Add Blast format and FastA Huge comment correction

.cproject

@@ -1,7 +1,5 @@
 <?xml version="1.0" encoding="UTF-8" standalone="no"?>
-<?fileVersion 4.0.0?>
-
-<cproject storage_type_id="org.eclipse.cdt.core.XmlProjectDescriptionStorage">
+<?fileVersion 4.0.0?><cproject storage_type_id="org.eclipse.cdt.core.XmlProjectDescriptionStorage">
 	<storageModule moduleId="org.eclipse.cdt.core.settings">
 		<cconfiguration id="cdt.managedbuild.config.gnu.lib.debug.1749414343">
 			<storageModule buildSystemId="org.eclipse.cdt.managedbuilder.core.configurationDataProvider" id="cdt.managedbuild.config.gnu.lib.debug.1749414343" moduleId="org.eclipse.cdt.core.settings" name="Debug">
@@ -13,11 +11,11 @@
 					</externalSetting>
 				</externalSettings>
 				<extensions>
-					<extension id="org.eclipse.cdt.core.ELF" point="org.eclipse.cdt.core.BinaryParser"/>
 					<extension id="org.eclipse.cdt.core.GmakeErrorParser" point="org.eclipse.cdt.core.ErrorParser"/>
 					<extension id="org.eclipse.cdt.core.CWDLocator" point="org.eclipse.cdt.core.ErrorParser"/>
 					<extension id="org.eclipse.cdt.core.GCCErrorParser" point="org.eclipse.cdt.core.ErrorParser"/>
 					<extension id="org.eclipse.cdt.core.GASErrorParser" point="org.eclipse.cdt.core.ErrorParser"/>
+					<extension id="org.eclipse.cdt.core.ELF" point="org.eclipse.cdt.core.BinaryParser"/>
 				</extensions>
 			</storageModule>
 			<storageModule moduleId="cdtBuildSystem" version="4.0.0">
@@ -37,8 +35,8 @@
 								<option id="gnu.cpp.compiler.lib.debug.option.debugging.level.1544253014" name="Debug Level" superClass="gnu.cpp.compiler.lib.debug.option.debugging.level" value="gnu.cpp.compiler.debugging.level.default" valueType="enumerated"/>
 								<option id="gnu.cpp.compiler.option.include.paths.1304733242" name="Include paths (-I)" superClass="gnu.cpp.compiler.option.include.paths" valueType="includePath">
 									<listOptionValue builtIn="false" value="&quot;${workspace_loc:/PlastLibrary/src}&quot;"/>
-									<listOptionValue builtIn="false" value="/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/include"/>
-									<listOptionValue builtIn="false" value="/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/include/linux"/>
+									<listOptionValue builtIn="false" value="/usr/lib/jvm/java/include"/>
+									<listOptionValue builtIn="false" value="/usr/lib/jvm/java/include/linux"/>
 								</option>
 								<option id="gnu.cpp.compiler.option.debugging.gprof.1253556430" name="Generate gprof information (-pg)" superClass="gnu.cpp.compiler.option.debugging.gprof" value="true" valueType="boolean"/>
 								<option id="gnu.cpp.compiler.option.preprocessor.def.1220974519" name="Defined symbols (-D)" superClass="gnu.cpp.compiler.option.preprocessor.def" valueType="definedSymbols">
@@ -97,11 +95,11 @@
 					</externalSetting>
 				</externalSettings>
 				<extensions>
-					<extension id="org.eclipse.cdt.core.ELF" point="org.eclipse.cdt.core.BinaryParser"/>
 					<extension id="org.eclipse.cdt.core.GmakeErrorParser" point="org.eclipse.cdt.core.ErrorParser"/>
 					<extension id="org.eclipse.cdt.core.CWDLocator" point="org.eclipse.cdt.core.ErrorParser"/>
 					<extension id="org.eclipse.cdt.core.GCCErrorParser" point="org.eclipse.cdt.core.ErrorParser"/>
 					<extension id="org.eclipse.cdt.core.GASErrorParser" point="org.eclipse.cdt.core.ErrorParser"/>
+					<extension id="org.eclipse.cdt.core.ELF" point="org.eclipse.cdt.core.BinaryParser"/>
 				</extensions>
 			</storageModule>
 			<storageModule moduleId="cdtBuildSystem" version="4.0.0">
@@ -116,8 +114,8 @@
 								<option id="gnu.cpp.compiler.lib.release.option.debugging.level.534390027" name="Debug Level" superClass="gnu.cpp.compiler.lib.release.option.debugging.level" value="gnu.cpp.compiler.debugging.level.none" valueType="enumerated"/>
 								<option id="gnu.cpp.compiler.option.include.paths.1267507508" name="Include paths (-I)" superClass="gnu.cpp.compiler.option.include.paths" valueType="includePath">
 									<listOptionValue builtIn="false" value="&quot;${workspace_loc:/PlastLibrary/src}&quot;"/>
-									<listOptionValue builtIn="false" value="/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/include"/>
-									<listOptionValue builtIn="false" value="/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/include/linux"/>
+									<listOptionValue builtIn="false" value="/usr/lib/jvm/java/include"/>
+									<listOptionValue builtIn="false" value="/usr/lib/jvm/java/include/linux"/>
 								</option>
 								<option id="gnu.cpp.compiler.option.preprocessor.def.650156252" name="Defined symbols (-D)" superClass="gnu.cpp.compiler.option.preprocessor.def" valueType="definedSymbols">
 									<listOptionValue builtIn="false" value="__LINUX__=1"/>

