Module: Nlpir

Defined in:
lib/nlpir.rb,
lib/nlpir/version.rb

Constant Summary collapse

NLPIR_FALSE = 
0
NLPIR_TRUE = 
1
POS_MAP_NUMBER = 
4
ICT_POS_MAP_FIRST =

计算所一级标注集

1
ICT_POS_MAP_SECOND =

计算所二级标注集

0
PKU_POS_MAP_SECOND =

北大二级标注集

2
PKU_POS_MAP_FIRST =

北大一级标注集

3
POS_SIZE = 
40
Result_t = 
struct ['int start','int length',"char  sPOS[#{POS_SIZE}]",'int iPOS',
'int word_ID','int word_type','double weight']
GBK_CODE =

默认支持GBK编码

0
UTF8_CODE =

UTF8编码

GBK_CODE + 1
BIG5_CODE =

BIG5编码

GBK_CODE + 2
GBK_FANTI_CODE =

GBK编码，里面包含繁体字

GBK_CODE + 3
NLPIR_Init_rb = 
Fiddle::Function.new(
  libm['_Z10NLPIR_InitPKci'],
  [Fiddle::TYPE_VOIDP,Fiddle::TYPE_INT],
  Fiddle::TYPE_INT
)
NLPIR_Exit_rb = 
Fiddle::Function.new(
  libm['_Z10NLPIR_Exitv'],
  [],
  Fiddle::TYPE_INT
)
NLPIR_ImportUserDict_rb = 
Fiddle::Function.new(
  libm['_Z20NLPIR_ImportUserDictPKc'],
  [Fiddle::TYPE_VOIDP],
  Fiddle::TYPE_INT
)
NLPIR_ParagraphProcess_rb = 
Fiddle::Function.new(
  libm['_Z22NLPIR_ParagraphProcessPKci'],
  [Fiddle::TYPE_VOIDP,Fiddle::TYPE_INT],
  Fiddle::TYPE_VOIDP
)
NLPIR_ParagraphProcessA_rb = 
Fiddle::Function.new(
  libm['_Z23NLPIR_ParagraphProcessAPKcPib'],
  [Fiddle::TYPE_VOIDP,Fiddle::TYPE_VOIDP],
  Fiddle::TYPE_VOIDP
)
NLPIR_FileProcess_rb = 
Fiddle::Function.new(
  libm['_Z17NLPIR_FileProcessPKcS0_i'],
  [Fiddle::TYPE_VOIDP,Fiddle::TYPE_VOIDP, Fiddle::TYPE_INT],
  Fiddle::TYPE_DOUBLE
)
NLPIR_GetParagraphProcessAWordCount_rb = 
Fiddle::Function.new(
  libm['_Z35NLPIR_GetParagraphProcessAWordCountPKc'],
  [Fiddle::TYPE_VOIDP],
  Fiddle::TYPE_INT
)
NLPIR_ParagraphProcessAW_rb = 
Fiddle::Function.new(
  libm['_Z24NLPIR_ParagraphProcessAWiP8result_t'],
  [Fiddle::TYPE_INT,Fiddle::TYPE_VOIDP],
  Fiddle::TYPE_INT
)
NLPIR_AddUserWord_rb = 
Fiddle::Function.new(
  libm['_Z17NLPIR_AddUserWordPKc'],
  [Fiddle::TYPE_VOIDP],
  Fiddle::TYPE_INT
)
NLPIR_SaveTheUsrDic_rb = 
Fiddle::Function.new(
  libm['_Z19NLPIR_SaveTheUsrDicv'],
  [],
  Fiddle::TYPE_INT
)
NLPIR_DelUsrWord_rb = 
Fiddle::Function.new(
  libm['_Z16NLPIR_DelUsrWordPKc'],
  [Fiddle::TYPE_VOIDP],
  Fiddle::TYPE_INT
)
NLPIR_GetKeyWords_rb = 
Fiddle::Function.new(
  libm['_Z17NLPIR_GetKeyWordsPKcib'],
  [Fiddle::TYPE_VOIDP,Fiddle::TYPE_INT,Fiddle::TYPE_INT],
  Fiddle::TYPE_VOIDP
)
NLPIR_GetFileKeyWords_rb = 
Fiddle::Function.new(
  libm['_Z21NLPIR_GetFileKeyWordsPKcib'],
  [Fiddle::TYPE_VOIDP,Fiddle::TYPE_INT,Fiddle::TYPE_INT],
  Fiddle::TYPE_VOIDP
)
NLPIR_GetNewWords_rb = 
Fiddle::Function.new(
  libm['_Z17NLPIR_GetNewWordsPKcib'],
  [Fiddle::TYPE_VOIDP,Fiddle::TYPE_INT,Fiddle::TYPE_INT],
  Fiddle::TYPE_VOIDP
)
NLPIR_GetFileNewWords_rb = 
Fiddle::Function.new(
  libm['_Z21NLPIR_GetFileNewWordsPKcib'],
  [Fiddle::TYPE_VOIDP,Fiddle::TYPE_INT,Fiddle::TYPE_INT],
  Fiddle::TYPE_VOIDP
)
NLPIR_FingerPrint_rb = 
Fiddle::Function.new(
  libm['_Z17NLPIR_FingerPrintPKc'],
  [Fiddle::TYPE_VOIDP],
  Fiddle::TYPE_LONG
)
NLPIR_SetPOSmap_rb = 
Fiddle::Function.new(
  libm['_Z15NLPIR_SetPOSmapi'],
  [Fiddle::TYPE_INT],
  Fiddle::TYPE_INT
)
NLPIR_NWI_Start_rb = 
Fiddle::Function.new(
  libm['_Z15NLPIR_NWI_Startv'],
  [],
  Fiddle::TYPE_INT
)
NLPIR_NWI_AddFile_rb = 
Fiddle::Function.new(
  libm['_Z17NLPIR_NWI_AddFilePKc'],
  [Fiddle::TYPE_VOIDP],
  Fiddle::TYPE_INT
)
NLPIR_NWI_AddMem_rb = 
Fiddle::Function.new(
  libm['_Z16NLPIR_NWI_AddMemPKc'],
  [Fiddle::TYPE_VOIDP],
  Fiddle::TYPE_INT
)
NLPIR_NWI_Complete_rb = 
Fiddle::Function.new(
  libm['_Z18NLPIR_NWI_Completev'],
  [],
  Fiddle::TYPE_INT
)
NLPIR_NWI_GetResult_rb = 
Fiddle::Function.new(
  libm['_Z19NLPIR_NWI_GetResultb'],
  [Fiddle::TYPE_INT],
  Fiddle::TYPE_VOIDP
)
NLPIR_NWI_Result2UserDict_rb = 
Fiddle::Function.new(
  libm['_Z25NLPIR_NWI_Result2UserDictv'],
  [],
  Fiddle::TYPE_VOIDP
)
VERSION = 
"0.0.3"

Instance Method Summary collapse

Instance Method Details

#NLPIR_AddUserWord(sWord) ⇒ Object 



214
215
216
 
# File 'lib/nlpir.rb', line 214

def NLPIR_AddUserWord(sWord)
  NLPIR_AddUserWord_rb.call(sWord)
end

#NLPIR_DelUsrWord(sWord) ⇒ Object 



222
223
224
 
# File 'lib/nlpir.rb', line 222

def NLPIR_DelUsrWord(sWord)
  NLPIR_DelUsrWord_rb.call(sWord)
end

#NLPIR_Exit ⇒ Object 



168
169
170
 
# File 'lib/nlpir.rb', line 168

def NLPIR_Exit()
  NLPIR_Exit_rb.call()
end

#NLPIR_FileProcess(sSourceFilename, sResultFilename, bPOStagged = NLPIR_TRUE) ⇒ Object 



193
194
195
 
# File 'lib/nlpir.rb', line 193

def NLPIR_FileProcess(sSourceFilename, sResultFilename, bPOStagged=NLPIR_TRUE)
  NLPIR_FileProcess_rb.call(sSourceFilename, sResultFilename, bPOStagged)
end

#NLPIR_FingerPrint(sLine) ⇒ Object 



242
243
244
 
# File 'lib/nlpir.rb', line 242

def NLPIR_FingerPrint(sLine)
  NLPIR_FingerPrint_rb.call(sLine)
end

#NLPIR_GetFileKeyWords(sTextFile, nMaxKeyLimit = 50, bWeightOut = NLPIR_FALSE) ⇒ Object 



230
231
232
 
# File 'lib/nlpir.rb', line 230

def NLPIR_GetFileKeyWords(sTextFile, nMaxKeyLimit=50, bWeightOut=NLPIR_FALSE)
  NLPIR_GetFileKeyWords_rb.call(sTextFile, nMaxKeyLimit, bWeightOut).to_s
end

#NLPIR_GetFileNewWords(sTextFile, nMaxKeyLimit = 50, bWeightOut = NLPIR_FALSE) ⇒ Object 



238
239
240
 
# File 'lib/nlpir.rb', line 238

def NLPIR_GetFileNewWords(sTextFile, nMaxKeyLimit=50, bWeightOut=NLPIR_FALSE)
  NLPIR_GetFileNewWords_rb.call(sTextFile, nMaxKeyLimit, bWeightOut).to_s
end

#NLPIR_GetKeyWords(sLine, nMaxKeyLimit = 50, bWeightOut = NLPIR_FALSE) ⇒ Object 



226
227
228
 
# File 'lib/nlpir.rb', line 226

def NLPIR_GetKeyWords(sLine, nMaxKeyLimit=50, bWeightOut=NLPIR_FALSE)
  NLPIR_GetKeyWords_rb.call(sLine, nMaxKeyLimit, bWeightOut).to_s
end

#NLPIR_GetNewWords(sLine, nMaxKeyLimit = 50, bWeightOut = NLPIR_FALSE) ⇒ Object 



234
235
236
 
# File 'lib/nlpir.rb', line 234

def NLPIR_GetNewWords(sLine, nMaxKeyLimit=50, bWeightOut=NLPIR_FALSE)
  NLPIR_GetNewWords_rb.call(sLine, nMaxKeyLimit, bWeightOut).to_s
end

#NLPIR_GetParagraphProcessAWordCount(sParagraph) ⇒ Object 



197
198
199
 
# File 'lib/nlpir.rb', line 197

def NLPIR_GetParagraphProcessAWordCount(sParagraph)
  NLPIR_GetParagraphProcessAWordCount_rb.call(sParagraph)
end

#NLPIR_ImportUserDict(sFilename) ⇒ Object 



172
173
174
 
# File 'lib/nlpir.rb', line 172

def NLPIR_ImportUserDict(sFilename)
  NLPIR_ImportUserDict_rb.call(sFilename)
end

#NLPIR_Init(sInitDirPath = nil, encoding = UTF8_CODE, filepath) ⇒ Object

–函数



151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
 
# File 'lib/nlpir.rb', line 151

def NLPIR_Init(sInitDirPath=nil , encoding=UTF8_CODE, filepath)
  filepath += "/Data/"
  if File.exist?(filepath)==false
    FileUtils.mkdir(filepath)
    filemother = File.expand_path("../Data/", __FILE__)
    list=Dir.entries(filemother)
    list.each_index do |x| 
      t = filemother+"/"+list[x]
      FileUtils.cp(t,filepath) if !File.directory?(t) 
    end
  end
  
  
  NLPIR_Init_rb.call(sInitDirPath,encoding)

end

#NLPIR_NWI_AddFile(sFilename) ⇒ Object 



254
255
256
 
# File 'lib/nlpir.rb', line 254

def NLPIR_NWI_AddFile(sFilename)
  NLPIR_NWI_AddFile_rb.call(sFilename)
end

#NLPIR_NWI_AddMem(sFilename) ⇒ Object 



258
259
260
 
# File 'lib/nlpir.rb', line 258

def NLPIR_NWI_AddMem(sFilename)
  NLPIR_NWI_AddMem_rb.call(sFilename)
end

#NLPIR_NWI_Complete ⇒ Object 



262
263
264
 
# File 'lib/nlpir.rb', line 262

def NLPIR_NWI_Complete()
  NLPIR_NWI_Complete_rb.call()
end

#NLPIR_NWI_GetResult(bWeightOut = NLPIR_FALSE) ⇒ Object 



266
267
268
 
# File 'lib/nlpir.rb', line 266

def NLPIR_NWI_GetResult( bWeightOut = NLPIR_FALSE)
  NLPIR_NWI_GetResult_rb.call(bWeightOut)
end

#NLPIR_NWI_Result2UserDict ⇒ Object 



270
271
272
 
# File 'lib/nlpir.rb', line 270

def NLPIR_NWI_Result2UserDict()
  NLPIR_NWI_Result2UserDict_rb.call()
end

#NLPIR_NWI_Start ⇒ Object 



250
251
252
 
# File 'lib/nlpir.rb', line 250

def NLPIR_NWI_Start()
  NLPIR_NWI_Start_rb.call()
end

#NLPIR_ParagraphProcess(sParagraph, bPOStagged = NLPIR_TRUE) ⇒ Object 



176
177
178
 
# File 'lib/nlpir.rb', line 176

def NLPIR_ParagraphProcess(sParagraph, bPOStagged=NLPIR_TRUE)
  NLPIR_ParagraphProcess_rb.call(sParagraph, bPOStagged).to_s
end

#NLPIR_ParagraphProcessA(sParagraph) ⇒ Object 



180
181
182
183
184
185
186
187
188
189
190
191
 
# File 'lib/nlpir.rb', line 180

def NLPIR_ParagraphProcessA(sParagraph)
  resultCount = NLPIR_GetParagraphProcessAWordCount(sParagraph)
  pResultCount = Fiddle::Pointer.to_ptr(resultCount)
  p = NLPIR_ParagraphProcessA_rb.call(sParagraph, pResultCount.ref.to_i)
  pVecResult = Fiddle::Pointer.new(p.to_i)
  words_list = []
  words_list << Result_t.new(pVecResult)
  for i in 1...resultCount  do
      words_list << Result_t.new(pVecResult += Result_t.size)
  end
  return words_list
end

#NLPIR_ParagraphProcessAW(sParagraph) ⇒ Object 



201
202
203
204
205
206
207
208
209
210
211
212
 
# File 'lib/nlpir.rb', line 201

def NLPIR_ParagraphProcessAW(sParagraph)
  free = Fiddle::Function.new(Fiddle::RUBY_FREE, [TYPE_VOIDP], TYPE_VOID)
  resultCount = NLPIR_GetParagraphProcessAWordCount(sParagraph)
  pVecResult = Pointer.malloc(Result_t.size*resultCount,free)
  NLPIR_ParagraphProcessAW_rb.call(resultCount,pVecResult)
  words_list = []
  words_list << Result_t.new(pVecResult)
  for i in 1...resultCount do
      words_list << Result_t.new(pVecResult+=Result_t.size)
  end
  return words_list
end

#NLPIR_SaveTheUsrDic ⇒ Object 



218
219
220
 
# File 'lib/nlpir.rb', line 218

def NLPIR_SaveTheUsrDic()
  NLPIR_SaveTheUsrDic_rb.call()
end

#NLPIR_SetPOSmap(nPOSmap) ⇒ Object 



246
247
248
 
# File 'lib/nlpir.rb', line 246

def NLPIR_SetPOSmap(nPOSmap)
  NLPIR_SetPOSmap_rb.call(nPOSmap)
end