Karoy Lorentey 06ea8b042f
Start using Rope for attribute run storage (#166)
* rdar://108152217 Start using Rope for attribute run storage

- Replace `Array<_InternalRun>` with `Rope<_InternalRun>`.
- Fix semantics of AttributeRunBoundaries.character. It no longer pretends that attributes can be tied to grapheme clusters — that never actually worked properly, and it cannot be supported without breaking the intended use case.
- Remove cache of latest run position; log(n) might be fast enough that we don’t need to worry about that.
- Switch AttributedString.Guts members to take/return BigString.Index values, not AttributedString indices. This cuts down on constant back-and-forth conversions.
- Discard unused Guts members; update remaining mutation methods to follow Swift naming conventions.
- Stop using the `public extension` language misfeature.

* rdar://108152217 Apply notes from code review

AttributedString.CharacterView and AttributedString.UnicodeScalarView
did not use the right semantics in the case when the underlying text
was not touched. It’s okay in that case to not mutate text storage,
but we still want to override the attributes within the subrange as if
the client actually did replace text — as in, we want to use precisely
the same attribute storage for the range on both paths.

Also, make sure we correctly enforce attribute constraints in the
`!hasStringChanges` case, in both views.

* rdar://108152217 Fix logic error in AttributedString.Runs.subscript

(Discovered during code review.)

* rdar://108152217 Clean up CharacterView/UnicodeScalarView mutations & update failing test

