module Sequel::Model::InstanceMethods
Sequel::Model instance methods that implement basic model functionality.
-
All of the methods in
HOOKS
andAROUND_HOOKS
create instance methods that are called by Sequel when the appropriate action occurs. For example, when destroying a model object, Sequel will callaround_destroy
, which will callbefore_destroy
, do the destroy, and then callafter_destroy
. -
The following instance_methods all call the class method of the same name: columns, db, primary_key, db_schema.
-
All of the methods in
BOOLEAN_SETTINGS
create attr_writers allowing you to set values for the attribute. It also creates instance getters returning the value of the setting. If the value has not yet been set, it gets the default value from the class by calling the class method of the same name.
Attributes
The hash of attribute values. Keys are symbols with the names of the underlying database columns. The returned hash is a reference to the receiver's values hash, and modifying it will also modify the receiver's values.
Artist.new(:name=>'Bob').values # => {:name=>'Bob'} Artist[1].values # => {:id=>1, :name=>'Jim', ...}
The hash of attribute values. Keys are symbols with the names of the underlying database columns. The returned hash is a reference to the receiver's values hash, and modifying it will also modify the receiver's values.
Artist.new(:name=>'Bob').values # => {:name=>'Bob'} Artist[1].values # => {:id=>1, :name=>'Jim', ...}
The hash of attribute values. Keys are symbols with the names of the underlying database columns. The returned hash is a reference to the receiver's values hash, and modifying it will also modify the receiver's values.
Artist.new(:name=>'Bob').values # => {:name=>'Bob'} Artist[1].values # => {:id=>1, :name=>'Jim', ...}
Public Class Methods
Creates new instance and passes the given values to set. If a block is given, yield the instance to the block.
Arguments:
- values
-
should be a hash to pass to set.
Artist.new(:name=>'Bob') Artist.new do |a| a.name = 'Bob' end
# File lib/sequel/model/base.rb, line 1377 def initialize(values = {}) @values = {} @new = true @modified = true initialize_set(values) changed_columns.clear yield self if block_given? end
Public Instance Methods
Alias of eql?
# File lib/sequel/model/base.rb, line 1414 def ==(obj) eql?(obj) end
If pk is not nil, true only if the objects have the same class and pk. If pk is nil, false.
Artist[1] === Artist[1] # true Artist.new === Artist.new # false Artist[1].set(:name=>'Bob') == Artist[1] # => true
# File lib/sequel/model/base.rb, line 1424 def ===(obj) pk.nil? ? false : (obj.class == model) && (obj.pk == pk) end
Returns value of the column's attribute.
Artist[1][:id] #=> 1
# File lib/sequel/model/base.rb, line 1389 def [](column) @values[column] end
Sets the value for the given column. If typecasting is enabled for this object, typecast the value based on the column's type. If this is a new record or the typecasted value isn't the same as the current value for the column, mark the column as changed.
a = Artist.new a[:name] = 'Bob' a.values #=> {:name=>'Bob'}
# File lib/sequel/model/base.rb, line 1401 def []=(column, value) # If it is new, it doesn't have a value yet, so we should # definitely set the new value. # If the column isn't in @values, we can't assume it is # NULL in the database, so assume it has changed. v = typecast_value(column, value) vals = @values if new? || !vals.include?(column) || v != (c = vals[column]) || v.class != c.class change_column_value(column, v) end end
The autoincrementing primary key for this model object. Should be overridden if you have a composite primary key with one part of it being autoincrementing.
# File lib/sequel/model/base.rb, line 1439 def autoincrementing_primary_key primary_key end
Cancel the current action. Should be called in before hooks to halt the
processing of the action. If a msg
argument is given and the
model instance is configured to raise exceptions on failure, sets the
message to use for the raised HookFailed
exception.
# File lib/sequel/model/base.rb, line 1447 def cancel_action(msg=nil) raise_hook_failure(msg) end
The columns that have been updated. This isn't completely accurate, as it could contain columns whose values have not changed.
a = Artist[1] a.changed_columns # => [] a.name = 'Bob' a.changed_columns # => [:name]
# File lib/sequel/model/base.rb, line 1458 def changed_columns @changed_columns ||= [] end
Deletes and returns self
. Does not run destroy hooks. Look
into using destroy
instead.
