biosyntax 0.1.0 → 0.1.2

data/README.md

@@ -1,7 +1,10 @@
 # ruby-biosyntax
 
 [![CI](https://github.com/kojix2/ruby-biosyntax/actions/workflows/ci.yml/badge.svg)](https://github.com/kojix2/ruby-biosyntax/actions/workflows/ci.yml)
+[![Gem Version](https://badge.fury.io/rb/biosyntax.svg)](https://badge.fury.io/rb/biosyntax)
 [![Lines of Code](https://img.shields.io/endpoint?url=https%3A%2F%2Ftokei.kojix2.net%2Fbadge%2Fgithub%2Fkojix2%2Fruby-biosyntax%2Flines)](https://tokei.kojix2.net/github/kojix2/ruby-biosyntax)
+[![DOI](https://zenodo.org/badge/1269890086.svg)](https://doi.org/10.5281/zenodo.20698478)
+
 
 
 :dna: [bioSyntax](https://github.com/bioSyntax/bioSyntax) - Syntax highlighting for biological data formats - for Ruby.
@@ -98,20 +101,18 @@ BioSyntax.guess_format("a.vcf.gz") # :vcf
 The metadata is generated from `libbiosyntax` at load time. The Ruby side does
 not maintain a separate hand-written table of formats or kinds.
 
-## Examples
+## Command line
 
-This gem does not install a CLI. See `examples/` for small scripts:
+Installing the gem also installs `biocat`:
 
 ```sh
-ruby examples/bcat.rb sample.vcf
-ruby examples/bcat.rb -l fastq reads.fastq
-ruby examples/bcat.rb -l
-ruby examples/inspect_spans.rb sample.vcf
+biocat sample.vcf
+biocat --format fastq reads.fastq
+biocat -l
 ```
 
-`bcat.rb` guesses the format from the file name when possible. Use `-l` /
-`--language` to pass a format explicitly. Calling `-l` without an argument
-prints the supported format names.
+`.gz`/`.bgz` are decompressed automatically. BAM/CRAM/BCF require optional
+`ruby-htslib` (`gem install htslib`).
 
 ## Development tasks
 
@@ -146,9 +147,10 @@ bundle exec rake
 
 ## License
 
-`biosyntax` vendors `libbiosyntax`, which is licensed under the GNU General
-Public License version 3 only. This gem is therefore distributed under
-`GPL-3.0-only`. See `LICENSE.md`.
+The Ruby code and native binding are licensed under the [MIT License](LICENSE.md).
+The vendored `libbiosyntax` source is licensed under
+[LGPL-2.1-or-later](LICENSE.libbiosyntax.md), and the compiled extension is
+subject to its terms.
 
 This project is inspired by the original bioSyntax project:
 <https://github.com/bioSyntax/bioSyntax>

