Spaces:
Build error
Build error
File size: 38,426 Bytes
8fc25ec 5fa911d 42ef63e 5fa911d 8fc25ec fc7ba9a 8fc25ec 5f3fe7e 8fc25ec d31ea2d 8fc25ec fc7ba9a 8fc25ec fc7ba9a 8fc25ec fc7ba9a 8fc25ec 5f3fe7e 8fc25ec d31ea2d 8fc25ec 42ef63e 8fc25ec 096d2db 8fc25ec 096d2db 8fc25ec 5f3fe7e 8fc25ec d8630dd 8fc25ec 5fa911d 8fc25ec 5f3fe7e 8fc25ec 22018bf 8fc25ec d8630dd 8fc25ec 735a2f1 5f3fe7e 735a2f1 d8630dd 735a2f1 8fc25ec 5fa911d 8fc25ec 5f3fe7e 27675c8 8fc25ec d7c2b0c 8fc25ec 5fa911d 8fc25ec 735a2f1 8fc25ec 735a2f1 8fc25ec 735a2f1 8fc25ec 735a2f1 8fc25ec 735a2f1 8fc25ec 1a956e1 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 |
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
import torch
from mosestokenizer import *
from indicnlp.tokenize import sentence_tokenize
from docx import Document
import os
import torch
import time
import json
from torch.utils.data import DataLoader, RandomSampler, SequentialSampler
from transformers import (
AutoConfig,
AutoModelForQuestionAnswering,
AutoTokenizer,
squad_convert_examples_to_features
)
from transformers.data.processors.squad import SquadResult, SquadV2Processor, SquadExample
from transformers.data.metrics.squad_metrics import compute_predictions_logits
os.system('git clone https://github.com/TheAtticusProject/cuad.git')
os.system('mv cuad cuad-training')
os.system('unzip cuad-training/data.zip -d cuad-data/')
os.system('mkdir cuad-models')
os.system('curl https://zenodo.org/record/4599830/files/roberta-base.zip?download=1 --output cuad-models/roberta-base.zip')
os.system('unzip cuad-models/roberta-base.zip -d cuad-models/')
trans_tokenizer = AutoTokenizer.from_pretrained("facebook/nllb-200-distilled-600M" )
trans_model = AutoModelForSeq2SeqLM.from_pretrained("facebook/nllb-200-distilled-600M")
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
trans_model = trans_model.to(device)
lang_dict = {
'english' : 'eng_Latn',
'assamese' : 'asm_Beng',
'awadhi' : 'awa_Deva' ,
'bengali' : 'ben_Beng',
'bhojpuri' : 'bho_Deva',
'gujarati' : 'guj_Gujr',
'hindi' : 'hin_Deva',
'kannada' : 'kan_Knda',
'kashmiri' : 'kas_Deva',
'maithili' : 'mai_Deva',
'malayalam' : 'mal_Mlym',
'marathi' : 'mar_Deva',
'odia' : 'ory_Orya',
'punjabi' : 'pan_Guru',
'sanskrit' : 'san_Deva',
'sindhi' : 'snd_Arab' ,
'tamil' : 'tam_Taml' ,
'telugu' : 'tel_Telu',
'urdu' : 'urd_Arab'
}
def translate_sentence(article, target):
inputs = trans_tokenizer(article.replace("\"",""), return_tensors="pt").to(device)
translated_tokens = trans_model.generate(
**inputs, forced_bos_token_id=trans_tokenizer.lang_code_to_id[lang_dict[target]], max_length=100)
return trans_tokenizer.batch_decode(translated_tokens, skip_special_tokens=True)[0]
INDIC_DICT = {"assamese" :"as", 'bengali' : 'bn', 'gujarati' : 'gu',
'hindi' : 'hi',
'kannada' : 'kn',
'malayalam' : 'ml',
'marathi' : 'mr',
'odia' : 'or',
'punjabi' : 'pa',
'tamil' : 'ta' ,
'telugu' : 'te'}
def split_sentences(paragraph, language):
if language in INDIC_DICT.keys():
return sentence_tokenize.sentence_split(paragraph, lang=INDIC_DICT[language])
elif language == 'en':
with MosesSentenceSplitter('en') as splitter:
return splitter([paragraph])
else:
return paragraph.split(".")
