1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/extensions/spellcheck/src/mozInlineSpellWordUtil.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,166 @@
1.4 +/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
1.5 +/* This Source Code Form is subject to the terms of the Mozilla Public
1.6 + * License, v. 2.0. If a copy of the MPL was not distributed with this
1.7 + * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
1.8 +
1.9 +#ifndef mozInlineSpellWordUtil_h
1.10 +#define mozInlineSpellWordUtil_h
1.11 +
1.12 +#include "nsCOMPtr.h"
1.13 +#include "nsIDOMDocument.h"
1.14 +#include "nsIDocument.h"
1.15 +#include "nsString.h"
1.16 +#include "nsTArray.h"
1.17 +
1.18 +//#define DEBUG_SPELLCHECK
1.19 +
1.20 +class nsRange;
1.21 +class nsINode;
1.22 +
1.23 +/**
1.24 + * This class extracts text from the DOM and builds it into a single string.
1.25 + * The string includes whitespace breaks whereever non-inline elements begin
1.26 + * and end. This string is broken into "real words", following somewhat
1.27 + * complex rules; for example substrings that look like URLs or
1.28 + * email addresses are treated as single words, but otherwise many kinds of
1.29 + * punctuation are treated as word separators. GetNextWord provides a way
1.30 + * to iterate over these "real words".
1.31 + *
1.32 + * The basic operation is:
1.33 + *
1.34 + * 1. Call Init with the weak pointer to the editor that you're using.
1.35 + * 2. Call SetEnd to set where you want to stop spellchecking. We'll stop
1.36 + * at the word boundary after that. If SetEnd is not called, we'll stop
1.37 + * at the end of the document's root element.
1.38 + * 3. Call SetPosition to initialize the current position inside the
1.39 + * previously given range.
1.40 + * 4. Call GetNextWord over and over until it returns false.
1.41 + */
1.42 +
1.43 +class mozInlineSpellWordUtil
1.44 +{
1.45 +public:
1.46 + struct NodeOffset {
1.47 + nsINode* mNode;
1.48 + int32_t mOffset;
1.49 +
1.50 + NodeOffset(nsINode* aNode, int32_t aOffset) :
1.51 + mNode(aNode), mOffset(aOffset) {}
1.52 + };
1.53 +
1.54 + mozInlineSpellWordUtil()
1.55 + : mRootNode(nullptr),
1.56 + mSoftBegin(nullptr, 0), mSoftEnd(nullptr, 0),
1.57 + mNextWordIndex(-1), mSoftTextValid(false) {}
1.58 +
1.59 + nsresult Init(nsWeakPtr aWeakEditor);
1.60 +
1.61 + nsresult SetEnd(nsINode* aEndNode, int32_t aEndOffset);
1.62 +
1.63 + // sets the current position, this should be inside the range. If we are in
1.64 + // the middle of a word, we'll move to its start.
1.65 + nsresult SetPosition(nsINode* aNode, int32_t aOffset);
1.66 +
1.67 + // Given a point inside or immediately following a word, this returns the
1.68 + // DOM range that exactly encloses that word's characters. The current
1.69 + // position will be at the end of the word. This will find the previous
1.70 + // word if the current position is space, so if you care that the point is
1.71 + // inside the word, you should check the range.
1.72 + //
1.73 + // THIS CHANGES THE CURRENT POSITION AND RANGE. It is designed to be called
1.74 + // before you actually generate the range you are interested in and iterate
1.75 + // the words in it.
1.76 + nsresult GetRangeForWord(nsIDOMNode* aWordNode, int32_t aWordOffset,
1.77 + nsRange** aRange);
1.78 +
1.79 + // Moves to the the next word in the range, and retrieves it's text and range.
1.80 + // An empty word and a nullptr range are returned when we are done checking.
1.81 + // aSkipChecking will be set if the word is "special" and shouldn't be
1.82 + // checked (e.g., an email address).
1.83 + nsresult GetNextWord(nsAString& aText, nsRange** aRange,
1.84 + bool* aSkipChecking);
1.85 +
1.86 + // Call to normalize some punctuation. This function takes an autostring
1.87 + // so we can access characters directly.
