static List<TNewType> TransformData<TType, TNewType>(this List<TType> data, Func<TType, TNewType> transformer)
{
var newList = new List<TNewType>();
foreach (TType dataItem in data)
{
newList.Add(transformer(dataItem));
}
return newList;
}
public static List<TNewType> TransformData<TType, TNewType>(this List<List<TType>> data, Func<List<TType>, TNewType> transformer)
{
var newList = new List<TNewType>();
foreach (List<TType> dataItemList in data)
{
newList.Add(transformer(dataItem));
}
return newList;
}
}

View File

@@ -0,0 +1,154 @@
using Parsing.Data;
namespace Parsing.Data;
public class SearchResult<TIndexType>
{
public SearchResult(IDataIndex<TIndexType> dataIndex)
{
this.DataIndex = dataIndex;
}
public IDataIndex<TIndexType>? DataIndex { get; set; }
}
public class DirectionalSearchResult<TIndexType> : SearchResult<TIndexType>
{
public DirectionalSearchResult(IDataIndex<TIndexType> dataIndex, Direction direction, int length): base(dataIndex)
{
this.Direction = direction;
this.Length = length;
}
public Direction Direction { get; set; }
public int Length { get; set; }
}
public abstract class DataSetManipulatorBase<TCollectedType, TDataType, TIndexType> where TDataType : IEquatable<TDataType>
{
protected IDataSetIndexer<TDataType, TIndexType> indexer;
protected List<TCollectedType> dataSet;
public DataSetManipulatorBase(List<TCollectedType> dataSet, IDataSetIndexer<TDataType, TIndexType> indexer)
{
this.indexer = indexer;
this.dataSet = dataSet;
}
// we do not know how to iterate a specific data set exactly, the implementation has to take care of validating directional input
protected abstract Direction ValidDirections();
protected void ValidateDirection(Direction d)
{
var allValidDirections = this.ValidDirections();
var isValid = ((d | allValidDirections) == allValidDirections) && ((d & allValidDirections) > 0);
if (!isValid)
{
throw new ArgumentException("Invalid search direction provided for given data set!");
}
}
protected List<Direction> SimplifyDirections(Direction d)
{
this.ValidateDirection(d);
var allDirections = DirectionProvider.GetAllDirections();
var singleDirections = new List<Direction>();
foreach (Direction direction in allDirections)
{
if ((direction & d) > 0)
{
singleDirections.Add(direction);
}
}
return singleDirections;
}
public List<Direction> GetValidDirectionList(Direction d)
{
return SimplifyDirections(this.ValidDirections());
}
// we do not know how to iterate a specific data set exactly, the implementation has to take care of ending traversal in any direction
public abstract bool IsValidIndex(IDataIndex<TIndexType> queryPosition);
// we do not know how to iterate a specific data set exactly, the implementation has to take care of traversing the set
public abstract IDataIndex<TIndexType> Move(IDataIndex<TIndexType> currentPosition, Direction direction);
public List<IDataIndex<TIndexType>> GetNeighborIndices(IDataIndex<TIndexType> currentPosition, Direction directions)
{
var singleDirections = this.SimplifyDirections(directions);
var neighbors = new List<IDataIndex<TIndexType>>();
foreach (var direction in singleDirections)
{
var newPosition = this.Move(currentPosition, direction);
if (this.IsValidIndex(newPosition))
{
neighbors.Add(newPosition);
}
}
return neighbors;
}
// we do not know how to iterate a specific data set exactly, but we only need to find specific items to be able to continue with any other algorithm
public abstract List<SearchResult<TIndexType>> FindInSet(TDataType data);
public List<DirectionalSearchResult<TIndexType>> FindAtPosition(IDataIndex<TIndexType> currentPosition, List<TDataType> data)
{
return this.FindAtPosition(currentPosition, data, this.ValidDirections());
}
public List<DirectionalSearchResult<TIndexType>> FindAtPosition(IDataIndex<TIndexType> currentPosition, List<TDataType> data, Direction directions)
{
var results = new List<DirectionalSearchResult<TIndexType>>();
var givenDirections = this.SimplifyDirections(directions);
if (EqualityComparer<TDataType>.Default.Equals(this.indexer.Get(this.dataSet, currentPosition), data[0]))
{
// found valid search start point, now validate each given direction
foreach (var direction in givenDirections)
{
int searchIndex = 1;
var searchPosition = this.Move(currentPosition, direction); ;
while (searchIndex < data.Count && this.IsValidIndex(searchPosition)
&& EqualityComparer<TDataType>.Default.Equals(this.indexer.Get(this.dataSet, searchPosition), data[searchIndex]))
{
searchPosition = this.Move(searchPosition, direction);
searchIndex++;
}
if (searchIndex == data.Count)
{
var result = new DirectionalSearchResult<TIndexType>(currentPosition, direction, searchIndex);
results.Add(result);
}
}
}
return results;
}
public List<DirectionalSearchResult<TIndexType>> FindInSet(List<TDataType> data, Direction directions)
{
var result = new List<DirectionalSearchResult<TIndexType>>();
// find valid starting points in set and perform search from there
var startingPoints = this.FindInSet(data[0]);
foreach (var startingPoint in startingPoints)
{
foreach (var results in this.FindAtPosition(startingPoint.DataIndex, data, directions))
{
result.AddRange(results);
}
}
return result;
}
public List<DirectionalSearchResult<TIndexType>> FindInSet(List<TDataType> data)
{
return this.FindInSet(data, this.ValidDirections());
}
}

