1 raise "Ruby 1.9 or later required" if RUBY_VERSION < "1.9"
3 # Copyright (C) 2007-2010 Martin A. Hansen.
5 # This program is free software; you can redistribute it and/or
6 # modify it under the terms of the GNU General Public License
7 # as published by the Free Software Foundation; either version 2
8 # of the License, or (at your option) any later version.
10 # This program is distributed in the hope that it will be useful,
11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 # GNU General Public License for more details.
15 # You should have received a copy of the GNU General Public License
16 # along with this program; if not, write to the Free Software
17 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
19 # http://www.gnu.org/copyleft/gpl.html
21 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
23 # This software is part of the Biopieces framework (www.biopieces.org).
25 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
33 # Biopieces are command line scripts and uses OptionParser to parse command line
34 # options according to a list of casts. Each cast prescribes the long and short
35 # name of the option, the type, if it is mandatory, the default value, and allowed
36 # and disallowed values. An optional list of extra casts can be supplied, and the
37 # integrity of the casts are checked. Following the command line parsing, the
38 # options are checked according to the casts. Methods are also included for handling
39 # the parsing and emitting of Biopiece records, which are ASCII text records consisting
40 # of lines with a key/value pair separated by a colon and a white space ': '.
41 # Each record is separated by a line with three dashes '---'.
45 attr_accessor :out # accessor for out stream _ios_
47 # Initialize a Biopiece and write the status to file.
48 # Options are for testing purposes only.
49 def initialize(test=nil, input=STDIN, output=STDOUT)
55 # Check the integrity of a list of casts, followed by parsion options from argv
56 # and finally checking the options according to the casts. Returns nil if argv
57 # is empty, otherwise an options hash.
58 def parse(argv, cast_list=[], script_path=$0)
59 casts = Casts.new(cast_list)
60 option_handler = OptionHandler.new(argv, casts, script_path, @test)
61 @options = option_handler.options_parse
64 # Open Biopiece input stream if not open and iterate over all Biopiece
65 # records in the stream.
67 @in = Stream::open(@options, mode="r", @input) unless @in.is_a? IO
72 @in.each_line do |line|
74 when /^([^:]+): (.*)$/
75 record[$1.to_sym] = $2
77 yield record unless record.empty?
80 raise "Bad record format: #{line}"
84 yield record unless record.empty?
89 alias :each :each_record
91 # Open Biopiece output stream if not open and puts record to the stream.
93 @out = Stream::open(@options, mode="w", @output) unless @out.is_a? IO
95 record.each do |key,value|
96 @out.print "#{key.to_s}: #{value}\n"
105 # Create a temporary directory inside the ENV["BP_TMP"] dir.
110 path = ENV["BP_TMP"] + "/" + [user, time + pid, pid, "bp_tmp"].join("_")
112 Status.new.set_tmpdir(path)
118 # Error class for all exceptions to do with option casts.
119 class CastError < StandardError; end
122 # Class to handle casts of command line options. Each cast prescribes the long and
123 # short name of the option, the type, if it is mandatory, the default value, and
124 # allowed and disallowed values. An optional list of extra casts can be supplied,
125 # and the integrity of the casts are checked.
127 TYPES = %w[flag string list int uint float file file! files files! dir dir! genome]
128 MANDATORY = %w[long short type mandatory default allowed disallowed]
130 # Initialize cast object with an optional options cast list to which
131 # ubiquitous casts are added after which all casts are checked.
132 def initialize(cast_list=[])
133 @cast_list = cast_list
137 self.push(*@cast_list)
142 # Add ubiquitous options casts.
144 @cast_list << {:long=>'help', :short=>'?', :type=>'flag', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
145 @cast_list << {:long=>'stream_in', :short=>'I', :type=>'files!', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
146 @cast_list << {:long=>'stream_out', :short=>'O', :type=>'file', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
147 @cast_list << {:long=>'verbose', :short=>'v', :type=>'flag', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
150 # Check integrity of the casts.
157 # Check if all mandatory keys are present in casts and raise if not.
159 @cast_list.each do |cast|
160 MANDATORY.each do |mandatory|
161 raise CastError, "Missing symbol in cast: '#{mandatory.to_sym}'" unless cast.has_key? mandatory.to_sym
166 # Check if all values in casts are valid.
168 @cast_list.each do |cast|
170 check_val_short(cast)
172 check_val_mandatory(cast)
173 check_val_default(cast)
174 check_val_allowed(cast)
175 check_val_disallowed(cast)
179 # Check if the values to long are legal and raise if not.
180 def check_val_long(cast)
181 unless cast[:long].is_a? String and cast[:long].length > 1
182 raise CastError, "Illegal cast of long: '#{cast[:long]}'"
186 # Check if the values to short are legal and raise if not.
187 def check_val_short(cast)
188 unless cast[:short].is_a? String and cast[:short].length == 1
189 raise CastError, "Illegal cast of short: '#{cast[:short]}'"
193 # Check if values to type are legal and raise if not.
194 def check_val_type(cast)
197 type_hash[type] = true
200 unless type_hash.has_key? cast[:type]
201 raise CastError, "Illegal cast of type: '#{cast[:type]}'"
205 # Check if values to mandatory are legal and raise if not.
206 def check_val_mandatory(cast)
207 unless cast[:mandatory] == true or cast[:mandatory] == false
208 raise CastError, "Illegal cast of mandatory: '#{cast[:mandatory]}'"
212 # Check if values to default are legal and raise if not.
213 def check_val_default(cast)
214 unless cast[:default].nil? or
215 cast[:default].is_a? String or
216 cast[:default].is_a? Integer or
217 cast[:default].is_a? Float
218 raise CastError, "Illegal cast of default: '#{cast[:default]}'"
222 # Check if values to allowed are legal and raise if not.
223 def check_val_allowed(cast)
224 unless cast[:allowed].is_a? String or cast[:allowed].nil?
225 raise CastError, "Illegal cast of allowed: '#{cast[:allowed]}'"
229 # Check if values to disallowed are legal and raise if not.
230 def check_val_disallowed(cast)
231 unless cast[:disallowed].is_a? String or cast[:disallowed].nil?
232 raise CastError, "Illegal cast of disallowed: '#{cast[:disallowed]}'"
236 # Check cast for duplicate long or short options names.
239 @cast_list.each do |cast|
240 raise CastError, "Duplicate argument: '--#{cast[:long]}'" if check_hash.has_key? cast[:long]
241 raise CastError, "Duplicate argument: '-#{cast[:short]}'" if check_hash.has_key? cast[:short]
242 check_hash[cast[:long]] = true
243 check_hash[cast[:short]] = true
247 # Convert values to :long keys to symbols for all casts.
249 @cast_list.each do |cast|
250 cast[:long] = cast[:long].to_sym
257 # Class for parsing argv using OptionParser according to given casts.
258 # Default options are set, file glob expressions expanded, and options are
259 # checked according to the casts. Usage information is printed and exit called
262 REGEX_LIST = /^(list|files|files!)$/
263 REGEX_INT = /^(int|uint)$/
264 REGEX_STRING = /^(file|file!|dir|dir!|genome)$/
266 def initialize(argv, casts, script_path, test=nil)
269 @script_path = script_path
273 # Parse options from argv using OptionParser and casts denoting long and
274 # short option names. Usage information is printed and exit called.
275 # A hash with options is returned.
279 option_parser = OptionParser.new do |option|
280 @casts.each do |cast|
283 option.on("-#{cast[:short]}", "--#{cast[:long]}") do |o|
284 @options[cast[:long]] = o
287 option.on("-#{cast[:short]}", "--#{cast[:long]} F", Float) do |f|
288 @options[cast[:long]] = f
291 option.on("-#{cast[:short]}", "--#{cast[:long]} S", String) do |s|
292 @options[cast[:long]] = s.to_sym # TODO: this to_sym - is that needed?
295 option.on( "-#{cast[:short]}", "--#{cast[:long]} A", Array) do |a|
296 @options[cast[:long]] = a
299 option.on("-#{cast[:short]}", "--#{cast[:long]} I", Integer) do |i|
300 @options[cast[:long]] = i
303 option.on("-#{cast[:short]}", "--#{cast[:long]} S", String) do |s|
304 @options[cast[:long]] = s
307 raise ArgumentError, "Unknown option type: '#{cast[:type]}'"
312 option_parser.parse!(@argv)
315 print_usage_and_exit(true)
316 elsif print_usage_short?
327 # Given the script name determine the path of the wiki file with the usage info.
