Merge pull request #172 from Hikari-Haru/compose-fix

Modernize GPT3Discord with gpt-index updates, async support, index renaming, bug fixes, QOL improvements

cogs/commands.py

@@ -523,6 +523,89 @@ class Commands(discord.Cog, name="Commands"):
     #
     # Index commands
     #
+    @add_to_group("index")
+    @discord.slash_command(
+        name="rename-user",
+        description="Select one of your saved indexes to rename",
+        guild_ids=ALLOWED_GUILDS,
+    )
+    @discord.guild_only()
+    @discord.option(
+        name="user_index",
+        description="Which user index to rename",
+        required=False,
+        autocomplete=File_autocompleter.get_user_indexes,
+    )
+    @discord.option(
+        name="new_name",
+        description="The new name",
+        required=False,
+        type=discord.SlashCommandOptionType.string,
+    )
+    async def rename_user_index(
+        self,
+        ctx: discord.ApplicationContext,
+        user_index: str,
+        new_name: str,
+    ):
+        await ctx.defer()
+        await self.index_cog.rename_user_index_command(ctx, user_index, new_name)
+
+    @add_to_group("index")
+    @discord.slash_command(
+        name="rename-server",
+        description="Select one of your saved server indexes to rename",
+        guild_ids=ALLOWED_GUILDS,
+    )
+    @discord.guild_only()
+    @discord.option(
+        name="server_index",
+        description="Which server index to rename",
+        required=False,
+        autocomplete=File_autocompleter.get_server_indexes,
+    )
+    @discord.option(
+        name="new_name",
+        description="The new name",
+        required=False,
+        type=discord.SlashCommandOptionType.string,
+    )
+    async def rename_server_index(
+        self,
+        ctx: discord.ApplicationContext,
+        server_index: str,
+        new_name: str,
+    ):
+        await ctx.defer()
+        await self.index_cog.rename_server_index_command(ctx, server_index, new_name)
+
+    @add_to_group("index")
+    @discord.slash_command(
+        name="rename-search",
+        description="Select one of your saved search indexes to rename",
+        guild_ids=ALLOWED_GUILDS,
+    )
+    @discord.guild_only()
+    @discord.option(
+        name="search_index",
+        description="Which search index to rename",
+        required=False,
+        autocomplete=File_autocompleter.get_user_search_indexes,
+    )
+    @discord.option(
+        name="new_name",
+        description="The new name",
+        required=False,
+        type=discord.SlashCommandOptionType.string,
+    )
+    async def rename_search_index(
+        self,
+        ctx: discord.ApplicationContext,
+        search_index: str,
+        new_name: str,
+    ):
+        await ctx.defer()
+        await self.index_cog.rename_search_index_command(ctx, search_index, new_name)
 
     @add_to_group("index")
     @discord.slash_command(
@@ -622,10 +705,21 @@ class Commands(discord.Cog, name="Commands"):
         required=False,
         input_type=discord.SlashCommandOptionType.channel,
     )
+    @discord.option(
+        name="message_limit",
+        description="The number of messages to index",
+        required=False,
+        input_type=discord.SlashCommandOptionType.integer,
+    )
     async def set_discord(
-        self, ctx: discord.ApplicationContext, channel: discord.TextChannel
+        self,
+        ctx: discord.ApplicationContext,
+        channel: discord.TextChannel,
+        message_limit: int,
     ):
-        await self.index_cog.set_discord_command(ctx, channel)
+        await self.index_cog.set_discord_command(
+            ctx, channel, message_limit=message_limit
+        )
 
     @add_to_group("index")
     @discord.slash_command(
@@ -634,9 +728,15 @@ class Commands(discord.Cog, name="Commands"):
         guild_ids=ALLOWED_GUILDS,
         checks=[Check.check_admin_roles(), Check.check_index_roles()],
     )
+    @discord.option(
+        name="message_limit",
+        description="The number of messages to index per channel",
+        required=False,
+        input_type=discord.SlashCommandOptionType.integer,
+    )
     @discord.guild_only()
-    async def discord_backup(self, ctx: discord.ApplicationContext):
-        await self.index_cog.discord_backup_command(ctx)
+    async def discord_backup(self, ctx: discord.ApplicationContext, message_limit: int):
+        await self.index_cog.discord_backup_command(ctx, message_limit=message_limit)
 