src/algo/core/api/IAlgoConfig.hpp

@@ -86,6 +86,16 @@ class IConfiguration : public dp::SmartPointer
      */
     virtual IParameters* createDefaultParameters (const std::string& algoName) = 0;
 
+    /** Create quick reader which permits to parse quickly the different file type
+     *  It permits to open the fasta format (.fa), the blast format (.pin for protein or .nin for nucleotid).
+     *  It permits also to open the alias file (.pal for protein and .nal for nucleotid).
+     *  This quick reader open the file and extract some information which is used to split the database reading
+     *  \param[in] uri : uri path
+     *  \param[in] shouldInferType : tells whether we should try to find the kind of genomic database we read( used only for fasta format)
+     *  \return a new IDatabaseQuickReader instance
+     */
+    virtual database::IDatabaseQuickReader* createDefaultQuickReader (const std::string& uri, bool shouldInferType) = 0;
+
     /** Create a command dispatcher instance. Such an instance can be used for parallelization (hits
      * iteration for instance, see IAlgorithm class).
      *  \return a new ICommandDispatcher instance
@@ -104,9 +114,10 @@ class IConfiguration : public dp::SmartPointer
     virtual os::impl::TimeInfo* createTimeInfo () = 0;
 
     /** Create a factory that builds ISequenceIterator objects.
+     * \param[in] uri : uri path to select the sequence iterator factory depending of the file type
      * \return the factory instance.
      */
-    virtual database::ISequenceIteratorFactory* createSequenceIteratorFactory () = 0;
+    virtual database::ISequenceIteratorFactory* createSequenceIteratorFactory (const std::string& uri) = 0;
 
     /** Create a database object (with means for retrieving sequence within the database) from an uri (likely
      *  a local file, but it should be a location on a remote computer). A Range can be provided for using only

src/algo/core/impl/AbstractAlgorithm.cpp

@@ -68,6 +68,9 @@ static const char* keyIter      = "iteration";
 static const char* keyOutput    = "output";
 static const char* keyAlgorithm = "algorithm";
 
+/*static u_int64_t checksum = 0;
+static u_int64_t nbDataSeq = 0;*/
+
 /********************************************************************************/
 
