htslib 0.2.0 → 0.2.3

data/lib/hts/libhts/vcf_funcs.rb

@@ -87,105 +87,130 @@ module HTS
     BCF_UN_ALL  = (BCF_UN_SHR | BCF_UN_FMT) # everything
 
     class << self
+      # Get number of samples
       def bcf_hdr_nsamples(hdr)
         hdr[:n][BCF_DT_SAMPLE]
       end
 
+      # Function for updating INFO fields
       def bcf_update_info_int32(hdr, line, key, values, n)
         bcf_update_info(hdr, line, key, values, n, BCF_HT_INT)
       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)
       end
 
+      # Function for updating INFO fields
       def bcf_update_info_flag(hdr, line, key, string, n)
         bcf_update_info(hdr, line, key, string, n, BCF_HT_FLAG)
       end
 
+      # Function for updating INFO fields
       def bcf_update_info_string(hdr, line, key, string)
         bcf_update_info(hdr, line, key, string, 1, BCF_HT_STR)
       end
 
+      # Function for updating FORMAT fields
       def bcf_update_format_int32(hdr, line, key, values, n)
         bcf_update_format(hdr, line, key, values, n,
                           BCF_HT_INT)
       end
 
+      # Function for updating FORMAT fields
       def bcf_update_format_float(hdr, line, key, values, n)
         bcf_update_format(hdr, line, key, values, n,
                           BCF_HT_REAL)
       end
 
+      # Function for updating FORMAT fields
       def bcf_update_format_char(hdr, line, key, values, n)
         bcf_update_format(hdr, line, key, values, n,
                           BCF_HT_STR)
       end
 
+      # Function for updating FORMAT fields
       def bcf_update_genotypes(hdr, line, gts, n)
         bcf_update_format(hdr, line, "GT", gts, n, BCF_HT_INT)
       end
 
+      # Macro for setting genotypes correctly
       def bcf_gt_phased(idx)
         ((idx + 1) << 1 | 1)
       end
 
+      # Macro for setting genotypes correctly
       def bcf_gt_unphased(idx)
         ((idx + 1) << 1)
       end
 
+      # Macro for setting genotypes correctly
       def bcf_gt_missing
         0
       end
 
+      # Macro for setting genotypes correctly
       def bcf_gt_is_missing(val)
-        ((val) >> 1 ? 0 : 1)
+        (val >> 1 ? 0 : 1)
       end
 
+      # Macro for setting genotypes correctly
       def bcf_gt_is_phased(idx)
-        ((idx) & 1)
+        (idx & 1)
       end
 
+      # Macro for setting genotypes correctly
       def bcf_gt_allele(val)
-        (((val) >> 1) - 1)
+        ((val >> 1) - 1)
       end
 
+      # Conversion between alleles indexes to Number=G genotype index (assuming diploid, all 0-based)
       def bcf_alleles2gt(a, b)
         ((a) > (b) ? (a * (a + 1) / 2 + b) : (b * (b + 1) / 2 + a))
       end
 
+      # Get INFO values
       def bcf_get_info_int32(hdr, line, tag, dst, ndst)
         bcf_get_info_values(hdr, line, tag, dst, ndst, BCF_HT_INT)
       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)
       end
 
+      # Get INFO values
       def bcf_get_info_string(hdr, line, tag, dst, ndst)
         bcf_get_info_values(hdr, line, tag, dst, ndst, BCF_HT_STR)
       end
 
+      # Get INFO values
       def bcf_get_info_flag(hdr, line, tag, dst, ndst)
         bcf_get_info_values(hdr, line, tag, dst, ndst, BCF_HT_FLAG)
       end
 
+      # Get FORMAT values
       def bcf_get_format_int32(hdr, line, tag, dst, ndst)
         bcf_get_format_values(hdr, line, tag, dst, ndst, BCF_HT_INT)
       end
 
+      # Get FORMAT values
       def bcf_get_format_float(hdr, line, tag, dst, ndst)
         bcf_get_format_values(hdr, line, tag, dst, ndst, BCF_HT_REAL)
       end
 
+      # Get FORMAT values
       def bcf_get_format_char(hdr, line, tag, dst, ndst)
         bcf_get_format_values(hdr, line, tag, dst, ndst, BCF_HT_STR)
       end
 
+      # Get FORMAT values
       def bcf_get_genotypes(hdr, line, dst, ndst)
         bcf_get_format_values(hdr, line, "GT", dst, ndst, BCF_HT_INT)
       end
 
+      # Translates numeric ID into string
       def bcf_hdr_int2id(hdr, type, int_id)
         LibHTS::BcfIdpair.new(
           hdr[:id][type].to_ptr +
@@ -193,10 +218,12 @@ module HTS
         )[:key]
       end
 
+      # Translates sequence names (chromosomes) into numeric ID
       def bcf_hdr_name2id(hdr, id)
         bcf_hdr_id2int(hdr, BCF_DT_CTG, id)
       end
 
+      # Translates numeric ID to sequence name
       def bcf_hdr_id2name(hdr, rid)
         return nil if hdr.null? || rid < 0 || rid >= hdr[:n][LibHTS::BCF_DT_CTG]
 
@@ -206,6 +233,7 @@ module HTS
         )[:key]
       end
 
+      # Macro for accessing bcf_idinfo_t
       def bcf_hdr_id2length(hdr, type, int_id)
         LibHTS::BcfIdpair.new(
           hdr[:id][LibHTS::BCF_DT_ID].to_ptr +
@@ -213,6 +241,7 @@ module HTS
         )[:val][:info][type] >> 8 & 0xf
       end
 
+      # Macro for accessing bcf_idinfo_t
       def bcf_hdr_id2number(hdr, type, int_id)
         LibHTS::BcfIdpair.new(
           hdr[:id][LibHTS::BCF_DT_ID].to_ptr +
@@ -220,6 +249,7 @@ module HTS
         )[:val][:info][type] >> 12
       end
 
+      # Macro for accessing bcf_idinfo_t
       def bcf_hdr_id2type(hdr, type, int_id)
         LibHTS::BcfIdpair.new(
           hdr[:id][LibHTS::BCF_DT_ID].to_ptr +
@@ -227,12 +257,60 @@ module HTS
         )[:val][:info][type] >> 4 & 0xf
       end
 
+      # Macro for accessing bcf_idinfo_t
       def bcf_hdr_id2coltype(hdr, type, int_id)
         LibHTS::BcfIdpair.new(
           hdr[:id][LibHTS::BCF_DT_ID].to_ptr +
           LibHTS::BcfIdpair.size * int_id # offset
         )[:val][:info][type] & 0xf
       end
+
+      # def bcf_hdr_idinfo_exists
+
+      # def bcf_hdr_id2hrec
+
+      alias bcf_itr_destroy hts_itr_destroy
+
+      def bcf_itr_queryi(idx, tid, beg, _end)
+        hts_itr_query(idx, tid, beg, _end, @@bcf_readrec)
+      end
+
+      @@bcf_hdr_name2id = proc do |hdr, id|
+        LibHTS.bcf_hdr_name2id(hdr, id)
+      end
+
+      def bcf_itr_querys(idx, hdr, s)
+        hts_itr_querys(idx, s, @@bcf_hdr_name2id, hdr, @@hts_itr_query, @@bcf_readrec)
+      end
+
+      # Load a BCF index
+      def bcf_index_load(fn)
+        hts_idx_load(fn, HTS_FMT_CSI)
+      end
+
+      # Load a BCF index
+      def bcf_index_seqnames(idx, hdr, nptr)
+        hts_idx_seqnames(idx, nptr, @@bcf_hdr_id2name, hdr)
+      end
+
+      # Typed value I/O
+      def bcf_int8_vector_end  = -127                       # INT8_MIN  + 1
+      def bcf_int16_vector_end = -32_767                    # INT16_MIN + 1
+      def bcf_int32_vector_end = -2_147_483_647             # INT32_MIN + 1
+      def bcf_int64_vector_end = -9_223_372_036_854_775_807 # INT64_MIN + 1
+      def bcf_str_vector_end   = 0
+      def bcf_int8_missing     = -128                             # INT8_MIN
+      def bcf_int16_missing    = (-32_767 - 1)                    # INT16_MIN
+      def bcf_int32_missing    = (-2_147_483_647 - 1)             # INT32_MIN
+      def bcf_int64_missing    = (-9_223_372_036_854_775_807 - 1) # INT64_MIN
+      def bcf_str_missing      = 0x07
+
+      BCF_MAX_BT_INT8  = 0x7f           # INT8_MAX  */
+      BCF_MAX_BT_INT16 = 0x7fff         # INT16_MAX */
+      BCF_MAX_BT_INT32 = 0x7fffffff     # INT32_MAX */
+      BCF_MIN_BT_INT8  = -120           # INT8_MIN  + 8 */
+      BCF_MIN_BT_INT16 = -32_760        # INT16_MIN + 8 */
+      BCF_MIN_BT_INT32 = -2_147_483_640 # INT32_MIN + 8 */
     end
   end
 end

data/lib/hts/libhts.rb

@@ -15,12 +15,13 @@ module HTS
     end
 
     # @!macro attach_function
+    #   @!scope class
     #   @!method $1(${2--2})
     #   @return [${-1}] the return value of $0
     def self.attach_function(*)
       super
     rescue FFI::NotFoundError => e
-      warn e.message
+      warn e.message if $VERBOSE
     end
   end
 end

data/lib/hts/tbx.rb

@@ -8,7 +8,7 @@ module HTS
   class Tbx < Hts
     include Enumerable
 
-    attr_reader :file_name
+    attr_reader :file_name, :index_name, :mode, :nthreads
 
     def self.open(*args, **kw)
       file = new(*args, **kw) # do not yield
@@ -22,24 +22,56 @@ module HTS
       file
     end
 
