Fix llm 01 patch

app/llm-poc-variant-01/README.md

@@ -8,15 +8,14 @@ It's simple and runs on the local machine with smaller sized and free LLMs.
 
 > Note: credits to this program goes to the original authors of [langchain-python-rag-privategpt](https://github.com/jmorganca/ollama/tree/main/examples/langchain-python-rag-privategpt) from Ivan Martinez who contributed to an example on [jmorganca/ollama](https://github.com/jmorganca/ollama).
 
-
 ## Table of Contents
 
 - [Requirements](#requirements)
 - [Installation](#installation)
 - [Setup](#setup)
-   - [Downloading Learning Path Index datasets](#downloading-learning-path-index-datasets)
-   - [Ingesting files](#ingesting-files)
-     - [via native shell CLI](#via-native-shell-cli)
+  - [Downloading Learning Path Index datasets](#downloading-learning-path-index-datasets)
+  - [Ingesting files](#ingesting-files)
+    - [via native shell CLI](#via-native-shell-cli)
 - [Usage](#usage)
   - [Ask questions](#ask-questions)
     - [via native shell CLI](#via-native-shell-cli-1)
@@ -42,9 +41,9 @@ List out the key requirements needed to run the project, such as:
   - or venv
   - or [pipenv](https://pipenv.pypa.io/en/latest/)
 - Docker (optional)
-- Ollama ([Download & Install(https://ollama.com/download))
+- Ollama ([Download & Install](https://ollama.com/download))
 - Windows:
-  -  Microsoft Visual C++ 14.0 or greater is required (needed when installing ```hnswlib``` )
+  - Microsoft Visual C++ 14.0 or greater is required (needed when installing `hnswlib` )
 
 ## Installation
 
@@ -91,7 +90,6 @@ and start the Ollama server
 ollama serve
 ```
 
-
 ## Setup
 
 ### Downloading Learning Path Index datasets
@@ -181,10 +179,11 @@ If this occurs then use the Docker container to run your commands, instructions
 
 #### via native shell CLI
 
-Before running ```lpiGPT.py```you need to specify the base URL for the Ollama API or the local instance of Ollama running on your machine. By default this will return a ```None``` value.
+Before running `lpiGPT.py`you need to specify the base URL for the Ollama API or the local instance of Ollama running on your machine. By default this will return a `None` value.
 
 - Windows:
   - This is typically http://localhost:11434 and can be set by using the following in command line:
+
 ```shell
 set OLLAMA_HOST=http://localhost:11434
 ```
@@ -218,8 +217,7 @@ Note: The results will be displayed in a table format with columns for Course Na
 
 To exit the GPT prompt, press Ctrl-C or Ctrl-D and it will return to the Linux/Command-prompt.
 
-
-```bash                                                             
+```bash
 > python3 lpiGPT.py --help
 usage: lpiGPT.py [-h] [--chat-model CHAT_MODEL] [--embeddings-model-name EMBEDDINGS_MODEL_NAME] [--persist-directory PERSIST_DIRECTORY]
                  [--target-source-chunks TARGET_SOURCE_CHUNKS] [--hide-source] [--mute-stream]
@@ -265,7 +263,6 @@ root@[your machine name]:/home#:
 
 in there, type the same commands as in the **via native shell CLI** sections of [Ingesting files](#ingesting-files) and [Ask questions](#ask-questions) respectively.
 
-
 ### Try a different model
 
 ```shell
@@ -280,7 +277,7 @@ Put any and all your files into the `source_documents` directory
 The supported extensions are:
 
 - `.csv`: CSV
-and others, we have trimmed them off from here to keep this example simple and concise.
+  and others, we have trimmed them off from here to keep this example simple and concise.
 
 [back to ToC](#table-of-contents)
 
@@ -301,6 +298,17 @@ For chat models, have a look at [this list](https://github.com/jmorganca/ollama/
 
 _Please share your resources on either or both of the Embeddings and Chat models with us_
 
+## Running the Chainlit UI
+
+![Preview of the Chainlit app](deploy/gcp/chainlit-app-demo.gif "Preview of the Chainlit app")
+
+Chainlit allows us to quickly spin up an LLM friendly UI to interact with different models.
+To start the app:
+
+```bash
+chainlit run chainlit_app.py --host 0.0.0.0 --port 8000
+```
+
 ## Contributing
 
 We are open to any or all of the below from your side in terms of contributions:

app/llm-poc-variant-01/docker/run-docker-container.sh

@@ -9,9 +9,9 @@ FULL_DOCKER_TAG_NAME="python-3.10-docker-env"
 echo "Running image ${FULL_DOCKER_TAG_NAME}"; echo ""
 
 pullImage() {
-	FULL_DOCKER_TAG_NAME="${DOCKER_USER_NAME}/${FULL_DOCKER_TAG_NAME}"
-
-	docker pull ${FULL_DOCKER_TAG_NAME} || true
+    FULL_DOCKER_TAG_NAME="${DOCKER_USER_NAME}/${FULL_DOCKER_TAG_NAME}"
+    
+    docker pull ${FULL_DOCKER_TAG_NAME} || true
 }
 
