1
0
Fork 0
mirror of https://github.com/ruby/ruby.git synced 2022-11-09 12:17:21 -05:00
ruby--ruby/lib/tempfile.rb

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

426 lines
14 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
#
# tempfile - manipulates temporary files
#
# $Id$
#
require 'delegate'
require 'tmpdir'
# A utility class for managing temporary files. When you create a Tempfile
# object, it will create a temporary file with a unique filename. A Tempfile
# objects behaves just like a File object, and you can perform all the usual
# file operations on it: reading data, writing data, changing its permissions,
# etc. So although this class does not explicitly document all instance methods
# supported by File, you can in fact call any File instance method on a
# Tempfile object.
#
# == Synopsis
#
# require 'tempfile'
#
# file = Tempfile.new('foo')
# file.path # => A unique filename in the OS's temp directory,
# # e.g.: "/tmp/foo.24722.0"
# # This filename contains 'foo' in its basename.
# file.write("hello world")
# file.rewind
# file.read # => "hello world"
# file.close
# file.unlink # deletes the temp file
#
# == Good practices
#
# === Explicit close
#
# When a Tempfile object is garbage collected, or when the Ruby interpreter
# exits, its associated temporary file is automatically deleted. This means
2021-08-31 12:58:42 -04:00
# that it's unnecessary to explicitly delete a Tempfile after use, though
# it's a good practice to do so: not explicitly deleting unused Tempfiles can
# potentially leave behind a large number of temp files on the filesystem
# until they're garbage collected. The existence of these temp files can make
# it harder to determine a new Tempfile filename.
#
# Therefore, one should always call #unlink or close in an ensure block, like
# this:
#
# file = Tempfile.new('foo')
# begin
# # ...do something with file...
# ensure
# file.close
# file.unlink # deletes the temp file
# end
#
# Tempfile.create { ... } exists for this purpose and is more convenient to use.
# Note that Tempfile.create returns a File instance instead of a Tempfile, which
# also avoids the overhead and complications of delegation.
#
# Tempfile.open('foo') do |file|
# # ...do something with file...
# end
#
# === Unlink after creation
#
# On POSIX systems, it's possible to unlink a file right after creating it,
# and before closing it. This removes the filesystem entry without closing
# the file handle, so it ensures that only the processes that already had
# the file handle open can access the file's contents. It's strongly
# recommended that you do this if you do not want any other processes to
# be able to read from or write to the Tempfile, and you do not need to
# know the Tempfile's filename either.
#
# For example, a practical use case for unlink-after-creation would be this:
# you need a large byte buffer that's too large to comfortably fit in RAM,
# e.g. when you're writing a web server and you want to buffer the client's
# file upload data.
#
# Please refer to #unlink for more information and a code example.
#
# == Minor notes
#
# Tempfile's filename picking method is both thread-safe and inter-process-safe:
# it guarantees that no other threads or processes will pick the same filename.
#
# Tempfile itself however may not be entirely thread-safe. If you access the
# same Tempfile object from multiple threads then you should protect it with a
# mutex.
class Tempfile < DelegateClass(File)