-    def initialize(file_name, threads: nil)
+    def initialize(file_name, mode = "r", index: nil, threads: nil, build_index: false)
       if block_given?
         message = "HTS::Tbx.new() dose not take block; Please use HTS::Tbx.open() instead"
         raise message
       end
 
-      @file_name = file_name
-
       # NOTE: Do not check for the existence of local files, since file_names may be remote URIs.
 
-      @mode     = "r"
+      @file_name  = file_name
+      @index_name = index
+      @mode       = mode
+      @nthreads   = threads
       @hts_file = LibHTS.hts_open(@file_name, @mode)
 
       raise Errno::ENOENT, "Failed to open #{@file_name}" if @hts_file.null?
 
       set_threads(threads) if threads
 
+      # return if @mode[0] == "w"
+      raise "Not implemented" if @mode[0] == "w"
+
+      # build_index(index) if build_index
+      @idx = load_index(index)
+
       super # do nothing
     end
+
+    def build_index
+      raise "Not implemented yet"
+    end
+
+    def load_index(index_name = nil)
+      if index_name
+        LibHTS.tbx_index_load2(@file_name, index_name)
+      else
+        LibHTS.tbx_index_load3(@file_name, nil, 2)
+      end
+    end
+
+    def tid(name)
+      LibHTS.tbx_name2id(@idx, name)
+    end
+
+    def seqnames
+      nseq = FFI::MemoryPointer.new(:int)
+      LibHTS.tbx_seqnames(@idx, nseq).then do |pts|
+        pts.read_array_of_pointer(nseq.read_int).map do |pt|
+          pt.read_string
+        end
+      end
+    end
   end
 end

data/lib/hts/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HTS
-  VERSION = "0.2.0"
+  VERSION = "0.2.3"
 end

data/lib/htslib.rb

@@ -30,9 +30,23 @@ module HTS
 
       warn "htslib shared library '#{name}' not found."
     end
+
+    def search_htslib_windows
+      ENV["HTSLIBDIR"] ||= [
+        RubyInstaller::Runtime.msys2_installation.msys_path,
+        RubyInstaller::Runtime.msys2_installation.mingwarch
+      ].join(File::ALT_SEPARATOR)
+      path = File.expand_path("bin/hts-3.dll", ENV["HTSLIBDIR"])
+      RubyInstaller::Runtime.add_dll_directory(File.dirname(path))
+      path
+    end
   end
 
-  self.lib_path = search_htslib
+  self.lib_path = if Object.const_defined?(:RubyInstaller)
+                    search_htslib_windows
+                  else
+                    search_htslib
+                  end
 
   # You can change the path of the shared library with `HTS.lib_path=`
   # before calling the LibHTS module.

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: htslib
 version: !ruby/object:Gem::Version
-  version: 0.2.0
+  version: 0.2.3
 platform: ruby
 authors:
 - kojix2
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2022-05-29 00:00:00.000000000 Z
+date: 2022-09-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ffi
@@ -132,7 +132,7 @@ files:
 - LICENSE.txt
 - README.md
 - lib/hts/bam.rb
-- lib/hts/bam/aux.rb
+- lib/hts/bam/auxi.rb
 - lib/hts/bam/cigar.rb
 - lib/hts/bam/flag.rb
 - lib/hts/bam/header.rb
@@ -140,6 +140,7 @@ files:
 - lib/hts/bcf.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
@@ -158,6 +159,7 @@ files:
 - lib/hts/libhts/sam.rb
 - lib/hts/libhts/sam_funcs.rb
 - lib/hts/libhts/tbx.rb
+- lib/hts/libhts/tbx_funcs.rb
 - lib/hts/libhts/thread_pool.rb
 - lib/hts/libhts/vcf.rb
 - lib/hts/libhts/vcf_funcs.rb
@@ -167,7 +169,8 @@ files:
 homepage: https://github.com/kojix2/ruby-htslib
 licenses:
 - MIT
-metadata: {}
+metadata:
+  msys2_mingw_dependencies: htslib
 post_install_message: 
 rdoc_options: []
 require_paths: