Merge pull request #841 from KelvinF97/master
Optimize some code and fix some bugs
This commit is contained in:
binary-husky
GitHub
@ -1,10 +1,10 @@
|
||||
|
||||
def check_proxy(proxies):
|
||||
def check_proxy(proxies: dict):
|
||||
import requests
|
||||
proxies_https = proxies['https'] if proxies is not None else '无'
|
||||
proxies_https = proxies.get('https') if proxies is not None else '无'
|
||||
try:
|
||||
response = requests.get("https://ipapi.co/json/",
|
||||
proxies=proxies, timeout=4)
|
||||
proxies=proxies, timeout=30)
|
||||
data = response.json()
|
||||
print(f'查询代理的地理位置，返回的结果是{data}')
|
||||
if 'country_name' in data:
|
||||
@ -16,8 +16,8 @@ def check_proxy(proxies):
|
||||
result = f"代理配置 {proxies_https}, 代理数据解析失败：{data}"
|
||||
print(result)
|
||||
return result
|
||||
except:
|
||||
result = f"代理配置 {proxies_https}, 代理所在地查询超时，代理可能无效"
|
||||
except Exception as e:
|
||||
result = f"代理 {proxies_https} 查询出现异常: {e}，代理可能无效"
|
||||
print(result)
|
||||
return result
|
||||
|
||||
|
||||
@ -1,16 +1,19 @@
|
||||
from toolbox import update_ui, get_conf, trimmed_format_exc
|
||||
import threading
|
||||
|
||||
|
||||
def input_clipping(inputs, history, max_token_limit):
|
||||
import numpy as np
|
||||
from request_llm.bridge_all import model_info
|
||||
enc = model_info["gpt-3.5-turbo"]['tokenizer']
|
||||
def get_token_num(txt): return len(enc.encode(txt, disallowed_special=()))
|
||||
|
||||
def get_token_num(txt):
|
||||
return len(enc.encode(txt, disallowed_special=()))
|
||||
|
||||
mode = 'input-and-history'
|
||||
# 当 输入部分的token占比 小于 全文的一半时，只裁剪历史
|
||||
input_token_num = get_token_num(inputs)
|
||||
if input_token_num < max_token_limit//2:
|
||||
if input_token_num < max_token_limit // 2:
|
||||
mode = 'only-history'
|
||||
max_token_limit = max_token_limit - input_token_num
|
||||
|
||||
@ -18,13 +21,13 @@ def input_clipping(inputs, history, max_token_limit):
|
||||
everything.extend(history)
|
||||
n_token = get_token_num('\n'.join(everything))
|
||||
everything_token = [get_token_num(e) for e in everything]
|
||||
delta = max(everything_token) // 16 # 截断时的颗粒度
|
||||
|
||||
delta = max(everything_token) // 16 # 截断时的颗粒度
|
||||
|
||||
while n_token > max_token_limit:
|
||||
where = np.argmax(everything_token)
|
||||
encoded = enc.encode(everything[where], disallowed_special=())
|
||||
clipped_encoded = encoded[:len(encoded)-delta]
|
||||
everything[where] = enc.decode(clipped_encoded)[:-1] # -1 to remove the may-be illegal char
|
||||
clipped_encoded = encoded[:len(encoded) - delta]
|
||||
everything[where] = enc.decode(clipped_encoded)[:-1] # -1 to remove the may-be illegal char
|
||||
everything_token[where] = get_token_num(everything[where])
|
||||
n_token = get_token_num('\n'.join(everything))
|
||||
|
||||
@ -35,12 +38,13 @@ def input_clipping(inputs, history, max_token_limit):
|
||||
history = everything[1:]
|
||||
return inputs, history
|
||||
|
||||
|
||||
def request_gpt_model_in_new_thread_with_ui_alive(
|
||||
inputs, inputs_show_user, llm_kwargs,
|
||||
inputs, inputs_show_user, llm_kwargs,
|
||||
chatbot, history, sys_prompt, refresh_interval=0.2,
|
||||
handle_token_exceed=True,
|
||||
handle_token_exceed=True,
|
||||
retry_times_at_unknown_error=2,
|
||||
):
|
||||
):
|
||||
"""
|
||||
Request GPT model，请求GPT模型同时维持用户界面活跃。
|
||||
|
||||
@ -64,15 +68,16 @@ def request_gpt_model_in_new_thread_with_ui_alive(
|
||||
from request_llm.bridge_all import predict_no_ui_long_connection
|
||||
# 用户反馈
|
||||
chatbot.append([inputs_show_user, ""])
|
||||
yield from update_ui(chatbot=chatbot, history=[]) # 刷新界面
|
||||
yield from update_ui(chatbot=chatbot, history=[]) # 刷新界面
|
||||
executor = ThreadPoolExecutor(max_workers=16)
|
||||
mutable = ["", time.time(), ""]
|
||||
|
||||
def _req_gpt(inputs, history, sys_prompt):
|
||||
retry_op = retry_times_at_unknown_error
|
||||
exceeded_cnt = 0
|
||||
while True:
|
||||
# watchdog error
|
||||
if len(mutable) >= 2 and (time.time()-mutable[1]) > 5:
|
||||
if len(mutable) >= 2 and (time.time() - mutable[1]) > 5:
|
||||
raise RuntimeError("检测到程序终止。")
|
||||
try:
|
||||
# 【第一种情况】：顺利完成
|
||||
@ -89,14 +94,14 @@ def request_gpt_model_in_new_thread_with_ui_alive(
|
||||
p_ratio, n_exceed = get_reduce_token_percent(str(token_exceeded_error))
|
||||
MAX_TOKEN = 4096
|
||||
EXCEED_ALLO = 512 + 512 * exceeded_cnt
|
||||
inputs, history = input_clipping(inputs, history, max_token_limit=MAX_TOKEN-EXCEED_ALLO)
|
||||
inputs, history = input_clipping(inputs, history, max_token_limit=MAX_TOKEN - EXCEED_ALLO)
|
||||
mutable[0] += f'[Local Message] 警告，文本过长将进行截断，Token溢出数：{n_exceed}。\n\n'
|
||||
continue # 返回重试
|
||||
continue # 返回重试
|
||||
else:
|
||||
# 【选择放弃】
|
||||
tb_str = '```\n' + trimmed_format_exc() + '```'
|
||||
mutable[0] += f"[Local Message] 警告，在执行过程中遭遇问题, Traceback：\n\n{tb_str}\n\n"
|
||||
return mutable[0] # 放弃
|
||||
return mutable[0] # 放弃
|
||||
except:
|
||||
# 【第三种情况】：其他错误：重试几次
|
||||
tb_str = '```\n' + trimmed_format_exc() + '```'
|
||||
@ -104,14 +109,15 @@ def request_gpt_model_in_new_thread_with_ui_alive(
|
||||
mutable[0] += f"[Local Message] 警告，在执行过程中遭遇问题, Traceback：\n\n{tb_str}\n\n"
|
||||
if retry_op > 0:
|
||||
retry_op -= 1
|
||||
mutable[0] += f"[Local Message] 重试中，请稍等 {retry_times_at_unknown_error-retry_op}/{retry_times_at_unknown_error}：\n\n"
|
||||
mutable[
|
||||
