Current File : //proc/thread-self/root/opt/alt/ruby33/share/ruby/open3.rb
# frozen_string_literal: true
#
# = open3.rb: Popen, but with stderr, too
#
# Author:: Yukihiro Matsumoto
# Documentation:: Konrad Meyer
#
# Open3 gives you access to stdin, stdout, and stderr when running other
# programs.
#
#
# Open3 grants you access to stdin, stdout, stderr and a thread to wait for the
# child process when running another program.
# You can specify various attributes, redirections, current directory, etc., of
# the program in the same way as for Process.spawn.
#
# - Open3.popen3 : pipes for stdin, stdout, stderr
# - Open3.popen2 : pipes for stdin, stdout
# - Open3.popen2e : pipes for stdin, merged stdout and stderr
# - Open3.capture3 : give a string for stdin; get strings for stdout, stderr
# - Open3.capture2 : give a string for stdin; get a string for stdout
# - Open3.capture2e : give a string for stdin; get a string for merged stdout and stderr
# - Open3.pipeline_rw : pipes for first stdin and last stdout of a pipeline
# - Open3.pipeline_r : pipe for last stdout of a pipeline
# - Open3.pipeline_w : pipe for first stdin of a pipeline
# - Open3.pipeline_start : run a pipeline without waiting
# - Open3.pipeline : run a pipeline and wait for its completion
#
require 'open3/version'
# \Module \Open3 supports creating child processes
# with access to their $stdin, $stdout, and $stderr streams.
#
# == What's Here
#
# Each of these methods executes a given command in a new process or subshell,
# or multiple commands in new processes and/or subshells:
#
# - Each of these methods executes a single command in a process or subshell,
# accepts a string for input to $stdin,
# and returns string output from $stdout, $stderr, or both:
#
# - Open3.capture2: Executes the command;
# returns the string from $stdout.
# - Open3.capture2e: Executes the command;
# returns the string from merged $stdout and $stderr.
# - Open3.capture3: Executes the command;
# returns strings from $stdout and $stderr.
#
# - Each of these methods executes a single command in a process or subshell,
# and returns pipes for $stdin, $stdout, and/or $stderr:
#
# - Open3.popen2: Executes the command;
# returns pipes for $stdin and $stdout.
# - Open3.popen2e: Executes the command;
# returns pipes for $stdin and merged $stdout and $stderr.
# - Open3.popen3: Executes the command;
# returns pipes for $stdin, $stdout, and $stderr.
#
# - Each of these methods executes one or more commands in processes and/or subshells,
# returns pipes for the first $stdin, the last $stdout, or both:
#
# - Open3.pipeline_r: Returns a pipe for the last $stdout.
# - Open3.pipeline_rw: Returns pipes for the first $stdin and the last $stdout.
# - Open3.pipeline_w: Returns a pipe for the first $stdin.
# - Open3.pipeline_start: Does not wait for processes to complete.
# - Open3.pipeline: Waits for processes to complete.
#
# Each of the methods above accepts:
#
# - An optional hash of environment variable names and values;
# see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
# - A required string argument that is a +command_line+ or +exe_path+;
# see {Argument command_line or exe_path}[rdoc-ref:Process@Argument+command_line+or+exe_path].
# - An optional hash of execution options;
# see {Execution Options}[rdoc-ref:Process@Execution+Options].
#
module Open3
# :call-seq:
# Open3.popen3([env, ] command_line, options = {}) -> [stdin, stdout, stderr, wait_thread]
# Open3.popen3([env, ] exe_path, *args, options = {}) -> [stdin, stdout, stderr, wait_thread]
# Open3.popen3([env, ] command_line, options = {}) {|stdin, stdout, stderr, wait_thread| ... } -> object
# Open3.popen3([env, ] exe_path, *args, options = {}) {|stdin, stdout, stderr, wait_thread| ... } -> object
#
# Basically a wrapper for
# {Process.spawn}[rdoc-ref:Process.spawn]
# that:
#
# - Creates a child process, by calling Process.spawn with the given arguments.
# - Creates streams +stdin+, +stdout+, and +stderr+,
# which are the standard input, standard output, and standard error streams
# in the child process.
# - Creates thread +wait_thread+ that waits for the child process to exit;
# the thread has method +pid+, which returns the process ID
# of the child process.
#
# With no block given, returns the array
# <tt>[stdin, stdout, stderr, wait_thread]</tt>.
# The caller should close each of the three returned streams.
#
# stdin, stdout, stderr, wait_thread = Open3.popen3('echo')
# # => [#<IO:fd 8>, #<IO:fd 10>, #<IO:fd 12>, #<Process::Waiter:0x00007f58d5428f58 run>]
# stdin.close
# stdout.close
# stderr.close
# wait_thread.pid # => 2210481
# wait_thread.value # => #<Process::Status: pid 2210481 exit 0>
#
# With a block given, calls the block with the four variables
# (three streams and the wait thread)
# and returns the block's return value.
# The caller need not close the streams:
#
# Open3.popen3('echo') do |stdin, stdout, stderr, wait_thread|
# p stdin
# p stdout
# p stderr
# p wait_thread
# p wait_thread.pid
# p wait_thread.value
# end
#
# Output:
#
# #<IO:fd 6>
# #<IO:fd 7>
# #<IO:fd 9>
# #<Process::Waiter:0x00007f58d53606e8 sleep>
# 2211047
# #<Process::Status: pid 2211047 exit 0>
#
# Like Process.spawn, this method has potential security vulnerabilities
# if called with untrusted input;
# see {Command Injection}[rdoc-ref:command_injection.rdoc@Command+Injection].
#
# Unlike Process.spawn, this method waits for the child process to exit
# before returning, so the caller need not do so.
#
# If the first argument is a hash, it becomes leading argument +env+
# in the call to Process.spawn;
# see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
#
# If the last argument is a hash, it becomes trailing argument +options+
# in the call to Process.spawn;
# see {Execution Options}[rdoc-ref:Process@Execution+Options].
#
# The single required argument is one of the following:
#
# - +command_line+ if it is a string,
# and if it begins with a shell reserved word or special built-in,
# or if it contains one or more metacharacters.
# - +exe_path+ otherwise.
#
# <b>Argument +command_line+</b>
#
# \String argument +command_line+ is a command line to be passed to a shell;
# it must begin with a shell reserved word, begin with a special built-in,
# or contain meta characters:
#
# Open3.popen3('if true; then echo "Foo"; fi') {|*args| p args } # Shell reserved word.
# Open3.popen3('echo') {|*args| p args } # Built-in.
# Open3.popen3('date > date.tmp') {|*args| p args } # Contains meta character.
#
# Output (similar for each call above):
#
# [#<IO:(closed)>, #<IO:(closed)>, #<IO:(closed)>, #<Process::Waiter:0x00007f58d52f28c8 dead>]
#
# The command line may also contain arguments and options for the command:
#
# Open3.popen3('echo "Foo"') { |i, o, e, t| o.gets }
# "Foo\n"
#
# <b>Argument +exe_path+</b>
#
# Argument +exe_path+ is one of the following:
#
# - The string path to an executable to be called.
# - A 2-element array containing the path to an executable
# and the string to be used as the name of the executing process.
#
# Example:
#
# Open3.popen3('/usr/bin/date') { |i, o, e, t| o.gets }
# # => "Wed Sep 27 02:56:44 PM CDT 2023\n"
#
# Ruby invokes the executable directly, with no shell and no shell expansion:
#
# Open3.popen3('doesnt_exist') { |i, o, e, t| o.gets } # Raises Errno::ENOENT
#
# If one or more +args+ is given, each is an argument or option
# to be passed to the executable:
#
# Open3.popen3('echo', 'C #') { |i, o, e, t| o.gets }
# # => "C #\n"
# Open3.popen3('echo', 'hello', 'world') { |i, o, e, t| o.gets }
# # => "hello world\n"
#
# Take care to avoid deadlocks.
# Output streams +stdout+ and +stderr+ have fixed-size buffers,
# so reading extensively from one but not the other can cause a deadlock
# when the unread buffer fills.
# To avoid that, +stdout+ and +stderr+ should be read simultaneously
# (using threads or IO.select).
#
# Related:
#
# - Open3.popen2: Makes the standard input and standard output streams
# of the child process available as separate streams,
# with no access to the standard error stream.
# - Open3.popen2e: Makes the standard input and the merge
# of the standard output and standard error streams
# of the child process available as separate streams.
#
def popen3(*cmd, &block)
if Hash === cmd.last
opts = cmd.pop.dup
else
opts = {}
end
in_r, in_w = IO.pipe
opts[:in] = in_r
in_w.sync = true
out_r, out_w = IO.pipe
opts[:out] = out_w
err_r, err_w = IO.pipe
opts[:err] = err_w
popen_run(cmd, opts, [in_r, out_w, err_w], [in_w, out_r, err_r], &block)
end
module_function :popen3
# :call-seq:
# Open3.popen2([env, ] command_line, options = {}) -> [stdin, stdout, wait_thread]
# Open3.popen2([env, ] exe_path, *args, options = {}) -> [stdin, stdout, wait_thread]
# Open3.popen2([env, ] command_line, options = {}) {|stdin, stdout, wait_thread| ... } -> object
# Open3.popen2([env, ] exe_path, *args, options = {}) {|stdin, stdout, wait_thread| ... } -> object
#
# Basically a wrapper for
# {Process.spawn}[rdoc-ref:Process.spawn]
# that:
#
# - Creates a child process, by calling Process.spawn with the given arguments.
# - Creates streams +stdin+ and +stdout+,
# which are the standard input and standard output streams
# in the child process.
# - Creates thread +wait_thread+ that waits for the child process to exit;
# the thread has method +pid+, which returns the process ID
# of the child process.
#
# With no block given, returns the array
# <tt>[stdin, stdout, wait_thread]</tt>.
# The caller should close each of the two returned streams.
#
# stdin, stdout, wait_thread = Open3.popen2('echo')
# # => [#<IO:fd 6>, #<IO:fd 7>, #<Process::Waiter:0x00007f58d52dbe98 run>]
# stdin.close
# stdout.close
# wait_thread.pid # => 2263572
# wait_thread.value # => #<Process::Status: pid 2263572 exit 0>
#
# With a block given, calls the block with the three variables
# (two streams and the wait thread)
# and returns the block's return value.
# The caller need not close the streams:
#
# Open3.popen2('echo') do |stdin, stdout, wait_thread|
# p stdin
# p stdout
# p wait_thread
# p wait_thread.pid
# p wait_thread.value
# end
#
# Output:
#
# #<IO:fd 6>
# #<IO:fd 7>
# #<Process::Waiter:0x00007f58d59a34b0 sleep>
# 2263636
# #<Process::Status: pid 2263636 exit 0>
#
# Like Process.spawn, this method has potential security vulnerabilities
# if called with untrusted input;
# see {Command Injection}[rdoc-ref:command_injection.rdoc@Command+Injection].
#
# Unlike Process.spawn, this method waits for the child process to exit
# before returning, so the caller need not do so.
#
# If the first argument is a hash, it becomes leading argument +env+
# in the call to Process.spawn;
# see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
#
# If the last argument is a hash, it becomes trailing argument +options+
# in the call to Process.spawn;
# see {Execution Options}[rdoc-ref:Process@Execution+Options].
#
# The single required argument is one of the following:
#
# - +command_line+ if it is a string,
# and if it begins with a shell reserved word or special built-in,
# or if it contains one or more metacharacters.
# - +exe_path+ otherwise.
#
# <b>Argument +command_line+</b>
#
# \String argument +command_line+ is a command line to be passed to a shell;
# it must begin with a shell reserved word, begin with a special built-in,
# or contain meta characters:
#
# Open3.popen2('if true; then echo "Foo"; fi') {|*args| p args } # Shell reserved word.
# Open3.popen2('echo') {|*args| p args } # Built-in.
# Open3.popen2('date > date.tmp') {|*args| p args } # Contains meta character.
#
# Output (similar for each call above):
#
# # => [#<IO:(closed)>, #<IO:(closed)>, #<Process::Waiter:0x00007f7577dfe410 dead>]
#
# The command line may also contain arguments and options for the command:
#
# Open3.popen2('echo "Foo"') { |i, o, t| o.gets }
# "Foo\n"
#
# <b>Argument +exe_path+</b>
#
# Argument +exe_path+ is one of the following:
#
# - The string path to an executable to be called.
# - A 2-element array containing the path to an executable
# and the string to be used as the name of the executing process.
#
# Example:
#
# Open3.popen2('/usr/bin/date') { |i, o, t| o.gets }
# # => "Thu Sep 28 09:41:06 AM CDT 2023\n"
#
# Ruby invokes the executable directly, with no shell and no shell expansion:
#
# Open3.popen2('doesnt_exist') { |i, o, t| o.gets } # Raises Errno::ENOENT
#
# If one or more +args+ is given, each is an argument or option
# to be passed to the executable:
#
# Open3.popen2('echo', 'C #') { |i, o, t| o.gets }
# # => "C #\n"
# Open3.popen2('echo', 'hello', 'world') { |i, o, t| o.gets }
# # => "hello world\n"
#
#
# Related:
#
# - Open3.popen2e: Makes the standard input and the merge
# of the standard output and standard error streams
# of the child process available as separate streams.
# - Open3.popen3: Makes the standard input, standard output,
# and standard error streams
# of the child process available as separate streams.
#
def popen2(*cmd, &block)
if Hash === cmd.last
opts = cmd.pop.dup
else
opts = {}
end
in_r, in_w = IO.pipe
opts[:in] = in_r
in_w.sync = true
out_r, out_w = IO.pipe
opts[:out] = out_w
popen_run(cmd, opts, [in_r, out_w], [in_w, out_r], &block)
end
module_function :popen2
# :call-seq:
# Open3.popen2e([env, ] command_line, options = {}) -> [stdin, stdout_and_stderr, wait_thread]
# Open3.popen2e([env, ] exe_path, *args, options = {}) -> [stdin, stdout_and_stderr, wait_thread]
# Open3.popen2e([env, ] command_line, options = {}) {|stdin, stdout_and_stderr, wait_thread| ... } -> object
# Open3.popen2e([env, ] exe_path, *args, options = {}) {|stdin, stdout_and_stderr, wait_thread| ... } -> object
#
# Basically a wrapper for
# {Process.spawn}[rdoc-ref:Process.spawn]
# that:
#
# - Creates a child process, by calling Process.spawn with the given arguments.
# - Creates streams +stdin+, +stdout_and_stderr+,
# which are the standard input and the merge of the standard output
# and standard error streams in the child process.
# - Creates thread +wait_thread+ that waits for the child process to exit;
# the thread has method +pid+, which returns the process ID
# of the child process.
#
# With no block given, returns the array
# <tt>[stdin, stdout_and_stderr, wait_thread]</tt>.
# The caller should close each of the two returned streams.
#
# stdin, stdout_and_stderr, wait_thread = Open3.popen2e('echo')
# # => [#<IO:fd 6>, #<IO:fd 7>, #<Process::Waiter:0x00007f7577da4398 run>]
# stdin.close
# stdout_and_stderr.close
# wait_thread.pid # => 2274600
# wait_thread.value # => #<Process::Status: pid 2274600 exit 0>
#
# With a block given, calls the block with the three variables
# (two streams and the wait thread)
# and returns the block's return value.
# The caller need not close the streams:
#
# Open3.popen2e('echo') do |stdin, stdout_and_stderr, wait_thread|
# p stdin
# p stdout_and_stderr
# p wait_thread
# p wait_thread.pid
# p wait_thread.value
# end
#
# Output:
#
# #<IO:fd 6>
# #<IO:fd 7>
# #<Process::Waiter:0x00007f75777578c8 sleep>
# 2274763
# #<Process::Status: pid 2274763 exit 0>
#
# Like Process.spawn, this method has potential security vulnerabilities
# if called with untrusted input;
# see {Command Injection}[rdoc-ref:command_injection.rdoc@Command+Injection].
#
# Unlike Process.spawn, this method waits for the child process to exit
# before returning, so the caller need not do so.
#
# If the first argument is a hash, it becomes leading argument +env+
# in the call to Process.spawn;
# see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
#
# If the last argument is a hash, it becomes trailing argument +options+
# in the call to Process.spawn;
# see {Execution Options}[rdoc-ref:Process@Execution+Options].
#
# The single required argument is one of the following:
#
# - +command_line+ if it is a string,
# and if it begins with a shell reserved word or special built-in,
# or if it contains one or more metacharacters.
# - +exe_path+ otherwise.
#
# <b>Argument +command_line+</b>
#
# \String argument +command_line+ is a command line to be passed to a shell;
# it must begin with a shell reserved word, begin with a special built-in,
# or contain meta characters:
#
# Open3.popen2e('if true; then echo "Foo"; fi') {|*args| p args } # Shell reserved word.
# Open3.popen2e('echo') {|*args| p args } # Built-in.
# Open3.popen2e('date > date.tmp') {|*args| p args } # Contains meta character.
#
# Output (similar for each call above):
#
# # => [#<IO:(closed)>, #<IO:(closed)>, #<Process::Waiter:0x00007f7577d8a1f0 dead>]
#
# The command line may also contain arguments and options for the command:
#
# Open3.popen2e('echo "Foo"') { |i, o_and_e, t| o_and_e.gets }
# "Foo\n"
#
# <b>Argument +exe_path+</b>
#
# Argument +exe_path+ is one of the following:
#
# - The string path to an executable to be called.
# - A 2-element array containing the path to an executable
# and the string to be used as the name of the executing process.
#
# Example:
#
# Open3.popen2e('/usr/bin/date') { |i, o_and_e, t| o_and_e.gets }
# # => "Thu Sep 28 01:58:45 PM CDT 2023\n"
#
# Ruby invokes the executable directly, with no shell and no shell expansion:
#
# Open3.popen2e('doesnt_exist') { |i, o_and_e, t| o_and_e.gets } # Raises Errno::ENOENT
#
# If one or more +args+ is given, each is an argument or option
# to be passed to the executable:
#
# Open3.popen2e('echo', 'C #') { |i, o_and_e, t| o_and_e.gets }
# # => "C #\n"
# Open3.popen2e('echo', 'hello', 'world') { |i, o_and_e, t| o_and_e.gets }
# # => "hello world\n"
#
# Related:
#
# - Open3.popen2: Makes the standard input and standard output streams
# of the child process available as separate streams,
# with no access to the standard error stream.
# - Open3.popen3: Makes the standard input, standard output,
# and standard error streams
# of the child process available as separate streams.
#
def popen2e(*cmd, &block)
if Hash === cmd.last
opts = cmd.pop.dup
else
opts = {}
end
in_r, in_w = IO.pipe
opts[:in] = in_r
in_w.sync = true
out_r, out_w = IO.pipe
opts[[:out, :err]] = out_w
popen_run(cmd, opts, [in_r, out_w], [in_w, out_r], &block)
ensure
if block
in_r.close
in_w.close
out_r.close
out_w.close
end
end
module_function :popen2e
def popen_run(cmd, opts, child_io, parent_io) # :nodoc:
pid = spawn(*cmd, opts)
wait_thr = Process.detach(pid)
child_io.each(&:close)
result = [*parent_io, wait_thr]
if defined? yield
begin
return yield(*result)
ensure
parent_io.each(&:close)
wait_thr.join
end
end
result
end
module_function :popen_run
class << self
private :popen_run
end
# :call-seq:
# Open3.capture3([env, ] command_line, options = {}) -> [stdout_s, stderr_s, status]
# Open3.capture3([env, ] exe_path, *args, options = {}) -> [stdout_s, stderr_s, status]
#
# Basically a wrapper for Open3.popen3 that:
#
# - Creates a child process, by calling Open3.popen3 with the given arguments
# (except for certain entries in hash +options+; see below).
# - Returns as strings +stdout_s+ and +stderr_s+ the standard output
# and standard error of the child process.
# - Returns as +status+ a <tt>Process::Status</tt> object
# that represents the exit status of the child process.
#
# Returns the array <tt>[stdout_s, stderr_s, status]</tt>:
#
# stdout_s, stderr_s, status = Open3.capture3('echo "Foo"')
# # => ["Foo\n", "", #<Process::Status: pid 2281954 exit 0>]
#
# Like Process.spawn, this method has potential security vulnerabilities
# if called with untrusted input;
# see {Command Injection}[rdoc-ref:command_injection.rdoc@Command+Injection].
#
# Unlike Process.spawn, this method waits for the child process to exit
# before returning, so the caller need not do so.
#
# If the first argument is a hash, it becomes leading argument +env+
# in the call to Open3.popen3;
# see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
#
# If the last argument is a hash, it becomes trailing argument +options+
# in the call to Open3.popen3;
# see {Execution Options}[rdoc-ref:Process@Execution+Options].
#
# The hash +options+ is given;
# two options have local effect in method Open3.capture3:
#
# - If entry <tt>options[:stdin_data]</tt> exists, the entry is removed
# and its string value is sent to the command's standard input:
#
# Open3.capture3('tee', stdin_data: 'Foo')
# # => ["Foo", "", #<Process::Status: pid 2319575 exit 0>]
#
# - If entry <tt>options[:binmode]</tt> exists, the entry is removed and
# the internal streams are set to binary mode.
#
# The single required argument is one of the following:
#
# - +command_line+ if it is a string,
# and if it begins with a shell reserved word or special built-in,
# or if it contains one or more metacharacters.
# - +exe_path+ otherwise.
#
# <b>Argument +command_line+</b>
#
# \String argument +command_line+ is a command line to be passed to a shell;
# it must begin with a shell reserved word, begin with a special built-in,
# or contain meta characters:
#
# Open3.capture3('if true; then echo "Foo"; fi') # Shell reserved word.
# # => ["Foo\n", "", #<Process::Status: pid 2282025 exit 0>]
# Open3.capture3('echo') # Built-in.
# # => ["\n", "", #<Process::Status: pid 2282092 exit 0>]
# Open3.capture3('date > date.tmp') # Contains meta character.
# # => ["", "", #<Process::Status: pid 2282110 exit 0>]
#
# The command line may also contain arguments and options for the command:
#
# Open3.capture3('echo "Foo"')
# # => ["Foo\n", "", #<Process::Status: pid 2282092 exit 0>]
#
# <b>Argument +exe_path+</b>
#
# Argument +exe_path+ is one of the following:
#
# - The string path to an executable to be called.
# - A 2-element array containing the path to an executable
# and the string to be used as the name of the executing process.
#
# Example:
#
# Open3.capture3('/usr/bin/date')
# # => ["Thu Sep 28 05:03:51 PM CDT 2023\n", "", #<Process::Status: pid 2282300 exit 0>]
#
# Ruby invokes the executable directly, with no shell and no shell expansion:
#
# Open3.capture3('doesnt_exist') # Raises Errno::ENOENT
#
# If one or more +args+ is given, each is an argument or option
# to be passed to the executable:
#
# Open3.capture3('echo', 'C #')
# # => ["C #\n", "", #<Process::Status: pid 2282368 exit 0>]
# Open3.capture3('echo', 'hello', 'world')
# # => ["hello world\n", "", #<Process::Status: pid 2282372 exit 0>]
#
def capture3(*cmd)
if Hash === cmd.last
opts = cmd.pop.dup
else
opts = {}
end
stdin_data = opts.delete(:stdin_data) || ''
binmode = opts.delete(:binmode)
popen3(*cmd, opts) {|i, o, e, t|
if binmode
i.binmode
o.binmode
e.binmode
end
out_reader = Thread.new { o.read }
err_reader = Thread.new { e.read }
begin
if stdin_data.respond_to? :readpartial
IO.copy_stream(stdin_data, i)
else
i.write stdin_data
end
rescue Errno::EPIPE
end
i.close
[out_reader.value, err_reader.value, t.value]
