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

492 lines
15 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
# = PStore -- Transactional File Storage for Ruby Objects
#
# pstore.rb -
# originally by matz
# documentation by Kev Jackson and James Edward Gray II
# improved by Hongli Lai
#
# See PStore for documentation.
require "digest"
#
# PStore implements a file based persistence mechanism based on a Hash. User
# code can store hierarchies of Ruby objects (values) into the data store file
# by name (keys). An object hierarchy may be just a single object. User code
# may later read values back from the data store or even update data, as needed.
#
# The transactional behavior ensures that any changes succeed or fail together.
# This can be used to ensure that the data store is not left in a transitory
# state, where some values were updated but others were not.
#
# Behind the scenes, Ruby objects are stored to the data store file with
# Marshal. That carries the usual limitations. Proc objects cannot be
# marshalled, for example.
#
# == Usage example:
#
# require "pstore"
#
# # a mock wiki object...
# class WikiPage
# def initialize( page_name, author, contents )
# @page_name = page_name
# @revisions = Array.new
#
# add_revision(author, contents)
# end
#
# attr_reader :page_name
#
# def add_revision( author, contents )
# @revisions << { :created => Time.now,
# :author => author,
# :contents => contents }
# end
#
# def wiki_page_references
# [@page_name] + @revisions.last[:contents].scan(/\b(?:[A-Z]+[a-z]+){2,}/)
# end
#
# # ...
# end
#
# # create a new page...
# home_page = WikiPage.new( "HomePage", "James Edward Gray II",
# "A page about the JoysOfDocumentation..." )
#
# # then we want to update page data and the index together, or not at all...
# wiki = PStore.new("wiki_pages.pstore")
# wiki.transaction do # begin transaction; do all of this or none of it
# # store page...
# wiki[home_page.page_name] = home_page
# # ensure that an index has been created...
# wiki[:wiki_index] ||= Array.new
# # update wiki index...
# wiki[:wiki_index].push(*home_page.wiki_page_references)
# end # commit changes to wiki data store file
#
# ### Some time later... ###
#
# # read wiki data...
# wiki.transaction(true) do # begin read-only transaction, no changes allowed
# wiki.roots.each do |data_root_name|
# p data_root_name
# p wiki[data_root_name]
# end
# end
#
# == Transaction modes
#
# By default, file integrity is only ensured as long as the operating system
# (and the underlying hardware) doesn't raise any unexpected I/O errors. If an
# I/O error occurs while PStore is writing to its file, then the file will
# become corrupted.
#
# You can prevent this by setting <em>pstore.ultra_safe = true</em>.
# However, this results in a minor performance loss, and only works on platforms
# that support atomic file renames. Please consult the documentation for
# +ultra_safe+ for details.
#
# Needless to say, if you're storing valuable data with PStore, then you should
# backup the PStore files from time to time.
class PStore
RDWR_ACCESS = {mode: IO::RDWR | IO::CREAT | IO::BINARY, encoding: Encoding::ASCII_8BIT}.freeze
RD_ACCESS = {mode: IO::RDONLY | IO::BINARY, encoding: Encoding::ASCII_8BIT}.freeze
WR_ACCESS = {mode: IO::WRONLY | IO::CREAT | IO::TRUNC | IO::BINARY, encoding: Encoding::ASCII_8BIT}.freeze
