Sendable Requests to Server

These requests allows you to initialize a session, upload images and PDF-like files, set instructions, start LLM sessions, analyse files, generate study guides, flashcards, worksheets, mindmaps, and retrieve history.

All communication is done via POST requests to the server endpoint:

https://6q4-meon3oytck2ykijcg-comzfeezo-custom.service.onethingrobot.com/upload

Each request uses FormData with a command key and optional additional fields.

1. Initialize Session

Command: init_session
This command initializes a new study session under a specific user. It verifies both the session ID and username, creates the required folder structure under the user’s directory, and sets up a default conversation history for the LLM.

FormData:

formData.append("command", "init_session");
formData.append("session", "<SESSION_ID>");
formData.append("user", "<USERNAME>");

Server Behavior:

Verifies both id (session ID) and user (username) are provided.

Creates the following directory structure:

ROOT_DIR
└── <USERNAME>/
    └── <SESSION_ID>/
        ├── pdfs/
        └── imgs/

Initializes a default message history (messages.json) in the session folder with a system instruction and a first user prompt:

[
  {
    "role": "system",
    "content": "Your task is to serve as a study tool - given the info that will be provided next, in form of either text or image, you will try your best to understand them. You will be asked to generate a summary in case we have to check your understanding. You will generate a study guide covering the given information. The guide shall contain 1) a descriptive summary of all given materials in the form of a lecture note, 2) a set of flashcard questions, short in form, along with corresponding answers, 3) an organized worksheet served as an exercise, supplied along with an answer sheet. Your responses will be listened by a front end, so answer only what is being asked, and follow the form. Good Luck!"
  },
  {
    "role": "user",
    "content": "You will be writing a note and prepare some questions on following materials. Materials will be given"
  },
  {
    "role": "assistant",
    "content": "<Assistant’s initial response from Ollama>"
  }
]

Sends the initial messages to the LLM model via ollama.chat() and appends the assistant’s first response.
Saves this conversation in messages.json under the session directory.

Status:

✅ Success:

{ "message": "Session '<SESSION_ID>' initialized successfully for user '<USERNAME>'" }

❌ Failure:

Missing session ID:
```
{ "error": "No session ID provided" }
```
Missing username:
```
{ "error": "No permission." }
```

Example Usage:

const formData = new FormData();
formData.append("command", "init_session");
formData.append("id", "session001");
formData.append("user", "jayfeng");

fetch(SERVER_URL, { method: "POST", body: formData })
  .then(res => res.json())
  .then(console.log);

Notes:

Both id and user are required — the session won’t initialize otherwise.
The structure now organizes all sessions under user folders, enabling multiple users and multiple sessions per user.
Always call init_session before any upload, analysis, or generation commands.

2. Upload Image

Command: append_image
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "append_image");
formData.append("file", <File Object>);

Status:

Success: {"message": "File saved at: <path>: Success"}
Failure: {"error": "...error details..."}

3. Upload PDF-like

Command: append_pdflike
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "append_pdflike");
formData.append("file", <File Object>);

Status: Same as Upload Image

4. Remove Image

Command: remove_img
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "remove_img");
formData.append("filename", "<IMAGE_FILENAME>");

Status:

Success: {"message": "File '<IMAGE_FILENAME>' removed successfully."}
Failure:
- Session not initialized: {"error": "Session not initialized."}
- Filename missing: {"error": "No filename provided."}
- File not found: {"error": "File not found: <IMAGE_FILENAME>"}
- Other errors: {"error": "...error details..."}

5. Remove PDF

Command: remove_pdf FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "remove_pdf");
formData.append("filename", "<PDF_FILENAME>");

Status:

Success:
- PDF removed only: {"message": "PDF '<PDF_FILENAME>' removed successfully."}
- PDF and associated image folder removed: {"message": "PDF '<PDF_FILENAME>' removed successfully. Associated image folder '<BASENAME>' removed."}
Failure:
- Session not initialized: {"error": "Session not initialized."}
- Filename missing: {"error": "No filename provided."}
- PDF not found: {"error": "File not found: <PDF_FILENAME>"}
- Other errors: {"error": "...error details..."}

6. Set Instruction Text

Command: set_instruct
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "set_instruct");
formData.append("instruction_text", "<INSTRUCTION_TEXT>");

Status:

Success: {"message": "Instruction Text Reset Successful"}
Failure: {"error": "...error details..."}

7. Analyse PDF

Command: analyse_pdf
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "analyse_pdf");

