add support for picard-style .bam/.bai indexes

.gitignore

@@ -0,0 +1,5 @@
+bin/
+build/
+include/
+lib/
+src/toolkit/bamtools_version.h

src/api/BamIndex.h

@@ -39,6 +39,7 @@ class API_EXPORT BamIndex {
         // list of supported BamIndex types
         enum IndexType { BAMTOOLS = 0
                        , STANDARD
+                       , PICARD
                        };
 
     // ctor & dtor

src/api/internal/index/BamIndexFactory_p.cpp

@@ -22,6 +22,11 @@ const string BamIndexFactory::CreateIndexFilename(const string& bamFilename,
     switch ( type ) {
         case ( BamIndex::STANDARD ) : return ( bamFilename + BamStandardIndex::Extension() );
         case ( BamIndex::BAMTOOLS ) : return ( bamFilename + BamToolsIndex::Extension() );
+    case ( BamIndex::PICARD ) : {
+            string picardIndexFilename = bamFilename;
+            picardIndexFilename[picardIndexFilename.size()-1] = 'i'; // .bai instead of .bam.bai
+            return picardIndexFilename;
+        }
         default :
             return string();
     }
@@ -37,8 +42,9 @@ BamIndex* BamIndexFactory::CreateIndexFromFilename(const string& indexFilename,
         return 0;
 
     // create index based on extension
-    if      ( extension == BamStandardIndex::Extension() ) return new BamStandardIndex(reader);
-    else if ( extension == BamToolsIndex::Extension()    ) return new BamToolsIndex(reader);
+    if      ( extension == BamStandardIndex::BamDotExtension() ) return new BamStandardIndex(reader);
+    else if ( extension == BamStandardIndex::Extension()       ) return new BamStandardIndex(reader);
+    else if ( extension == BamToolsIndex::BamDotExtension()    ) return new BamToolsIndex(reader);
     else
         return 0;
 }
@@ -69,8 +75,36 @@ const string BamIndexFactory::FileExtension(const string& filename) {
     if ( lastDotPosition == string::npos )
         return string();
 
+    // now determine if we have .bam.* or just .bai
+    const size_t secondLastDotPosition = filename.find_last_of('.', lastDotPosition);
+
+    string ending;
+
+    if ( secondLastDotPosition != string::npos ) {
+        // it's possible we have a .bam.* ending
+        if ( filename.size() - secondLastDotPosition == 8) {
+            // and we do
+            if ( filename.substr(secondLastDotPosition, 5) == ".bam." ) {
+                ending = filename.substr(secondLastDotPosition);
+            }
+        }
+    }
+
+    // we didn't find a .bam.* ending, so we'll assume a bare .bai
+    if ( ending.empty() )
+        ending = filename.substr(lastDotPosition);
+
     // return substring from last dot position
-    return filename.substr(lastDotPosition);
+    return ending;
+}
+
+// tests if a file exists
+const bool BamIndexFactory::FileExists(const std::string& filename) {
+    struct stat buf;
+    if (stat(filename.c_str(), &buf) != -1) {
+        return true;
+    }
+    return false;
 }
 
 // returns name of existing index file that corresponds to @bamFilename
@@ -86,19 +120,24 @@ const string BamIndexFactory::FindIndexFilename(const string& bamFilename,
     // try to find index of preferred type first
     // return index filename if found
     string indexFilename = CreateIndexFilename(bamFilename, preferredType);
-    if ( !indexFilename.empty() )
+    if ( !indexFilename.empty() && FileExists(indexFilename) )
         return indexFilename;
 
     // couldn't find preferred type, try the other supported types
     // return index filename if found
     if ( preferredType != BamIndex::STANDARD ) {
         indexFilename = CreateIndexFilename(bamFilename, BamIndex::STANDARD);
-        if ( !indexFilename.empty() )
+        if ( !indexFilename.empty() && FileExists(indexFilename) )
             return indexFilename;
     }
     if ( preferredType != BamIndex::BAMTOOLS ) {
         indexFilename = CreateIndexFilename(bamFilename, BamIndex::BAMTOOLS);
-        if ( !indexFilename.empty() )
+        if ( !indexFilename.empty() && FileExists(indexFilename) )
+            return indexFilename;
+    }
+    if ( preferredType != BamIndex::PICARD ) {
+        indexFilename = CreateIndexFilename(bamFilename, BamIndex::PICARD);
+        if ( !indexFilename.empty() && FileExists(indexFilename) )
             return indexFilename;
     }
 

src/api/internal/index/BamIndexFactory_p.h

@@ -12,6 +12,7 @@
 
 #include "api/BamIndex.h"
 #include <string>
+#include <sys/stat.h>
 
 namespace BamTools {
 namespace Internal {
@@ -41,6 +42,9 @@ class BamIndexFactory {
                                                      const BamIndex::IndexType& type);
         // retrieves file extension (including '.')
         static const std::string FileExtension(const std::string& filename);
+
+    private:
+        static const bool FileExists(const std::string& filename);
 };
 
