version 0.6.1

external/BamTools/api/BamAlignment.cpp → external/BamTools-2.0/api/BamAlignment.cpp

@@ -2,7 +2,7 @@
 // BamAlignment.cpp (c) 2009 Derek Barnett
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 17 October 2011 (DB)
+// Last modified: 4 April 2012 (DB)
 // ---------------------------------------------------------------------------
 // Provides the BamAlignment data structure
 // ***************************************************************************
@@ -25,12 +25,22 @@ using namespace std;
 */
 /*! \var BamAlignment::QueryBases
     \brief 'original' sequence (as reported from sequencing machine)
+
+    \note Setting this field to "*" indicates that the sequence is not to be stored on output.
+    In this case, the contents of the Qualities field should be invalidated as well (cleared or marked as "*").
 */
 /*! \var BamAlignment::AlignedBases
     \brief 'aligned' sequence (includes any indels, padding, clipping)
+
+    This field will be completely empty after reading from BamReader/BamMultiReader when
+    QueryBases is empty.
 */
 /*! \var BamAlignment::Qualities
     \brief FASTQ qualities (ASCII characters, not numeric values)
+
+    \note Setting this field to "*" indicates to BamWriter that the quality scores are not to be stored,
+    but instead will be output as a sequence of '0xFF'. Otherwise, QueryBases must not be a "*" and
+    the length of this field should equal the length of QueryBases.
 */
 /*! \var BamAlignment::TagData
     \brief tag data (use the provided methods to query/modify)
@@ -70,8 +80,12 @@ using namespace std;
     \brief constructor
 */
 BamAlignment::BamAlignment(void)
-    : RefID(-1)
+    : Length(0)
+    , RefID(-1)
     , Position(-1)
+    , Bin(0)
+    , MapQuality(0)
+    , AlignmentFlag(0)
     , MateRefID(-1)
     , MatePosition(-1)
     , InsertSize(0)
@@ -134,36 +148,39 @@ bool BamAlignment::BuildCharData(void) {
     const unsigned int tagDataLength  = dataLength - tagDataOffset;
 
     // check offsets to see what char data exists
-    const bool hasSeqData  = ( seqDataOffset  < dataLength );
-    const bool hasQualData = ( qualDataOffset < dataLength );
+    const bool hasSeqData  = ( seqDataOffset  < qualDataOffset );
+    const bool hasQualData = ( qualDataOffset < tagDataOffset );
     const bool hasTagData  = ( tagDataOffset  < dataLength );
 
-    // set up char buffers
-    const char* allCharData = SupportData.AllCharData.data();
-    const char* seqData     = ( hasSeqData  ? (((const char*)allCharData) + seqDataOffset)  : (const char*)0 );
-    const char* qualData    = ( hasQualData ? (((const char*)allCharData) + qualDataOffset) : (const char*)0 );
-          char* tagData     = ( hasTagData  ? (((char*)allCharData) + tagDataOffset)        : (char*)0 );
-
     // store alignment name (relies on null char in name as terminator)
-    Name.assign((const char*)(allCharData));
+    Name.assign(SupportData.AllCharData.data());
 
     // save query sequence
     QueryBases.clear();
     if ( hasSeqData ) {
+        const char* seqData = SupportData.AllCharData.data() + seqDataOffset;
         QueryBases.reserve(SupportData.QuerySequenceLength);
         for ( size_t i = 0; i < SupportData.QuerySequenceLength; ++i ) {
             const char singleBase = Constants::BAM_DNA_LOOKUP[ ( (seqData[(i/2)] >> (4*(1-(i%2)))) & 0xf ) ];
             QueryBases.append(1, singleBase);
         }
     }
 
-    // save qualities, converting from numeric QV to 'FASTQ-style' ASCII character
+    // save qualities
+
     Qualities.clear();
     if ( hasQualData ) {
-        Qualities.reserve(SupportData.QuerySequenceLength);
-        for ( size_t i = 0; i < SupportData.QuerySequenceLength; ++i ) {
-            const char singleQuality = static_cast<const char>(qualData[i]+33);
-            Qualities.append(1, singleQuality);
+        const char* qualData = SupportData.AllCharData.data() + qualDataOffset;
+
+        // if marked as unstored (sequence of 0xFF) - don't do conversion, just fill with 0xFFs
+        if ( qualData[0] == (char)0xFF )
+            Qualities.resize(SupportData.QuerySequenceLength, (char)0xFF);
+
+        // otherwise convert from numeric QV to 'FASTQ-style' ASCII character
+        else {
+            Qualities.reserve(SupportData.QuerySequenceLength);
+            for ( size_t i = 0; i < SupportData.QuerySequenceLength; ++i )
+                Qualities.append(1, qualData[i]+33);
         }
     }
 
