Create Report_Writer.py
Browse files- pages/Report_Writer.py +148 -0
pages/Report_Writer.py
ADDED
@@ -0,0 +1,148 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
import os
|
3 |
+
import streamlit as st
|
4 |
+
from llama_index.core import Settings
|
5 |
+
from llama_index.core import VectorStoreIndex, Document
|
6 |
+
from llama_index.embeddings.gemini import GeminiEmbedding
|
7 |
+
from llama_index.llms.gemini import Gemini
|
8 |
+
from llama_index.core import DocumentSummaryIndex
|
9 |
+
import google.generativeai as genai
|
10 |
+
import os
|
11 |
+
import PyPDF2
|
12 |
+
|
13 |
+
# Set up Google API key
|
14 |
+
|
15 |
+
# Configure Google Gemini
|
16 |
+
Settings.embed_model = GeminiEmbedding(api_key=os.getenv("GOOGLE_API_KEY"), model_name="models/embedding-001")
|
17 |
+
Settings.llm = Gemini(api_key=os.getenv("GOOGLE_API_KEY"), temperature=0.8, model_name="models/gemini-pro")
|
18 |
+
llm = Gemini(api_key=os.getenv("GOOGLE_API_KEY"), temperature=0.1, model_name="models/gemini-pro")
|
19 |
+
|
20 |
+
# Load and index the input data
|
21 |
+
def load_data(document_text):
|
22 |
+
document = [Document(text=document_text)]
|
23 |
+
|
24 |
+
#index = VectorStoreIndex.from_documents([document])
|
25 |
+
index = DocumentSummaryIndex.from_documents(document)
|
26 |
+
return index
|
27 |
+
|
28 |
+
# Default report format template
|
29 |
+
DEFAULT_REPORT_FORMAT = """
|
30 |
+
Title Page
|
31 |
+
|
32 |
+
Includes the report title, author's name, and date.
|
33 |
+
|
34 |
+
Abstract
|
35 |
+
|
36 |
+
A concise summary of the report, covering the background, objectives, methodology, key findings, and conclusions.
|
37 |
+
|
38 |
+
Table of Contents
|
39 |
+
|
40 |
+
Lists sections and subsections with corresponding page numbers for easy navigation.
|
41 |
+
|
42 |
+
Introduction
|
43 |
+
|
44 |
+
Provides background information, defines the scope of the report, and states the objectives.
|
45 |
+
|
46 |
+
Literature Review
|
47 |
+
|
48 |
+
Reviews relevant literature and previous research related to the report topic.
|
49 |
+
|
50 |
+
Methodology/Approach
|
51 |
+
|
52 |
+
Details the methods used to gather data or conduct experiments, including design and analytical techniques.
|
53 |
+
|
54 |
+
Results and Discussion
|
55 |
+
|
56 |
+
Presents findings in a clear format, often using tables, figures, and charts, followed by a discussion interpreting these results.
|
57 |
+
|
58 |
+
Conclusions
|
59 |
+
|
60 |
+
Summarizes the main findings and their implications, often linking back to the report's objectives.
|
61 |
+
|
62 |
+
Recommendations
|
63 |
+
|
64 |
+
Suggests actions based on the findings, highlighting potential future work or improvements.
|
65 |
+
|
66 |
+
References
|
67 |
+
|
68 |
+
Lists all sources cited in the report, adhering to a specific referencing style.
|
69 |
+
|
70 |
+
Appendices
|
71 |
+
|
72 |
+
Contains supplementary material that supports the main text, such as raw data, detailed calculations, or additional figures.
|
73 |
+
|
74 |
+
"""
|
75 |
+
|
76 |
+
# Generate report
|
77 |
+
def generate_report(index, report_format, additional_info):
|
78 |
+
query_engine = index.as_query_engine()
|
79 |
+
|
80 |
+
if not report_format.strip():
|
81 |
+
report_format = DEFAULT_REPORT_FORMAT
|
82 |
+
st.info("Using default report format.")
|
83 |
+
|
84 |
+
response = query_engine.query(f"""
|
85 |
+
You are a professional report writer. Your task is to create a comprehensive report based on the entire document provided.
|
86 |
+
|
87 |
+
First, thoroughly analyze and summarize the entire document. Then, use the input text to create a well-structured report following the format below:
|
88 |
+
|
89 |
+
Report Format:
|
90 |
+
{report_format}
|
91 |
+
|
92 |
+
Additional Information:
|
93 |
+
{additional_info}
|
94 |
+
|
95 |
+
Even if the input is shallow, generate a report
|
96 |
+
Guidelines:
|
97 |
+
1. Ensure you comprehend and summarize the entire document before starting the report.
|
98 |
+
2. The report should be comprehensive, covering all major points from the document.
|
99 |
+
3. Adapt the provided format as necessary to best fit the content and context of the document.
|
100 |
+
4. Incorporate any additional information provided into the relevant sections of the report.
|
101 |
+
5. Use clear, professional language throughout the report.
|
102 |
+
6. Provide specific examples or data from the document to support your analysis and conclusions.
|
103 |
+
7. If the document contains technical information, explain it in a way that's accessible to a general audience.
|
104 |
+
|
105 |
+
Generate a thorough, well-structured report that captures the essence of the entire document.
|
106 |
+
""")
|
107 |
+
return response.response
|
108 |
+
|
109 |
+
# Streamlit app
|
110 |
+
def main():
|
111 |
+
st.title("AI Report Writer")
|
112 |
+
st.write("Upload your document and our AI will generate a comprehensive report based on its contents!")
|
113 |
+
|
114 |
+
# File uploader
|
115 |
+
uploaded_file = st.file_uploader("Choose a file (PDF or TXT)", type=["txt", "pdf"])
|
116 |
+
|
117 |
+
# Report format input
|
118 |
+
report_format = st.text_area("Enter the desired report format (optional)", height=150,
|
119 |
+
help="Leave blank to use a default template")
|
120 |
+
|
121 |
+
# Additional information input
|
122 |
+
additional_info = st.text_area("Enter any additional information or context for the report", height=100)
|
123 |
+
|
124 |
+
if uploaded_file is not None:
|
125 |
+
# Read file contents
|
126 |
+
if uploaded_file.type == "application/pdf":
|
127 |
+
pdf_reader = PyPDF2.PdfReader(uploaded_file)
|
128 |
+
document_text = ""
|
129 |
+
for page in pdf_reader.pages:
|
130 |
+
document_text += page.extract_text()
|
131 |
+
else:
|
132 |
+
document_text = uploaded_file.getvalue().decode("utf-8")
|
133 |
+
|
134 |
+
if st.button("Generate Report"):
|
135 |
+
st.write("Analyzing document and generating report...")
|
136 |
+
|
137 |
+
# Load data and generate report
|
138 |
+
doc_list = document_text.split(".")
|
139 |
+
index = load_data(document_text)
|
140 |
+
report = generate_report(index, report_format, additional_info)
|
141 |
+
|
142 |
+
st.write("## Generated Report")
|
143 |
+
st.write(report)
|
144 |
+
else:
|
145 |
+
st.warning("Please upload a file to generate a report.")
|
146 |
+
|
147 |
+
if __name__ == "__main__":
|
148 |
+
main()
|