X-Git-Url: http://woldlab.caltech.edu/gitweb/?p=pysam.git;a=blobdiff_plain;f=pysam%2Fctabix.pyx;h=a31f3f244b37329286cc23760d5c2942b8628103;hp=25c1a1ee732df3e383a29e80c1ea2ae533b7925e;hb=ca46ef4ba4a883c57cea62d5bf1bc021f1185109;hpb=d02fe5283ed7a93a2f76a5d6dc6e37b40c11b9b1

diff --git a/pysam/ctabix.pyx b/pysam/ctabix.pyx
index 25c1a1e..a31f3f2 100644
--- a/pysam/ctabix.pyx
+++ b/pysam/ctabix.pyx
@@ -33,7 +33,9 @@ cdef class Tabixfile:
         if self.tabixfile != NULL: self.close()
         self.tabixfile = NULL
 
-        self.filename = filename
+        if self._filename != NULL: free(self._filename )
+        self._filename = strdup( filename )
+
         filename_index = filename + ".tbi"
 
         if mode[0] == 'w':
@@ -42,14 +44,14 @@ cdef class Tabixfile:
 
         elif mode[0] == "r":
             # open file for reading
-            if not os.path.exists( self.filename ):
-                raise IOError( "file `%s` not found" % self.filename)
+            if not os.path.exists( self._filename ):
+                raise IOError( "file `%s` not found" % self._filename)
 
             if not os.path.exists( filename_index ):
                 raise IOError( "index `%s` not found" % filename_index)
 
             # open file and load index
-            self.tabixfile = ti_open( self.filename, filename_index )
+            self.tabixfile = ti_open( self._filename, filename_index )
 
         if self.tabixfile == NULL:
             raise IOError("could not open file `%s`" % filename )
@@ -136,12 +138,26 @@ cdef class Tabixfile:
             else:
                 return TabixIteratorParsed( self, -1, 0, 0, parser )
 
+    ###############################################################
+    ###############################################################
+    ###############################################################
+    ## properties
+    ###############################################################
+    property filename:
+        '''filename associated with this object.'''
+        def __get__(self):
+            if not self._isOpen(): raise ValueError( "I/O operation on closed file" )
+            return self._filename
+
     property header:
+        '''the file header.
+          
+        .. note::
+            The header is returned as an iterator over lines without the
+            newline character.
+        '''
+        
         def __get__( self ):
-            '''return header lines as an iterator.
-
-            Note that the header lines do not contain the newline '\n' character.
-            '''
             return TabixHeaderIterator( self )
 
     property contigs:
@@ -158,6 +174,22 @@ cdef class Tabixfile:
                 result.append( sequences[x] )
             return result
             
+    def close( self ):
+        '''
+        closes the :class:`pysam.Tabixfile`.'''
+        if self.tabixfile != NULL:
+            ti_close( self.tabixfile )
+            self.tabixfile = NULL
+
+    def __dealloc__( self ):
+        # remember: dealloc cannot call other python methods
+        # note: no doc string
+        # note: __del__ is not called.
+        if self.tabixfile != NULL:
+            ti_close( self.tabixfile )
+            self.tabixfile = NULL
+        if self._filename != NULL: free( self._filename )
+
 cdef class TabixIterator:
     """iterates over rows in *tabixfile* in region
     given by *tid*, *start* and *end*.
@@ -265,7 +297,10 @@ cdef class Parser:
     pass
 
 cdef class asTuple(Parser):
-    '''converts a :term:`tabix row` into a python tuple.''' 
+    '''converts a :term:`tabix row` into a python tuple.
+
+    Access is by numeric index.
+    ''' 
     def __call__(self, char * buffer, int len):
         cdef TabProxies.TupleProxy r
         r = TabProxies.TupleProxy()
@@ -275,7 +310,36 @@ cdef class asTuple(Parser):
         return r
 
 cdef class asGTF(Parser):
-    '''converts a :term:`tabix row` into a GTF record.''' 
+    '''converts a :term:`tabix row` into a GTF record with the following 
+    fields:
+
+    contig
+       contig
+    feature
+       feature
+    source
+       source
+    start
+       genomic start coordinate (0-based)
+    end
+       genomic end coordinate plus one (0-based)
+    score
+       feature score
+    strand
+       strand
+    frame
+       frame
+    attributes
+       attribute string.
+
+    GTF formatted entries also defined the attributes:
+
+    gene_id
+       the gene identifier
+    transcript_ind
+       the transcript identifier
+    
+    ''' 
     def __call__(self, char * buffer, int len):
         cdef TabProxies.GTFProxy r
         r = TabProxies.GTFProxy()
@@ -283,7 +347,39 @@ cdef class asGTF(Parser):
         return r
 
 cdef class asBed( Parser ):
-    '''converts a :term:`tabix row` into a GTF record.''' 
+    '''converts a :term:`tabix row` into a bed record
+    with the following fields:
+
+    contig
+       contig
+    start
+       genomic start coordinate (zero-based)
+    end
+       genomic end coordinate plus one (zero-based)
+    name
+       name of feature.
+    score
+       score of feature
+    strand
+       strand of feature
+    thickStart
+       thickStart
+    thickEnd
+       thickEnd
+    itemRGB
+       itemRGB
+    blockCount
+       number of bocks
+    blockSizes
+       ',' separated string of block sizes
+    blockStarts
+       ',' separated string of block genomic start positions
+
+    Only the first three fields are required. Additional
+    fields are optional, but if one is defined, all the preceeding
+    need to be defined as well.
+
+    ''' 
     def __call__(self, char * buffer, int len):
         cdef TabProxies.BedProxy r
         r = TabProxies.BedProxy()
@@ -291,7 +387,35 @@ cdef class asBed( Parser ):
         return r
 
 cdef class asVCF( Parser ): 
-    '''converts a :term:`tabix row` into a VCF record.'''
+    '''converts a :term:`tabix row` into a VCF record with
+    the following fields:
+    
+    contig
+       contig
+    pos
+       chromosomal position, zero-based
+    id 
+       id
+    ref
+       reference
+    alt
+       alt
+    qual
+       qual
+    filter
+       filter
+    info
+       info
+    format
+       format specifier.
+
+    Access to genotypes is via index::
+
+        contig = vcf.contig
+        first_sample_genotype = vcf[0]
+        second_sample_genotype = vcf[1]
+
+    '''
     def __call__(self, char * buffer, int len ):
         cdef TabProxies.VCFProxy r
         r = TabProxies.VCFProxy()