@@ -172,7 +189,7 @@ bool BamAlignment::BuildCharData(void) {
 
     // if QueryBases has data, build AlignedBases using CIGAR data
     // otherwise, AlignedBases will remain empty (this case IS allowed)
-    if ( !QueryBases.empty() ) {
+    if ( !QueryBases.empty() && QueryBases != "*" ) {
 
         // resize AlignedBases
         AlignedBases.reserve(SupportData.QuerySequenceLength);
@@ -231,6 +248,9 @@ bool BamAlignment::BuildCharData(void) {
     // save tag data
     TagData.clear();
     if ( hasTagData ) {
+
+        char* tagData = (((char*)SupportData.AllCharData.data()) + tagDataOffset);
+
         if ( IsBigEndian ) {
             size_t i = 0;
             while ( i < tagDataLength ) {

external/BamTools/api/BamAux.h → external/BamTools-2.0/api/BamAux.h

@@ -2,7 +2,7 @@
 // BamAux.h (c) 2009 Derek Barnett, Michael Str�mberg
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 10 October 2011 (DB)
+// Last modified: 25 October 2011 (DB)
 // ---------------------------------------------------------------------------
 // Provides data structures & utility methods that are used throughout the API.
 // ***************************************************************************
@@ -11,6 +11,7 @@
 #define BAMAUX_H
 
 #include "api/api_global.h"
+#include <cstring>
 #include <fstream> 
 #include <iostream>
 #include <string>
@@ -441,13 +442,25 @@ API_EXPORT inline unsigned short UnpackUnsignedShort(char* buffer) {
     \internal
 */
 struct RaiiBuffer {
+
+    // data members
+    char* Buffer;
+    const size_t NumBytes;
+
+    // ctor & dtor
     RaiiBuffer(const size_t n)
         : Buffer( new char[n]() )
+        , NumBytes(n)
     { }
+
     ~RaiiBuffer(void) {
         delete[] Buffer;
     }
-    char* Buffer;
+
+    // add'l methods
+    void Clear(void) {
+        memset(Buffer, 0, NumBytes);
+    }
 };
 
 } // namespace BamTools

external/BamTools/api/BamIndex.h → external/BamTools-2.0/api/BamIndex.h

@@ -66,6 +66,9 @@ class API_EXPORT BamIndex {
         // loads existing data from file into memory
         virtual bool Load(const std::string& filename) =0;
 
+        // returns the 'type' enum for derived index format
+        virtual BamIndex::IndexType Type(void) const =0;
+
     //! \cond
 
     // internal methods

external/BamTools/api/BamMultiReader.cpp → external/BamTools-2.0/api/BamMultiReader.cpp

@@ -2,7 +2,7 @@
 // BamMultiReader.cpp (c) 2010 Erik Garrison, Derek Barnett
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 10 October 2011 (DB)
+// Last modified: 25 October 2011 (DB)
 // ---------------------------------------------------------------------------
 // Convenience class for reading multiple BAM files.
 //
@@ -13,7 +13,7 @@
 // ***************************************************************************
 
 #include "api/BamMultiReader.h"
-#include "api/internal/BamMultiReader_p.h"
+#include "api/internal/bam/BamMultiReader_p.h"
 using namespace BamTools;
 
 #include <string>

external/BamTools/api/BamMultiReader.h → external/BamTools-2.0/api/BamMultiReader.h

@@ -2,7 +2,7 @@
 // BamMultiReader.h (c) 2010 Erik Garrison, Derek Barnett
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 10 October 2011 (DB)
+// Last modified: 25 October 2011 (DB)
 // ---------------------------------------------------------------------------
 // Convenience class for reading multiple BAM files.
 // ***************************************************************************

external/BamTools/api/BamReader.cpp → external/BamTools-2.0/api/BamReader.cpp

@@ -2,13 +2,13 @@
 // BamReader.cpp (c) 2009 Derek Barnett, Michael Str�mberg
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 10 October 2011 (DB)
+// Last modified: 25 October 2011 (DB)
 // ---------------------------------------------------------------------------
 // Provides read access to BAM files.
 // ***************************************************************************
 