1.88 + static void NormalizeWord(nsSubstring& aWord);
1.89 +
1.90 + nsIDOMDocument* GetDOMDocument() const { return mDOMDocument; }
1.91 + nsIDocument* GetDocument() const { return mDocument; }
1.92 + nsINode* GetRootNode() { return mRootNode; }
1.93 +
1.94 +private:
1.95 +
1.96 + // cached stuff for the editor, set by Init
1.97 + nsCOMPtr<nsIDOMDocument> mDOMDocument;
1.98 + nsCOMPtr<nsIDocument> mDocument;
1.99 +
1.100 + // range to check, see SetPosition and SetEnd
1.101 + nsINode* mRootNode;
1.102 + NodeOffset mSoftBegin;
1.103 + NodeOffset mSoftEnd;
1.104 +
1.105 + // DOM text covering the soft range, with newlines added at block boundaries
1.106 + nsString mSoftText;
1.107 + // A list of where we extracted text from, ordered by mSoftTextOffset. A given
1.108 + // DOM node appears at most once in this list.
1.109 + struct DOMTextMapping {
1.110 + NodeOffset mNodeOffset;
1.111 + int32_t mSoftTextOffset;
1.112 + int32_t mLength;
1.113 +
1.114 + DOMTextMapping(NodeOffset aNodeOffset, int32_t aSoftTextOffset, int32_t aLength)
1.115 + : mNodeOffset(aNodeOffset), mSoftTextOffset(aSoftTextOffset),
1.116 + mLength(aLength) {}
1.117 + };
1.118 + nsTArray<DOMTextMapping> mSoftTextDOMMapping;
1.119 +
1.120 + // A list of the "real words" in mSoftText, ordered by mSoftTextOffset
1.121 + struct RealWord {
1.122 + int32_t mSoftTextOffset;
1.123 + int32_t mLength;
1.124 + bool mCheckableWord;
1.125 +
1.126 + RealWord(int32_t aOffset, int32_t aLength, bool aCheckable)
1.127 + : mSoftTextOffset(aOffset), mLength(aLength), mCheckableWord(aCheckable) {}
1.128 + int32_t EndOffset() const { return mSoftTextOffset + mLength; }
1.129 + };
1.130 + nsTArray<RealWord> mRealWords;
1.131 + int32_t mNextWordIndex;
1.132 +
1.133 + bool mSoftTextValid;
1.134 +
1.135 + void InvalidateWords() { mSoftTextValid = false; }
1.136 + void EnsureWords();
1.137 +
1.138 + int32_t MapDOMPositionToSoftTextOffset(NodeOffset aNodeOffset);
1.139 + // Map an offset into mSoftText to a DOM position. Note that two DOM positions
1.140 + // can map to the same mSoftText offset, e.g. given nodes A=aaaa and B=bbbb
1.141 + // forming aaaabbbb, (A,4) and (B,0) give the same string offset. So,
1.142 + // aHintBefore controls which position we return ... if aHint is eEnd
1.143 + // then the position indicates the END of a range so we return (A,4). Otherwise
1.144 + // the position indicates the START of a range so we return (B,0).
1.145 + enum DOMMapHint { HINT_BEGIN, HINT_END };
1.146 + NodeOffset MapSoftTextOffsetToDOMPosition(int32_t aSoftTextOffset,
1.147 + DOMMapHint aHint);
1.148 + // Finds the index of the real word containing aSoftTextOffset, or -1 if none
1.149 + // If it's exactly between two words, then if aHint is HINT_BEGIN, return the
1.150 + // later word (favouring the assumption that it's the BEGINning of a word),
1.151 + // otherwise return the earlier word (assuming it's the END of a word).
1.152 + // If aSearchForward is true, then if we don't find a word at the given
1.153 + // position, search forward until we do find a word and return that (if found).
1.154 + int32_t FindRealWordContaining(int32_t aSoftTextOffset, DOMMapHint aHint,
1.155 + bool aSearchForward);
1.156 +
1.157 + // build mSoftText and mSoftTextDOMMapping
1.158 + void BuildSoftText();
1.159 + // Build mRealWords array
1.160 + void BuildRealWords();
1.161 +
1.162 + void SplitDOMWord(int32_t aStart, int32_t aEnd);
1.163 +
1.164 + // Convenience functions, object must be initialized
1.165 + nsresult MakeRange(NodeOffset aBegin, NodeOffset aEnd, nsRange** aRange);
1.166 + nsresult MakeRangeForWord(const RealWord& aWord, nsRange** aRange);
1.167 +};
1.168 +
1.169 +#endif
