import { Connection } from './connection.js'
import { SourceMap } from './source-map.js'
// Check if the global.disableEventTarget is set to true, and if it is, don't
// use the EventTarget superclass. This can then be set by a client by loading
// the disable-event-target.js module before loading math-concept.js. The default
// for all other clients is to allow it. Note that we have to test if global is
// undeclared first, for when this module is being imported in a browser.
//
let Superclass =
((typeof global !== 'undefined' && global.disableEventTarget) ||
(typeof window !== 'undefined' && window.disableEventTarget))
? class { emit () { } }
: EventTarget
/**
* The MathConcept class, an n-ary tree of MathConcept instances, using functions
* like {@link MathConcept#parent parent()} and {@link MathConcept#children children()}
* to navigate the tree.
*
* In many ways, this is an abstract base class. Yes, you can construct instances of
* it, and the test suite does so to verify that all the features herein work. But in
* actual applications, it will primarily (perhaps exclusively) be subclasses of this
* class that are instantiated. There are two such types of subclasses:
*
* 1. LogicConcepts, which the LDE knows how to process for validating variable
* scoping and correctness of logical inference
* 2. MathConcepts that are not merely logical, and thus can be arbitrarily complex
* (such as a chain of equations, or a set of exercises), but which can be broken
* down into many LogicConcepts algorithmically, for processing by the LDE.
* This algorithmic breakdown is implemented in the
* {@link MathConcept#interpretation interpretation()} function, which in this
* abstract base class returns simply an empty list, meaning "no LogicConcepts."
*
* This second category of subclasses is not intended to be fully specified, but can
* grow and change over time, as new classes in that category are developed.
*/
export class MathConcept extends Superclass {
//////
//
// Constructor
//
//////
/**
* Create a new MathConcept. Any argument that is not a MathConcept is ignored.
* @constructor
* @param {...MathConcept} children - child MathConcepts to be added to this one
* (using {@link MathConcept#insertChild insertChild()})
*/
constructor ( ...children ) {
super()
this._dirty = false
this._parent = null
this._children = [ ]
this._attributes = new Map
for ( const child of children ) {
this.insertChild( child, this._children.length )
}
}
//////
//
// The dirty flag
//
//////
/**
* Getter for the "dirty" flag of this MathConcept. A MathConcept may be marked
* dirty by the client for any number of reasons. For instance, if a
* MathConcept changes and thus needs to be reprocessed (such as interpreted
* or validated) to reflect those most recent changes, it may be marked
* dirty until such processing takes place.
*
* MathConcept instances are constructed with their dirty flag set to false.
*
* @return {boolean} Whether this MathConcept is currently marked dirty
* @see {@link MathConcept#markDirty markDirty()}
*/
isDirty () { return this._dirty }
/**
* Setter for the "dirty" flag of this MathConcept. For information on the
* meaning of the flag, see {@link MathConcept#isDirty isDirty()}.
*
* @param {boolean} [on=true] Whether to mark it dirty (true)
* or clean (false). If this value is not boolean, it will be converted
* to one (with the `!!` idiom).
* @param {boolean} [propagate] Whether to propagate the change upwards to
* parent MathConcepts. By default, this happens if and only if the `on`
* member is true, so that dirtiness propagates upwards, but cleanness
* does not. This is appropriate because when a child needs reprocessing,
* this often requires reprocessing its parent as well, but when a child
* has been reprocessed, its parent may still need to be.
* @see {@link MathConcept#isDirty isDirty()}
*/
markDirty ( on = true, propagate ) {
this._dirty = !!on
if ( typeof( propagate ) == 'undefined' ) propagate = on
if ( propagate && this._parent ) this._parent.markDirty( on, propagate )
}
//////
//
// Functions about attributes
//
//////
/**
* Every MathConcept stores a dictionary of attributes as key-value pairs.
* All keys should be strings (or they will be converted into strings) and
* their associated values must be amenable to a JSON encoding.
*
* This function looks up and returns the value of an attribute in this
* MathConcept, the one with the given `key`.
*
* @param {*} key - name of the attribute to look up
* @param {*} defaultValue - the value that should be returned if the `key`
* does not appear as the name of an attribute in this MathConcept
* (defaults to undefined)
* @return {*} the value associated with the given `key`
* @see {@link MathConcept#setAttribute setAttribute()}
* @see {@link MathConcept#getAttributeKeys getAttributeKeys()}
*/
getAttribute ( key, defaultValue = undefined ) {
key = `${key}`
return this._attributes.has( key ) ? this._attributes.get( key )
: defaultValue
}
/**
* Get the list of keys used in the attributes dictionary within this
* MathConcept. For more details on the MathConcept attribution system, see the
* documentation for {@link MathConcept#getAttribute getAttribute()}.
*
* Each key must be atomic and will be converted into a string if it is not
* already one.
* @return {Array} A list of values used as keys
* @see {@link MathConcept#getAttribute getAttribute()}
*/
getAttributeKeys () { return Array.from( this._attributes.keys() ) }
/**
* Whether this MathConcept has an attribute with the given key. For more
* details on the MathConcept attribution system, see the documentation for
* {@link MathConcept#getAttribute getAttribute()}.
* @param {*} key - name of the attribute to look up; this should be atomic
* and will be converted into a string if it is not already one
* @see {@link MathConcept#getAttribute getAttribute()}
* @see {@link MathConcept#getAttributeKeys getAttributeKeys()}
*/
hasAttribute ( key ) {
key = `${key}`
return this._attributes.has( key )
}
/**
* For details on how MathConcepts store attributes, see the documentation for
* the {@link MathConcept#getAttribute getAttribute()} function.
*
* This function stores a new key-value pair in the MathConcept's attribute
* dictionary. See the restrictions on keys and values in the documentation
* linked to above. Calling this function overwrites any old value that was
* stored under the given `key`.
*
* The change events are fired only if the new value is different from the
* old value, according to `JSON.equals()`.
*
* @fires MathConcept#willBeChanged
* @fires MathConcept#wasChanged
* @param {*} key - The key that indexes the key-value pair we are about to
* insert or overwrite; this must be a string or will be converted into one
* @param {*} value - The value to associate with the given key; this must
* be a JavaScript value amenable to JSON encoding
* @see {@link MathConcept#attr attr()}
*/
setAttribute ( key, value ) {
key = `${key}`
const oldValue = this._attributes.get( key )
if ( !JSON.equals( value, oldValue ) ) {
/**
* An event of this type is fired in a MathConcept immediately before
* one of that MathConcept's attributes is changed.
*
* @event MathConcept#willBeChanged
* @type {Object}
* @property {MathConcept} concept - The MathConcept emitting the
* event, which will soon have one of its attributes changed
* @property {*} key - A string value, the key of the attribute
* that is about to change
* @property {*} oldValue - A JavaScript value amenable to JSON
* encoding, the value currently associated with the key; this is
* undefined if the value is being associated with an unused key
* @property {*} newValue - A JavaScript value amenable to JSON
* encoding, the value about to be associated with the key; this
* is undefined if the key-value pair is being removed rather than
* changed to have a new value
* @see {@link MathConcept#wasChanged wasChanged}
* @see {@link MathConcept#setAttribute setAttribute()}
*/
this.emit( 'willBeChanged', {
concept : this,
key : key,
oldValue : oldValue,
newValue : value
} )
this._attributes.set( key, value )
/**
* An event of this type is fired in a MathConcept immediately after
* one of that MathConcept's attributes is changed.
*
* @event MathConcept#wasChanged
* @type {Object}
* @property {MathConcept} concept - The MathConcept emitting the
* event, which just had one of its attributes changed
* @property {*} key - A string value, the key of the attribute
* that just changed
* @property {*} oldValue - A JavaScript value amenable to JSON
* encoding, the value formerly associated with the key; this is
* undefined if the value is being associated with an unused key
* @property {*} newValue - A JavaScript value amenable to JSON
* encoding, the value now associated with the key; this is
* undefined if the key-value pair is being removed rather than
* changed to have a new value
* @see {@link MathConcept#willBeChanged willBeChanged}
* @see {@link MathConcept#setAttribute setAttribute()}
*/
this.emit( 'wasChanged', {
concept : this,
key : key,
oldValue : oldValue,
newValue : value
} )
}
}
/**
* For details on how MathConcepts store attributes, see the documentation for
* the {@link MathConcept#getAttribute getAttribute()} function.
*
* This function removes zero or more key-value pairs from the MathConcept's
* attribute dictionary. See the restrictions on keys and values in the
* documentation linked to above.
*
* The change events are fired only if the given keys are actually currently
* in use by some key-value pairs in the MathConcept. If you pass multiple
* keys to be removed, each will generate a separate pair of
* {@link MathConcept#willBeChanged willBeChanged} and
* {@link MathConcept#wasChanged wasChanged} events.
*
* @fires MathConcept#willBeChanged
* @fires MathConcept#wasChanged
* @param {Array} keys - The list of keys indicating which key-value pairs
* should be removed from this MathConcept; each of these keys must be a
* string, or it will be converted into one; if this parameter is omitted,
* it defaults to all the keys for this MathConcept's attributes
* @see {@link MathConcept#getAttributeKeys getAttributeKeys()}
*/
clearAttributes ( ...keys ) {
if ( keys.length == 0 ) {
keys = this._attributes.keys()
}
for ( let key of keys ) {
key = `${key}`
if ( this._attributes.has( key ) ) {
const oldValue = this._attributes.get( key )
this.emit( 'willBeChanged', {
concept : this,
key : key,
oldValue : oldValue,
newValue : undefined
} )
this._attributes.delete( key )
this.emit( 'wasChanged', {
concept : this,
key : key,
oldValue : oldValue,
newValue : undefined
} )
}
}
}
/**
* Add attributes to a MathConcept and return the MathConcept. This function is
* a convenient form of repeated calls to
* {@link MathConcept#setAttribute setAttribute()}, and returns the MathConcept
* for ease of use in method chaining.
*
* Example use: `const S = new MathConcept().attr( { k1 : 'v1', k2 : 'v2' } )`
*
* Because this calls {@link MathConcept#setAttribute setAttribute()} zero or
* more times, as dictated by the contents of `attributes`, it may result in
* multiple firings of the events
* {@link MathConcept#willBeChanged willBeChanged} and
* {@link MathConcept#wasChanged wasChanged}.
*
* @param {Object|Map|Array} attributes - A collection of key-value pairs to
* add to this MathConcept's attributes. This can be a JavaScript Object,
* with keys and values in the usual `{'key':value,...}` form, a
* JavaScript `Map` object, or a JavaScript Array of key-value pairs, of
* the form `[['key',value],...]`. If this argument is not of any of
* these three forms (or is omitted), this function does not add any
* attributes to the MathConcept.
* @return {MathConcept} The MathConcept itself, for use in method chaining, as
* in the example shown above.
* @see {@link MathConcept#setAttribute setAttribute()}
*/
attr ( attributes = [ ] ) {
if ( attributes instanceof Array ) {
for ( const [ key, value ] of attributes ) {
this.setAttribute( key, value )
}
} else if ( attributes instanceof Map ) {
for ( const [ key, value ] of attributes ) {
this.setAttribute( key, value )
}
} else if ( attributes instanceof Object ) {
for ( const key of Object.keys( attributes ) ) {
this.setAttribute( key, attributes[key] )
}
}
return this
}
/**
* Copy all the attributes from another MathConcept instance to this one.
* The attributes are copied deeply, so that if the values are arrays or
* objects, they are not shared between the two MathConcepts. The
* attributes are copied using {@link MathConcept#attr attr()}, which calls
* {@link MathConcept#setAttribute setAttribute()} on each key separately,
* thus possibly generating many pairs of
* {@link MathConcept#willBeChanged willBeChanged} and
* {@link MathConcept#wasChanged wasChanged} events.
*
* If this MathConcept shares some attribute keys with the one passed as the
* parameter, the attributes of `mathConcept` will overwrite the attributes
* already in this object.
*
* @param {MathConcept} mathConcept - another MathConcept instance from
* which to copy all of its attributes
* @return {MathConcept} this object, for method chaining, as in
* {@link MathConcpet#attr attr()}
*/
copyAttributesFrom ( mathConcept ) {
return this.attr( mathConcept._attributes.deepCopy() )
}
/**
* Several functions internal to this object ({@link MathConcept#isA
* isA()}, {@link MathConcept#asA asA()}, and {@link MathConcept#makeIntoA
* makeIntoA()}) all take a type name as an argument, but do not use it
* directly as an attribute key, to avoid collisions among commonly used
* words. Rather, they use this function to slightly obfuscate the type
* name, thus making accidental name collisions less likely.