Artist[1].delete # DELETE FROM artists WHERE (id = 1) # => #<Artist {:id=>1, ...}>
# File lib/sequel/model/base.rb, line 1467 def delete raise Sequel::Error, "can't delete frozen object" if frozen? _delete self end
Like delete but runs hooks before and after delete. If before_destroy returns false, returns false without deleting the object from the database. Otherwise, deletes the item from the database and returns self. Uses a transaction if use_transactions is true or if the :transaction option is given and true.
Artist[1].destroy # BEGIN; DELETE FROM artists WHERE (id = 1); COMMIT; # => #<Artist {:id=>1, ...}>
# File lib/sequel/model/base.rb, line 1482 def destroy(opts = OPTS) raise Sequel::Error, "can't destroy frozen object" if frozen? checked_save_failure(opts){checked_transaction(opts){_destroy(opts)}} end
Iterates through all of the current values using each.
Album[1].each{|k, v| puts "#{k} => #{v}"} # id => 1 # name => 'Bob'
# File lib/sequel/model/base.rb, line 1492 def each(&block) @values.each(&block) end
Compares model instances by values.
Artist[1] == Artist[1] # => true Artist.new == Artist.new # => true Artist[1].set(:name=>'Bob') == Artist[1] # => false
# File lib/sequel/model/base.rb, line 1501 def eql?(obj) (obj.class == model) && (obj.values == @values) end
Returns the validation errors associated with this object. See
Errors
.
# File lib/sequel/model/base.rb, line 1507 def errors @errors ||= errors_class.new end
Returns true when current instance exists, false otherwise. Generally an object that isn't new will exist unless it has been deleted. Uses a database query to check for existence, unless the model object is new, in which case this is always false.
Artist[1].exists? # SELECT 1 FROM artists WHERE (id = 1) # => true Artist.new.exists? # => false
# File lib/sequel/model/base.rb, line 1521 def exists? new? ? false : !this.get(SQL::AliasedExpression.new(1, :one)).nil? end
Ignore the model's setter method cache when this instances extends a module, as the module may contain setter methods.
# File lib/sequel/model/base.rb, line 1527 def extend(mod) @singleton_setter_added = true super end
Freeze the object in such a way that it is still usable but not modifiable. Once an object is frozen, you cannot modify it's values, #changed_columns, errors, or dataset.
# File lib/sequel/model/base.rb, line 1535 def freeze values.freeze changed_columns.freeze unless errors.frozen? validate errors.freeze end this.freeze if !new? && model.primary_key super end
Value that should be unique for objects with the same class and pk (if pk is not nil), or the same class and values (if pk is nil).
Artist[1].hash == Artist[1].hash # true Artist[1].set(:name=>'Bob').hash == Artist[1].hash # true Artist.new.hash == Artist.new.hash # true Artist.new(:name=>'Bob').hash == Artist.new.hash # false
# File lib/sequel/model/base.rb, line 1553 def hash case primary_key when Array [model, !pk.all? ? @values : pk].hash when Symbol [model, pk.nil? ? @values : pk].hash else [model, @values].hash end end
Returns value for the :id attribute, even if the primary key is not id. To
get the primary key value, use pk
.
Artist[1].id # => 1
# File lib/sequel/model/base.rb, line 1568 def id @values[:id] end
Returns a string representation of the model instance including the class name and values.
# File lib/sequel/model/base.rb, line 1574 def inspect "#<#{model.name} @values=#{inspect_values}>" end
Returns the keys in values
. May not include all column names.
Artist.new.keys # => [] Artist.new(:name=>'Bob').keys # => [:name] Artist[1].keys # => [:id, :name]
# File lib/sequel/model/base.rb, line 1583 def keys @values.keys end
Refresh this record using for_update
(by default, or the
specified style when given) unless this is a new record. Returns self.
This can be used to make sure no other process is updating the record at
the same time.
If style is a string, it will be used directly. You should never pass a string to this method that is derived from user input, as that can lead to SQL injection.
A symbol may be used for database independent locking behavior, but all supported symbols have separate methods (e.g. for_update).
a = Artist[1] Artist.db.transaction do a.lock! a.update(:name=>'A') end a = Artist[2] Artist.db.transaction do a.lock!('FOR NO KEY UPDATE') a.update(:name=>'B') end
# File lib/sequel/model/base.rb, line 1610 def lock!(style=:update) _refresh(this.lock_style(style)) unless new? self end
Remove elements of the model object that make marshalling fail. Returns self.
a = Artist[1] a.marshallable! Marshal.dump(a)
# File lib/sequel/model/base.rb, line 1620 def marshallable! @this = nil self end
Explicitly mark the object as modified, so
save_changes
/update
will run callbacks even if no
columns have changed.
