Spaces:
Runtime error
Runtime error
Merge branch 'seperate_utils'
Browse files- app.py +52 -106
- openai_api.py +30 -0
- utils.py +35 -0
app.py
CHANGED
|
@@ -1,114 +1,46 @@
|
|
| 1 |
-
import gradio as gr
|
| 2 |
from easyocr import Reader
|
| 3 |
-
|
| 4 |
-
import io
|
| 5 |
-
import json
|
| 6 |
-
import csv
|
| 7 |
import openai
|
| 8 |
import ast
|
| 9 |
import os
|
| 10 |
-
from deta import Deta
|
| 11 |
-
|
| 12 |
|
| 13 |
-
|
| 14 |
-
import
|
| 15 |
-
import json
|
| 16 |
|
| 17 |
-
import os
|
| 18 |
-
import openai
|
| 19 |
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
class OpenAI_API:
|
| 23 |
-
def __init__(self):
|
| 24 |
-
self.openai_api_key = ''
|
| 25 |
-
|
| 26 |
-
def single_request(self, address_text):
|
| 27 |
-
|
| 28 |
-
openai.api_type = "azure"
|
| 29 |
-
openai.api_base = "https://damlaopenai.openai.azure.com/"
|
| 30 |
-
openai.api_version = "2022-12-01"
|
| 31 |
-
openai.api_key = os.getenv("API_KEY")
|
| 32 |
-
|
| 33 |
-
response = openai.Completion.create(
|
| 34 |
-
engine="Davinci-003",
|
| 35 |
-
prompt=address_text,
|
| 36 |
-
temperature=0.,#9,
|
| 37 |
-
max_tokens=300,
|
| 38 |
-
top_p=1.0,
|
| 39 |
-
# n=1,
|
| 40 |
-
# logprobs=0,
|
| 41 |
-
# echo=False,
|
| 42 |
-
# stop=None,
|
| 43 |
-
frequency_penalty=0,
|
| 44 |
-
presence_penalty=0,
|
| 45 |
-
stop=["\n"],
|
| 46 |
-
best_of=1)
|
| 47 |
-
|
| 48 |
-
return response
|
| 49 |
-
|
| 50 |
-
########################
|
| 51 |
-
|
| 52 |
-
openai.api_key = os.getenv('API_KEY')
|
| 53 |
reader = Reader(["tr"])
|
| 54 |
|
| 55 |
|
| 56 |
-
def get_parsed_address(input_img):
|
| 57 |
-
|
| 58 |
-
address_full_text = get_text(input_img)
|
| 59 |
-
return openai_response(address_full_text)
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
def preprocess_img(inp_image):
|
| 63 |
-
gray = cv2.cvtColor(inp_image, cv2.COLOR_BGR2GRAY)
|
| 64 |
-
gray_img = cv2.bitwise_not(gray)
|
| 65 |
-
return gray_img
|
| 66 |
-
|
| 67 |
-
|
| 68 |
def get_text(input_img):
|
| 69 |
result = reader.readtext(input_img, detail=0)
|
| 70 |
return " ".join(result)
|
| 71 |
|
| 72 |
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
with open("adress_book.csv", "a", encoding="utf-8") as f:
|
| 77 |
-
write = csv.writer(f)
|
| 78 |
-
write.writerow(adres_full)
|
| 79 |
-
return adres_full
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
def get_json(mahalle, il, sokak, apartman):
|
| 83 |
-
adres = {"mahalle": mahalle, "il": il, "sokak": sokak, "apartman": apartman}
|
| 84 |
-
dump = json.dumps(adres, indent=4, ensure_ascii=False)
|
| 85 |
-
return dump
|
| 86 |
-
|
| 87 |
-
def write_db(data_dict):
|
| 88 |
-
# 2) initialize with a project key
|
| 89 |
-
deta_key = os.getenv('DETA_KEY')
|
| 90 |
-
deta = Deta(deta_key)
|
| 91 |
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
users.insert(data_dict)
|
| 95 |
|
| 96 |
|
|
|
|
| 97 |
def text_dict(input):
|
| 98 |
eval_result = ast.literal_eval(input)
|
| 99 |
-
write_db(eval_result)
|
| 100 |
|
| 101 |
return (
|
| 102 |
-
str(eval_result[
|
| 103 |
-
str(eval_result[
|
| 104 |
-
str(eval_result[
|
| 105 |
-
str(eval_result[
|
| 106 |
-
str(eval_result[
|
| 107 |
-
str(eval_result[
|
| 108 |
-
str(eval_result[
|
| 109 |
-
str(eval_result[
|
| 110 |
)
|
| 111 |
-
|
|
|
|
| 112 |
def openai_response(ocr_input):
|
| 113 |
prompt = f"""Tabular Data Extraction You are a highly intelligent and accurate tabular data extractor from
|
| 114 |
plain text input and especially from emergency text that carries address information, your inputs can be text
|
|
@@ -129,19 +61,19 @@ def openai_response(ocr_input):
|
|
| 129 |
resp = eval(resp.replace("'{", "{").replace("}'", "}"))
|
| 130 |
resp["input"] = ocr_input
|
| 131 |
dict_keys = [