}
end
module_function :capture3
# :call-seq:
# Open3.capture2([env, ] command_line, options = {}) -> [stdout_s, status]
# Open3.capture2([env, ] exe_path, *args, options = {}) -> [stdout_s, status]
#
# Basically a wrapper for Open3.popen3 that:
#
# - Creates a child process, by calling Open3.popen3 with the given arguments
# (except for certain entries in hash +options+; see below).
# - Returns as string +stdout_s+ the standard output of the child process.
# - Returns as +status+ a <tt>Process::Status</tt> object
# that represents the exit status of the child process.
#
# Returns the array <tt>[stdout_s, status]</tt>:
#
# stdout_s, status = Open3.capture2('echo "Foo"')
# # => ["Foo\n", #<Process::Status: pid 2326047 exit 0>]
#
# Like Process.spawn, this method has potential security vulnerabilities
# if called with untrusted input;
# see {Command Injection}[rdoc-ref:command_injection.rdoc@Command+Injection].
#
# Unlike Process.spawn, this method waits for the child process to exit
# before returning, so the caller need not do so.
#
# If the first argument is a hash, it becomes leading argument +env+
# in the call to Open3.popen3;
# see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
#
# If the last argument is a hash, it becomes trailing argument +options+
# in the call to Open3.popen3;
# see {Execution Options}[rdoc-ref:Process@Execution+Options].
#
# The hash +options+ is given;
# two options have local effect in method Open3.capture2:
#
# - If entry <tt>options[:stdin_data]</tt> exists, the entry is removed
# and its string value is sent to the command's standard input:
#
# Open3.capture2('tee', stdin_data: 'Foo')
#
# # => ["Foo", #<Process::Status: pid 2326087 exit 0>]
#
# - If entry <tt>options[:binmode]</tt> exists, the entry is removed and
# the internal streams are set to binary mode.
#
# The single required argument is one of the following:
#
# - +command_line+ if it is a string,
# and if it begins with a shell reserved word or special built-in,
# or if it contains one or more metacharacters.
# - +exe_path+ otherwise.
#
# <b>Argument +command_line+</b>
#
# \String argument +command_line+ is a command line to be passed to a shell;
# it must begin with a shell reserved word, begin with a special built-in,
# or contain meta characters:
#
# Open3.capture2('if true; then echo "Foo"; fi') # Shell reserved word.
# # => ["Foo\n", #<Process::Status: pid 2326131 exit 0>]
# Open3.capture2('echo') # Built-in.
# # => ["\n", #<Process::Status: pid 2326139 exit 0>]
# Open3.capture2('date > date.tmp') # Contains meta character.
# # => ["", #<Process::Status: pid 2326174 exit 0>]
#
# The command line may also contain arguments and options for the command:
#
# Open3.capture2('echo "Foo"')
# # => ["Foo\n", #<Process::Status: pid 2326183 exit 0>]
#
# <b>Argument +exe_path+</b>
#
# Argument +exe_path+ is one of the following:
#
# - The string path to an executable to be called.
# - A 2-element array containing the path to an executable
# and the string to be used as the name of the executing process.
#
# Example:
#
# Open3.capture2('/usr/bin/date')
# # => ["Fri Sep 29 01:00:39 PM CDT 2023\n", #<Process::Status: pid 2326222 exit 0>]
#
# Ruby invokes the executable directly, with no shell and no shell expansion:
#
# Open3.capture2('doesnt_exist') # Raises Errno::ENOENT
#
# If one or more +args+ is given, each is an argument or option
# to be passed to the executable:
#
# Open3.capture2('echo', 'C #')
# # => ["C #\n", #<Process::Status: pid 2326267 exit 0>]
# Open3.capture2('echo', 'hello', 'world')
# # => ["hello world\n", #<Process::Status: pid 2326299 exit 0>]
#
def capture2(*cmd)
if Hash === cmd.last
opts = cmd.pop.dup
else
opts = {}
end
stdin_data = opts.delete(:stdin_data)
binmode = opts.delete(:binmode)
popen2(*cmd, opts) {|i, o, t|
if binmode
i.binmode
o.binmode
end
out_reader = Thread.new { o.read }
if stdin_data
begin
if stdin_data.respond_to? :readpartial
IO.copy_stream(stdin_data, i)
else
i.write stdin_data
end
rescue Errno::EPIPE
end
end
i.close
[out_reader.value, t.value]
