htslib 0.0.6 → 0.1.0

data/lib/hts/libhts/cram.rb

@@ -0,0 +1,292 @@
+# frozen_string_literal: true
+
+module HTS
+  module LibHTS
+    #
+    # attach_function \
+    #   :cram_fd_get_header,
+    #   [CramFd],
+    #   SamHdr.by_ref
+    #
+    # attach_function \
+    #   :cram_fd_set_header,
+    #   [CramFd, SamHdr.by_ref],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_fd_get_version,
+    #   [CramFd],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_fd_set_version,
+    #   [CramFd, :int],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_major_vers,
+    #   [CramFd],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_minor_vers,
+    #   [CramFd],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_fd_get_fp,
+    #   [CramFd],
+    #   HFile.by_ref
+    #
+    # attach_function \
+    #   :cram_fd_set_fp,
+    #   [CramFd, HFile],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_container_get_length,
+    #   [CramContainer],
+    #   :int32
+    #
+    # attach_function \
+    #   :cram_container_set_length,
+    #   [CramContainer, :int32],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_container_get_num_blocks,
+    #   [CramContainer],
+    #   :int32
+    #
+    # attach_function \
+    #   :cram_container_set_num_blocks,
+    #   [CramContainer, :int32],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_container_get_landmarks,
+    #   [CramContainer, :int32],
+    #   :pointer
+    #
+    # attach_function \
+    #   :cram_container_set_landmarks,
+    #   [CramContainer, :Int32, :pointer],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_container_is_empty,
+    #   [CramFd],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_block_get_content_id,
+    #   [CramBlock],
+    #   :int32
+    #
+    # attach_function \
+    #   :cram_block_get_comp_size,
+    #   [CramBlock],
+    #   :int32
+    #
+    # attach_function \
+    #   :cram_block_get_uncomp_size,
+    #   [CramBlock],
+    #   :int32
+    #
+    # attach_function \
+    #   :cram_block_get_crc32,
+    #   [CramBlock],
+    #   :int32
+    #
+    # attach_function \
+    #   :cram_block_get_data,
+    #   [CramBlock],
+    #   :pointer
+    #
+    # attach_function \
+    #   :cram_block_get_content_type,
+    #   [CramBlock],
+    #   CramContentType                # ?
+    #
+    # attach_function \
+    #   :cram_block_set_content_id,
+    #   [CramBlock, :int32],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_block_set_comp_size,
+    #   [CramBlock, :int32],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_block_set_uncomp_size,
+    #   [CramBlock, :int32],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_block_set_crc32,
+    #   [CramBlock, :int32],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_block_set_data,
+    #   [CramBlock, :pointer],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_block_append,
+    #   [CramBlock, :pointer, :int],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_block_update_size,
+    #   [CramBlock],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_block_get_offset,
+    #   [CramBlock],
+    #   :size_t
+    #
+    # attach_function \
+    #   :cram_block_set_offset,
+    #   [CramBlock, :size_t],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_block_size,
+    #   [CramBlock],
+    #   :uint32
+    #
+    # attach_function \
+    #   :cram_transcode_rg,
+    #   [CramFd, CramFd, CramContainer, :int, :pointer, :pointer],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_copy_slice,
+    #   [CramFd, CramFd, :int32],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_new_block,
+    #   [CramContentType, :int],
+    #   CramBlock
+    #
+    # attach_function \
+    #   :cram_read_block,
+    #   [CramFd],
+    #   CramBlock
+    #
+    # attach_function \
+    #   :cram_write_block,
+    #   [CramFd, CramBlock],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_free_block,
+    #   [CramBlock],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_uncompress_block,
+    #   [CramBlock],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_compress_block,
+    #   [CramFd, CramBlock, CramMetrics, :int, :int],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_compress_block2,
+    #   [CramFd, CramSlice, CramBlock, CramMetrics, :int, :int],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_new_container,
+    #   [:int, :int],
+    #   CramContainer
+    #
+    # attach_function \
+    #   :cram_free_container,
+    #   [CramContainer],
+    #   :void
+    #
+    # attach_function \
+    #   :cram_read_container,
+    #   [CramFd],
+    #   CramContainer
+    #
+    # attach_function \
+    #   :cram_write_container,
+    #   [CramFd, CramContainer],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_store_container,
+    #   [CramFd, CramContainer, :string, :pointer],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_container_size,
+    #   [CramContainer],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_open,
+    #   [:string, :string],
+    #   CramFd
+    #
+    # attach_function \
+    #   :cram_dopen,
+    #   [:pointer, :string, :string],
+    #   CramFd
+    #
+    # attach_function \
+    #   :cram_close,
+    #   [CramFd],
+    #   :int
+    #
+
+    attach_function \
+      :cram_seek,
+      %i[pointer off_t int], :int # FIXME: pointer should be CramFd
+
+    #
+    # attach_function \
+    #   :cram_flush,
+    #   [CramFd],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_eof,
+    #   [CramFd],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_set_option,
+    #   [CramFd, HtsFmtOption, ...], # vararg!
+    #   :int
+    #
+    # attach_function \
+    #   :cram_set_voption,
+    #   [CramFd, HtsFmtOption, VaList],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_set_header,
+    #   [CramFd, SamHdr.by_ref],
+    #   :int
+    #
+    # attach_function \
+    #   :cram_check_eof = :cram_check_EOF,
+    #   [CramFd], :int
+    #
+    # attach_function \
+    #   :cram_get_refs,
+    #   [HtsFile],
+    #   RefsT # what is RefsT
+    #
+  end
+end