a = Artist[1] a.save_changes # No callbacks run, as no changes a.modified! a.save_changes # Callbacks run, even though no changes made
If a column is given, specifically marked that column as modified, so that
save_changes
/update
will include that column in
the update. This should be used if you plan on mutating the column value
instead of assigning a new column value:
a.modified!(:name) a.name.gsub!(/[aeou]/, 'i')
# File lib/sequel/model/base.rb, line 1640 def modified!(column=nil) if column && !changed_columns.include?(column) changed_columns << column end @modified = true end
Whether this object has been modified since last saved, used by #save_changes to determine whether changes should be saved. New values are always considered modified.
a = Artist[1] a.modified? # => false a.set(:name=>'Jim') a.modified? # => true
If a column is given, specifically check if the given column has been modified:
a.modified?(:num_albums) # => false a.num_albums = 10 a.modified?(:num_albums) # => true
# File lib/sequel/model/base.rb, line 1662 def modified?(column=nil) if column changed_columns.include?(column) else @modified || !changed_columns.empty? end end
Returns true if the current instance represents a new record.
Artist.new.new? # => true Artist[1].new? # => false
# File lib/sequel/model/base.rb, line 1674 def new? defined?(@new) ? @new : (@new = false) end
Returns the primary key value identifying the model instance. Raises an
Error
if this model does not have a primary key. If the model
has a composite primary key, returns an array of values.
Artist[1].pk # => 1 Artist[[1, 2]].pk # => [1, 2]
# File lib/sequel/model/base.rb, line 1684 def pk raise(Error, "No primary key is associated with this model") unless key = primary_key if key.is_a?(Array) vals = @values key.map{|k| vals[k]} else @values[key] end end
Returns a hash mapping the receivers primary key column(s) to their values.
Artist[1].pk_hash # => {:id=>1} Artist[[1, 2]].pk_hash # => {:id1=>1, :id2=>2}
# File lib/sequel/model/base.rb, line 1698 def pk_hash model.primary_key_hash(pk) end
Returns a hash mapping the receivers qualified primary key column(s) to their values.
Artist[1].qualified_pk_hash # => {Sequel[:artists][:id]=>1} Artist[[1, 2]].qualified_pk_hash # => {Sequel[:artists][:id1]=>1, Sequel[:artists][:id2]=>2}
# File lib/sequel/model/base.rb, line 1708 def qualified_pk_hash(qualifier=model.table_name) model.qualified_primary_key_hash(pk, qualifier) end
Reloads attributes from database and returns self. Also clears all #changed_columns
information. Raises an Error
if the record no longer exists
in the database.
a = Artist[1] a.name = 'Jim' a.refresh a.name # => 'Bob'
# File lib/sequel/model/base.rb, line 1720 def refresh raise Sequel::Error, "can't refresh frozen object" if frozen? _refresh(this) self end
Alias of refresh, but not aliased directly to make overriding in a plugin easier.
# File lib/sequel/model/base.rb, line 1727 def reload refresh end
Creates or updates the record, after making sure the record is valid and before hooks execute successfully. Fails if:
-
the record is not valid, or
-
before_save returns false, or
-
the record is new and before_create returns false, or
-
the record is not new and before_update returns false.
If save
fails and either raise_on_save_failure or the
:raise_on_failure option is true, it raises ValidationFailed or HookFailed. Otherwise it returns nil.
If it succeeds, it returns self.
You can provide an optional list of columns to update, in which case it only updates those columns, or a options hash.
Takes the following options:
- :changed
-
save all changed columns, instead of all columns or the columns given
- :columns
-
array of specific columns that should be saved.
- :raise_on_failure
-
set to true or false to override the current
raise_on_save_failure
setting - :server
-
set the server/shard on the object before saving, and use that server/shard in any transaction.
- :transaction
-
set to true or false to override the current
use_transactions
setting - :validate
-
set to false to skip validation
# File lib/sequel/model/base.rb, line 1759 def save(opts=OPTS) raise Sequel::Error, "can't save frozen object" if frozen? set_server(opts[:server]) if opts[:server] _before_validation if opts[:validate] != false # SEQUEL5: Remove if unless checked_save_failure(opts){_valid?(opts)} raise(ValidationFailed.new(self)) if raise_on_failure?(opts) return end end checked_save_failure(opts){checked_transaction(opts){_save(opts)}} end
Saves only changed columns if the object has been modified. If the object
has not been modified, returns nil. If unable to save, returns false
unless raise_on_save_failure
is true.
a = Artist[1] a.save_changes # => nil a.name = 'Jim' a.save_changes # UPDATE artists SET name = 'Bob' WHERE (id = 1) # => #<Artist {:id=>1, :name=>'Jim', ...}
# File lib/sequel/model/base.rb, line 1781 def save_changes(opts=OPTS) save(Hash[opts].merge!(:changed=>true)) || false if modified? end
Updates the instance with the supplied values with support for virtual
attributes, raising an exception if a value is used that doesn't have a
setter method (or ignoring it if strict_param_setting =
false
). Does not save the record.
artist.set(:name=>'Jim') artist.name # => 'Jim'
# File lib/sequel/model/base.rb, line 1792 def set(hash) set_restricted(hash, :default) end
Set all values using the entries in the hash, ignoring any setting of allowed_columns in the model.
Artist.set_allowed_columns(:num_albums) artist.set_all(:name=>'Jim') artist.name # => 'Jim'
# File lib/sequel/model/base.rb, line 1802 def set_all(hash) set_restricted(hash, :all) end
For each of the fields in the given array fields
, call the
setter method with the value of that hash
entry for the field.
Returns self.
You can provide an options hash, with the following options currently respected:
- :missing
-
Can be set to :skip to skip missing entries or :raise to raise an Error for missing entries. The default behavior is not to check for missing entries, in which case the default value is used. To be friendly with most web frameworks, the missing check will also check for the string version of the argument in the hash if given a symbol.
Examples:
artist.set_fields({:name=>'Jim'}, [:name]) artist.name # => 'Jim' artist.set_fields({:hometown=>'LA'}, [:name]) artist.name # => nil artist.hometown # => 'Sac' artist.name # => 'Jim' artist.set_fields({}, [:name], :missing=>:skip) artist.name # => 'Jim' artist.name # => 'Jim' artist.set_fields({}, [:name], :missing=>:raise) # Sequel::Error raised
# File lib/sequel/model/base.rb, line 1832 def set_fields(hash, fields, opts=nil) opts = if opts Hash[model.default_set_fields_options].merge!(opts) else model.default_set_fields_options end case opts[:missing] when :skip fields.each do |f| if hash.has_key?(f) set_column_value("#{f}=", hash[f]) elsif f.is_a?(Symbol) && hash.has_key?(sf = f.to_s) set_column_value("#{sf}=", hash[sf]) end end when :raise fields.each do |f| if hash.has_key?(f) set_column_value("#{f}=", hash[f]) elsif f.is_a?(Symbol) && hash.has_key?(sf = f.to_s) set_column_value("#{sf}=", hash[sf]) else raise(Sequel::Error, "missing field in hash: #{f.inspect} not in #{hash.inspect}") end end else fields.each{|f| set_column_value("#{f}=", hash[f])} end self end
Set the values using the entries in the hash, only if the key is included
in only. It may be a better idea to use set_fields
instead of
this method.
artist.set_only({:name=>'Jim'}, :name) artist.name # => 'Jim' artist.set_only({:hometown=>'LA'}, :name) # Raise Error
# File lib/sequel/model/base.rb, line 1872 def set_only(hash, *only) set_restricted(hash, only.flatten) end
Set the shard that this object is tied to. Returns self.
# File lib/sequel/model/base.rb, line 1877 def set_server(s) @server = s @this = @this.server(s) if @this self end
Clear the #setter_methods cache when a method is added
# File lib/sequel/model/base.rb, line 1884 def singleton_method_added(meth) @singleton_setter_added = true if meth.to_s =~ SETTER_METHOD_REGEXP super end
Returns (naked) dataset that should return only this instance.
Artist[1].this # SELECT * FROM artists WHERE (id = 1) LIMIT 1
# File lib/sequel/model/base.rb, line 1893 def this return @this if @this raise Error, "No dataset for model #{model}" unless ds = model.instance_dataset cond = if ds.joined_dataset? qualified_pk_hash else pk_hash end @this = use_server(ds.where(cond)) end
Runs set with the passed hash and then runs save_changes.
artist.update(:name=>'Jim') # UPDATE artists SET name = 'Jim' WHERE (id = 1)
# File lib/sequel/model/base.rb, line 1909 def update(hash) update_restricted(hash, :default) end
Update all values using the entries in the hash, ignoring any setting of
allowed_columns
in the model.