data/exe/biocat

@@ -0,0 +1,131 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+require 'optparse'
+require 'zlib'
+require 'biosyntax'
+
+options = { format: nil, list_formats: false }
+
+BINARY_FORMATS = {
+  '.bam' => :sam,
+  '.cram' => :sam,
+  '.bcf' => :vcf
+}.freeze
+
+def binary_format_for(path, requested_format)
+  unless path == '-'
+    ext = File.extname(path).downcase
+    return BINARY_FORMATS[ext] if BINARY_FORMATS.key?(ext)
+  end
+
+  case requested_format&.to_s&.downcase
+  when 'bam', 'cram' then :sam
+  when 'bcf' then :vcf
+  end
+end
+
+def require_htslib!
+  require 'htslib'
+rescue LoadError => e
+  warn "[biocat] ruby-htslib is required to read BAM/CRAM/BCF files: #{e.message}"
+  warn '[biocat] install it with `gem install htslib`'
+  exit 1
+end
+
+def emit_colored(text, highlighter)
+  text.each_line do |line|
+    print(highlighter ? highlighter.colorize(line) : line)
+  end
+end
+
+def emit_record(record, highlighter)
+  line = record.to_s
+  line = "#{line}\n" unless line.end_with?("\n")
+  emit_colored(line, highlighter)
+end
+
+def emit_htslib_file(path, format, highlighter)
+  require_htslib!
+
+  case format
+  when :sam
+    HTS::Bam.open(path) do |bam|
+      emit_colored(bam.header.to_s, highlighter)
+      bam.each { |record| emit_record(record, highlighter) }
+    end
+  when :vcf
+    HTS::Bcf.open(path) do |bcf|
+      emit_colored(bcf.header.to_s, highlighter)
+      bcf.each { |record| emit_record(record, highlighter) }
+    end
+  end
+end
+
+parser = OptionParser.new do |opts|
+  opts.banner = 'usage: biocat [options] [FILE ...]'
+
+  opts.on('-f', '--format FORMAT', 'Highlight as FORMAT') do |format|
+    options[:format] = format
+  end
+
+  opts.on('-l', '--list-formats', 'Print supported format names') do
+    options[:list_formats] = true
+  end
+
+  opts.on('-h', '--help', 'Print this help') do
+    puts opts
+    exit
+  end
+end
+
+begin
+  parser.parse!
+
+  if options[:list_formats]
+    puts BioSyntax::FORMAT_NAMES.join("\n")
+    exit
+  end
+
+  paths = ARGV.empty? ? ['-'] : ARGV
+
+  paths.each do |path|
+    binary_format = binary_format_for(path, options[:format])
+    format = binary_format || options[:format]
+    format ||= BioSyntax.guess_format(path) unless path == '-'
+
+    highlighter = format ? BioSyntax[format] : nil
+    if binary_format
+      emit_htslib_file(path, binary_format, highlighter)
+      next
+    end
+
+    input =
+      if path == '-'
+        $stdin
+      elsif path.downcase.end_with?('.gz', '.bgz')
+        Zlib::GzipReader.open(path)
+      else
+        File.open(path, 'rb')
+      end
+
+    begin
+      emit_colored(input, highlighter)
+    ensure
+      input.close unless path == '-'
+    end
+  end
+rescue OptionParser::ParseError, BioSyntax::Error => e
+  warn "[biocat] #{e.message}"
+  warn parser
+  exit 2
+rescue SystemCallError => e
+  warn "[biocat] #{e.message}"
+  exit 1
+rescue Zlib::GzipFile::Error => e
+  warn "[biocat] #{e.message}"
+  exit 1
+rescue StandardError => e
+  warn "[biocat] #{e.message}"
+  exit 1
+end

data/ext/biosyntax/biosyntax.c

@@ -1,7 +1,7 @@
 
 /*
  * libbiosyntax: dependency-free C tokenizer/highlighter core for biological files.
- * SPDX-License-Identifier: GPL-3.0-only
+ * SPDX-License-Identifier: LGPL-2.1-or-later
  */
 #define BIOSYN_BUILDING_LIBRARY
 #include "biosyntax.h"
@@ -431,25 +431,25 @@ static uint32_t mapq_score(int64_t v) { if (v == 255 || v < 10) return BIOSYN_CL
 
 static void hi_fasta_scheme(writer_t *w, const char *s, size_t n, uint32_t scheme) { n = trim_eol(s, n); if (!n) return; if (s[0] == '>') emit(w, 0, n, BIOSYN_CLASS_HEADER); else if (s[0] == ';' || s[0] == '#') emit(w, 0, n, BIOSYN_CLASS_COMMENT); else seq_runs_scheme(w, s, 0, n, scheme); }
 static int is_start_codon(const char *s, size_t i, size_t n) {
-    unsigned char a, b, c;
+    int a, b, c;
     if (i + 3 > n) return 0;
     a = up((unsigned char)s[i]); b = up((unsigned char)s[i + 1]); c = up((unsigned char)s[i + 2]);
     return a == 'A' && b == 'T' && c == 'G';
 }
 static int is_stop_codon(const char *s, size_t i, size_t n) {
-    unsigned char a, b, c;
+    int a, b, c;
     if (i + 3 > n) return 0;
     a = up((unsigned char)s[i]); b = up((unsigned char)s[i + 1]); c = up((unsigned char)s[i + 2]);
     return a == 'T' && ((b == 'A' && (c == 'A' || c == 'G')) || (b == 'G' && c == 'A'));
 }
 static int is_rna_start_codon(const char *s, size_t i, size_t n) {
-    unsigned char a, b, c;
+    int a, b, c;
     if (i + 3 > n) return 0;
     a = up((unsigned char)s[i]); b = up((unsigned char)s[i + 1]); c = up((unsigned char)s[i + 2]);
     return a == 'A' && b == 'U' && c == 'G';
 }
 static int is_rna_stop_codon(const char *s, size_t i, size_t n) {
-    unsigned char a, b, c;
+    int a, b, c;
     if (i + 3 > n) return 0;
     a = up((unsigned char)s[i]); b = up((unsigned char)s[i + 1]); c = up((unsigned char)s[i + 2]);
     return a == 'U' && ((b == 'A' && (c == 'A' || c == 'G')) || (b == 'G' && c == 'A'));

