Merge pull request #15 from lfnovo/transform_ui

Transform UI

.gitignore

@@ -1,3 +1,4 @@
+prompts/patterns/user/
 notebooks/
 data/
 .uploads/

README.md

@@ -6,6 +6,8 @@ In a world dominated by Artificial Intelligence, having the ability to think 
 
 Open Notebook empowers you to manage your research, generate AI-assisted notes, and interact with your content—on your terms.
 
+Learn more about our project at [https://www.open-notebook.ai](https://www.open-notebook.ai)
+
 ## ⚙️ Setting Up
 
 Go to the [Setup Guide](docs/SETUP.md) to learn how to set up the tool in details.

docs/SETUP.md

@@ -41,44 +41,7 @@ volumes:
   notebook_data:
 ```
 
-or with the environment variables:
-
-```yaml
-version: '3'
-
-services:
-  surrealdb:
-    image: surrealdb/surrealdb:v2
-    ports:
-      - "8000:8000"
-    volumes:
-      - surreal_data:/mydata
-    command: start --log trace --user root --pass root rocksdb:/mydata/mydatabase.db
-    pull_policy: always
-    user: root
-
-  open_notebook:
-    image: lfnovo/open_notebook:latest
-    ports:
-      - "8080:8502"
-    environment:
-        - OPENAI_API_KEY=API_KEY
-        - SURREAL_ADDRESSsurrealdb
-        - SURREAL_PORT=8000
-        - SURREAL_USER=root
-        - SURREAL_PASS=root
-        - SURREAL_NAMESPACE=open_notebook
-        - SURREAL_DATABASE=staging
-    depends_on:
-      - surrealdb
-    pull_policy: always
-    volumes:
-      - notebook_data:/app/data
-
-volumes:
-  surreal_data:
-  notebook_data:
-```
+Take a look at the [Open Notebook Boilerplate](https://github.com/lfnovo/open-notebook-boilerplate) repo with a sample of how to set it up for maximum feature usability. 
 
 ### 📦 Installing from Source
 

docs/TRANSFORMATIONS.md

@@ -24,42 +24,47 @@ For example, you could start by summarizing a text, then use that summary to gen
 
 ### Setting Up Transformations
 
+Take a look at the [Open Notebook Boilerplate](https://github.com/lfnovo/open-notebook-boilerplate) repo with a sample of how to set it up for maximum feature usability. 
+
 To set up your own Transformations, you'll define them in the `transformations.yaml` file. Below is an example setup:
 
 ```yaml
 source_insights:
   - name: "Summarize"
     insight_type: "Content Summary"
     description: "Summarize the content"
-    transformations:
-      - patterns/makeitdense
-      - patterns/summarize
+    patterns:
+      - patterns/default/makeitdense
+      - patterns/default/summarize
   - name: "Key Insights"
     insight_type: "Key Insights"
     description: "Extracts a list of the Key Insights of the content"
-    transformations:
-      - patterns/keyinsights
+    patterns:
+      - patterns/default/keyinsights
   - name: "Make it Dense"
     insight_type: "Dense Representation"
     description: "Create a dense representation of the content"
-    transformations:
-      - patterns/makeitdense
+    patterns:
+      - patterns/default/makeitdense
   - name: "Analyze Paper"
     insight_type: "Paper Analysis"
     description: "Analyze the paper and provide a quick summary"
-    transformations:
-      - patterns/analyze_paper
+    patterns:
+      - patterns/default/analyze_paper
   - name: "Reflection"
     insight_type: "Reflection Questions"
     description: "Generates a list of insightful questions to provoke reflection"
-    transformations:
-      - patterns/reflection_questions
+    patterns:
+      - patterns/default/reflection_questions
 ```
 
-Once you've defined your transformation, make sure to add the corresponding prompts to the `prompts/patterns` folder. Here's an example of a transformation prompt:
+You can mount this file to the docker image to replace its default value. 
+
+
+Once you've defined your transformation, make sure to add the corresponding prompts to the `prompts/patterns/user` folder. Here's an example of a transformation prompt:
 
 ```jinja
-{% include 'patterns/common_text.jinja' %}
+{% include 'patterns/user/common_text.jinja' %}
 
 # IDENTITY and PURPOSE
 
@@ -95,7 +100,6 @@ You extract deep, thought-provoking, and meaningful reflections from text conten
 - Any item that doesn't follow the `patterns/` format will be interpreted as a command (refer to `command.jinja` for clarity).
 
 
-
 ### Call for Contributions
 
 Have an idea for an amazing Transformation? We'd love to see your creativity! Please submit a pull request with your favorite transformations to help expand our library. Whether it's summarization, content analysis, or something entirely unique, your contributions will help us all get more out of our research!Leveraging Transformations in Open Notebook

open_notebook/config.py

@@ -36,18 +36,20 @@
 os.makedirs(PODCASTS_FOLDER, exist_ok=True)
 
 
-DEFAULT_MODELS = DefaultModels.load()
-
-if DEFAULT_MODELS.default_embedding_model:
-    EMBEDDING_MODEL = get_model(
-        DEFAULT_MODELS.default_embedding_model, model_type="embedding"
+def load_default_models():
+    default_models = DefaultModels.load()
+    embedding_model = (
+        get_model(default_models.default_embedding_model, model_type="embedding")
+        if default_models.default_embedding_model
+        else None
     )
-else:
-    EMBEDDING_MODEL = None
 
-if DEFAULT_MODELS.default_speech_to_text_model:
-    SPEECH_TO_TEXT_MODEL = get_model(
-        DEFAULT_MODELS.default_speech_to_text_model, model_type="speech_to_text"
+    speech_to_text_model = (
+        get_model(
+            default_models.default_speech_to_text_model, model_type="speech_to_text"
+        )
+        if default_models.default_speech_to_text_model
+        else None
     )
-else:
-    SPEECH_TO_TEXT_MODEL = None
+
+    return default_models, embedding_model, speech_to_text_model

open_notebook/domain/base.py

@@ -68,24 +68,27 @@ def get_embedding_content(self) -> Optional[str]:
         return None
 
     def save(self) -> None:
-        from open_notebook.config import EMBEDDING_MODEL
+        from open_notebook.config import load_default_models
+
+        DEFAULT_MODELS, EMBEDDING_MODEL, SPEECH_TO_TEXT_MODEL = load_default_models()
 
         try:
             logger.debug(f"Validating {self.__class__.__name__}")
             self.model_validate(self.model_dump(), strict=True)
             data = self._prepare_save_data()
-            data["updated"] = datetime.now().isoformat()
+            data["updated"] = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
 
             if self.needs_embedding():
                 embedding_content = self.get_embedding_content()
                 if embedding_content:
                     data["embedding"] = EMBEDDING_MODEL.embed(embedding_content)
 
             if self.id is None:
-                data["created"] = datetime.now().isoformat()
+                data["created"] = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
                 logger.debug("Creating new record")
                 repo_result = repo_create(self.__class__.table_name, data)
             else:
+                data["created"] = self.created.strftime("%Y-%m-%d %H:%M:%S")
                 logger.debug(f"Updating record with id {self.id}")
                 repo_result = repo_update(self.id, data)
 

open_notebook/domain/notebook.py

@@ -5,7 +5,7 @@
 from loguru import logger
 from pydantic import BaseModel, Field, field_validator
 
-from open_notebook.config import EMBEDDING_MODEL
+from open_notebook.config import load_default_models
 from open_notebook.database.repository import (
     repo_create,
     repo_query,
@@ -140,6 +140,8 @@ def save_chunks(self, text: str) -> None:
             raise DatabaseOperationError(e)
 
     def vectorize(self) -> None:
+        DEFAULT_MODELS, EMBEDDING_MODEL, SPEECH_TO_TEXT_MODEL = load_default_models()
+
         try:
             if not self.full_text:
                 return
@@ -189,6 +191,8 @@ def search(cls, query: str) -> List[Dict[str, Any]]:
             raise DatabaseOperationError("Failed to search sources")
 
     def add_insight(self, insight_type: str, content: str) -> Any:
+        DEFAULT_MODELS, EMBEDDING_MODEL, SPEECH_TO_TEXT_MODEL = load_default_models()
+
         if not insight_type or not content:
             raise InvalidInputError("Insight type and content must be provided")
         try:
@@ -209,6 +213,8 @@ def add_insight(self, insight_type: str, content: str) -> Any:
 
     # todo: move this to content processing pipeline as a major graph
     def generate_toc_and_title(self) -> "Source":
+        DEFAULT_MODELS, EMBEDDING_MODEL, SPEECH_TO_TEXT_MODEL = load_default_models()
+
         try:
             config = RunnableConfig(configurable=dict(thread_id=self.id))
             result = toc_graph.invoke({"content": self.full_text}, config=config)

open_notebook/graphs/chat.py

@@ -9,10 +9,12 @@
 from langgraph.graph.message import add_messages
 from typing_extensions import TypedDict
 
-from open_notebook.config import DEFAULT_MODELS, LANGGRAPH_CHECKPOINT_FILE
+from open_notebook.config import LANGGRAPH_CHECKPOINT_FILE, load_default_models
 from open_notebook.domain.notebook import Notebook
 from open_notebook.graphs.utils import run_pattern
 
+DEFAULT_MODELS, EMBEDDING_MODEL, SPEECH_TO_TEXT_MODEL = load_default_models()
+
 
 class ThreadState(TypedDict):
     messages: Annotated[list, add_messages]
@@ -22,12 +24,12 @@ class ThreadState(TypedDict):
 
 
 def call_model_with_messages(state: ThreadState, config: RunnableConfig) -> dict:
-    model_name = config.get("configurable", {}).get(
-        "model_name", DEFAULT_MODELS.default_chat_model
+    model_id = config.get("configurable", {}).get(
+        "model_id", DEFAULT_MODELS.default_chat_model
     )
     ai_message = run_pattern(
         "chat",
-        model_name,
+        model_id,
         messages=state["messages"],
         state=state,
     )

open_notebook/graphs/content_processing/audio.py

@@ -4,7 +4,7 @@
 from loguru import logger
 from pydub import AudioSegment
 
-from open_notebook.config import SPEECH_TO_TEXT_MODEL
+from open_notebook.config import load_default_models
 from open_notebook.graphs.content_processing.state import SourceState
 
