1515HEADERS = {"Content-Type" : "application/json" }
1616NER_RESPONSE_KEY = 'body'
1717
18+
1819class PiperOperatorException (BaseException ):
1920 def __init__ (self , msg ):
2021 pass
2122 # logger.exception(msg)
22-
2323
2424
2525class FileLoadException (PiperOperatorException ):
2626 def __init__ (self , fn ):
2727 self .fn = fn
28- super ().__init__ (f'file { fn } can`t be loaded' )
28+ super ().__init__ (f'file { fn } can`t be loaded' )
2929
3030
3131class JSONGetKeyException (PiperOperatorException ):
3232 def __init__ (self , key ):
3333 self .key = key
34- super ().__init__ (f'can`t get JSON key { key } ' )
34+ super ().__init__ (f'can`t get JSON key { key } ' )
3535
3636
3737class NoAvailableModelsException (PiperOperatorException ):
@@ -46,7 +46,8 @@ def get_data_by_key_from_response(cur_response, k):
4646 v = j .get (k )
4747 return v
4848
49- def get_data_by_key_from_url (url , key , post = True , data = None , file_name = "" ):
49+
50+ def get_data_by_key_from_url (url , key , post = True , data = None , file_name = "" ):
5051 try :
5152 if post :
5253 if file_name :
@@ -68,7 +69,7 @@ def get_data_by_key_from_url(url, key, post=True, data=None, file_name=""):
6869 return val
6970 else :
7071 return cur_response
71-
72+
7273 else :
7374 cur_response = requests .get (url , headers = HEADERS , data = data )
7475 cur_response .raise_for_status ()
@@ -83,7 +84,7 @@ def get_data_by_key_from_url(url, key, post=True, data=None, file_name=""):
8384 logger .exception (f'can`t get key from response: { cjke } ' )
8485
8586 except Exception as e :
86- logger .exception (f'error while processing url { url } : { e } ' )
87+ logger .exception (f'error while processing url { url } : { e } ' )
8788
8889
8990class PiperNLPWorker ():
@@ -106,15 +107,14 @@ def __init__(self, base_url):
106107 # get named entitys from text url
107108 self .url_spacy_get_NE = f'{ self .base_url } /extract_named_ents'
108109
109-
110110 def get_available_ner_models (self ):
111111 return get_data_by_key_from_url (self .url_spacy_all_models , 'available_models' , post = False )
112112
113113 def set_current_spacy_model (self , model ):
114- return get_data_by_key_from_url (self .url_spacy_set_model , '' , post = True , data = json .dumps ({'model_name' :model }))
114+ return get_data_by_key_from_url (self .url_spacy_set_model , '' , post = True , data = json .dumps ({'model_name' : model }))
115115
116116 def get_named_ent_from_text (self , txt ):
117- resp = get_data_by_key_from_url (self .url_spacy_get_NE , 'result' , post = False , data = json .dumps ({'txt' :txt }))
117+ resp = get_data_by_key_from_url (self .url_spacy_get_NE , 'result' , post = False , data = json .dumps ({'txt' : txt }))
118118 logger .debug (f'url is { resp } , response is { resp } ' )
119119 if NER_RESPONSE_KEY in resp .keys ():
120120 named_ents = resp .get (NER_RESPONSE_KEY )
@@ -138,9 +138,9 @@ def get_text_from_file(self, fn):
138138 def set_tesseract_config (self , conf ):
139139 return get_data_by_key_from_url (self .url_tsrct_cfg , '' , post = True , data = json .dumps (conf ))
140140
141+
141142if __name__ == '__main__' :
142143 piper_worker = PiperNLPWorker ('http://localhost:8788' )
143-
144144
145145 amodels = piper_worker .get_available_ner_models ()
146146 print ('all models' , amodels )
@@ -162,16 +162,12 @@ def set_tesseract_config(self, conf):
162162 logger .exception (e )
163163 # pprint(resp)
164164
165-
166165 txt = piper_worker .get_text_from_file ('/home/pavel/repo/piper_new/piper/tests/ocr_data.pdf' )
167166 logger .info (f'txt { txt } ' )
168167
169-
170168 ts_conf = dict ()
171169 ts_conf ['ts_lang' ] = 'eng'
172170 ts_conf ['ts_config_row' ] = rf'--oem 1 --psm 6'
173171
174172 resp = piper_worker .set_tesseract_config (ts_conf )
175173 logger .info (resp )
176-
177-
0 commit comments