* rdar://108152217 Apply review notes
2023-06-26 13:51:15 -07:00

483 lines
20 KiB
Swift

//===----------------------------------------------------------------------===//
//
// This source file is part of the Swift.org open source project
//
// Copyright (c) 2020-2023 Apple Inc. and the Swift project authors
// Licensed under Apache License v2.0 with Runtime Library Exception
//
// See https://swift.org/LICENSE.txt for license information
// See https://swift.org/CONTRIBUTORS.txt for the list of Swift project authors
//
//===----------------------------------------------------------------------===//
#if FOUNDATION_FRAMEWORK
@_implementationOnly @_spi(Unstable) import CollectionsInternal
#else
package import _RopeModule
#endif
extension AttributedString {
internal final class Guts : @unchecked Sendable {
typealias Index = AttributedString.Index
typealias Runs = AttributedString.Runs
typealias AttributeMergePolicy = AttributedString.AttributeMergePolicy
typealias AttributeRunBoundaries = AttributedString.AttributeRunBoundaries
typealias _InternalRun = AttributedString._InternalRun
typealias _InternalRuns = AttributedString._InternalRuns
typealias _AttributeValue = AttributedString._AttributeValue
typealias _AttributeStorage = AttributedString._AttributeStorage
var string: BigString
var runs: _InternalRuns
// Note: the caller is responsible for performing attribute fix-ups if needed based on the source of the runs
init(string: BigString, runs: _InternalRuns) {
precondition(string.isEmpty == runs.isEmpty, "An empty attributed string should not contain any runs")
self.string = string
self.runs = runs
}
// Note: the caller is responsible for performing attribute fix-ups if needed based on the source of the runs
convenience init(string: String, runs: _InternalRuns) {
self.init(string: BigString(string), runs: runs)
}
convenience init() {
self.init(string: BigString(), runs: _InternalRuns())
}
}
}
extension AttributedString.Guts {
__consuming func copy() -> AttributedString.Guts {
AttributedString.Guts(string: self.string, runs: self.runs)
}
__consuming func copy(in range: Range<BigString.Index>) -> AttributedString.Guts {
let string = BigString(self.string.unicodeScalars[range])
let runs = self.runs.extract(utf8Offsets: range._utf8OffsetRange)
let copy = AttributedString.Guts(string: string, runs: runs)
// FIXME: Extracting a slice should not invalidate anything but .textChanged attribute runs on the edges
if range.lowerBound != string.startIndex || range.upperBound != string.endIndex {
var utf8Range = copy.stringBounds._utf8OffsetRange
utf8Range = copy.enforceAttributeConstraintsBeforeMutation(to: utf8Range)
copy.enforceAttributeConstraintsAfterMutation(in: utf8Range, type: .attributesAndCharacters)
}
return copy
}
}
extension AttributedString.Guts {
internal static func characterwiseIsEqual(
_ left: AttributedString.Guts,
to right: AttributedString.Guts
) -> Bool {
characterwiseIsEqual(left, in: left.stringBounds, to: right, in: right.stringBounds)
}
internal static func characterwiseIsEqual(
_ left: AttributedString.Guts, in leftRange: Range<BigString.Index>,
to right: AttributedString.Guts, in rightRange: Range<BigString.Index>
) -> Bool {
let leftRuns = AttributedString.Runs(left, in: leftRange)
let rightRuns = AttributedString.Runs(right, in: rightRange)
return _characterwiseIsEqual(leftRuns, to: rightRuns)
}
internal static func _characterwiseIsEqual(
_ left: AttributedString.Runs,
to right: AttributedString.Runs
) -> Bool {
// To decide if two attributed strings are equal, we need to logically split them up on
// run boundaries, then check that each pair of pieces contains the same attribute values
// and NFC-normalized string contents.
//
// Run lengths cannot be compared directly, as NFC normalization can change string length.
//
// We need to separately normalize each individual string piece. We cannot simply
// normalize the entire string up front, as that would blur attribute run boundaries
// (especially ones that fall inside Characters).
//
// Note: This implementation must be precisely in sync with the `characterwiseHash(in:into:)`
// implementation below.
if left._guts === right._guts, left._strBounds == right._strBounds { return true }
guard left.count == right.count else { return false }
var leftIndex = left._strBounds.lowerBound
var rightIndex = right._strBounds.lowerBound