 #include "api/BamReader.h"
-#include "api/internal/BamReader_p.h"
+#include "api/internal/bam/BamReader_p.h"
 using namespace BamTools;
 using namespace BamTools::Internal;
 

external/BamTools/api/BamWriter.cpp → external/BamTools-2.0/api/BamWriter.cpp

@@ -2,15 +2,15 @@
 // BamWriter.cpp (c) 2009 Michael Str�mberg, Derek Barnett
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 10 October 2011 (DB)
+// Last modified: 25 October 2011 (DB)
 // ---------------------------------------------------------------------------
 // Provides the basic functionality for producing BAM files
 // ***************************************************************************
 
 #include "api/BamAlignment.h"
 #include "api/BamWriter.h"
 #include "api/SamHeader.h"
-#include "api/internal/BamWriter_p.h"
+#include "api/internal/bam/BamWriter_p.h"
 using namespace BamTools;
 using namespace BamTools::Internal;
 using namespace std;

external/BamTools/api/CMakeLists.txt → external/BamTools-2.0/api/CMakeLists.txt

@@ -12,7 +12,10 @@ include_directories( ${BamTools_SOURCE_DIR}/src )
 add_definitions( -DBAMTOOLS_API_LIBRARY ) # (for proper exporting of library symbols)
 add_definitions( -fPIC ) # (attempt to force PIC compiling on CentOS, not being set on shared libs by CMake)
 
-# list of all BamTools API source (.cpp) files
+# fetch all internal source files
+add_subdirectory ( internal )
+
+# make list of all API source files
 set( BamToolsAPISources
         BamAlignment.cpp
         BamMultiReader.cpp
@@ -25,41 +28,30 @@ set( BamToolsAPISources
         SamReadGroupDictionary.cpp
         SamSequence.cpp
         SamSequenceDictionary.cpp
-        internal/BamDeviceFactory_p.cpp
-        internal/BamException_p.cpp
-        internal/BamFile_p.cpp
-        internal/BamFtp_p.cpp
-        internal/BamHeader_p.cpp
-        internal/BamHttp_p.cpp
-        internal/BamIndexFactory_p.cpp
-        internal/BamMultiReader_p.cpp
-        internal/BamPipe_p.cpp
-        internal/BamRandomAccessController_p.cpp
-        internal/BamReader_p.cpp
-        internal/BamStandardIndex_p.cpp
-        internal/BamToolsIndex_p.cpp
-        internal/BamWriter_p.cpp
-        internal/BgzfStream_p.cpp
-        internal/ILocalIODevice_p.cpp
-        internal/IRemoteIODevice_p.cpp
-        internal/SamFormatParser_p.cpp
-        internal/SamFormatPrinter_p.cpp
-        internal/SamHeaderValidator_p.cpp
+        ${InternalSources}
 )
 
 # create main BamTools API shared library
 add_library( BamTools SHARED ${BamToolsAPISources} )
-set_target_properties( BamTools PROPERTIES SOVERSION "2.0.5" )
-set_target_properties( BamTools PROPERTIES OUTPUT_NAME "bamtools" )
+set_target_properties( BamTools PROPERTIES
+                       SOVERSION "2.1.0"
+                       OUTPUT_NAME "bamtools" )
 
 # create main BamTools API static library
 add_library( BamTools-static STATIC ${BamToolsAPISources} )
-set_target_properties( BamTools-static PROPERTIES OUTPUT_NAME "bamtools" )
-set_target_properties( BamTools-static PROPERTIES PREFIX "lib" )
+set_target_properties( BamTools-static PROPERTIES 
+                       OUTPUT_NAME "bamtools" 
+                       PREFIX "lib" )
+
+# link libraries automatically with zlib (and Winsock2, if applicable)
+if( _WIN32 )
+    set( APILibs z ws2_32 )
+else( _WIN32 )
+    set( APILibs z )
+endif( _WIN32 )
 
-# link libraries with zlib automatically
-target_link_libraries( BamTools z )
-target_link_libraries( BamTools-static z )
+target_link_libraries( BamTools ${APILibs} )
+target_link_libraries( BamTools-static ${APILibs} )
 
 # set library install destinations
 install( TARGETS BamTools LIBRARY DESTINATION "lib/bamtools" RUNTIME DESTINATION "bin")
@@ -69,7 +61,7 @@ install( TARGETS BamTools-static ARCHIVE DESTINATION "lib/bamtools")
 include(../ExportHeader.cmake)
 set(ApiIncludeDir "api")
 ExportHeader(APIHeaders api_global.h             ${ApiIncludeDir})
-ExportHeader(APIHeaders BamAlgorithms.h           ${ApiIncludeDir})
+ExportHeader(APIHeaders BamAlgorithms.h          ${ApiIncludeDir})
 ExportHeader(APIHeaders BamAlignment.h           ${ApiIncludeDir})
 ExportHeader(APIHeaders BamAux.h                 ${ApiIncludeDir})
 ExportHeader(APIHeaders BamConstants.h           ${ApiIncludeDir})
@@ -87,5 +79,5 @@ ExportHeader(APIHeaders SamReadGroupDictionary.h ${ApiIncludeDir})
 ExportHeader(APIHeaders SamSequence.h            ${ApiIncludeDir})
 ExportHeader(APIHeaders SamSequenceDictionary.h  ${ApiIncludeDir})
 
-set(AlgorithmsIncludeDir "api/algorithms")
-ExportHeader(AlgorithmsHeaders algorithms/Sort.h ${AlgorithmsIncludeDir})
+set( AlgorithmsIncludeDir "api/algorithms" )
+ExportHeader( AlgorithmsHeaders algorithms/Sort.h ${AlgorithmsIncludeDir} )

external/BamTools/api/IBamIODevice.h → external/BamTools-2.0/api/IBamIODevice.h

@@ -2,7 +2,7 @@
 // IBamIODevice.h (c) 2011 Derek Barnett
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 10 October 2011 (DB)
+// Last modified: 10 November 2011 (DB)
 // ---------------------------------------------------------------------------
 // Base class for all BAM I/O devices (e.g. local file, pipe, HTTP, FTP, etc.)
 //
@@ -19,16 +19,18 @@
 #define IBAMIODEVICE_H
 
 #include "api/api_global.h"
+#include <cstdio>
 #include <string>
 
 namespace BamTools {
 
 class API_EXPORT IBamIODevice {
 
     // enums
-    public: enum OpenMode { NotOpen = 0
-                          , ReadOnly
-                          , WriteOnly
+    public: enum OpenMode { NotOpen   = 0x0000
+                          , ReadOnly  = 0x0001
+                          , WriteOnly = 0x0002
+                          , ReadWrite = ReadOnly | WriteOnly
                           };
 
     // ctor & dtor
@@ -38,14 +40,16 @@ class API_EXPORT IBamIODevice {
     // IBamIODevice interface
     public:
 
+        // TODO: add seek(pos, *from*)
+
         // pure virtuals
         virtual void Close(void) =0;
         virtual bool IsRandomAccess(void) const =0;
         virtual bool Open(const OpenMode mode) =0;
-        virtual size_t Read(char* data, const unsigned int numBytes) =0;
-        virtual bool Seek(const int64_t& position) =0;
+        virtual int64_t Read(char* data, const unsigned int numBytes) =0;
+        virtual bool Seek(const int64_t& position, const int origin = SEEK_SET) =0;
         virtual int64_t Tell(void) const =0;
-        virtual size_t Write(const char* data, const unsigned int numBytes) =0;
+        virtual int64_t Write(const char* data, const unsigned int numBytes) =0;
 
         // default implementation provided
         virtual std::string GetErrorString(void);

external/BamTools/api/SamConstants.h → external/BamTools-2.0/api/SamConstants.h

@@ -2,7 +2,7 @@
 // SamConstants.h (c) 2010 Derek Barnett
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 10 October 2011 (DB)
+// Last modified: 27 March 2012 (DB)
 // ---------------------------------------------------------------------------
 // Provides constants for SAM header
 // ***************************************************************************
@@ -24,6 +24,8 @@ const char SAM_STAR   = '*';
 const char SAM_TAB    = '\t';
 const std::string SAM_DIGITS = "0123456789";
 
+const std::string SAM_CURRENT_VERSION = "1.4";
+
 // HD entries
 const std::string SAM_HD_BEGIN_TOKEN    = "@HD";
 const std::string SAM_HD_VERSION_TAG    = "VN";