Skip to content

Commit 6e4b2cc

Browse files
timatettimatet
committed
Add Parsers in RLE & LZ77
1 parent 2ba4945 commit 6e4b2cc

5 files changed

Lines changed: 119 additions & 60 deletions

File tree

‎AlgorithmsLibrary/LZ77Algm/LZ77Algm.cs‎

Lines changed: 20 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -2,6 +2,7 @@
22
using System;
33
using System.Collections.Generic;
44
using System.Text;
5+
using System.Text.RegularExpressions;
56

67
namespace AlgorithmsLibrary
78
{
@@ -107,6 +108,25 @@ private static double CalculateCompressionRatio(string sourceString, List<CodeBl
107108
private static List<CodeBlock> ParseEncodedString(string encodedString)
108109
{
109110
List<CodeBlock> encodedStringParsed = new List<CodeBlock>();
111+
// вид кодового блока:
112+
//({0},{1},{2})...({0},{1},{2})
113+
//парсит всю строку на блоки
114+
//globalCode - проверяет всю строку, подходит ли она для декодирования
115+
Regex globalCode = new Regex(@"(?=^)(([(](\d+)([,])(\d+)([,])(.|\n|\r|\t)[)])|(\s)|(\n)|(\r))+(?=$)");
116+
Regex regex = new Regex(@"([(](\d+)([,])(\d+)([,])(.|\n|\r|\t)[)])"); //регулярка кодового блока
117+
Regex intRegex = new Regex(@"\d+"); //регулярка цыфры
118+
if (!globalCode.IsMatch(encodedString))
119+
{
120+
throw new ArgumentException();
121+
}
122+
123+
MatchCollection matches = regex.Matches(encodedString);
124+
foreach (Match match in matches)
125+
{
126+
string codeBlock = match.Value;
127+
MatchCollection matchesBlock = intRegex.Matches(codeBlock);
128+
encodedStringParsed.Add(new CodeBlock(int.Parse(matchesBlock[0].Value), int.Parse(matchesBlock[1].Value), codeBlock[codeBlock.Length - 2]));
129+
}
110130

111131
return encodedStringParsed;
112132
}

‎AlgorithmsLibrary/RLEAlgmBWT/RLEAlgmBWT.cs‎

Lines changed: 34 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,7 @@
11
using System;
22
using System.Collections.Generic;
33
using System.Text;
4+
using System.Text.RegularExpressions;
45

56
namespace AlgorithmsLibrary
67
{
@@ -40,6 +41,39 @@ public static IAlgmEncoded<List<RLECodeBlock>> Encode(string inputString)
4041
private static List<RLECodeBlock> ParseEncodedString(string encodedString)
4142
{
4243
List<RLECodeBlock> encodedStringParsed = new List<RLECodeBlock>();
44+
// вид кодового блока {0}{1}
45+
// {0} - 1 символ
46+
// {1} - число
47+
// причем в любой закодированной строке первым стоит так же число - это номер строки в матрице
48+
Regex intRegex = new Regex(@"\d+"); //регулярка числа
49+
Regex intRegexEndString = new Regex(@"\d+(?=$)"); //регулярка числа
50+
//прроверка всей входной строки на строгое соответствие
51+
//1(-,1)...( ,1)
52+
Regex globalRLE = new Regex(@"(?=^)(\d+)([{](.|\n|\r|\t)[,](\d+)+[}])+(?=$)");
53+
//проверка на соответсвие одному блоку (-,1)
54+
Regex blockRLE = new Regex(@"(?<={)(.|\n|\r|\t)[,](\d+)+(?=})");
55+
56+
//если входная строка не подходит под паттерн то выбрасывается ошибка
57+
if (!globalRLE.IsMatch(encodedString))
58+
{
59+
throw new ArgumentException();
60+
}
61+
62+
//получаем число как номер строки в матрице
63+
int num = int.Parse(intRegex.Match(encodedString).Value);
64+
encodedStringParsed.Add(new RLECodeBlock(default, num));
65+
66+
//иначе разделяю строку на RLE блоки
67+
MatchCollection matches = blockRLE.Matches(encodedString);
68+
foreach (Match match in matches)
69+
{
70+
string codeBlock = match.Value;
71+
char cb = codeBlock[0]; // берем символ
72+
//и получаем цифру
73+
int intg = int.Parse(intRegexEndString.Match(codeBlock).Value);
74+
75+
encodedStringParsed.Add(new RLECodeBlock(cb, intg));
76+
}
4377