Artist.set_allowed_columns(:num_albums) artist.update_all(:name=>'Jim') # UPDATE artists SET name = 'Jim' WHERE (id = 1)
# File lib/sequel/model/base.rb, line 1918 def update_all(hash) update_restricted(hash, :all) end
Update the instances values by calling set_fields
with the
arguments, then saves any changes to the record. Returns self.
artist.update_fields({:name=>'Jim'}, [:name]) # UPDATE artists SET name = 'Jim' WHERE (id = 1) artist.update_fields({:hometown=>'LA'}, [:name]) # UPDATE artists SET name = NULL WHERE (id = 1)
# File lib/sequel/model/base.rb, line 1930 def update_fields(hash, fields, opts=nil) set_fields(hash, fields, opts) save_changes end
Update the values using the entries in the hash, only if the key is
included in only. It may be a better idea to use
update_fields
instead of this method.
artist.update_only({:name=>'Jim'}, :name) # UPDATE artists SET name = 'Jim' WHERE (id = 1) artist.update_only({:hometown=>'LA'}, :name) # Raise Error
# File lib/sequel/model/base.rb, line 1943 def update_only(hash, *only) update_restricted(hash, only.flatten) end
Validates the object and returns true if no errors are reported.
artist.set(:name=>'Valid').valid? # => true artist.set(:name=>'Invalid').valid? # => false artist.errors.full_messages # => ['name cannot be Invalid']
# File lib/sequel/model/base.rb, line 1961 def valid?(opts = OPTS) _before_validation begin _valid?(opts) rescue HookFailed false end end
Validates the object. If the object is invalid, errors should be added to
the errors attribute. By default, does nothing, as all models are valid by
default. See the “Model
Validations” guide. for details about validation. Should not be called
directly by user code, call valid?
instead to check if an
object is valid.
# File lib/sequel/model/base.rb, line 1953 def validate end
Private Instance Methods
Run code directly after the INSERT query, before after_create. This is only a temporary API, it should not be overridden by external code.
# File lib/sequel/model/base.rb, line 1974 def _after_create(pk) # SEQUEL5: Remove @this = nil @new = false @was_new = true end
Run code after around_save returns, before calling after_commit. This is only a temporary API, it should not be overridden by external code.
# File lib/sequel/model/base.rb, line 1983 def _after_save(pk) # SEQUEL5: Remove if @was_new @was_new = nil pk ? _save_refresh : changed_columns.clear else @columns_updated = nil end @modified = false end
Run code directly after the UPDATE query, before after_update. This is only a temporary API, it should not be overridden by external code.
# File lib/sequel/model/base.rb, line 1996 def _after_update # SEQUEL5: Remove @this = nil end
Run code before any validation is done, but also run it before saving even if validation is skipped. This is a private hook. It exists so that plugins can set values automatically before validation (as the values need to be validated), but should be set even if validation is skipped. Unlike the regular before_validation hook, we do not skip the save/validation if this returns false.
# File lib/sequel/model/base.rb, line 2007 def _before_validation end
Do the deletion of the object's dataset, and check that the row was actually deleted.
# File lib/sequel/model/base.rb, line 2012 def _delete n = _delete_without_checking raise(NoExistingObject, "Attempt to delete object did not result in a single row modification (Rows Deleted: #{n}, SQL: #{_delete_dataset.delete_sql})") if require_modification && n != 1 n end
The dataset to use when deleting the object. The same as the object's dataset by default.
# File lib/sequel/model/base.rb, line 2020 def _delete_dataset this end
Actually do the deletion of the object's dataset. Return the number of rows modified.
# File lib/sequel/model/base.rb, line 2026 def _delete_without_checking if sql = (m = model).fast_instance_delete_sql sql = sql.dup ds = use_server(m.dataset) ds.literal_append(sql, pk) ds.with_sql_delete(sql) else _delete_dataset.delete end end
Internal destroy method, separted from destroy to allow running inside a transaction
# File lib/sequel/model/base.rb, line 2039 def _destroy(opts) sh = {:server=>this_server} uacr = use_after_commit_rollback if uacr.nil? ? (method(:after_destroy_rollback).owner != InstanceMethods) : uacr Sequel::Deprecation.deprecate("Model#after_destroy_rollback", "Instead, call db.after_rollback in Model#before_destroy") db.after_rollback(sh){after_destroy_rollback} end called = false around_destroy do called = true if before_destroy == false Sequel::Deprecation.deprecate("Having before_destroy return false to cancel the destroy", "Instead, call cancel_action inside before_destroy") raise_hook_failure(:before_destroy) end _destroy_delete after_destroy true end raise_hook_failure(:around_destroy) unless called if uacr.nil? ? (method(:after_destroy_commit).owner != InstanceMethods) : uacr Sequel::Deprecation.deprecate("Model#after_destroy_commit", "Instead, call db.after_commit in Model#after_destroy") db.after_commit(sh){after_destroy_commit} end self end
Internal delete method to call when destroying an object, separated from delete to allow you to override destroy's version without affecting delete.
