Пример #1
0
        /// чтение файла со словами
        /// path - полный путь к файлу
        /// nounType - тип существительного
        private void ReadWords(string filename, MorphoAttributeEnum nounType)
        {
            var lines = ReadFile(filename);

            foreach (var line in lines)
            {
                var array = line.Split(WORDS_DICTIONARY_SEPARATOR, StringSplitOptions.RemoveEmptyEntries);
                if (array.Length != 3)
                {
                    _ModelLoadingErrorCallback("Wrong line format", line);
                    continue;
                }

                MorphoType morphoType = GetMorphoTypeByName(array[1]);
                if (morphoType == null)
                {
                    _ModelLoadingErrorCallback("Unknown morpho-type", line);
                }
                else
                if (array[2] != _PartOfSpeechStringDictionary[morphoType.PartOfSpeech])
                {
                    _ModelLoadingErrorCallback("Wrong part-of-speech", line);
                }
                else
                {
                    var word = array[0];

                    var _nounType = default(MorphoAttributePair?);
                    if ((morphoType.MorphoAttributeGroup & MorphoAttributeGroupEnum.NounType) == MorphoAttributeGroupEnum.NounType)
                    {
                        _nounType = _MorphoAttributeList.GetMorphoAttributePair(MorphoAttributeGroupEnum.NounType, nounType);
                    }
                    _TreeDictionary.AddWord(word, morphoType, _nounType);
                }
            }
        }
Пример #2
0
            /// чтение файла со словами
            /// path - полный путь к файлу
            /// nounType - тип существи тельного
            private void ReadWords(string filename, MorphoAttributeEnum nounType)
            {
                var lines = ReadFile(filename);

                var plw = default(ParsedLineWords_unsafe);

                foreach (var line in lines)
                {
                    fixed(char *lineBase = line)
                    {
                        if (!ParseLineWords(lineBase, ref plw))
                        {
                            _ModelLoadingErrorCallback("Wrong line format", line);
                            continue;
                        }

                        MorphoTypeNative morphoType = GetMorphoTypeByName((IntPtr)plw.MorphoTypeName);

                        if (morphoType == null)
                        {
                            _ModelLoadingErrorCallback("Unknown morpho-type", line);
                            continue;
                        }

                        if (!StringsHelper.IsEqual((IntPtr)plw.PartOfSpeech, _PartOfSpeechToNativeStringMapper[morphoType.PartOfSpeech]))
                        {
                            _ModelLoadingErrorCallback("Wrong part-of-speech", line);
                            continue;
                        }

                        if (morphoType.HasMorphoForms)
                        {
                            var nounTypePair = default(MorphoAttributePair?);
                            if ((morphoType.MorphoAttributeGroup & MorphoAttributeGroupEnum.NounType) == MorphoAttributeGroupEnum.NounType)
                            {
                                nounTypePair = _MorphoAttributeList.GetMorphoAttributePair(MorphoAttributeGroupEnum.NounType, nounType);
                            }

                            #region Allocate native-memory for baseOfWord
                            var len = plw.WordLength - StringsHelper.GetLength(morphoType.FirstEnding);
                            len = ((0 <= len) ? len : plw.WordLength);

                            IntPtr lineBasePtr;
                            if (0 < len)
                            {
                                *(lineBase + len) = '\0';
                                lineBasePtr       = new IntPtr(lineBase);

                                if (_EndingDictionary.TryGetValue(lineBasePtr, out IntPtr existsPtr))
                                {
                                    lineBasePtr = existsPtr;
                                }
                                else
                                {
                                    AllocHGlobalAndCopy(lineBase, len, out lineBasePtr);
                                    _EndingDictionary.Add(lineBasePtr, lineBasePtr);
                                }
                            }
                            else
                            {
                                lineBasePtr = _EMPTY_STRING;
                            }
                            #endregion

                            _TreeDictionary.AddWord((char *)lineBasePtr, morphoType, ref nounTypePair);
                        }
                    }
                }
            }