# The error type thrown by all PStore methods.
class Error < StandardError
end
# Whether PStore should do its best to prevent file corruptions, even when under
# unlikely-to-occur error conditions such as out-of-space conditions and other
# unusual OS filesystem errors. Setting this flag comes at the price in the form
# of a performance loss.
#
# This flag only has effect on platforms on which file renames are atomic (e.g.
# all POSIX platforms: Linux, MacOS X, FreeBSD, etc). The default value is false.
attr_accessor :ultra_safe
#
# To construct a PStore object, pass in the _file_ path where you would like
# the data to be stored.
#
# PStore objects are always reentrant. But if _thread_safe_ is set to true,
# then it will become thread-safe at the cost of a minor performance hit.
#
def initialize(file, thread_safe = false)
dir = File::dirname(file)
unless File::directory? dir
raise PStore::Error, format("directory %s does not exist", dir)
end
if File::exist? file and not File::readable? file
raise PStore::Error, format("file %s not readable", file)
end
@filename = file
@abort = false
@ultra_safe = false
@thread_safe = thread_safe
@lock = Thread::Mutex.new
end
# Raises PStore::Error if the calling code is not in a PStore#transaction.
def in_transaction
raise PStore::Error, "not in transaction" unless @lock.locked?
end
#
# Raises PStore::Error if the calling code is not in a PStore#transaction or
# if the code is in a read-only PStore#transaction.
#
def in_transaction_wr
in_transaction
raise PStore::Error, "in read-only transaction" if @rdonly
end
private :in_transaction, :in_transaction_wr
#
# Retrieves a value from the PStore file data, by _name_. The hierarchy of
# Ruby objects stored under that root _name_ will be returned.
#
# *WARNING*: This method is only valid in a PStore#transaction. It will
# raise PStore::Error if called at any other time.
#
def [](name)
in_transaction
@table[name]
end
#
# This method is just like PStore#[], save that you may also provide a
# _default_ value for the object. In the event the specified _name_ is not
# found in the data store, your _default_ will be returned instead. If you do
# not specify a default, PStore::Error will be raised if the object is not
# found.
#
# *WARNING*: This method is only valid in a PStore#transaction. It will
# raise PStore::Error if called at any other time.
#
def fetch(name, default=PStore::Error)
in_transaction
unless @table.key? name
if default == PStore::Error
raise PStore::Error, format("undefined root name `%s'", name)
else
return default
end
end
@table[name]
end
#
# Stores an individual Ruby object or a hierarchy of Ruby objects in the data
# store file under the root _name_. Assigning to a _name_ already in the data
# store clobbers the old data.
#
# == Example:
#
# require "pstore"
#
# store = PStore.new("data_file.pstore")
# store.transaction do # begin transaction
# # load some data into the store...
# store[:single_object] = "My data..."
# store[:obj_hierarchy] = { "Kev Jackson" => ["rational.rb", "pstore.rb"],
# "James Gray" => ["erb.rb", "pstore.rb"] }
# end # commit changes to data store file
#
# *WARNING*: This method is only valid in a PStore#transaction and it cannot
# be read-only. It will raise PStore::Error if called at any other time.
#
def []=(name, value)
in_transaction_wr
@table[name] = value
end
#
# Removes an object hierarchy from the data store, by _name_.
#
# *WARNING*: This method is only valid in a PStore#transaction and it cannot
# be read-only. It will raise PStore::Error if called at any other time.
#
def delete(name)
in_transaction_wr
@table.delete name
end
#
# Returns the names of all object hierarchies currently in the store.
#
# *WARNING*: This method is only valid in a PStore#transaction. It will
# raise PStore::Error if called at any other time.
#
def roots
in_transaction
@table.keys
end
#
# Returns true if the supplied _name_ is currently in the data store.
#
# *WARNING*: This method is only valid in a PStore#transaction. It will
# raise PStore::Error if called at any other time.
#
def root?(name)