data/ext/biosyntax/biosyntax.h

@@ -4,7 +4,7 @@
 /*
  * libbiosyntax: dependency-free C tokenizer/highlighter core for biological files.
  * The core performs no IO; callers pass one already-read text line at a time.
- * SPDX-License-Identifier: GPL-3.0-only
+ * SPDX-License-Identifier: LGPL-2.1-or-later
  */
 
 #include <stddef.h>
@@ -32,7 +32,7 @@ extern "C" {
 
 #define BIOSYN_VERSION_MAJOR 0
 #define BIOSYN_VERSION_MINOR 1
-#define BIOSYN_VERSION_PATCH 0
+#define BIOSYN_VERSION_PATCH 1
 #define BIOSYN_ABI_VERSION   1u
 
 #define BIOSYN_STRINGIFY_DETAIL(x) #x

data/ext/biosyntax/biosyntax_ext.c

@@ -1,6 +1,6 @@
 /*
  * Ruby native extension for libbiosyntax.
- * SPDX-License-Identifier: GPL-3.0-only
+ * SPDX-License-Identifier: MIT
  */
 
 #include "ruby.h"

data/lib/biosyntax/version.rb

@@ -1,5 +1,5 @@
 module BioSyntax
   # Ruby gem version.
   # @return [String]
-  VERSION = '0.1.0'
+  VERSION = '0.1.2'
 end

data/lib/biosyntax.rb

@@ -436,8 +436,8 @@ module BioSyntax
               else
                 name = value.to_s.downcase
                 FORMATS[name.to_sym] ||
-                  FORMATS[name.tr('_', '-').to_sym] ||
-                  FORMATS_BY_ID[Native.format_id_from_name(name)]
+                FORMATS[name.tr('_', '-').to_sym] ||
+                FORMATS_BY_ID[Native.format_id_from_name(name)]
               end
 
       return found if found

metadata

@@ -1,31 +1,34 @@
 --- !ruby/object:Gem::Specification
 name: biosyntax
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.1.2
 platform: ruby
 authors:
 - kojix2
-bindir: bin
+bindir: exe
 cert_chain: []
 date: 1980-01-02 00:00:00.000000000 Z
 dependencies: []
 email: 2xijok@gmail.com
-executables: []
+executables:
+- biocat
 extensions:
 - ext/biosyntax/extconf.rb
 extra_rdoc_files: []
 files:
+- LICENSE.libbiosyntax.md
 - LICENSE.md
 - README.md
+- exe/biocat
 - ext/biosyntax/biosyntax.c
 - ext/biosyntax/biosyntax.h
 - ext/biosyntax/biosyntax_ext.c
 - ext/biosyntax/extconf.rb
 - lib/biosyntax.rb
 - lib/biosyntax/version.rb
-homepage: https://github.com/kojix2/biosyntax
+homepage: https://github.com/kojix2/ruby-biosyntax
 licenses:
-- GPL-3.0-only
+- MIT
 metadata: {}
 rdoc_options: []
 require_paths:
@@ -41,7 +44,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 4.0.10
+rubygems_version: 4.0.16
 specification_version: 4
 summary: Ruby native binding for libbiosyntax
 test_files: []