var it1 = left.makeIterator()
var it2 = right.makeIterator()
loop:
while true {
switch (it1.next(), it2.next()) {
case let (leftRun?, rightRun?):
guard leftRun.attributes == rightRun.attributes else { return false }
let leftNext = left._guts.string.utf8.index(leftIndex, offsetBy: leftRun._utf8Count)
let rightNext = right._guts.string.utf8.index(rightIndex, offsetBy: rightRun._utf8Count)
// FIXME: This doesn't handle sub-character runs correctly.
guard
left._guts.string[leftIndex ..< leftNext] == right._guts.string[rightIndex ..< rightNext]
else {
return false
}
leftIndex = leftNext
rightIndex = rightNext
case (nil, nil):
break loop
default:
assertionFailure() // We compared counts above
return false
}
}
assert(leftIndex == left._strBounds.upperBound)
assert(rightIndex == right._strBounds.upperBound)
return true
}
internal func characterwiseHash(
in range: Range<BigString.Index>,
into hasher: inout Hasher
) {
// Note: This implementation must be precisely in sync with the `_characterwiseIsEqual`
// implementation above.
let runs = AttributedString.Runs(self, in: range)
hasher.combine(runs.count) // Hash discriminator
for run in runs {
hasher.combine(run._attributes)
// FIXME: This doesn't handle sub-character runs correctly.
hasher.combine(string[run._range])
}
}
}
extension AttributedString.Guts {
internal func description(in range: Range<BigString.Index>) -> String {
var result = ""
let runs = Runs(self, in: range)
for run in runs {
let text = String(self.string.unicodeScalars[run.range._bstringRange])
if !result.isEmpty { result += "\n" }
result += "\(text) \(run._attributes)"
}
return result
}
}
extension AttributedString.Guts {
var stringBounds: Range<BigString.Index> {
Range(uncheckedBounds: (string.startIndex, string.endIndex))
}
var utf8OffsetRange: Range<Int> {
0 ..< string.utf8.count
}
func utf8Index(at offset: Int) -> BigString.Index {
string.utf8.index(string.startIndex, offsetBy: offset)
}
func utf8IndexRange(from offsets: Range<Int>) -> Range<BigString.Index> {
let lower = utf8Index(at: offsets.lowerBound)
let upper = string.utf8.index(lower, offsetBy: offsets.count)
return Range(uncheckedBounds: (lower, upper))
}
func unicodeScalarRange(roundingDown range: Range<BigString.Index>) -> Range<BigString.Index> {
let lower = string.unicodeScalars.index(roundingDown: range.lowerBound)
let upper = string.unicodeScalars.index(roundingDown: range.upperBound)
return Range(uncheckedBounds: (lower, upper))
}
func characterRange(roundingDown range: Range<BigString.Index>) -> Range<BigString.Index> {
let lower = string.index(roundingDown: range.lowerBound)
let upper = string.index(roundingDown: range.upperBound)
return Range(uncheckedBounds: (lower, upper))
}
func index(afterRun i: BigString.Index) -> BigString.Index {
// Expected semantics: Result is the end of the run that contains `i`.
let index = self.runs.index(atUTF8Offset: i.utf8Offset).index
let length = self.runs[index].length
let next = self.string.utf8.index(i, offsetBy: index.utf8Offset + length - i.utf8Offset)
assert(next > i)
return next
}
func index(beforeRun i: BigString.Index) -> BigString.Index {
// Expected semantics: result is the start of the run preceding the one that contains `i`.
// (I.e., `i` needs to get implicitly rounded down to the nearest run boundary before we
// step back.)
let prev = self.string.utf8.index(before: i)
let index = self.runs.index(atUTF8Offset: prev.utf8Offset).index
let length = self.runs[index].length
if index.utf8Offset + length <= i.utf8Offset {
// Fast path: `i` already addresses a run boundary.
return self.string.utf8.index(prev, offsetBy: index.utf8Offset - prev.utf8Offset)
}
precondition(index > self.runs.startIndex, "Can't advance below start index")
let index2 = self.runs.index(before: index)
return self.string.utf8.index(prev, offsetBy: index2.utf8Offset - prev.utf8Offset)
}
internal func findRun(
at i: BigString.Index
) -> (runIndex: _InternalRuns.Index, start: BigString.Index) {
let run = self.runs.index(atUTF8Offset: i.utf8Offset)
let start = self.string.utf8.index(i, offsetBy: -run.remainingUTF8)
return (run.index, start)
}
/// Returns all the runs in the receiver, in the given range.
func runs(in utf8Bounds: Range<Int>) -> AttributedString._InternalRunsSlice {
AttributedString._InternalRunsSlice(self, utf8Bounds: utf8Bounds)
}
func runs(in range: Range<BigString.Index>) -> AttributedString._InternalRunsSlice {