}
end
module_function :capture2
# :call-seq:
# Open3.capture2e([env, ] command_line, options = {}) -> [stdout_and_stderr_s, status]
# Open3.capture2e([env, ] exe_path, *args, options = {}) -> [stdout_and_stderr_s, status]
#
# Basically a wrapper for Open3.popen3 that:
#
# - Creates a child process, by calling Open3.popen3 with the given arguments
# (except for certain entries in hash +options+; see below).
# - Returns as string +stdout_and_stderr_s+ the merged standard output
# and standard error of the child process.
# - Returns as +status+ a <tt>Process::Status</tt> object
# that represents the exit status of the child process.
#
# Returns the array <tt>[stdout_and_stderr_s, status]</tt>:
#
# stdout_and_stderr_s, status = Open3.capture2e('echo "Foo"')
# # => ["Foo\n", #<Process::Status: pid 2371692 exit 0>]
#
# Like Process.spawn, this method has potential security vulnerabilities
# if called with untrusted input;
# see {Command Injection}[rdoc-ref:command_injection.rdoc@Command+Injection].
#
# Unlike Process.spawn, this method waits for the child process to exit
# before returning, so the caller need not do so.
#
# If the first argument is a hash, it becomes leading argument +env+
# in the call to Open3.popen3;
# see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
#
# If the last argument is a hash, it becomes trailing argument +options+
# in the call to Open3.popen3;
# see {Execution Options}[rdoc-ref:Process@Execution+Options].
#
# The hash +options+ is given;
# two options have local effect in method Open3.capture2e:
#
# - If entry <tt>options[:stdin_data]</tt> exists, the entry is removed
# and its string value is sent to the command's standard input:
#
# Open3.capture2e('tee', stdin_data: 'Foo')
# # => ["Foo", #<Process::Status: pid 2371732 exit 0>]
#
# - If entry <tt>options[:binmode]</tt> exists, the entry is removed and
# the internal streams are set to binary mode.
#
# The single required argument is one of the following:
#
# - +command_line+ if it is a string,
# and if it begins with a shell reserved word or special built-in,
# or if it contains one or more metacharacters.
# - +exe_path+ otherwise.
#
# <b>Argument +command_line+</b>
#
# \String argument +command_line+ is a command line to be passed to a shell;
# it must begin with a shell reserved word, begin with a special built-in,
# or contain meta characters:
#
# Open3.capture2e('if true; then echo "Foo"; fi') # Shell reserved word.
# # => ["Foo\n", #<Process::Status: pid 2371740 exit 0>]
# Open3.capture2e('echo') # Built-in.
# # => ["\n", #<Process::Status: pid 2371774 exit 0>]
# Open3.capture2e('date > date.tmp') # Contains meta character.
# # => ["", #<Process::Status: pid 2371812 exit 0>]
#
# The command line may also contain arguments and options for the command:
#
# Open3.capture2e('echo "Foo"')
# # => ["Foo\n", #<Process::Status: pid 2326183 exit 0>]
#
# <b>Argument +exe_path+</b>
#
# Argument +exe_path+ is one of the following:
#
# - The string path to an executable to be called.
# - A 2-element array containing the path to an executable
# and the string to be used as the name of the executing process.
#
# Example:
#
# Open3.capture2e('/usr/bin/date')
# # => ["Sat Sep 30 09:01:46 AM CDT 2023\n", #<Process::Status: pid 2371820 exit 0>]
#
# Ruby invokes the executable directly, with no shell and no shell expansion:
#
# Open3.capture2e('doesnt_exist') # Raises Errno::ENOENT
#
# If one or more +args+ is given, each is an argument or option
# to be passed to the executable:
#
# Open3.capture2e('echo', 'C #')
# # => ["C #\n", #<Process::Status: pid 2371856 exit 0>]
# Open3.capture2e('echo', 'hello', 'world')
# # => ["hello world\n", #<Process::Status: pid 2371894 exit 0>]
#
def capture2e(*cmd)
if Hash === cmd.last
opts = cmd.pop.dup
else
opts = {}
end
stdin_data = opts.delete(:stdin_data)
binmode = opts.delete(:binmode)
popen2e(*cmd, opts) {|i, oe, t|
if binmode
i.binmode
oe.binmode
end
outerr_reader = Thread.new { oe.read }
if stdin_data
begin
if stdin_data.respond_to? :readpartial
IO.copy_stream(stdin_data, i)
else
i.write stdin_data
end
rescue Errno::EPIPE
end
end
i.close
[outerr_reader.value, t.value]
}
end
module_function :capture2e
# :call-seq:
# Open3.pipeline_rw([env, ] *cmds, options = {}) -> [first_stdin, last_stdout, wait_threads]
#
# Basically a wrapper for
# {Process.spawn}[rdoc-ref:Process.spawn]
# that:
#
# - Creates a child process for each of the given +cmds+
# by calling Process.spawn.
# - Pipes the +stdout+ from each child to the +stdin+ of the next child,
# or, for the first child, from the caller's +stdin+,
# or, for the last child, to the caller's +stdout+.
#
# The method does not wait for child processes to exit,
# so the caller must do so.
#
# With no block given, returns a 3-element array containing:
#
# - The +stdin+ stream of the first child process.
# - The +stdout+ stream of the last child process.
# - An array of the wait threads for all of the child processes.
#
# Example:
#
# first_stdin, last_stdout, wait_threads = Open3.pipeline_rw('sort', 'cat -n')
# # => [#<IO:fd 20>, #<IO:fd 21>, [#<Process::Waiter:0x000055e8de29ab40 sleep>, #<Process::Waiter:0x000055e8de29a690 sleep>]]