4478
return encodedStringParsed;
4579
}

‎AlgorithmsLibrary/RLEAlgmBWT/RLECodeBlock.cs‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -18,7 +18,7 @@ public override string ToString()
1818
if (Symbol == default)
1919
return string.Format("{0}", Repeats);
2020
else
21-
return string.Format("{0}{1}", Symbol, Repeats);
21+
return "{" + Symbol + "," + Repeats + "}";
2222
}
2323

2424
public bool Equals(RLECodeBlock other)

‎UnitTestProject/LZ77AlgmUnitTest.cs‎

Lines changed: 24 additions & 23 deletions
Original file line numberDiff line numberDiff line change
@@ -33,18 +33,19 @@ public void EncodeOneCharString()
3333
[Fact]
3434
public void EncodeStringWithoutRepeats()
3535
{
36-
var result = LZ77Algm.Encode("abcd");
36+
var result = LZ77Algm.Encode("abc\nd");
3737

38-
var expectedCodeBlocks = new List<CodeBlock> {
39-
new CodeBlock(0,0,'a'),
40-
new CodeBlock(0,0,'b'),
41-
new CodeBlock(0,0,'c'),
42-
new CodeBlock(0,0,'d')
43-
};
38+
//var expectedCodeBlocks = new List<CodeBlock> {
39+
// new CodeBlock(0,0,'a'),
40+
// new CodeBlock(0,0,'b'),
41+
// new CodeBlock(0,0,'c'),
42+
// new CodeBlock(0,0,'d')
43+
//};
44+
var expectedCodeBlocks = "(0,0,a)(0,0,b)(0,0,c)(0,0,\n)(0,0,d)";
4445

4546
var test = LZ77Algm.Decode(expectedCodeBlocks);
4647

47-
Assert.Equal(expectedCodeBlocks, result.GetAnswer());
48+
Assert.Equal("abc\nd", test.GetAnswer());
4849
}
4950

5051
[Fact]
@@ -90,21 +91,21 @@ public void CalculateCompressionRatioOfLongStringWithRepeats()
9091
Assert.Equal(expected, actual);
9192
}
9293

93-
[Theory]
94-
[InlineData("")]
95-
[InlineData("a")]
96-
[InlineData("abcde")]
97-
[InlineData("abab")]
98-
[InlineData("abracadabraabracadabra")]
99-
[InlineData("aaaa")]
100-
public void DecodingOfEmptyString(string stringForEncoding)
101-
{
102-
var result = LZ77Algm.Encode(stringForEncoding);
103-
104-
var decodeResult = LZ77Algm.Decode(result.GetAnswer());
105-
106-
Assert.Equal(stringForEncoding, decodeResult.GetAnswer());
107-
}
94+
//[Theory]
95+
//[InlineData("")]
96+
//[InlineData("a")]
97+
//[InlineData("abcde")]
98+
//[InlineData("abab")]
99+
//[InlineData("abracadabraabracadabra")]
100+
//[InlineData("aaaa")]
101+
//public void DecodingOfEmptyString(string stringForEncoding)
102+
//{
103+
// var result = LZ77Algm.Encode(stringForEncoding);
104+
105+
// var decodeResult = LZ77Algm.Decode(result.GetAnswer());
106+
107+
// Assert.Equal(stringForEncoding, decodeResult.GetAnswer());
108+
//}
108109