data/lib/hts/libhts/hfile.rb

@@ -6,13 +6,13 @@ module HTS
     attach_function \
       :hopen,
       %i[string string varargs],
-      :HFILE
+      HFILE.by_ref
 
     # Associate a stream with an existing open file descriptor
     attach_function \
       :hdopen,
       %i[int string],
-      :HFILE
+      HFILE.by_ref
 
     # Report whether the file name or URL denotes remote storage
     attach_function \
@@ -29,55 +29,63 @@ module HTS
     # Flush (for output streams) and close the stream
     attach_function \
       :hclose,
-      [:HFILE],
+      [HFILE],
       :int
 
     # Close the stream, without flushing or propagating errors
     attach_function \
       :hclose_abruptly,
-      [:HFILE],
+      [HFILE],
       :void
 
     # Reposition the read/write stream offset
     attach_function \
       :hseek,
-      %i[HFILE off_t int],
+      [HFILE, :off_t, :int],
       :off_t
 
+    # 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)
+    end
+
     # Read from the stream until the delimiter, up to a maximum length
     attach_function \
       :hgetdelim,
-      %i[string size_t int HFILE],
+      [:string, :size_t, :int, HFILE],
       :ssize_t
 
     # Read a line from the stream, up to a maximum length
     attach_function \
       :hgets,
-      %i[string int HFILE],
+      [:string, :int, HFILE],
       :string
 
     # Peek at characters to be read without removing them from buffers
     attach_function \
       :hpeek,
-      %i[HFILE pointer size_t],
+      [HFILE, :pointer, :size_t],
       :ssize_t
 
     # For writing streams, flush buffered output to the underlying stream
     attach_function \
       :hflush,
-      [:HFILE],
+      [HFILE],
       :int
 
     # For hfile_mem: get the internal buffer and it's size from a hfile
     attach_function \
       :hfile_mem_get_buffer,
-      %i[HFILE pointer],
+      [HFILE, :pointer],
       :string
 
     # For hfile_mem: get the internal buffer and it's size from a hfile.
     attach_function \
       :hfile_mem_steal_buffer,
-      %i[HFILE pointer],
+      [HFILE, :pointer],
       :string
   end
 end

data/lib/hts/libhts/hts.rb

@@ -61,7 +61,7 @@ module HTS
     # Determine format by peeking at the start of a file
     attach_function \
       :hts_detect_format,
-      [:HFILE, HtsFormat],
+      [HFILE, HtsFormat],
       :int
 