Status:

Success: {"message": "Analyse PDFs Successful"}
Failure: {"error": "...error details..."}

8. Analyse Image

Command: analyse_img
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "analyse_img");

Status:

Success: {"message": "Analysing Images Successful"}
Failure: {"error": "...error details..."}

9. Generate Study Guide

Command: generate_study_guide
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "generate_study_guide");

Status:

Success: {"markdown": "<generated markdown content>", "mermaid": "<mindmap mermaid source>"}
Failure: {"error": "...error details..."}

10. Generate Flashcard Questions

Command: generate_flashcard_questions
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "generate_flashcard_questions");
formData.append("num_questions", "<NUMBER>");
formData.append("difficulty", "<easy|medium|hard>");

Status:

Success: {"flashcards": "<JSON flashcards>"}
Failure: {"error": "...error details..."}

11. Generate Worksheet Questions

Command: generate_worksheet_questions
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "generate_worksheet_questions");
formData.append("num_questions", "<NUMBER>");
formData.append("difficulty", "<easy|medium|hard>");

Status:

Success: {"worksheet": "<JSON worksheet>"}
Failure: {"error": "...error details..."}

12. Mark Worksheet Questions

Command: generate_worksheet_questions
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "mark_worksheet_questions");
formData.append("question", "<THE QUESTION FOR MARKING>");
formData.append("answer", "<USER ANSWER>");
formData.append("mark_scheme", "<MARK SCHEME>");   // This is optional, but `generate_worksheet_questions` will provide mark_scheme.

Status:

Success: {"marking": "<JSON marking>"}
Failure: {"error": "...error details..."}

13. Generate Podcast

Command: generate_podcast
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "mark_worksheet_questions");
formData.append("prompt", "An optional user prompt if need additional requirements");   // This is optional

Action:
When request is sent, a podcast script (refer to the schema later in the file) will be generated by the LLM, after which audios are generated via ElevenLabs API, along with a podcast cover image. The latter two will be uploaded to supabase at location <USER>/<SESSION_ID>/podcasts/<PODCAST_ID>/i, under the media basket (i represents the i-th audio). The script will be returned back over the POST request (refer to the following Status section).

Status:

Success: {"script": "<JSON script>"}
Failure: {"error": "...error details..."}

13. Generate Mindmap

Command: generate_mindmap
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "generate_mindmap");

Status:

Success: {"message": "Generate Mindmap Successful"}
Failure: {"error": "...error details..."}

14. Retrieve Full History

Command: retrieve_full_history
FormData:

formData.append("command", "retrieve_full_history");

Status:

Success: {"message": "Full History Uploaded."}
Failure: {"error": "...error details..."}

To actually fetch the history, call the separate endpoint:
GET /full_history → {"full_history": [...messages...]}

15. Inference from Prompt

Command: inference_from_prompt
FormData:

formData.append("user", <USER_ID>);
formData.append("session", <SESSION_ID>);
formData.append("command", "inference_from_prompt");
formData.append("prompt", "<PROMPT_TEXT>");

Status:

Success: {"last_response": "<generated content>"}
Failure: {"error": "...error details..."}

Notes

Always run init_session first before any other command.
Uploads require valid File objects.
Errors always come as JSON: {"error": "...message..."}.
generate_flashcard_answers and generate_worksheet_answers are not separate commands (handled inside their respective question-generation commands).

Worksheet JSON Format:

workspaceId,
title: 'Some title'`,
description: 'Some description for generated practice problems',
difficulty: 'EASY/MEDIUM/HARD',
estimatedTime: '30 min',
'problems": [
        {{
            "question": "<Question 1>",
            "answer": "<Answer 1>",
            "type": "TEXT",
            "options": [Keep this empty],
            "mark_scheme": {
                "points": [
                    {
                        "point": 1,
                        "requirements": "Description of what's needed for this point"
                    }
                ],
                "totalPoints": <integer value representing total points for this question>
            }
        }},
        {{
            "question": "<Question 2>",
            "answer": "<Answer 2>",
            "type": "TEXT",
            "options": [Keep this empty],
            "mark_scheme": {
                "points": [
                    {
                        "point": 1,
                        "requirements": "Description of what's needed for this point"
                    }
                ],
                "totalPoints": <integer value representing total points for this question>
            }
        }},
        {{
            "question": "<Question 3>",
            "answer": "<Answer 3>",
            "type": "MULTIPLE_CHOICE",
            "options": ["<Option 1>", "<Option 2>", "<Option 3>"],
            "mark_scheme": {
                "points": [
                    {
                        "point": 1,
                        "requirements": "Answer explanation for this MCQ"
                    }
                ],
                "totalPoints": <integer value representing total points for this question>
            },
            "points": <integer value representing total points for this question>
        }},
        {{
            "question": "<Question 4>",
            "answer": "<Answer 4>",
            "type": "NUMERIC",
            "options": [Keep this empty],
            "mark_scheme": {
                "points": [
                    {
                        "point": 1,
                        "requirements": "Answer explanation for this numeric question"
                    }
                ],
                "totalPoints": <integer value representing total points for this question>
            },
            "points": <integer value representing total points for this question>
        }},
        {{
            "question": "<Question 5>",
            "answer": "<Answer 5>",
            "type": "TRUE_FALSE",
            "options": [Keep this empty],
            "mark_scheme": {
                "points": [
                    {
                        "point": 1,
                        "requirements": "Answer explanation for this true/false question"
                    }
                ],
                "totalPoints": <integer value representing total points for this question>
            },
            "points": <integer value representing total points for this question>
        }},
        {{
            "question": "<Question 6>",
            "answer": "<Answer 6>",
            "type": "MATCHING",
            "options": ["Option 1", "Option 2", "Option 3"],
            "mark_scheme": {
                "points": [
                    {
                        "point": 1,
                        "requirements": "Answer explanation for this matching question"
                    }
                ],
                "totalPoints": <integer value representing total points for this question>
            },
            "points": <integer value representing total points for this question>
        }}]

Marking JSON Format:

{
  totalPoints: <sum of all achievedPoints (total points earned by student)>,
  points: [
    {
      point: <max points for this criteria>,
      requirements: <requirements text from mark scheme>,
      achievedPoints: <points earned for this criteria>,
      feedback: <specific feedback for this criteria>
    },
    ... (one object for each marking point in the mark scheme)
  ]
}

Postcast Script Format:

[
<Section One Script (str)>,
<Section Two Script (str)>,
<Section Three Script (str)>,
...
]

Notes

Make sure to initialize a session first (init_session) before sending any other commands to that session.
All file uploads require a File object from a <input type="file"> element.
All output JSON responses are displayed in <pre> blocks for readability in the frontend.
Errors will be returned as HTTP errors or caught exceptions.

Endpoint: `/session_files`

Returns a listing of all files grouped by session.
A session is represented by a directory in the server’s working directory that contains at least a pdfs/ or imgs/ subfolder.

Method

GET /session_files

Response

200 OK — JSON object containing session data.
400/500 — Error response if something went wrong.

JSON Format Example

{
  "user_count": 5,
  "users": {
    "FantasticJellyFish": {
      "61016": {
        "counts": {
          "all": 1,
          "imgs": 0,
          "pdfs": 1
        },
        "imgs": [],
        "pdfs": [
          {
            "modified_iso": "2025-10-11T21:20:50",
            "modified_ts": 1760217650,
            "name": "AP Lab 04 - Diffusion and Osmosis 2016.pdf",
            "path": "Data/FantasticJellyFish/61016/pdfs/AP Lab 04 - Diffusion and Osmosis 2016.pdf",
            "size_bytes": 320573,
            "type": "pdf"
          }
        ],
        "user": "FantasticJellyFish"
      }
    },
    "HappyCat": {
      "1": {
        "counts": {
          "all": 1,
          "imgs": 1,
          "pdfs": 0
        },
        "imgs": [
          {
            "modified_iso": "2025-10-11T20:21:58",
            "modified_ts": 1760214118,
            "name": "figure-05-02-04.jpeg",
            "path": "Data/HappyCat/1/imgs/figure-05-02-04.jpeg",
            "size_bytes": 162699,
            "type": "img"
          }
        ],
        "pdfs": [],
        "user": "HappyCat"
      },
      "100": {
        "counts": {
          "all": 1,
          "imgs": 1,
          "pdfs": 0
        },
        "imgs": [
          {
            "modified_iso": "2025-10-11T19:26:42",
            "modified_ts": 1760210802,
            "name": "figure-05-02-04.jpeg",
            "path": "Data/HappyCat/100/imgs/figure-05-02-04.jpeg",
            "size_bytes": 162699,
            "type": "img"
          }
        ],
        "pdfs": [],
        "user": "HappyCat"
      },
      "200": {
        "counts": {
          "all": 1,
          "imgs": 1,
          "pdfs": 0
        },
        "imgs": [
          {
            "modified_iso": "2025-10-11T19:55:12",
            "modified_ts": 1760212512,
            "name": "figure-05-02-04.jpeg",
            "path": "Data/HappyCat/200/imgs/figure-05-02-04.jpeg",
            "size_bytes": 162699,
            "type": "img"
          }
        ],
        "pdfs": [],
        "user": "HappyCat"
      }
    },
    "SadDog": {
      "100": {
        "counts": {
          "all": 0,
          "imgs": 0,
          "pdfs": 0
        },
        "imgs": [],
        "pdfs": [],
        "user": "SadDog"
      },
      "610": {
        "counts": {
          "all": 0,
          "imgs": 0,
          "pdfs": 0
        },
        "imgs": [],
        "pdfs": [],
        "user": "SadDog"
      }
    },
    "SadPig": {
      "101": {
        "counts": {
          "all": 0,
          "imgs": 0,
          "pdfs": 0
        },
        "imgs": [],
        "pdfs": [],
        "user": "SadPig"
      }
    },
    "cmesjxa2i0000ry9oyjmoasjk": {
      "cmesjxwhz0002ry9ozjg0gpvn": {
        "counts": {
          "all": 0,
          "imgs": 0,
          "pdfs": 0
        },
        "imgs": [],
        "pdfs": [],
        "user": "cmesjxa2i0000ry9oyjmoasjk"
      }
    }
  }
}

