Update app.py

app.py

@@ -55,6 +55,53 @@ def unzip_and_find_jpgs(file_path):
                     jpg_files.append(full_path)
     return jpg_files
 
+def get_random_pairs_list(shots, num_pairs=2):
+    keys = random.sample(list(shots.keys()), num_pairs)
+    return [(key, shots[key]) for key in keys]
+
+def construct_prompt(input_text, random_pairs):
+    # Example setup based on your specified format
+    prompt = \
+    """    
+    Follow the examples below. Your response should contain only JSON. If you
+    encounter two dates in an input, prefer the earliest. If the answer is not
+    exact, try your best, but do not use excess wording. If you are completely
+    unsure or there is no answer, insert UNKNOWN.
+
+    Input 1:
+    {random_pairs[0][0]}
+
+    Output 1:
+    {{"Collector":"{random_pairs[0][1]['Collector']}","Location":"{random_pairs[0][1]['Location']}","Taxon":"{random_pairs[0][1]['Taxon']}","Date":"{random_pairs[0][1]['Date']}"}}
+
+    Input 2:
+    {random_pairs[1][0]}
+
+    Output 2:
+    {{"Collector":"{random_pairs[1][1]['Collector']}","Location":"{random_pairs[1][1]['Location']}","Taxon":"{random_pairs[1][1]['Taxon']}","Date":"{random_pairs[1][1]['Date']}"}}
+
+    Input 3:
+    {input_text}
+    Output 3:
+    """
+    return prompt
+
+def process_responses(responses):
+    structured_responses = []
+    for response in responses:
+        try:
+            # Assuming response is a string of JSON data
+            parsed_json = json.loads(response.text)
+            structured_responses.append(parsed_json)
+        except json.JSONDecodeError:
+            structured_responses.append({
+                "Collector": "UNKNOWN",
+                "Location": "UNKNOWN",
+                "Taxon": "UNKNOWN",
+                "Date": "UNKNOWN"
+            })
+    return structured_responses
+
 def process_images(uploaded_file):
     global results_df
     results_df = results_df.iloc[0:0]  # Clear the DataFrame if re-running this cell
@@ -75,11 +122,58 @@ def process_images(uploaded_file):
                 "Translated Text": translated_text
             }])
             results_df = pd.concat([results_df, new_row], ignore_index=True)
+
+        # Configure the generative AI model
+        genai.configure(api_key='AIzaSyB9iHlqAgz5TEF36Kg_fJLJvoIDCJkqwJI')
+        model = genai.GenerativeModel('gemini-pro')
+
+        # Prepare data for few-shot learning
+        shots = \
+        {
+            "Chinese National Herbarium (PE) Plants of Xizang CHINA, Xizang, Lhoka City, Lhozhag County, Lhakang Town, Kharchhu Gompa vicinity 28°5'37.15"N, 91°7'24.74"E; 3934 m Herbs. Slopes near roadsides. PE-Xizang Expedition #PE6663 NCIL 14 September 2017 N° 2581259 TIBET PE CHINESE NATIONAL HERBARIUM (PE) 02334125 #PE6663 COMPOSITAE Aster albescens (DC.) Hand.-Mazz. A: it (Guo-Jin ZHANG) 01 April 2018"\
+            :{"Collector":"Guo-Jin, Zhang",
+              "Location":"Xizang, Tibet, China, Lhoka City, Lhozhag County, Lhakang Town, near Kharchhu Gompa",
+              "Taxon":"Aster albescens (DC.) Hand.-Mazz., Compositae (Asteraceae) family",
+              "Date":"14 September 2017"
+            },
+
+            "PE-Xizang Expedition #PE6673 9 NSIT Chinese National Herbarium (PE) Plants of Xizang CHINA, Xizang, Lhoka City, Lhozhag County, Lhakang Town, Kharchhu Gompa vicinity 28°5'37.15"N, 91°7'24.74"E; 3934 m Herbs. Slopes near roadsides. PE-Xizang Expedition #PE6673 9 NSIT Chinese National Herbarium (PE) Plants of Xizang CHINA, Xizang, Lhoka City, Lhozhag County, Lhakang Town, Kharchhu Gompa vicinity 28°5'37.15"N, 91°7'24.74"E; 3934 m Herbs. Slopes near roadsides. PE-Xizang Expedition #PE6673 9 NSIT Chinese National Herbarium (PE) Plants of Xizang Spiral Leaf Green 17 May 2018"
+            :{"Collector":"UNKNOWN",
+              "Location":"Xizang, Tibet, China, Lhoka City, Lhozhag County, Lhakang Town, near Kharchhu Gompa",
+              "Taxon":"Spiral Leaf Green",
+              "Date":"17 May 2018"
+            },
+
+            "Honey Plants Research Institute of the Chinese Academy of Agricultural Sciences Collection No.: 13687. May 7, 1993 Habitat Roadside Altitude: 1600 * Characters Shrub No. Herbarium of the Institute of Botany, Chinese Academy of Sciences Collector 3687 Scientific Name Height: m (cm) Diameter at breast height m (cm) Flower: White Fruit: Notes Blooming period: from January to July Honey: Scientific Name: Rosa Sericea Lindl. Appendix: Collector: cm 1 2 3 4 25 CHINESE NATIONAL HERBARUM ( 01833954 No 1479566 * Herbarium of the Institute of Botany, Chinese Academy of Sciences Sichuan SZECHUAN DET. Rosa sercea Lindl. var. Various Zhi 2009-02-16"
+            :{"Collector":"UNKNOWN",
+              "Location":"Sichuan, China",
+              "Taxon":"Rosa sericea Lindl., with possible variant identification as 'var. Various Zhi'",
+              "Date":"7 May 1993",
+            },
+        }
+
+        responses = []
+        for input_text in results_df["Translated Text"]:
+            random_pairs = get_random_pairs_list(shots)
+            prompt = construct_prompt(input_text, random_pairs)
+            response = model.generate_content(prompt)
+            responses.append(response)
+
+        # Processing responses
+        json_responses = process_responses(responses)
+        results_df = pd.concat([results_df, pd.DataFrame(json_responses)], axis=1)
+
     except Exception as e:
         return f"An error occurred: {str(e)}"
 
     return results_df.to_html()
 
+css = """
+    body { font-family: Arial, sans-serif; }
+    .input-container { width: 95%; margin: auto; }
+    .output-container { width: 95%; margin: auto; }
+"""
+
 interface = gr.Interface(
     fn=process_images,
     inputs="file",
@@ -89,4 +183,4 @@ interface = gr.Interface(
 )
 
 if __name__ == "__main__":
-    interface.launch(debug=True)
+    interface.launch(share=True)