*
* To give a specific example, if we wanted to designate a symbol as, say,
* a number, we might not want to set its attribute `"number"` to true,
* because some other piece of code might have a different meaning/intent
* for the common word `"number"` and might overwrite or misread our data.
* So for saying that a MathConcept *is* a number (or any other type), we
* use this function, which turns the text `"number"` into the text
* `"_type_number"`, which almost no one would accidentally also use.
*
* @param {String} type - The type that will be stored/queried using the
* resulting key
* @returns {String} The key to use for querying the given `type`
*/
static typeAttributeKey ( type ) { return `_type_${type}` }
/**
* MathConcepts can be categorized into types with simple string labels.
* For instance, we might want to say that some MathConcepts are assumptions,
* and flag that using an attribute. Some of these attributes have meanings
* that may be respected by methods in this class or its subclasses, but the
* client is free to use any type names they wish. A MathConcept may have
* zero, one, or more types.
*
* This convenience function, together with
* {@link MathConcept#makeIntoA makeIntoA()} and {@link MathConcept#asA asA()},
* makes it easy to use the MathConcept's attributes to store such
* information.
*
* Note that the word "type" is being used in the informal, English sense,
* here. There is no intended or implied reference to mathematical types,
* variable types in programming languages, or type theory in general.
* This suite of functions is for adding boolean flags to MathConcepts in an
* easy way.
*
* @param {string} type - The type we wish to query
* @return {boolean} Whether this MathConcept has that type
* @see {@link MathConcept#makeIntoA makeIntoA()}
* @see {@link MathConcept#unmakeIntoA unmakeIntoA()}
* @see {@link MathConcept#asA asA()}
*/
isA ( type ) {
return this.getAttribute(
MathConcept.typeAttributeKey( type ) ) === true
}
/**
* For a full explanation of the typing features afforded by this function,
* see the documentation for {@link MathConcept#isA isA()}.
*
* This function adds the requested type to the MathConcept's attributes and
* returns the MathConcept itself, for use in method chaining, as in
* `S.makeIntoA( 'fruit' ).setAttribute( 'color', 'green' )`.
*
* @param {string} type - The type to add to this MathConcept
* @return {MathConcept} This MathConcept, after the change has been made to it
* @see {@link MathConcept#isA isA()}
* @see {@link MathConcept#asA asA()}
* @see {@link MathConcept#unmakeIntoA unmakeIntoA()}
*/
makeIntoA ( type ) {
this.setAttribute( MathConcept.typeAttributeKey( type ), true )
return this
}
/**
* For a full explanation of the typing features afforded by this function,
* see the documentation for {@link MathConcept#isA isA()}.
*
* This function removes the requested type to the MathConcept's attributes
* and returns the MathConcept itself, for use in method chaining, as in
* `S.unmakeIntoA( 'fruit' ).setAttribute( 'sad', true )`.
*
* Admittedly, this is a pretty bad name for a function, but it is the
* reverse of {@link MathConcept#makeIntoA makeIntoA()}, so there you go.
*
* @param {string} type - The type to remove from this MathConcept
* @return {MathConcept} This MathConcept, after the change has been made to it
* @see {@link MathConcept#isA isA()}
* @see {@link MathConcept#asA asA()}
* @see {@link MathConcept#makeIntoA makeIntoA()}
*/
unmakeIntoA ( type ) {
this.clearAttributes( MathConcept.typeAttributeKey( type ) )
return this
}
/**
* Create a copy of this MathConcept, but with the given type added, using
* {@link MathConcept#makeIntoA makeIntoA()}.
*
* @param {string} type - The type to add to the copy
* @return {MathConcept} A copy of this MathConcept, with the given type added
* @see {@link MathConcept#isA isA()}
* @see {@link MathConcept#makeIntoA makeIntoA()}
*/
asA ( type ) { return this.copy().makeIntoA( type ) }
//////
//
// Functions querying tree structure
//
//////
/**
* This MathConcept's parent MathConcept, that is, the one enclosing it, if any
* @return {MathConcept} This MathConcept's parent node, or null if there isn't one
* @see {@link MathConcept#children children()}
* @see {@link MathConcept#child child()}
*/
parent () { return this._parent }
/**
* An array containing this MathConcept's children, in the correct order.
*
* To get a specific child, it is more efficient to use the
* {@link MathConcept.child()} function instead.
*
* @return {MathConcept[]} A shallow copy of the MathConcept's children array
* @see {@link MathConcept#parent parent()}
* @see {@link MathConcept#child child()}
* @see {@link MathConcept#setChildren setChildren()}
* @see {@link MathConcept#allButFirstChild allButFirstChild()}
* @see {@link MathConcept#allButLastChild allButLastChild()}
* @see {@link MathConcept#childrenSatisfying childrenSatisfying()}
*/
children () { return this._children.slice() }
/**
* Get the child of this MathConcept at index i.
*
* If the index is invalid (that is, it is anything other than one of
* {0,1,...,n-1\} if there are n children) then undefined will be
* returned instead.
*
* @param {number} i - The index of the child being fetched
* @return {MathConcept} The child at the given index, or undefined if none
* @see {@link MathConcept#parent parent()}
* @see {@link MathConcept#children children()}
* @see {@link MathConcept#firstChild firstChild()}
* @see {@link MathConcept#lastChild lastChild()}
*/
child ( ...indices ) {
return indices.reduce( (x,n) =>
typeof(x)=='undefined' ? undefined :
x.children()[n < 0 ? x.children().length + n : n],this)
}
/**
* The number of children of this MathConcept
* @return {number} A nonnegative integer indicating the number of children
* @see {@link MathConcept#children children()}
* @see {@link MathConcept#child child()}
*/
numChildren () { return this._children.length }
/**
* Returns the value `i` such that `this.parent().child(i)` is this object,
* provided that this MathConcept has a parent.
*
* @return {number} The index of this MathConcept in its parent's children list
* @see {@link MathConcept#parent parent()}
* @see {@link MathConcept#child child()}
*/
indexInParent () {
if ( this._parent != null && this._parent._children ) {
return this._parent._children.indexOf( this )
}
}
/**
* Find the previous sibling of this MathConcept in its parent, if any
* @return {MathConcept} The previous sibling, or undefined if there is none
* @see {@link MathConcept#children children()}
* @see {@link MathConcept#nextSibling nextSibling()}
*/
previousSibling () {
let index = this.indexInParent()
if ( index != null) {
return this._parent._children[index-1]
}
}
/**
* Find the next sibling of this MathConcept in its parent, if any
* @return {MathConcept} The next sibling, or undefined if there is none
* @see {@link MathConcept#children children()}
* @see {@link MathConcept#previousSibling previousSibling()}
*/
nextSibling () {
let index = this.indexInParent()
if ( index != null ) {
return this._parent._children[index+1]
}
}
/**
* A MathConcept is atomic if and only if it has no children. Thus this is a
* shorthand for `S.numChildren() == 0`.
* @return {boolean} Whether the number of children is zero
* @see {@link MathConcept#numChildren numChildren()}
*/
isAtomic () { return this.numChildren() == 0 }
/**
* Convenience function for fetching just the first child of this MathConcept
* @return {MathConcept} The first child of this MathConcept, or undefined if none
* @see {@link MathConcept#lastChild lastChild()}
* @see {@link MathConcept#allButFirstChild allButFirstChild()}
*/
firstChild () { return this._children[0] }
/**
* Convenience function for fetching just the last child of this MathConcept
* @return {MathConcept} The last child of this MathConcept, or undefined if none
* @see {@link MathConcept#firstChild firstChild()}
* @see {@link MathConcept#allButLastChild allButLastChild()}
*/
lastChild () { return this._children.last() }
/**
* Convenience function for fetching the array containing all children of
* this MathConcept except for the first
* @return {MathConcept[]} All but the first child of this MathConcept, or an
* empty array if there is one or fewer children
* @see {@link MathConcept#firstChild firstChild()}
* @see {@link MathConcept#allButLastChild allButLastChild()}
*/
allButFirstChild () { return this._children.slice( 1 ) }
/**
* Convenience function for fetching the array containing all children of
* this MathConcept except for the last
* @return {MathConcept[]} All but the last child of this MathConcept, or an
* empty array if there is one or fewer children
* @see {@link MathConcept#lastChild lastChild()}
* @see {@link MathConcept#allButFirstChild allButFirstChild()}
*/
allButLastChild () { return this._children.slice( 0, -1 ) }
/**
* My address within the given ancestor, as a sequence of indices
* `[i1,i2,...,in]` such that `ancestor.child(i1).child(i2)....child(in)` is
* this MathConcept.
*
* This is a kind of inverse to {@link MathConcept#index index()}.
*
* @param {MathConcept} [ancestor] - The ancestor in which to compute my
* address, which defaults to my highest ancestor. If this argument is
* not actually an ancestor of this MathConcept, then we treat it as if it
* had been omitted.
* @return {number[]} An array of numbers as described above, which will be
* empty in the degenerate case where this MathConcept has no parent or this
* MathConcept is the given ancestor
* @see {@link MathConcept#child child()}
* @see {@link MathConcept#indexInParent indexInParent()}
*/
address ( ancestor ) {
if ( ancestor === this || !this.parent() ) return [ ]
const lastStep = this.indexInParent()
return this.parent().address( ancestor ).concat( [ lastStep ] )
}
/**
* Performs repeated child indexing to find a specific descendant. If the
* address given as input is the array `[i1,i2,...,in]`, then this returns
* `this.child(i1).child(i2)....child(in)`.
*
* If the given address is the empty array, the result is this MathConcept.
*
* This is a kind of inverse to {@link MathConcept#address address()}.
*
* @param {number[]} address - A sequence of nonnegative indices, as
* described in the documentation for address()
* @return {MathConcept} A descendant MathConcept, following the definition
* above, or undefined if there is no such MathConcept
* @see {@link MathConcept#child child()}
* @see {@link MathConcept#address address()}
*/
index ( address ) {
if ( !( address instanceof Array ) ) return undefined
if ( address.length == 0 ) return this
const nextStep = this.child( address[0] )
if ( !( nextStep instanceof MathConcept ) ) return undefined
return nextStep.index( address.slice( 1 ) )
}
//////
//
// Advanced queries, including predicates and iterators
//
//////
/**
* The list of children of this MathConcept that satisfy the given predicate,
* in the same order that they appear as children. Obviously, not all
* children may be included in the result, depending on the predicate.
*
* @param {function(MathConcept):boolean} predicate - The predicate to use for
* testing children
* @return {MathConcept[]} The array of children satisfying the given predicate
* @see {@link MathConcept#children children()}
* @see {@link MathConcept#descendantsSatisfying descendantsSatisfying()}
* @see {@link MathConcept#hasChildSatisfying hasChildSatisfying()}
*/
childrenSatisfying ( predicate ) { return this._children.filter( predicate ) }
/**
* Whether this MathConcept has any children satisfying the given predicate.
* The predicate will be evaluated on each child in order until one passes
* or all fail; it may not be evaluated on all children, if not needed.
*
* @param {function(MathConcept):boolean} predicate - The predicate to use for
* testing children
* @return {boolean} True if and only if some child satisfies the given predicate
* @see {@link MathConcept#hasDescendantSatisfying hasDescendantSatisfying()}
* @see {@link MathConcept#childrenSatisfying childrenSatisfying()}
*/
hasChildSatisfying ( predicate ) { return this._children.some( predicate ) }
/**
* An iterator over all descendants of this MathConcept, in a pre-order tree
* traversal.
*
* This function is a generator that yields this MathConcept, then its first
* child, and so on down that branch of the tree, and onward in a pre-order
* traversal.
*
* @see {@link MathConcept#descendantsSatisfying descendantsSatisfying()}
* @see {@link MathConcept#hasDescendantSatisfying hasDescendantSatisfying()}
*/
*descendantsIterator () {
yield this
for ( let child of this._children ) yield* child.descendantsIterator()
}
/**
* An array of those descendants of this MathConcept that satisfy the given
* predicate. These are not copies, but the actual descendants; if you
* alter one, it changes the hierarchy beneath this MathConcept.