in_transaction
@table.key? name
end
# Returns the path to the data store file.
def path
@filename
end
#
# Ends the current PStore#transaction, committing any changes to the data
# store immediately.
#
# == Example:
#
# require "pstore"
#
# store = PStore.new("data_file.pstore")
# store.transaction do # begin transaction
# # load some data into the store...
# store[:one] = 1
# store[:two] = 2
#
# store.commit # end transaction here, committing changes
#
# store[:three] = 3 # this change is never reached
# end
#
# *WARNING*: This method is only valid in a PStore#transaction. It will
# raise PStore::Error if called at any other time.
#
def commit
in_transaction
@abort = false
throw :pstore_abort_transaction
end
#
# Ends the current PStore#transaction, discarding any changes to the data
# store.
#
# == Example:
#
# require "pstore"
#
# store = PStore.new("data_file.pstore")
# store.transaction do # begin transaction
# store[:one] = 1 # this change is not applied, see below...
# store[:two] = 2 # this change is not applied, see below...
#
# store.abort # end transaction here, discard all changes
#
# store[:three] = 3 # this change is never reached
# end
#
# *WARNING*: This method is only valid in a PStore#transaction. It will
# raise PStore::Error if called at any other time.
#
def abort
in_transaction
@abort = true
throw :pstore_abort_transaction
end
#
# Opens a new transaction for the data store. Code executed inside a block
# passed to this method may read and write data to and from the data store
# file.
#
# At the end of the block, changes are committed to the data store
# automatically. You may exit the transaction early with a call to either
# PStore#commit or PStore#abort. See those methods for details about how
# changes are handled. Raising an uncaught Exception in the block is
# equivalent to calling PStore#abort.
#
# If _read_only_ is set to +true+, you will only be allowed to read from the
# data store during the transaction and any attempts to change the data will
# raise a PStore::Error.
#
# Note that PStore does not support nested transactions.
#
def transaction(read_only = false) # :yields: pstore
value = nil
if !@thread_safe
raise PStore::Error, "nested transaction" unless @lock.try_lock
else
begin
@lock.lock
rescue ThreadError
raise PStore::Error, "nested transaction"
end
end
begin
@rdonly = read_only
@abort = false
file = open_and_lock_file(@filename, read_only)
if file
begin
@table, checksum, original_data_size = load_data(file, read_only)
catch(:pstore_abort_transaction) do
value = yield(self)
end
if !@abort && !read_only
save_data(checksum, original_data_size, file)
end
ensure
file.close
end
else
# This can only occur if read_only == true.
@table = {}
catch(:pstore_abort_transaction) do
value = yield(self)
end
end
ensure
@lock.unlock
end
value
end
private
# Constant for relieving Ruby's garbage collector.
CHECKSUM_ALGO = %w[SHA512 SHA384 SHA256 SHA1 RMD160 MD5].each do |algo|
begin
break Digest(algo)
rescue LoadError
end
end
EMPTY_STRING = ""
EMPTY_MARSHAL_DATA = Marshal.dump({})
EMPTY_MARSHAL_CHECKSUM = CHECKSUM_ALGO.digest(EMPTY_MARSHAL_DATA)