329 path = ENV["BP_DIR"] + "/bp_usage/" + File.basename(@script_path) + ".wiki"
330 raise "No such wiki file: #{path}" unless File.file? path
334 # Check if full "usage info" should be printed.
335 def print_usage_full?
339 # Check if short "usage info" should be printed.
340 def print_usage_short?
343 elsif @options[:stream_in]
345 elsif @options[:data_in]
347 elsif wiki_path =~ /^(list_biopieces|list_genomes|list_mysql_databases|biostat)$/ # TODO get rid of this!
354 # Print usage info by Calling an external script 'print_wiki'
355 # using a system() call and exit. An optional 'full' flag
356 # outputs the full usage info.
357 def print_usage_and_exit(full=nil)
362 system("print_wiki --data_in #{wiki_path} --help")
364 system("print_wiki --data_in #{wiki_path}")
367 raise "Failed printing wiki: #{wiki_path}" unless $?.success?
373 # Set default options value from cast unless a value is set.
375 @casts.each do |cast|
377 unless @options.has_key? cast[:long]
378 if cast[:type] == 'list'
379 @options[cast[:long]] = cast[:default].split ','
381 @options[cast[:long]] = cast[:default]
388 # Expands glob expressions to a full list of paths.
389 # Examples: "*.fna" or "foo.fna,*.fna" or "foo.fna,/bar/*.fna"
391 @casts.each do |cast|
392 if cast[:type] == 'files' or cast[:type] == 'files!'
393 if @options.has_key? cast[:long]
396 @options[cast[:long]].each do |path|
398 Dir.glob(path).each do |file|
399 files << file if File.file? file
406 @options[cast[:long]] = files
412 # Check all options according to casts.
414 @casts.each do |cast|
415 options_check_mandatory(cast)
416 options_check_int(cast)
417 options_check_uint(cast)
418 options_check_file(cast)
419 options_check_files(cast)
420 options_check_dir(cast)
421 options_check_allowed(cast)
422 options_check_disallowed(cast)
426 # Check if a mandatory option is set and raise if it isn't.
427 def options_check_mandatory(cast)
429 raise ArgumentError, "Mandatory argument: --#{cast[:long]}" unless @options.has_key? cast[:long]
433 # Check int type option and raise if not an integer.
434 def options_check_int(cast)
435 if cast[:type] == 'int' and @options.has_key? cast[:long]
436 unless @options[cast[:long]].is_a? Integer
437 raise ArgumentError, "Argument to --#{cast[:long]} must be an integer, not '#{@options[cast[:long]]}'"
442 # Check uint type option and raise if not an unsinged integer.
443 def options_check_uint(cast)
444 if cast[:type] == 'uint' and @options.has_key? cast[:long]
445 unless @options[cast[:long]].is_a? Integer and @options[cast[:long]] >= 0
446 raise ArgumentError, "Argument to --#{cast[:long]} must be an unsigned integer, not '#{@options[cast[:long]]}'"
451 # Check file! type argument and raise if file don't exists.
452 def options_check_file(cast)
453 if cast[:type] == 'file!' and @options.has_key? cast[:long]
454 raise ArgumentError, "No such file: '#{@options[cast[:long]]}'" unless File.file? @options[cast[:long]]
458 # Check files! type argument and raise if files don't exists.
459 def options_check_files(cast)
460 if cast[:type] == 'files!' and @options.has_key? cast[:long]
461 @options[cast[:long]].each do |path|
463 raise ArgumentError, "File not readable: '#{path}'" unless File.readable? path
468 # Check dir! type argument and raise if directory don't exist.
469 def options_check_dir(cast)
470 if cast[:type] == 'dir!' and @options.has_key? cast[:long]
471 raise ArgumentError, "No such directory: '#{@options[cast[:long]]}'" unless File.directory? @options[cast[:long]]
475 # Check options and raise unless allowed.
476 def options_check_allowed(cast)
477 if cast[:allowed] and @options.has_key? cast[:long]
479 cast[:allowed].split(',').each { |a| allowed_hash[a.to_s] = 1 }
481 raise ArgumentError, "Argument '#{@options[cast[:long]]}' to --#{cast[:long]} not allowed" unless allowed_hash.has_key? @options[cast[:long]].to_s
485 # Check disallowed argument values and raise if disallowed.