*
* Note that this MathConcept counts as a descendant of itself. To exclude
* this MathConcept from consideration, simply change your predicate, as in
* `X.descendantsSatisfying( d => X != d && predicate(d) )`.
*
* @param {function(MathConcept):boolean} predicate - The predicate to use for
* testing descendants
* @return {MathConcept[]} A list of descendants of this MathConcept, precisely
* those that satisfy the given predicate, listed in the order they would
* be visited in a depth-first traversal of the tree
* @see {@link MathConcept#hasDescendantSatisfying hasDescendantSatisfying()}
* @see {@link MathConcept#ancestorsSatisfying ancestorsSatisfying()}
* @see {@link MathConcept#childrenSatisfying childrenSatisfying()}
*/
descendantsSatisfying ( predicate ) {
let result = [ ]
for ( let descendant of this.descendantsIterator() )
if ( predicate( descendant ) ) result.push( descendant )
return result
}
/**
* Whether this MathConcept has any descendant satisfying the given predicate.
* The predicate will be evaluated on each descendant in depth-first order
* until one passes or all fail; it may not be evaluated on all descendants,
* if not needed.
*
* Note that this MathConcept counts as a descendant of itself. To ignore
* this MathConcept, simply change the predicate to do so, as in
* `X.descendantsSatisfying( d => X != d && predicate(d) )`.
*
* @param {function(MathConcept):boolean} predicate - The predicate to use for
* testing descendants
* @return {boolean} True if and only if some descendant satisfies the given predicate
* @see {@link MathConcept#hasChildSatisfying hasChildSatisfying()}
* @see {@link MathConcept#descendantsSatisfying descendantsSatisfying()}
* @see {@link MathConcept#hasAncestorSatisfying hasAncestorSatisfying()}
*/
hasDescendantSatisfying ( predicate ) {
for ( let descendant of this.descendantsIterator() )
if ( predicate( descendant ) ) return true
return false
}
/**
* An iterator through all the ancestors of this MathConcept, starting with
* itself as the first (trivial) ancestor, and walking upwards from there.
*
* This function is a generator that yields this MathConcept, then its
* parent, grandparent, etc.
*
* @see {@link MathConcept#ancestors ancestors()}
* @see {@link MathConcept#parent parent()}
*/
*ancestorsIterator () {
yield this
if ( this.parent() ) yield* this.parent().ancestorsIterator()
}
/**
* An array of all ancestors of this MathConcept, starting with itself. This
* array is the exact contents of
* {@link MathConcept#ancestorsIterator ancestorsIterator()}, but in array
* form rather than as an iterator.
*
* @return {MathConcept[]} An array beginning with this MathConcept, then its
* parent, grandparent, etc.
* @see {@link MathConcept#ancestorsIterator ancestorsIterator()}
* @see {@link MathConcept#parent parent()}
*/
ancestors () { return Array.from( this.ancestorsIterator() ) }
/**
* Find all ancestors of this MathConcept satisfying the given predicate.
* Note that this MathConcept counts as a trivial ancestor of itself, so if
* you don't want that, modify your predicate to exclude it.
*
* @param {function(MathConcept):boolean} predicate - Predicate to evaluate on
* each ancestor
* @return {MathConcept[]} The ancestors satisfying the predicate, which may
* be an empty array
* @see {@link MathConcept#ancestorsIterator ancestorsIterator()}
* @see {@link MathConcept#hasAncestorSatisfying hasAncestorSatisfying()}
* @see {@link MathConcept#descendantsSatisfying descendantsSatisfying()}
*/
ancestorsSatisfying ( predicate ) {
const result = [ ]
for ( let ancestor of this.ancestorsIterator() )
if ( predicate( ancestor ) ) result.push( ancestor )
return result
}
/**
* Whether this MathConcept has an ancestor (including itself) satisfying the
* given predicate.
*
* @param {function(MathConcept):boolean} predicate - Predicate to evaluate on
* each ancestor
* @return {boolean} Whether an ancestor satisfying the given predicate
* exists
* @see {@link MathConcept#ancestorsIterator ancestorsIterator()}
* @see {@link MathConcept#ancestorsSatisfying ancestorsSatisfying()}
* @see {@link MathConcept#hasDescendantSatisfying hasDescendantSatisfying()}
*/
hasAncestorSatisfying ( predicate ) {
for ( let ancestor of this.ancestorsIterator() )
if ( predicate( ancestor ) ) return true
return false
}
//////
//
// Functions altering tree structure
//
//////
/**
* Insert a child into this MathConcept's list of children.
*
* Any children at the given index or later will be moved one index later to
* make room for the new insertion. The index can be anything from 0 to the
* number of children (inclusive); this last value means insert at the end
* of the children array. The default insertion index is the beginning of
* the array.
*
* If the child to be inserted is an ancestor of this MathConcept, then we
* remove this MathConcept from its parent, to obey the insertion command given
* while still maintaining acyclicity in the tree structure. If the child to
* be inserted is this node itself, this function does nothing.
*
* @param {MathConcept} child - the child to insert
* @param {number} atIndex - the index at which the new child will be
* @fires MathConcept#willBeInserted
* @fires MathConcept#wasInserted
* @see {@link MathConcept#children children()}
* @see {@link MathConcept#setChildren setChildren()}
* @see {@link MathConcept#child child()}
* @see {@link MathConcept#pushChild pushChild()}
* @see {@link MathConcept#unshiftChild unshiftChild()}
*/
insertChild ( child, atIndex = 0 ) {
if ( !( child instanceof MathConcept ) ) return
if ( child === this ) return
if ( atIndex < 0 || atIndex > this._children.length ) return
let walk = this
while ( ( walk = walk.parent() ) != null ) {
if ( walk === child ) {
this.remove();
break;
}
}
child.remove()
/**
* An event of this type is fired in a MathConcept immediately before that
* MathConcept is inserted as a child within a new parent.
*
* @event MathConcept#willBeInserted
* @type {Object}
* @property {MathConcept} child - The MathConcept emitting the event, which
* will soon be a child of a new parent MathConcept
* @property {MathConcept} parent - The new parent the child will have
* after insertion
* @property {number} index - The new index the child will have after
* insertion
* @see {@link MathConcept#wasInserted wasInserted}
* @see {@link MathConcept#insertChild insertChild()}
*/
child.emit( 'willBeInserted', {
child : child,
parent : this,
index : atIndex
} )
this._children.splice( atIndex, 0, child )
child._parent = this
/**
* An event of this type is fired in a MathConcept immediately after that
* MathConcept is inserted as a child within a new parent.
*
* @event MathConcept#wasInserted
* @type {Object}
* @property {MathConcept} child - The MathConcept emitting the event, which
* just became a child of a new parent MathConcept
* @property {MathConcept} parent - The new parent the child now has
* @property {number} index - The index the child now has in its new
* parent
* @see {@link MathConcept#willBeInserted willBeInserted}
* @see {@link MathConcept#insertChild insertChild()}
*/
child.emit( 'wasInserted', {
child : child,
parent : this,
index : atIndex
} )
}
/**
* If this MathConcept has a parent, remove this from its parent's child list
* and set our parent pointer to null, thus severing the relationship. If
* this has no parent, do nothing.
*
* @fires MathConcept#willBeRemoved
* @see {@link MathConcept#parent parent()}
* @see {@link MathConcept#removeChild removeChild()}
*/
remove () {
if ( this._parent != null ) {
const parent = this._parent
const index = this.indexInParent()
/**
* This event is fired in a MathConcept immediately before that
* MathConcept is removed from its parent MathConcept. This could be
* from a simple removal, or it might be the first step in a
* re-parenting process that ends up with the MathConcept as the child
* of a new parent.
*
* @event MathConcept#willBeRemoved
* @type {Object}
* @property {MathConcept} child - The MathConcept emitting the event,
* which is about to be removed from its parent MathConcept
* @property {MathConcept} parent - The current parent MathConcept
* @property {number} index - The index the child has in its parent,
* before the removal
* @see {@link MathConcept#remove remove()}
*/
this.emit( 'willBeRemoved', {
child : this,
parent : parent,
index : index
} )
this._parent._children.splice( this.indexInParent(), 1 )
this._parent = null
/**
* This event is fired in a MathConcept immediately after that
* MathConcept is removed from its parent MathConcept. This could be
* from a simple removal, or it might be the first step in a
* re-parenting process that ends up with the MathConcept as the child
* of a new parent.
*
* @event MathConcept#wasRemoved
* @type {Object}
* @property {MathConcept} child - The MathConcept emitting the event,
* which was just removed from its parent MathConcept
* @property {MathConcept} parent - The old parent MathConcept from
* which the child was just removed
* @property {number} index - The index the child had in its parent,
* before the removal
* @see {@link MathConcept#remove remove()}
*/
this.emit( 'wasRemoved', {
child : this,
parent : parent,
index : index
} )
}
}
/**
* Calls {@link MathConcept#remove remove()} on the child with index `i`.
* Does nothing if the index is invalid.
*
* @param {number} i - the index of the child to remove
* @see {@link MathConcept#remove remove()}
* @see {@link MathConcept#child child()}
* @see {@link MathConcept#popChild popChild()}
* @see {@link MathConcept#shiftChild shiftChild()}
*/
removeChild ( i ) {
if ( i < 0 || i >= this._children.length ) return
this._children[i].remove()
}
/**
* Replace this MathConcept, exactly where it sits in its parent MathConcept,
* with the given one, thus deparenting this one.
*
* For example, if `A` is a child of `B` and we call `B.replaceWith(C)`,
* then `C` will now be a child of `A` at the same index that `B` formerly
* occupied, and `B` will now have no parent. If `C` had a parent before,
* it will have been removed from it (thus decreasing that parent's number
* of children by one).
*
* @param {MathConcept} other - the MathConcept with which to replace this one
* @see {@link MathConcept#remove remove()}
* @see {@link MathConcept#child child()}
* @see {@link MathConcept#parent parent()}
*/
replaceWith ( other ) {
let originalParent = this._parent;
if ( originalParent != null ) {
const originalIndex = this.indexInParent()
this.remove()
originalParent.insertChild( other, originalIndex )
}
}
/**
* Remove the last child of this MathConcept and return it. If there is no
* such child, take no action and return undefined.
* @return {MathConcept} The popped last child, or undefined if none
* @see {@link MathConcept#pushChild pushChild()}
* @see {@link MathConcept#shiftChild shiftChild()}
*/
popChild () {
const child = this.lastChild()
if ( !child ) return
child.remove()
return child
}
/**
* Remove the first child of this MathConcept and return it. If there is no
* such child, take no action and return undefined.
* @return {MathConcept} The popped first child, or undefined if none
* @see {@link MathConcept#popChild popChild()}
* @see {@link MathConcept#unshiftChild unshiftChild()}
*/
shiftChild () {
const child = this.firstChild()
if ( !child ) return
child.remove()
return child
}
/**
* Append a new child to the end of this MathConcept's list of children. This
* is equivalent to a call to `insertChild()` with the length of the current
* children array as the index at which to insert.
*
* @param {MathConcept} child - The new MathConcept to append
* @see {@link MathConcept#popChild popChild()}
* @see {@link MathConcept#unshiftChild unshiftChild()}
*/
pushChild ( child ) { this.insertChild( child, this._children.length ) }
/**
* Prepend a new child to the beginning of this MathConcept's list of children.
* This is equivalent to a call to `insertChild()` with the default second
* parameter (i.e., insert at index zero), and thus this function is here
* only for convenience, to fit with shiftChild().
*
* @param {MathConcept} child - The new MathConcept to prepend
* @see {@link MathConcept#shiftChild shiftChild()}
* @see {@link MathConcept#pushChild pushChild()}
*/
unshiftChild ( child ) { this.insertChild( child ) }
/**
* Replace the entire children array of this MathConcept with a new one.
*
* This is equivalent to removing all the current children of this MathConcept
* in order from lowest index to highest, then inserting all the children in
* the given array, again from lowest index to highest.
*
* The intent is not for any of the elements of the given array to be
* ancestors or descendants of one another, but even if they are, the action
* taken here still follows the explanation given in the previous paragraph.
