zouguojun souljoy commited on
Commit
bbfae1f
0 Parent(s):

Duplicate from souljoy/chatPDF

Browse files

Co-authored-by: 小爷毛毛-卓寿杰 <[email protected]>

Files changed (4) hide show
  1. .gitattributes +34 -0
  2. README.md +14 -0
  3. app.py +166 -0
  4. requirements.txt +3 -0
.gitattributes ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tflite filter=lfs diff=lfs merge=lfs -text
29
+ *.tgz filter=lfs diff=lfs merge=lfs -text
30
+ *.wasm filter=lfs diff=lfs merge=lfs -text
31
+ *.xz filter=lfs diff=lfs merge=lfs -text
32
+ *.zip filter=lfs diff=lfs merge=lfs -text
33
+ *.zst filter=lfs diff=lfs merge=lfs -text
34
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ title: ChatPDF
3
+ emoji: 💻
4
+ colorFrom: gray
5
+ colorTo: yellow
6
+ sdk: gradio
7
+ sdk_version: 3.20.0
8
+ app_file: app.py
9
+ pinned: false
10
+ license: apache-2.0
11
+ duplicated_from: souljoy/chatPDF
12
+ ---
13
+
14
+ Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
app.py ADDED
@@ -0,0 +1,166 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import requests
2
+ import json
3
+ import gradio as gr
4
+ # from concurrent.futures import ThreadPoolExecutor
5
+ import pdfplumber
6
+ import pandas as pd
7
+ import time
8
+ from cnocr import CnOcr
9
+ from sentence_transformers import SentenceTransformer, models, util
10
+ word_embedding_model = models.Transformer('uer/sbert-base-chinese-nli', do_lower_case=True)
11
+ pooling_model = models.Pooling(word_embedding_model.get_word_embedding_dimension(), pooling_mode='cls')
12
+ embedder = SentenceTransformer(modules=[word_embedding_model, pooling_model])
13
+ ocr = CnOcr()
14
+ # chat_url = 'https://souljoy-my-api.hf.space/sale'
15
+ chat_url = 'https://souljoy-my-api.hf.space/chatpdf'
16
+ headers = {
17
+ 'Content-Type': 'application/json',
18
+ }
19
+ # thread_pool_executor = ThreadPoolExecutor(max_workers=4)
20
+ history_max_len = 500
21
+ all_max_len = 3000
22
+
23
+
24
+ def get_emb(text):
25
+ emb_url = 'https://souljoy-my-api.hf.space/embeddings'
26
+ data = {"content": text}
27
+ try:
28
+ result = requests.post(url=emb_url,
29
+ data=json.dumps(data),
30
+ headers=headers
31
+ )
32
+ return result.json()['data'][0]['embedding']
33
+ except Exception as e:
34
+ print('data', data, 'result json', result.json())
35
+
36
+
37
+ def doc_emb(doc: str):
38
+ texts = doc.split('\n')
39
+ # futures = []
40
+ emb_list = embedder.encode(texts)
41
+ # for text in texts:
42
+ # futures.append(thread_pool_executor.submit(get_emb, text))
43
+ # for f in futures:
44
+ # emb_list.append(f.result())
45
+ print('\n'.join(texts))
46
+ return texts, emb_list, gr.Textbox.update(visible=True), gr.Button.update(visible=True), gr.Markdown.update(
47
+ value="""操作说明 step 3:PDF解析提交成功! 🙋 可以开始对话啦~"""), gr.Chatbot.update(visible=True)
48
+
49
+
50
+ def get_response(msg, bot, doc_text_list, doc_embeddings):
51
+ # future = thread_pool_executor.submit(get_emb, msg)
52
+ now_len = len(msg)
53
+ req_json = {'question': msg}
54
+ his_bg = -1
55
+ for i in range(len(bot) - 1, -1, -1):
56
+ if now_len + len(bot[i][0]) + len(bot[i][1]) > history_max_len:
57
+ break
58
+ now_len += len(bot[i][0]) + len(bot[i][1])
59
+ his_bg = i
60
+ req_json['history'] = [] if his_bg == -1 else bot[his_bg:]
61
+ # query_embedding = future.result()
62
+ query_embedding = embedder.encode([msg])
63
+ cos_scores = util.cos_sim(query_embedding, doc_embeddings)[0]
64
+ score_index = [[score, index] for score, index in zip(cos_scores, [i for i in range(len(cos_scores))])]
65
+ score_index.sort(key=lambda x: x[0], reverse=True)
66
+ print('score_index:\n', score_index)
67
+ index_set, sub_doc_list = set(), []
68
+ for s_i in score_index:
69
+ doc = doc_text_list[s_i[1]]
70
+ if now_len + len(doc) > all_max_len:
71
+ break
72
+ index_set.add(s_i[1])
73
+ now_len += len(doc)
74
+ # 可能段落截断错误,所以把上下段也加入进来
75
+ if s_i[1] > 0 and s_i[1] -1 not in index_set:
76
+ doc = doc_text_list[s_i[1]-1]
77
+ if now_len + len(doc) > all_max_len:
78
+ break
79
+ index_set.add(s_i[1]-1)
80
+ now_len += len(doc)
81
+ if s_i[1] + 1 < len(doc_text_list) and s_i[1] + 1 not in index_set:
82
+ doc = doc_text_list[s_i[1]+1]
83
+ if now_len + len(doc) > all_max_len:
84
+ break
85
+ index_set.add(s_i[1]+1)
86
+ now_len += len(doc)
87
+
88
+ index_list = list(index_set)
89
+ index_list.sort()
90
+ for i in index_list:
91
+ sub_doc_list.append(doc_text_list[i])
92
+ req_json['doc'] = '' if len(sub_doc_list) == 0 else '\n'.join(sub_doc_list)
93
+ data = {"content": json.dumps(req_json)}
94
+ print('data:\n', req_json)
95
+ result = requests.post(url=chat_url,
96
+ data=json.dumps(data),
97
+ headers=headers
98
+ )
99
+ res = result.json()['content']
100
+ bot.append([msg, res])
101
+ return bot[max(0, len(bot) - 3):]
102
+
103
+
104
+ def up_file(files):
105
+ doc_text_list = []
106
+ for idx, file in enumerate(files):
107
+ print(file.name)
108
+ with pdfplumber.open(file.name) as pdf:
109
+ for i in range(len(pdf.pages)):
110
+ # 读取PDF文档第i+1页
111
+ page = pdf.pages[i]
112
+ res_list = page.extract_text().split('\n')[:-1]
113
+
114
+ for j in range(len(page.images)):
115
+ # 获取图片的二进制流
116
+ img = page.images[j]
117
+ file_name = '{}-{}-{}.png'.format(str(time.time()), str(i), str(j))
118
+ with open(file_name, mode='wb') as f:
119
+ f.write(img['stream'].get_data())
120
+ try:
121
+ res = ocr.ocr(file_name)
122
+ except Exception as e:
123
+ res = []
124
+ if len(res) > 0:
125
+ res_list.append(' '.join([re['text'] for re in res]))
126
+
127
+ tables = page.extract_tables()
128
+ for table in tables:
129
+ # 第一��当成表头:
130
+ df = pd.DataFrame(table[1:], columns=table[0])
131
+ try:
132
+ records = json.loads(df.to_json(orient="records", force_ascii=False))
133
+ for rec in records:
134
+ res_list.append(json.dumps(rec, ensure_ascii=False))
135
+ except Exception as e:
136
+ res_list.append(str(df))
137
+
138
+ doc_text_list += res_list
139
+ doc_text_list = [str(text).strip() for text in doc_text_list if len(str(text).strip()) > 0]
140
+ print(doc_text_list)
141
+ return gr.Textbox.update(value='\n'.join(doc_text_list), visible=True), gr.Button.update(
142
+ visible=True), gr.Markdown.update(
143
+ value="操作说明 step 2:确认PDF解析结果(可修正),点击“提交解析结果”,随后进行对话")
144
+
145
+
146
+ with gr.Blocks() as demo:
147
+ with gr.Row():
148
+ with gr.Column():
149
+ file = gr.File(file_types=['.pdf'], label='点击上传PDF,进行解析(支持多文档、表格、OCR)', file_count='multiple')
150
+ doc_bu = gr.Button(value='提交解析结果', visible=False)
151
+ txt = gr.Textbox(label='PDF解析结果', visible=False)
152
+ doc_text_state = gr.State([])
153
+ doc_emb_state = gr.State([])
154
+ with gr.Column():
155
+ md = gr.Markdown("""操作说明 step 1:点击左侧区域,上传PDF,进行解析""")
156
+ chat_bot = gr.Chatbot(visible=False)
157
+ msg_txt = gr.Textbox(label='消息框', placeholder='输入消息,点击发送', visible=False)
158
+ chat_bu = gr.Button(value='发送', visible=False)
159
+
160
+ file.change(up_file, [file], [txt, doc_bu, md])
161
+ doc_bu.click(doc_emb, [txt], [doc_text_state, doc_emb_state, msg_txt, chat_bu, md, chat_bot])
162
+ chat_bu.click(get_response, [msg_txt, chat_bot, doc_text_state, doc_emb_state], [chat_bot])
163
+
164
+ if __name__ == "__main__":
165
+ demo.queue().launch()
166
+ # demo.queue().launch(share=False, server_name='172.22.2.54', server_port=9191)
requirements.txt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ pdfplumber
2
+ sentence_transformers
3
+ cnocr