added files

InferenceServer.py

@@ -0,0 +1,17 @@
+from rpunct import RestorePuncts
+print("Loading Model...")
+rpunct = RestorePuncts()
+
+from fastapi import FastAPI
+
+app = FastAPI()
+print("Models loaded !")
+
+@app.get("/")
+def read_root():
+    return {"Homepage!"}
+
+@app.get("/{restore}")
+def get_correction(input_sentence):
+    '''Returns sentence with correct punctuations and case'''
+    return {"corrected_sentence": rpunct.punctuate(input_sentence, lang="en")}

app.py

@@ -0,0 +1,88 @@
+import streamlit as st
+from multiprocessing import Process
+import json
+import requests
+import time
+import os
+
+
+def start_server():   
+    '''Helper to start to service through Unicorn '''
+    os.system("uvicorn InferenceServer:app --port 8080 --host 0.0.0.0 --workers 2")
+
+def load_models():
+    '''One time loading/ Init of models and starting server as a seperate process'''
+    if not is_port_in_use(8080):
+        with st.spinner(text="Loading model, please wait..."):
+            proc = Process(target=start_server, args=(), daemon=True)
+            proc.start()
+            while not is_port_in_use(8080):
+                time.sleep(1)
+            st.success("Model server started.")
+    else:
+        st.success("Model server already running...")
+    st.session_state['models_loaded'] = True
+
+def is_port_in_use(port):
+    '''Helper to check if service already running'''
+    import socket
+    with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+        return s.connect_ex(('0.0.0.0', port)) == 0
+
+if 'models_loaded' not in st.session_state:
+    st.session_state['models_loaded'] = False
+
+def get_correction(input_text):
+    '''Invokes the inference service'''
+    correct_request = "http://0.0.0.0:8080/restore?input_sentence="+input_text
+    correct_response = requests.get(correct_request)
+    correct_json = json.loads(correct_response.text)
+    corrected_sentence = correct_json["corrected_sentence"]
+    result = diff_strings(corrected_sentence,input_text)
+    st.markdown(f'##### Corrected text:')
+    st.write('')
+    st.markdown(result, unsafe_allow_html=True)
+
+def diff_strings(output_text, input_text):
+    '''Highlights corrections'''
+    c_text = ""
+    for x in output_text.split(" "):
+      if x in input_text.split(" "):
+        c_text = c_text + x + " "
+      else:
+        c_text = c_text + '<span style="font-weight:bold; color:rgb(142, 208, 129);">' + x + '</span>' + " "
+    return c_text   
+        
+if __name__ == "__main__":
+    
+        st.title('Rpunct')
+        st.subheader('For Punctuation and Upper Case restoration')
+        st.markdown("Spaces for [felflare/bert-restore-punctuation](https://huggingface.co/felflare/bert-restore-punctuation) using [Fork with CPU support](https://github.com/anuragshas/rpunct) | [Original repo](https://github.com/Felflare/rpunct)", unsafe_allow_html=True)
+        st.markdown("Model restores the following punctuations -- [! ? . , - : ; ' ] and also the upper-casing of words.")
+        st.markdown("Integrate with just few lines of code", unsafe_allow_html=True)
+        st.markdown("""
+                    ```python 
+                    from rpunct import RestorePuncts
+                    rpunct = RestorePuncts()
+                    rpunct.punctuate('''my name is clara and i live in berkeley california''')
+                    ```    
+                    """)
+        examples = [
+                    "my name is clara and i live in berkeley california",
+                    "in 2018 cornell researchers built a high-powered detector",
+                    "lorem ipsum has been the industrys standard dummy text ever since the 1500s when an unknown printer took a galley of type and scrambled it to make a type specimen book"
+                    ]
+        if not st.session_state['models_loaded']:
+            load_models()
+        
+        input_text = st.selectbox(
+        label="Choose an example",
+        options=examples
+        )
+        st.write("(or)")
+        input_text = st.text_input(
+            label="Input sentence",
+            value=input_text
+        )
+        if input_text.strip(): 
+            get_correction(input_text)

requirements.txt

@@ -0,0 +1,3 @@
+git+https://github.com/anuragshas/rpunct.git
+fastapi
+uvicorn