View File

@@ -0,0 +1,62 @@
public class DefaultDataSetIndexer<TDataType> : IDataSetIndexer<TDataType, int>
{
public TDataType Get<TGenericCollectionContentType>(List<TGenericCollectionContentType> collection, IDataIndex<int> index)
{
var indices = index.GetIndices();
return this.GetInternal(collection, indices.ToArray());
}
private TDataType GetInternal<TGenericCollectionContentType>(List<TGenericCollectionContentType> collection, int[] indices)
{
if (indices.Length == 3)
{
return this.GetAtIndex((collection as List<List<List<TDataType>>>), indices[0], indices[1], indices[2]);
}
else if (indices.Length == 2)
{
return this.GetAtIndex((collection as List<List<TDataType>>), indices[0], indices[1]);
}
else if (indices.Length == 1)
{
return this.GetAtIndex((collection as List<TDataType>), indices[0]);
}
else
{
throw new ArgumentException("Invalid Data Set access!");
}
}
public TDataType Get<TGenericCollectionContentType>(List<TGenericCollectionContentType> collection, params int[] indices)
{
return this.GetInternal(collection, indices);
}
public TDataType GetAtIndex(List<TDataType> collection, int index)
{
if (collection == null)
{
throw new ArgumentException("Invalid data set provided for access");
}
return collection[index];
}
public TDataType GetAtIndex(List<List<TDataType>> collection, int x, int y)
{
if (collection == null)
{
throw new ArgumentException("Invalid data set provided for access");
}
return collection[collection.Count - y - 1][x];
}
public TDataType GetAtIndex(List<List<List<TDataType>>> collection, int x, int y, int z)
{
if (collection == null)
{
throw new ArgumentException("Invalid data set provided for access");
}
return collection[z][y][x];
}
}