0] += f"[Local Message] 重试中，请稍等 {retry_times_at_unknown_error - retry_op}/{retry_times_at_unknown_error}：\n\n"
|
||||
if ("Rate limit reached" in tb_str) or ("Too Many Requests" in tb_str):
|
||||
time.sleep(30)
|
||||
time.sleep(5)
|
||||
continue # 返回重试
|
||||
continue # 返回重试
|
||||
else:
|
||||
time.sleep(5)
|
||||
return mutable[0] # 放弃
|
||||
return mutable[0] # 放弃
|
||||
|
||||
# 提交任务
|
||||
future = executor.submit(_req_gpt, inputs, history, sys_prompt)
|
||||
@ -123,21 +129,21 @@ def request_gpt_model_in_new_thread_with_ui_alive(
|
||||
if future.done():
|
||||
break
|
||||
chatbot[-1] = [chatbot[-1][0], mutable[0]]
|
||||
yield from update_ui(chatbot=chatbot, history=[]) # 刷新界面
|
||||
yield from update_ui(chatbot=chatbot, history=[]) # 刷新界面
|
||||
|
||||
final_result = future.result()
|
||||
chatbot[-1] = [chatbot[-1][0], final_result]
|
||||
yield from update_ui(chatbot=chatbot, history=[]) # 如果最后成功了，则删除报错信息
|
||||
yield from update_ui(chatbot=chatbot, history=[]) # 如果最后成功了，则删除报错信息
|
||||
return final_result
|
||||
|
||||
|
||||
def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
|
||||
inputs_array, inputs_show_user_array, llm_kwargs,
|
||||
chatbot, history_array, sys_prompt_array,
|
||||
inputs_array, inputs_show_user_array, llm_kwargs,
|
||||
chatbot, history_array, sys_prompt_array,
|
||||
refresh_interval=0.2, max_workers=-1, scroller_max_len=30,
|
||||
handle_token_exceed=True, show_user_at_complete=False,
|
||||
retry_times_at_unknown_error=2,
|
||||
):
|
||||
):
|
||||
"""
|
||||
Request GPT model using multiple threads with UI and high efficiency
|
||||
请求GPT模型的[多线程]版。
|
||||
@ -170,19 +176,21 @@ def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
|
||||
from request_llm.bridge_all import predict_no_ui_long_connection
|
||||
assert len(inputs_array) == len(history_array)
|
||||
assert len(inputs_array) == len(sys_prompt_array)
|
||||
if max_workers == -1: # 读取配置文件
|
||||
try: max_workers, = get_conf('DEFAULT_WORKER_NUM')
|
||||
except: max_workers = 8
|
||||
if max_workers == -1: # 读取配置文件
|
||||
try:
|
||||
max_workers, = get_conf('DEFAULT_WORKER_NUM')
|
||||
except:
|
||||
max_workers = 8
|
||||
if max_workers <= 0: max_workers = 3
|
||||
# 屏蔽掉 chatglm的多线程，可能会导致严重卡顿
|
||||
if not (llm_kwargs['llm_model'].startswith('gpt-') or llm_kwargs['llm_model'].startswith('api2d-')):
|
||||
max_workers = 1
|
||||
|
||||
|
||||
executor = ThreadPoolExecutor(max_workers=max_workers)
|
||||
n_frag = len(inputs_array)
|
||||
# 用户反馈
|
||||
chatbot.append(["请开始多线程操作。", ""])
|
||||
yield from update_ui(chatbot=chatbot, history=[]) # 刷新界面
|
||||
yield from update_ui(chatbot=chatbot, history=[]) # 刷新界面
|
||||
# 跨线程传递
|
||||
mutable = [["", time.time(), "等待中"] for _ in range(n_frag)]
|
||||
|
||||
@ -194,13 +202,13 @@ def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
|
||||
mutable[index][2] = "执行中"
|
||||
while True:
|
||||
# watchdog error
|
||||
if len(mutable[index]) >= 2 and (time.time()-mutable[index][1]) > 5:
|
||||
if len(mutable[index]) >= 2 and (time.time() - mutable[index][1]) > 5:
|
||||
raise RuntimeError("检测到程序终止。")
|
||||
try:
|
||||
# 【第一种情况】：顺利完成
|
||||
# time.sleep(10); raise RuntimeError("测试")
|
||||
gpt_say = predict_no_ui_long_connection(
|
||||
inputs=inputs, llm_kwargs=llm_kwargs, history=history,
|
||||
inputs=inputs, llm_kwargs=llm_kwargs, history=history,
|
||||
sys_prompt=sys_prompt, observe_window=mutable[index], console_slience=True
|
||||
)
|
||||
mutable[index][2] = "已成功"
|
||||
@ -214,24 +222,26 @@ def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
|
||||
p_ratio, n_exceed = get_reduce_token_percent(str(token_exceeded_error))
|
||||
MAX_TOKEN = 4096
|
||||
EXCEED_ALLO = 512 + 512 * exceeded_cnt
|
||||
inputs, history = input_clipping(inputs, history, max_token_limit=MAX_TOKEN-EXCEED_ALLO)
|
||||
inputs, history = input_clipping(inputs, history, max_token_limit=MAX_TOKEN - EXCEED_ALLO)
|
||||
gpt_say += f'[Local Message] 警告，文本过长将进行截断，Token溢出数：{n_exceed}。\n\n'
|
||||
mutable[index][2] = f"截断重试"
|
||||
continue # 返回重试
|
||||
continue # 返回重试
|
||||
else:
|
||||
# 【选择放弃】
|
||||
tb_str = '```\n' + trimmed_format_exc() + '```'
|
||||
gpt_say += f"[Local Message] 警告，线程{index}在执行过程中遭遇问题, Traceback：\n\n{tb_str}\n\n"
|
||||
if len(mutable[index][0]) > 0: gpt_say += "此线程失败前收到的回答：\n\n" + mutable[index][0]
|
||||
if len(mutable[index][0]) > 0:
|
||||
gpt_say += "此线程失败前收到的回答：\n\n" + mutable[index][0]
|
||||
mutable[index][2] = "输入过长已放弃"
|
||||
return gpt_say # 放弃
|
||||
except:
|
||||
return gpt_say # 放弃
|
||||
except Exception as e:
|
||||
# 【第三种情况】：其他错误
|
||||
tb_str = '```\n' + trimmed_format_exc() + '```'
|
||||
print(tb_str)
|
||||
print(f"发生异常：{e}, 调用栈信息：{tb_str}")
|
||||
gpt_say += f"[Local Message] 警告，线程{index}在执行过程中遭遇问题, Traceback：\n\n{tb_str}\n\n"
|
||||
if len(mutable[index][0]) > 0: gpt_say += "此线程失败前收到的回答：\n\n" + mutable[index][0]
|
||||
if retry_op > 0:
|
||||
if len(mutable[index][0]) > 0:
|
||||
gpt_say += "此线程失败前收到的回答：\n\n" + mutable[index][0]
|
||||
if retry_op > 0:
|
||||
retry_op -= 1
|
||||
wait = random.randint(5, 20)
|
||||
if ("Rate limit reached" in tb_str) or ("Too Many Requests" in tb_str):
|
||||
@ -241,19 +251,22 @@ def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
|
||||
fail_info = ""
|
||||
# 也许等待十几秒后，情况会好转
|
||||
for i in range(wait):
|
||||
mutable[index][2] = f"{fail_info}等待重试 {wait-i}"; time.sleep(1)
|
||||
