Update file_utils.py
Browse files- file_utils.py +19 -105
file_utils.py
CHANGED
|
@@ -1,112 +1,26 @@
|
|
| 1 |
-
import
|
| 2 |
-
from
|
| 3 |
-
from
|
| 4 |
-
import os
|
| 5 |
|
| 6 |
-
# === GPT-4 Client Setup ===
|
| 7 |
-
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
|
| 8 |
-
MAX_TOKENS = 7000 # Safety buffer below GPT-4's 8192 limit
|
| 9 |
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
approx_tokens = len(prompt) // 4
|
| 14 |
-
if approx_tokens > MAX_TOKENS:
|
| 15 |
-
prompt = prompt[:MAX_TOKENS * 4] # truncate by character length
|
| 16 |
-
prompt += "\n[NOTE: Truncated due to length limit.]"
|
| 17 |
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
messages=[
|
| 21 |
-
{"role": "system", "content": system_message},
|
| 22 |
-
{"role": "user", "content": prompt}
|
| 23 |
-
],
|
| 24 |
-
temperature=temperature
|
| 25 |
-
)
|
| 26 |
-
return response.choices[0].message.content
|
| 27 |
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
text = ""
|
| 31 |
-
with fitz.open(pdf_path) as doc:
|
| 32 |
-
for page in doc:
|
| 33 |
-
text += page.get_text()
|
| 34 |
-
return text
|
| 35 |
|
|
|
|
|
|
|
| 36 |
|
| 37 |
-
def
|
| 38 |
-
|
| 39 |
-
|
|
|
|
|
|
|
|
|
|
| 40 |
|
| 41 |
-
|
| 42 |
-
def extract_text_from_md(file_path):
|
| 43 |
-
with open(file_path, 'r', encoding='utf-8') as f:
|
| 44 |
-
return f.read()
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
def extract_text_from_pptx(file_path):
|
| 48 |
-
text = ""
|
| 49 |
-
prs = Presentation(file_path)
|
| 50 |
-
for slide in prs.slides:
|
| 51 |
-
for shape in slide.shapes:
|
| 52 |
-
if hasattr(shape, "text"):
|
| 53 |
-
text += shape.text + "\n"
|
| 54 |
-
return text
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
def extract_text_from_file(file):
|
| 58 |
-
if file is None:
|
| 59 |
-
return ""
|
| 60 |
-
|
| 61 |
-
name = file.name.lower()
|
| 62 |
-
if name.endswith(".pdf"):
|
| 63 |
-
return extract_text_from_pdf(file.name)
|
| 64 |
-
elif name.endswith(".txt"):
|
| 65 |
-
return extract_text_from_txt(file.name)
|
| 66 |
-
elif name.endswith(".md"):
|
| 67 |
-
return extract_text_from_md(file.name)
|
| 68 |
-
elif name.endswith(".pptx"):
|
| 69 |
-
return extract_text_from_pptx(file.name)
|
| 70 |
-
else:
|
| 71 |
-
return "Unsupported file type."
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
# === GPT-4 Powered Content Generators ===
|
| 75 |
-
def generate_summary(text):
|
| 76 |
-
prompt = f"""Please summarize the following study material in a concise and organized manner:
|
| 77 |
-
|
| 78 |
-
{text}
|
| 79 |
-
"""
|
| 80 |
-
return call_llm(prompt, system_message="You are a helpful study assistant.")
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
def generate_flashcards(text):
|
| 84 |
-
prompt = f"""Based on the content below, create a set of helpful flashcards.
|
| 85 |
-
Each flashcard should be formatted as:
|
| 86 |
-
Q: Question?
|
| 87 |
-
A: Answer.
|
| 88 |
-
|
| 89 |
-
{text}
|
| 90 |
-
"""
|
| 91 |
-
return call_llm(prompt, system_message="You are a flashcard generator assistant.")
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
def generate_quiz(text):
|
| 95 |
-
prompt = f"""Create a short quiz (3-5 questions) based on the content below.
|
| 96 |
-
Include a mix of multiple choice and short answer questions.
|
| 97 |
-
|
| 98 |
-
{text}
|
| 99 |
-
"""
|
| 100 |
-
return call_llm(prompt, system_message="You are a quiz generator assistant.")
|
| 101 |
-
|
| 102 |
-
|
| 103 |
-
def answer_question(text, question):
|
| 104 |
-
prompt = f"""You are an AI tutor. Answer the following question based only on the content below.
|
| 105 |
-
|
| 106 |
-
Content:
|
| 107 |
-
{text}
|
| 108 |
-
|
| 109 |
-
Question:
|
| 110 |
-
{question}
|
| 111 |
-
"""
|
| 112 |
-
return call_llm(prompt, system_message="You are a helpful and accurate tutor that stays grounded in provided content.")
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from file_module import file_dashboard as file_viewer_dashboard
|
| 3 |
+
from file_module import file_study_guide_dashboard
|
|
|
|
| 4 |
|
|
|
|
|
|
|
|
|
|
| 5 |
|
| 6 |
+
def file_based_learning_router():
|
| 7 |
+
with gr.Column(scale=4, elem_id="main-column") as router:
|
| 8 |
+
view = gr.State("study") # default view
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
|
| 10 |
+
file_view_panel = gr.Column(visible=False)
|
| 11 |
+
file_study_panel = gr.Column(visible=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
|
| 13 |
+
with file_study_panel:
|
| 14 |
+
file_study_guide_dashboard()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
|
| 16 |
+
with file_view_panel:
|
| 17 |
+
file_viewer_dashboard()
|
| 18 |
|
| 19 |
+
def show_view(mode):
|
| 20 |
+
return (
|
| 21 |
+
gr.update(visible=(mode == "study")),
|
| 22 |
+
gr.update(visible=(mode == "file")),
|
| 23 |
+
mode
|
| 24 |
+
)
|
| 25 |
|
| 26 |
+
return router, show_view, file_study_panel, file_view_panel
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|