     @add_to_group("index")
     @discord.slash_command(
@@ -650,7 +750,7 @@ class Commands(discord.Cog, name="Commands"):
         required=False,
         default=1,
         min_value=1,
-        max_value=3,
+        max_value=5,
         input_type=discord.SlashCommandOptionType.integer,
     )
     @discord.option(
@@ -661,15 +761,27 @@ class Commands(discord.Cog, name="Commands"):
         default="default",
         choices=["default", "compact", "tree_summarize"],
     )
+    @discord.option(
+        name="child_branch_factor",
+        description="Only for deep indexes, how deep to go, higher is expensive.",
+        required=False,
+        default=1,
+        min_value=1,
+        max_value=3,
+        input_type=discord.SlashCommandOptionType.integer,
+    )
     async def query(
         self,
         ctx: discord.ApplicationContext,
         query: str,
         nodes: int,
         response_mode: str,
+        child_branch_factor: int,
     ):
         await ctx.defer()
-        await self.index_cog.query_command(ctx, query, nodes, response_mode)
+        await self.index_cog.query_command(
+            ctx, query, nodes, response_mode, child_branch_factor
+        )
 
     #
     # DALLE commands
@@ -859,7 +971,7 @@ class Commands(discord.Cog, name="Commands"):
         description="The higher the number, the more accurate the results, but more expensive",
         required=False,
         input_type=discord.SlashCommandOptionType.integer,
-        max_value=5,
+        max_value=8,
         min_value=1,
     )
     @discord.option(

cogs/index_service_cog.py

@@ -1,4 +1,5 @@
 import traceback
+from pathlib import Path
 
 import discord
 
@@ -24,6 +25,52 @@ class IndexService(discord.Cog, name="IndexService"):
         self.bot = bot
         self.index_handler = Index_handler(bot, usage_service)
 
+    async def rename_user_index_command(self, ctx, user_index, new_name):
+        """Command handler to rename a user index"""
+
+        if not new_name:
+            await ctx.respond("Please provide a new name for this index")
+            return
+
+        if await self.index_handler.rename_index(
+            ctx,
+            f"indexes/{ctx.user.id}/{user_index}",
+            f"indexes/{ctx.user.id}/{new_name}",
+        ):
+            await ctx.respond(f"Your index has been renamed to `{new_name}`")
+        else:
+            await ctx.respond("Something went wrong while renaming your index")
+
+    async def rename_server_index_command(self, ctx, server_index, new_name):
+        """Command handler to rename a user index"""
+
+        if not new_name:
+            await ctx.respond("Please provide a new name for this index")
+            return
+
+        if await self.index_handler.rename_index(
+            ctx,
+            f"indexes/{ctx.guild.id}/{server_index}",
+            f"indexes/{ctx.guild.id}/{new_name}",
+        ):
+            await ctx.respond(f"Your index has been renamed to `{new_name}`")
+        else:
+            await ctx.respond("Something went wrong while renaming your index")
+
+    async def rename_search_index_command(self, ctx, search_index, new_name):
+        if not new_name:
+            await ctx.respond("Please provide a new name for this index")
+            return
+
+        if await self.index_handler.rename_index(
+            ctx,
+            f"indexes/{ctx.user.id}_search/{search_index}",
+            f"indexes/{ctx.user.id}_search/{new_name}",
+        ):
+            await ctx.respond(f"Your index has been renamed to `{new_name}`")
+        else:
+            await ctx.respond("Something went wrong while renaming your index")
+
     async def set_index_command(
         self, ctx, file: discord.Attachment = None, link: str = None
     ):
@@ -56,7 +103,9 @@ class IndexService(discord.Cog, name="IndexService"):
                 ctx, link, user_api_key=user_api_key
             )
 
-    async def set_discord_command(self, ctx, channel: discord.TextChannel = None):
+    async def set_discord_command(
+        self, ctx, channel: discord.TextChannel = None, message_limit: int = 2500
+    ):
         """Command handler to set a channel as your personal index"""
         await ctx.defer()
 