486 def options_check_disallowed(cast)
487 if cast[:disallowed] and @options.has_key? cast[:long]
488 cast[:disallowed].split(',').each do |val|
489 raise ArgumentError, "Argument '#{@options[cast[:long]]}' to --#{cast[:long]} is disallowed" if val.to_s == @options[cast[:long]].to_s
495 # Class for manipulating the execution status of Biopieces by setting a
496 # status file with a time stamp, process id, and command arguments. The
497 # status file is used for creating log entries and for displaying the
498 # runtime status of Biopieces.
500 # Write the status to a status file.
502 time0 = Time.new.strftime("%Y-%m-%d %X")
504 File.open(path, mode="w") do |fh|
505 fh.puts [time0, ARGV.join(" ")].join(";")
509 # Append the a temporary directory path to the status file.
510 def set_tmpdir(tmpdir_path)
513 File.open(path, mode="r") do |fh|
514 status = fh.read.chomp
517 status = "#{status};#{tmpdir_path}\n"
519 File.open(path, mode="w") do |fh|
524 # Extract the temporary directory path from the status file,
525 # and return this or nil if not found.
527 File.open(path, mode="r") do |fh|
528 tmpdir_path = fh.read.chomp.split(";").last
529 return tmpdir_path if File.directory?(tmpdir_path)
535 # Write the Biopiece status to the log file.
537 time1 = Time.new.strftime("%Y-%m-%d %X")
539 script = File.basename($0)
541 stream = File.open(path)
542 time0, args, tmp_dir = stream.first.split(";")
545 elap = time_diff(time0, time1)
546 command = [script, args].join(" ")
547 log_file = ENV["BP_LOG"] + "/biopieces.log"
549 File.open(log_file, mode = "a") { |file| file.puts [time0, time1, elap, user, exit_status, command].join("\t") }
552 # Delete status file.
559 # Path to status file
562 script = File.basename($0)
564 path = ENV["BP_TMP"] + "/" + [user, script, pid, "status"].join(".")
567 # Get the elapsed time from the difference between two time stamps.
568 def time_diff(t0, t1)
569 Time.at((DateTime.parse(t1).to_time - DateTime.parse(t0).to_time).to_i).gmtime.strftime('%X')
575 # Open Biopieces output data stream for reading from stdin or a file
576 # specified in options[:stream_in] OR writing to stdout or a file
577 # specified in options[:stream_out] or options[:data_out].
578 def self.open(options, mode, stdio)
580 if options[:data_in] and options[:data_in].first == "-"
583 $stdin.tty? ? read(options[:stream_in]) : stdio
586 options[:stream_out] ? self.write(options[:stream_out], options[:compress]) : stdio
588 raise "Bad mode #{mode}"
594 # Opens a reads stream to a list of files.
596 return if files.nil? #TODO case/when
597 self.zipped?(files) ? self.zread(files) : self.nread(files)
600 # Opens a write stream to a file and returns a _io_ object.
601 def self.write(file, zip=nil)
602 zip ? self.zwrite(file) : self.nwrite(file)
605 # Opens a list of gzipped files for reading and return an _io_ object.
606 def self.zread(files)
607 stdin, stdout, stderr = Open3.popen3("zcat " + files.join(' '));
613 # Opens a file for gzipped writing and return an _io_ object.
614 def self.zwrite(file)
615 stdin, stdout, stderr = Open3.popen3("gzip -f > #{file}")
621 # Opens a list of files for reading and return an _io_ object.
622 def self.nread(files)
623 stdin, stdout, stderr = Open3.popen3("cat " + files.join(' '));
629 # Opens a file for writing and return an _io_ object.
630 def self.nwrite(file)
631 File.open(file, mode="w")
634 # Test if a list of files are gzipped or not.
635 # Raises if files are mixed zipped and unzipped.
636 def self.zipped?(files)
640 type = `file #{file}`
642 if type =~ /gzip compressed/
643 type_hash[:gzip] = true
645 type_hash[:ascii] = true
649 raise "Mixture of zipped and unzipped files" if type_hash.size == 2
659 exit_status = $! ? $!.inspect : "OK"
663 exit_status = "ERROR"
665 exit_status = "INTERRUPTED"
667 exit_status = "TERMINATED"
673 tmpdir = status.get_tmpdir
674 FileUtils.remove_entry_secure(tmpdir) unless tmpdir.nil?
675 status.log(exit_status)