#
# Open the specified filename (either in read-only mode or in
# read-write mode) and lock it for reading or writing.
#
# The opened File object will be returned. If _read_only_ is true,
# and the file does not exist, then nil will be returned.
#
# All exceptions are propagated.
#
def open_and_lock_file(filename, read_only)
if read_only
begin
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
file = File.new(filename, **RD_ACCESS)
* array.c: replace rb_protect_inspect() and rb_inspecting_p() by rb_exec_recursive() in eval.c. * eval.c (rb_exec_recursive): new function. * array.c (rb_ary_join): use rb_exec_recursive(). * array.c (rb_ary_inspect, rb_ary_hash): ditto. * file.c (rb_file_join): ditto. * hash.c (rb_hash_inspect, rb_hash_to_s, rb_hash_hash): ditto. * io.c (rb_io_puts): ditto. * object.c (rb_obj_inspect): ditto * struct.c (rb_struct_inspect): ditto. * lib/set.rb (SortedSet::setup): a hack to shut up warning. [ruby-talk:132866] * lib/time.rb (Time::strptime): add new function. inspired by [ruby-talk:132815]. * lib/parsedate.rb (ParseDate::strptime): ditto. * regparse.c: move st_*_strend() functions from st.c. fixed some potential memory leaks. * exception error messages updated. [ruby-core:04497] * ext/socket/socket.c (Init_socket): add bunch of Socket constants. Patch from Sam Roberts <sroberts@uniserve.com>. [ruby-core:04409] * array.c (rb_ary_s_create): no need for negative argc check. [ruby-core:04463] * array.c (rb_ary_unshift_m): ditto. * lib/xmlrpc/parser.rb (XMLRPC::FaultException): make it subclass of StandardError class, not Exception class. [ruby-core:04429] * parse.y (fcall_gen): lvar(arg) will be evaluated as lvar.call(arg) when lvar is a defined local variable. [new] * object.c (rb_class_initialize): call inherited method before calling initializing block. * eval.c (rb_thread_start_1): initialize newly pushed frame. * lib/open3.rb (Open3::popen3): $? should not be EXIT_FAILURE. fixed: [ruby-core:04444] * eval.c (is_defined): NODE_IASGN is an assignment. * ext/readline/readline.c (Readline.readline): use rl_outstream and rl_instream. [ruby-dev:25699] * ext/etc/etc.c (Init_etc): sGroup needs HAVE_ST_GR_PASSWD check [ruby-dev:25675] * misc/ruby-mode.el: [ruby-core:04415] * lib/rdoc/generators/html_generator.rb: [ruby-core:04412] * lib/rdoc/generators/ri_generator.rb: ditto. * struct.c (make_struct): fixed: [ruby-core:04402] * ext/curses/curses.c (window_color_set): [ruby-core:04393] * ext/socket/socket.c (Init_socket): SO_REUSEPORT added. [ruby-talk:130092] * object.c: [ruby-doc:818] * parse.y (open_args): fix too verbose warnings for the space before argument parentheses. [ruby-dev:25492] * parse.y (parser_yylex): ditto. * parse.y (parser_yylex): the first expression in the parentheses should not be a command. [ruby-dev:25492] * lib/irb/context.rb (IRB::Context::initialize): [ruby-core:04330] * object.c (Init_Object): remove Object#type. [ruby-core:04335] * st.c (st_foreach): report success/failure by return value. [ruby-Bugs-1396] * parse.y: forgot to initialize parser struct. [ruby-dev:25492] * parse.y (parser_yylex): no tLABEL on EXPR_BEG. [ruby-talk:127711] * document updates - [ruby-core:04296], [ruby-core:04301], [ruby-core:04302], [ruby-core:04307] * dir.c (rb_push_glob): should work for NUL delimited patterns. * dir.c (rb_glob2): should aware of offset in the pattern. * string.c (rb_str_new4): should propagate taintedness. * env.h: rename member names in struct FRAME; last_func -> callee, orig_func -> this_func, last_class -> this_class. * struct.c (rb_struct_set): use original method name, not callee name, to retrieve member slot. [ruby-core:04268] * time.c (time_strftime): protect from format modification from GC finalizers. * object.c (Init_Object): remove rb_obj_id_obsolete() * eval.c (rb_mod_define_method): incomplete subclass check. [ruby-dev:25464] * gc.c (rb_data_object_alloc): klass