htslib 0.3.1 → 0.4.0

data/lib/hts/libhts/fai.rb

@@ -48,24 +48,24 @@ module HTS
     attach_function \
       :fai_fetch,
       [Faidx, :string, :pointer],
-      :string
+      :pointer
 
     # Fetch the sequence in a region
     attach_function \
       :fai_fetch64,
       [Faidx, :string, :pointer],
-      :string
+      :pointer
 
     # Fetch the quality string for a region for FASTQ files
     attach_function \
       :fai_fetchqual,
       [Faidx, :string, :pointer],
-      :string
+      :pointer
 
     attach_function \
       :fai_fetchqual64,
       [Faidx, :string, :pointer],
-      :string
+      :pointer
 
     # Fetch the number of sequences
     attach_function \
@@ -77,25 +77,25 @@ module HTS
     attach_function \
       :faidx_fetch_seq,
       [Faidx, :string, :int, :int, :pointer],
-      :string
+      :pointer
 
     # Fetch the sequence in a region
     attach_function \
       :faidx_fetch_seq64,
       [Faidx, :string, :int64, :int64, :pointer],
-      :string
+      :pointer
 
     # Fetch the quality string in a region for FASTQ files
     attach_function \
       :faidx_fetch_qual,
       [Faidx, :string, :int, :int, :pointer],
-      :string
+      :pointer
 
     # Fetch the quality string in a region for FASTQ files
     attach_function \
       :faidx_fetch_qual64,
       [Faidx, :string, :int64, :int64, :pointer],
-      :string
+      :pointer
 
     # Query if sequence is present
     attach_function \
@@ -121,6 +121,11 @@ module HTS
       [Faidx, :string],
       :int
 
+    attach_function \
+      :faidx_seq_len64,
+      [Faidx, :string],
+      :int64
+
     # Parses a region string.
     attach_function \
       :fai_parse_region,

data/lib/hts/libhts/hfile.rb

@@ -46,10 +46,10 @@ module HTS
 
     # Report the current stream offset
     def self.htell(fp)
-      # TODO: This is a hack. Is this OK?
-      bg = FFI::Pointer.new(:int, fp.pointer.address + fp.offset_of(:begin)).read_int
-      bf = FFI::Pointer.new(:int, fp.pointer.address + fp.offset_of(:buffer)).read_int
-      fp[:offset] + (bg - bf)
+      base_ptr   = fp.pointer
+      begin_ptr  = base_ptr.get_pointer(fp.offset_of(:begin))
+      buffer_ptr = base_ptr.get_pointer(fp.offset_of(:buffer))
+      fp[:offset] + (begin_ptr.address - buffer_ptr.address)
     end
 
     # Read from the stream until the delimiter, up to a maximum length

data/lib/hts/libhts/hts.rb

@@ -467,5 +467,11 @@ module HTS
       :hts_md5_destroy,
       [:pointer],
       :void
+
+    # Computes CRC32 for a buffer with an initial crc value
+    attach_function \
+      :hts_crc32,
+      %i[uint32 pointer size_t],
+      :uint32
   end
 end

data/lib/hts/libhts/sam.rb

@@ -382,11 +382,15 @@ module HTS
 
     # Get the next read from a SAM/BAM/CRAM iterator
     def self.sam_itr_next(htsfp, itr, r)
-      # FIXME: check if htsfp is compressed BGZF
+      raise("#{htsfp[:fn] || 'File'} not BGZF compressed") unless htsfp[:is_bgzf] == 1 || htsfp[:is_cram] == 1
       raise("Null iterator") if itr.null?
 
-      # FIXME: check multi
-      hts_itr_next(htsfp[:fp][:bgzf], itr, r, htsfp)
+      if itr[:multi] == 1
+        hts_itr_multi_next(htsfp, itr, r)
+      else
+        bgzf = htsfp[:is_bgzf] == 1 ? htsfp[:fp][:bgzf] : FFI::Pointer::NULL
+        hts_itr_next(bgzf, itr, r, htsfp)
+      end
     end
 
     attach_function \
@@ -508,7 +512,7 @@ module HTS
     # Append tag data to a bam record
     attach_function \
       :bam_aux_append,
-      [Bam1, :string, :string, :int, :pointer],
+      [Bam1, :string, :char, :int, :pointer],
       :int
 