@@ -69,7 +118,7 @@ class IndexService(discord.Cog, name="IndexService"):
                 return
 
         await self.index_handler.set_discord_index(
-            ctx, channel, user_api_key=user_api_key
+            ctx, channel, user_api_key=user_api_key, message_limit=message_limit
         )
 
     async def reset_command(self, ctx):
@@ -83,7 +132,7 @@ class IndexService(discord.Cog, name="IndexService"):
                 "Something went wrong while resetting your indexes. Contact the server admin."
             )
 
-    async def discord_backup_command(self, ctx):
+    async def discord_backup_command(self, ctx, message_limit: int = 2500):
         """Command handler to backup the entire server"""
         await ctx.defer()
 
@@ -94,7 +143,9 @@ class IndexService(discord.Cog, name="IndexService"):
             )
             if not user_api_key:
                 return
-        await self.index_handler.backup_discord(ctx, user_api_key=user_api_key)
+        await self.index_handler.backup_discord(
+            ctx, user_api_key=user_api_key, message_limit=message_limit
+        )
 
     async def load_index_command(self, ctx, user_index, server_index, search_index):
         """Command handler to load indexes"""
@@ -137,7 +188,9 @@ class IndexService(discord.Cog, name="IndexService"):
                 return
         await self.index_handler.load_index(ctx, index, server, search, user_api_key)
 
-    async def query_command(self, ctx, query, nodes, response_mode):
+    async def query_command(
+        self, ctx, query, nodes, response_mode, child_branch_factor
+    ):
         """Command handler to query your index"""
 
         user_api_key = None
@@ -153,7 +206,9 @@ class IndexService(discord.Cog, name="IndexService"):
             if await Moderation.simple_moderate_and_respond(query, ctx):
                 return
 
-        await self.index_handler.query(ctx, query, response_mode, nodes, user_api_key)
+        await self.index_handler.query(
+            ctx, query, response_mode, nodes, user_api_key, child_branch_factor
+        )
 
     async def compose_command(self, ctx, name):
         """Command handler to compose from your index"""

cogs/text_service_cog.py

@@ -793,6 +793,13 @@ class GPT3ComCon(discord.Cog, name="GPT3ComCon"):
         user = ctx.user if is_context else ctx.author
         prompt = await self.mention_to_username(ctx, prompt.strip())
 
+        if len(prompt) < self.model.prompt_min_length:
+            alias = ctx.respond if is_context else ctx.send
+            await alias(
+                f"Prompt must be greater than {self.model.prompt_min_length} characters, it is currently: {len(prompt)} characters"
+            )
+            return
+
         user_api_key = None
         if USER_INPUT_API_KEYS:
             user_api_key = await TextService.get_user_api_key(user.id, ctx, USER_KEY_DB)
@@ -846,6 +853,13 @@ class GPT3ComCon(discord.Cog, name="GPT3ComCon"):
         text = await self.mention_to_username(ctx, text.strip())
         instruction = await self.mention_to_username(ctx, instruction.strip())
 
+        # Validate that  all the parameters are in a good state before we send the request
+        if len(instruction) < self.model.prompt_min_length:
+            await ctx.respond(
+                f"Instruction must be at least {self.model.prompt_min_length} characters long"
+            )
+            return
+
         user_api_key = None
         if USER_INPUT_API_KEYS:
             user_api_key = await TextService.get_user_api_key(user.id, ctx, USER_KEY_DB)

detailed_guides/CUSTOM-INDEXES.md

@@ -12,4 +12,8 @@ Supported filetypes:
 Index Compositions:  
 Indexes can be combined with other indexes through a composition. To combine indexes, you can run the `/index compose` command, and select the indexes that you want to combine together. You should only combine relevant indexes together, combining irrelevant indexes together will result in poor results (for example, don't upload a math textbook and then upload a large set of poems and combine them together). When creating a composition, you will be given the option to do a "Deep" composition, deep compositions are more detailed and will give you better results, but are incredibly costly and will sometimes take multiple minutes to compose.  
   
