Added GetTagNames method to BamAlignment

diff --git a/src/api/BamAlignment.cpp b/src/api/BamAlignment.cpp
index 251c5e003e7aa4bf2effcef17fd6919b33f20910..b8482f69d45806156578e8d2aacb18be3ea628f6 100644 (file)
--- a/src/api/BamAlignment.cpp
+++ b/src/api/BamAlignment.cpp
@@ -2,7 +2,7 @@
 // BamAlignment.cpp (c) 2009 Derek Barnett
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 4 April 2012 (DB)
+// Last modified: 18 November 2012 (DB)
 // ---------------------------------------------------------------------------
 // Provides the BamAlignment data structure
 // ***************************************************************************
@@ -551,6 +551,45 @@ bool BamAlignment::GetSoftClips(vector<int>& clipSizes,
     return softClipFound;
 }
 
+/*! \fn std::vector<std::string> BamAlignment::GetTagNames(void) const
+    \brief Retrieves the BAM tag names.
+
+    When paired with GetTagType() and GetTag(), this method allows you
+    to iterate over an alignment's tag data without knowing the names (or types)
+    beforehand.
+
+    \return \c vector containing all tag names found (empty if none available)
+    \sa \samSpecURL for more details on reserved tag names, supported tag types, etc.
+*/
+std::vector<std::string> BamAlignment::GetTagNames(void) const {
+
+    std::vector<std::string> result;
+    if ( SupportData.HasCoreOnly || TagData.empty() )
+        return result;
+
+    char* pTagData = (char*)TagData.data();
+    const unsigned int tagDataLength = TagData.size();
+    unsigned int numBytesParsed = 0;
+    while ( numBytesParsed < tagDataLength ) {
+
+        // get current tag name & type
+        const char* pTagName = pTagData;
+        const char* pTagType = pTagData + 2;
+        pTagData       += 3;
+        numBytesParsed +=3;
+
+        // store tag name
+        result.push_back( std::string(pTagName, 2)  );
+
+        // find the next tag
+        if ( *pTagType == '\0' ) break;
+        if ( !SkipToNextTag(*pTagType, pTagData, numBytesParsed) ) break;
+        if ( *pTagData == '\0' ) break;
+    }
+
+    return result;
+}
+
 /*! \fn bool BamAlignment::GetTagType(const std::string& tag, char& type) const
     \brief Retrieves the BAM tag type-code associated with requested tag name.
 

diff --git a/src/api/BamAlignment.h b/src/api/BamAlignment.h
index a2349ea9ce9b254066450dc953b99925a8f03b56..d18b239d65f50b44703682184e760927fd181f81 100644 (file)
--- a/src/api/BamAlignment.h
+++ b/src/api/BamAlignment.h
@@ -2,7 +2,7 @@
 // BamAlignment.h (c) 2009 Derek Barnett
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 16 October 2011 (DB)
+// Last modified: 18 November 2012 (DB)
 // ---------------------------------------------------------------------------
 // Provides the BamAlignment data structure
 // ***************************************************************************
@@ -80,6 +80,9 @@ struct API_EXPORT BamAlignment {
         template<typename T> bool GetTag(const std::string& tag, T& destination) const;
         template<typename T> bool GetTag(const std::string& tag, std::vector<T>& destination) const;
 
+        // retrieves all current tag names
+        std::vector<std::string> GetTagNames(void) const;
+
         // retrieves the SAM/BAM type-code for requested tag name
         bool GetTagType(const std::string& tag, char& type) const;