369be69fa0
This patch refactors the find functionality in a few ways: - A C++-only method `FindFromRangeBoundaries()` is introduced, which uses `RangeBoundary`s instead of collapsed `nsRange`s to describe the search range. - Use `nsContentUtils::ComparePoints()` to check whether the end of the search range has been reached. This is more robust than the previous check and finds the end point also when it's in a node which is being skipped (eg. a non-text-node). - Clarify the interface: The search start and end points are now always in document read order, and independent of the search direction. Therefore, the start point must always be before the end (this is asserted). All usages of the find code are updated accordingly. - Some bugs in the tests, which have surfaced because of the broken end point implementation have been fixed. Differential Revision: https://phabricator.services.mozilla.com/D237497
82 lines
2.7 KiB
C++
82 lines
2.7 KiB
C++
/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
|
|
/* vim: set ts=8 sts=2 et sw=2 tw=80: */
|
|
/* This Source Code Form is subject to the terms of the Mozilla Public
|
|
* License, v. 2.0. If a copy of the MPL was not distributed with this
|
|
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
|
|
|
|
#ifndef nsFind_h__
|
|
#define nsFind_h__
|
|
|
|
#include "nsIFind.h"
|
|
|
|
#include "nsCOMPtr.h"
|
|
#include "nsCycleCollectionParticipant.h"
|
|
#include "nsINode.h"
|
|
#include "mozilla/intl/Segmenter.h"
|
|
#include "mozilla/RangeBoundary.h"
|
|
|
|
#define NS_FIND_CONTRACTID "@mozilla.org/embedcomp/rangefind;1"
|
|
|
|
#define NS_FIND_CID \
|
|
{0x471f4944, 0x1dd2, 0x11b2, {0x87, 0xac, 0x90, 0xbe, 0x0a, 0x51, 0xd6, 0x09}}
|
|
|
|
class nsFind : public nsIFind {
|
|
public:
|
|
NS_DECL_CYCLE_COLLECTING_ISUPPORTS
|
|
NS_DECL_NSIFIND
|
|
NS_DECL_CYCLE_COLLECTION_CLASS(nsFind)
|
|
|
|
// The IDL interface allows to toggle the "find entire words" search mode
|
|
// by calling `SetEntireWord()`.
|
|
// Internally, it is possible to specify word boundaries at the beginning
|
|
// and/or the end of the search pattern.
|
|
// `GetEntireWord()` returns true if `mWordStartBounded` and `mWordEndBounded`
|
|
// are true.
|
|
void SetWordStartBounded(bool aWordStartBounded) {
|
|
mWordStartBounded = aWordStartBounded;
|
|
}
|
|
void SetWordEndBounded(bool aWordEndBounded) {
|
|
mWordEndBounded = aWordEndBounded;
|
|
}
|
|
|
|
void SetNodeIndexCache(nsContentUtils::NodeIndexCache* aCache) {
|
|
mNodeIndexCache = aCache;
|
|
}
|
|
|
|
already_AddRefed<nsRange> FindFromRangeBoundaries(
|
|
const nsAString& aPatText, const mozilla::RangeBoundary& aStartPoint,
|
|
const mozilla::RangeBoundary& aEndPoint);
|
|
|
|
protected:
|
|
virtual ~nsFind() = default;
|
|
|
|
// Parameters set from the interface:
|
|
bool mFindBackward = false;
|
|
bool mCaseSensitive = false;
|
|
bool mMatchDiacritics = false;
|
|
|
|
bool mWordStartBounded = false;
|
|
bool mWordEndBounded = false;
|
|
mozilla::intl::WordBreakIteratorUtf16 mWordBreakIter{nullptr};
|
|
nsContentUtils::NodeIndexCache* mNodeIndexCache = nullptr;
|
|
struct State;
|
|
class StateRestorer;
|
|
|
|
// Extract a character from a string, handling surrogate pairs and
|
|
// incrementing the index if a surrogate pair is encountered
|
|
char32_t DecodeChar(const char16_t* t2b, int32_t* index) const;
|
|
|
|
// Determine if a line break can occur between two characters
|
|
//
|
|
// This could be improved because some languages require more context than two
|
|
// characters to determine where line breaks can occur
|
|
bool BreakInBetween(char32_t x, char32_t y);
|
|
|
|
// Get the first character from the next node (last if mFindBackward).
|
|
//
|
|
// This will mutate the state, but then restore it afterwards.
|
|
char32_t PeekNextChar(State&, bool aAlreadyMatching) const;
|
|
};
|
|
|
|
#endif // nsFind_h__
|