|
| 132 |
-
|
| 133 |
-
|
| 134 |
-
|
| 135 |
-
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
|
| 140 |
-
|
| 141 |
]
|
| 142 |
for key in dict_keys:
|
| 143 |
if key not in resp.keys():
|
| 144 |
-
resp[key] =
|
| 145 |
return resp
|
| 146 |
|
| 147 |
def ner_response(ocr_input):
|
|
@@ -157,12 +89,16 @@ def ner_response(ocr_input):
|
|
| 157 |
})
|
| 158 |
return output
|
| 159 |
|
|
|
|
| 160 |
with gr.Blocks() as demo:
|
| 161 |
gr.Markdown(
|
| 162 |
-
|
| 163 |
# Enkaz Bildirme Uygulaması
|
| 164 |
-
"""
|
| 165 |
-
|
|
|
|
|
|
|
|
|
|
| 166 |
with gr.Row():
|
| 167 |
img_area = gr.Image(label="Ekran Görüntüsü yükleyin 👇")
|
| 168 |
ocr_result = gr.Textbox(label="Metin yükleyin 👇 ")
|
|
@@ -183,13 +119,23 @@ with gr.Blocks() as demo:
|
|
| 183 |
with gr.Row():
|
| 184 |
no = gr.Textbox(label="Kapı No")
|
| 185 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 186 |
|
| 187 |
-
|
| 188 |
-
|
| 189 |
-
|
| 190 |
|
| 191 |
-
open_api_text.change(
|
|
|
|
|
|
|
|
|
|
|
|
|
| 192 |
|
| 193 |
|
| 194 |
if __name__ == "__main__":
|
| 195 |
-
demo.launch()
|
|
|
|
|
|
|
| 1 |
from easyocr import Reader
|
| 2 |
+
import gradio as gr
|
|
|
|
|
|
|
|
|
|
| 3 |
import openai
|
| 4 |
import ast
|
| 5 |
import os
|
|
|
|
|
|
|
| 6 |
|
| 7 |
+
from openai_api import OpenAI_API
|
| 8 |
+
import utils
|
|
|
|
| 9 |
|
|
|
|
|
|
|
| 10 |
|
| 11 |
+
openai.api_key = os.getenv("API_KEY")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
reader = Reader(["tr"])
|
| 13 |
|
| 14 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
def get_text(input_img):
|
| 16 |
result = reader.readtext(input_img, detail=0)
|
| 17 |
return " ".join(result)
|
| 18 |
|
| 19 |
|
| 20 |
+
# Submit button
|
| 21 |
+
def get_parsed_address(input_img):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
+
address_full_text = get_text(input_img)
|
| 24 |
+
return openai_response(address_full_text)
|
|
|
|
| 25 |
|
| 26 |
|
| 27 |
+
# Open API on change
|
| 28 |
def text_dict(input):
|
| 29 |
eval_result = ast.literal_eval(input)
|
| 30 |
+
utils.write_db(eval_result)
|
| 31 |
|
| 32 |
return (
|
| 33 |
+
str(eval_result["city"]),
|
| 34 |
+
str(eval_result["distinct"]),
|
| 35 |
+
str(eval_result["neighbourhood"]),
|
| 36 |
+
str(eval_result["street"]),
|
| 37 |
+
str(eval_result["address"]),
|
| 38 |
+
str(eval_result["tel"]),
|
| 39 |
+
str(eval_result["name_surname"]),
|
| 40 |
+
str(eval_result["no"]),
|
| 41 |
)
|
| 42 |
+
|
| 43 |
+
|
| 44 |
def openai_response(ocr_input):
|
| 45 |
prompt = f"""Tabular Data Extraction You are a highly intelligent and accurate tabular data extractor from
|
| 46 |
plain text input and especially from emergency text that carries address information, your inputs can be text
|
|
|
|
| 61 |
resp = eval(resp.replace("'{", "{").replace("}'", "}"))
|
| 62 |
resp["input"] = ocr_input
|
| 63 |
dict_keys = [
|
| 64 |
+
"city",
|
| 65 |
+
"distinct",
|
| 66 |
+
"neighbourhood",
|
| 67 |
+
"street",
|
| 68 |
+
"no",
|
| 69 |
+
"tel",
|
| 70 |
+
"name_surname",
|
| 71 |
+
"address",
|
| 72 |
+
"input",
|
| 73 |
]
|
| 74 |
for key in dict_keys:
|
| 75 |
if key not in resp.keys():
|
| 76 |
+
resp[key] = ""
|
| 77 |
return resp
|
| 78 |
|
| 79 |
def ner_response(ocr_input):
|
|
|
|
| 89 |
})
|
| 90 |
return output
|
| 91 |
|
| 92 |
+
# User Interface
|
| 93 |
with gr.Blocks() as demo:
|
| 94 |
gr.Markdown(
|
| 95 |
+
"""
|
| 96 |
# Enkaz Bildirme Uygulaması
|
| 97 |
+
"""
|
| 98 |
+
)
|
| 99 |
+
gr.Markdown(
|
| 100 |
+
"Bu uygulamada ekran görüntüsü sürükleyip bırakarak AFAD'a enkaz bildirimi yapabilirsiniz. Mesajı metin olarak da girebilirsiniz, tam adresi ayrıştırıp döndürür. API olarak kullanmak isterseniz sayfanın en altında use via api'ya tıklayın."