 # future: parallelize the transcription process
@@ -72,6 +72,8 @@ def split_audio(input_file, segment_length_minutes=15, output_prefix=None):
 
 
 def extract_audio(data: SourceState):
+    DEFAULT_MODELS, EMBEDDING_MODEL, SPEECH_TO_TEXT_MODEL = load_default_models()
+
     input_audio_path = data.get("file_path")
     audio_files = []
 

open_notebook/graphs/doc_query.py

@@ -1,14 +1,15 @@
-import os
-
 from langchain_core.runnables import (
     RunnableConfig,
 )
 from langgraph.graph import END, START, StateGraph
 from typing_extensions import TypedDict
 
+from open_notebook.config import load_default_models
 from open_notebook.domain.notebook import Note, Notebook, Source
 from open_notebook.graphs.utils import run_pattern
 
+DEFAULT_MODELS, EMBEDDING_MODEL, SPEECH_TO_TEXT_MODEL = load_default_models()
+
 
 class DocQueryState(TypedDict):
     doc_id: str
@@ -19,10 +20,10 @@ class DocQueryState(TypedDict):
 
 
 def call_model(state: dict, config: RunnableConfig) -> dict:
-    model_name = config.get("configurable", {}).get(
-        "model_name", os.environ.get("RETRIEVAL_MODEL")
+    model_id = config.get("configurable", {}).get(
+        "model_id", DEFAULT_MODELS.default_transformation_model
     )
-    return {"answer": run_pattern("doc_query", model_name, state)}
+    return {"answer": run_pattern("doc_query", model_id, state)}
 
 
 # todo: there is probably a better way to do this and avoid repetition

open_notebook/graphs/multipattern.py

@@ -7,48 +7,50 @@
 from langgraph.graph import END, START, StateGraph
 from typing_extensions import Annotated, TypedDict
 
-from open_notebook.config import DEFAULT_MODELS
+from open_notebook.config import load_default_models
 from open_notebook.graphs.utils import run_pattern
 
+DEFAULT_MODELS, EMBEDDING_MODEL, SPEECH_TO_TEXT_MODEL = load_default_models()
+
 
 class PatternChainState(TypedDict):
     content_stack: Annotated[Sequence[str], operator.add]
-    transformations: List[str]
+    patterns: List[str]
     output: str
 
 
 def call_model(state: dict, config: RunnableConfig) -> dict:
-    model_name = config.get("configurable", {}).get(
-        "model_name", DEFAULT_MODELS.default_transformation_model
+    model_id = config.get("configurable", {}).get(
+        "model_id", DEFAULT_MODELS.default_transformation_model
     )
-    transformations = state["transformations"]
-    current_transformation = transformations.pop(0)
+    patterns = state["patterns"]
+    current_transformation = patterns.pop(0)
     if current_transformation.startswith("patterns/"):
         input_args = {"input_text": state["content_stack"][-1]}
     else:
         input_args = {
             "input_text": state["content_stack"][-1],
             "command": current_transformation,
         }
-        current_transformation = "patterns/custom"
+        current_transformation = "patterns/default/command"
 
     transformation_result = run_pattern(
         pattern_name=current_transformation,
-        model_name=model_name,
+        model_id=model_id,
         state=input_args,
     )
     return {
         "content_stack": [transformation_result.content],
         "output": transformation_result.content,
-        "transformations": state["transformations"],
+        "patterns": state["patterns"],
     }
 
 
 def transform_condition(state: PatternChainState) -> Literal["agent", END]:  # type: ignore
     """
     Checks whether there are more chunks to process.
     """
-    if len(state["transformations"]) > 0:
+    if len(state["patterns"]) > 0:
         return "agent"
     return END
 

open_notebook/graphs/pattern.py

@@ -4,9 +4,11 @@
 from langgraph.graph import END, START, StateGraph
 from typing_extensions import TypedDict
 
-from open_notebook.config import DEFAULT_MODELS
+from open_notebook.config import load_default_models
 from open_notebook.graphs.utils import run_pattern
 
+DEFAULT_MODELS, EMBEDDING_MODEL, SPEECH_TO_TEXT_MODEL = load_default_models()
+
 
 class PatternState(TypedDict):
     input_text: str
@@ -15,13 +17,13 @@ class PatternState(TypedDict):
 
 
 def call_model(state: dict, config: RunnableConfig) -> dict:
-    model_name = config.get("configurable", {}).get(
-        "model_name", DEFAULT_MODELS.default_transformation_model
+    model_id = config.get("configurable", {}).get(
+        "model_id", DEFAULT_MODELS.default_transformation_model
     )
     return {
         "output": run_pattern(
             pattern_name=state["pattern"],
-            model_name=model_name,
+            model_id=model_id,
             state=state,
         )
     }