add features

.gitignore

@@ -1 +1,2 @@
-.env
+.env
+test.py

app.py

@@ -3,6 +3,7 @@ import requests
 from decouple import config
 import gradio as gr
 import spacy
+import re
 
 # for generating requirements.txt, use:
 # pip install pipreqs
@@ -10,7 +11,7 @@ import spacy
 
 RAPIDAPI_KEY = config('RAPIDAPI_KEY')
 
-nlp = gr.Blocks.load(name='models/victorialslocum/en_reciparse_model')
+ner = gr.Blocks.load(name='models/victorialslocum/en_reciparse_model')
 
 def get_recipe_data(recipe_url):
     rapid_api_url = "https://mycookbook-io1.p.rapidapi.com/recipes/rapidapi"
@@ -41,29 +42,64 @@ def get_recipe_data(recipe_url):
 
     return instructions, ingredients, recipe_title
 
-def get_ingredient_data(instructions, ingredients):
-    doc = nlp(instructions)
-    print(doc)
+def get_ingredient_data(ingredients):
+    nlp = spacy.load("en_core_web_sm")
+
+    doc = ner(ingredients)
     
     ingredients_list = []
 
-    # TODO: add lemmatizer
     for item in doc:
         if item[1] == 'INGREDIENT':
-            if item[0] not in ingredients_list:
-                ingredients_list.append(item[0])
+            item_lemma = ' ' .join([tok.lemma_ for tok in nlp(item[0])])
+            if item_lemma not in ingredients_list:
+                ingredients_list.append(item_lemma)
 
-    ingredients = ""
+    ingredients_out = ""
 
     for i in range(len(ingredients_list)):
-        ingredients += f'{i+1}. {ingredients_list[i]} \n'
+        ingredients_out += f'{i+1}. {ingredients_list[i]} \n'
+
+    return ingredients_out
+
+def get_instruction_data(instructions):
+    nlp = spacy.load("en_core_web_sm")
+
+    doc = ner(instructions)
+    
+    ingredients_list = []
+
+    for item in doc:
+        if item[1] == 'INGREDIENT':
+            item_lemma = ' ' .join([tok.lemma_ for tok in nlp(item[0])])
+            if item_lemma not in ingredients_list:
+                ingredients_list.append(item_lemma)
 
-    return ingredients
+    ingredients_out = ""
+
+    for i in range(len(ingredients_list)):
+        ingredients_out += f'{i+1}. {ingredients_list[i]} \n'
+
+    return ingredients_out
+
+def final_ingredients(inst_ingredients_list, ingr_ingredients_list):
+    instructions_guess = [re.sub(r'[0-9]+', '', item)[2:-1] for item in inst_ingredients_list.split('\n')if len(item) > 3]
+    ingredients_guess = [re.sub(r'[0-9]+', '', item)[2:-1] for item in ingr_ingredients_list.split('\n')if len(item) > 3]
+
+    final_guess_list = [item for item in instructions_guess if item in ingredients_guess]
+
+    final_guess = ""
+
+    for i in range(len(final_guess_list)):
+        final_guess += f'{i+1}. {final_guess_list[i]} \n'
+
+    return final_guess
 
 with gr.Blocks() as demo:
     gr.Markdown("# Reciparse Visualizer")
     gr.Markdown("### Powered by spaCy and Gradio")
     with gr.Box():
+        ingredients_list = gr.Variable()
         with gr.Column():
             gr.Markdown("## Recipe Info")
             with gr.Row():
@@ -78,9 +114,16 @@ with gr.Blocks() as demo:
         with gr.Column():
             gr.Markdown("## Ingredient Info")
             ingredient_btn = gr.Button("Get the list of ingredients")
-            ingredient_list = gr.Textbox(label="Ingredients in recipe")
+            final_btn = gr.Button("Create the final list")
+            with gr.Row():
+                ingr_ingredients_list = gr.Textbox(label="Ingredients in recipe based off of ingredients list")
+                inst_ingredients_list = gr.Textbox(label="Ingredients in recipe based off of instructions")
+                final_ingredients_list = gr.Textbox(label="Final ingredient guess")
+
 
     recipe_btn.click(fn=get_recipe_data, inputs=recipe_link, outputs=[instructions, ingredients, recipe_title])
-    ingredient_btn.click(fn=get_ingredient_data, inputs=[instructions, ingredients], outputs=[ingredient_list])
+    ingredient_btn.click(fn=get_ingredient_data, inputs=[ingredients], outputs=[ingr_ingredients_list])
+    ingredient_btn.click(fn=get_instruction_data, inputs=[instructions], outputs=[inst_ingredients_list])
+    final_btn.click(fn=final_ingredients, inputs=[inst_ingredients_list, ingr_ingredients_list], outputs=[final_ingredients_list])
 
 demo.launch()