may be NULL. [ruby-list:40498] * bignum.c (rb_big_rand): should return positive random number. [ruby-dev:25401] * bignum.c (rb_big_rand): do not use rb_big_modulo to generate random bignums. [ruby-dev:25396] * variable.c (rb_autoload): [ruby-dev:25373] * eval.c (svalue_to_avalue): [ruby-dev:25366] * string.c (rb_str_justify): [ruby-dev:25367] * io.c (rb_f_select): [ruby-dev:25312] * ext/socket/socket.c (sock_s_getservbyport): [ruby-talk:124072] * struct.c (make_struct): [ruby-dev:25249] * dir.c (dir_open_dir): new function. [ruby-dev:25242] * io.c (rb_f_open): add type check for return value from to_open. * lib/pstore.rb (PStore#transaction): Use the empty content when a file is not found. [ruby-dev:24561] git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@8068 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2005-03-04 01:47:45 -05:00
begin
file.flock(File::LOCK_SH)
return file
rescue
file.close
raise
* array.c: replace rb_protect_inspect() and rb_inspecting_p() by rb_exec_recursive() in eval.c. * eval.c (rb_exec_recursive): new function. * array.c (rb_ary_join): use rb_exec_recursive(). * array.c (rb_ary_inspect, rb_ary_hash): ditto. * file.c (rb_file_join): ditto. * hash.c (rb_hash_inspect, rb_hash_to_s, rb_hash_hash): ditto. * io.c (rb_io_puts): ditto. * object.c (rb_obj_inspect): ditto * struct.c (rb_struct_inspect): ditto. * lib/set.rb (SortedSet::setup): a hack to shut up warning. [ruby-talk:132866] * lib/time.rb (Time::strptime): add new function. inspired by [ruby-talk:132815]. * lib/parsedate.rb (ParseDate::strptime): ditto. * regparse.c: move st_*_strend() functions from st.c. fixed some potential memory leaks. * exception error messages updated. [ruby-core:04497] * ext/socket/socket.c (Init_socket): add bunch of Socket constants. Patch from Sam Roberts <sroberts@uniserve.com>. [ruby-core:04409] * array.c (rb_ary_s_create): no need for negative argc check. [ruby-core:04463] * array.c (rb_ary_unshift_m): ditto. * lib/xmlrpc/parser.rb (XMLRPC::FaultException): make it subclass of StandardError class, not Exception class. [ruby-core:04429] * parse.y (fcall_gen): lvar(arg) will be evaluated as lvar.call(arg) when lvar is a defined local variable. [new] * object.c (rb_class_initialize): call inherited method before calling initializing block. * eval.c (rb_thread_start_1): initialize newly pushed frame. * lib/open3.rb (Open3::popen3): $? should not be EXIT_FAILURE. fixed: [ruby-core:04444] * eval.c (is_defined): NODE_IASGN is an assignment. * ext/readline/readline.c (Readline.readline): use rl_outstream and rl_instream. [ruby-dev:25699] * ext/etc/etc.c (Init_etc): sGroup needs HAVE_ST_GR_PASSWD check [ruby-dev:25675] * misc/ruby-mode.el: [ruby-core:04415] * lib/rdoc/generators/html_generator.rb: [ruby-core:04412] * lib/rdoc/generators/ri_generator.rb: ditto. * struct.c (make_struct): fixed: [ruby-core:04402] * ext/curses/curses.c (window_color_set): [ruby-core:04393] * ext/socket/socket.c (Init_socket): SO_REUSEPORT added. [ruby-talk:130092] * object.c: [ruby-doc:818] * parse.y (open_args): fix too verbose warnings for the space before argument parentheses. [ruby-dev:25492] * parse.y (parser_yylex): ditto. * parse.y (parser_yylex): the first expression in the parentheses should not be a command. [ruby-dev:25492] * lib/irb/context.rb (IRB::Context::initialize): [ruby-core:04330] * object.c (Init_Object): remove Object#type. [ruby-core:04335] * st.c (st_foreach): report success/failure by return value. [ruby-Bugs-1396] * parse.y: forgot to initialize parser struct. [ruby-dev:25492] * parse.y (parser_yylex): no tLABEL on EXPR_BEG. [ruby-talk:127711] * document updates - [ruby-core:04296], [ruby-core:04301], [ruby-core:04302], [ruby-core:04307] * dir.c (rb_push_glob): should work for NUL delimited patterns. * dir.c (rb_glob2): should aware of offset in the pattern. * string.c (rb_str_new4): should