Initial Upload

.gitignore

@@ -0,0 +1,7 @@
+/__pycache__
+
+/src/base/__pycache__
+/src/book_rec/__pycache__
+/src/cat_and_dog/__pycache__
+/src/movie_rec/__pycache__
+/src/movie_reviews/__pycache__

Dockerfile

@@ -0,0 +1,14 @@
+# read the doc: https://huggingface.co/docs/hub/spaces-sdks-docker
+# you will also find guides on how best to write your Dockerfile
+
+FROM python:3.9
+
+WORKDIR /code
+
+COPY ./requirements.txt /code/requirements.txt
+
+RUN pip install --no-cache-dir --upgrade -r /code/requirements.txt
+
+COPY . .
+
+CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]

main.py

@@ -0,0 +1,50 @@
+from fastapi import FastAPI
+
+# Importing Models and Schemas
+from src.movie_reviews.main import movie_reviews, Schema as MovieReviewsSchema
+from src.cat_and_dog.main import cat_and_dog, Schema as CatAndDogSchema
+from src.book_rec.main import book_rec, Schema as BookRecSchema
+from src.movie_rec.main import movie_rec, Schema as MovieRecSchema
+
+# Initializing App
+app = FastAPI()
+
+# Allowing Cross Origins
+from fastapi.middleware.cors import CORSMiddleware
+origins = [
+    "http://localhost.tiangolo.com",
+    "https://localhost.tiangolo.com",
+    "http://localhost",
+    "http://localhost:8080",
+]
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+print(" ........... App Started ........... ")
+
+# Endpoints
+
+@app.get("/")
+def index():
+	return "Welcome to the API of PyModelsAI"
+
+@app.post("/movie_reviews")
+def endpoint_movie_reviews(req: MovieReviewsSchema):
+    return movie_reviews(req)
+
+@app.post("/cat_and_dog")
+def endpoint_cat_and_dog(req: CatAndDogSchema):
+    return cat_and_dog(req)
+
+@app.post("/book_rec")
+def endpoint_book_rec(req: BookRecSchema):
+    return book_rec(req)
+
+@app.post("/movie_rec")
+def endpoint_movie_rec(req: MovieRecSchema):
+    return movie_rec(req)

requirements.txt

@@ -0,0 +1,8 @@
+fastapi
+uvicorn
+joblib
+scikit-learn
+numpy
+tensorflow-cpu
+keras
+Pillow

src/base/recommender.py

@@ -0,0 +1,39 @@
+import json
+from pydantic import BaseModel
+
+# SCHEMA
+class Schema(BaseModel):
+	title: str
+	n: int = 5
+
+# Request Handler
+def recommender(req, data):
+	title = req.title
+	n = req.n
+	output = predict(title, n, data)
+	return output
+
+def predict(title, n, data):
+	index = data['titles'].index(title)
+	recs = data['recs'][index][:n]
+	output = []
+
+	for rec in [[index, 0]] + recs:
+		i, score = rec
+		new_rec = {
+			"title": data['titles'][i],
+			"score": score,
+			"img": None,
+			"info": None
+		}
+		if (data['imgs']):
+			new_rec['img'] = data['imgs'][i]
+		if (data['infos']):
+			new_rec['info'] = data['infos'][i],
+		
+		if type(new_rec['info']) == tuple:
+			new_rec['info'] = new_rec['info'][0]
+
+		output.append(new_rec)
+
+	return output

src/book_rec/main.py

@@ -0,0 +1,12 @@
+from src.base.recommender import recommender
+from src.base.recommender import Schema
+
+import json
+data_path = "./src/book_rec/data.json"
+with open(data_path, 'rb') as f:
+	data = json.load(f)
+
+# Request Handler
+def book_rec(req):
+	# Sending to Base
+	return recommender(req, data)

