Added app.py

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.pdf filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,169 @@
+import gradio as gr
+import requests
+import json
+from jiwer import cer, wer
+import re
+
+pdf_file_path = 'dummy.pdf'
+
+with open("page_transcriptions.json", encoding="utf-8") as f:
+    
+    data = json.load(f)
+
+def send_request(url):
+    
+    try:
+        
+        with open(pdf_file_path, 'rb') as pdf_file:
+            
+            files = {
+                'file': (
+                    'dummy.pdf',
+                    pdf_file,
+                    'application/pdf'
+                )
+            }
+        
+            response = requests.post(url, files=files)
+    
+    except Exception as e:
+        
+        return {"Error message: "f"Error occurred while sending request. Error message: {e}"}
+    
+    try:
+        
+        response_json = response.json()
+        
+    except Exception as e:
+        
+        return {
+            "Error message": e,
+            "Response": response.content
+            }
+    
+    if isinstance(response_json, list):
+        
+        for page in response_json:
+            
+            if isinstance(page, dict):
+            
+                if "page_number" not in page.keys() or "MD_text" not in page.keys():
+                    
+                    return {
+                        "Error message": "Response is not in desired structure. Desired structure: [{'page_number': 1, 'MD_text': 'Extracted text'}]",
+                        "Response": response_json
+                    }
+                    
+                if isinstance(page["page_number"], int) and isinstance(page["MD_text"], str):
+                    
+                    continue
+                
+                else:
+                    
+                    return {
+                        "Error message": "'page_number' should be integer and 'MD_text' should be string.",
+                        "Response": response_json
+                    }
+            
+            else:
+                
+                return {
+                    "Error message": "List should include only dictionaries.",
+                    "Response": response_json
+                }
+            
+        if len(response_json) != len(data):
+            
+            return {
+                "Error message": "The number of pages are not equal between transcription and ground truth.",
+                "Response": response_json
+            }
+            
+        final_metrics = []
+        total_reference = ""
+        total_hypothesis = ""
+        
+        for page in response_json:
+            
+            for transcription in data:
+                
+                if page["page_number"] == transcription["page_number"]:
+                    
+                    reference = transcription['MD_text'].strip()
+                    hypothesis = page['MD_text'].strip()
+                    
+                    reference = reference.lower()
+                    hypothesis = hypothesis.lower()
+                    
+                    reference = reference.replace("\n", " ")
+                    hypothesis = hypothesis.replace("\n", " ")
+                    
+                    reference = re.sub(r'\s+', ' ', reference)
+                    hypothesis = re.sub(r'\s+', ' ', hypothesis)
+                    
+                    total_reference += reference
+                    total_reference += " "
+                    total_hypothesis += hypothesis
+                    total_hypothesis += " "
+                    
+                    cer_value = max(1 - cer(reference, hypothesis), 0)
+                    wer_value = max(1 - wer(reference, hypothesis), 0)
+                    
+                    final_metrics.append({"page_number": page["page_number"], "Character Success Rate (CSR)": round(cer_value, 4), "Word Success Rate (WSR)": round(wer_value, 4), "MD_text_used_for_metrics": hypothesis, "Ground_Truth_used_for_metrics": reference})
+            
+        global_cer = max(1 - cer(total_reference.strip(), total_hypothesis.strip()), 0)
+        global_wer = max(1 - wer(total_reference.strip(), total_hypothesis.strip()), 0)
+        
+        final_metrics.append({"Global CSR": global_cer, "Global WSR": global_wer, "MD_text_used_for_metrics": total_hypothesis.strip(), "Ground_Truth_used_for_metrics": total_reference.strip()})
+        
+        return final_metrics
+
+    else:
+        
+        return {
+            "Error message": "Response should be list of dictionaries.",
+            "Response": response_json
+        }
+
+with gr.Blocks() as demo:
+    
+    gr.Markdown(
+        """
+        # OCR Endpoint Response Validator and Quality Checker
+        
+        Character Success Rate (CSR) and Word Success Rate (WSR) are metrics that will be provided for each page and total.
+        They are calculated by simply subtracting CER and WER from 1 respectively.
+        If CER or WER is > 1, CSR or WSR is considered as 0. 
+        
+        Enter your endpoint below and click **Send** to get the result.
+
+        Format:
+        ```http://<host>/<endpoint>```
+        """
+    )
+    
+    output = gr.JSON(
+        label="Output"
+    )
+
+    input_box = gr.Textbox(
+        label="Input",
+        lines=1,
+        placeholder="Type your text here..."
+    )
+
+    send_btn = gr.Button("Send")
+
+    send_btn.click(
+        fn=send_request,
+        inputs=input_box,
+        outputs=output
+    )
+
+    input_box.submit(
+        fn=send_request,
+        inputs=input_box,
+        outputs=output
+    )
+
+demo.launch()