     # Delete tag data from a bam record
@@ -738,6 +742,18 @@ module HTS
       :bam_mods_recorded,
       [HtsBaseModState, :pointer],
       :pointer
+
+    # Sets the header to the file
+    attach_function \
+      :sam_hdr_set,
+      [HtsFile, SamHdr, :int],
+      :int
+
+    # Get the header from the file pointer
+    attach_function \
+      :sam_hdr_get,
+      [HtsFile],
+      SamHdr
   end
 end
 

data/lib/hts/libhts/vcf.rb

@@ -270,12 +270,12 @@ module HTS
     attach_function \
       :bcf_hdr_get_hrec,
       [BcfHdr, :int, :string, :string, :string],
-      BcfHrec.by_ref
+      :pointer
 
     # Duplicate a header record
     attach_function \
       :bcf_hrec_dup,
-      [BcfHrec],
+      [:pointer],
       BcfHrec.by_ref
 
     # Add a new header record key
@@ -304,7 +304,7 @@ module HTS
     # Free up a header record and associated structures
     attach_function \
       :bcf_hrec_destroy,
-      [BcfHrec],
+      [:pointer],
       :void
 
     # Individual record querying and manipulation routines
@@ -539,15 +539,18 @@ module HTS
       [HtsFile],
       :int
 
-    attach_function \
+    attach_variable \
       :bcf_float_vector_end,
-      [],
       :uint32
 
-    attach_function \
+    attach_variable \
       :bcf_float_missing,
-      [],
       :uint32
+
+    attach_function \
+      :bcf_format_gt_v2,
+      [BcfHdr, BcfFmt, :int, KString],
+      :int
   end
 end
 

data/lib/hts/libhts/vcf_funcs.rb

@@ -97,6 +97,11 @@ module HTS
         bcf_update_info(hdr, line, key, values, n, BCF_HT_INT)
       end
 
+      # Function for updating INFO fields (int64; VCF only)
+      def bcf_update_info_int64(hdr, line, key, values, n)
+        bcf_update_info(hdr, line, key, values, n, BCF_HT_LONG)
+      end
+
       # Function for updating INFO fields
       def bcf_update_info_float(hdr, line, key, values, n)
         bcf_update_info(hdr, line, key, values, n, BCF_HT_REAL)
@@ -150,9 +155,19 @@ module HTS
         0
       end
 
+      # Macro for setting genotypes correctly
+      def bcf_gt_vector_end
+        bcf_int32_vector_end
+      end
+
       # Macro for setting genotypes correctly
       def bcf_gt_is_missing(val)
-        (val >> 1 ? 0 : 1)
+        ((val >> 1) == 0 ? 1 : 0)
+      end
+
+      # Macro for setting genotypes correctly
+      def bcf_gt_is_vector_end(val)
+        val == bcf_gt_vector_end ? 1 : 0
       end
 
       # Macro for setting genotypes correctly
@@ -175,6 +190,11 @@ module HTS
         bcf_get_info_values(hdr, line, tag, dst, ndst, BCF_HT_INT)
       end
 
+      # Get INFO values (int64; VCF only)
+      def bcf_get_info_int64(hdr, line, tag, dst, ndst)
+        bcf_get_info_values(hdr, line, tag, dst, ndst, BCF_HT_LONG)
+      end
+
       # Get INFO values
       def bcf_get_info_float(hdr, line, tag, dst, ndst)
         bcf_get_info_values(hdr, line, tag, dst, ndst, BCF_HT_REAL)
@@ -265,7 +285,16 @@ module HTS
         )[:val][:info][type] & 0xf
       end
 
-      # def bcf_hdr_idinfo_exists
+      def bcf_hdr_idinfo_exists(hdr, type, int_id)
+        return false if int_id.negative? || int_id >= hdr[:n][LibHTS::BCF_DT_ID]
+
+        pair = LibHTS::BcfIdpair.new(
+          hdr[:id][LibHTS::BCF_DT_ID].to_ptr +
+          LibHTS::BcfIdpair.size * int_id # offset
+        )
+
+        !pair[:val].null? && bcf_hdr_id2coltype(hdr, type, int_id) != 0xf
+      end
 