*
* @param {MathConcept[]} children - New list of children
* @see {@link MathConcept#children children()}
* @see {@link MathConcept#removeChild removeChild()}
* @see {@link MathConcept#insertChild insertChild()}
*/
setChildren ( children ) {
while ( this._children.length > 0 ) {
this.firstChild().remove()
}
for ( const child of children ) {
this.pushChild( child )
}
}
//////
//
// Order relations and traversals
//
//////
/**
* Under pre-order tree traversal, which of two MathConcept comes first? We
* call the first "earlier than" the other MathConcept, because we will use
* MathConcept hierarchies to represent documents, and first in a pre-order
* tree traversal would then mean earlier in the document.
*
* Note that this is a strict ordering, so a MathConcept is not earlier than
* itself.
*
* @param {MathConcept} other - The MathConcept with which to compare this one.
* (The result is undefined if this is not a MathConcept.)
* @return {boolean} Whether this MathConcept is earlier than the other, or
* undefined if they are incomparable (not in the same tree)
* @see {@link MathConcept#isLaterThan isLaterThan()}
* @see {@link MathConcept#preOrderTraversal preOrderTraversal()}
* @see {@link MathConcept#nextInTree nextInTree()}
* @see {@link MathConcept#previousInTree previousInTree()}
*/
isEarlierThan ( other ) {
// type check
if ( !( other instanceof MathConcept ) ) return undefined
// base case
if( other === this ) return false
// we will need to compare ancestors
const myAncestors = this.ancestors().reverse()
const otherAncestors = other.ancestors().reverse()
// if we have no common ancestor, we are incomparable
if ( otherAncestors[0] != myAncestors[0] ) return undefined
// we have a common top-level ancestor; find our least common ancestor
let lowest = null
while ( myAncestors[0] == otherAncestors[0] ) {
myAncestors.shift()
lowest = otherAncestors.shift()
}
// if either of us is an ancestor of the other, then that one is earlier
if ( lowest === this ) return true
if ( lowest === other ) return false
// otherwise, compare child indices within the common ancestor
return myAncestors[0].indexInParent()
< otherAncestors[0].indexInParent()
}
/**
* This is the opposite of {@link MathConcept#isEarlierThan isEarlierThan()}.
* We have `A.isLaterThan(B)` if and only if `B.isEarlierThan(A)`. This is
* therefore just a convenience function.
*
* @param {MathConcept} other - The MathConcept with which to compare this one.
* (The result is undefined if this is not a MathConcept.)
* @return {boolean} Whether this MathConcept is later than the other, or
* undefined if they are incomparable (not in the same tree)
* @see {@link MathConcept#isEarlierThan isEarlierThan()}
* @see {@link MathConcept#preOrderTraversal preOrderTraversal()}
* @see {@link MathConcept#nextInTree nextInTree()}
* @see {@link MathConcept#previousInTree previousInTree()}
*/
isLaterThan ( other ) {
if ( !( other instanceof MathConcept ) ) return undefined
return other.isEarlierThan( this )
}
/**
* Finds the next node in the same tree as this one, where "next" is defined
* in terms of a pre-order tree traversal. If there is no such node, this
* will return undefined.
*
* Therefore this function also returns the earliest node later than this
* one, in the sense of {@link MathConcept#isEarlierThan isEarlierThan()} and
* {@link MathConcept#isLaterThan isLaterThan()}.
*
* For example, in a parent node with several atomic children, the next node
* of the parent is the first child, and the next node of each child is the
* one after, but the last child has no next node.
*
* @return {MathConcept} The next node in pre-order traversal after this one
* @see {@link MathConcept#isEarlierThan isEarlierThan()}
* @see {@link MathConcept#isLaterThan isLaterThan()}
* @see {@link MathConcept#preOrderTraversal preOrderTraversal()}
* @see {@link MathConcept#previousInTree previousInTree()}
*/
nextInTree () {
// if I have a first child, that's my next node.
if ( this._children.length > 0 )
return this._children[0]
// if I have a next sibling, that's my next node.
// otherwise, use my parent's next sibling, or my grandparent's, ...
for ( let ancestor of this.ancestorsIterator() ) {
if ( ancestor.nextSibling() ) {
return ancestor.nextSibling()
}
}
// no nodes after me, so return undefined
}
/**
* Finds the previous node in the same tree as this one, where "previous" is
* defined in terms of a pre-order tree traversal. If there is no such
* node, this will return undefined.
*
* Therefore this function also returns the latest node earlierr than this
* one, in the sense of {@link MathConcept#isEarlierThan isEarlierThan()} and
* {@link MathConcept#isLaterThan isLaterThan()}.
*
* This is the reverse of {@link MathConcept#nextInTree nextInTree()}, in the
* sense that `X.nextInTree().previousInTree()` and
* `X.previousInTree().nextInTree()` will, in general, be `X`, unless one of
* the computations involved is undefined.
*
* @return {MathConcept} The previous node in pre-order traversal before this
* one
* @see {@link MathConcept#nextInTree nextInTree()}
* @see {@link MathConcept#isEarlierThan isEarlierThan()}
* @see {@link MathConcept#isLaterThan isLaterThan()}
* @see {@link MathConcept#preOrderTraversal preOrderTraversal()}
*/
previousInTree () {
// if I have a previous sibling, then its latest descendant is my
// previous node
let beforeMe = this.previousSibling()
while ( beforeMe && beforeMe._children.length > 0 ) {
beforeMe = beforeMe.lastChild()
}
if ( beforeMe ) return beforeMe
// otherwise, my previous node is my parent (which may be null if
// I'm the earliest node in my tree, which we convert to undefined)
return this._parent || undefined
}
/**
* An iterator that walks through the entire tree from this node onward, in
* a pre-order tree traversal, yielding each node in turn.
*
* This function is a generator that yields the next node after this one in
* pre-order tree traversal, just as {@link MathConcept#nextInTree
* nextInTree()} would yield, then the next after that, and so on.
*
* @param {boolean} inThisTreeOnly - Set this to true to limit the iterator
* to return only descendants of this MathConcept. Set it to false to
* permit the iterator to proceed outside of this tree into its context,
* once all nodes within this tree have been exhausted. If this MathConcept
* has no parent, then this parameter is irrelevant.
*
* @see {@link MathConcept#nextInTree nextInTree()}
* @see {@link MathConcept#isEarlierThan isEarlierThan()}
* @see {@link MathConcept#isLaterThan isLaterThan()}
* @see {@link MathConcept#preOrderTraversal preOrderTraversal()}
*/
*preOrderIterator ( inThisTreeOnly = true ) {
// compute the last descendant of this tree (or undefined if they did
// not limit us to traversing only this subtree)
let stopHere = inThisTreeOnly ? this : undefined
while ( stopHere && stopHere._children.length > 0 ) {
stopHere = stopHere.lastChild()
}
// now iterate over all the nexts (stopping only if we encounter the
// final descendant computed above, if any)
let nextResult = this
while ( nextResult ) {
yield nextResult
if ( nextResult === stopHere ) break
nextResult = nextResult.nextInTree()
}
}
/**
* The same as {@link MathConcept#preOrderIterator preOrderIterator()}, but
* already computed into array form for convenience (usually at a cost of
* efficiency).
*
* @param {boolean} inThisTreeOnly - Has the same meaning as it does in
* {@link MathConcept#preOrderIterator preOrderIterator()}
* @return {MathConcept[]} The array containing a pre-order tree traversal
* starting with this node, beginning with
* {@link MathConcept#nextInTree nextInTree()}, then the next after that,
* and so on.
* @see {@link MathConcept#preOrderIterator preOrderIterator()}
* @see {@link MathConcept#nextInTree nextInTree()}
*/
preOrderTraversal ( inThisTreeOnly = true ) {
return Array.from( this.preOrderIterator( inThisTreeOnly ) )
}
/**
* In computer programming, the notion of variable scope is common. A line
* of code can "see" a variable (or is in the scope of that variable) if it
* appears later than the variable's declaration and at a deeper level of
* block nesting. We have the same concept within MathConcepts, and we call
* it both "scope" and "accessibility." We say that any later MathConcept is
* "in the scope of" an earlier one, or equivalently, the earlier one "is
* accessible to" the later one, if the nesting of intermediate MathConcepts
* permits it in the usual way.
*
* More specifically, a MathConcept `X` is in the scope of precisely the
* following other MathConcepts: all of `X`'s previous siblings, all of
* `X.parent()`'s previous siblings (if `X.parent()` exists), all of
* `X.parent().parent()`'s previous siblings (if `X.parent().parent()`
* exists), and so on. In particular, a MathConcept is not in its own scope,
* nor in the scope of any of its other ancestors.
*
* The one exception to what's stated above is the reflexive case, whether
* `X.isAccessibleTo(X)`. By default, this is false, because we typically
* think of `X.isAccessibleTo(Y)` as answering the question, "Can `Y`
* justify itself by citing `X`?" and we do not wish that relation to be
* reflexive. However, `X.isInTheScopeOf(X)` would typically be considered
* true, because a variable declaration is the beginning of the scope of
* that variable. So we provide the second parameter, `reflexive`, for
* customizing this behavior, and we have that, for any boolean value `b`,
* `X.isAccessibleTo(Y,b)` if and only if `Y.isInTheScopeOf(X,b)`.
*
* @param {MathConcept} other - The MathConcept to which we're asking whether
* the current one is accessible. If this parameter is not a MathConcept,
* the result is undefined.
* @param {boolean} reflexive - Whether the relation should be reflexive,
* that is, whether it should judge `X.isAccessibleTo(X)` to be true.
* @return {boolean} Whether this MathConcept is accessible to `other`.
* @see {@link MathConcept#isInTheScopeOf isInTheScopeOf()}
*/
isAccessibleTo ( other, reflexive = false ) {
if ( this === other ) return reflexive
if ( !( other instanceof MathConcept ) ) return undefined
if ( other.parent() === null ) return false
if ( this.parent() === other.parent() ) {
return this.indexInParent() < other.indexInParent()
}
return this.isAccessibleTo( other.parent() )
}
/**
* A full definition of both
* {@link MathConcept#isAccessibleTo isAccessibleTo()} and
* {@link MathConcept#isInTheScopeOf isInTheScopeOf()} appears in the
* documentation for {@link MathConcept#isAccessibleTo isAccessibleTo()}.
* Refer there for details.
*
* @param {MathConcept} other - The MathConcept in whose scope we're asking
* whether this one lies. If this parameter is not a MathConcept, the
* result is undefined.
* @param {boolean} reflexive - Whether the relation should be reflexive,
* that is, whether it should judge `X.isInTheScopeOf(X)` to be true.
* @return {boolean} Whether this MathConcept is in the scope of `other`.
* @see {@link MathConcept#isAccessibleTo isAccessibleTo()}
*/
isInTheScopeOf ( other, reflexive = true ) {
if ( !( other instanceof MathConcept ) ) return undefined
return other.isAccessibleTo( this, reflexive )
}
/**
* For a definition of accessibility, refer to the documentation for the
* {@link MathConcept#isAccessibleTo isAccessibleTo()} function.
*
* In short, the accessibles of a node are its previous siblings, the
* previous siblings of its parent, the previous siblings of its
* grandparent, and so on, where each node yielded
* {@link MathConcept#isLaterThan isLaterThan()} all nodes yielded thereafter.
* You can limit the list to only those accessibles within a given ancestor
* by using the `inThis` parameter, documented below.
*
* This function is a generator that yields each MathConcept accessible to
* this one, beginning with the one closest to this one (often its previous
* sibling) and proceeding back through the hierarchy, so that each new
* result is accessible to (and earlier than) the previous).
*
* @param {boolean} reflexive - Functions analogously to the `reflexive`
* parameter for {@link MathConcept#isAccessibleTo isAccessibleTo()}; that
* is, do we include this MathConcept on its list of accessibles? The
* default value is false.
* @param {MathConcept} inThis - The container MathConcept in which to list
* accessibles. No accessible outside this ancestor will be returned.
* (If this is not actually an ancestor, it is ignored, and all accessibles
* are returned, which is the default.)