mutable[index][2] = f"{fail_info}等待重试 {wait - i}";
|
||||
time.sleep(1)
|
||||
# 开始重试
|
||||
mutable[index][2] = f"重试中 {retry_times_at_unknown_error-retry_op}/{retry_times_at_unknown_error}"
|
||||
continue # 返回重试
|
||||
mutable[index][
|
||||
2] = f"重试中 {retry_times_at_unknown_error - retry_op}/{retry_times_at_unknown_error}"
|
||||
continue # 返回重试
|
||||
else:
|
||||
mutable[index][2] = "已失败"
|
||||
wait = 5
|
||||
time.sleep(5)
|
||||
return gpt_say # 放弃
|
||||
return gpt_say # 放弃
|
||||
|
||||
# 异步任务开始
|
||||
futures = [executor.submit(_req_gpt, index, inputs, history, sys_prompt) for index, inputs, history, sys_prompt in zip(
|
||||
range(len(inputs_array)), inputs_array, history_array, sys_prompt_array)]
|
||||
futures = [executor.submit(_req_gpt, index, inputs, history, sys_prompt) for index, inputs, history, sys_prompt in
|
||||
zip(
|
||||
range(len(inputs_array)), inputs_array, history_array, sys_prompt_array)]
|
||||
cnt = 0
|
||||
while True:
|
||||
# yield一次以刷新前端页面
|
||||
@ -267,17 +280,17 @@ def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
|
||||
mutable[thread_index][1] = time.time()
|
||||
# 在前端打印些好玩的东西
|
||||
for thread_index, _ in enumerate(worker_done):
|
||||
print_something_really_funny = "[ ...`"+mutable[thread_index][0][-scroller_max_len:].\
|
||||
print_something_really_funny = "[ ...`" + mutable[thread_index][0][-scroller_max_len:]. \
|
||||
replace('\n', '').replace('```', '...').replace(
|
||||
' ', '.').replace('<br/>', '.....').replace('$', '.')+"`... ]"
|
||||
' ', '.').replace('<br/>', '.....').replace('$', '.') + "`... ]"
|
||||
observe_win.append(print_something_really_funny)
|
||||
# 在前端打印些好玩的东西
|
||||
stat_str = ''.join([f'`{mutable[thread_index][2]}`: {obs}\n\n'
|
||||
if not done else f'`{mutable[thread_index][2]}`\n\n'
|
||||
stat_str = ''.join([f'`{mutable[thread_index][2]}`: {obs}\n\n'
|
||||
if not done else f'`{mutable[thread_index][2]}`\n\n'
|
||||
for thread_index, done, obs in zip(range(len(worker_done)), worker_done, observe_win)])
|
||||
# 在前端打印些好玩的东西
|
||||
chatbot[-1] = [chatbot[-1][0], f'多线程操作已经开始，完成情况: \n\n{stat_str}' + ''.join(['.']*(cnt % 10+1))]
|
||||
yield from update_ui(chatbot=chatbot, history=[]) # 刷新界面
|
||||
chatbot[-1] = [chatbot[-1][0], f'多线程操作已经开始，完成情况: \n\n{stat_str}' + ''.join(['.'] * (cnt % 10 + 1))]
|
||||
yield from update_ui(chatbot=chatbot, history=[]) # 刷新界面
|
||||
if all(worker_done):
|
||||
executor.shutdown()
|
||||
break
|
||||
@ -287,13 +300,13 @@ def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
|
||||
for inputs_show_user, f in zip(inputs_show_user_array, futures):
|
||||
gpt_res = f.result()
|
||||
gpt_response_collection.extend([inputs_show_user, gpt_res])
|
||||
|
||||
|
||||
# 是否在结束时，在界面上显示结果
|
||||
if show_user_at_complete:
|
||||
for inputs_show_user, f in zip(inputs_show_user_array, futures):
|
||||
gpt_res = f.result()
|
||||
chatbot.append([inputs_show_user, gpt_res])
|
||||
yield from update_ui(chatbot=chatbot, history=[]) # 刷新界面
|
||||
yield from update_ui(chatbot=chatbot, history=[]) # 刷新界面
|
||||
time.sleep(0.3)
|
||||
return gpt_response_collection
|
||||
|
||||
@ -306,6 +319,7 @@ def breakdown_txt_to_satisfy_token_limit(txt, get_token_fn, limit):
|
||||
lines = txt_tocut.split('\n')
|
||||
estimated_line_cut = limit / get_token_fn(txt_tocut) * len(lines)
|
||||
estimated_line_cut = int(estimated_line_cut)
|
||||
cnt = 0
|
||||
for cnt in reversed(range(estimated_line_cut)):
|
||||
if must_break_at_empty_line:
|
||||
if lines[cnt] != "":
|
||||
@ -322,6 +336,7 @@ def breakdown_txt_to_satisfy_token_limit(txt, get_token_fn, limit):
|
||||
result = [prev]
|
||||
result.extend(cut(post, must_break_at_empty_line))
|
||||
return result
|
||||
|
||||
try:
|
||||
return cut(txt, must_break_at_empty_line=True)
|
||||
except RuntimeError:
|
||||
@ -337,9 +352,10 @@ def force_breakdown(txt, limit, get_token_fn):
|
||||
return txt[:i], txt[i:]
|
||||
return "Tiktoken未知错误", "Tiktoken未知错误"
|
||||
|
||||
|
||||
def breakdown_txt_to_satisfy_token_limit_for_pdf(txt, get_token_fn, limit):
|
||||
# 递归
|
||||
def cut(txt_tocut, must_break_at_empty_line, break_anyway=False):
|
||||
def cut(txt_tocut, must_break_at_empty_line, break_anyway=False):
|
||||
if get_token_fn(txt_tocut) <= limit:
|
||||
return [txt_tocut]
|
||||
else:
|
||||
@ -365,6 +381,7 @@ def breakdown_txt_to_satisfy_token_limit_for_pdf(txt, get_token_fn, limit):
|
||||
result = [prev]
|
||||
result.extend(cut(post, must_break_at_empty_line, break_anyway=break_anyway))
|
||||
return result
|
||||
|
||||
try:
|
||||
# 第1次尝试，将双空行（\n\n）作为切分点
|
||||
return cut(txt, must_break_at_empty_line=True)
|
||||
@ -375,7 +392,7 @@ def breakdown_txt_to_satisfy_token_limit_for_pdf(txt, get_token_fn, limit):
|
||||
except RuntimeError:
|
||||
try:
|
||||
# 第3次尝试，将英文句号（.）作为切分点
|
||||
res = cut(txt.replace('.', '。\n'), must_break_at_empty_line=False) # 这个中文的句号是故意的，作为一个标识而存在
|
||||
res = cut(txt.replace('.', '。\n'), must_break_at_empty_line=False) # 这个中文的句号是故意的，作为一个标识而存在
|
||||
return [r.replace('。\n', '.') for r in res]
|
||||
except RuntimeError as e:
|
||||
try:
|
||||
@ -387,7 +404,6 @@ def breakdown_txt_to_satisfy_token_limit_for_pdf(txt, get_token_fn, limit):
|
||||
return cut(txt, must_break_at_empty_line=False, break_anyway=True)
|
||||
|
||||
|
||||
|
||||
def read_and_clean_pdf_text(fp):
|
||||
"""
|
||||
这个函数用于分割pdf，用了很多trick，逻辑较乱，效果奇好
|
||||
@ -415,8 +431,9 @@ def read_and_clean_pdf_text(fp):
|
||||
fc = 0 # Index 0 文本
|
||||
fs = 1 # Index 1 字体
|
||||
fb = 2 # Index 2 框框
|
||||
REMOVE_FOOT_NOTE = True # 是否丢弃掉 不是正文的内容 （比正文字体小，如参考文献、脚注、图注等）
|
||||
REMOVE_FOOT_FFSIZE_PERCENT = 0.95 # 小于正文的？时，判定为不是正文（有些文章的正文部分字体大小不是100%统一的，有肉眼不可见的小变化）
|
||||
REMOVE_FOOT_NOTE = True # 是否丢弃掉 不是正文的内容 （比正文字体小，如参考文献、脚注、图注等）
|
||||
REMOVE_FOOT_FFSIZE_PERCENT = 0.95 # 小于正文的？时，判定为不是正文（有些文章的正文部分字体大小不是100%统一的，有肉眼不可见的小变化）
|
||||
|
||||
def primary_ffsize(l):
|
||||
"""
|
||||
提取文本块主字体
|
||||
@ -426,12 +443,12 @@ def read_and_clean_pdf_text(fp):
|
||||
if wtf['size'] not in fsize_statiscs: fsize_statiscs[wtf['size']] = 0
|
||||
fsize_statiscs[wtf['size']] += len(wtf['text'])
|
||||
return max(fsize_statiscs, key=fsize_statiscs.get)
|
||||
|
||||
def ffsize_same(a,b):
|
||||
|
||||
def ffsize_same(a, b):
|
||||
"""
|
||||
提取字体大小是否近似相等
|
||||
"""
|
||||
return abs((a-b)/max(a,b)) < 0.02
|
||||
return abs((a - b) / max(a, b)) < 0.02
|
||||
|
||||
with fitz.open(fp) as doc:
|
||||
meta_txt = []
|
||||
@ -451,18 +468,19 @@ def read_and_clean_pdf_text(fp):
|
||||
if len(txt_line) == 0: continue
|
||||
pf = primary_ffsize(l)
|
||||
meta_line.append([txt_line, pf, l['bbox'], l])
|
||||
for wtf in l['spans']: # for l in t['lines']:
|
||||
for wtf in l['spans']: # for l in t['lines']:
|
||||
meta_span.append([wtf['text'], wtf['size'], len(wtf['text'])])
|
||||
# meta_line.append(["NEW_BLOCK", pf])
|
||||
# 块元提取 for each word segment with in line for each line cross-line words for each block
|
||||
# 块元提取 for each word segment with in line for each line
|
||||
# cross-line words for each block
|
||||
meta_txt.extend([" ".join(["".join([wtf['text'] for wtf in l['spans']]) for l in t['lines']]).replace(
|
||||
'- ', '') for t in text_areas['blocks'] if 'lines' in t])
|
||||
meta_font.extend([np.mean([np.mean([wtf['size'] for wtf in l['spans']])
|
||||
for l in t['lines']]) for t in text_areas['blocks'] if 'lines' in t])
|
||||
for l in t['lines']]) for t in text_areas['blocks'] if 'lines' in t])
|
||||
if index == 0:
|
||||
page_one_meta = [" ".join(["".join([wtf['text'] for wtf in l['spans']]) for l in t['lines']]).replace(
|
||||
'- ', '') for t in text_areas['blocks'] if 'lines' in t]
|
||||
|
||||
|
||||
############################## <第 2 步，获取正文主字体> ##################################
|
||||
fsize_statiscs = {}
|
||||
for span in meta_span:
|
||||
@ -476,32 +494,33 @@ def read_and_clean_pdf_text(fp):
|
||||
mega_sec = []
|
||||
sec = []
|
||||
for index, line in enumerate(meta_line):
|
||||
if index == 0:
|
||||
if index == 0:
|
||||
sec.append(line[fc])
|
||||
continue
|
||||
if REMOVE_FOOT_NOTE:
|
||||
if meta_line[index][fs] <= give_up_fize_threshold:
|
||||
continue
|
||||
if ffsize_same(meta_line[index][fs], meta_line[index-1][fs]):
|
||||
if ffsize_same(meta_line[index][fs], meta_line[index - 1][fs]):
|
||||
# 尝试识别段落
|
||||
if meta_line[index][fc].endswith('.') and\
|
||||
(meta_line[index-1][fc] != 'NEW_BLOCK') and \
|
||||
(meta_line[index][fb][2] - meta_line[index][fb][0]) < (meta_line[index-1][fb][2] - meta_line[index-1][fb][0]) * 0.7:
|
||||
if meta_line[index][fc].endswith('.') and \
|
||||
(meta_line[index - 1][fc] != 'NEW_BLOCK') and \
|
||||
(meta_line[index][fb][2] - meta_line[index][fb][0]) < (
|
||||
meta_line[index - 1][fb][2] - meta_line[index - 1][fb][0]) * 0.7:
|
||||
sec[-1] += line[fc]
|
||||
sec[-1] += "\n\n"
|
||||
else:
|
||||
sec[-1] += " "
|
||||
sec[-1] += line[fc]
|
||||
else:
|
||||
if (index+1 < len(meta_line)) and \
|
||||
meta_line[index][fs] > main_fsize:
|
||||
if (index + 1 < len(meta_line)) and \
|
||||
meta_line[index][fs] > main_fsize:
|
||||
# 单行 + 字体大
|
||||
mega_sec.append(copy.deepcopy(sec))
|
||||
sec = []
|
||||
sec.append("# " + line[fc])
|
||||
else:
|
||||
# 尝试识别section
|
||||
if meta_line[index-1][fs] > meta_line[index][fs]:
|
||||
if meta_line[index - 1][fs] > meta_line[index][fs]:
|
||||
sec.append("\n" + line[fc])
|
||||
else:
|
||||
sec.append(line[fc])
|
||||
@ -520,13 +539,15 @@ def read_and_clean_pdf_text(fp):
|
||||
if len(block_txt) < 100:
|
||||
meta_txt[index] = '\n'
|
||||
return meta_txt
|
||||
|
||||
meta_txt = 把字符太少的块清除为回车(meta_txt)
|
||||
|
||||
def 清理多余的空行(meta_txt):
|
||||
for index in reversed(range(1, len(meta_txt))):
|
||||
if meta_txt[index] == '\n' and meta_txt[index-1] == '\n':
|
||||
if meta_txt[index] == '\n' and meta_txt[index - 1] == '\n':
|
||||
meta_txt.pop(index)
|
||||
return meta_txt
|
||||
|
||||
meta_txt = 清理多余的空行(meta_txt)
|
||||
|
||||
def 合并小写开头的段落块(meta_txt):
|
||||
@ -537,16 +558,18 @@ def read_and_clean_pdf_text(fp):
|
||||
return True
|
||||
else:
|
||||
return False
|
||||
|
||||
for _ in range(100):
|
||||
for index, block_txt in enumerate(meta_txt):
|
||||
if starts_with_lowercase_word(block_txt):
|
||||
if meta_txt[index-1] != '\n':
|
||||
meta_txt[index-1] += ' '
|
||||
if meta_txt[index - 1] != '\n':
|
||||
meta_txt[index - 1] += ' '
|
||||
else:
|
||||
meta_txt[index-1] = ''
|
||||
meta_txt[index-1] += meta_txt[index]
|
||||
meta_txt[index - 1] = ''
|
||||
meta_txt[index - 1] += meta_txt[index]
|
||||
meta_txt[index] = '\n'
|
||||
return meta_txt
|
||||
|
||||
meta_txt = 合并小写开头的段落块(meta_txt)
|
||||
meta_txt = 清理多余的空行(meta_txt)
|
||||
|
||||
@ -566,7 +589,7 @@ def read_and_clean_pdf_text(fp):
|
||||
return meta_txt, page_one_meta
|
||||
|
||||
|
||||
def get_files_from_everything(txt, type): # type='.md'
|
||||
def get_files_from_everything(txt, type): # type='.md'
|
||||
"""
|
||||
这个函数是用来获取指定目录下所有指定类型（如.md）的文件，并且对于网络上的文件，也可以获取它。
|
||||
下面是对每个参数和返回值的说明：
|
||||
@ -588,9 +611,10 @@ def get_files_from_everything(txt, type): # type='.md'
|
||||
from toolbox import get_conf
|
||||
proxies, = get_conf('proxies')
|
||||
r = requests.get(txt, proxies=proxies)
|
||||
with open('./gpt_log/temp'+type, 'wb+') as f: f.write(r.content)
|
||||
with open('./gpt_log/temp' + type, 'wb+') as f:
|
||||
f.write(r.content)
|
||||
project_folder = './gpt_log/'
|
||||
file_manifest = ['./gpt_log/temp'+type]
|
||||
file_manifest = ['./gpt_log/temp' + type]
|
||||
elif txt.endswith(type):
|
||||
# 直接给定文件
|
||||
file_manifest = [txt]
|
||||
@ -598,7 +622,7 @@ def get_files_from_everything(txt, type): # type='.md'
|
||||
elif os.path.exists(txt):
|
||||
# 本地路径，递归搜索
|
||||
project_folder = txt
|
||||
file_manifest = [f for f in glob.glob(f'{project_folder}/**/*'+type, recursive=True)]
|
||||
file_manifest = [f for f in glob.glob(f'{project_folder}/**/*' + type, recursive=True)]
|
||||
if len(file_manifest) == 0:
|
||||
success = False
|
||||
else:
|
||||
@ -609,16 +633,14 @@ def get_files_from_everything(txt, type): # type='.md'
|
||||
return success, file_manifest, project_folder
|
||||
|
||||
|
||||
|
||||
|
||||
def Singleton(cls):
|
||||
_instance = {}
|
||||
|
||||
|
||||
def _singleton(*args, **kargs):
|
||||
if cls not in _instance:
|
||||
_instance[cls] = cls(*args, **kargs)
|
||||
return _instance[cls]
|
||||
|
||||
|
||||
return _singleton
|
||||
|
||||
|
||||
@ -637,31 +659,30 @@ class knowledge_archive_interface():
|
||||
from toolbox import ProxyNetworkActivate
|
||||
print('Checking Text2vec ...')
|
||||
from langchain.embeddings.huggingface import HuggingFaceEmbeddings
|
||||
with ProxyNetworkActivate(): # 临时地激活代理网络
|
||||
with ProxyNetworkActivate(): # 临时地激活代理网络
|
||||
self.text2vec_large_chinese = HuggingFaceEmbeddings(model_name="GanymedeNil/text2vec-large-chinese")
|
||||
|
||||
return self.text2vec_large_chinese
|
||||
|
||||
|
||||
def feed_archive(self, file_manifest, id="default"):
|
||||
self.threadLock.acquire()
|
||||
# import uuid
|
||||
self.current_id = id
|
||||
from zh_langchain import construct_vector_store
|
||||
self.qa_handle, self.kai_path = construct_vector_store(
|
||||
vs_id=self.current_id,
|
||||
files=file_manifest,
|
||||
self.qa_handle, self.kai_path = construct_vector_store(
|
||||
vs_id=self.current_id,
|
||||
files=file_manifest,
|
||||
sentence_size=100,
|
||||
history=[],
|
||||
one_conent="",
|
||||
one_content_segmentation="",
|
||||
text2vec = self.get_chinese_text2vec(),
|
||||
text2vec=self.get_chinese_text2vec(),
|
||||
)
|
||||
self.threadLock.release()
|
||||
|
||||
def get_current_archive_id(self):
|
||||
return self.current_id
|
||||
|
||||
|
||||
def get_loaded_file(self):
|
||||
return self.qa_handle.get_loaded_file()
|
||||
|
||||
@ -670,30 +691,31 @@ class knowledge_archive_interface():
|
||||
if not self.current_id == id:
|
||||
self.current_id = id
|
||||
from zh_langchain import construct_vector_store
|
||||
self.qa_handle, self.kai_path = construct_vector_store(
|
||||
vs_id=self.current_id,
|
||||
files=[],
|
||||
self.qa_handle, self.kai_path = construct_vector_store(
|
||||
vs_id=self.current_id,
|
||||
files=[],
|
||||
sentence_size=100,
|
||||
history=[],
|
||||
one_conent="",
|
||||
one_content_segmentation="",
|
||||
text2vec = self.get_chinese_text2vec(),
|
||||
text2vec=self.get_chinese_text2vec(),
|
||||
)
|
||||
VECTOR_SEARCH_SCORE_THRESHOLD = 0
|
||||
VECTOR_SEARCH_TOP_K = 4
|
||||
CHUNK_SIZE = 512
|
||||
resp, prompt = self.qa_handle.get_knowledge_based_conent_test(
|
||||
query = txt,
|
||||
vs_path = self.kai_path,
|
||||
query=txt,
|
||||
vs_path=self.kai_path,
|
||||
score_threshold=VECTOR_SEARCH_SCORE_THRESHOLD,
|
||||
vector_search_top_k=VECTOR_SEARCH_TOP_K,
|
||||
vector_search_top_k=VECTOR_SEARCH_TOP_K,
|
||||
chunk_conent=True,
|
||||
chunk_size=CHUNK_SIZE,
|
||||
text2vec = self.get_chinese_text2vec(),
|
||||
text2vec=self.get_chinese_text2vec(),
|
||||
)
|
||||
self.threadLock.release()
|
||||
return resp, prompt
|
||||
|
||||
|
||||
def try_install_deps(deps):
|
||||
for dep in deps:
|
||||
import subprocess, sys
|
||||
|
||||
@ -1,67 +1,78 @@
|
||||
from toolbox import update_ui
|
||||
from toolbox import CatchException, report_execption, write_results_to_file
|
||||
from toolbox import update_ui
|
||||
from .crazy_utils import request_gpt_model_in_new_thread_with_ui_alive
|
||||
|
||||
fast_debug = False
|
||||
|
||||
|
||||
def 解析Paper(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt):
|
||||
import time, glob, os
|
||||
import time
|
||||
import os
|
||||
print('begin analysis on:', file_manifest)
|
||||
for index, fp in enumerate(file_manifest):
|
||||
with open(fp, 'r', encoding='utf-8', errors='replace') as f:
|
||||
file_content = f.read()
|
||||
|
||||
prefix = "接下来请你逐文件分析下面的论文文件，概括其内容" if index==0 else ""
|
||||
prefix = "接下来请你逐文件分析下面的论文文件，概括其内容" if index == 0 else ""
|
||||
i_say = prefix + f'请对下面的文章片段用中文做一个概述，文件名是{os.path.relpath(fp, project_folder)}，文章内容是 ```{file_content}```'
|
||||
i_say_show_user = prefix + f'[{index}/{len(file_manifest)}] 请对下面的文章片段做一个概述: {os.path.abspath(fp)}'
|
||||
chatbot.append((i_say_show_user, "[Local Message] waiting gpt response."))