# Creates a file in the underlying file system;
# returns a new \Tempfile object based on that file.
#
# If possible, consider instead using Tempfile.create, which:
#
# - Avoids the performance cost of delegation,
# incurred when Tempfile.new calls its superclass <tt>DelegateClass(File)</tt>.
# - Does not rely on a finalizer to close and unlink the file,
# which can be unreliable.
#
# Creates and returns file whose:
#
# - Class is \Tempfile (not \File, as in Tempfile.create).
# - Directory is the system temporary directory (system-dependent).
# - Generated filename is unique in that directory.
# - Permissions are <tt>0600</tt>;
# see {File Permissions}[https://docs.ruby-lang.org/en/master/File.html#label-File+Permissions].
# - Mode is <tt>'w+'</tt> (read/write mode, positioned at the end).
#
# The underlying file is removed when the \Tempfile object dies
# and is reclaimed by the garbage collector.
#
# Example:
#
# f = Tempfile.new # => #<Tempfile:/tmp/20220505-17839-1s0kt30>
# f.class # => Tempfile
# f.path # => "/tmp/20220505-17839-1s0kt30"
# f.stat.mode.to_s(8) # => "100600"
# File.exist?(f.path) # => true
# File.unlink(f.path) #
# File.exist?(f.path) # => false
#
# Argument +basename+, if given, may be one of:
#
# - A string: the generated filename begins with +basename+:
#
# Tempfile.new('foo') # => #<Tempfile:/tmp/foo20220505-17839-1whk2f>
#
# - An array of two strings <tt>[prefix, suffix]</tt>:
# the generated filename begins with +prefix+ and ends with +suffix+:
#
# Tempfile.new(%w/foo .jpg/) # => #<Tempfile:/tmp/foo20220505-17839-58xtfi.jpg>
#
# With arguments +basename+ and +tmpdir+, the file is created in directory +tmpdir+:
#
# Tempfile.new('foo', '.') # => #<Tempfile:./foo20220505-17839-xfstr8>
#
# Keyword arguments +mode+ and +options+ are passed directly to method
# {File.open}[https://docs.ruby-lang.org/en/master/File.html#method-c-open]:
#
# - The value given with +mode+ must be an integer,
# and may be expressed as the logical OR of constants defined in
# {File::Constants}[https://docs.ruby-lang.org/en/master/File/Constants.html].
# - For +options+, see {Open Options}[https://docs.ruby-lang.org/en/master/IO.html#class-IO-label-Open+Options].
#
# Related: Tempfile.create.
#
def initialize(basename="", tmpdir=nil, mode: 0, **options)
warn "Tempfile.new doesn't call the given block.", uplevel: 1 if block_given?
@unlinked = false
@mode = mode|File::RDWR|File::CREAT|File::EXCL
::Dir::Tmpname.create(basename, tmpdir, **options) do |tmpname, n, opts|
opts[:perm] = 0600
Make rb_scan_args handle keywords more similar to Ruby methods (#2460) Cfuncs that use rb_scan_args with the : entry suffer similar keyword argument separation issues that Ruby methods suffer if the cfuncs accept optional or variable arguments. This makes the following changes to : handling. * Treats as **kw, prompting keyword argument separation warnings if called with a positional hash. * Do not look for an option hash if empty keywords are provided. For backwards compatibility, treat an empty keyword splat as a empty mandatory positional hash argument, but emit a a warning, as this behavior will be removed in Ruby 3. The argument number check needs to be moved lower so it can correctly handle an empty positional argument being added. * If the last argument is nil and it is necessary to treat it as an option hash in order to make sure all arguments are processed, continue to treat the last argument as the option hash. Emit a warning in this case, as this behavior will be removed in Ruby 3. * If splitting the keyword hash into two hashes, issue a warning, as we will not be splitting hashes in Ruby 3. * If the keyword argument is required to fill a mandatory positional argument, continue to do so, but emit a warning as this behavior will be going away in Ruby 3. * If keyword arguments are provided and the last argument is not a hash, that indicates something wrong. This can happen if a cfunc is calling rb_scan_args multiple times, and providing arguments that were not passed to it from Ruby. Callers need to switch to the new rb_scan_args_kw function, which allows passing of whether keywords were provided. This commit fixes all warnings caused by the changes above. It switches some function calls to *_kw versions with appropriate kw_splat flags. If delegating arguments, RB_PASS_CALLED_KEYWORDS is used. If creating new arguments, RB_PASS_KEYWORDS is used if the last argument is a hash to be treated as keywords. In open_key_args in io.c, use rb_scan_args_kw. In this case, the arguments provided come from another C function, not Ruby. The last argument may or may not be a hash, so we can't set keyword argument mode. However, if it is a hash, we don't want to warn when treating it as keywords. In Ruby files, make sure to appropriately use keyword splats or literal keywords when calling Cfuncs that now issue keyword argument separation warnings through rb_scan_args. Also, make sure not to pass nil in place of an option hash. Work around Kernel#warn warnings due to problems in the Rubygems override of the method. There is an open pull request to fix these issues in Rubygems, but part of the Rubygems tests for their override fail on ruby-head due to rb_scan_args not recognizing empty keyword splats, which this commit fixes. Implementation wise, adding rb_scan_args_kw is kind of a pain, because rb_scan_args takes a variable number of arguments. In order to not duplicate all the code, the function internals need to be split into two functions taking a va_list, and to avoid passing in a ton of arguments, a single struct argument is used to handle the variables previously local to the function.
