feat: BingImageCreator output four images

This commit is contained in:
hibobmaster 2023-05-20 09:46:16 +08:00
parent 451c37d321
commit 6ff585d402
No known key found for this signature in database
GPG Key ID: 316B77D7914D713C
6 changed files with 106 additions and 52 deletions

View File

@ -11,5 +11,6 @@ BARD_TOKEN="xxxxxxxxxxxxxxxxxxxx", # Optional, for !bard command
JAILBREAKENABLED="true" # Optional JAILBREAKENABLED="true" # Optional
BING_AUTH_COOKIE="xxxxxxxxxxxxxxxxxxx" # _U cookie, Optional, for Bing Image Creator BING_AUTH_COOKIE="xxxxxxxxxxxxxxxxxxx" # _U cookie, Optional, for Bing Image Creator
MARKDOWN_FORMATTED="true" # Optional MARKDOWN_FORMATTED="true" # Optional
OUTPUT_FOUR_IMAGES="true" # Optional
IMPORT_KEYS_PATH="element-keys.txt" # Optional IMPORT_KEYS_PATH="element-keys.txt" # Optional
IMPORT_KEYS_PASSWORD="xxxxxxx" # Optional IMPORT_KEYS_PASSWORD="xxxxxxx" # Optional

View File

@ -2,6 +2,7 @@
Code derived from: Code derived from:
https://github.com/acheong08/EdgeGPT/blob/f940cecd24a4818015a8b42a2443dd97c3c2a8f4/src/ImageGen.py https://github.com/acheong08/EdgeGPT/blob/f940cecd24a4818015a8b42a2443dd97c3c2a8f4/src/ImageGen.py
""" """
from log import getlogger from log import getlogger
from uuid import uuid4 from uuid import uuid4
import os import os
@ -20,16 +21,17 @@ FORWARDED_IP = (
f"13.{random.randint(104, 107)}.{random.randint(0, 255)}.{random.randint(0, 255)}" f"13.{random.randint(104, 107)}.{random.randint(0, 255)}.{random.randint(0, 255)}"
) )
HEADERS = { HEADERS = {
"accept": "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.7", "accept": "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.7", # noqa: E501
"accept-language": "en-US,en;q=0.9", "accept-language": "en-US,en;q=0.9",
"cache-control": "max-age=0", "cache-control": "max-age=0",
"content-type": "application/x-www-form-urlencoded", "content-type": "application/x-www-form-urlencoded",
"referrer": "https://www.bing.com/images/create/", "referrer": "https://www.bing.com/images/create/",
"origin": "https://www.bing.com", "origin": "https://www.bing.com",
"user-agent": "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/110.0.0.0 Safari/537.36 Edg/110.0.1587.63", "user-agent": "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/110.0.0.0 Safari/537.36 Edg/110.0.1587.63", # noqa: E501
"x-forwarded-for": FORWARDED_IP, "x-forwarded-for": FORWARDED_IP,
} }
class ImageGenAsync: class ImageGenAsync:
""" """
Image generation by Microsoft Bing Image generation by Microsoft Bing
@ -53,7 +55,7 @@ class ImageGenAsync:
def __del__(self): def __del__(self):
try: try:
loop = asyncio.get_running_loop() loop = asyncio.get_running_loop()
except RuntimeError as e: except RuntimeError:
loop = asyncio.new_event_loop() loop = asyncio.new_event_loop()
asyncio.set_event_loop(loop) asyncio.set_event_loop(loop)
loop.run_until_complete(self._close()) loop.run_until_complete(self._close())
@ -76,7 +78,7 @@ class ImageGenAsync:
content = await response.text() content = await response.text()
if "this prompt has been blocked" in content.lower(): if "this prompt has been blocked" in content.lower():
raise Exception( raise Exception(
"Your prompt has been blocked by Bing. Try to change any bad words and try again.", "Your prompt has been blocked by Bing. Try to change any bad words and try again.", # noqa: E501
) )
if response.status != 302: if response.status != 302:
# if rt4 fails, try rt3 # if rt4 fails, try rt3
@ -97,7 +99,7 @@ class ImageGenAsync:
request_id = redirect_url.split("id=")[-1] request_id = redirect_url.split("id=")[-1]
await self.session.get(f"{BING_URL}{redirect_url}") await self.session.get(f"{BING_URL}{redirect_url}")
# https://www.bing.com/images/create/async/results/{ID}?q={PROMPT} # https://www.bing.com/images/create/async/results/{ID}?q={PROMPT}
polling_url = f"{BING_URL}/images/create/async/results/{request_id}?q={url_encoded_prompt}" polling_url = f"{BING_URL}/images/create/async/results/{request_id}?q={url_encoded_prompt}" # noqa: E501
# Poll for results # Poll for results
if not self.quiet: if not self.quiet:
print("Waiting for results...") print("Waiting for results...")
@ -134,31 +136,40 @@ class ImageGenAsync:
raise Exception("No images") raise Exception("No images")
return normal_image_links return normal_image_links
async def save_images(self, links: list, output_dir: str) -> str: async def save_images(self, links: list, output_dir: str,
output_four_images: bool) -> list:
""" """
Saves images to output directory Saves images to output directory
""" """
if not self.quiet:
print("\nDownloading images...")
with contextlib.suppress(FileExistsError): with contextlib.suppress(FileExistsError):
os.mkdir(output_dir) os.mkdir(output_dir)
# image name image_path_list = []
image_name = str(uuid4())
# since matrix only support one media attachment per message, we just need one link
if links:
link = links.pop()
if output_four_images:
for link in links:
image_name = str(uuid4())
image_path = os.path.join(output_dir, f"{image_name}.jpeg") image_path = os.path.join(output_dir, f"{image_name}.jpeg")
try: try:
async with self.session.get(link, raise_for_status=True) as response: async with self.session.get(link, raise_for_status=True) as response:
# save response to file
with open(image_path, "wb") as output_file: with open(image_path, "wb") as output_file:
async for chunk in response.content.iter_chunked(8192): async for chunk in response.content.iter_chunked(8192):
output_file.write(chunk) output_file.write(chunk)
return f"{output_dir}/{image_name}.jpeg" image_path_list.append(image_path)
except aiohttp.client_exceptions.InvalidURL as url_exception: except aiohttp.client_exceptions.InvalidURL as url_exception:
raise Exception( raise Exception("Inappropriate contents found in the generated images. Please try again or try another prompt.") from url_exception # noqa: E501
"Inappropriate contents found in the generated images. Please try again or try another prompt.", else:
) from url_exception image_name = str(uuid4())
if links:
link = links.pop()
try:
async with self.session.get(link, raise_for_status=True) as response:
image_path = os.path.join(output_dir, f"{image_name}.jpeg")
with open(image_path, "wb") as output_file:
async for chunk in response.content.iter_chunked(8192):
output_file.write(chunk)
image_path_list.append(image_path)
except aiohttp.client_exceptions.InvalidURL as url_exception:
raise Exception("Inappropriate contents found in the generated images. Please try again or try another prompt.") from url_exception # noqa: E501
return image_path_list