propagate taintedness. * env.h: rename member names in struct FRAME; last_func -> callee, orig_func -> this_func, last_class -> this_class. * struct.c (rb_struct_set): use original method name, not callee name, to retrieve member slot. [ruby-core:04268] * time.c (time_strftime): protect from format modification from GC finalizers. * object.c (Init_Object): remove rb_obj_id_obsolete() * eval.c (rb_mod_define_method): incomplete subclass check. [ruby-dev:25464] * gc.c (rb_data_object_alloc): klass may be NULL. [ruby-list:40498] * bignum.c (rb_big_rand): should return positive random number. [ruby-dev:25401] * bignum.c (rb_big_rand): do not use rb_big_modulo to generate random bignums. [ruby-dev:25396] * variable.c (rb_autoload): [ruby-dev:25373] * eval.c (svalue_to_avalue): [ruby-dev:25366] * string.c (rb_str_justify): [ruby-dev:25367] * io.c (rb_f_select): [ruby-dev:25312] * ext/socket/socket.c (sock_s_getservbyport): [ruby-talk:124072] * struct.c (make_struct): [ruby-dev:25249] * dir.c (dir_open_dir): new function. [ruby-dev:25242] * io.c (rb_f_open): add type check for return value from to_open. * lib/pstore.rb (PStore#transaction): Use the empty content when a file is not found. [ruby-dev:24561] git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@8068 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2005-03-04 01:47:45 -05:00
end
rescue Errno::ENOENT
return nil
end
else
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
file = File.new(filename, **RDWR_ACCESS)
file.flock(File::LOCK_EX)
return file
end
end
# Load the given PStore file.
# If +read_only+ is true, the unmarshalled Hash will be returned.
# If +read_only+ is false, a 3-tuple will be returned: the unmarshalled
# Hash, a checksum of the data, and the size of the data.
def load_data(file, read_only)
if read_only
begin
table = load(file)
raise Error, "PStore file seems to be corrupted." unless table.is_a?(Hash)
rescue EOFError
# This seems to be a newly-created file.
table = {}
end
table
else
data = file.read
if data.empty?
# This seems to be a newly-created file.
table = {}
checksum = empty_marshal_checksum
size = empty_marshal_data.bytesize
else
table = load(data)
checksum = CHECKSUM_ALGO.digest(data)
size = data.bytesize
raise Error, "PStore file seems to be corrupted." unless table.is_a?(Hash)
end
data.replace(EMPTY_STRING)
[table, checksum, size]
end
end
def on_windows?
is_windows = RUBY_PLATFORM =~ /mswin|mingw|bccwin|wince/
self.class.__send__(:define_method, :on_windows?) do
is_windows
end
is_windows
end
def save_data(original_checksum, original_file_size, file)
new_data = dump(@table)
if new_data.bytesize != original_file_size || CHECKSUM_ALGO.digest(new_data) != original_checksum
if @ultra_safe && !on_windows?
# Windows doesn't support atomic file renames.
save_data_with_atomic_file_rename_strategy(new_data, file)
else
save_data_with_fast_strategy(new_data, file)
end
end
new_data.replace(EMPTY_STRING)
end
def save_data_with_atomic_file_rename_strategy(data, file)
temp_filename = "#{@filename}.tmp.#{Process.pid}.#{rand 1000000}"
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
temp_file = File.new(temp_filename, **WR_ACCESS)
begin
temp_file.flock(File::LOCK_EX)
temp_file.write(data)
temp_file.flush
File.rename(temp_filename, @filename)
rescue
File.unlink(temp_file) rescue nil
raise
ensure
temp_file.close
end
end
def save_data_with_fast_strategy(data, file)
file.rewind
file.write(data)
file.truncate(data.bytesize)
end
# This method is just a wrapped around Marshal.dump
# to allow subclass overriding used in YAML::Store.
def dump(table) # :nodoc:
Marshal::dump(table)
end
# This method is just a wrapped around Marshal.load.
# to allow subclass overriding used in YAML::Store.
def load(content) # :nodoc:
Marshal::load(content)
end
def empty_marshal_data
EMPTY_MARSHAL_DATA
end
def empty_marshal_checksum
EMPTY_MARSHAL_CHECKSUM
end
end