2019-09-25 14:18:49 -04:00
@tmpfile = File.open(tmpname, @mode, **opts)
@opts = opts.freeze
end
ObjectSpace.define_finalizer(self, Remover.new(@tmpfile))
super(@tmpfile)
end
# Opens or reopens the file with mode "r+".
def open
_close
mode = @mode & ~(File::CREAT|File::EXCL)
Make rb_scan_args handle keywords more similar to Ruby methods (#2460) Cfuncs that use rb_scan_args with the : entry suffer similar keyword argument separation issues that Ruby methods suffer if the cfuncs accept optional or variable arguments. This makes the following changes to : handling. * Treats as **kw, prompting keyword argument separation warnings if called with a positional hash. * Do not look for an option hash if empty keywords are provided. For backwards compatibility, treat an empty keyword splat as a empty mandatory positional hash argument, but emit a a warning, as this behavior will be removed in Ruby 3. The argument number check needs to be moved lower so it can correctly handle an empty positional argument being added. * If the last argument is nil and it is necessary to treat it as an option hash in order to make sure all arguments are processed, continue to treat the last argument as the option hash. Emit a warning in this case, as this behavior will be removed in Ruby 3. * If splitting the keyword hash into two hashes, issue a warning, as we will not be splitting hashes in Ruby 3. * If the keyword argument is required to fill a mandatory positional argument, continue to do so, but emit a warning as this behavior will be going away in Ruby 3. * If keyword arguments are provided and the last argument is not a hash, that indicates something wrong. This can happen if a cfunc is calling rb_scan_args multiple times, and providing arguments that were not passed to it from Ruby. Callers need to switch to the new rb_scan_args_kw function, which allows passing of whether keywords were provided. This commit fixes all warnings caused by the changes above. It switches some function calls to *_kw versions with appropriate kw_splat flags. If delegating arguments, RB_PASS_CALLED_KEYWORDS is used. If creating new arguments, RB_PASS_KEYWORDS is used if the last argument is a hash to be treated as keywords. In open_key_args in io.c, use rb_scan_args_kw. In this case, the arguments provided come from another C function, not Ruby. The last argument may or may not be a hash, so we can't set keyword argument mode. However, if it is a hash, we don't want to warn when treating it as keywords. In Ruby files, make sure to appropriately use keyword splats or literal keywords when calling Cfuncs that now issue keyword argument separation warnings through rb_scan_args. Also, make sure not to pass nil in place of an option hash. Work around Kernel#warn warnings due to problems in the Rubygems override of the method. There is an open pull request to fix these issues in Rubygems, but part of the Rubygems tests for their override fail on ruby-head due to rb_scan_args not recognizing empty keyword splats, which this commit fixes. Implementation wise, adding rb_scan_args_kw is kind of a pain, because rb_scan_args takes a variable number of arguments. In order to not duplicate all the code, the function internals need to be split into two functions taking a va_list, and to avoid passing in a ton of arguments, a single struct argument is used to handle the variables previously local to the function.
2019-09-25 14:18:49 -04:00
@tmpfile = File.open(@tmpfile.path, mode, **@opts)
__setobj__(@tmpfile)
end
def _close # :nodoc:
@tmpfile.close
end
protected :_close
# Closes the file. If +unlink_now+ is true, then the file will be unlinked
# (deleted) after closing. Of course, you can choose to later call #unlink
# if you do not unlink it now.
#
# If you don't explicitly unlink the temporary file, the removal
# will be delayed until the object is finalized.
def close(unlink_now=false)
_close
unlink if unlink_now
end
# Closes and unlinks (deletes) the file. Has the same effect as called
# <tt>close(true)</tt>.
def close!
close(true)