# File lib/sequel/model/base.rb, line 2068 def _destroy_delete delete end
Insert the record into the database, returning the primary key if the record should be refreshed from the database.
# File lib/sequel/model/base.rb, line 2074 def _insert ds = _insert_dataset if _use_insert_select?(ds) && (h = _insert_select_raw(ds)) _save_set_values(h) nil else iid = _insert_raw(ds) # if we have a regular primary key and it's not set in @values, # we assume it's the last inserted id if (pk = autoincrementing_primary_key) && pk.is_a?(Symbol) && !(vals = @values)[pk] vals[pk] = iid end pk end end
The dataset to use when inserting a new object. The same as the model's dataset by default.
# File lib/sequel/model/base.rb, line 2092 def _insert_dataset use_server(model.instance_dataset) end
Insert into the given dataset and return the primary key created (if any).
# File lib/sequel/model/base.rb, line 2097 def _insert_raw(ds) ds.insert(_insert_values) end
Insert into the given dataset and return the hash of column values.
# File lib/sequel/model/base.rb, line 2102 def _insert_select_raw(ds) ds.insert_select(_insert_values) end
Refresh using a particular dataset, used inside save to make sure the same server is used for reading newly inserted values from the database
# File lib/sequel/model/base.rb, line 2111 def _refresh(dataset) _refresh_set_values(_refresh_get(dataset) || raise(NoExistingObject, "Record not found")) changed_columns.clear end
Get the row of column data from the database.
# File lib/sequel/model/base.rb, line 2117 def _refresh_get(dataset) if (sql = model.fast_pk_lookup_sql) && !dataset.opts[:lock] sql = sql.dup ds = use_server(dataset) ds.literal_append(sql, pk) ds.with_sql_first(sql) else dataset.first end end
Set the refreshed values after
# File lib/sequel/model/base.rb, line 2129 def _refresh_set_values(h) @values = h end
Internal version of save, split from save to allow running inside it's own transaction.
# File lib/sequel/model/base.rb, line 2135 def _save(opts) sh = {:server=>this_server} uacr = use_after_commit_rollback if uacr.nil? ? (method(:after_rollback).owner != InstanceMethods) : uacr Sequel::Deprecation.deprecate("Model#after_rollback", "Instead, call db.after_rollback in Model#before_save.") db.after_rollback(sh){after_rollback} end pk = nil called_save = false called_cu = false around_save do called_save = true if before_save == false Sequel::Deprecation.deprecate("Having before_save return false to cancel the save", "Instead, call cancel_action inside before_save") raise_hook_failure(:before_save) end if new? around_create do called_cu = true if before_create == false Sequel::Deprecation.deprecate("Having before_create return false to cancel the create", "Instead, call cancel_action inside before_create") raise_hook_failure(:before_create) end pk = _insert _after_create(pk) # SEQUEL5: Remove # SEQUEL5 # @this = nil # @new = false # @modified = false # pk ? _save_refresh : changed_columns.clear after_create true end raise_hook_failure(:around_create) unless called_cu else around_update do called_cu = true if before_update == false Sequel::Deprecation.deprecate("Having before_update return false to cancel the update", "Instead, call cancel_action inside before_update") raise_hook_failure(:before_update) end columns = opts[:columns] if columns.nil? columns_updated = if opts[:changed] # SEQUEL5: Use local variable instead of instance variable @values.reject{|k,v| !changed_columns.include?(k)} else _save_update_all_columns_hash end changed_columns.clear else # update only the specified columns columns = Array(columns) columns_updated = @values.reject{|k, v| !columns.include?(k)} changed_columns.reject!{|c| columns.include?(c)} end _update_columns(columns_updated) _after_update # SEQUEL5: Remove # SEQUEL5 # @this = nil # @modified = false after_update true end raise_hook_failure(:around_update) unless called_cu end after_save true end raise_hook_failure(:around_save) unless called_save _after_save(pk) # SEQUEL5: Remove if uacr.nil? ? (method(:after_commit).owner != InstanceMethods) : uacr Sequel::Deprecation.deprecate("Model#after_commit", "Instead, call db.after_commit in Model#after_save.") db.after_commit(sh){after_commit} end self end
Refresh the object after saving it, used to get default values of all columns. Separated from _save so it can be overridden to avoid the refresh.