View File

@@ -0,0 +1,59 @@
using System.Runtime.InteropServices;
using Parsing.Data;
namespace Parsing.Data;
public static class DefaultOneDimensionalManipulator
{
public static DefaultOneDimensionalManipulator<TDataType> Create<TDataType>(List<TDataType> dataSet) where TDataType : IEquatable<TDataType>
{
return new DefaultOneDimensionalManipulator<TDataType>(dataSet);
}
}
public class DefaultOneDimensionalManipulator<TDataType> : DataSetManipulatorBase<TDataType, TDataType, int> where TDataType : IEquatable<TDataType>
{
public DefaultOneDimensionalManipulator(List<TDataType> dataSet) : base(dataSet, new DefaultDataSetIndexer<TDataType>())
{
}
protected override Direction ValidDirections()
{
return (Direction.Left | Direction.Right);
}
public override bool IsValidIndex(IDataIndex<int> queryPosition)
{
var index = queryPosition.GetIndices()[0];
return (index >= 0) && (index < this.dataSet.Count);
}
public override IDataIndex<int> Move(IDataIndex<int> currentPosition, Direction direction)
{
switch (direction)
{
case Direction.Forward:
return new DefaultPositionalDataIndex(currentPosition.GetIndices()[0] + 1);
case Direction.Backward:
return new DefaultPositionalDataIndex(currentPosition.GetIndices()[0] - 1);
default:
throw new ArgumentOutOfRangeException("Direction was not accounted for move for current data set!");
}
}
public override List<SearchResult<int>> FindInSet(TDataType data)
{
var results = new List<SearchResult<int>>();
for (int i = 0; i < this.dataSet.Count; i++)
{
if (EqualityComparer<TDataType>.Default.Equals(this.dataSet[i], data))
{
var singleResult = new SearchResult<int>(new DefaultPositionalDataIndex(i));
results.Add(singleResult);
}
}
return results;
}
}

View File

@@ -0,0 +1,14 @@
public class DefaultPositionalDataIndex : IDataIndex<int>
{
private List<int> indices = new List<int>();
public DefaultPositionalDataIndex(params int[] indices)
{
this.indices.AddRange(indices);
}
public IList<int> GetIndices()
{
return indices;
}
}

View File

@@ -0,0 +1,82 @@
using System.Runtime.InteropServices;
using Parsing.Data;
namespace Parsing.Data;
public static class DefaultTwoDimensionalManipulator
{
public static DefaultTwoDimensionalManipulator<TDataType> Create<TDataType>(List<List<TDataType>> dataSet) where TDataType : IEquatable<TDataType>
{
return new DefaultTwoDimensionalManipulator<TDataType>(dataSet);
}
}
public class DefaultTwoDimensionalManipulator<TDataType> : DataSetManipulatorBase<List<TDataType>, TDataType, int> where TDataType : IEquatable<TDataType>
{
public DefaultTwoDimensionalManipulator(List<List<TDataType>> dataSet) : base(dataSet, new DefaultDataSetIndexer<TDataType>())
{
}
protected override Direction ValidDirections()
{
return (Direction.N
| Direction.NE
| Direction.E
| Direction.SE
| Direction.S
| Direction.SW
| Direction.W
| Direction.NW);
}
public override bool IsValidIndex(IDataIndex<int> queryPosition)
{
var xIndex = queryPosition.GetIndices()[0];
var yIndex = queryPosition.GetIndices()[1];
return (yIndex >= 0) && (yIndex < this.dataSet.Count) && (xIndex >= 0) && (xIndex < this.dataSet[yIndex].Count);
}
public override IDataIndex<int> Move(IDataIndex<int> currentPosition, Direction direction)
{
switch (direction)
{
case Direction.N:
return new DefaultPositionalDataIndex(currentPosition.GetIndices()[0], currentPosition.GetIndices()[1] + 1);
case Direction.NE:
return new DefaultPositionalDataIndex(currentPosition.GetIndices()[0] + 1, currentPosition.GetIndices()[1] + 1);
case Direction.E:
return new DefaultPositionalDataIndex(currentPosition.GetIndices()[0] + 1, currentPosition.GetIndices()[1]);
case Direction.SE:
return new DefaultPositionalDataIndex(currentPosition.GetIndices()[0] + 1, currentPosition.GetIndices()[1] - 1);
case Direction.S:
return new DefaultPositionalDataIndex(currentPosition.GetIndices()[0], currentPosition.GetIndices()[1] - 1);
case Direction.SW:
return new DefaultPositionalDataIndex(currentPosition.GetIndices()[0] - 1, currentPosition.GetIndices()[1] - 1);
case Direction.W:
return new DefaultPositionalDataIndex(currentPosition.GetIndices()[0] - 1, currentPosition.GetIndices()[1]);
case Direction.NW:
return new DefaultPositionalDataIndex(currentPosition.GetIndices()[0] - 1, currentPosition.GetIndices()[1] + 1);
default:
throw new ArgumentOutOfRangeException("Direction was not accounted for move for current data set!");
}
}
public override List<SearchResult<int>> FindInSet(TDataType data)
{
var results = new List<SearchResult<int>>();
for (int y = 0; y < this.dataSet.Count; y++)
{
for (int x = 0; x < this.dataSet[this.dataSet.Count - y - 1].Count; x++)
{
if (EqualityComparer<TDataType>.Default.Equals(this.dataSet[this.dataSet.Count - y - 1][x], data))
{
var singleResult = new SearchResult<int>(new DefaultPositionalDataIndex(x, y));
results.Add(singleResult);
}
}
}
return results;
}
}