View File

@ -26,6 +26,7 @@ sudo docker compose up -d
<hr> <hr>
Normal Method:<br> Normal Method:<br>
system dependece: `libolm-dev`
1. Clone the repository and create virtual environment: 1. Clone the repository and create virtual environment:
@ -74,6 +75,7 @@ python main.py
## Usage ## Usage
To interact with the bot, simply send a message to the bot in the Matrix room with one of the two prompts:<br> To interact with the bot, simply send a message to the bot in the Matrix room with one of the two prompts:<br>
- `!help` help message
- `!gpt` To generate a one time response: - `!gpt` To generate a one time response:
@ -115,6 +117,8 @@ https://github.com/hibobmaster/matrix_chatgpt_bot/wiki/ <br>
1. [matrix-nio](https://github.com/poljar/matrix-nio) 1. [matrix-nio](https://github.com/poljar/matrix-nio)
2. [acheong08](https://github.com/acheong08) 2. [acheong08](https://github.com/acheong08)
3. [node-chatgpt-api](https://github.com/waylaidwanderer/node-chatgpt-api) 3. [node-chatgpt-api](https://github.com/waylaidwanderer/node-chatgpt-api)
4. [8go](https://github.com/8go/)
<a href="https://jb.gg/OpenSourceSupport" target="_blank"> <a href="https://jb.gg/OpenSourceSupport" target="_blank">
<img src="https://resources.jetbrains.com/storage/products/company/brand/logos/jb_beam.png" alt="JetBrains Logo (Main) logo." width="200" height="200"> <img src="https://resources.jetbrains.com/storage/products/company/brand/logos/jb_beam.png" alt="JetBrains Logo (Main) logo." width="200" height="200">
</a> </a>

69
bot.py
View File

@ -42,6 +42,7 @@ class Bot:
jailbreakEnabled: Union[bool, None] = True, jailbreakEnabled: Union[bool, None] = True,
bing_auth_cookie: Union[str, None] = '', bing_auth_cookie: Union[str, None] = '',
markdown_formatted: Union[bool, None] = False, markdown_formatted: Union[bool, None] = False,
output_four_images: Union[bool, None] = False,
import_keys_path: Optional[str] = None, import_keys_path: Optional[str] = None,
import_keys_password: Optional[str] = None, import_keys_password: Optional[str] = None,
): ):
@ -88,6 +89,11 @@ class Bot:
else: else:
self.markdown_formatted = markdown_formatted self.markdown_formatted = markdown_formatted
if output_four_images is None:
self.output_four_images = False
else:
self.output_four_images = output_four_images
# initialize AsyncClient object # initialize AsyncClient object
self.store_path = os.getcwd() self.store_path = os.getcwd()
self.config = AsyncClientConfig(store=SqliteStore, self.config = AsyncClientConfig(store=SqliteStore,
@ -95,7 +101,8 @@ class Bot:
store_sync_tokens=True, store_sync_tokens=True,
encryption_enabled=True, encryption_enabled=True,
) )
self.client = AsyncClient(homeserver=self.homeserver, user=self.user_id, device_id=self.device_id, self.client = AsyncClient(homeserver=self.homeserver, user=self.user_id,
device_id=self.device_id,
config=self.config, store_path=self.store_path,) config=self.config, store_path=self.store_path,)
if self.access_token is not None: if self.access_token is not None:
@ -111,7 +118,7 @@ class Bot:
self.client.add_to_device_callback( self.client.add_to_device_callback(
self.to_device_callback, (KeyVerificationEvent, )) self.to_device_callback, (KeyVerificationEvent, ))
# regular expression to match keyword [!gpt {prompt}] [!chat {prompt}] [!bing {prompt}] [!pic {prompt}] [!bard {prompt}] # regular expression to match keyword commands
self.gpt_prog = re.compile(r"^\s*!gpt\s*(.+)$") self.gpt_prog = re.compile(r"^\s*!gpt\s*(.+)$")
self.chat_prog = re.compile(r"^\s*!chat\s*(.+)$") self.chat_prog = re.compile(r"^\s*!chat\s*(.+)$")
self.bing_prog = re.compile(r"^\s*!bing\s*(.+)$") self.bing_prog = re.compile(r"^\s*!bing\s*(.+)$")
@ -149,7 +156,7 @@ class Bot:
def __del__(self): def __del__(self):
try: try:
loop = asyncio.get_running_loop() loop = asyncio.get_running_loop()
except RuntimeError as e: except RuntimeError:
loop = asyncio.new_event_loop() loop = asyncio.new_event_loop()
asyncio.set_event_loop(loop) asyncio.set_event_loop(loop)
loop.run_until_complete(self._close()) loop.run_until_complete(self._close())
@ -202,10 +209,12 @@ class Bot:
) )
except Exception as e: except Exception as e:
logger.error(e, exc_info=True) logger.error(e, exc_info=True)
await send_room_message(self.client, room_id, reply_message=str(e)) await send_room_message(self.client, room_id,
reply_message=str(e))
else: else:
logger.warning("No API_KEY provided") logger.warning("No API_KEY provided")
await send_room_message(self.client, room_id, reply_message="API_KEY not provided") await send_room_message(self.client, room_id,
reply_message="API_KEY not provided")
m = self.gpt_prog.match(content_body) m = self.gpt_prog.match(content_body)
if m: if m:
@ -239,7 +248,8 @@ class Bot:
) )
except Exception as e: except Exception as e:
logger.error(e, exc_info=True) logger.error(e, exc_info=True)
await send_room_message(self.client, room_id, reply_message=str(e)) await send_room_message(self.client, room_id,
reply_message=str(e))
# Image Generation by Microsoft Bing # Image Generation by Microsoft Bing
if self.bing_auth_cookie != '': if self.bing_auth_cookie != '':
@ -250,7 +260,8 @@ class Bot:
asyncio.create_task(self.pic(room_id, prompt)) asyncio.create_task(self.pic(room_id, prompt))
except Exception as e: except Exception as e:
logger.error(e, exc_info=True) logger.error(e, exc_info=True)
await send_room_message(self.client, room_id, reply_message=str(e)) await send_room_message(self.client, room_id,
reply_message=str(e))
# Google's Bard # Google's Bard
if self.bard_token is not None: if self.bard_token is not None:
@ -281,7 +292,8 @@ class Bot:
return return
logger.error( logger.error(
f"Failed to decrypt message: {event.event_id} from {event.sender} in {room.room_id}\n" + f"Failed to decrypt message: {event.event_id} \
from {event.sender} in {room.room_id}\n" +
"Please make sure the bot current session is verified" "Please make sure the bot current session is verified"
) )
@ -501,7 +513,8 @@ class Bot:
logger.info(estr) logger.info(estr)
# !chat command # !chat command
async def chat(self, room_id, reply_to_event_id, prompt, sender_id, raw_user_message): async def chat(self, room_id, reply_to_event_id, prompt, sender_id,
raw_user_message):
await self.client.room_typing(room_id, timeout=120000) await self.client.room_typing(room_id, timeout=120000)
try: try:
text = await self.chatbot.ask_async(prompt) text = await self.chatbot.ask_async(prompt)
@ -511,17 +524,23 @@ class Bot:
try: try:
text = text.strip() text = text.strip()
await send_room_message(self.client, room_id, reply_message=text, await send_room_message(self.client, room_id, reply_message=text,
reply_to_event_id="", sender_id=sender_id, user_message=raw_user_message, markdown_formatted=self.markdown_formatted) reply_to_event_id="", sender_id=sender_id,
user_message=raw_user_message,
markdown_formatted=self.markdown_formatted)
except Exception as e: except Exception as e:
logger.error(f"Error: {e}", exc_info=True) logger.error(f"Error: {e}", exc_info=True)
# !gpt command # !gpt command
async def gpt(self, room_id, reply_to_event_id, prompt, sender_id, raw_user_message) -> None: async def gpt(self, room_id, reply_to_event_id, prompt, sender_id,
raw_user_message) -> None:
try: try:
# sending typing state # sending typing state
await self.client.room_typing(room_id, timeout=240000) await self.client.room_typing(room_id, timeout=240000)
# timeout 240s # timeout 240s
text = await asyncio.wait_for(self.askgpt.oneTimeAsk(prompt, self.chatgpt_api_endpoint, self.headers), timeout=240) text = await asyncio.wait_for(self.askgpt.oneTimeAsk(prompt,
self.chatgpt_api_endpoint,
self.headers),
timeout=240)
except TimeoutError: except TimeoutError:
logger.error("TimeoutException", exc_info=True) logger.error("TimeoutException", exc_info=True)
raise Exception("Timeout error") raise Exception("Timeout error")
@ -531,12 +550,15 @@ class Bot:
try: try:
text = text.strip() text = text.strip()
await send_room_message(self.client, room_id, reply_message=text, await send_room_message(self.client, room_id, reply_message=text,
reply_to_event_id="", sender_id=sender_id, user_message=raw_user_message, markdown_formatted=self.markdown_formatted) reply_to_event_id="", sender_id=sender_id,
user_message=raw_user_message,
markdown_formatted=self.markdown_formatted)
except Exception as e: except Exception as e:
logger.error(f"Error: {e}", exc_info=True) logger.error(f"Error: {e}", exc_info=True)
# !bing command # !bing command
async def bing(self, room_id, reply_to_event_id, prompt, sender_id, raw_user_message) -> None: async def bing(self, room_id, reply_to_event_id, prompt, sender_id,
raw_user_message) -> None:
try: try:
# sending typing state # sending typing state
await self.client.room_typing(room_id, timeout=180000) await self.client.room_typing(room_id, timeout=180000)
@ -551,12 +573,15 @@ class Bot:
try: try:
text = text.strip() text = text.strip()
await send_room_message(self.client, room_id, reply_message=text, await send_room_message(self.client, room_id, reply_message=text,
reply_to_event_id="", sender_id=sender_id, user_message=raw_user_message, markdown_formatted=self.markdown_formatted) reply_to_event_id="", sender_id=sender_id,
user_message=raw_user_message,
markdown_formatted=self.markdown_formatted)
except Exception as e: except Exception as e:
logger.error(e, exc_info=True) logger.error(e, exc_info=True)
# !bard command # !bard command
async def bard(self, room_id, reply_to_event_id, prompt, sender_id, raw_user_message) -> None: async def bard(self, room_id, reply_to_event_id, prompt, sender_id,
raw_user_message) -> None:
try: try:
# sending typing state # sending typing state
await self.client.room_typing(room_id) await self.client.room_typing(room_id)
@ -567,7 +592,9 @@ class Bot:
try: try:
content = str(response['content']).strip() content = str(response['content']).strip()
await send_room_message(self.client, room_id, reply_message=content, await send_room_message(self.client, room_id, reply_message=content,
reply_to_event_id="", sender_id=sender_id, user_message=raw_user_message, markdown_formatted=self.markdown_formatted) reply_to_event_id="", sender_id=sender_id,
user_message=raw_user_message,
markdown_formatted=self.markdown_formatted)
except Exception as e: except Exception as e:
logger.error(e, exc_info=True) logger.error(e, exc_info=True)
@ -580,13 +607,15 @@ class Bot:
try: try:
links = await self.imageGen.get_images(prompt) links = await self.imageGen.get_images(prompt)
image_path = await self.imageGen.save_images(links, "images") image_path_list = await self.imageGen.save_images(links, "images",
self.output_four_images)
except Exception as e: except Exception as e:
logger.error(f"Image Generation error: {e}", exc_info=True) logger.error(f"Image Generation error: {e}", exc_info=True)
raise Exception(e) raise Exception(e)
# send image # send image
try: try:
for image_path in image_path_list:
await send_room_image(self.client, room_id, image_path) await send_room_image(self.client, room_id, image_path)
await self.client.room_typing(room_id, typing_state=False) await self.client.room_typing(room_id, typing_state=False)
except Exception as e: except Exception as e:
@ -605,7 +634,7 @@ class Bot:
"!bing [content], chat with context conversation powered by Bing AI\n" + \ "!bing [content], chat with context conversation powered by Bing AI\n" + \
"!bard [content], chat with Google's Bard\n" + \ "!bard [content], chat with Google's Bard\n" + \
"!pic [prompt], Image generation by Microsoft Bing\n" + \ "!pic [prompt], Image generation by Microsoft Bing\n" + \
"!help, help message" "!help, help message" # noqa: E501
await send_room_message(self.client, room_id, reply_message=help_info) await send_room_message(self.client, room_id, reply_message=help_info)
except Exception as e: except Exception as e:
@ -635,7 +664,7 @@ class Bot:
logger.error(f"import_keys failed with {resp}") logger.error(f"import_keys failed with {resp}")
else: else:
logger.info( logger.info(
f"import_keys success, please remove import_keys configuration!!!") "import_keys success, please remove import_keys configuration!!!")
# sync messages in the room # sync messages in the room
async def sync_forever(self, timeout=30000, full_state=True) -> None: async def sync_forever(self, timeout=30000, full_state=True) -> None:

View File

@ -11,6 +11,7 @@
"bard_token": "xxxxxxx", "bard_token": "xxxxxxx",
"bing_auth_cookie": "xxxxxxxxxxx", "bing_auth_cookie": "xxxxxxxxxxx",
"markdown_formatted": true, "markdown_formatted": true,
"output_four_images": true,
"import_keys_path": "element-keys.txt", "import_keys_path": "element-keys.txt",
"import_keys_password": "xxxxxxxxx" "import_keys_password": "xxxxxxxxx"
} }

16
main.py
View File

@ -25,11 +25,13 @@ async def main():
jailbreakEnabled=config.get('jailbreakEnabled'), jailbreakEnabled=config.get('jailbreakEnabled'),
bing_auth_cookie=config.get('bing_auth_cookie'), bing_auth_cookie=config.get('bing_auth_cookie'),
markdown_formatted=config.get('markdown_formatted'), markdown_formatted=config.get('markdown_formatted'),
output_four_images=config.get('output_four_images'),
import_keys_path=config.get('import_keys_path'), import_keys_path=config.get('import_keys_path'),
import_keys_password=config.get( import_keys_password=config.get(
'import_keys_password'), 'import_keys_password'),
) )
if config.get('import_keys_path') and config.get('import_keys_password') is not None: if config.get('import_keys_path') and \
config.get('import_keys_password') is not None:
need_import_keys = True need_import_keys = True
else: else:
@ -43,15 +45,21 @@ async def main():
access_token=os.environ.get("ACCESS_TOKEN"), access_token=os.environ.get("ACCESS_TOKEN"),
bard_token=os.environ.get("BARD_TOKEN"), bard_token=os.environ.get("BARD_TOKEN"),
jailbreakEnabled=os.environ.get( jailbreakEnabled=os.environ.get(
"JAILBREAKENABLED", "false").lower() in ('true', '1', 't'), "JAILBREAKENABLED", "false").lower() \
in ('true', '1', 't'),
bing_auth_cookie=os.environ.get("BING_AUTH_COOKIE"), bing_auth_cookie=os.environ.get("BING_AUTH_COOKIE"),
markdown_formatted=os.environ.get( markdown_formatted=os.environ.get(
"MARKDOWN_FORMATTED", "false").lower() in ('true', '1', 't'), "MARKDOWN_FORMATTED", "false").lower() \
in ('true', '1', 't'),
output_four_images=os.environ.get(
"OUTPUT_FOUR_IMAGES", "false").lower() \
in ('true', '1', 't'),
import_keys_path=os.environ.get("IMPORT_KEYS_PATH"), import_keys_path=os.environ.get("IMPORT_KEYS_PATH"),
import_keys_password=os.environ.get( import_keys_password=os.environ.get(
"IMPORT_KEYS_PASSWORD"), "IMPORT_KEYS_PASSWORD"),
) )
if os.environ.get("IMPORT_KEYS_PATH") and os.environ.get("IMPORT_KEYS_PASSWORD") is not None: if os.environ.get("IMPORT_KEYS_PATH") \
and os.environ.get("IMPORT_KEYS_PASSWORD") is not None:
need_import_keys = True need_import_keys = True
await matrix_bot.login() await matrix_bot.login()