return runs(in: range._utf8OffsetRange)
}
func run(at index: _InternalRuns.Index) -> (attributes: _AttributeStorage, utf8Range: Range<Int>) {
let run = runs[index]
let length = self.runs[index].length
let utf8Range = Range(uncheckedBounds: (index.utf8Offset, index.utf8Offset + length))
return (run.attributes, utf8Range)
}
/// Update the attribute dictionary at the specified index by invoking a closure on it.
/// If the index addresses a partial run at the beginning or end of the given slice, then the
/// underlying run is automatically split to accommodate the change.
///
/// Once `body` returns, the resulting item is coalesced with its neighbors if needed;
/// such coalescing may change the number of items in the slice.
///
/// On return, this function updates `index` to address the run that includes the UTF-8 range
/// of the original element. If the update needed to coalesce runs, the new `index` may address
/// a wider run than the original did.
///
/// If `body` does not end up actually mutating the attributes passed to it, then it may signal
/// this fact by setting its `mutated` argument to `false`. This avoid coalescing, so it will
/// slightly speed up execution.
func updateRun(
at index: inout _InternalRuns.Index,
within utf8Bounds: Range<Int>,
with body: (
_ attributes: inout _AttributeStorage,
_ utf8Range: Range<Int>,
_ mutated: inout Bool
) -> Void
) {
var (attributes, fullUTF8Range) = run(at: index)
let clampedUTF8Range = fullUTF8Range.clamped(to: utf8Bounds)
precondition(!clampedUTF8Range.isEmpty, "Index out of bounds")
var mutated = true
if clampedUTF8Range == fullUTF8Range {
// Allow in-place mutations of the attribute dictionary.
self.runs._update(at: &index) { $0.attributes = _AttributeStorage() }
body(&attributes, clampedUTF8Range, &mutated)
if mutated {
self.runs.updateAndCoalesce(at: &index) { $0 = attributes }
} else {
self.runs._update(at: &index) { $0.attributes = attributes }
}
} else {
body(&attributes, clampedUTF8Range, &mutated)
if mutated {
let run = _InternalRun(length: clampedUTF8Range.count, attributes: attributes)
self.runs.replaceUTF8Subrange(clampedUTF8Range, with: CollectionOfOne(run))
index = self.runs.index(atUTF8Offset: index.utf8Offset).index
} else {
return // nothing mutated
}
}
}
/// Get the uniform value for the specified key across the given range of indices.
/// Returns nil if the given range includes multiple different values for the same key.
func getUniformValue<K: AttributedStringKey>(
in range: Range<BigString.Index>, key: K.Type
) -> _AttributeValue? {
var result: _AttributeValue? = nil
for run in self.runs(in: range._utf8OffsetRange) {
guard let value = run.attributes[K.name] else {
return nil
}
if let previous = result, value != previous {
return nil
}
result = value
}
return result
}
/// Get all attributes that have consistent values across the given range of indices.
/// Attributes that have multiple different values are not included in the returned storage.
func getUniformValues(in range: Range<BigString.Index>) -> _AttributeStorage {
var attributes = _AttributeStorage()
var first = true
for run in self.runs(in: range._utf8OffsetRange) {
guard !first else {
attributes = run.attributes
first = false
continue
}
attributes = attributes.filterWithoutInvalidatingDependents {
guard let value = run.attributes[$0.key] else { return false }
return value == $0.value
}
if attributes.isEmpty {
break
}
}
return attributes
}
func setAttributeValue(
_ value: _AttributeValue,
forKey key: String,
in range: Range<BigString.Index>
) {
let utf8Range = unicodeScalarRange(roundingDown: range)._utf8OffsetRange
self.runs(in: utf8Range).updateEach { attributes, range, mutated in
attributes[key] = value
}
if value.hasConstrainedAttributes {
self.enforceAttributeConstraintsAfterMutation(
in: utf8Range, type: .attributes, constraintsInvolved: value.constraintsInvolved)
}
}
func setAttributeValue<K: AttributedStringKey>(
_ value: K.Value, forKey key: K.Type, in range: Range<BigString.Index>
) where K.Value : Sendable {
let value = _AttributeValue(value, for: K.self)
self.setAttributeValue(value, forKey: K.name, in: range)
}
func mergeAttributes(
_ attributes: AttributeContainer,
in range: Range<BigString.Index>,
mergePolicy: AttributeMergePolicy = .keepNew
) {
let new = attributes.storage