* @see {@link MathConcept#isAccessibleTo isAccessibleTo()}
* @see {@link MathConcept#accessibles accessibles()}
*/
*accessiblesIterator ( reflexive = false, inThis = null ) {
// return myself if reflexive, unless I'm the limiting ancestor
if ( inThis == this ) return
if ( reflexive ) yield this
// yield all previous siblings of myself
for ( let previous = this.previousSibling() ; previous ;
previous = previous.previousSibling() )
yield previous
// if there is no parent, or we are not allowed to use it, we're done
if ( !this._parent || this._parent == inThis ) return
// otherwise, recur on the parent
yield* this._parent.accessiblesIterator( false, inThis )
}
/**
* The full contents of
* {@link MathConcept#accessiblesIterator accessiblesIterator()}, but put into
* an array rather than an iterator, for convenience, possibly at the cost
* of efficiency.
*
* @param {boolean} reflexive - Passed directly to
* {@link MathConcept#accessiblesIterator accessiblesIterator()}; see that
* function for more information
* @param {MathConcept} inThis - Passed directly to
* {@link MathConcept#accessiblesIterator accessiblesIterator()}; see that
* function for more information
* @return {MathConcept[]} All MathConcepts accessible to this one, with the
* latest (closest to this MathConcept) first, proceeding on to the earliest
* at the end of the array
* @see {@link MathConcept#accessiblesIterator accessiblesIterator()}
* @see {@link MathConcept#isAccessibleTo isAccessibleTo()}
*/
accessibles ( reflexive = false, inThis = null ) {
return Array.from( this.accessiblesIterator( reflexive, inThis ) )
}
/**
* For a definition of scope, refer to the documentation for the
* {@link MathConcept#isAccessibleTo isAccessibleTo()} function.
*
* In short, the scope of a node is itself, all of its later siblings, and
* all their descendants, where each node yielded by the iterator
* {@link MathConcept#isEarlierThan isEarlierThan()} all nodes yielded
* thereafter.
*
* This function is a generator that yields each MathConcept in the scope of
* this one, beginning with the one closest to this one (often its previous
* sibling) and proceeding forward through the hierarchy, so that each new
* result {@link MathConcept#isLaterThan isLaterThan()} the previous.
*
* @param {boolean} reflexive - Functions analogously to the `reflexive`
* parameter for {@link MathConcept#isInTheScopeOf isInTheScopeOf()}; that
* is, do we include this MathConcept on its list of things in its scope?
* The default value is true.
*
* @see {@link MathConcept#isInTheScopeOf isInTheScopeOf()}
* @see {@link MathConcept#scope scope()}
*/
*scopeIterator ( reflexive = true ) {
for ( let sibling = this ; sibling ; sibling = sibling.nextSibling() ) {
if ( sibling === this ) {
if ( reflexive ) yield this
} else {
yield* sibling.descendantsIterator()
}
}
}
/**
* The full contents of {@link MathConcept#scopeIterator scopeIterator()}, but
* put into an array rather than an iterator, for convenience, possibly at
* the cost of efficiency.
*
* @param {boolean} reflexive - Passed directly to
* {@link MathConcept#scopeIterator scopeIterator()}; see that function for
* more information
* @return {MathConcept[]} All MathConcepts in the scope of to this one, with
* the earliest (closest to this MathConcept) first, proceeding on to the
* latest at the end of the array
* @see {@link MathConcept#scopeIterator scopeIterator()}
* @see {@link MathConcept#isInTheScopeOf isInTheScopeOf()}
*/
scope ( reflexive = true ) {
return Array.from( this.scopeIterator( reflexive ) )
}
//////
//
// Interpretation
//
//////
/**
* Any MathConcept can be interpreted, which means converting its high-level
* concepts into lower-level concepts that are only logical. For example,
* in mathematics, we my write A=B=C, but logically, this is two separate
* statements, A=B and B=C.
*
* The interpretation function defined here can be used by any subclass to
* implement its specific means of interpretation of mathematical concepts
* into logical ones. In this abstract base class, the default is simply
* to return an empty list, meaning "no logic concepts." Subclasses should
* override this with an implementation specific to their actual mathematical
* meaning.
*
* @return {LogicConcept[]} The ordered list of LogicConcepts whose combined
* meaning is equal to the meaning of this MathConcept
*/
interpretation () { return [ ] }
//////
//
// Functions for copying and serialization
//
//////
/**
* In order for a hierarchy of MathConcepts to be able to be serialized and
* deserialized, we need to track the class of each MathConcept in the
* hierarchy. We cannot reconstitute an object from its serialized state if
* we do not know which class to construct. So we track all subclasses of
* this class in a single static map, here.
*
* This class and each of its subclasses should add themselves to this map
* and save the corresponding name in a static `className` variable in their
* class.
*
* @see {@link MathConcept#className className}
* @see {@link MathConcept.addSubclass addSubclass}
*/
static subclasses = new Map
/**
* Adds a subclass to the static {@link MathConcept#subclasses subclasses} map
* tracked by this object, for use in reconsituting objects correctly from
* their serialized forms.
*
* This method should be called once per subclass of `MathConcept`. To see
* how, see the code that initializes {@link MathConcept#className className}.
*
* @param {string} name - The name of the class, as it appears in code
* @param {class} classObject - The class itself, such as `MathConcept`, or
* any of its subclasses, that is, the JavaScript object used when
* constructing new instances.
* @return {string} The value of the `name` parameter, for convenience in
* initializing each class's static `className` field
* @see {@link MathConcept#className className}
* @see {@link MathConcept#subclasses subclasses}
*/
static addSubclass ( name, classObject ) {
MathConcept.subclasses.set( name, classObject )
return name
}
/**
* The name of this class, as a JavaScript string. For the MathConcept class,
* this is, of course, `"MathConcept"`, but for subclasses, it will vary.
*
* See the code initializing this member to see how subclasses should
* initialize their `className` members. This is used in deserialization,
* to correctly reconstitute objects of the appropriate class.
* @see {@link MathConcept#subclasses subclasses}
* @see {@link MathConcept.addSubclass addSubclass}
*/
static className = MathConcept.addSubclass( 'MathConcept', MathConcept )
/**
* A deep copy of this MathConcept. It will have no subtree in common with
* this one, and yet it will satisfy an {@link MathConcept#equals equals()}
* check with this MathConcept.
*
* In order to ensure that the copy has the same class as the original (even
* if that is a proper subclass of MathConcept), this function depends upon
* that subclass's having registered itself with the
* {@link MathConcept#subclasses subclasses} static member.
*
* @return {MathConcept} A deep copy
* @see {@link MathConcept#equals equals()}
* @see {@link MathConcept#subclasses subclasses}
*/
copy () {
const className = this.constructor.className
const classObject = MathConcept.subclasses.get( className )
const childCopies = this._children.map( child => child.copy() )
const copy = new classObject( ...childCopies )
copy._attributes = this._attributes.deepCopy()
return copy
}
/**
* Whether this MathConcept is structurally equal to the one passed as
* parameter. In particular, this means that this function will return
* true if and only if all the following are true.
*
* * `other` is an instance of the MathConcept class
* * `other` has the same set of attribute keys as this instance
* * each of those keys maps to the same data in each instance (where
* comparison of attribute values is done by
* {@link JSON.equals JSON.equals()})
* * `other` has the same number of children as this instance
* * each of `other`'s children passes a recursive
* {@link MathConcept#equals equals()} check with the corresponding
* child of this instance
*
* @param {MathConcept} other
* @returns {boolean} true if and only if this MathConcept equals `other`
* @see {@link MathConcept#copy copy()}
*/
equals ( other, attributesToIgnore = [ ] ) {
// other must be a MathConcept with same specific subclass
if ( !( other instanceof MathConcept ) ) return false
if ( this.constructor !== other.constructor ) return false
// other must have the same number of attribute keys
const keys1 = Array.from( this._attributes.keys() ).filter(
key => !attributesToIgnore.includes( key ) )
const keys2 = Array.from( other._attributes.keys() ).filter(
key => !attributesToIgnore.includes( key ) )
if ( keys1.length != keys2.length ) return false
// other must have the same set of attribute keys
keys1.sort()
keys2.sort()
if ( !JSON.equals( keys1, keys2 ) ) return false
// other must have the same value for each attribute key
for ( let key of keys1 )
if ( !JSON.equals( this.getAttribute( key ),
other.getAttribute( key ) ) )
return false
// other must have the same number of children
if ( this._children.length != other._children.length ) return false
// other must have the same children, structurally, recursively compared
for ( let i = 0 ; i < this._children.length ; i++ )
if ( !this.child( i ).equals( other.child( i ), attributesToIgnore ) )
return false
// that is the complete set of requirements for equality
return true
}
/**
* Convert this object to JavaScript data ready for JSON serialization.
* Note that the result of this function is *not* a string, but is ready to
* be converted into one through `JSON.stringify()` or (preferably),
* {@link predictableStringify predictableStringify()}.
*
* The resulting object has some of its attributes directly re-used (not
* copied) from within this MathConcept (notably the values of many
* attributes), for the sake of efficiency. Thus you should *not* modify
* the contents of the returned MathConcept. If you want a completely
* independent copy, call `JSON.copy()` on the return value.
*
* The particular classes of this MathConcept and any of its children are
* stored in the result, so that a deep copy of this MathConcept can be
* recreated from that object using {@link Strucure.fromJSON fromJSON()}.
*
* If the serialized result will later be deserialized after the original
* has been destroyed, then you may wish to preserve the unique IDs of each
* MathConcept in the hierarchy in the serialization. But if the original
* will still exist, you may not. Thus the parameter lets you choose which
* of these behaviors you need. By default, IDs are included.
*
* @param {boolean} includeIDs - Whether to include the IDs of the
* MathConcept and its descendants in the serialized form (as part of the
* MathConcept's attributes)
* @return {Object} A serialized version of this MathConcept
* @see {@link Strucure.fromJSON fromJSON()}
* @see {@link Strucure#subclasses subclasses}
*/
toJSON ( includeIDs = true ) {
let serializedAttributes = [ ...this._attributes ]
if ( !includeIDs )
serializedAttributes = serializedAttributes.filter(
pair => pair[0] != '_id' )
return {
className : this.constructor.className,
attributes : serializedAttributes,
children : this._children.map( child => child.toJSON( includeIDs ) )
}
}
/**
* Deserialize the data in the argument, producing a new MathConcept instance
* (or, more specifically, sometimes an instance of one of its subclasses).
*
* Note that because this function is static, clients access it as
* `MathConcept.fromJSON(...)`.
*
* @param {Object} data - A JavaScript Object of the form produced by
* {@link MathConcept#toJSON toJSON()}
* @return {MathConcept} A new MathConcept instance (which may actually be an
* instance of a proper subclass of MathConcept) as encoded in the given
* `data`
* @see {@link MathConcept#toJSON toJSON()}
*/
static fromJSON ( data ) {
const classObject = MathConcept.subclasses.get( data.className )
const result = new classObject(
...data.children.map( MathConcept.fromJSON ) )
result._attributes = new Map( JSON.copy( data.attributes ) )
return result
}
/**
* A simple string representation that represents any MathConcept using
* an S-expression (that is, `(a b c ...)`) of the string representations of
* its children. This produces LISP-like results, although they will
* contain only parentheses if all are MathConcept instances. But
* subclasses can override this method to specialize it.
*
* @return {string} A simple string representation
* @see {@link MathConcept#toJSON toJSON()}
*/
toString () {
return '('
+ this._children.map( child => child.toString() ).join( ' ' )
+ ')'
}
//////
//
// Bound and free identifiers
//
//////
/**
* By default, MathConcepts do not bind symbols. Subclasses of
* MathConcept may import the {@link BindingInterface BindingInterface} and
* therefore override the following function, but in the base case, it
* simply returns false to indicate that no symbols are bound.
*
* @return {boolean} the constant false
*
* @see {@link BindingInterface.binds binds()}
*/
binds () { return false }
/**
* A {@link Symbol} X is free in an ancestor Y if and only if no MathConcept
* that is an ancestor A of X inside of (or equal to) Y satisfies
* `A.binds( X.text() )`. This function returns an array of all symbol
* names that appear within this MathConcept and, at the point where they
* appear, are free in this ancestor MathConcept.
*
* If, instead of just the names of the symbols, you wish to have the
* {@link Symbol Symbol} instances themselves, you can couple the
* {@link MathConcept#isFree isFree()} function with the
* {@link MathConcept#descendantsSatisfying descendantsSatisfying()}
* function to achieve that.
*
* @return {string[]} an array of names of free symbols appearing as
* descendants of this MathConcept
*
* @see {@link BindingInterface.binds binds()}
* @see {@link BindingInterface.boundSymbols boundSymbols()}
*/
freeSymbolNames () {
// a single symbol is free in itself
if ( this instanceof MathConcept.subclasses.get( 'Symbol' ) )
return [ this.text() ]
// otherwise we collect all the free variables in all children...
const result = new Set
this.children().forEach( child =>
child.freeSymbolNames().forEach( name => result.add( name ) ) )
// ...excepting any that this MathConcept binds, if any
if ( this.binds() )
this.boundSymbolNames().forEach( name => result.delete( name ) )
return Array.from( result )
}
/**
* Is this MathConcept free in one of its ancestors? If the ancestor is not
* specified, it defaults to the MathConcept's topmost ancestor. Otherwise,
* you can specify it with the parameter.
*
* A MathConcept is free in an ancestor if none of the MathConcept's free
* identifiers are bound within that ancestor.
*
* Note the one rare corner case that the head of a binding (even if it is
* a compound expression) is not bound by the binding. For instance, if
* we have an expression like $\sum_{i=a}^b A_i$, then $i$ is bound in
* $A_i$, but not in either $a$ or $b$, which are part of the compound head
* symbol, written in LISP notation something like `((sum a b) i (A i))`.
* This corner case rarely arises, because it would be very confusing for
* $i$ to appear free in either $a$ or $b$, but is important to document.
*
* @param {MathConcept} [inThis] - The ancestor in which the question takes
* place, as described above
* @return {boolean} Whether this MathConcept is free in the specified
* ancestor (or its topmost ancestor if none is specified)
*
* @see {@link BindingInterface.binds binds()}
*/
isFree ( inThis ) {
// compute the free identifiers in me that an ancestor might bind
const myFreeSymbolNames = this.freeSymbolNames()
// walk upwards to the appropriate ancestor and see if any bind any of
// those identifiers; if so, I am not free in that ancestor
let walk = this
while ( walk && walk != inThis ) {
const parent = walk.parent()
if ( parent
&& myFreeSymbolNames.some( name => parent.binds( name ) ) )
return false
walk = parent
}
// none bound me, so I am free
return true
}
/**
* Does a copy of the given MathConcept `concept` occur free anywhere in this
* MathConcept? More specifically, is there a descendant D of this MathConcept
* such that `D.equals( concept )` and `D.isFree( inThis )`?
*
* @param {MathConcept} concept - This function looks for copies of this
* MathConcept
* @param {MathConcept} [inThis] - The notion of "free" is relative to this
* MathConcept, in the same sense of the `inThis` parameter to
* {@link MathConcept#isFree isFree()}
* @return {boolean} True if and only if there is a copy of `concept` as a
* descendant of this MathConcept satisfying `.isFree( inThis )`
* @see {@link MathConcept#isFree isFree()}
*
* @see {@link BindingInterface.binds binds()}
*/
occursFree ( concept, inThis ) {
return this.hasDescendantSatisfying( descendant =>
descendant.equals( concept ) && descendant.isFree( inThis ) )
}
/**
* A MathConcept A is free to replace a MathConcept B if no identifier free in A
* becomes bound when B is replaced by A.
*
* @param {MathConcept} original - The MathConcept to be replaced with this one
* @param {MathConcept} [inThis] - The ancestor we use as a context in which to
* gauge bound/free identifiers, as in the `inThis` parameter to
* {@link MathConcept#isFree isFree()}. If omitted, the context defaults to
* the top-level ancestor of `original`.
*
* @return {boolean} True if this MathConcept is free to replace `original`,
* and false if it is not.
*
* @see {@link BindingInterface.binds binds()}
*/
isFreeToReplace ( original, inThis ) {
// this implementation is an exact copy of isFree(), with one exception:
// while the free identifiers are computed from this MathConcept, freeness
// is computed from original.
const freeSymbolNames = this.freeSymbolNames()
let walk = original
while ( walk && walk != inThis ) {
const parent = walk.parent()
if ( parent
&& freeSymbolNames.some( name => parent.binds( name ) ) )
return false
walk = parent
}
return true
}
/**
* Consider every free occurrence of `original` within this MathConcept, and
* replace each with a copy of `replacement` if and only if `replacement` is
* free to replace that instance. Each instance is judged separately, so
* there may be any number of replacements, from zero up to the number of
* free occurrences of `original`.
*
* @param {MathConcept} original - Replace copies of this MathConcept with
* copies of `replacement`
* @param {MathConcept} replacement - Replace copies of `original` with
* copies of this MathConcept
* @param {MathConcept} [inThis] - When judging free/bound identifiers, judge
* them relative to this ancestor context, in the same sense of the
* `inThis` parameter to {@link MathConcept#isFree isFree()}
*
* @see {@link BindingInterface.binds binds()}
*/
replaceFree ( original, replacement, inThis ) {
this.descendantsSatisfying(
descendant => descendant.equals( original )
).forEach( instance => {
if ( replacement.isFreeToReplace( instance, inThis ) )
instance.replaceWith( replacement.copy() )
} )
}
//////
//
// Unique IDs
//
//////
/**
* We want the capability of assigning each MathConcept in a given hierarchy a
* globally unique ID. We therefore need a global place to store the
* mapping of IDs to instances, and thus we create this Map in the MathConcept
* class.
*
* Each key in the map is an ID and the corresponding value is the instance
* with that ID. Each ID is a string.
*
* This data structure should not be accessed by clients; it is private to
* this class. Use {@link MathConcept.instanceWithID instanceWithID()} and
* {@link MathConcept#trackIDs trackIDs()} instead.
*
* @see {@link MathConcept.instanceWithID instanceWithID()}
* @see {@link MathConcept#trackIDs trackIDs()}
*/
static IDs = new Map
/**
* Find a MathConcept instance from a given string ID. This assumes that the
* assignment of ID to MathConcept has been recorded in the global mapping in
* {@link MathConcept#IDs IDs}, by the function
* {@link MathConcept#trackIDs trackIDs()}. If it has not been so recorded,
* then this function will not find the instance and will return undefined.
*
* Note that because this function is static, clients access it as
* `MathConcept.instanceWithID("...")`.
*
* @param {string} id - The MathConcept ID to look up
* @return {MathConcept} The MathConcept that has the given ID, if any, or
* undefined if no MathConcept has the given ID
*
* @see {@link MathConcept#IDs IDs}
* @see {@link MathConcept#trackIDs trackIDs()}
*/
static instanceWithID ( id ) { return MathConcept.IDs.get( `${id}` ) }
/**
* The ID of this MathConcept, if it has one, or undefined otherwise. An ID
* is always a string; this is ensured by the
* {@link MathConcept#setId setId()} function.
*
* @return {string} The ID of this MathConcept, or undefined if there is none
*
* @see {@link MathConcept#setId setID()}
*/
ID () { return this.getAttribute( '_id' ) }
/**
* Set the ID of this MathConcept. Note that this does not change the
* global tracking of IDs, because one could easily call this function to
* assign an already-in-use ID. To ensure that the IDs in a hierarchy are
* tracked, call {@link MathConcept#trackIDs trackIDs()}, and if that has
* already been called, then to change a MathConcept's ID assignment, call
* {@link MathConcept#changeID changeID()}.
*
* @param {string} id - The new ID to assign. If this is not a string, it
* will be converted into one.
*
* @see {@link MathConcept#ID ID()}
* @see {@link MathConcept#trackIDs trackIDs()}
* @see {@link MathConcept#changeID changeID()}
*/
setID ( id ) { this.setAttribute( '_id', `${id}` ) }
/**
* Store in the global {@link MathConcept#IDs IDs} mapping the association of
* this MathConcept's ID with this MathConcept instance itself. If the
* parameter is set to true (the default), then do the same recursively to
* all of its descendants.
*
* Calling this function then enables you to call
* {@link MathConcept.instanceWithID instanceWithID()} on any of the IDs of a
* descendant and get that descendant in return. Note that this does not
* check to see if a MathConcept with the given ID has already been recorded;
* it will overwrite any past data in the {@link MathConcept#IDs IDs} mapping.
*
* This function also makes a call to
* {@link MathConcept#trackConnections trackConnections()}, because IDs are
* required in order for connections to exist, and enabling IDs almost
* always coincides with enabling connections as well.
*
* **Important:**
* To prevent memory leaks, whenever a MathConcept hierarchy is no longer used
* by the client, you should call {@link MathConcept#untrackIDs untrackIDs()}
* on it.
*
* @param {boolean} recursive - Whether to recursively track IDs of all
* child, grandchild, etc. MathConcepts. (If false, only this MathConcept's
* ID is tracked, not those of its descendants.)
*
* @see {@link MathConcept#IDs IDs}
* @see {@link MathConcept#untrackIDs untrackIDs()}
* @see {@link MathConcept#trackConnections trackConnections()}
*/
trackIDs ( recursive = true ) {
this.trackConnections()
if ( this.hasAttribute( '_id' ) ) MathConcept.IDs.set( this.ID(), this )
if ( recursive ) for ( let child of this._children ) child.trackIDs()
}
/**
* This removes the ID of this MathConcept (and, if requested, all descendant
* MathConcepts) from the global {@link MathConcept#IDs IDs} mapping. It is the
* reverse of {@link MathConcept#trackIDs trackIDs()}, and should always be
* called once the client is finished using a MathConcept, to prevent memory
* leaks.
*
* Because connections use the ID system, any connections that this
* MathConcept is a part of will also be severed, by a call to
* {@link MathConcept#removeConnections removeConnections()}.
*
* @param {boolean} recursive - Whether to recursively apply this function
* to all child, grandchild, etc. MathConcepts. (If false, only this
* MathConcept's ID is untracked, not those of its descendants.)
*
* @see {@link MathConcept#IDs IDs}
* @see {@link MathConcept#trackIDs trackIDs()}
* @see {@link MathConcept#clearIDs clearIDs()}
*/
untrackIDs ( recursive = true ) {
this.removeConnections()
if ( this.hasAttribute( '_id' ) ) MathConcept.IDs.delete( this.ID() )
if ( recursive ) for ( let child of this._children ) child.untrackIDs()
}
/**
* Check whether this MathConcept's ID is currently tracked and associated
* with this MathConcept itself.
*
* @return {boolean} Whether the ID of this MathConcept is currently tracked
* by the global {@link MathConcept#IDs IDs} mapping *and* that it is
* associated, by that mapping, with this MathConcept
*
* @see {@link MathConcept#IDs IDs}
* @see {@link MathConcept#trackIDs trackIDs()}
*/
idIsTracked () {
return this.hasAttribute( '_id' )
&& this == MathConcept.instanceWithID( this.ID() )
}
/**
* Remove the ID of this MathConcept and, if requested, all of its
* descendants. This does not change anything about the global
* {@link MathConcept#IDs IDs} mapping, so if this MathConcept's IDs are
* tracked, you should call {@link MathConcept#untrackIDs untrackIDs()} first.
*
* Because connections use the ID system, any connections that this
* MathConcept is a part of will also be severed, by a call to
* {@link MathConcept#removeConnections removeConnections()}.
*
* @param {boolean} recursive - Whether to clear IDs from all descendants of
* this MathConcept as well
*
* @see {@link MathConcept#IDs IDs}
* @see {@link MathConcept#untrackIDs untrackIDs()}
*/
clearIDs ( recursive = true ) {
this.removeConnections()
this.clearAttributes( '_id' )
if ( recursive ) for ( let child of this._children ) child.clearIDs()
}
/**
* If a MathConcept wishes to change its ID, then we may need to update the
* internal {@link MathConcept#IDs IDs} mapping. The following function
* changes the ID and updates that mapping if needed all in one action, to
* make it easy for the client to change a MathConcept's ID, just by calling
* this function.
*
* If for some reason the change was not possible, then this function will
* take no action and return false. Possible reasons include:
* * the old ID isn't tracked in the {@link MathConcept#IDs IDs} mapping
* * the new ID is already associated with another MathConcept
* * the new ID is the same as the old ID
*
* This function also updates *other* MathConcepts that connect to this one,
* changing their connections to use this MathConcept's new ID, so that all
* connections are preserved across the use of this function.
*
* @param {string} newID - The ID to use as the replacement for this
* MathConcept's existing ID. It will be treated as a string if it is not
* already one.
* @return {boolean} True if the operation succeeded, false if it could not
* be performed (and thus no action was taken)
*/
changeID ( newID ) {
// verify that we can do the job:
const oldID = this.ID()
newID = `${newID}`
if ( MathConcept.IDs.has( newID )
|| this != MathConcept.instanceWithID( oldID ) ) return false
// change my ID in connections:
for ( const connection of this.getConnections() )
connection.handleIDChange( oldID, newID )
// change my ID:
MathConcept.IDs.delete( oldID )
this.setID( newID )
MathConcept.IDs.set( newID, this )
return true
}
//////
//
// Sending feedback
//
//////
/**
* This implementation of the feedback function is a stub. It does nothing
* except dump the data to the console. However, it serves as the central
* method that all MathConcepts should use to transmit feedback, so that when
* this class is used in the LDE, which has a mechanism for transmitting
* feedback messages to its clients, the LDE can override this
* implementation with a real one, and all calls that use this central
* channel will then be correctly routed.
*
* Documentation will be forthcoming later about the required form and
* content of the `feedbackData` parameter.
*
* @param {Object} feedbackData - Any data that can be encoded using
* `JSON.stringify()` (or
* {@link predictableStringify predictableStringify()}), to be transmitted
* @see {@link MathConcept#feedback feedback() method for instances}
* @see {@link LogicConcept#feedback feedback() for LogicConcepts}
*/
static feedback ( feedbackData ) {
console.log( 'MathConcept class feedback not implemented:', feedbackData )
}
/**
* Send feedback on this particular MathConcept instance. This takes the
* given feedback data, adds to it the fact that this particular instance is
* the subject of the feedback (by using its {@link MathConcept#id id()},
* and then asks the static {@link MathConcept.feedback feedback()} function
* to send that feedback to the LDE.
*
* @param {Object} feedbackData - Any data that can be encoded using
* `JSON.stringify()` (or
* {@link predictableStringify predictableStringify()}), to be transmitted
* @see {@link MathConcept.feedback static feedback() method}
* @see {@link LogicConcept#feedback feedback() for LogicConcepts}
*/
feedback ( feedbackData ) {
feedbackData.subject = this.ID()
MathConcept.feedback( feedbackData )
}
//////
//
// Connections
//
//////
/**
* Get the IDs of all connections into or out of this MathConcept.
*
* @return {string[]} An array of all the IDs of all the connections into or
* out of this MathConcept. These unique IDs can be used to get a
* {@link Connection Connection} object; see that class's
* {@link Connection.withID withID()} function.
* @see {@link MathConcept#getConnections getConnections()}
* @see {@link MathConcept#getConnectionIDsIn getConnectionIDsIn()}
* @see {@link MathConcept#getConnectionIDsOut getConnectionIDsOut()}
*/
getConnectionIDs () {
return this.getAttributeKeys().filter( key =>
key.substring( 0, 13 ) == '_conn target ' ||
key.substring( 0, 13 ) == '_conn source ' )
.map( key => key.substring( 13 ) )
}
/**
* Get the IDs of all connections into this MathConcept.
*
* @return {string[]} An array of all the IDs of all the connections into
* this MathConcept. These unique IDs can be used to get a
* {@link Connection Connection} object; see that class's
* {@link Connection.withID withID()} function.
* @see {@link MathConcept#getConnectionsIn getConnectionsIn()}
* @see {@link MathConcept#getConnectionIDs getConnectionIDs()}
* @see {@link MathConcept#getConnectionIDsOut getConnectionIDsOut()}
*/
getConnectionIDsIn () {
return this.getAttributeKeys().filter( key =>
key.substring( 0, 13 ) == '_conn source ' )
.map( key => key.substring( 13 ) )
}
/**
* Get the IDs of all connections out of this MathConcept.
*
* @return {string[]} An array of all the IDs of all the connections out of
* this MathConcept. These unique IDs can be used to get a
* {@link Connection Connection} object; see that class's
* {@link Connection.withID withID()} function.
* @see {@link MathConcept#getConnectionsOut getConnectionsOut()}
* @see {@link MathConcept#getConnectionIDs getConnectionIDs()}
* @see {@link MathConcept#getConnectionIDsIn getConnectionIDsIn()}
*/
getConnectionIDsOut () {
return this.getAttributeKeys().filter( key =>
key.substring( 0, 13 ) == '_conn target ' )
.map( key => key.substring( 13 ) )
}
/**
* Get all connections into or out of this MathConcept, as
* {@link Connection Connection} instances. This function simply maps the
* {@link Connection.withID withID()} function over the result of
* {@link MathConcept#getConnectionIDs getConnectionIDs()}.
*
* @return {Connection[]} An array of all the Connections into or out of
* this MathConcept.
* @see {@link MathConcept#getConnectionIDs getConnectionIDs()}
* @see {@link MathConcept#getConnectionsIn getConnectionsIn()}
* @see {@link MathConcept#getConnectionsOut getConnectionsOut()}
*/
getConnections () { return this.getConnectionIDs().map( Connection.withID ) }
/**
* Get all connections into this MathConcept, as
* {@link Connection Connection} instances. This function simply maps the
* {@link Connection.withID withID()} function over the result of
* {@link MathConcept#getConnectionIDsIn getConnectionIDsIn()}.
*
* @return {Connection[]} An array of all the Connections into this
* MathConcept.
* @see {@link MathConcept#getConnectionIDsIn getConnectionIDsIn()}
* @see {@link MathConcept#getConnections getConnections()}
* @see {@link MathConcept#getConnectionsOut getConnectionsOut()}
*/
getConnectionsIn () { return this.getConnectionIDsIn().map( Connection.withID ) }
/**
* Get all connections out of this MathConcept, as
* {@link Connection Connection} instances. This function simply maps the
* {@link Connection.withID withID()} function over the result of
* {@link MathConcept#getConnectionIDsOut getConnectionIDsOut()}.
*
* @return {Connection[]} An array of all the Connections out of this
* MathConcept.
* @see {@link MathConcept#getConnectionIDsOut getConnectionIDsOut()}
* @see {@link MathConcept#getConnections getConnections()}
* @see {@link MathConcept#getConnectionsIn getConnectionsIn()}
*/
getConnectionsOut () { return this.getConnectionIDsOut().map( Connection.withID ) }
/**
* Connect this MathConcept to another, called the *target,* optionally
* attaching some data to the connection as well. This function just calls
* {@link Connection.create Connection.create()}, and is thus here just for
* convenience.
*
* @param {MathConcept} target - The target of the new connection
* @param {string} connectionID - The unique ID to use for the new
* connection we are to create
* @param {*} data - The optional data to attach to the new connection. See
* the {@link Connection.create create()} function in the
* {@link Connection Connection} class for the acceptable formats of this
* data.
* @return {Connection} A {@link Connection Connection} instance for the
* newly created connection between this MathConcept and the target. This
* return value can be safely ignored, because the connection data is
* stored in the source and target MathConcepts, and is not dependent on the
* Connection object itself. However, the return value will be false if
* the chosen connection ID is in use or if this MathConcept or the target
* does not pass {@link MathConcept#idIsTracked idIsTracked()}.
*/
connectTo ( target, connectionID, data = null ) {
return Connection.create( connectionID, this.ID(), target.ID(), data )
}
/**
* Remove all connections into or out of this MathConcept. This deletes the
* relevant data from this MathConcept's attributes as well as those of the
* MathConcepts on the other end of each connection. For documentation on the
* data format for this stored data, see the {@link Connection Connection}
* class.
*
* This function simply runs {@link Connection#remove remove()} on every
* connection in {@link MathConcept#getConnections getConnections()}.
*
* @see {@link Connection#remove remove()}
* @see {@link MathConcept#getConnections getConnections()}
*/
removeConnections () {
this.getConnections().forEach( connection => connection.remove() )
}
/**
* There are some situations in which a MathConcept hierarchy will have data
* in it about connections, and yet those connections were not created with
* the API in the {@link Connection Connection}s class. For example, if a
* MathConcept hierarchy has been saved in serialized form and then
* deserialized at a later date. Thus we need a way to place into the
* {@link Connection.IDs IDs member of the Connection class} all the IDs of
* the connections in any given MathConcept hierarchy. This function does so.
* In that way, it is very similar to {@link MathConcept#trackIDs trackIDs()}.
*
* Connections are processed only at the source node, so that we do not
* process each one twice. Thus any connection into this MathConcept from
* outside will not be processed by this function, but connections from this
* one out or among this one's descendants in either direction will be
* processed.
*
* @return {boolean} True if and only if every connection ID that appers in
* this MathConcept and its descendants was able to be added to the global
* mapping in {@link Connection.IDs IDs}. If any fail (because the ID was
* already in use), this returns false. Even if it returns false, it
* still adds as many connections as it can to that global mapping.
*/
trackConnections () {
let success = true
for ( const id of this.getConnectionIDsOut() ) {
if ( Connection.IDs.has( id ) ) {
success = false
} else {
Connection.IDs.set( id, this )
}
}
for ( const child of this.children() ) {
if ( !child.trackConnections() ) success = false
}
return success
}
/**
* When replacing a MathConcept in a hierarchy with another, we often want to
* transfer all connections that went into or out of the old MathConcept to
* its replacement instead. This function performs that task.
*
* This function is merely a convenient interface that just calls
* {@link Connection.transferConnections Connection.transferConnections()}
* on your behalf.
*
* @param {MathConcept} recipient - The MathConcept to which to transfer all of
* this one's connections
* @see {@link Connection.transferConnections Connection.transferConnections()}
*/
transferConnectionsTo ( recipient ) {
return Connection.transferConnections( this, recipient )
}
//////
//
// Interpretation and smackdown notation
//
//////
// for internal use by fromSmackdown(), interpret(), and
// attemptReverseInterpret()
static interpretationKey = 'Interpret as'
/**
* MathConcept trees can be represented using a notation called
* "smackdown," which is a superset of the "putdown" notation used to
* represent LogicConcepts, {@link LogicConcept.fromPutdown as documented
* here}. (Both of these are, of course, plays on the name of the famous
* format "markdown" by John Gruber.)
*
* Smackdown supports all notation used in putdown (so readers may wish to
* begin learning about smackdown by following the link above to first
* learn about putdown) plus the following additional features:
*
* * The notation `$...$` can be used to represent a {@link LogicConcept
* LogicConcept}, for example, `$x^2-1$`. The use of dollar signs is
* intentionally reminiscent of $\LaTeX$ notation for in-line math.
* * Note! `$x^2-1$` is merely an example stating that mathematics
* *of some sort* can be placed between the dollar signs; it is
* *not* an indication that the specific notation `x^2-1` is
* supported.
* * At present, *no* notation is supported, and any text between
* dollar signs is simply stored as a {@link MathConcept
* MathConcept} instance with attribute "Interpret as" set to
* `["notation","X"]`, where X is the contents of the `$...$`
* (without the dollar signs)---e.g., `["notation","x^2-1"]`.
* * We will later add support for defining custom mathematical
* notation for use in `$...$` expressions, and then create a more
* robust way to {@link MathConcept#interpret interpret()}
* MathConcept trees into {@link LogicConcept LogicConcepts}, which
* will parse such notation to create real expressions.
* * To include a literal `$` inside a `$...$` section, escape it as
* `\$`.
* * The notation `\command{argument1}...{argumentN}` is also
* intentionally reminiscent of $\LaTeX$ notation, and has an analogous
* meaning: some command applied to a sequence of $N$ arguments. For
* now, only the following commands are supported.
* * `\begin{proof}` and `\end{proof}` are replaced with ` { ` and
* ` } `, respectively, so that they can be used to construct {@link
* Environment Environments}, which is the meaning one would expect.
* * `\label{X}` is interpreted as if it were putdown notation for
* adding a JSON attribute to the preceding {@link LogicConcept
* LogicConcept}, associating the key "label" with the value X.
* * `\ref{X}` is exactly like the previous, but using attribute key
* "ref" instead of "label."
* * Any other command is stored as a {@link MathConcept MathConcept}
* instance with attribute "Interpret as" having the form
* `["command",...]`, for example, `\foo{bar}{baz}` would become
* `["command","foo","bar","baz"]`.
*
* The two notations given above will work hand-in-hand more over time.
* Specifically, we will create types of `\command`s that can define new
* notation to appear inside `$...$` blocks.
*
* For now, this routine fully supports parsing smackdown notation, but
* does not yet obey a robust set of commands, only those shown above.
*
* @param {string} string the smackdown code to be interpreted
* @returns {MathConcept[]} an array of MathConcept instances, the meaning
* of the smackdown code provided as input
*
* @see {@link MathConcept#toSmackdown toSmackdown()}
*/
static fromSmackdown ( string ) {
const map = new SourceMap( string )
// Regular expressions for key features of smackdown:
const notationRE = /(?<!\\)(?:\\\\)*\$((?:[^\\$\n\r]|\\\$|\\\\)*)\$/
const notationErrors = [
// open dollar with no close dollar on the same line:
/(?<!\\)(?:\\\\)*\$((?:[^\\$\n\r]|\\\$|\\\\)*)(?:\n|\r|\/\/|$)/
]
const commandRE =
/\\([a-zA-Z]+)(?:(?![a-zA-Z{])|(?:\{((?:[^{}\\\n\r]|\\\\|\\.|\}\{)*)\}))/
const commandErrors = [
// start of command with no ending on the same line:
/\\([a-z]+)\{((?:[^{}\\\n\r]|\\.|\}\{)*)(?:\n|\r|\/\/|$)/,
// escaping errors:
/\\([a-z]+)\{((?:[^{}\\\n\r]|\\.|\}\{)*)\}/
]
const unescape = ( text, escapables ) => {
let result = ''
escapables += '\\'
for ( let i = 0 ; i < text.length ; i++ ) {
const char = text.charAt( i )
if ( char != '\\' ) { result += char } else {
if ( i == text.length - 1 )
throw new Error(
`Backslash at end of escaped text: ${text}` )
const next = text.charAt( i + 1 )
if ( escapables.indexOf( next ) == -1 )
throw new Error(
`Cannot escape this character: ${next}` )
result += next
i++
}
}
return result
}
// Walk through the text using the above regular expressions:
while ( string.length > 0 ) {
const notationPos = string.search( notationRE )
const commandPos = string.search( commandRE )
// If there's a notation error before anything else, quit now:
notationErrors.forEach( badRE => {
const badPos = string.search( badRE )
if ( badPos > -1
&& ( notationPos == -1 || notationPos > badPos )
&& ( commandPos == -1 || commandPos > badPos ) )
throw new Error( 'Invalid notation: '
+ string.substring( badPos, badPos + 10 ) )
} )
// If there's a command error before anything else, quit now:
commandErrors.forEach( badRE => {
const badPos = string.search( badRE )
if ( badPos > -1
&& ( notationPos == -1 || notationPos > badPos )
&& ( commandPos == -1 || commandPos > badPos ) )
throw new Error( 'Invalid command: '
+ string.substring( badPos, badPos + 10 ) )
} )
// If there are no special smackdown features, the putdown is fine:
if ( notationPos == -1 && commandPos == -1 )
break
// If the next thing is (some putdown and then) a command, then
// process (the putdown and then) that command next:
if ( notationPos == -1 ||
( commandPos > -1 && commandPos < notationPos ) ) {
string = string.substring( commandPos )
const match = commandRE.exec( string )
string = string.substring( match[0].length )
const command = match[1]
if (typeof match[2] === 'undefined') match[2]=''
const args = match[2].split( '}{' ).map(
arg => unescape( arg, '{}' ) )
let replacement =
MathConcept.attemptTextCommand( command, ...args )
if ( replacement === undefined )
replacement = map.nextMarker()
map.modify( map.nextModificationPosition() + commandPos,
match[0].length, replacement, {
type : 'command',
operator : command,
operands : args
} )
continue
}
// So the next thing is (some putdown and then) $...$ notation,
// so we will process (the putdown and then) that notation:
string = string.substring( notationPos )
const match = notationRE.exec( string )
string = string.substring( match[0].length )
map.modify( map.nextModificationPosition() + notationPos,
match[0].length, map.nextMarker(), {
type : 'notation',
notation : unescape( match[1], '$' )
} )
}
// How to reverse interpret that LC tree back into a MathConcept tree:
const reverseInterpret = LC => {
// If it was created by one of the aspects of smackdown that's not
// part of putdown, then create a special MathConcept for it:
if ( LC.constructor.className == 'Symbol'
&& SourceMap.isMarker( LC.text() ) ) {
const result = MathConcept.attemptReverseInterpret(
map.dataForMarker( LC.text() ) )
for ( let key of LC.getAttributeKeys() )
if ( key != 'symbol text' )
result.setAttribute( key,
JSON.copy( LC.getAttribute( key ) ) )
return result
}
// Otherwise just create a MathConcept that does a simple
// imitation of the LogicConcept created from the putdown:
const result = new MathConcept(
...LC.children().map( reverseInterpret ) )
result._attributes = LC._attributes.deepCopy()
result.setAttribute( MathConcept.interpretationKey,
[ 'class', LC.constructor.className ] )
return result
}
// Parse the putdown and reverse-interpret it:
try {
return MathConcept.subclasses.get( 'LogicConcept' )
.fromPutdown( map.modified() ).map( reverseInterpret )
} catch ( e ) {
// Convert any error containing "line n col m" to use correct #s:
const message = e.message ? e.message : e
const match = /line ([0-9]+) col ([0-9]+)/.exec( message )
if ( !match ) throw e
const pair = map.sourceLineAndColumn(
parseInt( match[1] ), parseInt( match[2] ) )
if ( !pair ) throw e
const [ origLine, origCol ] = pair
throw new Error( message.replace(
match[0], `line ${origLine} col ${origCol}` ) )
}
}
// For internal use by fromSmackdown().
// Briefly, its purpose: Take the operator and operands from some
// smackdown command of the form \operator{operand1}...{operandN} and
// return either a text replacement that obeys the command, or undefined
// to indicate that the command in question is not a simple
// text-replacement command.
static attemptTextCommand ( operator, ...operands ) {
const err = () => {
throw new Error(
`Invalid command use: \\${operator}{${operands.join('}{')}}` )
}
// handle label/ref text replacements
if ( operator == 'label' || operator == 'ref' ) {
if ( operands.length != 1 ) err()
return ` +{"${operator}":${JSON.stringify(operands[0])}}\n`
}
// handle \begin{proof}...\end{proof} text replacements
if ( operator == 'begin' || operator == 'end' ) {
if ( operands.length != 1 || operands[0] != 'proof' ) err()
return operator == 'begin' ? ' { ' : ' } '
}
// no other text replacements defined at this time
}
// For internal use by fromSmackdown().
// Briefly, its purpose: Take a data object associated with some section
// of the smackdown source and build a corresponding MathConcept instance
// X such that X.interpret() yields the meaning of the given data, as a
// LogicConcept instance.
static attemptReverseInterpret ( data ) {
// notation type not yet implemented; this is a placeholder
if ( data.type == 'notation' ) return new MathConcept().attr( [
[ MathConcept.interpretationKey, [ 'notation', data.notation ] ]
] )
// command type not yet implemented; this is a placeholder
if ( data.type == 'command' ) return new MathConcept().attr( [
[ MathConcept.interpretationKey, [ 'command', data.operator,
...data.operands ] ]
] )
// no other types yet implemented
throw new Error( `Unknown MathConcept type: ${data.type}` )
}
/**
* This function is a temporary placeholder. Later, a sophisticated
* interpretation mechanism will be developed to convert a user's
* representation of their document into a hierarchy of
* {@link LogicConcept LogicConcept} instances.
*
* (In fact, we will actually use the
* {@link MathConcept#interpret interpret()} function when we do so, and
* remove this one. That one is the official permanent interpretation API.)
*
* For now, we have this simple version in which many features are not yet
* implemented. Its behavior is as follows.
*
* 1. The method of interpretation that should be followed is extracted
* from the "Interpret as" attribute of this object. If there is no
* such attribute, an error is thrown. The attribute value should be
* an array, call it $[a_1,\ldots,a_n]$, where $a_1$ is the method of
* interpretation and each other $a_i$ is some parameter to it.
* 2. If $a_1$ is "class" then $a_2$ must be the name of some subclass of
* {@link LogicConcept LogicConcept}, and this routine will contruct a
* new instance of that class, copy all this object's attributes to
* it, and give it a children list built by recursively interpreting
* this MathConcept's children.
* 3. If $a_1$ is "notation" or "command" then a single {@link Symbol
* Symbol} will be created whose text content states that support for
* interpreting the notation or command in question has not yet been
* implemented.
* 4. If $a_1$ is anything else, an error is thrown.
*
* @returns {LogicConcept} the meaning of this MathConcept, subject to the
* limitations documented above
*
* @see {@link MathConcept.fromSmackdown fromSmackdown()} (which creates
* MathConcept hierarchies intended for interpretation)
* @see {@link MathConcept#toSmackdown toSmackdown()}
*/
interpret () {
const method = this.getAttribute( MathConcept.interpretationKey )
const fromPutdown = text =>
MathConcept.subclasses.get( 'LogicConcept' ).fromPutdown( text )
if ( method[0] == 'class' ) {
const classObject = MathConcept.subclasses.get( method[1] )
let constructorArgs = this.children().map( x => x.interpret() )
if ( method[1] == 'Declaration' )
constructorArgs = [
constructorArgs.slice( 0, constructorArgs.length - 1 ),
constructorArgs[constructorArgs.length - 1]
]
const result = new classObject( ...constructorArgs )
for ( let key of this.getAttributeKeys() )
if ( key != MathConcept.interpretationKey )
result.setAttribute( key,
JSON.copy( this.getAttribute( key ) ) )
return result
}
if ( method[0] == 'notation' )
return fromPutdown( JSON.stringify(
'notation interpretation not yet implemented: $'
+ method[1] + '$' ) )[0]
if ( method[0] == 'command' )
return fromPutdown( JSON.stringify(
'command interpretation not yet implemented: \\'
+ method[1] + '{' + method.slice( 2 ).join( '}{' )
+ '}' ) )[0]
throw new Error( `Invalid interpretation method: ${method[0]}` )
}
/**
* This function reverses the operation of
* {@link MathConcept.fromSmackdown fromSmackdown()}. It requires this
* MathConcept to be of the particular form created by that function; it
* cannot operate on arbitrary MathConcepts, because not all can be
* represented by smackdown notation. (For instance, a MathConcept
* created by a call to `new MathConcept()` is too vague to be
* representable using smackdown notation.)
*
* @returns {string} smackdown notation for this MathConcept
*
* @see {@link MathConcept.fromSmackdown fromSmackdown()}
* @see {@link MathConcept#interpret interpret()}
*/
toSmackdown () {
const LurchSymbol = MathConcept.subclasses.get( 'Symbol' )
const prePutdown = mc => {
const method = mc.getAttribute( MathConcept.interpretationKey )
if ( method[0] == 'class' ) {
const classObject = MathConcept.subclasses.get( method[1] )
let constructorArgs = mc.children().map( prePutdown )
if ( method[1] == 'Declaration' )
constructorArgs = [
constructorArgs.slice( 0, constructorArgs.length - 1 ),
constructorArgs[constructorArgs.length - 1]
]
const result = new classObject( ...constructorArgs )
for ( let key of mc.getAttributeKeys() )
if ( key != MathConcept.interpretationKey )
result.setAttribute( key,
JSON.copy( mc.getAttribute( key ) ) )
return result
}
if ( method[0] == 'notation' ) {
const result = new LurchSymbol( 'temp' )
const notation = '$' + method[1].replace( /\\/g, '\\\\' )
.replace( /\$/g, '\\$' ) + '$'
result.toPutdown = () => notation
return result
}
if ( method[0] == 'command' ) {
const result = new LurchSymbol( 'temp' )
const operator = method[1]
const operands = method.slice( 2 ).map( operand =>
operand.replace( /\{/g, '\\{' ).replace( /\}/g, '\\}' ) )
result.toPutdown = () =>
`\\${operator}{${operands.join('}{')}}`
return result
}
throw new Error( 'Cannot convert to smackdown: '
+ JSON.stringify( mc.toJSON() ) )
}
return prePutdown( this ).toPutdown()
}
}
source