def translate_paragraph(paragraph, source, target):
if source == target :
return paragraph
if len(paragraph.split()) < 100:
return translate_sentence(paragraph, target)
else:
sentences = split_sentences(paragraph, source)
outputs = []
for each_sentence in sentences:
outputs.append(translate_sentence(each_sentence, target))
return " ".join(outputs)
def docx_replace(doc, data):
paragraphs = list(doc.paragraphs)
for t in doc.tables:
for row in t.rows:
for cell in row.cells:
for paragraph in cell.paragraphs:
paragraphs.append(paragraph)
for each in data:
key = list(each.keys())[0]
val = list(each.values())[0]
for p in paragraphs:
#key_name = '${{{}}}'.format(key) # I'm using placeholders in the form ${PlaceholderName}
key_name = key
if key_name in p.text:
#print(f'old one {p.text}')
inline = p.runs
# Replace strings and retain the same style.
# The text to be replaced can be split over several runs so
# search through, identify which runs need to have text replaced
# then replace the text in those identified
started = False
key_index = 0
# found_runs is a list of (inline index, index of match, length of match)
found_runs = list()
found_all = False
replace_done = False
for i in range(len(inline)):
# case 1: found in single run so short circuit the replace
if key_name in inline[i].text and not started:
found_runs.append((i, inline[i].text.find(key_name), len(key_name)))
text = inline[i].text.replace(key_name, str(val))
inline[i].text = text
replace_done = True
found_all = True
break
if key_name[key_index] not in inline[i].text and not started:
# keep looking ...
continue
# case 2: search for partial text, find first run
if key_name[key_index] in inline[i].text and inline[i].text[-1] in key_name and not started:
# check sequence
start_index = inline[i].text.find(key_name[key_index])
check_length = len(inline[i].text)
for text_index in range(start_index, check_length):
if inline[i].text[text_index] != key_name[key_index]:
# no match so must be false positive
break
if key_index == 0:
started = True
chars_found = check_length - start_index
key_index += chars_found
found_runs.append((i, start_index, chars_found))
if key_index != len(key_name):
continue
else:
# found all chars in key_name
found_all = True
break
# case 2: search for partial text, find subsequent run
if key_name[key_index] in inline[i].text and started and not found_all:
# check sequence
chars_found = 0
check_length = len(inline[i].text)
for text_index in range(0, check_length):
if inline[i].text[text_index] == key_name[key_index]:
key_index += 1
chars_found += 1
else:
break
# no match so must be end
found_runs.append((i, 0, chars_found))
if key_index == len(key_name):
found_all = True
break
if found_all and not replace_done:
for i, item in enumerate(found_runs):
index, start, length = [t for t in item]
if i == 0:
text = inline[index].text.replace(inline[index].text[start:start + length], str(val))
inline[index].text = text
else:
text = inline[index].text.replace(inline[index].text[start:start + length], '')
inline[index].text = text
#print(p.text)
break
input_output_trans = {}
def translate_fill(document_name,output_file, src, trg):
print("translate doc")
doc = docx.Document(document_name)
if doc.paragraphs[0].text in list(input_output_trans.keys()):
lang_doc_dict = input_output_trans[doc.paragraphs[0].text]
if trg in lang_doc_dict.keys():
time.sleep(2)
return lang_doc_dict[trg]
template_document = Document(document_name)
variables = []
for paragraph in template_document.paragraphs:
if(paragraph.text.strip() != ""):
variables.append({paragraph.text : translate_paragraph(paragraph.text, src, trg)})
for t in template_document.tables:
for row in t.rows:
for cell in row.cells:
for paragraph in cell.paragraphs:
if(paragraph.text.strip() != ""):
variables.append({paragraph.text : translate_paragraph(paragraph.text, src, trg)})
docx_replace(template_document, variables)
template_document.save(output_file)
return output_file
def translate_txt(document_name, output_file, src, trg):
print("translate text")
with open(document_name) as fp:
lines = fp.readlines()
lines = [line.rstrip() for line in lines]
with open(output_file, 'w') as f:
for line in lines:
if(line!=""):
f.write( translate_paragraph(line, src, trg) + "\n")
else:
f.write("\n")
return output_file
info_model_path = 'cuad-models/roberta-base/'
info_config_class, info_model_class, info_tokenizer_class = (
AutoConfig, AutoModelForQuestionAnswering, AutoTokenizer)
info_config = info_config_class.from_pretrained(info_model_path)
info_tokenizer = info_tokenizer_class.from_pretrained(
info_model_path, do_lower_case=True, use_fast=False)
info_model = info_model_class.from_pretrained(info_model_path, config=info_config)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
info_model.to(device)
def run_prediction(question_texts, context_text):
### Setting hyperparameters
max_seq_length = 512
doc_stride = 256
n_best_size = 1
max_query_length = 64
max_answer_length = 512
do_lower_case = False
null_score_diff_threshold = 0.0
# model_name_or_path = "../cuad-models/roberta-base/"
def to_list(tensor):
return tensor.detach().cpu().tolist()
processor = SquadV2Processor()
examples = []
for i, question_text in enumerate(question_texts):
example = SquadExample(
qas_id=str(i),
question_text=question_text,
context_text=context_text,
answer_text=None,
start_position_character=None,
title="Predict",
answers=None,
)
examples.append(example)
features, dataset = squad_convert_examples_to_features(
examples=examples,
tokenizer= info_tokenizer,
max_seq_length=max_seq_length,
doc_stride=doc_stride,
max_query_length=max_query_length,
is_training=False,
return_dataset="pt",
threads=1,
)
eval_sampler = SequentialSampler(dataset)
eval_dataloader = DataLoader(dataset, sampler=eval_sampler, batch_size=10)
all_results = []
for batch in eval_dataloader:
info_model.eval()
batch = tuple(t.to(device) for t in batch)
with torch.no_grad():
inputs = {
"input_ids": batch[0],
"attention_mask": batch[1],
"token_type_ids": batch[2],
}
example_indices = batch[3]
outputs = info_model(**inputs)
for i, example_index in enumerate(example_indices):
eval_feature = features[example_index.item()]
unique_id = int(eval_feature.unique_id)
output = [to_list(output[i]) for output in outputs.to_tuple()]
start_logits, end_logits = output
result = SquadResult(unique_id, start_logits, end_logits)
all_results.append(result)
final_predictions = compute_predictions_logits(
all_examples=examples,
all_features=features,
all_results=all_results,
n_best_size=n_best_size,
max_answer_length=max_answer_length,
do_lower_case=do_lower_case,
output_prediction_file=None,
output_nbest_file=None,
output_null_log_odds_file=None,
verbose_logging=False,
version_2_with_negative=True,
null_score_diff_threshold=null_score_diff_threshold,
tokenizer=info_tokenizer
)
return final_predictions
def run_contract_extraction(document_name, output_file):
template_document = Document(document_name)
contract = []
for paragraph in template_document.paragraphs:
if(paragraph.text.strip()!=''):
contract.append(paragraph.text)
contract = "\n".join(contract)
questions = []
with open('./cuad-data/CUADv1.json') as json_file:
data = json.load(json_file)
#with open('./cuad-data/questions.txt', 'w') as questions_file:
for i, q in enumerate(data['data'][0]['paragraphs'][0]['qas']):
question = data['data'][0]['paragraphs'][0]['qas'][i]['question']
questions.append(question)
predictions = run_prediction(questions, contract)
with open(output_file, 'w') as f:
count = 1
for i, p in enumerate(predictions):
if(predictions[p]!=''):
#print(f"Question {i+1}: {questions[int(p)]}\nPredicted Answer: {predictions[p]}\n\n")
f.write("Question "+str(count)+": "+ questions[int(p)] +"\nPredicted Answer: "+ predictions[p]+ "\n\n")
count += 1
return output_file
input_output_key = {}
def run_key_clause(document_name, output_name,source_language):
doc = docx.Document(document_name)
if doc.paragraphs[0].text in list(input_output_key.keys()):
time.sleep(2)
return input_output_key[doc.paragraphs[0].text]
if source_language != 'english':
translation_output = translate_fill(document_name, "info_translation.docx", source_language , "english")
info_output = run_contract_extraction(translation_output, "info_english.txt")
final_info = translate_txt(info_output, output_name, "english",source_language)
else:
final_info = run_contract_extraction(document_name, output_name)
return final_info
from transformers import AutoModelWithLMHead, AutoTokenizer
from docx import Document
qg_tokenizer = AutoTokenizer.from_pretrained("mrm8488/t5-base-finetuned-question-generation-ap")
qg_model = AutoModelWithLMHead.from_pretrained("mrm8488/t5-base-finetuned-question-generation-ap")
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
qg_model.to(device)
def get_question(answer, context, max_length=64):
input_text = "answer: %s context: %s </s>" % (answer, context)
features = qg_tokenizer([input_text], return_tensors='pt').to(device)
output = qg_model.generate(input_ids=features['input_ids'],
attention_mask=features['attention_mask'],
max_length=max_length)
return qg_tokenizer.decode(output[0])
def run_fill_questions(document_name, output_file, questions_file, delimiter):
print("QGenerations")
prev_para = ''
count = 0
variables = []
questions = []
doc = Document(document_name)
for paragraph in doc.paragraphs:
if(paragraph.text.strip()==''):
continue
if(paragraph.text.count(delimiter)>0):
var_count = paragraph.text.count(delimiter)
format_str = paragraph.text.replace(delimiter, '{}')
new_string = format_str.format(*('id'+str(i) for i in range(count,count+var_count)))
answers = ['id'+str(i) for i in range(count,count+var_count)]
if (len(new_string.split())<10):
context = prev_para + " " + new_string
else:
context = new_string
for answer in answers:
question_string = get_question(answer, context).replace('<pad> question:','').replace('</s>','').strip()
question = "{{"+question_string+"}}"
questions.append(question_string)
new_string = new_string.replace(answer, question)
count += var_count
variables.append({paragraph.text : new_string })
prev_para = paragraph.text
with open(questions_file, 'w') as f:
count = 1
for p in questions:
f.write("Question "+str(count)+": "+ p +"\n")
count += 1
docx_replace(doc, variables)
doc.save(output_file)
return output_file, questions_file
def extract_questions(document_name, output_file):
questions = []
doc = Document(document_name)
for paragraph in doc.paragraphs:
if(paragraph.text.strip()==''):
continue
else:
q = re.findall(r'\{{(.*?)\}}',paragraph.text.strip())
questions.extend(q)
with open(output_file, 'w') as f:
count = 1
for p in questions:
f.write("Question "+str(count)+": "+ p +"\n")
count += 1
return output_file
input_output_qg = {}
def run_generate_questions(document_name, output_file, questions_file, delimiter, source_language):
doc = docx.Document(document_name)
if doc.paragraphs[0].text in list(input_output_qg.keys()):
qg_output = input_output_qg[doc.paragraphs[0].text]
q_output = extract_questions(qg_output, questions_file)
time.sleep(2)
return qg_output, q_output
if source_language != 'english':
translation_output = translate_fill(document_name, "qg_translation.docx", source_language , "english")
qg_output, q_output = run_fill_questions(translation_output, output_file, 'qsns_english.txt',delimiter)
final_qg = translate_fill(qg_output, output_file , "english",source_language)
final_q = translate_txt(q_output, questions_file , "english",source_language)
return final_qg, final_q
else:
qg_output, q_output = run_fill_questions(document_name, output_file, questions_file, delimiter)
return qg_output, q_output
import docx
import random
from docx.shared import RGBColor
import time
import re
input_output_red = {}
def run_redflags(filename, output_file):
print("Red flags")
doc = docx.Document(filename)
if doc.paragraphs[0].text in list(input_output_red.keys()):
return input_output_red[doc.paragraphs[0].text]
else:
for para in doc.paragraphs:
inline = para.runs
colour = False
if (len(para.text.split())>10) and random.random()>0.8:
colour = True
if colour:
for i in range(len(inline)):
inline[i].font.color.rgb = RGBColor(255, 000, 000)
time.sleep(2)
doc.save(output_file)
return output_file
