htslib 0.0.0 → 0.0.4

data/lib/hts/{ffi → libhts}/sam.rb

@@ -1,7 +1,24 @@
 # frozen_string_literal: true
 
 module HTS
-  module FFI
+  module LibHTS
+    # constants
+    BAM_CMATCH     = 0
+    BAM_CINS       = 1
+    BAM_CDEL       = 2
+    BAM_CREF_SKIP  = 3
+    BAM_CSOFT_CLIP = 4
+    BAM_CHARD_CLIP = 5
+    BAM_CPAD       = 6
+    BAM_CEQUAL     = 7
+    BAM_CDIFF      = 8
+    BAM_CBACK      = 9
+
+    BAM_CIGAR_STR = "MIDNSHP=XB"
+    BAM_CIGAR_SHIFT = 4
+    BAM_CIGAR_MASK  = 0xf
+    BAM_CIGAR_TYPE  = 0x3C1A7
+
     # macros
     class << self
       def bam_cigar_op(c)
@@ -12,8 +29,8 @@ module HTS
         c >> BAM_CIGAR_SHIFT
       end
 
-      def bam_cigar_opchar(c)
-        _BAM_CIGAR_STR_PADDED[bam_cigar_op(c)]
+      def bam_cigar_opchr(c)
+        ("#{BAM_CIGAR_STR}??????")[bam_cigar_op(c)]
       end
 
       def bam_cigar_gen(l, o)
@@ -25,6 +42,19 @@ module HTS
       end
     end
 
+    BAM_FPAIRED        =    1
+    BAM_FPROPER_PAIR   =    2
+    BAM_FUNMAP         =    4
+    BAM_FMUNMAP        =    8
+    BAM_FREVERSE       =   16
+    BAM_FMREVERSE      =   32
+    BAM_FREAD1         =   64
+    BAM_FREAD2         =  128
+    BAM_FSECONDARY     =  256
+    BAM_FQCFAIL        =  512
+    BAM_FDUP           = 1024
+    BAM_FSUPPLEMENTARY = 2048
+
     # macros
     # function-like macros
     class << self
@@ -60,7 +90,9 @@ module HTS
         b[:l_data] - (b[:core][:n_cigar] << 2) - b[:core][:l_qname] - b[:core][:l_qseq] - ((b[:core][:l_qseq] + 1) >> 1)
       end
 
-      # def bam_seqi(s, i)
+      def bam_seqi(s, i)
+        s[(i) >> 1].read_uint8 >> ((~i & 1) << 2) & 0xf
+      end
 
       # def bam_set_seqi(s, i, b)
     end
@@ -146,13 +178,13 @@ module HTS
     # Returns a complete line of formatted text for a given type and ID.
     attach_function \
       :sam_hdr_find_line_id,
-      [SamHdr, :string, :string, :string, Kstring],
+      [SamHdr, :string, :string, :string, KString],
       :int
 
     # Returns a complete line of formatted text for a given type and index.
     attach_function \
       :sam_hdr_find_line_pos,
-      [SamHdr, :string, :int, Kstring],
+      [SamHdr, :string, :int, KString],
       :int
 
     # Remove a line with given type / id from a header
@@ -206,13 +238,13 @@ module HTS
     # Return the value associated with a key for a header line identified by ID_key:ID_val
     attach_function \
       :sam_hdr_find_tag_id,
-      [SamHdr, :string, :string, :string, :string, Kstring],
+      [SamHdr, :string, :string, :string, :string, KString],
       :int
 
     # Return the value associated with a key for a header line identified by position
     attach_function \
       :sam_hdr_find_tag_pos,
-      [SamHdr, :string, :int, :string, Kstring],
+      [SamHdr, :string, :int, :string, KString],
       :int
 
     # Remove the key from the line identified by type, ID_key and ID_value.
@@ -405,6 +437,14 @@ module HTS
       [HtsIdx, SamHdr, :pointer, :uint],
       HtsItr.by_ref
 
+    # 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
+      hts_log_error("Null iterator") if itr.null?
+      # FIXME: check multi
+      hts_itr_next(htsfp[:fp][:bgzf], itr, r, htsfp)
+    end
+
     attach_function \
       :sam_parse_region,
       [SamHdr, :string, :pointer, :pointer, :pointer, :int],
@@ -435,12 +475,12 @@ module HTS
 
     attach_function \
       :sam_parse1,
-      [Kstring, SamHdr, Bam1],
+      [KString, SamHdr, Bam1],
       :int
 
     attach_function \
       :sam_format1,
-      [SamHdr, Bam1, Kstring],
+      [SamHdr, Bam1, KString],
       :int
 
     # Read a record from a file
@@ -543,52 +583,52 @@ module HTS
     attach_function \
       :bam_plp_init,
       %i[bam_plp_auto_f pointer],
-      BamPlp
+      :bam_plp
 
     attach_function \
       :bam_plp_destroy,
-      [BamPlp],
+      [:bam_plp],
       :void
 
     attach_function \
       :bam_plp_push,
-      [BamPlp, Bam1],
+      [:bam_plp, Bam1],
       :int
 
     attach_function \
       :bam_plp_next,
-      [BamPlp, :pointer, :pointer, :pointer],
-      :pointer
+      %i[bam_plp pointer pointer pointer],
+      BamPileup1.by_ref
 
     attach_function \
       :bam_plp_auto,
-      [BamPlp, :pointer, :pointer, :pointer],
-      :pointer
+      %i[bam_plp pointer pointer pointer],
+      BamPileup1.by_ref
 
     attach_function \
       :bam_plp64_next,
-      [BamPlp, :pointer, :pointer, :pointer],
-      :pointer
+      %i[bam_plp pointer pointer pointer],
+      BamPileup1.by_ref
 
     attach_function \
       :bam_plp64_auto,
-      [BamPlp, :pointer, :pointer, :pointer],
-      :pointer
+      %i[bam_plp pointer pointer pointer],
+      BamPileup1.by_ref
 
     attach_function \
       :bam_plp_set_maxcnt,
-      [BamPlp, :int],
+      %i[bam_plp int],
       :void
 
     attach_function \
       :bam_plp_reset,
-      [BamPlp],
+      [:bam_plp],
       :void
 
     # sets a callback to initialise any per-pileup1_t fields.
     attach_function \
       :bam_plp_insertion,
-      [:pointer, Kstring, :pointer],
+      [BamPileup1, KString, :pointer],
       :int
 
     # sets a callback to initialise any per-pileup1_t fields.
@@ -602,36 +642,36 @@ module HTS
     attach_function \
       :bam_mplp_init,
       %i[int bam_plp_auto_f pointer],
-      BamMplp.by_ref
+      :bam_mplp
 
     attach_function \
       :bam_mplp_init_overlaps,
-      [BamMplp],
+      [:bam_mplp],
       :int
 
     attach_function \
       :bam_mplp_destroy,
-      [BamMplp],
+      [:bam_mplp],
       :void
 
     attach_function \
       :bam_mplp_set_maxcnt,
-      [BamMplp, :int],
+      %i[bam_mplp int],
       :void
 
     attach_function \
       :bam_mplp_auto,
-      [BamMplp, :pointer, :pointer, :pointer, :pointer],
+      %i[bam_mplp pointer pointer pointer pointer], # BamPileup1T
       :int
 
     attach_function \
       :bam_mplp64_auto,
-      [BamMplp, :pointer, :pointer, :pointer, :pointer],
+      %i[bam_mplp pointer pointer pointer pointer], # BamPileup1T
       :int
 
     attach_function \
       :bam_mplp_reset,
-      [BamMplp],
+      [:bam_mplp],
       :void
 
     # bam_mplp_constructor

data/lib/hts/{ffi → libhts}/tbx.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 module HTS
-  module FFI
+  module LibHTS
     attach_function \
       :tbx_name2id,
       [Tbx, :string],

data/lib/hts/{ffi → libhts}/vcf.rb

@@ -1,7 +1,183 @@
 # frozen_string_literal: true
 
 module HTS
-  module FFI
+  module LibHTS
+    # constants
+    BCF_HL_FLT    = 0 # header line
+    BCF_HL_INFO   = 1
+    BCF_HL_FMT    = 2
+    BCF_HL_CTG    = 3
+    BCF_HL_STR    = 4 # structured header line TAG=<A=..,B=..>
+    BCF_HL_GEN    = 5 # generic header line
+    BCF_HT_FLAG   = 0 # header type
+
+    BCF_HT_INT    = 1
+    BCF_HT_REAL   = 2
+    BCF_HT_STR    = 3
+    BCF_HT_LONG   = (BCF_HT_INT | 0x100) # BCF_HT_INT, but for int64_t values; VCF only!
+
+    BCF_VL_FIXED  = 0 # variable length
+    BCF_VL_VAR    = 1
+    BCF_VL_A      = 2
+    BCF_VL_G      = 3
+    BCF_VL_R      = 4
+
+    BCF_DT_ID     = 0 # dictionary type
+    BCF_DT_CTG    = 1
+    BCF_DT_SAMPLE = 2
+
+    BCF_BT_NULL   = 0
+    BCF_BT_INT8   = 1
+    BCF_BT_INT16  = 2
+    BCF_BT_INT32  = 3
+    BCF_BT_INT64  = 4 # Unofficial, for internal use only.
+    BCF_BT_FLOAT  = 5
+    BCF_BT_CHAR   = 7
+
+    VCF_REF       = 0
+    VCF_SNP       = 1
+    VCF_MNP       = 2
+    VCF_INDEL     = 4
+    VCF_OTHER     = 8
+    VCF_BND       = 16 # breakend
+    VCF_OVERLAP   = 32 # overlapping deletion, ALT=*
+
+    BCF1_DIRTY_ID  = 1
+    BCF1_DIRTY_ALS = 2
+    BCF1_DIRTY_FLT = 4
+    BCF1_DIRTY_INF = 8
+
+    BCF_ERR_CTG_UNDEF   = 1
+    BCF_ERR_TAG_UNDEF   = 2
+    BCF_ERR_NCOLS       = 4
+    BCF_ERR_LIMITS      = 8
+    BCF_ERR_CHAR        = 16
+    BCF_ERR_CTG_INVALID = 32
+    BCF_ERR_TAG_INVALID = 64
+
+    # macros
+    class << self
+      alias bcf_open hts_open
+      alias vcf_open hts_open
+      alias bcf_close hts_close
+      alias vcf_close hts_close
+
+      def bcf_hdr_nsamples(hdr)
+        hdr[:n][BCF_DT_SAMPLE]
+      end
+
+      def bcf_update_info_int32(hdr, line, key, values, n)
+        bcf_update_info(hdr, line, key, values, n, BCF_HT_INT)
+      end
+
+      def bcf_update_info_float(hdr, line, key, values, n)
+        bcf_update_info(hdr, line, key, values, n, BCF_HT_REAL)
+      end
+
+      def bcf_update_info_flag(hdr, line, key, string, n)
+        bcf_update_info(hdr, line, key, string, n, BCF_HT_FLAG)
+      end
+
+      def bcf_update_info_string(hdr, line, key, string)
+        bcf_update_info(hdr, line, key, string, 1, BCF_HT_STR)
+      end
+
+      def bcf_update_format_int32(hdr, line, key, values, n)
+        bcf_update_format(hdr, line, key, values, n,
+                          BCF_HT_INT)
+      end
+
+      def bcf_update_format_float(hdr, line, key, values, n)
+        bcf_update_format(hdr, line, key, values, n,
+                          BCF_HT_REAL)
+      end
+
+      def bcf_update_format_char(hdr, line, key, values, n)
+        bcf_update_format(hdr, line, key, values, n,
+                          BCF_HT_STR)
+      end
+
+      def bcf_update_genotypes(hdr, line, gts, n)
+        bcf_update_format(hdr, line, "GT", gts, n, BCF_HT_INT)
+      end
+
+      def bcf_gt_phased(idx)
+        ((idx + 1) << 1 | 1)
+      end
+
+      def bcf_gt_unphased(idx)
+        ((idx + 1) << 1)
+      end
+
+      def bcf_gt_missing
+        0
+      end
+
+      def bcf_gt_is_missing(val)
+        ((val) >> 1 ? 0 : 1)
+      end
+
+      def bcf_gt_is_phased(idx)
+        ((idx) & 1)
+      end
+
+      def bcf_gt_allele(val)
+        (((val) >> 1) - 1)
+      end
+
+      def bcf_alleles2gt(a, b)
+        ((a) > (b) ? (a * (a + 1) / 2 + b) : (b * (b + 1) / 2 + a))
+      end
+
+      def bcf_get_info_int32(hdr, line, tag, dst, ndst)
+        bcf_get_info_values(hdr, line, tag, dst, ndst, BCF_HT_INT)
+      end
+
+      def bcf_get_info_float(hdr, line, tag, dst, ndst)
+        bcf_get_info_values(hdr, line, tag, dst, ndst, BCF_HT_REAL)
+      end
+
+      def bcf_get_info_string(hdr, line, tag, dst, ndst)
+        bcf_get_info_values(hdr, line, tag, dst, ndst, BCF_HT_STR)
+      end
+
+      def bcf_get_info_flag(hdr, line, tag, dst, ndst)
+        bcf_get_info_values(hdr, line, tag, dst, ndst, BCF_HT_FLAG)
+      end
+
+      def bcf_get_format_int32(hdr, line, tag, dst, ndst)
+        bcf_get_format_values(hdr, line, tag, dst, ndst, BCF_HT_INT)
+      end
+
+      def bcf_get_format_float(hdr, line, tag, dst, ndst)
+        bcf_get_format_values(hdr, line, tag, dst, ndst, BCF_HT_REAL)
+      end
+
+      def bcf_get_format_char(hdr, line, tag, dst, ndst)
+        bcf_get_format_values(hdr, line, tag, dst, ndst, BCF_HT_STR)
+      end
+
+      def bcf_get_genotypes(hdr, line, dst, ndst)
+        bcf_get_format_values(hdr, line, "GT", dst, ndst, BCF_HT_INT)
+      end
+
+      def bcf_hdr_int2id(hdr, type, int_id)
+        LibHTS::BcfIdpair.new(
+          hdr[:id][type].to_ptr +
+          LibHTS::BcfIdpair.size * int_id # offsets
+        )[:key]
+      end
+    end
+
+    # constants
+    BCF_UN_STR  = 1 # up to ALT inclusive
+    BCF_UN_FLT  = 2 # up to FILTER
+    BCF_UN_INFO = 4 # up to INFO
+    BCF_UN_SHR  = (BCF_UN_STR | BCF_UN_FLT | BCF_UN_INFO) # all shared information
+    BCF_UN_FMT  = 8 # unpack format and each sample
+    BCF_UN_IND  = BCF_UN_FMT # a synonym of BCF_UN_FMT
+    BCF_UN_ALL  = (BCF_UN_SHR | BCF_UN_FMT) # everything
+
     attach_function \
       :bcf_hdr_init,
       [:string],
@@ -14,7 +190,7 @@ module HTS
 
     attach_function \
       :bcf_init,
-      [:void],
+      [],
       Bcf1.by_ref
 
     attach_function \
@@ -54,7 +230,7 @@ module HTS
 
     attach_function \
       :vcf_parse,
-      [Kstring, BcfHdr, Bcf1],
+      [KString, BcfHdr, Bcf1],
       :int
 
     attach_function \
@@ -64,7 +240,7 @@ module HTS
 
     attach_function \
       :vcf_format,
-      [BcfHdr, Bcf1, Kstring],
+      [BcfHdr, Bcf1, KString],
       :int
 
     attach_function \
@@ -119,7 +295,7 @@ module HTS
 
     attach_function \
       :vcf_write_line,
-      [HtsFile, Kstring],
+      [HtsFile, KString],
       :int
 
     attach_function \
@@ -144,7 +320,7 @@ module HTS
 
     attach_function \
       :bcf_hdr_format,
-      [BcfHdr, :int, Kstring],
+      [BcfHdr, :int, KString],
       :int
 
     attach_function \
@@ -204,7 +380,7 @@ module HTS
 
     attach_function \
       :bcf_hrec_format,
-      [BcfHrec, Kstring],
+      [BcfHrec, KString],
       :int
 
     attach_function \
@@ -369,27 +545,27 @@ module HTS
 
     attach_function \
       :bcf_fmt_array,
-      [Kstring, :int, :int, :pointer],
+      [KString, :int, :int, :pointer],
       :int
 
     attach_function \
       :bcf_fmt_sized_array,
-      [Kstring, :pointer],
+      [KString, :pointer],
       :uint8_t
 
     attach_function \
       :bcf_enc_vchar,
-      [Kstring, :int, :string],
+      [KString, :int, :string],
       :int
 
     attach_function \
       :bcf_enc_vint,
-      [Kstring, :int, :pointer, :int],
+      [KString, :int, :pointer, :int],
       :int
 
     attach_function \
       :bcf_enc_vfloat,
-      [Kstring, :int, :pointer],
+      [KString, :int, :pointer],
       :int
 
     attach_function \

data/lib/hts/libhts.rb

@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+
+require_relative "ffi_ext/struct"
+
+module HTS
+  module LibHTS
+    extend FFI::Library
+
+    begin
+      ffi_lib HTS.lib_path
+    rescue LoadError => e
+      raise LoadError, "#{e}\nCould not find #{HTS.lib_path}"
+    end
+
+    def self.attach_function(*)
+      super
+    rescue FFI::NotFoundError => e
+      warn e.message
+    end
+  end
+end
+
+require_relative "libhts/constants"
+
+# This is alphabetical order.
+require_relative "libhts/bgzf"
+require_relative "libhts/faidx"
+require_relative "libhts/hfile"
+require_relative "libhts/hts"
+require_relative "libhts/sam"
+require_relative "libhts/kfunc"
+require_relative "libhts/tbx"
+require_relative "libhts/vcf"

data/lib/hts/tabix.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+# Based on hts-python
+# https://github.com/quinlan-lab/hts-python
+
+require_relative "utils/open_method"
+
+module HTS
+  class Tabix
+    extend Utils::OpenMethod
+
+    def initialize; end
+
+    def build; end
+
+    def sequences; end
+
+    # def __call__\
+  end
+end

data/lib/hts/utils/open_method.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module HTS
+  module Utils
+    module OpenMethod
+      def open(path)
+        object = new(path)
+        if block_given?
+          yield(object)
+          object.close
+        else
+          object
+        end
+      end
+    end
+  end
+end

data/lib/hts/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HTS
-  VERSION = '0.0.0'
+  VERSION = "0.0.4"
 end

data/lib/htslib.rb

@@ -1,18 +1,45 @@
 # frozen_string_literal: true
 
-require 'ffi'
+require "ffi"
 
-require 'hts/version'
+require "hts/version"
 
 module HTS
   class Error < StandardError; end
 
   class << self
-    attr_accessor :ffi_lib
+    attr_accessor :lib_path
+
+    def search_htslib(name = nil)
+      name ||= "libhts.#{FFI::Platform::LIBSUFFIX}"
+      lib_path = if ENV["HTSLIBDIR"]
+                   File.expand_path(name, ENV["HTSLIBDIR"])
+                 else
+                   File.expand_path("../vendor/#{name}", __dir__)
+                 end
+      return lib_path if File.exist?(lib_path)
+
+      begin
+        require "pkg-config"
+        lib_dir = PKGConfig.variable("htslib", "libdir")
+        lib_path = File.expand_path(name, lib_dir)
+      rescue PackageConfig::NotFoundError
+        warn "htslib.pc was not found in the pkg-config search path."
+      end
+      return lib_path if File.exist?(lib_path)
+
+      warn "htslib shared library '#{name}' not found."
+    end
   end
-  self.ffi_lib = File.expand_path("libhts.#{FFI::Platform::LIBSUFFIX}", ENV['HTSLIBDIR'])
-  autoload :FFI, 'hts/ffi'
+
+  self.lib_path = search_htslib
+
+  # You can change the path of the shared library with `HTS.lib_path=`
+  # before calling the LibHTS module.
+  autoload :LibHTS, "hts/libhts"
 end
 
-# alias
-HTSlib = HTS
+require_relative "hts/bam"
+require_relative "hts/faidx"
+require_relative "hts/tabix"
+require_relative "hts/bcf"