|
||||
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
||||
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
||||
|
||||
if not fast_debug:
|
||||
if not fast_debug:
|
||||
msg = '正常'
|
||||
# ** gpt request **
|
||||
gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(i_say, i_say_show_user, llm_kwargs, chatbot, history=[], sys_prompt=system_prompt) # 带超时倒计时
|
||||
gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(i_say, i_say_show_user, llm_kwargs,
|
||||
chatbot, history=[],
|
||||
sys_prompt=system_prompt) # 带超时倒计时
|
||||
|
||||
chatbot[-1] = (i_say_show_user, gpt_say)
|
||||
history.append(i_say_show_user); history.append(gpt_say)
|
||||
yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
|
||||
history.append(i_say_show_user);
|
||||
history.append(gpt_say)
|
||||
yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
|
||||
if not fast_debug: time.sleep(2)
|
||||
|
||||
all_file = ', '.join([os.path.relpath(fp, project_folder) for index, fp in enumerate(file_manifest)])
|
||||
i_say = f'根据以上你自己的分析，对全文进行概括，用学术性语言写一段中文摘要，然后再写一段英文摘要（包括{all_file}）。'
|
||||
chatbot.append((i_say, "[Local Message] waiting gpt response."))
|
||||
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
||||
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
||||
|
||||
if not fast_debug:
|
||||
if not fast_debug:
|
||||
msg = '正常'
|
||||
# ** gpt request **
|
||||
gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(i_say, i_say, llm_kwargs, chatbot, history=history, sys_prompt=system_prompt) # 带超时倒计时
|
||||
gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(i_say, i_say, llm_kwargs, chatbot,
|
||||
history=history,
|
||||
sys_prompt=system_prompt) # 带超时倒计时
|
||||
|
||||
chatbot[-1] = (i_say, gpt_say)
|
||||
history.append(i_say); history.append(gpt_say)
|
||||
yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
|
||||
history.append(i_say)
|
||||
history.append(gpt_say)
|
||||
yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
|
||||
res = write_results_to_file(history)
|
||||
chatbot.append(("完成了吗？", res))
|
||||
yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
|
||||
|
||||
yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
|
||||
|
||||
|
||||
@CatchException
|
||||
def 读文章写摘要(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
|
||||
history = [] # 清空历史，以免输入溢出
|
||||
import glob, os
|
||||
def 读文章写摘要(txt, llm_kwargs, plugin_kwargs, chatbot, system_prompt, web_port, history=None):
|
||||
# history = [] # 清空历史，以免输入溢出
|
||||
if history is None:
|
||||
history = [] # 清空历史，以免输入溢出
|
||||
import glob
|
||||
import os
|
||||
if os.path.exists(txt):
|
||||
project_folder = txt
|
||||
else:
|
||||
if txt == "": txt = '空空如也的输入栏'
|
||||
report_execption(chatbot, history, a = f"解析项目: {txt}", b = f"找不到本地项目或无权访问: {txt}")
|
||||
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
||||
if txt == "":
|
||||
txt = '空空如也的输入栏'
|
||||
report_execption(chatbot, history, a=f"解析项目: {txt}", b=f"找不到本地项目或无权访问: {txt}")
|
||||
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
||||
return
|
||||
file_manifest = [f for f in glob.glob(f'{project_folder}/**/*.tex', recursive=True)] # + \
|
||||
# [f for f in glob.glob(f'{project_folder}/**/*.cpp', recursive=True)] + \
|
||||
# [f for f in glob.glob(f'{project_folder}/**/*.c', recursive=True)]
|
||||
file_manifest = [f for f in glob.glob(f'{project_folder}/**/*.tex', recursive=True)] # + \
|
||||
# [f for f in glob.glob(f'{project_folder}/**/*.cpp', recursive=True)] + \
|
||||
# [f for f in glob.glob(f'{project_folder}/**/*.c', recursive=True)]
|
||||
if len(file_manifest) == 0:
|
||||
report_execption(chatbot, history, a = f"解析项目: {txt}", b = f"找不到任何.tex文件: {txt}")
|
||||
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
||||
report_execption(chatbot, history, a=f"解析项目: {txt}", b=f"找不到任何.tex文件: {txt}")
|
||||
yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
|
||||
return
|
||||
yield from 解析Paper(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt)
|
||||
|
||||
@ -28,6 +28,7 @@ proxies, API_KEY, TIMEOUT_SECONDS, MAX_RETRY = \
|
||||
timeout_bot_msg = '[Local Message] Request timeout. Network error. Please check proxy settings in config.py.' + \
|
||||
'网络错误，检查代理服务器是否可用，以及代理设置的格式是否正确，格式须是[协议]://[地址]:[端口]，缺一不可。'
|
||||
|
||||
|
||||
def get_full_error(chunk, stream_response):
|
||||
"""
|
||||
获取完整的从Openai返回的报错
|
||||
@ -40,7 +41,9 @@ def get_full_error(chunk, stream_response):
|
||||
return chunk
|
||||
|
||||
|
||||
def predict_no_ui_long_connection(inputs, llm_kwargs, history=[], sys_prompt="", observe_window=None, console_slience=False):
|
||||
def predict_no_ui_long_connection(
|
||||
inputs, llm_kwargs, history=None, sys_prompt="", observe_window=None, console_slience=False
|
||||
):
|
||||
"""
|
||||
发送至chatGPT，等待回复，一次性完成，不显示中间过程。但内部用stream的方法避免中途网线被掐。
|
||||
inputs：
|
||||
@ -54,45 +57,59 @@ def predict_no_ui_long_connection(inputs, llm_kwargs, history=[], sys_prompt="",
|
||||
observe_window = None：
|
||||
用于负责跨越线程传递已经输出的部分，大部分时候仅仅为了fancy的视觉效果，留空即可。observe_window[0]：观测窗。observe_window[1]：看门狗
|
||||
"""
|
||||
watch_dog_patience = 5 # 看门狗的耐心, 设置5秒即可
|
||||
if history is None:
|
||||
history = []
|
||||
watch_dog_patience = 5 # 看门狗的耐心, 设置5秒即可
|
||||
headers, payload = generate_payload(inputs, llm_kwargs, history, system_prompt=sys_prompt, stream=True)
|
||||
retry = 0
|
||||
from bridge_all import model_info
|
||||
while True:
|
||||
try:
|
||||
# make a POST request to the API endpoint, stream=False
|
||||
from .bridge_all import model_info
|
||||
endpoint = model_info[llm_kwargs['llm_model']]['endpoint']
|
||||
response = requests.post(endpoint, headers=headers, proxies=proxies,
|
||||
json=payload, stream=True, timeout=TIMEOUT_SECONDS); break
|
||||
except requests.exceptions.ReadTimeout as e:
|
||||
json=payload, stream=True, timeout=TIMEOUT_SECONDS)
|
||||
stream_response = response.iter_lines()
|
||||
break
|
||||
except (requests.exceptions.ReadTimeout, requests.exceptions.ConnectionError):
|
||||
retry += 1
|
||||
traceback.print_exc()
|
||||
