Mercurial > repos > aaronquinlan > multi_intersect

comparison BEDTools-Version-2.14.3/src/utils/bedGraphFile/bedGraphFile.h @ 0:dfcd8b6c1bda

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author aaronquinlan
date Thu, 03 Nov 2011 10:25:04 -0400
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:dfcd8b6c1bda
1 /*****************************************************************************
2 bedGraphFile.cpp
3
4 (c) 2010 - Assaf Gordon
5 Hall Laboratory
6 Department of Biochemistry and Molecular Genetics
7 University of Virginia
8 aaronquinlan@gmail.com
9
10 Licenced under the GNU General Public License 2.0 license.
11 ******************************************************************************/
12 #ifndef BEDGRAPHFILE_H
13 #define BEDGRAPHFILE_H
14
15 #include "gzstream.h"
16 #include "lineFileUtilities.h"
17 #include "fileType.h"
18 #include <vector>
19 #include <map>
20 #include <set>
21 #include <string>
22 #include <iostream>
23 #include <fstream>
24 #include <sstream>
25 #include <cstring>
26 #include <algorithm>
27 #include <limits.h>
28 #include <stdint.h>
29 #include <cstdio>
30
31 using namespace std;
32
33 //*************************************************
34 // Data type tydedef
35 //*************************************************
36 #ifndef CHRPOS
37 typedef uint32_t CHRPOS;
38 #endif
39
40 #ifndef DEPTH
41 typedef uint32_t DEPTH;
42 #endif
43
44 /*
45 Structure for regular BedGraph records
46 */
47 template <typename T>
48 class BEDGRAPH
49 {
50 public:
51 std::string chrom;
52 CHRPOS start;
53 CHRPOS end;
54 T depth;
55
56 public:
57 typedef T DEPTH_TYPE;
58 // constructors
59
60 // Null
61 BEDGRAPH() :
62 start(0),
63 end(0),
64 depth(T())
65 {}
66
67 // BEDGraph
68 BEDGRAPH(string _chrom, CHRPOS _start, CHRPOS _end, T _depth) :
69 chrom(_chrom),
70 start(_start),
71 end(_end),
72 depth(_depth)
73 {}
74 }; // BEDGraph
75
76 typedef BEDGRAPH<int32_t> BEDGRAPH_INT;
77 typedef BEDGRAPH<std::string> BEDGRAPH_STR;
78 typedef BEDGRAPH<double> BEDGRAPH_FLOAT;
79
80 template <typename T>
81 std::ostream& operator<< (std::ostream& strm, const BEDGRAPH<T>& bg)
82 {
83 strm << bg.chrom << "\t"
84 << bg.start << "\t"
85 << bg.end << "\t"
86 << bg.depth;
87 return strm;
88 }
89
90 // enum to flag the state of a given line in a BEDGraph file.
91 enum BedGraphLineStatus
92 {
93 BEDGRAPH_INVALID = -1,
94 BEDGRAPH_HEADER = 0,
95 BEDGRAPH_BLANK = 1,
96 BEDGRAPH_VALID = 2
97 };
98
99
100 //************************************************
101 // BedGraphFile Class methods and elements
102 //************************************************
103 class BedGraphFile {
104
105 public:
106
107 // Constructor
108 BedGraphFile(string &);
109
110 // Destructor
111 ~BedGraphFile(void);
112
113 // Open a BEDGraph file for reading (creates an istream pointer)
114 void Open(void);
115
116 // Close an opened BED file.
117 void Close(void);
118
119 // Get the next BED entry in an opened BED file.
120 template <typename T>
121 BedGraphLineStatus GetNextBedGraph (BEDGRAPH<T> &bedgraph, int &lineNum)
122 {
123 // make sure there are still lines to process.
124 // if so, tokenize, validate and return the BED entry.
125 if (_bedGraphStream->good()) {
126 string bedGraphLine;
127 vector<string> bedGraphFields;
128
129 // parse the bedStream pointer
130 getline(*_bedGraphStream, bedGraphLine);
131 if (_bedGraphStream->eof())
132 return BEDGRAPH_INVALID;
133 if (_bedGraphStream->bad()) {
134 cerr << "Error while reading file '" << bedGraphFile << "' : "
135 << strerror(errno) << endl;
136 exit(1);
137 }
138 lineNum++;
139
140 // split into a string vector.
141 Tokenize(bedGraphLine,bedGraphFields);
142
143 // load the BED struct as long as it's a valid BED entry.
144 return parseLine(bedgraph, bedGraphFields, lineNum);
145 }
146
147 // default if file is closed or EOF
148 return BEDGRAPH_INVALID;
149 }
150
151 // the bedfile with which this instance is associated
152 string bedGraphFile;
153
154 private:
155 // data
156 istream *_bedGraphStream;
157
158 template <typename T>
159 BedGraphLineStatus parseLine (BEDGRAPH<T> &bg, const vector<string> &lineVector, int &lineNum)
160 {
161 if (lineVector.size() == 0)
162 return BEDGRAPH_BLANK;
163
164 if (lineVector[0].find("track") != string::npos ||
165 lineVector[0].find("browser") != string::npos ||
166 lineVector[0].find("#") != string::npos)
167 return BEDGRAPH_HEADER;
168
169 if (lineVector.size() != 4)
170 return BEDGRAPH_INVALID;
171
172 bg.chrom = lineVector[0];
173
174 stringstream str_start(lineVector[1]);
175 if (! (str_start >> bg.start) ) {
176 cerr << "Input error, failed to extract start value from '" << lineVector[1]
177 << "' (column 2) in " << bedGraphFile << " line " << lineNum << endl;
178 exit(1);
179 }
180
181 stringstream str_end(lineVector[2]);
182 if (! (str_end >> bg.end) ) {
183 cerr << "Input error, failed to extract end value from '" << lineVector[2]
184 << "' (column 3) in " << bedGraphFile << " line " << lineNum << endl;
185 exit(1);
186 }
187
188 stringstream str_depth(lineVector[3]);
189 if (! (str_depth >> bg.depth) ) {
190 cerr << "Input error, failed to extract depth value from '" << lineVector[3]
191 << "' (column 4) in " << bedGraphFile << " line " << lineNum << endl;
192 exit(1);
193 }
194
195 return BEDGRAPH_VALID;
196 }
197 };
198
199 #endif /* BEDFILE_H */