# first_stdin.puts("foo\nbar\nbaz")
# first_stdin.close # Send EOF to sort.
# puts last_stdout.read
# wait_threads.each do |wait_thread|
# wait_thread.join
# end
#
# Output:
#
# 1 bar
# 2 baz
# 3 foo
#
# With a block given, calls the block with the +stdin+ stream of the first child,
# the +stdout+ stream of the last child,
# and an array of the wait processes:
#
# Open3.pipeline_rw('sort', 'cat -n') do |first_stdin, last_stdout, wait_threads|
# first_stdin.puts "foo\nbar\nbaz"
# first_stdin.close # send EOF to sort.
# puts last_stdout.read
# wait_threads.each do |wait_thread|
# wait_thread.join
# end
# end
#
# Output:
#
# 1 bar
# 2 baz
# 3 foo
#
# Like Process.spawn, this method has potential security vulnerabilities
# if called with untrusted input;
# see {Command Injection}[rdoc-ref:command_injection.rdoc@Command+Injection].
#
# If the first argument is a hash, it becomes leading argument +env+
# in each call to Process.spawn;
# see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
#
# If the last argument is a hash, it becomes trailing argument +options+
# in each call to Process.spawn;
# see {Execution Options}[rdoc-ref:Process@Execution+Options].
#
# Each remaining argument in +cmds+ is one of:
#
# - A +command_line+: a string that begins with a shell reserved word
# or special built-in, or contains one or more metacharacters.
# - An +exe_path+: the string path to an executable to be called.
# - An array containing a +command_line+ or an +exe_path+,
# along with zero or more string arguments for the command.
#
# See {Argument command_line or exe_path}[rdoc-ref:Process@Argument+command_line+or+exe_path].
#
def pipeline_rw(*cmds, &block)
if Hash === cmds.last
opts = cmds.pop.dup
else
opts = {}
end
in_r, in_w = IO.pipe
opts[:in] = in_r
in_w.sync = true
out_r, out_w = IO.pipe
opts[:out] = out_w
pipeline_run(cmds, opts, [in_r, out_w], [in_w, out_r], &block)
end
module_function :pipeline_rw
# :call-seq:
# Open3.pipeline_r([env, ] *cmds, options = {}) -> [last_stdout, wait_threads]
#
# Basically a wrapper for
# {Process.spawn}[rdoc-ref:Process.spawn]
# that:
#
# - Creates a child process for each of the given +cmds+
# by calling Process.spawn.
# - Pipes the +stdout+ from each child to the +stdin+ of the next child,
# or, for the last child, to the caller's +stdout+.
#
# The method does not wait for child processes to exit,
# so the caller must do so.
#
# With no block given, returns a 2-element array containing:
#
# - The +stdout+ stream of the last child process.
# - An array of the wait threads for all of the child processes.
#
# Example:
#
# last_stdout, wait_threads = Open3.pipeline_r('ls', 'grep R')
# # => [#<IO:fd 5>, [#<Process::Waiter:0x000055e8de2f9898 dead>, #<Process::Waiter:0x000055e8de2f94b0 sleep>]]
# puts last_stdout.read
# wait_threads.each do |wait_thread|
# wait_thread.join
# end
#
# Output:
#
# Rakefile
# README.md
#
# With a block given, calls the block with the +stdout+ stream
# of the last child process,
# and an array of the wait processes:
#
# Open3.pipeline_r('ls', 'grep R') do |last_stdout, wait_threads|
# puts last_stdout.read
# wait_threads.each do |wait_thread|
# wait_thread.join
# end
# end
#
# Output:
#
# Rakefile
# README.md
#
# Like Process.spawn, this method has potential security vulnerabilities
# if called with untrusted input;
# see {Command Injection}[rdoc-ref:command_injection.rdoc@Command+Injection].
#
# If the first argument is a hash, it becomes leading argument +env+
# in each call to Process.spawn;
# see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
#
# If the last argument is a hash, it becomes trailing argument +options+
# in each call to Process.spawn;
# see {Execution Options}[rdoc-ref:Process@Execution+Options].
#
# Each remaining argument in +cmds+ is one of:
#
# - A +command_line+: a string that begins with a shell reserved word
# or special built-in, or contains one or more metacharacters.
# - An +exe_path+: the string path to an executable to be called.
# - An array containing a +command_line+ or an +exe_path+,
# along with zero or more string arguments for the command.
#
# See {Argument command_line or exe_path}[rdoc-ref:Process@Argument+command_line+or+exe_path].
#
def pipeline_r(*cmds, &block)
if Hash === cmds.last
opts = cmds.pop.dup
else
opts = {}
end
out_r, out_w = IO.pipe
opts[:out] = out_w
pipeline_run(cmds, opts, [out_w], [out_r], &block)