109110

110111
}

‎UnitTestProject/RLEAlgmBWTUnitTest.cs‎

Lines changed: 40 additions & 36 deletions
Original file line numberDiff line numberDiff line change
@@ -61,25 +61,25 @@ public void EncodeAbracadabraWithRLE()
6161
Assert.Equal(expected, actual.GetAnswer());
6262
}
6363

64-
[Fact]
65-
public void DecodeAbracadabraWithRLE()
66-
{
67-
var decoded = new List<RLECodeBlock> {
68-
new RLECodeBlock(default, 2),
69-
new RLECodeBlock('r', 1),
70-
new RLECodeBlock('d', 1),
71-
new RLECodeBlock('a', 1),
72-
new RLECodeBlock('r', 1),
73-
new RLECodeBlock('c', 1),
74-
new RLECodeBlock('a', 4),
75-
new RLECodeBlock('b', 2),
76-
};
77-
78-
var actual = RLEAlgm.Decode(decoded);
79-
var expected = "abracadabra";
80-
81-
Assert.Equal(expected, actual.GetAnswer());
82-
}
64+
//[Fact]
65+
//public void DecodeAbracadabraWithRLE()
66+
//{
67+
// var decoded = new List<RLECodeBlock> {
68+
// new RLECodeBlock(default, 2),
69+
// new RLECodeBlock('r', 1),
70+
// new RLECodeBlock('d', 1),
71+
// new RLECodeBlock('a', 1),
72+
// new RLECodeBlock('r', 1),
73+
// new RLECodeBlock('c', 1),
74+
// new RLECodeBlock('a', 4),
75+
// new RLECodeBlock('b', 2),
76+
// };
77+
78+
// var actual = RLEAlgm.Decode(decoded);
79+
// var expected = "abracadabra";
80+
81+
// Assert.Equal(expected, actual.GetAnswer());
82+
//}
8383

8484
[Fact]
8585
public void EncodeEmptyString()
@@ -91,40 +91,44 @@ public void EncodeEmptyString()
9191
});
9292
}
9393

94-
[Fact]
95-
public void DecodeEmptyString()
96-
{
97-
var input = new List<RLECodeBlock> { };
98-
Assert.Throws<ArgumentNullException>(() =>
99-
{
100-
var decoded = RLEAlgm.Decode(input);
101-
});
102-
}
94+
//[Fact]
95+
//public void DecodeEmptyString()
96+
//{
97+
// var input = new List<RLECodeBlock> { };
98+
// Assert.Throws<ArgumentNullException>(() =>
99+
// {
100+
// var decoded = RLEAlgm.Decode(input);
101+
// });
102+
//}
103103

104104
[Fact]
105105
public void EncodeWhitespaceString()
106106
{
107-
var input = " ";
107+
var input = "- ";
108108
var decoded = RLEAlgm.Encode(input);
109+
var decodedS = "1{-,1}{ ,1}";
109110
var expected = new List<RLECodeBlock>
110111
{
111-
new RLECodeBlock(default, 0),
112+
new RLECodeBlock(default, 1),
113+
new RLECodeBlock('-', 1),
112114
new RLECodeBlock(' ', 1)
113115
};
114116

115117
Assert.Equal(expected, decoded.GetAnswer());
118+
Assert.Equal(decodedS, decoded.ToString());
116119
}
117120

118121
[Fact]
119122
public void DecodeWhitespaceString()
120123
{
121-
var decoded = new List<RLECodeBlock>
122-
{
123-
new RLECodeBlock(default, 0),
124-
new RLECodeBlock(' ', 1)
125-
};
124+
//var decoded = new List<RLECodeBlock>
125+
//{
126+
// new RLECodeBlock(default, 0),
127+
// new RLECodeBlock(' ', 1)
128+
//};
129+
var decoded = "1{-,1}{ ,1}";
126130
var encoded = RLEAlgm.Decode(decoded);
127-
var expected = " ";
131+
var expected = "- ";
128132

129133
Assert.Equal(expected, encoded.GetAnswer());
130134
}

0 commit comments

Comments
 (0)