 WORKDIR="/home/"
@@ -23,16 +23,16 @@ HF_CACHE_SHARED="--volume ${LOCAL_MODEL_FOLDER}/.cache:/root/.cache"
 set -x
 
 # pullImage
-time docker run --rm  -it --network="host" \
-                --platform="linux/amd64"   \
-                --network="host"           \
-                --add-host=host.docker.internal:host-gateway \
-                --workdir "${WORKDIR}"     \
-                --env OLLAMA_HOST="http://host.docker.internal:11434" \
-                ${HF_CACHE_SHARED}         \
-                ${MODEL_VOLUME_SHARED}     \
-                ${OLLAMA_VOLUME_SHARED}    \
-                "${FULL_DOCKER_TAG_NAME}"
+time docker run --rm  -it                 \
+--platform="linux/amd64"   \
+--network="host"           \
+--add-host=host.docker.internal:host-gateway \
+--workdir "${WORKDIR}"     \
+--env OLLAMA_HOST="http://host.docker.internal:11434" \
+${HF_CACHE_SHARED}         \
+${MODEL_VOLUME_SHARED}     \
+${OLLAMA_VOLUME_SHARED}    \
+"${FULL_DOCKER_TAG_NAME}"
 set +x
 
 echo "* Finished running docker image ${FULL_DOCKER_TAG_NAME}"

app/llm-poc-variant-01/requirements.txt

@@ -16,7 +16,7 @@ chainlit==1.1.404
 charset-normalizer==3.4.0
 chevron==0.14.0
 chroma-hnswlib==0.7.6
-chromadb==0.5.20
+chromadb==0.3.26
 click==8.1.7
 coloredlogs==15.0.1
 dataclasses-json==0.5.14

app/qooper-scraper/.env.example

@@ -0,0 +1,5 @@
+QOOPER_EMAIL=user@email.com
+QOOPER_PASSWORD=password
+
+CHROME_DRIVER_PATH=/chromedriver-linux64/chromedriver
+CHROME_PATH=/usr/bin/google-chrome

app/qooper-scraper/.gitignore

@@ -0,0 +1,3 @@
+output/
+chromedriver*
+qooper-token.txt

app/qooper-scraper/README.md

@@ -0,0 +1,20 @@
+- Setup Selenium (Chrome Driver, Chrome install)
+Grab the Chrome installer and driver from https://googlechromelabs.github.io/chrome-for-testing/
+
+```bash
+wget https://storage.googleapis.com/chrome-for-testing-public/131.0.6778.85/linux64/chromedriver-linux64.zip
+```
+
+Selenium and Chrome in WSL
+```bash
+https://www.gregbrisebois.com/posts/chromedriver-in-wsl2/
+```
+
+- Setup environment variables
+.env.example is a template to become .env
+
+
+- Works by signing in to the site using Selenium (sign-in involves multiple APIs and services, and so would be difficult to achieve with vanilla web scrapping)
+
+- Uses the tokens retrieved to query Qooper API
+

