| import gradio as gr | |
| from transformers import pipeline, BertTokenizer, BertForSequenceClassification | |
| import os | |
| import pickle | |
| import pandas as pd | |
| from service_dops_api.dops_config import ServiceDopsConfig | |
| from service_dops_api.dops_classifier import DopsClassifier | |
| import json | |
| HF_TOKEN = os.getenv('HF_TOKEN') | |
| tokenizer_cat = BertTokenizer.from_pretrained("warleagle/service_name_categorizer", | |
| token=HF_TOKEN) | |
| model_cat = BertForSequenceClassification.from_pretrained('warleagle/service_name_categorizer',token=HF_TOKEN) | |
| tokenizer_spec = BertTokenizer.from_pretrained("warleagle/specialists_categorizer_model", | |
| token=HF_TOKEN) | |
| model_spec = BertForSequenceClassification.from_pretrained('warleagle/specialists_categorizer_model',token=HF_TOKEN) | |
| def categoriser_predict(input_text): | |
| clf = pipeline("text-classification", model=model_cat, tokenizer=tokenizer_cat) | |
| predictions = clf(input_text) | |
| numeric_label = int(predictions[0]['label'].split("_")[1]) | |
| id2label = pd.read_pickle('id2label_service_categoriser.pickle') | |
| text_label = id2label[numeric_label] | |
| return text_label | |
| def doctor_spec_predict(input_text): | |
| clf = pipeline("text-classification", model=model_spec, tokenizer=tokenizer_spec) | |
| predictions = clf(input_text) | |
| numeric_label = int(predictions[0]['label'].split("_")[1]) | |
| id2label = pd.read_pickle('id2label_spec_categoriser.pickle') | |
| text_label = id2label[numeric_label] | |
| return text_label | |
| def dops_predict(input_text): | |
| cfg = ServiceDopsConfig() | |
| model = DopsClassifier(config=cfg) | |
| result = model.run_all_dops(input_text) | |
| return result | |
| def service_pipeline(input_text): | |
| categoriser_result = categoriser_predict(input_text) | |
| if categoriser_result!='Консультация специалиста': | |
| return 'Эта услуга не относится к приему специалиста','-','-' | |
| else: | |
| doctor_spec_result = doctor_spec_predict(input_text) | |
| dops_result = dops_predict(input_text) | |
| dops_result = json.dumps(dops_result,indent=4,ensure_ascii=False) | |
| return categoriser_result,doctor_spec_result,dops_result | |
| demo = gr.Interface(fn=service_pipeline,inputs=gr.components.Textbox(label='Название услуги'), | |
| outputs=[gr.components.Textbox(label='Относится ли данная услуга к приёму специалиста'), | |
| gr.components.Textbox(label='Специальность врача'), | |
| gr.Textbox(label='Дополнительные параметры услуги')], | |
| examples=[ | |
| ['Врач-офтальмолог (высшая категория/кандидат медицинских наук), первичный приём'], | |
| ['Прием (осмотр, консультация) - врача -оториноларинголога Первичный, рекомендации'], | |
| ['Прием врача специалиста ЛОР']]) | |
| if __name__ == "__main__": | |
| demo.launch() |