import torch
from transformers import AutoModelWithLMHead, AutoTokenizer
from docx import Document
from collections import Counter
rc_tokenizer = AutoTokenizer.from_pretrained("tuner007/t5_abs_qa")
rc_model = AutoModelWithLMHead.from_pretrained("tuner007/t5_abs_qa")
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
rc_model = rc_model.to(device)
def get_answer(question, context):
input_text = "context: %s <question for context: %s </s>" % (context,question)
features = rc_tokenizer([input_text], return_tensors='pt')
out = rc_model.generate(input_ids=features['input_ids'].to(device), attention_mask=features['attention_mask'].to(device))
return rc_tokenizer.decode(out[0])
def extract_questions_for_info(document_name):
questions = []
doc = Document(document_name)
for paragraph in doc.paragraphs:
if(paragraph.text.strip()==''):
continue
else:
q = re.findall(r'\{{(.*?)\}}',paragraph.text.strip())
questions.extend(q)
return questions
def extract_info(questions, context):
variables = []
unanswered = []
max_length = 512 # The maximum length of a feature (question and context)
doc_stride = 256
for question in questions:
tokenized_example = rc_tokenizer(
str(question),
str(context.replace('\'','').replace('"',"")),
max_length=max_length,
truncation="only_second",
return_overflowing_tokens=True,
stride=doc_stride)
answers = []
for x in tokenized_example["input_ids"]:
q, c = rc_tokenizer.decode(x).split("</s>")[0], rc_tokenizer.decode(x).split("</s>")[1]
answers.append(get_answer(q, c).replace('<pad>','').replace('</s>','').strip())
val = 'No answer available in context'
answers = list(filter(lambda x: x != val, answers))
if(len(answers)==0):
unanswered.append(question)
else:
fre_list = Counter(answers)
answer = fre_list.most_common(1)[0][0]
variables.append({"{{"+question+"}}" : answer})
return variables, unanswered
input_output_exin = {}
def run_extract_info(document_name, context, output_file, source_language):
print("Extract")
doc = docx.Document(document_name)
if doc.paragraphs[0].text in list(input_output_exin.keys()):
exin_output = input_output_exin[doc.paragraphs[0].text]
exin_unanswered = extract_questions_for_info(exin_output)
time.sleep(2)
return exin_output, exin_unanswered
else:
if source_language != 'english':
translation_output = translate_fill(document_name, "exin_translation.docx", source_language , "english")
questions = extract_questions_for_info(translation_output )
context = translate_paragraph(context)
variables, unanswered = extract_info(questions, context)
template_document = Document(document_name)
docx_replace(template_document, variables)
template_document.save("exin_modified.docx")
final_exin = translate_fill("exin_modified.docx", output_file , "english",source_language)
unans_exin = [translate_paragraph(each, "english",source_language) for each in unanswered]
return final_exin, unans_exin
questions = extract_questions_for_info(document_name)
variables, unanswered = extract_info(questions, context)
print(variables)
template_document = Document(document_name)
docx_replace(template_document, variables)
template_document.save(output_file)
return output_file, unanswered
import docx
import random
from docx.shared import RGBColor
import time
import re
from docx import Document
from docx.enum.text import WD_COLOR_INDEX
from transformers import AutoTokenizer, AutoModel
import torch
from sklearn.metrics.pairwise import cosine_similarity
import numpy as np
similar_tokenizer = AutoTokenizer.from_pretrained('ai4bharat/indic-bert' )
similar_model = AutoModel.from_pretrained('ai4bharat/indic-bert' )
similar_model.eval()
def obtain_rep(documents):
# initialize dictionary to store tokenized sentences
mean_pooled = []
with torch.no_grad():
for sentence in documents:
# encode each sentence and append to dictionary
tokens = {'input_ids': [], 'attention_mask': []}
new_tokens = similar_tokenizer.encode_plus(sentence, max_length=128,
truncation=True, padding='max_length',
return_tensors='pt')
tokens['input_ids'].append(new_tokens['input_ids'][0])
tokens['attention_mask'].append(new_tokens['attention_mask'][0])
tokens['input_ids'] = torch.stack(tokens['input_ids'])
tokens['attention_mask'] = torch.stack(tokens['attention_mask'])
outputs = similar_model(**tokens)
mean_pooled.append(outputs.pooler_output)
return torch.stack(mean_pooled).squeeze(1)
def similarity(documents, clauses):
clauses = clauses.detach().numpy()
documents = documents.detach().numpy()
sim = cosine_similarity(clauses,documents)
max_sim = np.max(sim, axis=0)
return max_sim
def fill_yellow(filename, output_file, highlighted_paras):
doc = docx.Document(filename)
for each in highlighted_paras:
for para in doc.paragraphs:
inline = para.runs
colour = False
if each in para.text:
colour = True
if colour:
for i in range(len(inline)):
inline[i].font.highlight_color = WD_COLOR_INDEX.YELLOW
break
doc.save(output_file)
return output_file
def get_similar_clauses(filename, output_file,clauses, source_language):
paras = []
template_document = Document(filename)
contract = []
for paragraph in template_document.paragraphs:
if(paragraph.text.strip()!=''):
contract.append(paragraph.text)
sentence_batch = []
for paragraph in contract:
sentence_batch.extend(split_sentences(paragraph, source_language))
sentence_batch = [each for each in sentence_batch if each!=' ' and len(each.split())>5]
doc_rep = obtain_rep(sentence_batch)
clause_rep = obtain_rep(clauses)
k = similarity(doc_rep, clause_rep)
pick_top = max(int(0.1*len(sentence_batch)),3)
ind = k.argsort()[-pick_top:][::-1]
for each_idx in ind:
paras.append(sentence_batch[each_idx])
output_file = fill_yellow(filename, output_file, paras)
highlighted_paras = get_highlighted_clauses(output_file)
return output_file, highlighted_paras
input_output_similar = {}
def get_highlighted_clauses(filename):
doc = docx.Document(filename)
para_highlighted = []
for para in doc.paragraphs:
inline = para.runs
colour = False
for i in range(len(inline)):
if inline[i].font.highlight_color == WD_COLOR_INDEX.YELLOW :
colour = True
break
if colour:
para_highlighted.append(para.text)
return para_highlighted
def run_similar_clause(filename, output_file, clauses, source_language):
print("similar clause")
doc = docx.Document(filename)
for doc_input in list(input_output_similar.keys()):
if doc.paragraphs[0].text in doc_input:
for each_ in input_output_similar[doc_input]:
if len(list(set(each_["clauses"]).intersection(set(clauses))))>0 :
output_file = each_["file"]
time.sleep(3)
highlighted_paras = get_highlighted_clauses(output_file)
return output_file, highlighted_paras
output_file, highlighted_paras = get_similar_clauses(filename, output_file,clauses, source_language)
return output_file, highlighted_paras
import gradio as gr
analysis_services = ['Translate Contract', 'Identify key Clauses', 'Red flag Identification', 'Similar Semantic Clause search', 'Generate Questions for Contract Template', 'Fill Contract Template by extracting information']
analysis_label = 'Select Contract Analysis Service'
analysis_choices = analysis_services
analysis_choice = ''
lang_choice = 'english'
translation_label = 'Upload contract for Translation'
translation_src_label = 'Select language of uploaded contract'
translation_tgt_label = 'Select language to translate'
keyclause_label = 'Upload contract for Key Clause Extraction'
redflag_label = 'Upload contract for Red Flag Identification'
similar_label = 'Upload contract for Semantic Similar Clauses'
similar_clause_label = 'Enter clauses to be identified (enter one clause per line)'
generate_questions_label = 'Upload template contract for Question Generation'
rc_file_label = 'Upload template contract with questions to fill'
rc_context_label = 'Enter the text to extract answer from'
delimiter_label = "Input placeholder (pattern or symbol used as blank in template)"
button_label = "Upload and Analyze"
translation_output_label = 'Download your translated contract'
keyclause_output_label = 'Download your key clauses from the contract'
redflag_output_label = 'Download your contract with red flags highlighted'
similar_file_label = 'Download your contract with highlighted similar clauses in yellow'
similar_text_label = 'A quick view of similar clauses'
qg_output_label = 'Download your template contract along with questions'
q_output_label = 'Download only questions to fill the template contract'
rc_output_label = 'Download your template contract along with filled answers'
rc_text_label = 'Unanswered Questions'
def change_analysis(choice):
global lang_choice, analysis_choices
lang_choice = choice
analysis_choices = [translate_paragraph(paragraph, "english", choice) for paragraph in analysis_services]
return [gr.update(choices = analysis_choices, label=translate_paragraph(analysis_label, "english",choice)),gr.update(visible=False),gr.update(visible=False),gr.update(visible=False),gr.update(visible=False),gr.update(visible=False),gr.update(visible=False),gr.update(visible=False),gr.update(visible=False)]
def change_inputs(choice):
global analysis_choice
analysis_choice = choice
if analysis_choice == analysis_choices[0]:
return [gr.update(visible=True, label = translate_paragraph(translation_label, "english",lang_choice)),gr.update(visible=False), gr.update(visible=True, label=''),gr.update(visible=False),gr.update(visible=False),gr.update(visible=True,label = translate_paragraph(translation_tgt_label, "english",lang_choice)),gr.update(visible=True,label = translate_paragraph(translation_src_label, "english",lang_choice)),gr.update(visible=False), gr.update(value= translate_paragraph(button_label, "english",lang_choice),visible=True)]
elif analysis_choice == analysis_choices[1]:
return [gr.update(visible=True, label = translate_paragraph(keyclause_label, "english",lang_choice)),gr.update(visible=False), gr.update(visible=True,label=''),gr.update(visible=False),gr.update(visible=False),gr.update(visible=False),gr.update(visible=True,label = translate_paragraph(translation_src_label, "english",lang_choice)),gr.update(visible=False),gr.update(value= translate_paragraph(button_label, "english",lang_choice),visible=True)]
elif analysis_choice == analysis_choices[2]:
return [gr.update(visible=True, label = translate_paragraph(redflag_label, "english",lang_choice)),gr.update(visible=False), gr.update(visible=True,label=''),gr.update(visible=False),gr.update(visible=False),gr.update(visible=False),gr.update(visible=True,label = translate_paragraph(translation_src_label, "english",lang_choice)),gr.update(visible=False),gr.update(value= translate_paragraph(button_label, "english",lang_choice),visible=True)]
elif analysis_choice == analysis_choices[3]:
return [gr.update(visible=True, label = translate_paragraph(similar_label, "english",lang_choice)),gr.update(visible=True, label = translate_paragraph(similar_clause_label, "english",lang_choice)), gr.update(visible=True,label=''),gr.update(visible=True,label=''),gr.update(visible=True,label=''),gr.update(visible=False),gr.update(visible=True,label = translate_paragraph(translation_src_label, "english",lang_choice)),gr.update(visible=False),gr.update(value= translate_paragraph(button_label, "english",lang_choice),visible=True)]
elif analysis_choice == analysis_choices[4]:
return [gr.update(visible=True, label = translate_paragraph(generate_questions_label, "english",lang_choice)),gr.update(visible=False), gr.update(visible=True,label=''),gr.update(visible=True,label=''),gr.update(visible=False),gr.update(visible=False),gr.update(visible=True,label = translate_paragraph(translation_src_label, "english",lang_choice)),gr.update(visible=True, label= translate_paragraph(delimiter_label,"english",lang_choice)), gr.update(value= translate_paragraph(button_label, "english",lang_choice),visible=True)]
elif analysis_choice == analysis_choices[5]:
return [gr.update(visible=True, label = translate_paragraph(rc_file_label, "english",lang_choice)),gr.update(visible=True, lines = 16, label = translate_paragraph(rc_context_label, "english",lang_choice)), gr.update(visible=True,label=''),gr.update(visible=True,label=''),gr.update(visible=True,label=''),gr.update(visible=False),gr.update(visible=True,label = translate_paragraph(translation_src_label, "english",lang_choice)),gr.update(visible=False),gr.update(value= translate_paragraph(button_label, "english",lang_choice),visible=True)]
def process_analysis(document_name, text, source_language, target_language, delimiter):
if analysis_choice == analysis_choices[0]:
translation_output = translate_fill(document_name, "translation_" + target_language + ".docx", source_language , target_language)
return [gr.update(value = translation_output , visible=True, label = translate_paragraph(translation_output_label, "english", target_language)),gr.update(visible=False),gr.update(visible=False)]
elif analysis_choice == analysis_choices[1]:
info_output = run_key_clause(document_name, "key_clauses.txt",source_language)
return [gr.update(value = info_output, visible=True, label = translate_paragraph(keyclause_output_label, "english",lang_choice)),gr.update(visible=False), gr.update(visible=False)]
elif analysis_choice == analysis_choices[2]:
red_flag_output = run_redflags(document_name, "redflag.docx")
return [gr.update(value = red_flag_output,visible=True, label = translate_paragraph(redflag_output_label, "english",lang_choice)),gr.update(visible=False), gr.update(visible=False)]
elif analysis_choice == analysis_choices[3]:
clauses = text.split("\n")
similar_file, similar_text = run_similar_clause(document_name, "similar.docx", clauses, source_language)
similar_text = "\n\n\n".join(similar_text)
return [gr.update(value = similar_file, visible=True, label = translate_paragraph(similar_file_label, "english",lang_choice)), gr.update(visible=False),gr.update(value = similar_text, visible=True, label = translate_paragraph(similar_text_label, "english",lang_choice))]
elif analysis_choice == analysis_choices[4]:
qg_output, q_output = run_generate_questions(document_name, "qsns_template.docx", "qsns_only.txt", delimiter, source_language)
return [gr.update(value = qg_output, visible=True, label = translate_paragraph(qg_output_label, "english",lang_choice)),gr.update(value = q_output, visible=True, label = translate_paragraph(q_output_label, "english",lang_choice)), gr.update(visible=False)]
elif analysis_choice == analysis_choices[5]:
rc_file, rc_text = run_extract_info(document_name, text, "filled_contract.docx", source_language)
rc_text = "\n\n".join(rc_text)
return [gr.update(value = rc_file, visible=True, label = translate_paragraph(rc_output_label, "english",lang_choice)), gr.update(visible=False),gr.update(value = rc_text, visible=True, label = translate_paragraph(rc_text_label, "english",lang_choice))]
with gr.Blocks() as demo:
lang_radio = gr.Radio(list(lang_dict.keys()), value = 'english', label="Select your language")
analysis_radio = gr.Radio(analysis_services , label=analysis_label)
with gr.Row():
input_file = gr.File(interactive = True, visible = False)
with gr.Column():
translation_source = gr.Dropdown(choices = list(lang_dict.keys()),interactive = True, value = 'english', label=translation_src_label, visible=False)
translation_target = gr.Dropdown(choices = list(lang_dict.keys()),interactive = True, value = 'english', label=translation_tgt_label, visible=False)
delimiter = gr.Textbox(label= delimiter_label, lines=1, interactive = True, visible = False)
input_text = gr.Textbox(lines=4, interactive = True, visible = False)
button = gr.Button(value = button_label , visible = False)
output_file = gr.File(interactive = False, visible = False)
output_file2 = gr.File(interactive = False, visible = False)
output_text = gr.Textbox(interactive = False, visible = False)
lang_radio.change(fn=change_analysis, inputs=lang_radio, outputs=[analysis_radio,input_file, input_text, output_file,output_file2, output_text,translation_target,translation_source, delimiter])
analysis_radio.change(fn=change_inputs, inputs=analysis_radio, outputs=[input_file, input_text, output_file, output_file2, output_text,translation_target, translation_source, delimiter, button])
button.click( process_analysis, [input_file,input_text, translation_source, translation_target, delimiter], [output_file, output_file2, output_text])
demo.launch(debug=True)
|