     # Get a human-readable description of the file format
@@ -86,7 +86,7 @@ module HTS
     # Open an existing stream as a SAM/BAM/CRAM/VCF/BCF/etc file
     attach_function \
       :hts_hopen,
-      %i[HFILE string string],
+      [HFILE, :string, :string],
       HtsFile.by_ref
 
     # For output streams, flush any buffered data
@@ -145,7 +145,7 @@ module HTS
     # Create extra threads to aid compress/decompression for this file
     attach_function \
       :hts_set_thread_pool,
-      [HtsFile, HtsThreadPool],
+      [HtsFile, HtsTpool],
       :int
 
     # Adds a cache of decompressed blocks, potentially speeding up seeks.

data/lib/hts/libhts/sam_funcs.rb

@@ -345,7 +345,8 @@ 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
-      hts_log_error("Null iterator") if itr.null?
+      raise("Null iterator") if itr.null?
+
       # FIXME: check multi
       hts_itr_next(htsfp[:fp][:bgzf], itr, r, htsfp)
     end

data/lib/hts/libhts/thread_pool.rb

@@ -0,0 +1,139 @@
+# frozen_string_literal: true
+
+module HTS
+  module LibHTS
+    attach_function \
+      :hts_tpool_init,
+      [:int],
+      HtsTpool.by_ref
+
+    attach_function \
+      :hts_tpool_size,
+      [HtsTpool],
+      :int
+
+    # FIXME: struct
+    HtsTpoolProcess = :pointer
+    HtsTpoolResult  = :pointer
+
+    attach_function \
+      :hts_tpool_dispatch,
+      [HtsTpool, HtsTpoolProcess, :pointer, :pointer],
+      :int
+
+    attach_function \
+      :hts_tpool_dispatch2,
+      [HtsTpool, HtsTpoolProcess, :pointer, :pointer, :int],
+      :int
+
+    attach_function \
+      :hts_tpool_dispatch3,
+      [HtsTpool, HtsTpoolProcess, :pointer, :pointer, :pointer, :pointer, :int],
+      :int
+
+    attach_function \
+      :hts_tpool_wake_dispatch,
+      [HtsTpoolProcess],
+      :void
+
+    attach_function \
+      :hts_tpool_process_flush,
+      [HtsTpoolProcess],
+      :int
+
+    attach_function \
+      :hts_tpool_process_reset,
+      [HtsTpoolProcess, :int],
+      :int
+
+    attach_function \
+      :hts_tpool_process_qsize,
+      [HtsTpoolProcess],
+      :int
+
+    attach_function \
+      :hts_tpool_destroy,
+      [HtsTpool],
+      :void
+
+    attach_function \
+      :hts_tpool_kill,
+      [HtsTpool],
+      :void
+
+    attach_function \
+      :hts_tpool_next_result,
+      [HtsTpoolProcess],
+      HtsTpoolResult # .by_ref
+
+    attach_function \
+      :hts_tpool_next_result_wait,
+      [HtsTpoolProcess],
+      HtsTpoolResult # .by_ref
+
+    attach_function \
+      :hts_tpool_delete_result,
+      [HtsTpoolResult, :int],
+      :void
+
+    attach_function \
+      :hts_tpool_result_data,
+      [HtsTpoolResult],
+      :pointer
+
+    attach_function \
+      :hts_tpool_process_init,
+      [HtsTpool, :int, :int],
+      HtsTpoolProcess # .by_ref
+
+    attach_function \
+      :hts_tpool_process_destroy,
+      [HtsTpoolProcess],
+      :void
+
+    attach_function \
+      :hts_tpool_process_empty,
+      [HtsTpoolProcess],
+      :int
+
+    attach_function \
+      :hts_tpool_process_len,
+      [HtsTpoolProcess],
+      :int
+
+    attach_function \
+      :hts_tpool_process_sz,
+      [HtsTpoolProcess],
+      :int
+
+    attach_function \
+      :hts_tpool_process_shutdown,
+      [HtsTpoolProcess],
+      :void
+
+    attach_function \
+      :hts_tpool_process_is_shutdown,
+      [HtsTpoolProcess],
+      :int
+
+    attach_function \
+      :hts_tpool_process_attach,
+      [HtsTpool, HtsTpoolProcess],
+      :void
+
+    attach_function \
+      :hts_tpool_process_detach,
+      [HtsTpool, HtsTpoolProcess],
+      :void
+
+    attach_function \
+      :hts_tpool_process_ref_incr,
+      [HtsTpoolProcess],
+      :void
+
+    attach_function \
+      :hts_tpool_process_ref_decr,
+      [HtsTpoolProcess],
+      :void
+  end
+end