end
# Unlinks (deletes) the file from the filesystem. One should always unlink
# the file after using it, as is explained in the "Explicit close" good
# practice section in the Tempfile overview:
#
# file = Tempfile.new('foo')
# begin
# # ...do something with file...
# ensure
# file.close
# file.unlink # deletes the temp file
# end
#
# === Unlink-before-close
#
# On POSIX systems it's possible to unlink a file before closing it. This
# practice is explained in detail in the Tempfile overview (section
# "Unlink after creation"); please refer there for more information.
#
# However, unlink-before-close may not be supported on non-POSIX operating
# systems. Microsoft Windows is the most notable case: unlinking a non-closed
# file will result in an error, which this method will silently ignore. If
# you want to practice unlink-before-close whenever possible, then you should
# write code like this:
#
# file = Tempfile.new('foo')
# file.unlink # On Windows this silently fails.
# begin
# # ... do something with file ...
# ensure
# file.close! # Closes the file handle. If the file wasn't unlinked
# # because #unlink failed, then this method will attempt
# # to do so again.
# end
def unlink
return if @unlinked
begin
File.unlink(@tmpfile.path)
rescue Errno::ENOENT
rescue Errno::EACCES
# may not be able to unlink on Windows; just ignore
return
end
ObjectSpace.undefine_finalizer(self)
@unlinked = true
end
alias delete unlink
# Returns the full path name of the temporary file.
# This will be nil if #unlink has been called.
def path
@unlinked ? nil : @tmpfile.path
end
# Returns the size of the temporary file. As a side effect, the IO
# buffer is flushed before determining the size.
def size
if !@tmpfile.closed?
@tmpfile.size # File#size calls rb_io_flush_raw()
else
File.size(@tmpfile.path)
end
end
alias length size
# :stopdoc:
def inspect
if @tmpfile.closed?
"#<#{self.class}:#{path} (closed)>"
else
"#<#{self.class}:#{path}>"
end
end
class Remover # :nodoc:
def initialize(tmpfile)
@pid = Process.pid
@tmpfile = tmpfile
end
def call(*args)
return if @pid != Process.pid
$stderr.puts "removing #{@tmpfile.path}..." if $DEBUG
@tmpfile.close
begin
File.unlink(@tmpfile.path)
rescue Errno::ENOENT
end
$stderr.puts "done" if $DEBUG
end
end
class << self
# :startdoc:
# Creates a new Tempfile.
#
# This method is not recommended and exists mostly for backward compatibility.
# Please use Tempfile.create instead, which avoids the cost of delegation,
# does not rely on a finalizer, and also unlinks the file when given a block.
#
# Tempfile.open is still appropriate if you need the Tempfile to be unlinked
# by a finalizer and you cannot explicitly know where in the program the
# Tempfile can be unlinked safely.
#
# If no block is given, this is a synonym for Tempfile.new.
#
# If a block is given, then a Tempfile object will be constructed,
# and the block is run with the Tempfile object as argument. The Tempfile
# object will be automatically closed after the block terminates.
# However, the file will *not* be unlinked and needs to be manually unlinked
# with Tempfile#close! or Tempfile#unlink. The finalizer will try to unlink
# but should not be relied upon as it can keep the file on the disk much
# longer than intended. For instance, on CRuby, finalizers can be delayed
# due to conservative stack scanning and references left in unused memory.
#
# The call returns the value of the block.
#
# In any case, all arguments (<code>*args</code>) will be passed to Tempfile.new.
#
# Tempfile.open('foo', '/home/temp') do |f|
# # ... do something with f ...
# end
#
# # Equivalent:
# f = Tempfile.open('foo', '/home/temp')
# begin
# # ... do something with f ...
# ensure
# f.close
# end
def open(*args, **kw)
tempfile = new(*args, **kw)
if block_given?
begin
yield(tempfile)