View File

@@ -0,0 +1,43 @@
[Flags]
public enum Direction
{
N = 1,
NE = 2,
E = 4,
SE = 8,
S = 16,
SW = 32,
W = 64,
NW = 128,
Horizontal = E | W,
Vertical = N | S,
Cardinal = Horizontal | Vertical,
RisingDiagonal = NE | SW,
FallingDiagonal = NW | SE,
Diagonal = RisingDiagonal | FallingDiagonal,
All = Cardinal | Diagonal,
Left = W,
Right = E,
Up = N,
Down = S,
Forward = Right,
Backward = Left
}
public static class DirectionProvider
{
public static Direction[] GetAllDirections()
{
var directions = new Direction[] {
Direction.N,
Direction.NE,
Direction.E,
Direction.SE,
Direction.S,
Direction.SW,
Direction.W,
Direction.NW
};
return directions;
}
}

View File

@@ -0,0 +1,4 @@
public interface IDataIndex<TIndexType>
{
public IList<TIndexType> GetIndices();
}

View File

@@ -0,0 +1,10 @@
public interface IDataSetIndexer<TDataType, TIndexType>
{
public TDataType Get<TGenericCollectionContentType>(List<TGenericCollectionContentType> collection, IDataIndex<TIndexType> index);
public TDataType Get<TGenericCollectionContentType>(List<TGenericCollectionContentType> collection, params TIndexType[] indices);
public TDataType GetAtIndex(List<TDataType> collection, TIndexType index);
public TDataType GetAtIndex(List<List<TDataType>> collection, TIndexType x, TIndexType y);
public TDataType GetAtIndex(List<List<List<TDataType>>> collection, TIndexType x, TIndexType y, TIndexType z);
}