|
| 101 |
+
)
|
| 102 |
with gr.Row():
|
| 103 |
img_area = gr.Image(label="Ekran Görüntüsü yükleyin 👇")
|
| 104 |
ocr_result = gr.Textbox(label="Metin yükleyin 👇 ")
|
|
|
|
| 119 |
with gr.Row():
|
| 120 |
no = gr.Textbox(label="Kapı No")
|
| 121 |
|
| 122 |
+
submit_button.click(
|
| 123 |
+
get_parsed_address,
|
| 124 |
+
inputs=img_area,
|
| 125 |
+
outputs=open_api_text,
|
| 126 |
+
api_name="upload_image",
|
| 127 |
+
)
|
| 128 |
|
| 129 |
+
ocr_result.change(
|
| 130 |
+
openai_response, ocr_result, open_api_text, api_name="upload-text"
|
| 131 |
+
)
|
| 132 |
|
| 133 |
+
open_api_text.change(
|
| 134 |
+
text_dict,
|
| 135 |
+
open_api_text,
|
| 136 |
+
[city, distinct, neighbourhood, street, address, tel, name_surname, no],
|
| 137 |
+
)
|
| 138 |
|
| 139 |
|
| 140 |
if __name__ == "__main__":
|
| 141 |
+
demo.launch()
|
openai_api.py
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import openai
|
| 2 |
+
import os
|
| 3 |
+
|
| 4 |
+
class OpenAI_API:
|
| 5 |
+
def __init__(self):
|
| 6 |
+
self.openai_api_key = ""
|
| 7 |
+
|
| 8 |
+
def single_request(self, address_text):
|
| 9 |
+
|
| 10 |
+
openai.api_type = "azure"
|
| 11 |
+
openai.api_base = "https://damlaopenai.openai.azure.com/"
|
| 12 |
+
openai.api_version = "2022-12-01"
|
| 13 |
+
openai.api_key = os.getenv("API_KEY")
|
| 14 |
+
|
| 15 |
+
response = openai.Completion.create(
|
| 16 |
+
engine="Davinci-003",
|
| 17 |
+
prompt=address_text,
|
| 18 |
+
temperature=0.9,
|
| 19 |
+
max_tokens=256,
|
| 20 |
+
top_p=1.0,
|
| 21 |
+
n=1,
|
| 22 |
+
logprobs=0,
|
| 23 |
+
echo=False,
|
| 24 |
+
stop=None,
|
| 25 |
+
frequency_penalty=0,
|
| 26 |
+
presence_penalty=0,
|
| 27 |
+
best_of=1,
|
| 28 |
+
)
|
| 29 |
+
|
| 30 |
+
return response
|
utils.py
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import cv2
|
| 2 |
+
import csv
|
| 3 |
+
import json
|
| 4 |
+
from deta import Deta
|
| 5 |
+
import os
|
| 6 |
+
|
| 7 |
+
|
| 8 |
+
def preprocess_img(inp_image):
|
| 9 |
+
gray = cv2.cvtColor(inp_image, cv2.COLOR_BGR2GRAY)
|
| 10 |
+
gray_img = cv2.bitwise_not(gray)
|
| 11 |
+
return gray_img
|
| 12 |
+
|
| 13 |
+
def save_csv(mahalle, il, sokak, apartman):
|
| 14 |
+
adres_full = [mahalle, il, sokak, apartman]
|
| 15 |
+
|
| 16 |
+
with open("adress_book.csv", "a", encoding="utf-8") as f:
|
| 17 |
+
write = csv.writer(f)
|
| 18 |
+
write.writerow(adres_full)
|
| 19 |
+
return adres_full
|
| 20 |
+
|
| 21 |
+
|
| 22 |
+
def get_json(mahalle, il, sokak, apartman):
|
| 23 |
+
adres = {"mahalle": mahalle, "il": il, "sokak": sokak, "apartman": apartman}
|
| 24 |
+
dump = json.dumps(adres, indent=4, ensure_ascii=False)
|
| 25 |
+
return dump
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
def write_db(data_dict):
|
| 29 |
+
# 2) initialize with a project key
|
| 30 |
+
deta_key = os.getenv('DETA_KEY')
|
| 31 |
+
deta = Deta(deta_key)
|
| 32 |
+
|
| 33 |
+
# 3) create and use as many DBs as you want!
|
| 34 |
+
users = deta.Base("deprem-ocr")
|
| 35 |
+
users.insert(data_dict)
|