-You can also compose a singular index with itself with "Deep Compose", this will give you a more detailed version of the index, but will be costly and will sometimes take multiple minutes to compose. **Deep compositions are useless for very short documents!**  
+You can also compose a singular index with itself with "Deep Compose", this will give you a more detailed version of the index, but will be costly and will sometimes take multiple minutes to compose. **Deep compositions are useless for very short documents!**  
+
+Doing a deep composition will also allow you to use the `child_branch_factor` parameter for `/index query`, increasing this past 1 will take a much longer time to query and will be much more expensive for large documents, so be wary.
+
+**When doing Deep Compositions, it's highly reccomended to keep the document size small, or only do deep compositions on single documents.** This is because a deep composition reorganizes the simple index into a tree structure and uses GPT3 to summarize different nodes of the tree, which will lead to high costs. For example, a deep composition of a 300 page lab manual and the contents of my personal website at https://kaveenk.com cost me $2 USD roughly.

gpt3discord.py

@@ -31,7 +31,7 @@ from services.environment_service import EnvService
 from models.openai_model import Model
 
 
-__version__ = "10.6.3"
+__version__ = "10.7.0"
 
 
 PID_FILE = Path("bot.pid")

models/index_model.py

@@ -1,3 +1,4 @@
+import functools
 import os
 import random
 import tempfile
@@ -18,6 +19,7 @@ from langchain import OpenAI
 
 from gpt_index.readers import YoutubeTranscriptReader
 from gpt_index.readers.schema.base import Document
+from gpt_index.langchain_helpers.text_splitter import TokenTextSplitter
 
 from gpt_index import (
     GPTSimpleVectorIndex,
@@ -46,7 +48,14 @@ SHORT_TO_LONG_CACHE = {}
 
 
 def get_and_query(
-    user_id, index_storage, query, response_mode, nodes, llm_predictor, embed_model
+    user_id,
+    index_storage,
+    query,
+    response_mode,
+    nodes,
+    llm_predictor,
+    embed_model,
+    child_branch_factor,
 ):
     index: [GPTSimpleVectorIndex, ComposableGraph] = index_storage[
         user_id
@@ -54,9 +63,10 @@ def get_and_query(
     if isinstance(index, GPTTreeIndex):
         response = index.query(
             query,
-            child_branch_factor=2,
+            child_branch_factor=child_branch_factor,
             llm_predictor=llm_predictor,
             embed_model=embed_model,
+            use_async=True,
         )
     else:
         response = index.query(
@@ -65,6 +75,7 @@ def get_and_query(
             llm_predictor=llm_predictor,
             embed_model=embed_model,
             similarity_top_k=nodes,
+            use_async=True,
         )
     return response
 
@@ -116,7 +127,8 @@ class IndexData:
             # First, clear all the files inside it
             for file in os.listdir(f"{app_root_path()}/indexes/{user_id}"):
                 os.remove(f"{app_root_path()}/indexes/{user_id}/{file}")
-
+            for file in os.listdir(f"{app_root_path()}/indexes/{user_id}_search"):
+                os.remove(f"{app_root_path()}/indexes/{user_id}_search/{file}")
         except Exception:
             traceback.print_exc()
 
@@ -139,6 +151,22 @@ class Index_handler:
         )
         self.EMBED_CUTOFF = 2000
 
+    async def rename_index(self, ctx, original_path, rename_path):
+        """Command handler to rename a user index"""
+
+        index_file = EnvService.find_shared_file(original_path)
+        if not index_file:
+            return False
+
+        # Rename the file at f"indexes/{ctx.user.id}/{user_index}" to f"indexes/{ctx.user.id}/{new_name}" using Pathlib
+        try:
+            if not rename_path.endswith(".json"):
+                rename_path = rename_path + ".json"
+            Path(original_path).rename(rename_path)
+            return True
+        except Exception as e:
+            return False
+
     async def paginate_embed(self, response_text):
         """Given a response text make embed pages and return a list of the pages. Codex makes it a codeblock in the embed"""
 
@@ -165,22 +193,26 @@ class Index_handler:
 
         return pages
 
-    # TODO We need to do predictions below for token usage.
     def index_file(self, file_path, embed_model) -> GPTSimpleVectorIndex:
         document = SimpleDirectoryReader(file_path).load_data()
-        index = GPTSimpleVectorIndex(document, embed_model=embed_model)
+        index = GPTSimpleVectorIndex(document, embed_model=embed_model, use_async=True)
         return index
 
     def index_gdoc(self, doc_id, embed_model) -> GPTSimpleVectorIndex:
         document = GoogleDocsReader().load_data(doc_id)
-        index = GPTSimpleVectorIndex(document, embed_model=embed_model)
+        index = GPTSimpleVectorIndex(document, embed_model=embed_model, use_async=True)
         return index
 
     def index_youtube_transcript(self, link, embed_model):
-        documents = YoutubeTranscriptReader().load_data(ytlinks=[link])
+        try:
+            documents = YoutubeTranscriptReader().load_data(ytlinks=[link])
+        except Exception as e:
+            raise ValueError(f"The youtube transcript couldn't be loaded: {e}")
+
         index = GPTSimpleVectorIndex(
             documents,
             embed_model=embed_model,
+            use_async=True,
         )
         return index
 
@@ -202,6 +234,7 @@ class Index_handler:
         index = GPTSimpleVectorIndex(
             documents,
             embed_model=embed_model,
+            use_async=True,
         )
         return index
 
@@ -216,6 +249,7 @@ class Index_handler:
         index = GPTSimpleVectorIndex(
             document,
             embed_model=embed_model,
+            use_async=True,
         )
         return index
 
@@ -252,10 +286,16 @@ class Index_handler:
                         # Detect if the link is a PDF, if it is, we load it differently
                         if response.headers["Content-Type"] == "application/pdf":
                             documents = await self.index_pdf(url)
-                            index = GPTSimpleVectorIndex(
-                                documents,
-                                embed_model=embed_model,
+                            index = await self.loop.run_in_executor(
+                                None,
+                                functools.partial(
+                                    GPTSimpleVectorIndex,
+                                    documents=documents,
+                                    embed_model=embed_model,
+                                    use_async=True,
+                                ),
                             )
+
                             return index
         except:
             raise ValueError("Could not load webpage")
@@ -263,7 +303,16 @@ class Index_handler:
         documents = BeautifulSoupWebReader(
             website_extractor=DEFAULT_WEBSITE_EXTRACTOR
         ).load_data(urls=[url])
-        index = GPTSimpleVectorIndex(documents, embed_model=embed_model)
+        # index = GPTSimpleVectorIndex(documents, embed_model=embed_model, use_async=True)
+        index = await self.loop.run_in_executor(
+            None,
+            functools.partial(
+                GPTSimpleVectorIndex,
+                documents=documents,
+                embed_model=embed_model,
+                use_async=True,
+            ),
+        )
         return index
 
     def reset_indexes(self, user_id):
@@ -331,7 +380,6 @@ class Index_handler:
         else:
             os.environ["OPENAI_API_KEY"] = user_api_key
 
-        # TODO Link validation
         try:
             embedding_model = OpenAIEmbedding()
 
@@ -380,6 +428,11 @@ class Index_handler:
 
             self.index_storage[ctx.user.id].add_index(index, ctx.user.id, file_name)
 
+        except ValueError as e:
+            await ctx.respond(str(e))
+            traceback.print_exc()
+            return
+
         except Exception:
             await ctx.respond("Failed to set index")
             traceback.print_exc()
@@ -392,6 +445,7 @@ class Index_handler:
         ctx: discord.ApplicationContext,
         channel: discord.TextChannel,
         user_api_key,
+        message_limit: int = 2500,
     ):
         if not user_api_key:
             os.environ["OPENAI_API_KEY"] = self.openai_key
@@ -400,7 +454,7 @@ class Index_handler:
 
         try:
             document = await self.load_data(
-                channel_ids=[channel.id], limit=1000, oldest_first=False
+                channel_ids=[channel.id], limit=message_limit, oldest_first=False
             )
             embedding_model = OpenAIEmbedding()
             index = await self.loop.run_in_executor(
@@ -445,6 +499,33 @@ class Index_handler:
             traceback.print_exc()
             await ctx.respond(e)
 