       # def bcf_hdr_id2hrec
 

data/lib/hts/tabix.rb

@@ -68,6 +68,7 @@ module HTS
     end
 
     def load_index(index_name = nil)
+      check_closed
       if index_name
         LibHTS.tbx_index_load2(@file_name, index_name)
       else
@@ -76,17 +77,23 @@ module HTS
     end
 
     def index_loaded?
+      check_closed
       !@idx.null?
     end
 
     def name2id(name)
+      check_closed
       LibHTS.tbx_name2id(@idx, name)
     end
 
     def seqnames
+      check_closed
       nseq = FFI::MemoryPointer.new(:int)
-      LibHTS.tbx_seqnames(@idx, nseq).then do |pts|
+      pts = LibHTS.tbx_seqnames(@idx, nseq)
+      begin
         pts.read_array_of_pointer(nseq.read_int).map(&:read_string)
+      ensure
+        LibHTS.hts_free(pts) unless pts.null?
       end
     end
 
@@ -101,6 +108,18 @@ module HTS
       end
     end
 
+    def close
+      return if closed?
+
+      @idx.close if @idx && !@idx.null?
+      @idx = nil
+      super
+    end
+
+    def closed?
+      @hts_file.nil? || @hts_file.null?
+    end
+
     private
 
     def queryi(id, start, end_, &block)
@@ -126,10 +145,18 @@ module HTS
     def query_yield(qiter)
       r = LibHTS::KString.new
       begin
-        yield r[:s].split("\t") while LibHTS.tbx_itr_next(@hts_file, @idx, qiter, r) > 0
+        while (slen = LibHTS.tbx_itr_next(@hts_file, @idx, qiter, r)) >= 0
+          yield r.read_string_copy.split("\t")
+        end
+        raise if slen < -1
       ensure
+        r.free_buffer
         LibHTS.hts_itr_destroy(qiter)
       end
     end
+
+    def check_closed
+      raise IOError, "closed Tabix" if closed?
+    end
   end
 end

data/lib/hts/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HTS
-  VERSION = "0.3.1"
+  VERSION = "0.4.0"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: htslib
 version: !ruby/object:Gem::Version
-  version: 0.3.1
+  version: 0.4.0
 platform: ruby
 authors:
 - kojix2
@@ -71,13 +71,13 @@ files:
 - lib/hts/bam/pileup.rb
 - lib/hts/bam/record.rb
 - lib/hts/bcf.rb
+- lib/hts/bcf/errors.rb
 - lib/hts/bcf/format.rb
 - lib/hts/bcf/header.rb
 - lib/hts/bcf/header_record.rb
 - lib/hts/bcf/info.rb
 - lib/hts/bcf/record.rb
 - lib/hts/faidx.rb
-- lib/hts/faidx/sequence.rb
 - lib/hts/ffi_ext/README.md
 - lib/hts/ffi_ext/pointer.rb
 - lib/hts/ffi_ext/struct.rb
@@ -119,7 +119,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.6.9
+rubygems_version: 4.0.10
 specification_version: 4
 summary: HTSlib bindings for Ruby
 test_files: []

data/lib/hts/faidx/sequence.rb

@@ -1,62 +0,0 @@
-module HTS
-  class Faidx
-    class Sequence
-      attr_reader :name, :faidx
-
-      def initialize(faidx, name)
-        raise unless faidx.has_key?(name)
-
-        @faidx = faidx
-        @name = name
-      end
-
-      def length
-        faidx.seq_len(name)
-      end
-      alias size length
-
-      def seq(start = nil, stop = nil)
-        faidx.seq(name, start, stop)
-      end
-
-      def qual(start = nil, stop = nil)
-        faidx.qual(name, start, stop)
-      end
-
-      def [](arg)
-        case arg
-        when Integer
-          if arg >= 0
-            start = arg
-            stop = arg
-          else
-            start = length + arg
-            stop = length + arg
-          end
-        when Range
-          arg = Range.new(arg.begin, arg.end + length, arg.exclude_end?) if arg.end&.<(0)
-          arg = Range.new(arg.begin + length, arg.end, arg.exclude_end?) if arg.begin&.<(0)
-          if arg.begin.nil?
-            if arg.end.nil?
-              start = nil
-              stop = nil
-            else
-              start = 0
-              stop = arg.exclude_end? ? arg.end - 1 : arg.end
-            end
-          elsif arg.end.nil?
-            # always include the first base
-            start = arg.begin
-            stop = length - 1
-          else
-            start = arg.begin
-            stop = arg.exclude_end? ? arg.end - 1 : arg.end
-          end
-        else
-          raise ArgumentError
-        end
-        seq(start, stop)
-      end
-    end
-  end
-end