ensure
tempfile.close
end
else
tempfile
end
end
end
end
# Creates a file in the underlying file system;
# returns a new \File object based on that file.
#
# With no block given and no arguments, creates and returns file whose:
#
# - Class is {File}[https://docs.ruby-lang.org/en/master/File.html] (not \Tempfile).
# - Directory is the system temporary directory (system-dependent).
# - Generated filename is unique in that directory.
# - Permissions are <tt>0600</tt>;
# see {File Permissions}[https://docs.ruby-lang.org/en/master/File.html#label-File+Permissions].
# - Mode is <tt>'w+'</tt> (read/write mode, positioned at the end).
#
# With no block, the file is not removed automatically,
# and so should be explicitly removed.
#
# Example:
#
# f = Tempfile.create # => #<File:/tmp/20220505-9795-17ky6f6>
# f.class # => File
# f.path # => "/tmp/20220505-9795-17ky6f6"
# f.stat.mode.to_s(8) # => "100600"
# File.exist?(f.path) # => true
# File.unlink(f.path)
# File.exist?(f.path) # => false
#
# Argument +basename+, if given, may be one of:
#
# - A string: the generated filename begins with +basename+:
#
# Tempfile.create('foo') # => #<File:/tmp/foo20220505-9795-1gok8l9>
#
# - An array of two strings <tt>[prefix, suffix]</tt>:
# the generated filename begins with +prefix+ and ends with +suffix+:
#
# Tempfile.create(%w/foo .jpg/) # => #<File:/tmp/foo20220505-17839-tnjchh.jpg>
#
# With arguments +basename+ and +tmpdir+, the file is created in directory +tmpdir+:
#
# Tempfile.create('foo', '.') # => #<File:./foo20220505-9795-1emu6g8>
#
# Keyword arguments +mode+ and +options+ are passed directly to method
# {File.open}[https://docs.ruby-lang.org/en/master/File.html#method-c-open]:
#
# - The value given with +mode+ must be an integer,
# and may be expressed as the logical OR of constants defined in
# {File::Constants}[https://docs.ruby-lang.org/en/master/File/Constants.html].
# - For +options+, see {Open Options}[https://docs.ruby-lang.org/en/master/IO.html#class-IO-label-Open+Options].
#
# With a block given, creates the file as above, passes it to the block,
# and returns the block's value;
# before the return, the file object is closed and the underlying file is removed:
#
# Tempfile.create {|file| file.path } # => "/tmp/20220505-9795-rkists"
#
# Related: Tempfile.new.
#
def Tempfile.create(basename="", tmpdir=nil, mode: 0, **options)
tmpfile = nil
Dir::Tmpname.create(basename, tmpdir, **options) do |tmpname, n, opts|
mode |= File::RDWR|File::CREAT|File::EXCL
opts[:perm] = 0600
Make rb_scan_args handle keywords more similar to Ruby methods (#2460) Cfuncs that use rb_scan_args with the : entry suffer similar keyword argument separation issues that Ruby methods suffer if the cfuncs accept optional or variable arguments. This makes the following changes to : handling. * Treats as **kw, prompting keyword argument separation warnings if called with a positional hash. * Do not look for an option hash if empty keywords are provided. For backwards compatibility, treat an empty keyword splat as a empty mandatory positional hash argument, but emit a a warning, as this behavior will be removed in Ruby 3. The argument number check needs to be moved lower so it can correctly handle an empty positional argument being added. * If the last argument is nil and it is necessary to treat it as an option hash in order to make sure all arguments are processed, continue to treat the last argument as the option hash. Emit a warning in this case, as this behavior will be removed in Ruby 3. * If splitting the keyword hash into two hashes, issue a warning, as we will not be splitting hashes in Ruby 3. * If the keyword argument is required to fill a mandatory positional argument, continue to do so, but emit a warning as this behavior will be going away in Ruby 3. * If keyword arguments are provided and the last argument is not a hash, that indicates something wrong. This can happen if a cfunc is calling rb_scan_args multiple times, and providing arguments that were not passed to it from Ruby. Callers need to switch to the new rb_scan_args_kw function, which allows passing of whether keywords were provided. This commit fixes all warnings caused by the changes above. It switches some function calls to *_kw versions with appropriate kw_splat flags. If delegating arguments, RB_PASS_CALLED_KEYWORDS is used. If creating new arguments, RB_PASS_KEYWORDS is used if the last argument is a hash to be treated as keywords. In open_key_args in io.c, use rb_scan_args_kw. In this case, the arguments provided come from another C function, not Ruby. The last argument may or may not be a hash, so we can't set keyword argument mode. However, if it is a hash, we don't want to warn when treating it as keywords. In Ruby files, make sure to appropriately use keyword splats or literal keywords when calling Cfuncs that now issue keyword argument separation warnings through rb_scan_args. Also, make sure not to pass nil in place of an option hash. Work around Kernel#warn warnings due to problems in the Rubygems override of the method. There is an open pull request to fix these issues in Rubygems, but part of the Rubygems tests for their override fail on ruby-head due to rb_scan_args not recognizing empty keyword splats, which this commit fixes. Implementation wise, adding rb_scan_args_kw is kind of a pain, because rb_scan_args takes a variable number of arguments. In order to not duplicate all the code, the function internals need to be split into two functions taking a va_list, and to avoid passing in a ton of arguments, a single struct argument is used to handle the variables previously local to the function.
2019-09-25 14:18:49 -04:00
tmpfile = File.open(tmpname, mode, **opts)
end
if block_given?
begin
yield tmpfile
ensure
unless tmpfile.closed?
if File.identical?(tmpfile, tmpfile.path)
unlinked = File.unlink tmpfile.path rescue nil
end
tmpfile.close
end
unless unlinked
begin
File.unlink tmpfile.path
rescue Errno::ENOENT
end
end
end
else
tmpfile
end
end