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 "nsSemanticUnitScanner.h"
michael@0: 
michael@0: NS_IMPL_ISUPPORTS(nsSemanticUnitScanner, nsISemanticUnitScanner)
michael@0: 
michael@0: nsSemanticUnitScanner::nsSemanticUnitScanner() : nsSampleWordBreaker()
michael@0: {
michael@0:   /* member initializers and constructor code */
michael@0: }
michael@0: 
michael@0: nsSemanticUnitScanner::~nsSemanticUnitScanner()
michael@0: {
michael@0:   /* destructor code */
michael@0: }
michael@0: 
michael@0: 
michael@0: /* void start (in string characterSet); */
michael@0: NS_IMETHODIMP nsSemanticUnitScanner::Start(const char *characterSet)
michael@0: {
michael@0:     // do nothing for now.
michael@0:     return NS_OK;
michael@0: }
michael@0: 
michael@0: /* void next (in wstring text, in long length, in long pos, out boolean hasMoreUnits, out long begin, out long end); */
michael@0: NS_IMETHODIMP nsSemanticUnitScanner::Next(const char16_t *text, int32_t length, int32_t pos, bool isLastBuffer, int32_t *begin, int32_t *end, bool *_retval)
michael@0: {
michael@0:     // xxx need to bullet proff and check input pointer 
michael@0:     //  make sure begin, end and _retval is not nullptr here
michael@0: 
michael@0:     // if we reach the end, just return
michael@0:     if (pos >= length) {
michael@0:        *begin = pos;
michael@0:        *end = pos;
michael@0:        *_retval = false;
michael@0:        return NS_OK;
michael@0:     }
michael@0: 
michael@0:     uint8_t char_class = nsSampleWordBreaker::GetClass(text[pos]);
michael@0: 
michael@0:     // if we are in chinese mode, return one han letter at a time
michael@0:     // we should not do this if we are in Japanese or Korean mode
michael@0:     if (kWbClassHanLetter == char_class) {
michael@0:        *begin = pos;
michael@0:        *end = pos+1;
michael@0:        *_retval = true;
michael@0:        return NS_OK;
michael@0:     }
michael@0: 
michael@0:     int32_t next;
michael@0:     // find the next "word"
michael@0:     next = NextWord(text, (uint32_t) length, (uint32_t) pos);
michael@0: 
michael@0:     // if we don't have enough text to make decision, return 
michael@0:     if (next == NS_WORDBREAKER_NEED_MORE_TEXT) {
michael@0:        *begin = pos;
michael@0:        *end = isLastBuffer ? length : pos;
michael@0:        *_retval = isLastBuffer;
michael@0:        return NS_OK;
michael@0:     } 
michael@0:     
michael@0:     // if what we got is space or punct, look at the next break
michael@0:     if ((char_class == kWbClassSpace) || (char_class == kWbClassPunct)) {
michael@0:         // if the next "word" is not letters, 
michael@0:         // call itself recursively with the new pos
michael@0:         return Next(text, length, next, isLastBuffer, begin, end, _retval);
michael@0:     }
michael@0: 
michael@0:     // for the rest, return 
michael@0:     *begin = pos;
michael@0:     *end = next;
michael@0:     *_retval = true;
michael@0:     return NS_OK;
michael@0: }
michael@0: