Minor cleanup

[bamtools.git] / src / utils / bamtools_utilities.cpp

diff --git a/src/utils/bamtools_utilities.cpp b/src/utils/bamtools_utilities.cpp
index 7772587e84c10fe2289535244cd4c723f3986f54..f79faacfaec80917b6e6cc726794563d7ea76c7a 100644 (file)
--- a/src/utils/bamtools_utilities.cpp
+++ b/src/utils/bamtools_utilities.cpp
@@ -1,26 +1,69 @@
 // ***************************************************************************
 // bamtools_utilities.cpp (c) 2010 Derek Barnett, Erik Garrison
 // Marth Lab, Department of Biology, Boston College
-// All rights reserved.
 // ---------------------------------------------------------------------------
-// Last modified: 2 June 2010
+// Last modified: 9 June 2011
 // ---------------------------------------------------------------------------
 // Provides general utilities used by BamTools sub-tools.
 // ***************************************************************************
 
-#include <cstdlib>
-#include <sys/stat.h> 
-#include "bamtools_utilities.h"
-#include "BamReader.h"
-#include "BamMultiReader.h"
+#include <api/BamMultiReader.h>
+#include <api/BamReader.h>
+#include <utils/bamtools_utilities.h>
+using namespace BamTools;
 
+#include <algorithm>
+#include <cstdlib>
+#include <cstring>
+#include <fstream>
+#include <iostream>
+#include <sstream>
 using namespace std;
-using namespace BamTools;
+
+namespace BamTools {
+  
+const char REVCOMP_LOOKUP[] = {'T',  0,  'G', 'H',
+                                0,   0,  'C', 'D',
+                                0,   0,   0,   0,
+                               'K', 'N',  0,   0,
+                                0,  'Y', 'W', 'A',
+                               'A', 'B', 'S', 'X',
+                               'R',  0 };
+  
+} // namespace BamTools 
+  
+// returns true if 'source' contains 'pattern'
+bool Utilities::Contains(const string& source, const string& pattern) {
+    return ( source.find(pattern) != string::npos );
+}
+
+// returns true if 'source' contains 'c'
+bool Utilities::Contains(const std::string &source, const char c) {
+    return ( source.find(c) != string::npos );
+}
+
+// returns true if 'source' ends with 'pattern'
+bool Utilities::EndsWith(const string& source, const string& pattern) {
+    return ( source.find(pattern) == (source.length() - pattern.length()) );
+}
+
+// returns true if 'source' ends with 'c'
+bool Utilities::EndsWith(const std::string& source, const char c) {
+    return ( source.find(c) == (source.length() - 1) );
+}
+
+// check if a file exists
+bool Utilities::FileExists(const string& filename) {
+    ifstream f(filename.c_str(), ifstream::in);
+    return !f.fail();
+}
 
 // Parses a region string, does validation (valid ID's, positions), stores in Region struct
 // Returns success (true/false)
-bool Utilities::ParseRegionString(const std::string& regionString, const BamReader& reader, BamRegion& region) {
-  
+bool Utilities::ParseRegionString(const string& regionString,
+                                  const BamReader& reader,
+                                  BamRegion& region)
+{
     // -------------------------------
     // parse region string
   
@@ -126,8 +169,10 @@ bool Utilities::ParseRegionString(const std::string& regionString, const BamRead
 }
 
 // Same as ParseRegionString() above, but accepts a BamMultiReader
-bool Utilities::ParseRegionString(const std::string& regionString, const BamMultiReader& reader, BamRegion& region) {
-  
+bool Utilities::ParseRegionString(const string& regionString,
+                                  const BamMultiReader& reader,
+                                  BamRegion& region)
+{
     // -------------------------------
     // parse region string
   
@@ -229,13 +274,60 @@ bool Utilities::ParseRegionString(const std::string& regionString, const BamMult
     region.LeftPosition = startPos;
     region.RightRefID = stopRefID;;
     region.RightPosition = stopPos;
-
     return true;
 }
 
-bool Utilities::FileExists(const std::string& filename) { 
+void Utilities::Reverse(string& sequence) {
+    reverse(sequence.begin(), sequence.end());
+}
+
+void Utilities::ReverseComplement(string& sequence) {
+    
+    // do complement, in-place
+    size_t seqLength = sequence.length();
+    for ( size_t i = 0; i < seqLength; ++i )
+        sequence.replace(i, 1, 1, REVCOMP_LOOKUP[(int)sequence.at(i) - 65]);
+    
+    // reverse it
+    Reverse(sequence);
+}
+
+vector<string> Utilities::Split(const string& source, const char delim) {
+
+    stringstream ss(source);
+    string field;
+    vector<string> fields;
+
+    while ( getline(ss, field, delim) )
+        fields.push_back(field);
+    return fields;
+}
+
+vector<string> Utilities::Split(const string& source, const string& delims) {
+
+    vector<string> fields;
+
+    char* tok;
+    char* cchars = new char[source.size()+1];
+    char* cstr = &cchars[0];
+    strcpy(cstr, source.c_str());
+    tok = strtok(cstr, delims.c_str());
+    while (tok != NULL) {
+        fields.push_back(tok);
+        tok = strtok(NULL, delims.c_str());
+    }
 
-    struct stat fileInfo; 
-    return stat(filename.c_str(), &fileInfo) == 0;
+    delete[] cchars;
+
+    return fields;
+}
+
+// returns true if 'source' starts with 'pattern'
+bool Utilities::StartsWith(const string& source, const string& pattern) {
+    return ( source.find(pattern) == 0 );
+}
 
+// returns true if 'source' starts with 'c'
+bool Utilities::StartsWith(const std::string &source, const char c) {
+    return ( source.find(c) == 0 );
 }