src/cat_and_dog/main.py

@@ -0,0 +1,40 @@
+import numpy as np
+import tensorflow as tf
+import io, base64, requests
+from pydantic import BaseModel
+
+# SCHEMA
+class Schema(BaseModel):
+	resized_img_base64:str = None,
+	img_url:str = None
+
+# Request Handler
+def cat_and_dog(req):
+	resized_img_base64 = req.resized_img_base64
+	img_url = req.img_url
+	output = predict(resized_img_base64, img_url)
+	return output
+
+model_path = "./src/cat_and_dog/model_85.9.h5"
+"""
+This Model has an accuracy of 85.9%
+"""
+
+model = tf.keras.models.load_model(model_path)
+
+def predict(img_data, img_url):
+	if img_url == None:
+		content = img_data.replace(" ", "+")
+		converted = bytes(content, "utf-8")
+		img = base64.decodebytes(converted)
+	else:
+		img = requests.get(img_url).content
+
+	img = io.BytesIO(img)
+	img = tf.keras.preprocessing.image.load_img(img, target_size=model.input_shape[1:])
+	img = np.array(img)
+	img = img.reshape(1, *img.shape)
+	img = img / 255.
+	pred = model.predict(img)[0, 0]
+	pred = float(pred)
+	return [round(1-pred, 3), round(pred, 3)]

src/cat_and_dog/model_85.9.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84e57acb8a4be38f5aed70b36a2c4b6f18aa87c755df651121b6bca1097556e0
+size 440216

src/movie_rec/main.py

@@ -0,0 +1,12 @@
+from src.base.recommender import recommender
+from src.base.recommender import Schema
+
+import json
+data_path = "./src/movie_rec/data.json"
+with open(data_path, 'rb') as f:
+	data = json.load(f)
+
+# Request Handler
+def movie_rec(req):
+	# Sending to Base
+	return recommender(req, data)

src/movie_reviews/main.py

@@ -0,0 +1,70 @@
+import string
+import re
+import joblib
+from pydantic import BaseModel
+
+# SCHEMA
+class Schema(BaseModel):
+	text: str
+        
+# Request Handler
+def movie_reviews(req):
+	text = req.text
+	output = predict(text)
+	return output
+
+# PREPROCESSING
+punc = string.punctuation
+abbv = {
+    "AFAIK":"as far as I know",
+	"IMO":	"in my opinion",
+	"IMHO":	"in my humble opinion",
+	"LGTM":	"look good to me",
+	"AKA":	"also know as",
+	"ASAP":	"as sone as possible",
+	"BTW":	"by the way",
+	"FAQ":	"frequently asked questions",
+	"DIY":	"do it yourself",
+	"DM":	"direct message",
+	"FYI":	"for your information",
+	"IC":	"i see",
+	"IOW":	"in other words",
+	"IIRC":	"If I Remember Correctly",
+	"icymi":"In case you missed it",
+	"CUZ":	"because",
+	"COS":	"because",
+	"nv":	"nevermind",
+	"PLZ":	"please",
+}
+html_pattern = re.compile('<.*?>')
+urls_pattern = re.compile(r'https?://\S+|www\.\S+')
+emoji_pattern = re.compile("["
+	u"\U0001F600-\U0001F64F"  # emoticons
+	u"\U0001F300-\U0001F5FF"  # symbols & pictographs
+	u"\U0001F680-\U0001F6FF"  # transport & map symbols
+	u"\U0001F1E0-\U0001F1FF"  # flags (iOS)
+"]+", flags=re.UNICODE)
+
+# PIPELINE
+pipeline = joblib.load("./src/movie_reviews/pipeline.pkl")
+
+def predict(text):
+	cleaned = preprocess(text)
+	pred = pipeline.predict([cleaned])[0]
+	output = [0, 0]
+	output[pred] = 0.8
+	output[1-pred] = 0.2
+	return output
+
+def preprocess(text):
+    text = text.lower()	# Lowercase
+    text = html_pattern.sub(r'', text)	# HTML Tags
+    text = urls_pattern.sub(r'', text)	# urls
+    text = text.translate(str.maketrans("", "", punc))	# punctuations
+    text = emoji_pattern.sub(r'', text)	# Emojis
+    new_text = []
+    for word in text.split(" "):
+        word = abbv.get(word.upper(), word)	# abbreviations
+        new_text.append(word)
+    text = " ".join(new_text)
+    return text

src/movie_reviews/pipeline.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8cace6b6bc66a701a90b8c4e19f6ae96e87b57c4be47ad4accf772380e3536a3
+size 2182254