+    async def index_to_docs(
+        self, old_index, chunk_size: int = 4000, chunk_overlap: int = 200
+    ) -> List[Document]:
+        documents = []
+        for doc_id in old_index.docstore.docs.keys():
+            text = ""
+            if isinstance(old_index, GPTSimpleVectorIndex):
+                nodes = old_index.docstore.get_document(doc_id).get_nodes(
+                    old_index.docstore.docs[doc_id].id_map
+                )
+                for node in nodes:
+                    extra_info = node.extra_info
+                    text += f"{node.text} "
+            if isinstance(old_index, GPTTreeIndex):
+                nodes = old_index.docstore.get_document(doc_id).all_nodes.items()
+                for node in nodes:
+                    extra_info = node[1].extra_info
+                    text += f"{node[1].text} "
+            text_splitter = TokenTextSplitter(
+                separator=" ", chunk_size=chunk_size, chunk_overlap=chunk_overlap
+            )
+            text_chunks = text_splitter.split_text(text)
+            for text in text_chunks:
+                document = Document(text, extra_info=extra_info)
+                documents.append(document)
+        return documents
+
     async def compose_indexes(self, user_id, indexes, name, deep_compose):
         # Load all the indexes first
         index_objects = []
@@ -459,11 +540,7 @@ class Index_handler:
         if deep_compose:
             documents = []
             for _index in index_objects:
-                [
-                    documents.append(_index.docstore.get_document(doc_id))
-                    for doc_id in [docmeta for docmeta in _index.docstore.docs.keys()]
-                    if isinstance(_index.docstore.get_document(doc_id), Document)
-                ]
+                documents.extend(await self.index_to_docs(_index, 256, 20))
             llm_predictor = LLMPredictor(
                 llm=OpenAI(model_name="text-davinci-003", max_tokens=-1)
             )
@@ -476,6 +553,7 @@ class Index_handler:
                     documents=documents,
                     llm_predictor=llm_predictor,
                     embed_model=embedding_model,
+                    use_async=True,
                 ),
             )
 
@@ -497,11 +575,7 @@ class Index_handler:
         else:
             documents = []
             for _index in index_objects:
-                [
-                    documents.append(_index.docstore.get_document(doc_id))
-                    for doc_id in [docmeta for docmeta in _index.docstore.docs.keys()]
-                    if isinstance(_index.docstore.get_document(doc_id), Document)
-                ]
+                documents.extend(await self.index_to_docs(_index))
 
             embedding_model = OpenAIEmbedding()
 
@@ -511,6 +585,7 @@ class Index_handler:
                     GPTSimpleVectorIndex,
                     documents=documents,
                     embed_model=embedding_model,
+                    use_async=True,
                 ),
             )
 
@@ -525,7 +600,9 @@ class Index_handler:
             simple_index.save_to_disk(f"indexes/{user_id}/{name}.json")
             self.index_storage[user_id].queryable_index = simple_index
 
-    async def backup_discord(self, ctx: discord.ApplicationContext, user_api_key):
+    async def backup_discord(
+        self, ctx: discord.ApplicationContext, user_api_key, message_limit
+    ):
         if not user_api_key:
             os.environ["OPENAI_API_KEY"] = self.openai_key
         else:
@@ -536,7 +613,7 @@ class Index_handler:
             for c in ctx.guild.text_channels:
                 channel_ids.append(c.id)
             document = await self.load_data(
-                channel_ids=channel_ids, limit=3000, oldest_first=False
+                channel_ids=channel_ids, limit=message_limit, oldest_first=False
             )
             embedding_model = OpenAIEmbedding()
             index = await self.loop.run_in_executor(
@@ -567,6 +644,7 @@ class Index_handler:
         response_mode,
         nodes,
         user_api_key,
+        child_branch_factor,
     ):
         if not user_api_key:
             os.environ["OPENAI_API_KEY"] = self.openai_key
@@ -588,6 +666,7 @@ class Index_handler:
                     nodes,
                     llm_predictor,
                     embedding_model,
+                    child_branch_factor,
                 ),
             )
             print("The last token usage was ", llm_predictor.last_token_usage)

models/openai_model.py

@@ -649,13 +649,6 @@ class Model:
         codex=False,
         custom_api_key=None,
     ):
-        # Validate that  all the parameters are in a good state before we send the request
-        if len(instruction) < self.prompt_min_length:
-            raise ValueError(
-                "Instruction must be greater than 8 characters, it is currently "
-                + str(len(instruction))
-            )
-
         print(
             f"The text about to be edited is [{text}] with instructions [{instruction}] codex [{codex}]"
         )
@@ -831,10 +824,6 @@ class Model:
         Tuple[dict, bool]
     ):  # The response, and a boolean indicating whether or not the context limit was reached.
         # Validate that  all the parameters are in a good state before we send the request
-        if len(prompt) < self.prompt_min_length:
-            raise ValueError(
-                f"Prompt must be greater than {self.prompt_min_length} characters, it is currently: {len(prompt)} characters"
-            )
 
         if not max_tokens_override:
             if model:

pyproject.toml

@@ -37,6 +37,7 @@ dependencies = [
   "sentencepiece",
   "protobuf",
   "python-pptx",
+  "langchain",
 ]
 
 dynamic = ["version"]

requirements.txt

@@ -17,4 +17,5 @@ sentencepiece==0.1.97
 protobuf==3.20.2
 python-pptx==0.6.21
 sentence-transformers==2.2.2
-openai-whisper
+langchain==0.0.93
+openai-whisper

requirements_base.txt

@@ -15,4 +15,5 @@ PyPDF2==3.0.1
 youtube_transcript_api==0.5.0
 sentencepiece==0.1.97
 protobuf==3.20.2
-python-pptx==0.6.21
+python-pptx==0.6.21
+langchain==0.0.93

sample.env

@@ -2,11 +2,11 @@
 
 OPENAI_TOKEN = "<openai_api_token>"
 DISCORD_TOKEN = "<discord_bot_token>"
-#PINECONE_TOKEN = "<pinecone_token>" # pinecone token if you have it enabled. See readme
-#PINECONE_REGION = "<pinecone_region>" # add your region here if it's not us-west1-gcp
-#GOOGLE_SEARCH_API_KEY: "<google_api_key>"
-#GOOGLE_SEARCH_ENGINE_ID: "<google_engine_id>"
-#DEEPL_TOKEN: "<deepl_token>"
+# PINECONE_TOKEN = "<pinecone_token>" # pinecone token if you have it enabled. See readme
+# PINECONE_REGION = "<pinecone_region>" # add your region here if it's not us-west1-gcp
+# GOOGLE_SEARCH_API_KEY: "<google_api_key>"
+# GOOGLE_SEARCH_ENGINE_ID: "<google_engine_id>"
+# DEEPL_TOKEN: "<deepl_token>"
 
 DEBUG_GUILD = "974519864045756446"  # discord_server_id
 DEBUG_CHANNEL = "977697652147892304"  # discord_chanel_id
@@ -29,7 +29,7 @@ SEARCH_ROLES: "Admin,Owner"
 CUSTOM_BOT_NAME: "GPT3Discord"
 
 # If True, users must use their own API keys for OpenAI. If False, the bot will use the API key in the .env file.
-USER_INPUT_API_KEYS="False"
+USER_INPUT_API_KEYS = "False"
 
 # Moderations Service alert channel, this is where moderation alerts will be sent as a default if enabled
 MODERATIONS_ALERT_CHANNEL = "977697652147892304"
@@ -44,4 +44,4 @@ PRE_MODERATE = "False"
 FORCE_ENGLISH = "False"
 
 # The welcome message to send it the welcome setting is set to true
-WELCOME_MESSAGE = "Hi There! Welcome to our Discord server. We hope you'll enjoy our server and we look forward to engaging with you!"  # This is a fallback message if gpt3 fails to generate a welcome message.
+WELCOME_MESSAGE = "Hi There! Welcome to our Discord server. We hope you'll enjoy our server and we look forward to engaging with you!"  # This is a fallback message if gpt3 fails to generate a welcome message.