# File lib/sequel/model/base.rb, line 2215 def _save_refresh _save_set_values(_refresh_get(this.server?(:default)) || raise(NoExistingObject, "Record not found")) changed_columns.clear end
Set values to the provided hash. Called after a create, to set the full values from the database in the model instance.
# File lib/sequel/model/base.rb, line 2222 def _save_set_values(h) @values = h end
Return a hash of values used when saving all columns of an existing object (i.e. not passing specific columns to save or using update/save_changes). Defaults to all of the object's values except unmodified primary key columns, as some databases don't like you setting primary key values even to their existing values.
# File lib/sequel/model/base.rb, line 2232 def _save_update_all_columns_hash v = Hash[@values] Array(primary_key).each{|x| v.delete(x) unless changed_columns.include?(x)} v end
Update this instance's dataset with the supplied column hash, checking that only a single row was modified.
# File lib/sequel/model/base.rb, line 2248 def _update(columns) n = _update_without_checking(columns) raise(NoExistingObject, "Attempt to update object did not result in a single row modification (SQL: #{_update_dataset.update_sql(columns)})") if require_modification && n != 1 n end
Call _update with the given columns, if any are present. Plugins can override this method in order to update with additional columns, even when the column hash is initially empty.
# File lib/sequel/model/base.rb, line 2241 def _update_columns(columns) @columns_updated ||= DeprecatedColumnsUpdated.new(columns) # SEQUEL5: Remove _update(columns) unless columns.empty? end
The dataset to use when updating an object. The same as the object's dataset by default.
# File lib/sequel/model/base.rb, line 2256 def _update_dataset this end
Update this instances dataset with the supplied column hash.
# File lib/sequel/model/base.rb, line 2261 def _update_without_checking(columns) _update_dataset.update(columns) end
Whether to use insert_select when inserting a new row.
# File lib/sequel/model/base.rb, line 2266 def _use_insert_select?(ds) (!ds.opts[:select] || ds.opts[:returning]) && ds.supports_insert_select? end
Internal validation method, running validation hooks.
# File lib/sequel/model/base.rb, line 2271 def _valid?(opts) return errors.empty? if frozen? errors.clear called = false # skip_validate = opts[:validate] == false # SEQUEL5 around_validation do called = true if before_validation == false Sequel::Deprecation.deprecate("Having before_validation return false to mark the object as invalid", "Instead, call cancel_action inside before_validation") raise_hook_failure(:before_validation) else validate # unless skip_validate # SEQUEL5 after_validation end end # return true if skip_validate # SEQUEL5 if called errors.empty? else raise_hook_failure(:around_validation) end end
Change the value of the column to given value, recording the change.
# File lib/sequel/model/base.rb, line 2314 def change_column_value(column, value) cc = changed_columns cc << column unless cc.include?(column) @values[column] = value end
If not raising on failure, check for HookFailed being raised by yielding and swallow it.
# File lib/sequel/model/base.rb, line 2296 def checked_save_failure(opts) if raise_on_failure?(opts) yield else begin yield rescue HookFailed nil end end end
If transactions should be used, wrap the yield in a transaction block.
# File lib/sequel/model/base.rb, line 2309 def checked_transaction(opts=OPTS) use_transaction?(opts) ? db.transaction({:server=>this_server}.merge!(opts)){yield} : yield end
:nocov: Ruby 1.8 doesn't support #initialize_clone, so override clone to dup and freeze.
# File lib/sequel/model/base.rb, line 2336 def clone o = dup o.freeze if frozen? o end
Default error class used for errors.
# File lib/sequel/model/base.rb, line 2321 def errors_class Errors end
Clone constructor – freeze internal data structures if the original's are frozen.
# File lib/sequel/model/base.rb, line 2328 def initialize_clone(other) super freeze if other.frozen? self end
Copy constructor – Duplicate internal data structures.
# File lib/sequel/model/base.rb, line 2346 def initialize_copy(other) super @values = Hash[@values] @changed_columns = @changed_columns.dup if @changed_columns @errors = @errors.dup if @errors @this = @this.dup if @this self end
Set the columns with the given hash. By default, the same as
set
, but exists so it can be overridden. This is called only
for new records, before #changed_columns
is cleared.
# File lib/sequel/model/base.rb, line 2358 def initialize_set(h) set(h) unless h.empty? end
Default inspection output for the values hash, overwrite to change what inspect displays.
# File lib/sequel/model/base.rb, line 2363 def inspect_values @values.inspect end
Raise an error appropriate to the hook type. May be swallowed by #checked_save_failure depending on the raise_on_failure? setting.
# File lib/sequel/model/base.rb, line 2377 def raise_hook_failure(type=nil) msg = case type when String type when Symbol "the #{type} hook failed" else "a hook failed" end raise HookFailed.new(msg, self) end
Whether to raise or return false if this action fails. If the :raise_on_failure option is present in the hash, use that, otherwise, fallback to the object's raise_on_save_failure (if set), or class's default (if not).
# File lib/sequel/model/base.rb, line 2371 def raise_on_failure?(opts) opts.fetch(:raise_on_failure, raise_on_save_failure) end
Get the ruby class or classes related to the given column's type.
# File lib/sequel/model/base.rb, line 2391 def schema_type_class(column) if (sch = db_schema[column]) && (type = sch[:type]) db.schema_type_class(type) end end
Call setter methods based on keys in hash, with the appropriate values. Restrict which methods can be called based on the provided type.
# File lib/sequel/model/base.rb, line 2399 def set_restricted(hash, type) return self if hash.empty? meths = setter_methods(type) strict = strict_param_setting hash.each do |k,v| m = "#{k}=" if meths.include?(m) set_column_value(m, v) elsif strict # Avoid using respond_to? or creating symbols from user input if public_methods.map(&:to_s).include?(m) if Array(model.primary_key).map(&:to_s).member?(k.to_s) && model.restrict_primary_key? raise MassAssignmentRestriction, "#{k} is a restricted primary key" else raise MassAssignmentRestriction, "#{k} is a restricted column" end else raise MassAssignmentRestriction, "method #{m} doesn't exist" end end end self end
Returns all methods that can be used for attribute assignment (those that end with =), depending on the type:
- :default
-
Use the default methods allowed in the model class.
- :all
-
Allow setting all setters, except those specifically restricted (such as ==).
- Array
-
Only allow setting of columns in the given array.
# File lib/sequel/model/base.rb, line 2429 def setter_methods(type) if type == :default if !@singleton_setter_added || model.allowed_columns return model.setter_methods end end if type.is_a?(Array) type.map{|x| "#{x}="} else meths = methods.collect(&:to_s).grep(SETTER_METHOD_REGEXP) - RESTRICTED_SETTER_METHODS meths -= Array(primary_key).map{|x| "#{x}="} if type != :all && primary_key && model.restrict_primary_key? meths end end
The server/shard that the model object's dataset uses, or :default if the model object's dataset does not have an associated shard.
# File lib/sequel/model/base.rb, line 2447 def this_server if (s = @server) s elsif (t = @this) t.opts[:server] || :default else model.dataset.opts[:server] || :default end end
Typecast the value to the column's type if typecasting. Calls the database's #typecast_value method, so database adapters can override/augment the handling for database specific column types.
# File lib/sequel/model/base.rb, line 2460 def typecast_value(column, value) return value unless typecast_on_assignment && db_schema && (col_schema = db_schema[column]) value = nil if '' == value and typecast_empty_string_to_nil and col_schema[:type] and ![:string, :blob].include?(col_schema[:type]) raise(InvalidValue, "nil/NULL is not allowed for the #{column} column") if raise_on_typecast_failure && value.nil? && (col_schema[:allow_null] == false) begin model.db.typecast_value(col_schema[:type], value) rescue InvalidValue raise_on_typecast_failure ? raise : value end end
Set the columns, filtered by the only and except arrays.
# File lib/sequel/model/base.rb, line 2472 def update_restricted(hash, type) set_restricted(hash, type) save_changes end
Set the given dataset to use the current object's shard.
# File lib/sequel/model/base.rb, line 2478 def use_server(ds) @server ? ds.server(@server) : ds end
Whether to use a transaction for this action. If the :transaction option is present in the hash, use that, otherwise, fallback to the object's default (if set), or class's default (if not).
# File lib/sequel/model/base.rb, line 2485 def use_transaction?(opts = OPTS) opts.fetch(:transaction, use_transactions) end