added python poc

frontend/src/models/constants.ts

@@ -0,0 +1,2 @@
+export const dotnetApiUrl = "https://cf-contract-analysis-mvp.azurewebsites.net";
+// export const dotnetApiUrl = "https://localhost:7254";

frontend/src/pages/ContractAnalysis.tsx

@@ -1,6 +1,7 @@
 import React, { useState } from "react";
 import axios, { AxiosResponse } from "axios";
 import ContractResults from "../components/ContractResults";
+import { dotnetApiUrl } from "../models/constants";
 
 const ContractAnalysis = () => {
   const [loading, setLoading] = useState<boolean>(false);
@@ -31,7 +32,7 @@ const ContractAnalysis = () => {
     formData.append("file", file);
 
     axios
-      .post("https://cf-contract-analysis-mvp.azurewebsites.net/api/azure/analyze-contract", formData, {
+      .post(`${dotnetApiUrl}/api/azure/analyze-contract`, formData, {
         headers: {
           "Content-Type": "multipart/form-data",
           "Access-Control-Allow-Origin": "https://cf-contract-analysis-mvp.azurewebsites.net",

frontend/src/pages/DocumentAnalysis.tsx

@@ -1,6 +1,7 @@
 import React, { useState } from "react";
 import axios, { AxiosResponse } from "axios";
 import DocumentResults from "../components/DocumentResults";
+import { dotnetApiUrl } from "../models/constants";
 
 const DocumentAnalysis = () => {
   const [loading, setLoading] = useState<boolean>(false);
@@ -31,7 +32,7 @@ const DocumentAnalysis = () => {
     formData.append("file", file);
 
     axios
-      .post("https://cf-contract-analysis-mvp.azurewebsites.net/api/azure/analyze-document", formData, {
+      .post(`${dotnetApiUrl}/api/azure/analyze-document`, formData, {
         headers: {
           "Content-Type": "multipart/form-data",
         },

frontend/src/pages/OpenAiAnalysis.tsx

@@ -1,6 +1,7 @@
 import React, { useState } from "react";
 import axios, { AxiosResponse } from "axios";
 import { CopyBlock, dracula } from "react-code-blocks";
+import { dotnetApiUrl } from "../models/constants";
 
 const OpenAiAnalysis = () => {
   const [loading, setLoading] = useState<boolean>(false);
@@ -31,7 +32,7 @@ const OpenAiAnalysis = () => {
     formData.append("file", file);
 
     axios
-      .post("https://cf-contract-analysis-mvp.azurewebsites.net/api/open-ai/analyze-contract", formData, {
+      .post(`${dotnetApiUrl}/api/open-ai/analyze-contract`, formData, {
         headers: {
           "Content-Type": "multipart/form-data",
         },

python-poc/main.py

@@ -0,0 +1,62 @@
+import os
+from dotenv import load_dotenv
+from azure.core.credentials import AzureKeyCredential
+from azure.ai.formrecognizer import DocumentAnalysisClient
+from openai import OpenAI
+
+from prompt import system_prompt
+
+load_dotenv()
+azure_key = os.getenv("AZURE_KEY")
+azure_service_endpoint = os.getenv("AZURE_SERVICE_ENDPOINT")
+open_ai_org_id = os.getenv("OPEN_AI_ORG_ID")
+
+document_analysis_client = DocumentAnalysisClient(
+    endpoint=azure_service_endpoint, credential=AzureKeyCredential(azure_key)
+)
+
+client = OpenAI()
+
+
+def analyze_contract(path_to_file):
+    print("Uploading to Azure Document AI...")
+    print("________________________________________")
+
+    with open(path_to_file, "rb") as f:
+        poller = document_analysis_client.begin_analyze_document(
+            "prebuilt-read", document=f
+        )
+
+    result = poller.result()
+
+    print("File read by Azure Document AI")
+    print("Uploading file content to Open AI API...")
+    print("________________________________________")
+
+    messages = system_prompt
+
+    messages.append({"role": "user", "content": "Upload Starting Now."})
+    for paragraph in result.paragraphs:
+        messages.append({"role": "user", "content": paragraph.content})
+
+    messages.append({"role": "user", "content": "Please Start Summary Results."})
+
+    response = client.chat.completions.create(model="gpt-3.5-turbo", messages=messages)
+
+    print("Contract Summary Results:")
+    print("________________________________________")
+
+    print(response.choices[0].message.content)
+
+
+if __name__ == "__main__":
+    analyze_contract(
+        "/Users/kaden/dev/cf/sample-contract-files/AmendmentToContract.pdf"
+    )
+    # analyze_contract(
+    #     "/Users/kaden/dev/cf/sample-contract-files/FarmAndRanchContract.pdf"
+    # )
+    # analyze_contract(
+    #     "/Users/kaden/dev/cf/sample-contract-files/Sample_Utah_Real_Estate_Contract.pdf"
+    # )
+    # analyze_contract("/Users/kaden/dev/cf/sample-contract-files/SimpleContract.pdf")

python-poc/parse_attempts.py

@@ -0,0 +1,17 @@
+import PyPDF2
+
+
+def try_parse_pdf(path_to_file):
+    pdfFileObj = open(path_to_file, "rb")
+
+    pdfReader = PyPDF2.PdfReader(pdfFileObj)
+    print(len(pdfReader.pages))
+
+    pageObj = pdfReader.pages[0]
+    print(pageObj.extract_text())
+
+    pdfFileObj.close()
+
+
+if __name__ == "__main__":
+    try_parse_pdf("/Users/kaden/dev/cf/sample-contract-files/AmendmentToContract.pdf")

python-poc/prompt.py

@@ -0,0 +1,26 @@
+system_prompt = [
+    {
+        "role": "system",
+        "content": "You are a document research assistant that reads long files and summarizes them.",
+    },
+    {
+        "role": "system",
+        "content": "The conversation will start by the user saying 'Upload Starting Now.' The user will then provide you with document paragraphs.",
+    },
+    {
+        "role": "system",
+        "content": "You will not answer until the user has asked, 'Please Start Summary Results.'",
+    },
+    {
+        "role": "system",
+        "content": "You will then provide the most accurate details about the current information given to you.",
+    },
+    {
+        "role": "system",
+        "content": "You will start by listing in bullet points the following information: Buyer Name (if applicable), Seller Name (if applicable), Contract Amount (if applicable), Contract Key Dates, and Contract Property Address (if applicable).",
+    },
+    {
+        "role": "system",
+        "content": "After providing that information, you will give a summary of all other major details in paragraph form.",
+    },
+]