 /** Command that launch 'iterate' method on a IHitIterator instance. Using a Command allows to
@@ -294,6 +297,17 @@ void AbstractAlgorithm::execute (void)
             /** Shortcuts. */
             ISequenceDatabase* subjectDb = subjectDbIt->currentItem();
 
+            /*ISequence seq;
+            for (u_int64_t nbSeq=0;nbSeq<subjectDb->getSequencesNumber();nbSeq++)
+            {
+            	subjectDb->getSequenceByIndex(nbSeq,seq);
+            	for (u_int64_t nData=0;nData<seq.data.letters.size;nData++)
+            	{
+            		checksum = (checksum + seq.data.letters.data[nData]) % (1<<16);
+            		nbDataSeq++;
+            	}
+            }*/
+
             DEBUG (("AbstractAlgorithm::execute : subjectSeqNb=%ld  querySeqNb=%ld\n",
                 subjectDb->getSequencesNumber(),
                 queryDb->getSequencesNumber()
@@ -363,6 +377,7 @@ void AbstractAlgorithm::execute (void)
         }  /* end of for (subjectDbIt.first(); ... */
 
     }  /* end of for (queryDbIt.first(); ... */
+    //printf ("checksum=%ld nbData=%ld\n", checksum,	nbDataSeq);
 }
 
 /*********************************************************************

src/algo/core/impl/DefaultAlgoConfig.cpp

@@ -22,6 +22,10 @@
 
 #include <database/impl/BufferedSequenceDatabase.hpp>
 #include <database/impl/FastaSequenceIterator.hpp>
+#include <database/impl/FastaDatabaseQuickReader.hpp>
+#include <database/impl/BlastdbSequenceIterator.hpp>
+#include <database/impl/BlastdbDatabaseQuickReader.hpp>
+#include <database/impl/DatabaseUtility.hpp>
 
 #include <seed/impl/BasicSeedModel.hpp>
 #include <seed/impl/SubSeedModel.hpp>
@@ -136,6 +140,31 @@ DefaultConfiguration::~DefaultConfiguration ()
     setProperties (0);
 }
 
+/*********************************************************************
+** METHOD  :
+** PURPOSE :
+** INPUT   :
+** OUTPUT  :
+** RETURN  :
+** REMARKS :
+*********************************************************************/
+database::IDatabaseQuickReader* DefaultConfiguration::createDefaultQuickReader (const std::string& uri, bool shouldInferType)
+{
+	DatabaseLookupType::QuickReaderType_e databaseType = DatabaseLookupType::ENUM_TYPE_UNKNOWN;
+
+	if (uri!="foo")
+	{
+		databaseType = DatabaseLookupType::quickReaderType(uri);
+		if ((databaseType==DatabaseLookupType::ENUM_BLAST_PIN)||(databaseType==DatabaseLookupType::ENUM_BLAST_NIN)
+				||(databaseType==DatabaseLookupType::ENUM_BLAST_PAL)||(databaseType==DatabaseLookupType::ENUM_BLAST_NAL))
+			return new BlastdbDatabaseQuickReader (uri, shouldInferType);
+		else
+			return new FastaDatabaseQuickReader (uri, shouldInferType);
+	}
+	else
+		return new FastaDatabaseQuickReader (uri, shouldInferType);
+}
+
 /*********************************************************************
 ** METHOD  :
 ** PURPOSE :
@@ -355,9 +384,18 @@ os::impl::TimeInfo* DefaultConfiguration::createTimeInfo ()
 ** RETURN  :
 ** REMARKS :
 *********************************************************************/
-database::ISequenceIteratorFactory* DefaultConfiguration::createSequenceIteratorFactory ()
+database::ISequenceIteratorFactory* DefaultConfiguration::createSequenceIteratorFactory (const string& uri)
 {
-    return new FastaSequenceIteratorFactory ();
+	DatabaseLookupType::QuickReaderType_e databaseType = DatabaseLookupType::ENUM_TYPE_UNKNOWN;
+
+	databaseType = DatabaseLookupType::quickReaderType(uri);
+	if ((databaseType==DatabaseLookupType::ENUM_BLAST_PIN)||(databaseType==DatabaseLookupType::ENUM_BLAST_NIN)
+		||(databaseType==DatabaseLookupType::ENUM_BLAST_PAL)||(databaseType==DatabaseLookupType::ENUM_BLAST_NAL))
+	{
+		return new BlastdbSequenceIteratorFactory ();
+	}
+	else
+		return new FastaSequenceIteratorFactory();
 }
 