app/qooper-scraper/get_token.py

@@ -0,0 +1,107 @@
+import os
+import sys
+import json
+
+from selenium import webdriver
+from selenium.webdriver.common.by import By
+from selenium.webdriver.chrome.service import Service
+from selenium.webdriver.common.keys import Keys
+from selenium.webdriver.support import expected_conditions as EC
+from selenium.webdriver.remote.webdriver import BaseWebDriver
+from selenium.webdriver.support.wait import WebDriverWait
+from dotenv import load_dotenv
+
+load_dotenv()
+
+
+QOOPER_HOME_PAGE = "https://mentoring.qooper.io/"
+
+try:
+    CHROME_DRIVER_PATH = os.environ.get(
+        "CHROME_DRIVER_PATH", r"chromedriver-linux64/chromedriver"
+    )
+    CHROME_PATH = os.environ.get("CHROME_PATH", r"/usr/bin/google-chrome")
+    QOOPER_EMAIL = os.environ["QOOPER_EMAIL"]
+    QOOPER_PASSWORD = os.environ["QOOPER_PASSWORD"]
+except KeyError as error:
+    print(f"Set environment variables: {error.with_traceback()}")
+    sys.exit(1)
+
+
+def wait_for_element_visible(
+    driver: BaseWebDriver, element_identifier, by: By = By.XPATH, timeout=10
+):
+    return WebDriverWait(driver, timeout).until(
+        EC.presence_of_element_located((by, element_identifier))
+    )
+
+
+def wait_for_element_clickable(
+    driver: BaseWebDriver, element_identifier, by: By = By.XPATH, timeout=10
+):
+    return WebDriverWait(driver, timeout).until(
+        EC.element_to_be_clickable((by, element_identifier))
+    )
+
+
+def login_to_qooper(driver: webdriver.Chrome):
+    driver.get(QOOPER_HOME_PAGE)
+
+    input_email = wait_for_element_visible(
+        driver,
+        "input[type=email]",
+        By.CSS_SELECTOR,
+    )
+    input_email.send_keys(QOOPER_EMAIL)
+    next_button = driver.find_element(By.CSS_SELECTOR, "button.btn.bg-blue")
+    next_button.click()
+
+    input_password = wait_for_element_visible(
+        driver,
+        "input[type=password]",
+        By.CSS_SELECTOR,
+    )
+    input_password.send_keys(QOOPER_PASSWORD)
+    next_button = driver.find_element(By.CSS_SELECTOR, "button.btn.bg-blue")
+    next_button.click()
+
+    radio_input_kagglex = wait_for_element_clickable(
+        driver,
+        'label[for="KaggleX Fellowship Program - Cohort 4"]',
+        By.CSS_SELECTOR,
+        timeout=30,
+    )
+
+    radio_input_kagglex.click()
+    next_button = driver.find_element(By.CSS_SELECTOR, 'button[label="Continue"]')
+    next_button.click()
+
+    wait_for_element_visible(
+        driver,
+        "p",
+        By.TAG_NAME,
+    )
+
+    auth_token = driver.execute_script("return localStorage.getItem('qooper_atfu');")
+    auth_token = json.loads(auth_token)
+    print('Auth token: ', f"{auth_token[:4]}...{auth_token[-4:]}")
+    with open('qooper-token.txt', 'w') as f:
+        f.write(auth_token)
+
+
+options = webdriver.ChromeOptions()
+options.page_load_strategy = "eager"
+options.add_argument("--no-sandbox")
+service = Service(executable_path=CHROME_DRIVER_PATH)
+
+browser = webdriver.Chrome(
+    options=options,
+    service=service,
+)
+
+try:
+    login_to_qooper(browser)
+
+finally:
+    browser.quit()
+

app/qooper-scraper/requirements.txt

@@ -0,0 +1,21 @@
+annotated-types==0.7.0
+attrs==24.2.0
+certifi==2024.8.30
+charset-normalizer==3.4.0
+h11==0.14.0
+idna==3.10
+outcome==1.3.0.post0
+pydantic==2.10.2
+pydantic_core==2.27.1
+PySocks==1.7.1
+python-dotenv==1.0.1
+requests==2.32.3
+selenium==4.26.1
+sniffio==1.3.1
+sortedcontainers==2.4.0
+trio==0.27.0
+trio-websocket==0.11.1
+typing_extensions==4.12.2
+urllib3==2.2.3
+websocket-client==1.8.0
+wsproto==1.2.0

app/qooper-scraper/schemas.py

@@ -0,0 +1,79 @@
+from pydantic import BaseModel, RootModel
+from typing import Optional, Generic, TypeVar
+
+T = TypeVar("T")
+
+
+class Group(BaseModel, Generic[T]):
+    id: int
+    name: str
+    description: Optional[str] = ""
+    image_url: Optional[str]
+    kind: str
+    tags: list[str] = []
+    is_joined: bool = True
+    members_count: int = 0
+    resources_count: int = 0
+    discussions_count: int = 0
+    past_events_count: int = 0
+    upcoming_events_count: int = 0
+    created_at: str
+
+
+class GroupList(RootModel):
+    root: list[Group]
+
+
+class GroupMember(BaseModel, Generic[T]):
+    id: int
+    first_name: str
+    last_name: str
+    image_url: Optional[str]
+    current_position: Optional[str]
+    current_organization: Optional[str]
+    roles: list[str] = []
+
+
+class GroupMemberList(RootModel):
+    root: list[GroupMember]
+
+
+class GroupEvent(BaseModel, Generic[T]):
+    id: int
+    title: str
+    timezone: str
+    start_time: str
+    end_time: str
+    image_url: Optional[str]
+    updated_at: str
+    created_at: str
+    address: Optional[str]
+    # When scraping discussions or events,
+    # store their publishers in a separate group member table,
+    # Then keep only the ID on the relevant record
+    publisher: GroupMember["id"]
+
+
+class GroupEventList(RootModel):
+    root: list[GroupEvent]
+
+
+class Discussion(BaseModel, Generic[T]):
+    id: int
+    # If comments_count > 2,
+    # query the discussion detail page to get all comments
+    comments_count: Optional[int] = None
+    upvotes_count: Optional[int] = None
+    attachments_count: Optional[int] = None
+    title: Optional[str] = None
+    description: Optional[str] = None
+    created_at: str
+    publisher: GroupMember["id"]
+    tags: list[str] = []
+    upvotes_user_ids: list[int] = []
+    comments: list[int] = []
+    is_comment: bool = False
+
+
+class DiscussionList(RootModel):
+    root: list[Discussion]