1 /* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
2 /* This Source Code Form is subject to the terms of the Mozilla Public
3 * License, v. 2.0. If a copy of the MPL was not distributed with this
4 * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
5
6 #include "mozEnglishWordUtils.h"
7 #include "nsComponentManagerUtils.h"
8 #include "nsReadableUtils.h"
9 #include "nsUnicharUtils.h"
10 #include "nsUnicodeProperties.h"
11 #include "nsCRT.h"
12 #include "mozilla/Likely.h"
13 #include "nsMemory.h"
14
NS_IMPL_CYCLE_COLLECTION_ROOT_NATIVE(mozEnglishWordUtils,AddRef)15 NS_IMPL_CYCLE_COLLECTION_ROOT_NATIVE(mozEnglishWordUtils, AddRef)
16 NS_IMPL_CYCLE_COLLECTION_UNROOT_NATIVE(mozEnglishWordUtils, Release)
17
18 NS_IMPL_CYCLE_COLLECTION(mozEnglishWordUtils, mURLDetector)
19
20 mozEnglishWordUtils::mozEnglishWordUtils() {
21 mURLDetector = do_CreateInstance(MOZ_TXTTOHTMLCONV_CONTRACTID);
22 }
23
~mozEnglishWordUtils()24 mozEnglishWordUtils::~mozEnglishWordUtils() {}
25
26 // This needs vast improvement
27
28 // static
ucIsAlpha(char16_t aChar)29 bool mozEnglishWordUtils::ucIsAlpha(char16_t aChar) {
30 // XXX we have to fix callers to handle the full Unicode range
31 return nsUGenCategory::kLetter == mozilla::unicode::GetGenCategory(aChar);
32 }
33
FindNextWord(const nsAString & aWord,uint32_t offset,int32_t * begin,int32_t * end)34 bool mozEnglishWordUtils::FindNextWord(const nsAString& aWord, uint32_t offset,
35 int32_t* begin, int32_t* end) {
36 if (offset >= aWord.Length()) {
37 *begin = -1;
38 *end = -1;
39 return false;
40 }
41
42 const char16_t* word = aWord.BeginReading();
43 uint32_t length = aWord.Length();
44 const char16_t* p = word + offset;
45 const char16_t* endbuf = word + length;
46 const char16_t* startWord = p;
47
48 // XXX These loops should be modified to handle non-BMP characters.
49 // if previous character is a word character, need to advance out of the
50 // word
51 if (offset > 0 && ucIsAlpha(*(p - 1))) {
52 while (p < endbuf && ucIsAlpha(*p)) {
53 p++;
54 }
55 }
56 while ((p < endbuf) && (!ucIsAlpha(*p))) {
57 p++;
58 }
59 startWord = p;
60 while ((p < endbuf) && ((ucIsAlpha(*p)) || (*p == '\''))) {
61 p++;
62 }
63
64 // we could be trying to break down a url, we don't want to break a url into
65 // parts, instead we want to find out if it really is a url and if so, skip
66 // it, advancing startWord to a point after the url.
67
68 // before we spend more time looking to see if the word is a url, look for a
69 // url identifer and make sure that identifer isn't the last character in
70 // the word fragment.
71 if ((p < endbuf - 1) && (*p == ':' || *p == '@' || *p == '.')) {
72 // ok, we have a possible url...do more research to find out if we really
73 // have one and determine the length of the url so we can skip over it.
74
75 if (mURLDetector) {
76 int32_t startPos = -1;
77 int32_t endPos = -1;
78
79 mURLDetector->FindURLInPlaintext(startWord, endbuf - startWord,
80 p - startWord, &startPos, &endPos);
81
82 // ok, if we got a url, adjust the array bounds, skip the current url
83 // text and find the next word again
84 if (startPos != -1 && endPos != -1) {
85 startWord = p + endPos + 1; // skip over the url
86
87 // now recursively call FindNextWord to search for the next word now
88 // that we have skipped the url
89 return FindNextWord(aWord, startWord - word, begin, end);
90 }
91 }
92 }
93
94 while ((p > startWord) && (*(p - 1) == '\'')) { // trim trailing apostrophes
95 p--;
96 }
97
98 if (startWord == endbuf) {
99 *begin = -1;
100 *end = -1;
101 return false;
102 }
103 *begin = startWord - word;
104 *end = p - word;
105 return true;
106 }
107