michael@0: /* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
michael@0: /* This Source Code Form is subject to the terms of the Mozilla Public
michael@0:  * License, v. 2.0. If a copy of the MPL was not distributed with this
michael@0:  * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
michael@0: 
michael@0: #include "nsHyphenator.h"
michael@0: #include "nsIFile.h"
michael@0: #include "nsUTF8Utils.h"
michael@0: #include "nsUnicodeProperties.h"
michael@0: #include "nsUnicharUtilCIID.h"
michael@0: #include "nsIURI.h"
michael@0: 
michael@0: #include "hyphen.h"
michael@0: 
michael@0: nsHyphenator::nsHyphenator(nsIURI *aURI)
michael@0:   : mDict(nullptr)
michael@0: {
michael@0:   nsCString uriSpec;
michael@0:   nsresult rv = aURI->GetSpec(uriSpec);
michael@0:   if (NS_FAILED(rv)) {
michael@0:     return;
michael@0:   }
michael@0:   mDict = hnj_hyphen_load(uriSpec.get());
michael@0: #ifdef DEBUG
michael@0:   if (mDict) {
michael@0:     printf("loaded hyphenation patterns from %s\n", uriSpec.get());
michael@0:   }
michael@0: #endif
michael@0: }
michael@0: 
michael@0: nsHyphenator::~nsHyphenator()
michael@0: {
michael@0:   if (mDict != nullptr) {
michael@0:     hnj_hyphen_free((HyphenDict*)mDict);
michael@0:     mDict = nullptr;
michael@0:   }
michael@0: }
michael@0: 
michael@0: bool
michael@0: nsHyphenator::IsValid()
michael@0: {
michael@0:   return (mDict != nullptr);
michael@0: }
michael@0: 
michael@0: nsresult
michael@0: nsHyphenator::Hyphenate(const nsAString& aString,
michael@0:                         FallibleTArray<bool>& aHyphens)
michael@0: {
michael@0:   if (!aHyphens.SetLength(aString.Length())) {
michael@0:     return NS_ERROR_OUT_OF_MEMORY;
michael@0:   }
michael@0:   memset(aHyphens.Elements(), false, aHyphens.Length());
michael@0: 
michael@0:   bool inWord = false;
michael@0:   uint32_t wordStart = 0, wordLimit = 0;
michael@0:   uint32_t chLen;
michael@0:   for (uint32_t i = 0; i < aString.Length(); i += chLen) {
michael@0:     uint32_t ch = aString[i];
michael@0:     chLen = 1;
michael@0: 
michael@0:     if (NS_IS_HIGH_SURROGATE(ch)) {
michael@0:       if (i + 1 < aString.Length() && NS_IS_LOW_SURROGATE(aString[i+1])) {
michael@0:         ch = SURROGATE_TO_UCS4(ch, aString[i+1]);
michael@0:         chLen = 2;
michael@0:       } else {
michael@0:         NS_WARNING("unpaired surrogate found during hyphenation");
michael@0:       }
michael@0:     }
michael@0: 
michael@0:     nsIUGenCategory::nsUGenCategory cat = mozilla::unicode::GetGenCategory(ch);
michael@0:     if (cat == nsIUGenCategory::kLetter || cat == nsIUGenCategory::kMark) {
michael@0:       if (!inWord) {
michael@0:         inWord = true;
michael@0:         wordStart = i;
michael@0:       }
michael@0:       wordLimit = i + chLen;
michael@0:       if (i + chLen < aString.Length()) {
michael@0:         continue;
michael@0:       }
michael@0:     }
michael@0: 
michael@0:     if (inWord) {
michael@0:       const char16_t *begin = aString.BeginReading();
michael@0:       NS_ConvertUTF16toUTF8 utf8(begin + wordStart,
michael@0:                                  wordLimit - wordStart);
michael@0:       nsAutoTArray<char,200> utf8hyphens;
michael@0:       utf8hyphens.SetLength(utf8.Length() + 5);
michael@0:       char **rep = nullptr;
michael@0:       int *pos = nullptr;
michael@0:       int *cut = nullptr;
michael@0:       int err = hnj_hyphen_hyphenate2((HyphenDict*)mDict,
michael@0:                                       utf8.BeginReading(), utf8.Length(),
michael@0:                                       utf8hyphens.Elements(), nullptr,
michael@0:                                       &rep, &pos, &cut);
michael@0:       if (!err) {
michael@0:         // Surprisingly, hnj_hyphen_hyphenate2 converts the 'hyphens' buffer
michael@0:         // from utf8 code unit indexing (which would match the utf8 input
michael@0:         // string directly) to Unicode character indexing.
michael@0:         // We then need to convert this to utf16 code unit offsets for Gecko.
michael@0:         const char *hyphPtr = utf8hyphens.Elements();
michael@0:         const char16_t *cur = begin + wordStart;
michael@0:         const char16_t *end = begin + wordLimit;
michael@0:         while (cur < end) {
michael@0:           if (*hyphPtr & 0x01) {
michael@0:             aHyphens[cur - begin] = true;
michael@0:           }
michael@0:           cur++;
michael@0:           if (cur < end && NS_IS_LOW_SURROGATE(*cur) &&
michael@0:               NS_IS_HIGH_SURROGATE(*(cur-1)))
michael@0:           {
michael@0:             cur++;
michael@0:           }
michael@0:           hyphPtr++;
michael@0:         }
michael@0:       }
michael@0:     }
michael@0:     
michael@0:     inWord = false;
michael@0:   }
michael@0: 
michael@0:   return NS_OK;
michael@0: }