dhtslib.fastq source code

1 module dhtslib.fastq;
2 import dhtslib.bgzf;
3 import std.range.interfaces : InputRange, inputRangeObject; 
4 
5 /*
6  *  Represents a complete fastq read record.
7  */ 
8 struct FastqRecord
9 {
10     string id;
11     string sequence;
12     string extra;
13     string qscores;
14 }
15 
16 /*
17  *  Parses a bgzipped, gzipped, or plain-text fastq file 
18  *  into a range of FastqRecords.
19  */ 
20 struct FastqFile
21 {
22     BGZFile f;
23     InputRange!string lines;
24     FastqRecord rec;
25     bool last;
26 
27     this(string fn)
28     {
29         f = BGZFile(fn);
30         lines = f.byLineCopy.inputRangeObject;
31         popFront;
32     }
33 
34     FastqRecord front()
35     {
36         return rec;
37     }
38 
39     void popFront()
40     {
41         //get id line
42         rec.id = lines.front;
43         lines.popFront;
44 
45         //get seq line
46         rec.sequence = lines.front;
47         lines.popFront;
48 
49         //get extra line
50         rec.extra = lines.front;
51         lines.popFront;
52 
53         //get qscore line
54         rec.qscores = lines.front;
55         if(!lines.empty)
56             lines.popFront;
57         else
58             last = true;
59     }
60 
61     bool empty()
62     {
63         return lines.empty && last;
64     }
65 
66 }
67 
68 ///
69 debug(dhtslib_unittest) unittest
70 {
71     import std.stdio;
72     import htslib.hts_log;
73     import std.algorithm : map;
74     import std.array : array;
75     import std.path : buildPath,dirName;
76     hts_set_log_level(htsLogLevel.HTS_LOG_INFO);
77     hts_log_info(__FUNCTION__, "Testing FastqFile");
78     hts_log_info(__FUNCTION__, "Loading test file");
79 
80     auto fqs = FastqFile(buildPath(dirName(dirName(dirName(__FILE__))),"htslib","test","fastqs.fq"));
81     assert(fqs.array.length == 125);
82     // assert(bg.array == ["122333444455555"]);
83 }