 } // namespace Internal

src/api/internal/index/BamStandardIndex_p.cpp

@@ -29,6 +29,7 @@ using namespace std;
 const int BamStandardIndex::MAX_BIN               = 37450;  // =(8^6-1)/7+1
 const int BamStandardIndex::BAM_LIDX_SHIFT        = 14;
 const string BamStandardIndex::BAI_EXTENSION      = ".bai";
+const string BamStandardIndex::BAM_BAI_EXTENSION  = ".bam.bai";
 const char* const BamStandardIndex::BAI_MAGIC     = "BAI\1";
 const int BamStandardIndex::SIZEOF_ALIGNMENTCHUNK = sizeof(uint64_t)*2;
 const int BamStandardIndex::SIZEOF_BINCORE        = sizeof(uint32_t) + sizeof(int32_t);
@@ -420,6 +421,11 @@ bool BamStandardIndex::Create(void) {
 }
 
 // returns format's file extension
+const string BamStandardIndex::BamDotExtension(void) {
+    return BamStandardIndex::BAM_BAI_EXTENSION;
+}
+
+// alternative file extension
 const string BamStandardIndex::Extension(void) {
     return BamStandardIndex::BAI_EXTENSION;
 }
@@ -934,7 +940,7 @@ void BamStandardIndex::WriteLinearOffsets(const int& refId, BaiLinearOffsetVecto
     // make sure linear offsets are sorted before writing & saving summary
     SortLinearOffsets(linearOffsets);
 
-    int64_t numBytesWritten = 0;
+    uint64_t numBytesWritten = 0;
 
     // write number of linear offsets
     int32_t offsetCount = linearOffsets.size();

src/api/internal/index/BamStandardIndex_p.h

@@ -127,6 +127,7 @@ class BamStandardIndex : public BamIndex {
     public:
         // returns format's file extension
         static const std::string Extension(void);
+        static const std::string BamDotExtension(void);
 
     // internal methods
     private:
@@ -225,6 +226,7 @@ class BamStandardIndex : public BamIndex {
         static const int MAX_BIN;
         static const int BAM_LIDX_SHIFT;
         static const std::string BAI_EXTENSION;
+        static const std::string BAM_BAI_EXTENSION;
         static const char* const BAI_MAGIC;
         static const int SIZEOF_ALIGNMENTCHUNK;
         static const int SIZEOF_BINCORE;

src/api/internal/index/BamToolsIndex_p.cpp

@@ -30,7 +30,8 @@ using namespace std;
 // --------------------------------
 
 const uint32_t BamToolsIndex::DEFAULT_BLOCK_LENGTH = 1000;
-const string BamToolsIndex::BTI_EXTENSION     = ".bti";
+const string BamToolsIndex::BTI_EXTENSION     = ".bam.bti";
+const string BamToolsIndex::BAM_BTI_EXTENSION = ".bam.bti";
 const char* const BamToolsIndex::BTI_MAGIC    = "BTI\1";
 const int BamToolsIndex::SIZEOF_BLOCK         = sizeof(int32_t)*2 + sizeof(int64_t);
 
@@ -275,6 +276,11 @@ bool BamToolsIndex::Create(void) {
 }
 
 // returns format's file extension
+const std::string BamToolsIndex::BamDotExtension(void) {
+    return BamToolsIndex::BAM_BTI_EXTENSION;
+}
+
+// just the bare extension
 const std::string BamToolsIndex::Extension(void) {
     return BamToolsIndex::BTI_EXTENSION;
 }

src/api/internal/index/BamToolsIndex_p.h

@@ -122,6 +122,7 @@ class BamToolsIndex : public BamIndex {
     public:
         // returns format's file extension
         static const std::string Extension(void);
+        static const std::string BamDotExtension(void);
 
     // internal methods
     private:
@@ -176,6 +177,7 @@ class BamToolsIndex : public BamIndex {
     private:
         static const uint32_t DEFAULT_BLOCK_LENGTH;
         static const std::string BTI_EXTENSION;
+        static const std::string BAM_BTI_EXTENSION;
         static const char* const BTI_MAGIC;
         static const int SIZEOF_BLOCK;
 };