let utf8Range = unicodeScalarRange(roundingDown: range)._utf8OffsetRange
self.runs(in: utf8Range).updateEach { attributes, range, mutated in
attributes.mergeIn(new, mergePolicy: mergePolicy)
}
if new.hasConstrainedAttributes {
self.enforceAttributeConstraintsAfterMutation(
in: utf8Range, type: .attributes, constraintsInvolved: new.constraintsInvolved)
}
}
func setAttributes(_ attributes: _AttributeStorage, in range: Range<BigString.Index>) {
let utf8Range = unicodeScalarRange(roundingDown: range)._utf8OffsetRange
let run = _InternalRun(length: utf8Range.count, attributes: attributes)
self.runs.replaceUTF8Subrange(utf8Range, with: CollectionOfOne(run))
self.enforceAttributeConstraintsAfterMutation(
in: utf8Range,
type: .attributes,
constraintsInvolved: attributes.constraintsInvolved)
}
func removeAttributeValue<K: AttributedStringKey>(
forKey key: K.Type, in range: Range<BigString.Index>
) where K.Value: Sendable {
let utf8Range = unicodeScalarRange(roundingDown: range)._utf8OffsetRange
self.runs(in: utf8Range).updateEach { attributes, range, mutated in
attributes[K.self] = nil
}
if K.runBoundaries != nil {
self.enforceAttributeConstraintsAfterMutation(
in: utf8Range, type: .attributes, constraintsInvolved: K.constraintsInvolved)
}
}
func removeAttributeValue(forKey key: String, in range: Range<BigString.Index>) {
let utf8Range = unicodeScalarRange(roundingDown: range)._utf8OffsetRange
removeAttributeValue(forKey: key, in: utf8Range)
}
func removeAttributeValue(
forKey key: String, in utf8Range: Range<Int>, adjustConstrainedAttributes: Bool = true
) {
self.runs(in: utf8Range).updateEach { attributes, range, mutated in
attributes[key] = nil
}
if adjustConstrainedAttributes {
// FIXME: Collect boundary constraints.
self.enforceAttributeConstraintsAfterMutation(in: utf8Range, type: .attributes)
}
}
func _prepareStringMutation(
in range: Range<BigString.Index>
) -> (oldUTF8Count: Int, invalidationRange: Range<Int>) {
let utf8TargetRange = range._utf8OffsetRange
let invalidationRange = self.enforceAttributeConstraintsBeforeMutation(to: utf8TargetRange)
assert(invalidationRange.lowerBound <= utf8TargetRange.lowerBound)
assert(invalidationRange.upperBound >= utf8TargetRange.upperBound)
return (self.string.utf8.count, invalidationRange)
}
func _finalizeStringMutation(
_ state: (oldUTF8Count: Int, invalidationRange: Range<Int>)
) {
let utf8Delta = self.string.utf8.count - state.oldUTF8Count
let lower = state.invalidationRange.lowerBound
let upper = state.invalidationRange.upperBound + utf8Delta
self.enforceAttributeConstraintsAfterMutation(
in: lower ..< upper,
type: .attributesAndCharacters)
}
func replaceSubrange(
_ range: Range<BigString.Index>,
with replacement: some AttributedStringProtocol
) {
let replacementScalars = replacement.unicodeScalars._unicodeScalars
// Determine if this replacement is going to actively change character data, or if this is
// purely an attributes update, by seeing if the replacement string slice is identical to
// our own storage. (If it is identical, then we need to update attributes surrounding the
// affected bounds in a different way.)
//
// Note: this is intentionally not comparing actual string data.
let hasStringChanges = !replacementScalars.isIdentical(to: string.unicodeScalars[range])
let utf8SourceRange = Range(uncheckedBounds: (
replacementScalars.startIndex.utf8Offset,
replacementScalars.endIndex.utf8Offset
))
let replacementRuns = replacement.__guts.runs(in: utf8SourceRange)
let utf8TargetRange = range._utf8OffsetRange
if hasStringChanges {
let state = _prepareStringMutation(in: range)
self.string.unicodeScalars.replaceSubrange(range, with: replacementScalars)
self.runs.replaceUTF8Subrange(utf8TargetRange, with: replacementRuns)
_finalizeStringMutation(state)
} else {
self.runs.replaceUTF8Subrange(utf8TargetRange, with: replacementRuns)
self.enforceAttributeConstraintsAfterMutation(in: range._utf8OffsetRange, type: .attributes)
}
}
func attributesToUseForTextReplacement(in range: Range<BigString.Index>) -> _AttributeStorage {
guard !self.string.isEmpty else { return _AttributeStorage() }
var position = range.lowerBound
if range.isEmpty, position.utf8Offset > 0 {
position = self.string.utf8.index(before: position)
}
let runIndex = self.runs.index(atUTF8Offset: position.utf8Offset).index
let attributes = self.runs[runIndex].attributes
return attributes.attributesForAddedText()
}
}