Notes

sessions is a dictionary keyed by session name.
Each session includes:
- counts: number of PDFs, images, and total files.
- pdfs, imgs, all: arrays of file metadata.
File metadata includes:
- name: filename only.
- path: relative path to the file.
- type: "pdf" or "img".
- size_bytes: file size.
- modified_ts: last modified timestamp (epoch).
- modified_iso: last modified time (ISO format).

Name		Name	Last commit message	Last commit date
Latest commit History 65 Commits
Data/cmesjxa2i0000ry9oyjmoasjk		Data/cmesjxa2i0000ry9oyjmoasjk
__pycache__		__pycache__
app		app
.gitignore		.gitignore
FILE_PROCESSING.md		FILE_PROCESSING.md
LICENSE		LICENSE
LLM.py		LLM.py
Procfile		Procfile
README.md		README.md
SCHEMA.md		SCHEMA.md
WORKSPACE_CONTEXT_USAGE.md		WORKSPACE_CONTEXT_USAGE.md
create_messages_table.sql		create_messages_table.sql
inferenceCLI.py		inferenceCLI.py
listener.py		listener.py
login.html		login.html
markdownConvertor.py		markdownConvertor.py
mermaid.html		mermaid.html
requirements.txt		requirements.txt
runtime.txt		runtime.txt
setup.py		setup.py
simpleFrontEnd.html		simpleFrontEnd.html
update_messages_to_db.py		update_messages_to_db.py

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Repository files navigation

Sendable Requests to Server

1. Initialize Session

2. Upload Image

3. Upload PDF-like

4. Remove Image

5. Remove PDF

6. Set Instruction Text

7. Analyse PDF

8. Analyse Image

9. Generate Study Guide

10. Generate Flashcard Questions

11. Generate Worksheet Questions

12. Mark Worksheet Questions

13. Generate Podcast

13. Generate Mindmap

14. Retrieve Full History

15. Inference from Prompt

Notes

Worksheet JSON Format:

Marking JSON Format:

Postcast Script Format:

Notes

Endpoint: `/session_files`

Method

Response

JSON Format Example

Notes

About

Uh oh!

Releases

Packages

Contributors 2

Uh oh!

Languages

License

goscribe/inference-backend

Folders and files

Latest commit

History

Repository files navigation

Sendable Requests to Server

1. Initialize Session

2. Upload Image

3. Upload PDF-like

4. Remove Image

5. Remove PDF

6. Set Instruction Text

7. Analyse PDF

8. Analyse Image

9. Generate Study Guide

10. Generate Flashcard Questions

11. Generate Worksheet Questions

12. Mark Worksheet Questions

13. Generate Podcast

13. Generate Mindmap

14. Retrieve Full History

15. Inference from Prompt

Notes

Worksheet JSON Format:

Marking JSON Format:

Postcast Script Format:

Notes

Endpoint: /session_files

Method

Response

JSON Format Example

Notes

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Contributors 2

Uh oh!

Languages

Endpoint: `/session_files`

Packages