end
module_function :pipeline_r
# :call-seq:
# Open3.pipeline_w([env, ] *cmds, options = {}) -> [first_stdin, wait_threads]
#
# Basically a wrapper for
# {Process.spawn}[rdoc-ref:Process.spawn]
# that:
#
# - Creates a child process for each of the given +cmds+
# by calling Process.spawn.
# - Pipes the +stdout+ from each child to the +stdin+ of the next child,
# or, for the first child, pipes the caller's +stdout+ to the child's +stdin+.
#
# The method does not wait for child processes to exit,
# so the caller must do so.
#
# With no block given, returns a 2-element array containing:
#
# - The +stdin+ stream of the first child process.
# - An array of the wait threads for all of the child processes.
#
# Example:
#
# first_stdin, wait_threads = Open3.pipeline_w('sort', 'cat -n')
# # => [#<IO:fd 7>, [#<Process::Waiter:0x000055e8de928278 run>, #<Process::Waiter:0x000055e8de923e80 run>]]
# first_stdin.puts("foo\nbar\nbaz")
# first_stdin.close # Send EOF to sort.
# wait_threads.each do |wait_thread|
# wait_thread.join
# end
#
# Output:
#
# 1 bar
# 2 baz
# 3 foo
#
# With a block given, calls the block with the +stdin+ stream
# of the first child process,
# and an array of the wait processes:
#
# Open3.pipeline_w('sort', 'cat -n') do |first_stdin, wait_threads|
# first_stdin.puts("foo\nbar\nbaz")
# first_stdin.close # Send EOF to sort.
# wait_threads.each do |wait_thread|
# wait_thread.join
# end
# end
#
# Output:
#
# 1 bar
# 2 baz
# 3 foo
#
# Like Process.spawn, this method has potential security vulnerabilities
# if called with untrusted input;
# see {Command Injection}[rdoc-ref:command_injection.rdoc@Command+Injection].
#
# If the first argument is a hash, it becomes leading argument +env+
# in each call to Process.spawn;
# see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
#
# If the last argument is a hash, it becomes trailing argument +options+
# in each call to Process.spawn;
# see {Execution Options}[rdoc-ref:Process@Execution+Options].
#
# Each remaining argument in +cmds+ is one of:
#
# - A +command_line+: a string that begins with a shell reserved word
# or special built-in, or contains one or more metacharacters.
# - An +exe_path+: the string path to an executable to be called.
# - An array containing a +command_line+ or an +exe_path+,
# along with zero or more string arguments for the command.
#
# See {Argument command_line or exe_path}[rdoc-ref:Process@Argument+command_line+or+exe_path].
#
def pipeline_w(*cmds, &block)
if Hash === cmds.last
opts = cmds.pop.dup
else
opts = {}
end
in_r, in_w = IO.pipe
opts[:in] = in_r
in_w.sync = true
pipeline_run(cmds, opts, [in_r], [in_w], &block)
end
module_function :pipeline_w
# :call-seq:
# Open3.pipeline_start([env, ] *cmds, options = {}) -> [wait_threads]
#
# Basically a wrapper for
# {Process.spawn}[rdoc-ref:Process.spawn]
# that:
#
# - Creates a child process for each of the given +cmds+
# by calling Process.spawn.
# - Does not wait for child processes to exit.
#
# With no block given, returns an array of the wait threads
# for all of the child processes.
#
# Example:
#
# wait_threads = Open3.pipeline_start('ls', 'grep R')
# # => [#<Process::Waiter:0x000055e8de9d2bb0 run>, #<Process::Waiter:0x000055e8de9d2890 run>]
# wait_threads.each do |wait_thread|
# wait_thread.join
# end
#
# Output:
#
# Rakefile
# README.md
#
# With a block given, calls the block with an array of the wait processes:
#
# Open3.pipeline_start('ls', 'grep R') do |wait_threads|
# wait_threads.each do |wait_thread|
# wait_thread.join
# end
# end
#
# Output:
#
# Rakefile
# README.md
#
# Like Process.spawn, this method has potential security vulnerabilities
# if called with untrusted input;
# see {Command Injection}[rdoc-ref:command_injection.rdoc@Command+Injection].
#
# If the first argument is a hash, it becomes leading argument +env+
# in each call to Process.spawn;
# see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
#
# If the last argument is a hash, it becomes trailing argument +options+
# in each call to Process.spawn;
# see {Execution Options}[rdoc-ref:Process@Execution+Options].
#
# Each remaining argument in +cmds+ is one of:
#
# - A +command_line+: a string that begins with a shell reserved word
# or special built-in, or contains one or more metacharacters.
# - An +exe_path+: the string path to an executable to be called.
# - An array containing a +command_line+ or an +exe_path+,
# along with zero or more string arguments for the command.
#
# See {Argument command_line or exe_path}[rdoc-ref:Process@Argument+command_line+or+exe_path].
#
def pipeline_start(*cmds, &block)
if Hash === cmds.last
opts = cmds.pop.dup
else
opts = {}
end
if block
pipeline_run(cmds, opts, [], [], &block)
else
ts, = pipeline_run(cmds, opts, [], [])
ts
end
end
module_function :pipeline_start
# :call-seq:
# Open3.pipeline([env, ] *cmds, options = {}) -> array_of_statuses
#
# Basically a wrapper for
# {Process.spawn}[rdoc-ref:Process.spawn]
# that:
#
# - Creates a child process for each of the given +cmds+
# by calling Process.spawn.
# - Pipes the +stdout+ from each child to the +stdin+ of the next child,
# or, for the last child, to the caller's +stdout+.
# - Waits for the child processes to exit.
# - Returns an array of Process::Status objects (one for each child).
#
# Example:
#
# wait_threads = Open3.pipeline('ls', 'grep R')
# # => [#<Process::Status: pid 2139200 exit 0>, #<Process::Status: pid 2139202 exit 0>]
#
# Output:
#
# Rakefile
# README.md
#
# Like Process.spawn, this method has potential security vulnerabilities
# if called with untrusted input;
# see {Command Injection}[rdoc-ref:command_injection.rdoc@Command+Injection].
#
# If the first argument is a hash, it becomes leading argument +env+
# in each call to Process.spawn;
# see {Execution Environment}[rdoc-ref:Process@Execution+Environment].
#
# If the last argument is a hash, it becomes trailing argument +options+
# in each call to Process.spawn'
# see {Execution Options}[rdoc-ref:Process@Execution+Options].
#
# Each remaining argument in +cmds+ is one of:
#
# - A +command_line+: a string that begins with a shell reserved word
# or special built-in, or contains one or more metacharacters.
# - An +exe_path+: the string path to an executable to be called.
# - An array containing a +command_line+ or an +exe_path+,
# along with zero or more string arguments for the command.
#
# See {Argument command_line or exe_path}[rdoc-ref:Process@Argument+command_line+or+exe_path].
#
def pipeline(*cmds)
if Hash === cmds.last
opts = cmds.pop.dup
else
opts = {}
end
pipeline_run(cmds, opts, [], []) {|ts|
ts.map(&:value)
}
end
module_function :pipeline
def pipeline_run(cmds, pipeline_opts, child_io, parent_io) # :nodoc:
if cmds.empty?
raise ArgumentError, "no commands"
end
opts_base = pipeline_opts.dup
opts_base.delete :in
opts_base.delete :out
wait_thrs = []
r = nil
cmds.each_with_index {|cmd, i|
cmd_opts = opts_base.dup
if String === cmd
cmd = [cmd]
else
cmd_opts.update cmd.pop if Hash === cmd.last
end
if i == 0
if !cmd_opts.include?(:in)
if pipeline_opts.include?(:in)
cmd_opts[:in] = pipeline_opts[:in]
end
end
else
cmd_opts[:in] = r
end
if i != cmds.length - 1
r2, w2 = IO.pipe
cmd_opts[:out] = w2
else
if !cmd_opts.include?(:out)
if pipeline_opts.include?(:out)
cmd_opts[:out] = pipeline_opts[:out]
end
end
end
pid = spawn(*cmd, cmd_opts)
wait_thrs << Process.detach(pid)
r&.close
w2&.close
r = r2
}
result = parent_io + [wait_thrs]
child_io.each(&:close)
if defined? yield
begin
return yield(*result)
ensure
parent_io.each(&:close)
wait_thrs.each(&:join)
end
end
result
end
module_function :pipeline_run
class << self
private :pipeline_run
end
end
# JRuby uses different popen logic on Windows, require it here to reuse wrapper methods above.
require 'open3/jruby_windows' if RUBY_ENGINE == 'jruby' && JRuby::Util::ON_WINDOWS
Mini Shell