data/lib/hts/libhts/vcf.rb

@@ -221,6 +221,20 @@ module HTS
           LibHTS::BcfIdpair.size * int_id # offset
         )[:val][:info][type] >> 12
       end
+
+      def bcf_hdr_id2type(hdr, type, int_id)
+        LibHTS::BcfIdpair.new(
+          hdr[:id][LibHTS::BCF_DT_ID].to_ptr +
+          LibHTS::BcfIdpair.size * int_id # offset
+        )[:val][:info][type] >> 4 & 0xf
+      end
+
+      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
     end
   end
 end

data/lib/hts/libhts.rb

@@ -1,6 +1,7 @@
 # frozen_string_literal: true
 
 require_relative "ffi_ext/struct"
+require_relative "ffi_ext/pointer"
 
 module HTS
   module LibHTS
@@ -12,6 +13,9 @@ module HTS
       raise LoadError, "#{e}\nCould not find #{HTS.lib_path}"
     end
 
+    # @!macro attach_function
+    #   @!method $1(${2--2})
+    #   @return [${-1}] the return value of $0
     def self.attach_function(*)
       super
     rescue FFI::NotFoundError => e
@@ -23,11 +27,13 @@ end
 require_relative "libhts/constants"
 
 # This is alphabetical order.
+require_relative "libhts/kfunc"
 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/cram"
 require_relative "libhts/vcf"
+require_relative "libhts/tbx"
+require_relative "libhts/faidx"
+require_relative "libhts/thread_pool"

data/lib/hts/tbx.rb

@@ -0,0 +1,46 @@
+# frozen_string_literal: true
+
+require_relative "../htslib"
+
+require_relative "hts"
+
+module HTS
+  class Tbx < Hts
+    include Enumerable
+
+    attr_reader :file_name
+
+    def self.open(*args, **kw)
+      file = new(*args, **kw) # do not yield
+      return file unless block_given?
+
+      begin
+        yield file
+      ensure
+        file.close
+      end
+      file
+    end
+
+    def initialize(file_name, threads: nil)
+      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"
+      @hts_file = LibHTS.hts_open(@file_name, @mode)
+
+      raise Errno::ENOENT, "Failed to open #{@file_name}" if @hts_file.null?
+
+      if threads&.> 0
+        r = LibHTS.hts_set_threads(@hts_file, threads)
+        raise "Failed to set number of threads: #{threads}" if r < 0
+      end
+    end
+  end
+end

data/lib/hts/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HTS
-  VERSION = "0.0.6"
+  VERSION = "0.1.0"
 end

data/lib/htslib.rb

@@ -37,9 +37,9 @@ module HTS
   # You can change the path of the shared library with `HTS.lib_path=`
   # before calling the LibHTS module.
   autoload :LibHTS, "hts/libhts"
-end
 
-require_relative "hts/bam"
-require_relative "hts/faidx"
-require_relative "hts/tabix"
-require_relative "hts/bcf"
+  autoload :Bam,    "hts/bam"
+  autoload :Bcf,    "hts/bcf"
+  autoload :Tbx,    "hts/tbx"
+  autoload :Faidx,  "hts/faidx"
+end