if retry > MAX_RETRY: raise TimeoutError
|
||||
if MAX_RETRY!=0: print(f'请求超时，正在重试 ({retry}/{MAX_RETRY}) ……')
|
||||
if retry > MAX_RETRY:
|
||||
raise TimeoutError
|
||||
if MAX_RETRY != 0:
|
||||
print(f'请求超时，正在重试 ({retry}/{MAX_RETRY}) ……')
|
||||
except Exception as e:
|
||||
print(f"出现异常：{e}")
|
||||
raise e
|
||||
|
||||
stream_response = response.iter_lines()
|
||||
result = ''
|
||||
while True:
|
||||
try: chunk = next(stream_response).decode()
|
||||
try:
|
||||
chunk = next(stream_response).decode()
|
||||
except StopIteration:
|
||||
break
|
||||
except requests.exceptions.ConnectionError:
|
||||
chunk = next(stream_response).decode() # 失败了，重试一次？再失败就没办法了。
|
||||
if len(chunk)==0: continue
|
||||
# except requests.exceptions.ConnectionError:
|
||||
# chunk = next(stream_response).decode() # 失败了，重试一次？再失败就没办法了。
|
||||
if len(chunk) == 0:
|
||||
continue
|
||||
if not chunk.startswith('data:'):
|
||||
error_msg = get_full_error(chunk.encode('utf8'), stream_response).decode()
|
||||
if "reduce the length" in error_msg:
|
||||
raise ConnectionAbortedError("OpenAI拒绝了请求:" + error_msg)
|
||||
else:
|
||||
raise RuntimeError("OpenAI拒绝了请求：" + error_msg)
|
||||
if ('data: [DONE]' in chunk): break # api2d 正常完成
|
||||
if 'data: [DONE]' in chunk:
|
||||
break # api2d 正常完成
|
||||
json_data = json.loads(chunk.lstrip('data:'))['choices'][0]
|
||||
delta = json_data["delta"]
|
||||
if len(delta) == 0: break
|
||||
if "role" in delta: continue
|
||||
if len(delta) == 0:
|
||||
break
|
||||
if "role" in delta:
|
||||
continue
|
||||
if "content" in delta:
|
||||
result += delta["content"]
|
||||
if not console_slience: print(delta["content"], end='')
|
||||
if not console_slience:
|
||||
print(delta["content"], end='')
|
||||
if observe_window is not None:
|
||||
# 观测窗，把已经获取的数据显示出去
|
||||
if len(observe_window) >= 1: observe_window[0] += delta["content"]
|
||||
@ -100,7 +117,8 @@ def predict_no_ui_long_connection(inputs, llm_kwargs, history=[], sys_prompt="",
|
||||
if len(observe_window) >= 2:
|
||||
if (time.time()-observe_window[1]) > watch_dog_patience:
|
||||
raise RuntimeError("用户取消了程序。")
|
||||
else: raise RuntimeError("意外Json结构："+delta)
|
||||
else:
|
||||
raise RuntimeError("意外Json结构："+delta)
|
||||
if json_data['finish_reason'] == 'length':
|
||||
raise ConnectionAbortedError("正常结束，但显示Token不足，导致输出不完整，请削减单次输入的文本量。")
|
||||
return result
|
||||
@ -228,6 +246,7 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
|
||||
yield from update_ui(chatbot=chatbot, history=history, msg="Json异常" + error_msg) # 刷新界面
|
||||
return
|
||||
|
||||
|
||||
def generate_payload(inputs, llm_kwargs, history, system_prompt, stream):
|
||||
"""
|
||||
整合所有信息，选择LLM模型，生成http请求，为发送请求做准备
|
||||
@ -247,23 +266,19 @@ def generate_payload(inputs, llm_kwargs, history, system_prompt, stream):
|
||||
messages = [{"role": "system", "content": system_prompt}]
|
||||
if conversation_cnt:
|
||||
for index in range(0, 2*conversation_cnt, 2):
|
||||
what_i_have_asked = {}
|
||||
what_i_have_asked["role"] = "user"
|
||||
what_i_have_asked["content"] = history[index]
|
||||
what_gpt_answer = {}
|
||||
what_gpt_answer["role"] = "assistant"
|
||||
what_gpt_answer["content"] = history[index+1]
|
||||
what_i_have_asked = {"role": "user", "content": history[index]}
|
||||
what_gpt_answer = {"role": "assistant", "content": history[index + 1]}
|
||||
if what_i_have_asked["content"] != "":
|
||||
if what_gpt_answer["content"] == "": continue
|
||||
if what_gpt_answer["content"] == timeout_bot_msg: continue
|
||||
if what_gpt_answer["content"] == "":
|
||||
continue
|
||||
if what_gpt_answer["content"] == timeout_bot_msg:
|
||||
continue
|
||||
messages.append(what_i_have_asked)
|
||||
messages.append(what_gpt_answer)
|
||||
else:
|
||||
messages[-1]['content'] = what_gpt_answer['content']
|
||||
|
||||
what_i_ask_now = {}
|
||||
what_i_ask_now["role"] = "user"
|
||||
what_i_ask_now["content"] = inputs
|
||||
what_i_ask_now = {"role": "user", "content": inputs}
|
||||
messages.append(what_i_ask_now)
|
||||
|
||||
payload = {
|
||||
@ -278,8 +293,8 @@ def generate_payload(inputs, llm_kwargs, history, system_prompt, stream):
|
||||
}
|
||||
try:
|
||||
print(f" {llm_kwargs['llm_model']} : {conversation_cnt} : {inputs[:100]} ..........")
|
||||
except:
|
||||
print('输入中可能存在乱码。')
|
||||
return headers,payload
|
||||
except Exception as e:
|
||||
print(f'输入中可能存在乱码。抛出异常: {e}')
|
||||
return headers, payload
|
||||
|
||||
|
||||
|
||||
@ -1,4 +1,4 @@
|
||||
./docs/gradio-3.32.2-py3-none-any.whl
|
||||
gradio>=3.33.1
|
||||
tiktoken>=0.3.3
|
||||
requests[socks]
|
||||
transformers
|
||||
@ -15,4 +15,6 @@ pymupdf
|
||||
openai
|
||||
numpy
|
||||
arxiv
|
||||
rich
|
||||
rich
|
||||
langchain
|
||||
zh_langchain
|
||||
37
toolbox.py
37
toolbox.py
@ -21,6 +21,7 @@ pj = os.path.join
|
||||
========================================================================
|
||||
"""
|
||||
|
||||
|
||||
class ChatBotWithCookies(list):
|
||||
def __init__(self, cookie):
|
||||
self._cookies = cookie
|
||||
@ -71,11 +72,13 @@ def update_ui(chatbot, history, msg='正常', **kwargs): # 刷新界面
|
||||
assert isinstance(chatbot, ChatBotWithCookies), "在传递chatbot的过程中不要将其丢弃。必要时，可用clear将其清空，然后用for+append循环重新赋值。"
|
||||
yield chatbot.get_cookies(), chatbot, history, msg
|
||||
|
||||
|
||||
def update_ui_lastest_msg(lastmsg, chatbot, history, delay=1): # 刷新界面
|
||||
"""
|
||||
刷新用户界面
|
||||
"""
|
||||
if len(chatbot) == 0: chatbot.append(["update_ui_last_msg", lastmsg])
|
||||
if len(chatbot) == 0:
|
||||
chatbot.append(["update_ui_last_msg", lastmsg])
|
||||
chatbot[-1] = list(chatbot[-1])
|
||||
chatbot[-1][-1] = lastmsg
|
||||
yield from update_ui(chatbot=chatbot, history=history)
|
||||
@ -83,24 +86,25 @@ def update_ui_lastest_msg(lastmsg, chatbot, history, delay=1): # 刷新界面
|
||||
|
||||
|
||||
def trimmed_format_exc():
|
||||
import os, traceback
|
||||
str = traceback.format_exc()
|
||||
import os
|
||||
import traceback
|
||||
_str = traceback.format_exc()
|
||||
current_path = os.getcwd()
|
||||
replace_path = "."
