instead of returning the image bytes to the api user, we are moving towards hosting them on the backend but we need to be careful moving forwards and make sure that we delete the images after use

app.py

@@ -12,10 +12,7 @@ app = Flask(
     template_folder="templates",
 )
 
-CORS(
-    app,
-    origins="*",
-)
+CORS(app, origins="*")
 
 
 @app.route("/", methods=["GET"])

cluster/clusterer.py

@@ -5,7 +5,7 @@ from typing import Callable
 @dataclass
 class Clusterer:
     cluster_func: Callable
-    plot = None
+    plot_key = None
 
     def eval(
         self,

cluster/kmeans.py

@@ -74,5 +74,5 @@ class Kmeans(Clusterer):
             "k": self.k,
             "max_iter": self.max_iter,
             "clusters": cluster_data,
-            "plot": self.plot,
+            "plot_key": self.plot_key,
         }

cluster/plot.py

@@ -1,14 +1,17 @@
-import io
-import base64
-import numpy as np
 import matplotlib
-import matplotlib.pyplot as plt 
+import matplotlib.pyplot as plt
 import seaborn as sns
+from plt_id import generate_image_key
+import os
 
 
 matplotlib.use("Agg")
 sns.set()
 
+# Replace with the desired upload folder path
+UPLOAD_FOLDER = '/path/to/upload/folder'
+
+
 def plot(clusterer, X) -> None:
     cluster_data = clusterer.to_dict(X)["clusters"]
     # plot the clusters and data points
@@ -32,11 +35,12 @@ def plot(clusterer, X) -> None:
     ax.set_title("K-means Clustering")
     ax.set_ylabel("Normalized Petal Length (cm)")
     ax.set_xlabel("Normalized Petal Length (cm)")
-    clusterer.plot = plt_bytes(fig)
 
+    image_key = generate_image_key()  # Generate a unique key for the image
 
-def plt_bytes(fig):
-    buf = io.BytesIO()
-    fig.savefig(buf, format="png")
+    # Save the plot as an image file with the key in the filename
+    plot_filename = os.path.join(UPLOAD_FOLDER, f"{image_key}.png")
+    fig.savefig(plot_filename, format="png")
     plt.close(fig)
-    return base64.b64encode(buf.getvalue()).decode("utf-8")
+
+    clusterer.plot_key = image_key

example/neural_network.py

@@ -4,10 +4,9 @@ import requests
 import json
 
 
-ENDPOINT: str = "http://127.0.0.1:5000/"
+ENDPOINT: str = "http://127.0.0.1:5000/neural-network"
 
 request_params = {
-    "algorithm": "neural-network",
     "arguments": {
         "epochs": 100,
         "activation_func": "tanh",

neural_network/neural_network.py

@@ -19,7 +19,7 @@ class NeuralNetwork:
     loss_history: list = field(
         default_factory=lambda: [],
     )
-    plot = None
+    plot_key = None
 
     def predict(self, x: np.array) -> np.array:
         n1 = self.compute_node(x, self.w1, self.b1, self.activation_func)
@@ -54,5 +54,5 @@ class NeuralNetwork:
             # not returning this because we are making our own
             # plots and this can be a lot of data
             # "loss_history": self.loss_history,
-            "plot": self.plot,
+            "plot_id": self.plot_id,
         }

neural_network/plot.py

@@ -1,12 +1,15 @@
 import numpy as np
-import base64
-import io
 import seaborn as sns
 import matplotlib
 import matplotlib.pyplot as plt
+from plt_id import generate_image_key
+import os
 
 matplotlib.use("Agg")
 
+UPLOAD_FOLDER = "/plots"
+
+
 def plot(model) -> None:
     sns.set()
     fig, ax = plt.subplots()
@@ -18,8 +21,11 @@ def plot(model) -> None:
     plt.ylabel("Loss")
     plt.xlabel("Epoch")
     plt.title("Loss / Epoch")
-    buf = io.BytesIO() 
-    fig.savefig(buf, format="png")
+
+    image_key = generate_image_key()
+
+    plot_filename = os.path.join(UPLOAD_FOLDER, f"{image_key}.png")
+    fig.savefig(plot_filename, format="png")
     plt.close(fig)
-    plot_data = base64.b64encode(buf.getvalue()).decode("utf-8")
-    model.plot = plot_data
+
+    model.plot_key = image_key

plt_id.py

@@ -0,0 +1,5 @@
+import uuid
+
+
+def generate_image_key() -> str:
+    return str(uuid.uuid4())