interim.py

import os
from datetime import datetime
import streamlit as st
from patentwiz import preprocess_data, qa_agent

# Check if the API key is loaded
api_key = os.getenv("OPENAI_API_KEY")
if not api_key:
    st.error("OPENAI_API_KEY not found! Please set it in the environment variables or Hugging Face Secrets.")
    st.stop()

PROMPT = """
Task: Carefully review the given patent text and extract as much physical measurements information such as length/distance, mass/weight, time, temperature, Volume, area, speed, pressure, energy, power, electric current 
and voltage, frequency, force, acceleration, density, resistivity, magnetic field strength, and luminous intensity as much as possible. 
We are particularly interested in physical measurements including substance that was measured, Value of the measurement, and Unit of the measurement, and measurement type mentioned in the text. 
For each measurement, please provide the following details:
- The substance that was measured. (substance)
- The specific value or range that was measured. (Measured Value)
- The unit of the measurement, if provided. (Unit)
- The type of measurement being conducted (e.g., diameter, size, etc.) 
Format your response in a structured JSON-like format, as follows:
{"Content": [
    {
      "Measurement_substance": "substance",
      "Measured_value": "value",
      "Measured_unit": "unit",
      "measurement_type": "type"
    },
    // ... additional measurements, if present
  ]
}
If multiple measurements are present in the text, each should be listed as a separate object within the "Content" array.
Example: If the text includes the sentence, "The resulting BaCO3 had a crystallite size of between about 20 and 40 nm", the output should be:
{"Content": [
    {
      "Measurement_substance": "BaCO3",
      "Measured_value": "between about 20 and 40",
      "Measured_unit": "nm",
      "measurement_type": "crystallite size"
    }
  ]
}
Try to provide as complete and accurate information as possible. Print only the formatted JSON response.
"""

# Title and description
st.title("Technical Measurements Extractor for Patents")
st.write(
    "Analyze patents to extract physical measurements such as length, mass, time, and more. "
    "Provide a date to download patents, and analyze them using GPT models."
)

# User Input Section
st.header("Enter Details for Patent Analysis")
user_date_input = st.text_input("Enter a date in the format 'YYYY-MM-DD':", value="2024-06-16")

num_patents_to_analyze = st.number_input(
    "Number of patents to analyze:", min_value=1, value=1, step=1, help="Specify how many patents you want to analyze."
)

model_choice = st.selectbox(
    "Select a model for analysis:", ["gpt-3.5-turbo", "gpt-4"], help="Choose the OpenAI GPT model for the analysis."
)

logging_enabled = st.checkbox("Enable logging?", value=False, help="Toggle logging for debugging purposes.")

# Run Analysis Button
if st.button("Analyze Patents"):
    if not user_date_input:
        st.error("Please enter a valid date!")
    else:
        try:
            # Parse date input
            input_date = datetime.strptime(user_date_input, "%Y-%m-%d")
            year, month, day = input_date.year, input_date.month, input_date.day

            # Step 1: Download and preprocess patents
            with st.spinner("Downloading and extracting patents..."):
                saved_patent_names = preprocess_data.parse_and_save_patents(
                    year, month, day, logging_enabled
                )
                if not saved_patent_names:
                    st.error("No patents found for the given date.")
                    st.stop()
                st.success(f"{len(saved_patent_names)} patents found and processed!")

            # Step 2: Analyze patents using GPT
            random_patents = saved_patent_names[:num_patents_to_analyze]
            total_cost = 0
            results = []

            st.write("Starting patent analysis...")
            for i, patent_file in enumerate(random_patents):
                cost, output = qa_agent.call_QA_to_json(
                    PROMPT,
                    year,
                    month,
                    day,
                    saved_patent_names,
                    i,
                    logging_enabled,
                    model_choice,
                )
                total_cost += cost
                results.append(output)

            # Step 3: Display results
            st.write(f"**Total Cost:** ${total_cost:.4f}")
            st.write("### Analysis Results:")
            for idx, result in enumerate(results):
                st.subheader(f"Patent {idx + 1}")
                st.json(result)

        except ValueError as ve:
            st.error(f"Invalid date format: {ve}")
        except Exception as e:
            st.error(f"An unexpected error occurred: {e}")