|
||||
return str.replace(current_path, replace_path)
|
||||
return _str.replace(current_path, replace_path)
|
||||
|
||||
|
||||
def CatchException(f):
|
||||
"""
|
||||
装饰器函数，捕捉函数f中的异常并封装到一个生成器中返回，并显示到聊天当中。
|
||||
"""
|
||||
|
||||
@wraps(f)
|
||||
def decorated(txt, top_p, temperature, chatbot, history, systemPromptTxt, WEB_PORT=-1):
|
||||
try:
|
||||
yield from f(txt, top_p, temperature, chatbot, history, systemPromptTxt, WEB_PORT)
|
||||
except Exception as e:
|
||||
from check_proxy import check_proxy
|
||||
from toolbox import get_conf
|
||||
# from toolbox import get_conf # 不需要导入本文件内容
|
||||
proxies, = get_conf('proxies')
|
||||
tb_str = '```\n' + trimmed_format_exc() + '```'
|
||||
if len(chatbot) == 0:
|
||||
@ -108,7 +112,7 @@ def CatchException(f):
|
||||
chatbot.append(["插件调度异常", "异常原因"])
|
||||
chatbot[-1] = (chatbot[-1][0],
|
||||
f"[Local Message] 实验性函数调用出错: \n\n{tb_str} \n\n当前代理可用性: \n\n{check_proxy(proxies)}")
|
||||
yield from update_ui(chatbot=chatbot, history=history, msg=f'异常 {e}') # 刷新界面
|
||||
yield from update_ui(chatbot=chatbot, history=history, msg=f'异常 {e}') # 刷新界面
|
||||
return decorated
|
||||
|
||||
|
||||
@ -148,6 +152,7 @@ def HotReload(f):
|
||||
========================================================================
|
||||
"""
|
||||
|
||||
|
||||
def get_reduce_token_percent(text):
|
||||
"""
|
||||
* 此函数未来将被弃用
|
||||
@ -207,8 +212,6 @@ def regular_txt_to_markdown(text):
|
||||
return text
|
||||
|
||||
|
||||
|
||||
|
||||
def report_execption(chatbot, history, a, b):
|
||||
"""
|
||||
向chatbot中添加错误信息
|
||||
@ -238,6 +241,7 @@ def text_divide_paragraph(text):
|
||||
text = "</br>".join(lines)
|
||||
return pre + text + suf
|
||||
|
||||
|
||||
@lru_cache(maxsize=128) # 使用 lru缓存 加快转换速度
|
||||
def markdown_convertion(txt):
|
||||
"""
|
||||
@ -440,6 +444,7 @@ def find_recent_files(directory):
|
||||
|
||||
return recent_files
|
||||
|
||||
|
||||
def promote_file_to_downloadzone(file, rename_file=None, chatbot=None):
|
||||
# 将文件复制一份到下载区
|
||||
import shutil
|
||||
@ -452,6 +457,7 @@ def promote_file_to_downloadzone(file, rename_file=None, chatbot=None):
|
||||
else: current = []
|
||||
chatbot._cookies.update({'file_to_promote': [new_path] + current})
|
||||
|
||||
|
||||
def on_file_uploaded(files, chatbot, txt, txt2, checkboxes):
|
||||
"""
|
||||
当文件被上传时的回调函数
|
||||
@ -505,17 +511,20 @@ def on_report_generated(cookies, files, chatbot):
|
||||
chatbot.append(['报告如何远程获取？', f'报告已经添加到右侧“文件上传区”（可能处于折叠状态），请查收。{file_links}'])
|
||||
return cookies, report_files, chatbot
|
||||
|
||||
|
||||
def is_openai_api_key(key):
|
||||
API_MATCH_ORIGINAL = re.match(r"sk-[a-zA-Z0-9]{48}$", key)
|
||||
API_MATCH_AZURE = re.match(r"[a-zA-Z0-9]{32}$", key)
|
||||
return bool(API_MATCH_ORIGINAL) or bool(API_MATCH_AZURE)
|
||||
|
||||
|
||||
def is_api2d_key(key):
|
||||
if key.startswith('fk') and len(key) == 41:
|
||||
return True
|
||||
else:
|
||||
return False
|
||||
|
||||
|
||||
def is_any_api_key(key):
|
||||
if ',' in key:
|
||||
keys = key.split(',')
|
||||
@ -525,6 +534,7 @@ def is_any_api_key(key):
|
||||
else:
|
||||
return is_openai_api_key(key) or is_api2d_key(key)
|
||||
|
||||
|
||||
def what_keys(keys):
|
||||
avail_key_list = {'OpenAI Key':0, "API2D Key":0}
|
||||
key_list = keys.split(',')
|
||||
@ -539,6 +549,7 @@ def what_keys(keys):
|
||||
|
||||
return f"检测到： OpenAI Key {avail_key_list['OpenAI Key']} 个，API2D Key {avail_key_list['API2D Key']} 个"
|
||||
|
||||
|
||||
def select_api_key(keys, llm_model):
|
||||
import random
|
||||
avail_key_list = []
|
||||
@ -558,6 +569,7 @@ def select_api_key(keys, llm_model):
|
||||
api_key = random.choice(avail_key_list) # 随机负载均衡
|
||||
return api_key
|
||||
|
||||
|
||||
def read_env_variable(arg, default_value):
|
||||
"""
|
||||
环境变量可以是 `GPT_ACADEMIC_CONFIG`(优先)，也可以直接是`CONFIG`
|
||||
@ -612,6 +624,7 @@ def read_env_variable(arg, default_value):
|
||||
print亮绿(f"[ENV_VAR] 成功读取环境变量{arg}")
|
||||
return r
|
||||
|
||||
|
||||
@lru_cache(maxsize=128)
|
||||
def read_single_conf_with_lru_cache(arg):
|
||||
from colorful import print亮红, print亮绿, print亮蓝
|
||||
@ -676,6 +689,7 @@ class DummyWith():
|
||||
def __exit__(self, exc_type, exc_value, traceback):
|
||||
return
|
||||
|
||||
|
||||
def run_gradio_in_subpath(demo, auth, port, custom_path):
|
||||
"""
|
||||
把gradio的运行地址更改到指定的二次路径上
|
||||
@ -770,6 +784,7 @@ def clip_history(inputs, history, tokenizer, max_token_limit):
|
||||
========================================================================
|
||||
"""
|
||||
|
||||
|
||||
def zip_folder(source_folder, dest_folder, zip_name):
|
||||
import zipfile
|
||||
import os
|
||||
@ -801,6 +816,7 @@ def zip_folder(source_folder, dest_folder, zip_name):
|
||||
|
||||
print(f"Zip file created at {zip_file}")
|
||||
|
||||
|
||||
def zip_result(folder):
|
||||
import time
|
||||
t = time.strftime("%Y-%m-%d-%H-%M-%S", time.localtime())
|
||||
@ -811,6 +827,7 @@ def gen_time_str():
|
||||
import time
|
||||
return time.strftime("%Y-%m-%d-%H-%M-%S", time.localtime())
|
||||
|
||||
|
||||
class ProxyNetworkActivate():
|
||||
"""
|
||||
这段代码定义了一个名为TempProxy的空上下文管理器, 用于给一小段代码上代理
|
||||
@ -830,12 +847,14 @@ class ProxyNetworkActivate():
|
||||
if 'HTTPS_PROXY' in os.environ: os.environ.pop('HTTPS_PROXY')
|
||||
return
|
||||
|
||||
|
||||
def objdump(obj, file='objdump.tmp'):
|
||||
import pickle
|
||||
with open(file, 'wb+') as f:
|
||||
pickle.dump(obj, f)
|
||||
return
|
||||
|
||||
|
||||
def objload(file='objdump.tmp'):
|
||||
import pickle, os
|
||||
if not os.path.exists(file):
|
||||
|
||||
Reference in New Issue
Block a user