View File

View File

@@ -0,0 +1,176 @@
namespace Parsing.Data;
using System;
using System.Collections.Generic;
using Parsing.Schema;
using Parsing.Tokenization;
public class TokenConverter
{
protected List<List<IToken>> rawTokens = new List<List<IToken>>();
public TokenConverter()
{
}
private List<T> AsGenericCollection<T, U>() where T : ICollection<U>, new()
{
List<T> returnData = new List<T>();
foreach (var tokenRow in this.rawTokens)
{
T newRow = new T();
foreach (IToken token in tokenRow)
{
if (token == null)
{
throw new Exception("No token was provided, but token was expected!");
}
IValueToken<U>? valueToken = token as IValueToken<U>;
if (valueToken == null)
{
throw new Exception("Provided token is not a ValueToken");
}
newRow.Add(valueToken.GetValue());
}
returnData.Add(newRow);
}
return returnData;
}
private void CheckConversionPrerequisites()
{
// in order to convert rows to columns or grid we require every row to have the same length
int rowLength = this.rawTokens[0].Count;
foreach (var tokenRow in this.rawTokens)
{
if (tokenRow.Count != rowLength)
{
throw new Exception("Attempted to convert token dataset that is not able to be converted!");
}
}
}
public List<T> AsSingleStream<T>()
{
List<T> returnData = new List<T>();
foreach (var tokenRow in this.rawTokens)
{
foreach (IToken token in tokenRow)
{
if (token == null)
{
throw new Exception("No token was provided, but token was expected!");
}
IValueToken<T>? valueToken = token as IValueToken<T>;
if (valueToken == null)
{
throw new Exception("Provided token is not a ValueToken");
}
returnData.Add(valueToken.GetValue());
}
}
return returnData;
}
public List<T[]> AsRows<T>()
{
var listRows = this.AsListRows<T>();
var newList = new List<T[]>();
foreach (var rowList in listRows)
{
newList.Add(rowList.ToArray());
}
return newList;
}
public List<List<T>> AsListRows<T>()
{
return this.AsGenericCollection<List<T>, T>();
}
public List<T[]> AsColumns<T>()
{
var listColumns = this.AsListColumns<T>();
var newList = new List<T[]>();
foreach (var columnList in listColumns)
{
newList.Add(columnList.ToArray());
}
return newList;
}
public List<List<T>> AsListColumns<T>()
{
this.CheckConversionPrerequisites();
var rows = AsListRows<T>();
var columns = new List<List<T>>();
for (int i = 0; i < rows[0].Count; i++)
{
columns.Add(new List<T>());
}
foreach (var row in rows)
{
for (int i = 0; i < row.Count; i++)
{
columns[i].Add(row[i]);
}
}
return columns;
}
public T[][] AsGrid<T>()
{
this.CheckConversionPrerequisites();
var rowsList = AsRows<T>();
return rowsList.ToArray();
}
public List<List<IToken>> AsRawData()
{
return this.rawTokens;
}
public List<Fragment> AsFragments()
{
var items = this.AsSingleStream<Fragment>();
var newList = new List<Fragment>();
foreach (var item in items)
{
var typedItem = item as Fragment;
if (typedItem == null)
{
throw new Exception("Invalid token type encountered");
}
newList.Add(typedItem);
}
return newList;
}
public TokenConverter Filter<T>(params InputType[] inputTypes)
{
var newTokenList = new List<List<IToken>>()
foreach(var token in rawTokens)
{
if(inputTypes.Contains(token.GetInputType()))
{
newTokenList.Add(token);
}
}
this.rawTokens = newTokenList;
return this;
}
}

View File

@@ -4,10 +4,11 @@
public enum BlockType
{
Integer = 1,
String = 2,
Char = 2,
String = 4,
// technically not a block type but keeping it here for consistency/having all types in one place
Fragment = 4,
FixedRepetition = 8,
GreedyRepetition = 16,
NonZeroRepetition = 32,
Fragment = 8,
FixedRepetition = 16,
GreedyRepetition = 32,
NonZeroRepetition = 64,
}

View File

@@ -8,7 +8,7 @@ abstract class BuildingBlockBase : IBuildingBlock
{
}
public abstract IToken ParseWord(InputProvider inputs);
public abstract List<IToken> ParseWord(InputProvider inputs);
public abstract bool CanParseWord(InputProvider inputs);

View File

@@ -0,0 +1,49 @@
namespace Parsing.Schema.BuildingBlocks;
using Parsing.Tokenization;
class CharBlock : BuildingBlockBase
{
public CharBlock()
{
}
public override List<IToken> ParseWord(InputProvider inputs)
{
var tokenList = new List<IToken>();
foreach (char c in inputs.YieldWord())
{
tokenList.Add(new StringToken(c.ToString()));
}
return tokenList;
}
public override bool CanParseWord(InputProvider inputs)
{
string word = string.Empty;
using (inputs.GetLookaheadContext())
{
word = inputs.YieldWord();
}
return this.CanParseWord(word);
}
public override bool CanParseWord(string word)
{
// Here we need to ensure we are not matching any non-string tokens, since string can match pretty much anything
IntegerBlock intBlock = new IntegerBlock();
if (intBlock.CanParseWord(word))
{
return false;
}
return true;
}
public override BlockType GetBlockType()
{
return BlockType.String;
}
}

View File

@@ -20,7 +20,7 @@ class FixedRepetitionBlock : BuildingBlockBase
this.context = this.inputSchema.CreateContext();
}
public override IToken ParseWord(InputProvider inputs)
public override List<IToken> ParseWord(InputProvider inputs)
{
var result = inputSchema.ProcessNextWord(context, inputs);
if (context.HasFinished)
@@ -31,7 +31,7 @@ class FixedRepetitionBlock : BuildingBlockBase
this.context = this.inputSchema.CreateContext();
}
}
return result.Single();
return result;
}
public override bool CanParseWord(InputProvider inputs)

View File

@@ -15,14 +15,14 @@ class GreedyRepetitionBlock : BuildingBlockBase
this.context = this.inputSchema.CreateContext();
}
public override IToken ParseWord(InputProvider inputs)
public override List<IToken> ParseWord(InputProvider inputs)
{
var result = inputSchema.ProcessNextWord(context, inputs);
if (!this.CanParseWord(inputs))
{
this.context = this.inputSchema.CreateContext();
}
return result.Single();
return result;
}
public override bool CanParseWord(InputProvider inputs)

View File

@@ -4,7 +4,7 @@ using Parsing.Tokenization;
public interface IBuildingBlock
{
public IToken ParseWord(InputProvider inputs);
public List<IToken> ParseWord(InputProvider inputs);
public bool CanParseWord(InputProvider inputs);

View File

@@ -9,9 +9,9 @@ class IntegerBlock : BuildingBlockBase
{
}
public override IToken ParseWord(InputProvider inputs)
public override List<IToken> ParseWord(InputProvider inputs)
{
return new IntegerToken(inputs.YieldWord());
return new List<IToken>() { new IntegerToken(inputs.YieldWord()) };
}
public override bool CanParseWord(InputProvider inputs)

View File

@@ -9,9 +9,9 @@ class StringBlock : BuildingBlockBase
{
}
public override IToken ParseWord(InputProvider inputs)
public override List<IToken> ParseWord(InputProvider inputs)
{
return new StringToken(inputs.YieldWord());
return new List<IToken>() { new StringToken(inputs.YieldWord()) };
}
public override bool CanParseWord(InputProvider inputs)

View File

@@ -28,15 +28,13 @@ public class InputSchema : ISchema<InputSchemaContext>
public List<IToken> ProcessNextWord(InputSchemaContext currentContext, InputProvider inputs)
{
var nextBlock = this.buildingBlocks[currentContext.lastProcessedBlockIndex];
var token = nextBlock.ParseWord(inputs);
var tokens = nextBlock.ParseWord(inputs);
if (!nextBlock.IsRepetitionType() || nextBlock.CheckIsDoneParsingAndReset(inputs))
{
currentContext.lastProcessedBlockIndex++;
currentContext.HasFinished = currentContext.lastProcessedBlockIndex >= this.buildingBlocks.Count;
}
var newTokenList = new List<IToken>();
newTokenList.Add(token);
return newTokenList;
return tokens;
}
public bool CanProcessNextWord(InputSchemaContext currentContext, InputProvider inputs)

View File

@@ -21,6 +21,9 @@ public class InputSchemaBuilder : RepetitionSchemaBuilder<InputSchemaBuilder, In
case InputType.Integer:
block = new IntegerBlock();
break;
case InputType.Char:
block = new CharBlock();
break;
default:
throw new Exception("Unrecognized InputType");
}

View File

@@ -6,4 +6,5 @@ public enum InputType
Integer = BlockType.Integer,
String = BlockType.String,
Fragment = BlockType.Fragment,
Char = BlockType.Char,
}

View File

@@ -2,6 +2,7 @@
using System;
using System.Collections.Generic;
using Parsing.Data;
using Parsing.Schema;
using Parsing.Tokenization;

View File

@@ -1 +1 @@
0.6.0
0.9.0