 /*********************************************************************
@@ -376,11 +414,13 @@ ISequenceDatabase*  DefaultConfiguration::createDatabase (
 )
 {
     LOCAL (sequenceIteratorFactory);
+    ISequenceIteratorFactory* tempFactory = createSequenceIteratorFactory(uri);
+    LOCAL (tempFactory);
 
     /** We create the sequence iterator. */
     ISequenceIterator* seqIterator =  sequenceIteratorFactory ?
         sequenceIteratorFactory->createSequenceIterator (uri, range) :
-        new FastaSequenceIterator (uri.c_str(), 2*1024, range.begin, range.end);
+        tempFactory->createSequenceIterator (uri, range);
 
     /** We create the database. */
     return new BufferedSequenceDatabase (seqIterator, filtering);

src/algo/core/impl/DefaultAlgoConfig.hpp

@@ -59,6 +59,9 @@ class DefaultConfiguration : public IConfiguration
     /** Destructor. */
     virtual ~DefaultConfiguration ();
 
+    /** \copydoc IConfiguration::createDefaultQuickReader */
+    database::IDatabaseQuickReader* createDefaultQuickReader (const std::string& uri, bool shouldInferType);
+
     /** \copydoc IConfiguration::createDefaultParameters */
     IParameters* createDefaultParameters (const std::string& algoName);
 
@@ -72,7 +75,7 @@ class DefaultConfiguration : public IConfiguration
     os::impl::TimeInfo* createTimeInfo ();
 
     /** \copydoc IConfiguration::createSequenceIteratorFactory */
-    database::ISequenceIteratorFactory* createSequenceIteratorFactory ();
+    database::ISequenceIteratorFactory* createSequenceIteratorFactory (const std::string& uri);
 
     /** \copydoc IConfiguration::createDatabase */
     database::ISequenceDatabase*  createDatabase (

src/algo/core/impl/DefaultAlgoEnvironment.cpp

@@ -625,7 +625,8 @@ void DefaultEnvironment::setSubjectBank (dp::IProperties* properties, u_int64_t
     IProperty* subjectUriProp = properties->getProperty (STR_OPTION_SUBJECT_URI);
 
     /** We create the quick reader instance for the subject bank. */
-    setQuickSubjectDbReader (new FastaDatabaseQuickReader (subjectUriProp->value, true));
+    setQuickSubjectDbReader (_config->createDefaultQuickReader(subjectUriProp->value, true));
+    subjectUriProp->value = _quickSubjectDbReader->getUri();
 
     /** We check whether it is an 'info' file or a fasta file. */
     bool isInfoFile = FastaDatabaseQuickReader::isQuickReaderFile (subjectUriProp->value);

src/algo/core/impl/PlastnAlgoConfig.cpp

@@ -194,11 +194,13 @@ ISequenceDatabase*  PlastnConfiguration::createDatabase (
     ));
 
     LOCAL (sequenceIteratorFactory);
+    ISequenceIteratorFactory* tempFactory = createSequenceIteratorFactory(uri);
+    LOCAL (tempFactory);
 
