Spaces:
Running
Running
File size: 41,336 Bytes
19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 19dc0f3 33473a0 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 |
import base64
import copy
import functools
import html
import json
import pprint
import re
from datetime import datetime
from functools import partial
from pathlib import Path
import gradio as gr
import yaml
from jinja2.sandbox import ImmutableSandboxedEnvironment
from PIL import Image
import modules.shared as shared
from modules import utils
from modules.extensions import apply_extensions
from modules.html_generator import (
chat_html_wrapper,
convert_to_markdown,
make_thumbnail
)
from modules.logging_colors import logger
from modules.text_generation import (
generate_reply,
get_encoded_length,
get_max_prompt_length
)
from modules.utils import delete_file, get_available_characters, save_file
# Copied from the Transformers library
jinja_env = ImmutableSandboxedEnvironment(trim_blocks=True, lstrip_blocks=True)
def str_presenter(dumper, data):
"""
Copied from https://github.com/yaml/pyyaml/issues/240
Makes pyyaml output prettier multiline strings.
"""
if data.count('\n') > 0:
return dumper.represent_scalar('tag:yaml.org,2002:str', data, style='|')
return dumper.represent_scalar('tag:yaml.org,2002:str', data)
yaml.add_representer(str, str_presenter)
yaml.representer.SafeRepresenter.add_representer(str, str_presenter)
def get_generation_prompt(renderer, impersonate=False, strip_trailing_spaces=True):
'''
Given a Jinja template, reverse-engineers the prefix and the suffix for
an assistant message (if impersonate=False) or an user message
(if impersonate=True)
'''
if impersonate:
messages = [
{"role": "user", "content": "<<|user-message-1|>>"},
{"role": "user", "content": "<<|user-message-2|>>"},
]
else:
messages = [
{"role": "assistant", "content": "<<|user-message-1|>>"},
{"role": "assistant", "content": "<<|user-message-2|>>"},
]
prompt = renderer(messages=messages)
suffix_plus_prefix = prompt.split("<<|user-message-1|>>")[1].split("<<|user-message-2|>>")[0]
suffix = prompt.split("<<|user-message-2|>>")[1]
prefix = suffix_plus_prefix[len(suffix):]
if strip_trailing_spaces:
prefix = prefix.rstrip(' ')
return prefix, suffix
def generate_chat_prompt(user_input, state, **kwargs):
impersonate = kwargs.get('impersonate', False)
_continue = kwargs.get('_continue', False)
also_return_rows = kwargs.get('also_return_rows', False)
history = kwargs.get('history', state['history'])['internal']
# Templates
chat_template_str = state['chat_template_str']
if state['mode'] != 'instruct':
chat_template_str = replace_character_names(chat_template_str, state['name1'], state['name2'])
instruction_template = jinja_env.from_string(state['instruction_template_str'])
chat_template = jinja_env.from_string(chat_template_str)
instruct_renderer = partial(
instruction_template.render,
builtin_tools=None,
tools=None,
tools_in_user_message=False,
add_generation_prompt=False
)
chat_renderer = partial(
chat_template.render,
add_generation_prompt=False,
name1=state['name1'],
name2=state['name2'],
user_bio=replace_character_names(state['user_bio'], state['name1'], state['name2']),
)
messages = []
if state['mode'] == 'instruct':
renderer = instruct_renderer
if state['custom_system_message'].strip() != '':
messages.append({"role": "system", "content": state['custom_system_message']})
else:
renderer = chat_renderer
if state['context'].strip() != '' or state['user_bio'].strip() != '':
context = replace_character_names(state['context'], state['name1'], state['name2'])
messages.append({"role": "system", "content": context})
insert_pos = len(messages)
for user_msg, assistant_msg in reversed(history):
user_msg = user_msg.strip()
assistant_msg = assistant_msg.strip()
if assistant_msg:
messages.insert(insert_pos, {"role": "assistant", "content": assistant_msg})
if user_msg not in ['', '<|BEGIN-VISIBLE-CHAT|>']:
messages.insert(insert_pos, {"role": "user", "content": user_msg})
user_input = user_input.strip()
if user_input and not impersonate and not _continue:
messages.append({"role": "user", "content": user_input})
def remove_extra_bos(prompt):
for bos_token in ['<s>', '<|startoftext|>', '<BOS_TOKEN>', '<|endoftext|>']:
while prompt.startswith(bos_token):
prompt = prompt[len(bos_token):]
return prompt
def make_prompt(messages):
if state['mode'] == 'chat-instruct' and _continue:
prompt = renderer(messages=messages[:-1])
else:
prompt = renderer(messages=messages)
if state['mode'] == 'chat-instruct':
outer_messages = []
if state['custom_system_message'].strip() != '':
outer_messages.append({"role": "system", "content": state['custom_system_message']})
prompt = remove_extra_bos(prompt)
command = state['chat-instruct_command']
command = command.replace('<|character|>', state['name2'] if not impersonate else state['name1'])
command = command.replace('<|prompt|>', prompt)
command = replace_character_names(command, state['name1'], state['name2'])
if _continue:
prefix = get_generation_prompt(renderer, impersonate=impersonate, strip_trailing_spaces=False)[0]
prefix += messages[-1]["content"]
else:
prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
if not impersonate:
prefix = apply_extensions('bot_prefix', prefix, state)
outer_messages.append({"role": "user", "content": command})
outer_messages.append({"role": "assistant", "content": prefix})
prompt = instruction_template.render(messages=outer_messages)
suffix = get_generation_prompt(instruct_renderer, impersonate=False)[1]
if len(suffix) > 0:
prompt = prompt[:-len(suffix)]
else:
if _continue:
suffix = get_generation_prompt(renderer, impersonate=impersonate)[1]
if len(suffix) > 0:
prompt = prompt[:-len(suffix)]
else:
prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
if state['mode'] == 'chat' and not impersonate:
prefix = apply_extensions('bot_prefix', prefix, state)
prompt += prefix
prompt = remove_extra_bos(prompt)
return prompt
prompt = make_prompt(messages)
# Handle truncation
if shared.tokenizer is not None:
max_length = get_max_prompt_length(state)
encoded_length = get_encoded_length(prompt)
while len(messages) > 0 and encoded_length > max_length:
# Remove old message, save system message
if len(messages) > 2 and messages[0]['role'] == 'system':
messages.pop(1)
# Remove old message when no system message is present
elif len(messages) > 1 and messages[0]['role'] != 'system':
messages.pop(0)
# Resort to truncating the user input
else:
user_message = messages[-1]['content']
# Bisect the truncation point
left, right = 0, len(user_message) - 1
while right - left > 1:
mid = (left + right) // 2
messages[-1]['content'] = user_message[:mid]
prompt = make_prompt(messages)
encoded_length = get_encoded_length(prompt)
if encoded_length <= max_length:
left = mid
else:
right = mid
messages[-1]['content'] = user_message[:left]
prompt = make_prompt(messages)
encoded_length = get_encoded_length(prompt)
if encoded_length > max_length:
logger.error(f"Failed to build the chat prompt. The input is too long for the available context length.\n\nTruncation length: {state['truncation_length']}\nmax_new_tokens: {state['max_new_tokens']} (is it too high?)\nAvailable context length: {max_length}\n")
raise ValueError
else:
logger.warning(f"The input has been truncated. Context length: {state['truncation_length']}, max_new_tokens: {state['max_new_tokens']}, available context length: {max_length}.")
break
prompt = make_prompt(messages)
encoded_length = get_encoded_length(prompt)
if also_return_rows:
return prompt, [message['content'] for message in messages]
else:
return prompt
def get_stopping_strings(state):
stopping_strings = []
renderers = []
if state['mode'] in ['instruct', 'chat-instruct']:
template = jinja_env.from_string(state['instruction_template_str'])
renderer = partial(template.render, add_generation_prompt=False)
renderers.append(renderer)
if state['mode'] in ['chat', 'chat-instruct']:
template = jinja_env.from_string(state['chat_template_str'])
renderer = partial(template.render, add_generation_prompt=False, name1=state['name1'], name2=state['name2'])
renderers.append(renderer)
for renderer in renderers:
prefix_bot, suffix_bot = get_generation_prompt(renderer, impersonate=False)
prefix_user, suffix_user = get_generation_prompt(renderer, impersonate=True)
stopping_strings += [
suffix_user + prefix_bot,
suffix_user + prefix_user,
suffix_bot + prefix_bot,
suffix_bot + prefix_user,
]
# Try to find the EOT token
for item in stopping_strings.copy():
item = item.strip()
if item.startswith("<") and ">" in item:
stopping_strings.append(item.split(">")[0] + ">")
elif item.startswith("[") and "]" in item:
stopping_strings.append(item.split("]")[0] + "]")
if 'stopping_strings' in state and isinstance(state['stopping_strings'], list):
stopping_strings += state.pop('stopping_strings')
# Remove redundant items that start with another item
result = [item for item in stopping_strings if not any(item.startswith(other) and item != other for other in stopping_strings)]
result = list(set(result))
if shared.args.verbose:
logger.info("STOPPING_STRINGS=")
pprint.PrettyPrinter(indent=4, sort_dicts=False).pprint(result)
print()
return result
def chatbot_wrapper(text, state, regenerate=False, _continue=False, loading_message=True, for_ui=False):
history = state['history']
output = copy.deepcopy(history)
output = apply_extensions('history', output)
state = apply_extensions('state', state)
visible_text = None
stopping_strings = get_stopping_strings(state)
is_stream = state['stream']
# Prepare the input
if not (regenerate or _continue):
visible_text = html.escape(text)
# Apply extensions
text, visible_text = apply_extensions('chat_input', text, visible_text, state)
text = apply_extensions('input', text, state, is_chat=True)
output['internal'].append([text, ''])
output['visible'].append([visible_text, ''])
# *Is typing...*
if loading_message:
yield {
'visible': output['visible'][:-1] + [[output['visible'][-1][0], shared.processing_message]],
'internal': output['internal']
}
else:
text, visible_text = output['internal'][-1][0], output['visible'][-1][0]
if regenerate:
if loading_message:
yield {
'visible': output['visible'][:-1] + [[visible_text, shared.processing_message]],
'internal': output['internal'][:-1] + [[text, '']]
}
elif _continue:
last_reply = [output['internal'][-1][1], output['visible'][-1][1]]
if loading_message:
yield {
'visible': output['visible'][:-1] + [[visible_text, last_reply[1] + '...']],
'internal': output['internal']
}
# Generate the prompt
kwargs = {
'_continue': _continue,
'history': output if _continue else {k: v[:-1] for k, v in output.items()}
}
prompt = apply_extensions('custom_generate_chat_prompt', text, state, **kwargs)
if prompt is None:
prompt = generate_chat_prompt(text, state, **kwargs)
# Generate
reply = None
for j, reply in enumerate(generate_reply(prompt, state, stopping_strings=stopping_strings, is_chat=True, for_ui=for_ui)):
# Extract the reply
visible_reply = reply
if state['mode'] in ['chat', 'chat-instruct']:
visible_reply = re.sub("(<USER>|<user>|{{user}})", state['name1'], reply)
visible_reply = html.escape(visible_reply)
if shared.stop_everything:
output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
yield output
return
if _continue:
output['internal'][-1] = [text, last_reply[0] + reply]
output['visible'][-1] = [visible_text, last_reply[1] + visible_reply]
if is_stream:
yield output
elif not (j == 0 and visible_reply.strip() == ''):
output['internal'][-1] = [text, reply.lstrip(' ')]
output['visible'][-1] = [visible_text, visible_reply.lstrip(' ')]
if is_stream:
yield output
output['visible'][-1][1] = apply_extensions('output', output['visible'][-1][1], state, is_chat=True)
yield output
def impersonate_wrapper(text, state):
static_output = chat_html_wrapper(state['history'], state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
prompt = generate_chat_prompt('', state, impersonate=True)
stopping_strings = get_stopping_strings(state)
yield text + '...', static_output
reply = None
for reply in generate_reply(prompt + text, state, stopping_strings=stopping_strings, is_chat=True):
yield (text + reply).lstrip(' '), static_output
if shared.stop_everything:
return
def generate_chat_reply(text, state, regenerate=False, _continue=False, loading_message=True, for_ui=False):
history = state['history']
if regenerate or _continue:
text = ''
if (len(history['visible']) == 1 and not history['visible'][0][0]) or len(history['internal']) == 0:
yield history
return
for history in chatbot_wrapper(text, state, regenerate=regenerate, _continue=_continue, loading_message=loading_message, for_ui=for_ui):
yield history
def character_is_loaded(state, raise_exception=False):
if state['mode'] in ['chat', 'chat-instruct'] and state['name2'] == '':
logger.error('It looks like no character is loaded. Please load one under Parameters > Character.')
if raise_exception:
raise ValueError
return False
else:
return True
def generate_chat_reply_wrapper(text, state, regenerate=False, _continue=False):
'''
Same as above but returns HTML for the UI
'''
if not character_is_loaded(state):
return
if state['start_with'] != '' and not _continue:
if regenerate:
text, state['history'] = remove_last_message(state['history'])
regenerate = False
_continue = True
send_dummy_message(text, state)
send_dummy_reply(state['start_with'], state)
history = state['history']
for i, history in enumerate(generate_chat_reply(text, state, regenerate, _continue, loading_message=True, for_ui=True)):
yield chat_html_wrapper(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu']), history
save_history(history, state['unique_id'], state['character_menu'], state['mode'])
def remove_last_message(history):
if len(history['visible']) > 0 and history['internal'][-1][0] != '<|BEGIN-VISIBLE-CHAT|>':
last = history['visible'].pop()
history['internal'].pop()
else:
last = ['', '']
return html.unescape(last[0]), history
def send_last_reply_to_input(history):
if len(history['visible']) > 0:
return html.unescape(history['visible'][-1][1])
else:
return ''
def replace_last_reply(text, state):
history = state['history']
if len(text.strip()) == 0:
return history
elif len(history['visible']) > 0:
history['visible'][-1][1] = html.escape(text)
history['internal'][-1][1] = apply_extensions('input', text, state, is_chat=True)
return history
def send_dummy_message(text, state):
history = state['history']
history['visible'].append([html.escape(text), ''])
history['internal'].append([apply_extensions('input', text, state, is_chat=True), ''])
return history
def send_dummy_reply(text, state):
history = state['history']
if len(history['visible']) > 0 and not history['visible'][-1][1] == '':
history['visible'].append(['', ''])
history['internal'].append(['', ''])
history['visible'][-1][1] = html.escape(text)
history['internal'][-1][1] = apply_extensions('input', text, state, is_chat=True)
return history
def redraw_html(history, name1, name2, mode, style, character, reset_cache=False):
return chat_html_wrapper(history, name1, name2, mode, style, character, reset_cache=reset_cache)
def start_new_chat(state):
mode = state['mode']
history = {'internal': [], 'visible': []}
if mode != 'instruct':
greeting = replace_character_names(state['greeting'], state['name1'], state['name2'])
if greeting != '':
history['internal'] += [['<|BEGIN-VISIBLE-CHAT|>', greeting]]
history['visible'] += [['', apply_extensions('output', html.escape(greeting), state, is_chat=True)]]
unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
save_history(history, unique_id, state['character_menu'], state['mode'])
return history
def get_history_file_path(unique_id, character, mode):
if mode == 'instruct':
p = Path(f'logs/instruct/{unique_id}.json')
else:
p = Path(f'logs/chat/{character}/{unique_id}.json')
return p
def save_history(history, unique_id, character, mode):
if shared.args.multi_user:
return
p = get_history_file_path(unique_id, character, mode)
if not p.parent.is_dir():
p.parent.mkdir(parents=True)
with open(p, 'w', encoding='utf-8') as f:
f.write(json.dumps(history, indent=4, ensure_ascii=False))
def rename_history(old_id, new_id, character, mode):
if shared.args.multi_user:
return
old_p = get_history_file_path(old_id, character, mode)
new_p = get_history_file_path(new_id, character, mode)
if new_p.parent != old_p.parent:
logger.error(f"The following path is not allowed: \"{new_p}\".")
elif new_p == old_p:
logger.info("The provided path is identical to the old one.")
elif new_p.exists():
logger.error(f"The new path already exists and will not be overwritten: \"{new_p}\".")
else:
logger.info(f"Renaming \"{old_p}\" to \"{new_p}\"")
old_p.rename(new_p)
def get_paths(state):
if state['mode'] == 'instruct':
return Path('logs/instruct').glob('*.json')
else:
character = state['character_menu']
# Handle obsolete filenames and paths
old_p = Path(f'logs/{character}_persistent.json')
new_p = Path(f'logs/persistent_{character}.json')
if old_p.exists():
logger.warning(f"Renaming \"{old_p}\" to \"{new_p}\"")
old_p.rename(new_p)
if new_p.exists():
unique_id = datetime.now().strftime('%Y%m%d-%H-%M-%S')
p = get_history_file_path(unique_id, character, state['mode'])
logger.warning(f"Moving \"{new_p}\" to \"{p}\"")
p.parent.mkdir(exist_ok=True)
new_p.rename(p)
return Path(f'logs/chat/{character}').glob('*.json')
def find_all_histories(state):
if shared.args.multi_user:
return ['']
paths = get_paths(state)
histories = sorted(paths, key=lambda x: x.stat().st_mtime, reverse=True)
return [path.stem for path in histories]
def find_all_histories_with_first_prompts(state):
if shared.args.multi_user:
return []
paths = get_paths(state)
histories = sorted(paths, key=lambda x: x.stat().st_mtime, reverse=True)
result = []
for i, path in enumerate(histories):
filename = path.stem
if re.match(r'^[0-9]{8}-[0-9]{2}-[0-9]{2}-[0-9]{2}$', filename):
with open(path, 'r', encoding='utf-8') as f:
data = json.load(f)
first_prompt = ""
if data and 'visible' in data and len(data['visible']) > 0:
if data['internal'][0][0] == '<|BEGIN-VISIBLE-CHAT|>':
if len(data['visible']) > 1:
first_prompt = html.unescape(data['visible'][1][0])
elif i == 0:
first_prompt = "New chat"
else:
first_prompt = html.unescape(data['visible'][0][0])
elif i == 0:
first_prompt = "New chat"
else:
first_prompt = filename
first_prompt = first_prompt.strip()
# Truncate the first prompt if it's longer than 32 characters
if len(first_prompt) > 32:
first_prompt = first_prompt[:29] + '...'
result.append((first_prompt, filename))
return result
def load_latest_history(state):
'''
Loads the latest history for the given character in chat or chat-instruct
mode, or the latest instruct history for instruct mode.
'''
if shared.args.multi_user:
return start_new_chat(state)
histories = find_all_histories(state)
if len(histories) > 0:
history = load_history(histories[0], state['character_menu'], state['mode'])
else:
history = start_new_chat(state)
return history
def load_history_after_deletion(state, idx):
'''
Loads the latest history for the given character in chat or chat-instruct
mode, or the latest instruct history for instruct mode.
'''
if shared.args.multi_user:
return start_new_chat(state)
histories = find_all_histories_with_first_prompts(state)
idx = min(int(idx), len(histories) - 1)
idx = max(0, idx)
if len(histories) > 0:
history = load_history(histories[idx][1], state['character_menu'], state['mode'])
else:
history = start_new_chat(state)
histories = find_all_histories_with_first_prompts(state)
return history, gr.update(choices=histories, value=histories[idx][1])
def update_character_menu_after_deletion(idx):
characters = utils.get_available_characters()
idx = min(int(idx), len(characters) - 1)
idx = max(0, idx)
return gr.update(choices=characters, value=characters[idx])
def load_history(unique_id, character, mode):
p = get_history_file_path(unique_id, character, mode)
f = json.loads(open(p, 'rb').read())
if 'internal' in f and 'visible' in f:
history = f
else:
history = {
'internal': f['data'],
'visible': f['data_visible']
}
return history
def load_history_json(file, history):
try:
file = file.decode('utf-8')
f = json.loads(file)
if 'internal' in f and 'visible' in f:
history = f
else:
history = {
'internal': f['data'],
'visible': f['data_visible']
}
return history
except:
return history
def delete_history(unique_id, character, mode):
p = get_history_file_path(unique_id, character, mode)
delete_file(p)
def replace_character_names(text, name1, name2):
text = text.replace('{{user}}', name1).replace('{{char}}', name2)
return text.replace('<USER>', name1).replace('<BOT>', name2)
def generate_pfp_cache(character):
cache_folder = Path(shared.args.disk_cache_dir)
if not cache_folder.exists():
cache_folder.mkdir()
for path in [Path(f"characters/{character}.{extension}") for extension in ['png', 'jpg', 'jpeg']]:
if path.exists():
original_img = Image.open(path)
original_img.save(Path(f'{cache_folder}/pfp_character.png'), format='PNG')
thumb = make_thumbnail(original_img)
thumb.save(Path(f'{cache_folder}/pfp_character_thumb.png'), format='PNG')
return thumb
return None
def load_character(character, name1, name2):
context = greeting = ""
greeting_field = 'greeting'
picture = None
filepath = None
for extension in ["yml", "yaml", "json"]:
filepath = Path(f'characters/{character}.{extension}')
if filepath.exists():
break
if filepath is None or not filepath.exists():
logger.error(f"Could not find the character \"{character}\" inside characters/. No character has been loaded.")
raise ValueError
file_contents = open(filepath, 'r', encoding='utf-8').read()
data = json.loads(file_contents) if extension == "json" else yaml.safe_load(file_contents)
cache_folder = Path(shared.args.disk_cache_dir)
for path in [Path(f"{cache_folder}/pfp_character.png"), Path(f"{cache_folder}/pfp_character_thumb.png")]:
if path.exists():
path.unlink()
picture = generate_pfp_cache(character)
# Finding the bot's name
for k in ['name', 'bot', '<|bot|>', 'char_name']:
if k in data and data[k] != '':
name2 = data[k]
break
# Find the user name (if any)
for k in ['your_name', 'user', '<|user|>']:
if k in data and data[k] != '':
name1 = data[k]
break
if 'context' in data:
context = data['context'].strip()
elif "char_persona" in data:
context = build_pygmalion_style_context(data)
greeting_field = 'char_greeting'
greeting = data.get(greeting_field, greeting)
return name1, name2, picture, greeting, context
def load_instruction_template(template):
if template == 'None':
return ''
for filepath in [Path(f'instruction-templates/{template}.yaml'), Path('instruction-templates/Alpaca.yaml')]:
if filepath.exists():
break
else:
return ''
file_contents = open(filepath, 'r', encoding='utf-8').read()
data = yaml.safe_load(file_contents)
if 'instruction_template' in data:
return data['instruction_template']
else:
return jinja_template_from_old_format(data)
@functools.cache
def load_character_memoized(character, name1, name2):
return load_character(character, name1, name2)
@functools.cache
def load_instruction_template_memoized(template):
return load_instruction_template(template)
def upload_character(file, img, tavern=False):
decoded_file = file if isinstance(file, str) else file.decode('utf-8')
try:
data = json.loads(decoded_file)
except:
data = yaml.safe_load(decoded_file)
if 'char_name' in data:
name = data['char_name']
greeting = data['char_greeting']
context = build_pygmalion_style_context(data)
yaml_data = generate_character_yaml(name, greeting, context)
else:
name = data['name']
yaml_data = generate_character_yaml(data['name'], data['greeting'], data['context'])
outfile_name = name
i = 1
while Path(f'characters/{outfile_name}.yaml').exists():
outfile_name = f'{name}_{i:03d}'
i += 1
with open(Path(f'characters/{outfile_name}.yaml'), 'w', encoding='utf-8') as f:
f.write(yaml_data)
if img is not None:
img.save(Path(f'characters/{outfile_name}.png'))
logger.info(f'New character saved to "characters/{outfile_name}.yaml".')
return gr.update(value=outfile_name, choices=get_available_characters())
def build_pygmalion_style_context(data):
context = ""
if 'char_persona' in data and data['char_persona'] != '':
context += f"{data['char_name']}'s Persona: {data['char_persona']}\n"
if 'world_scenario' in data and data['world_scenario'] != '':
context += f"Scenario: {data['world_scenario']}\n"
if 'example_dialogue' in data and data['example_dialogue'] != '':
context += f"{data['example_dialogue'].strip()}\n"
context = f"{context.strip()}\n"
return context
def upload_tavern_character(img, _json):
_json = {'char_name': _json['name'], 'char_persona': _json['description'], 'char_greeting': _json['first_mes'], 'example_dialogue': _json['mes_example'], 'world_scenario': _json['scenario']}
return upload_character(json.dumps(_json), img, tavern=True)
def check_tavern_character(img):
if "chara" not in img.info:
return "Not a TavernAI card", None, None, gr.update(interactive=False)
decoded_string = base64.b64decode(img.info['chara']).replace(b'\\r\\n', b'\\n')
_json = json.loads(decoded_string)
if "data" in _json:
_json = _json["data"]
return _json['name'], _json['description'], _json, gr.update(interactive=True)
def upload_your_profile_picture(img):
cache_folder = Path(shared.args.disk_cache_dir)
if not cache_folder.exists():
cache_folder.mkdir()
if img is None:
if Path(f"{cache_folder}/pfp_me.png").exists():
Path(f"{cache_folder}/pfp_me.png").unlink()
else:
img = make_thumbnail(img)
img.save(Path(f'{cache_folder}/pfp_me.png'))
logger.info(f'Profile picture saved to "{cache_folder}/pfp_me.png"')
def generate_character_yaml(name, greeting, context):
data = {
'name': name,
'greeting': greeting,
'context': context,
}
data = {k: v for k, v in data.items() if v} # Strip falsy
return yaml.dump(data, sort_keys=False, width=float("inf"))
def generate_instruction_template_yaml(instruction_template):
data = {
'instruction_template': instruction_template
}
return my_yaml_output(data)
def save_character(name, greeting, context, picture, filename):
if filename == "":
logger.error("The filename is empty, so the character will not be saved.")
return
data = generate_character_yaml(name, greeting, context)
filepath = Path(f'characters/{filename}.yaml')
save_file(filepath, data)
path_to_img = Path(f'characters/{filename}.png')
if picture is not None:
picture.save(path_to_img)
logger.info(f'Saved {path_to_img}.')
def delete_character(name, instruct=False):
for extension in ["yml", "yaml", "json"]:
delete_file(Path(f'characters/{name}.{extension}'))
delete_file(Path(f'characters/{name}.png'))
def jinja_template_from_old_format(params, verbose=False):
MASTER_TEMPLATE = """
{%- set ns = namespace(found=false) -%}
{%- for message in messages -%}
{%- if message['role'] == 'system' -%}
{%- set ns.found = true -%}
{%- endif -%}
{%- endfor -%}
{%- if not ns.found -%}
{{- '<|PRE-SYSTEM|>' + '<|SYSTEM-MESSAGE|>' + '<|POST-SYSTEM|>' -}}
{%- endif %}
{%- for message in messages %}
{%- if message['role'] == 'system' -%}
{{- '<|PRE-SYSTEM|>' + message['content'] + '<|POST-SYSTEM|>' -}}
{%- else -%}
{%- if message['role'] == 'user' -%}
{{-'<|PRE-USER|>' + message['content'] + '<|POST-USER|>'-}}
{%- else -%}
{{-'<|PRE-ASSISTANT|>' + message['content'] + '<|POST-ASSISTANT|>' -}}
{%- endif -%}
{%- endif -%}
{%- endfor -%}
{%- if add_generation_prompt -%}
{{-'<|PRE-ASSISTANT-GENERATE|>'-}}
{%- endif -%}
"""
if 'context' in params and '<|system-message|>' in params['context']:
pre_system = params['context'].split('<|system-message|>')[0]
post_system = params['context'].split('<|system-message|>')[1]
else:
pre_system = ''
post_system = ''
pre_user = params['turn_template'].split('<|user-message|>')[0].replace('<|user|>', params['user'])
post_user = params['turn_template'].split('<|user-message|>')[1].split('<|bot|>')[0]
pre_assistant = '<|bot|>' + params['turn_template'].split('<|bot-message|>')[0].split('<|bot|>')[1]
pre_assistant = pre_assistant.replace('<|bot|>', params['bot'])
post_assistant = params['turn_template'].split('<|bot-message|>')[1]
def preprocess(string):
return string.replace('\n', '\\n').replace('\'', '\\\'')
pre_system = preprocess(pre_system)
post_system = preprocess(post_system)
pre_user = preprocess(pre_user)
post_user = preprocess(post_user)
pre_assistant = preprocess(pre_assistant)
post_assistant = preprocess(post_assistant)
if verbose:
print(
'\n',
repr(pre_system) + '\n',
repr(post_system) + '\n',
repr(pre_user) + '\n',
repr(post_user) + '\n',
repr(pre_assistant) + '\n',
repr(post_assistant) + '\n',
)
result = MASTER_TEMPLATE
if 'system_message' in params:
result = result.replace('<|SYSTEM-MESSAGE|>', preprocess(params['system_message']))
else:
result = result.replace('<|SYSTEM-MESSAGE|>', '')
result = result.replace('<|PRE-SYSTEM|>', pre_system)
result = result.replace('<|POST-SYSTEM|>', post_system)
result = result.replace('<|PRE-USER|>', pre_user)
result = result.replace('<|POST-USER|>', post_user)
result = result.replace('<|PRE-ASSISTANT|>', pre_assistant)
result = result.replace('<|PRE-ASSISTANT-GENERATE|>', pre_assistant.rstrip(' '))
result = result.replace('<|POST-ASSISTANT|>', post_assistant)
result = result.strip()
return result
def my_yaml_output(data):
'''
pyyaml is very inconsistent with multiline strings.
for simple instruction template outputs, this is enough.
'''
result = ""
for k in data:
result += k + ": |-\n"
for line in data[k].splitlines():
result += " " + line.rstrip(' ') + "\n"
return result
def handle_replace_last_reply_click(text, state):
history = replace_last_reply(text, state)
save_history(history, state['unique_id'], state['character_menu'], state['mode'])
html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
return [history, html, ""]
def handle_send_dummy_message_click(text, state):
history = send_dummy_message(text, state)
save_history(history, state['unique_id'], state['character_menu'], state['mode'])
html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
return [history, html, ""]
def handle_send_dummy_reply_click(text, state):
history = send_dummy_reply(text, state)
save_history(history, state['unique_id'], state['character_menu'], state['mode'])
html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
return [history, html, ""]
def handle_remove_last_click(state):
last_input, history = remove_last_message(state['history'])
save_history(history, state['unique_id'], state['character_menu'], state['mode'])
html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
return [history, html, last_input]
def handle_unique_id_select(state):
history = load_history(state['unique_id'], state['character_menu'], state['mode'])
html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
convert_to_markdown.cache_clear()
return [history, html]
def handle_start_new_chat_click(state):
history = start_new_chat(state)
histories = find_all_histories_with_first_prompts(state)
html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
convert_to_markdown.cache_clear()
return [history, html, gr.update(choices=histories, value=histories[0][1])]
def handle_delete_chat_confirm_click(state):
index = str(find_all_histories(state).index(state['unique_id']))
delete_history(state['unique_id'], state['character_menu'], state['mode'])
history, unique_id = load_history_after_deletion(state, index)
html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
convert_to_markdown.cache_clear()
return [
history,
html,
unique_id,
gr.update(visible=False),
gr.update(visible=True),
gr.update(visible=False)
]
def handle_rename_chat_click():
return [
gr.update(visible=True, value="My New Chat"),
gr.update(visible=True),
gr.update(visible=True)
]
def handle_rename_chat_confirm(rename_to, state):
rename_history(state['unique_id'], rename_to, state['character_menu'], state['mode'])
histories = find_all_histories_with_first_prompts(state)
return [
gr.update(choices=histories, value=rename_to),
gr.update(visible=False),
gr.update(visible=False),
gr.update(visible=False)
]
def handle_upload_chat_history(load_chat_history, state):
history = start_new_chat(state)
history = load_history_json(load_chat_history, history)
histories = find_all_histories_with_first_prompts(state)
save_history(history, state['unique_id'], state['character_menu'], state['mode'])
html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
convert_to_markdown.cache_clear()
return [
history,
html,
gr.update(choices=histories, value=histories[0][1])
]
def handle_character_menu_change(state):
name1, name2, picture, greeting, context = load_character(state['character_menu'], state['name1'], state['name2'])
state['name1'] = name1
state['name2'] = name2
state['character_picture'] = picture
state['greeting'] = greeting
state['context'] = context
history = load_latest_history(state)
histories = find_all_histories_with_first_prompts(state)
html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
convert_to_markdown.cache_clear()
return [
history,
html,
name1,
name2,
picture,
greeting,
context,
gr.update(choices=histories, value=histories[0][1]),
]
def handle_mode_change(state):
history = load_latest_history(state)
histories = find_all_histories_with_first_prompts(state)
html = redraw_html(history, state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'])
convert_to_markdown.cache_clear()
return [
history,
html,
gr.update(visible=state['mode'] != 'instruct'),
gr.update(visible=state['mode'] == 'chat-instruct'),
gr.update(choices=histories, value=histories[0][1])
]
def handle_save_character_click(name2):
return [
name2,
gr.update(visible=True)
]
def handle_load_template_click(instruction_template):
output = load_instruction_template(instruction_template)
return [
output,
"Select template to load..."
]
def handle_save_template_click(instruction_template_str):
contents = generate_instruction_template_yaml(instruction_template_str)
return [
"My Template.yaml",
"instruction-templates/",
contents,
gr.update(visible=True)
]
def handle_delete_template_click(template):
return [
f"{template}.yaml",
"instruction-templates/",
gr.update(visible=True)
]
def handle_your_picture_change(picture, state):
upload_your_profile_picture(picture)
html = redraw_html(state['history'], state['name1'], state['name2'], state['mode'], state['chat_style'], state['character_menu'], reset_cache=True)
return html
def handle_send_instruction_click(state):
state['mode'] = 'instruct'
state['history'] = {'internal': [], 'visible': []}
output = generate_chat_prompt("Input", state)
return output
def handle_send_chat_click(state):
output = generate_chat_prompt("", state, _continue=True)
return output
|