     /** We create the sequence iterator. */
     ISequenceIterator* seqIterator =  sequenceIteratorFactory ?
         sequenceIteratorFactory->createSequenceIterator (uri, range) :
-        new FastaSequenceIterator (uri.c_str(), 2*1024, range.begin, range.end);
+        tempFactory->createSequenceIterator (uri, range);
 
     return new BufferedCachedSequenceDatabase (seqIterator, filtering);
 }

src/alignment/visitors/impl/RawOutputVisitor.hpp

@@ -62,7 +62,8 @@ class RawOutputVisitor : public FileVisitor
     /** \copydoc AbstractAlignmentResultVisitor::visitSubjectSequence */
     void visitSubjectSequence (const database::ISequence* seq, const misc::ProgressInfo& progress)
     {
-    	getFile()->print ("S %d %s\n",  seq->getLength(), seq->comment);
+    	//getFile()->print ("S %d %s\n",  seq->getLength(), seq->comment);
+    	getFile()->print ("S %d %s\n",  seq->getLength(), seq->getComment(seq->comment).c_str());
     }
 
     /** \copydoc AbstractAlignmentResultVisitor::visitAlignment */

src/alignment/visitors/impl/TabulatedOutputVisitor.cpp

@@ -93,8 +93,11 @@ void TabulatedOutputVisitor::dumpLine (core::Alignment* align)
     snprintf (queryName,   sizeof(queryName),   "%s", _currentQuery->comment);
     if ( (locate = database::ISequence::searchIdSeparator (queryName)) != 0)  { *locate = 0; }
 
-    char subjectName[128];
-    snprintf (subjectName, sizeof(subjectName), "%s", _currentSubject->comment);
+    char subjectName[128] = "";
+    //snprintf (subjectName, sizeof(subjectName), "%s", _currentSubject->comment);
+    //snprintf (subjectName, sizeof(subjectName), "TEST");
+    snprintf (subjectName, sizeof(subjectName), "%s", _currentSubject->getComment(_currentSubject->comment).c_str());
+
     if ( (locate = database::ISequence::searchIdSeparator (subjectName)) != 0)  { *locate = 0; }
 
     char evalueStr[32];

src/alignment/visitors/impl/XmlOutputVisitor.cpp

@@ -150,7 +150,7 @@ void XmlOutputVisitor::visitSubjectSequence (
 
     printline (4, "<Hit>");
     printline (5, "<Hit_num>%d</Hit_num>", _nbSubject);
-    printline (5, "<Hit_def>%s</Hit_def>", _currentSubject->comment);
+    printline (5, "<Hit_def>%s</Hit_def>", _currentSubject->getComment(_currentSubject->comment).c_str());
     printline (5, "<Hit_len>%d</Hit_len>", _currentSubject->data.letters.size);
     printline (5, "<Hit_hsps>");
 }

src/database/api/IAlphabet.hpp

@@ -101,6 +101,8 @@ enum Encoding
     SUBSEED,
     ASCII,
     NCBI,
+    NCBI_DNA_NO_AMB,
+    NCBI_DNA_WITH_AMB,
     UNKNOWN
 };
 

src/database/api/IDatabaseQuickReader.hpp

@@ -33,6 +33,8 @@
 
 #include <designpattern/api/SmartPointer.hpp>
 #include <designpattern/api/IProperty.hpp>
+#include <designpattern/impl/TokenizerIterator.hpp>
+#include <os/impl/CommonOsImpl.hpp>
 #include <misc/api/types.hpp>
 #include <vector>
 #include <string>
@@ -68,7 +70,7 @@ class IDatabaseQuickReader : public dp::SmartPointer
         ENUM_UNKNOWN
     };
 
-    /** Read the database. Must be called before using getters.
+     /** Read the database. Must be called before using getters.
      *  A 'maxblocksize' parameter can be provided; if not null, it is used for splitting the database
      *  into small sequences sets of maximum size; as a result, one can retrieve a vector of offsets,
      *  each offset pointing to a set of sequences.
@@ -118,6 +120,7 @@ class IDatabaseQuickReader : public dp::SmartPointer
 
     /** */
     virtual int save (const std::string& uri) = 0;
+
 };
 
 /********************************************************************************/