fast microservice to accompany n8n workflow
Create main.py
authored by
Jasper Mayone
and committed by
GitHub
2b2c07c1
+115
+115
main.py
+115
main.py
···
1
+
# main.py
2
+
from fastapi import FastAPI, File, UploadFile
3
+
from fastapi.middleware.cors import CORSMiddleware
4
+
import pytesseract
5
+
from PIL import Image
6
+
import io
7
+
import cv2
8
+
import numpy as np
9
+
import uvicorn
10
+
11
+
app = FastAPI()
12
+
13
+
# Add CORS middleware
14
+
app.add_middleware(
15
+
CORSMiddleware,
16
+
allow_origins=["*"], # Allows all origins
17
+
allow_credentials=True,
18
+
allow_methods=["*"], # Allows all methods
19
+
allow_headers=["*"], # Allows all headers
20
+
)
21
+
22
+
def process_image(image_bytes):
23
+
# Convert bytes to numpy array
24
+
nparr = np.frombuffer(image_bytes, np.uint8)
25
+
img = cv2.imdecode(nparr, cv2.IMREAD_COLOR)
26
+
27
+
# Convert to grayscale
28
+
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
29
+
30
+
# Apply thresholding
31
+
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)[1]
32
+
33
+
# Convert back to PIL Image
34
+
pil_image = Image.fromarray(thresh)
35
+
36
+
# Perform OCR
37
+
text = pytesseract.image_to_string(pil_image)
38
+
39
+
return text
40
+
41
+
def parse_schedule(text):
42
+
lines = text.split('\n')
43
+
schedule = {}
44
+
current_person = None
45
+
dates = {}
46
+
47
+
for line in lines:
48
+
# Skip empty lines
49
+
if not line.strip():
50
+
continue
51
+
52
+
parts = line.split()
53
+
if not parts:
54
+
continue
55
+
56
+
# If we find the date row (contains multiple '1/' patterns)
57
+
if sum('1/' in part for part in parts) > 1:
58
+
dates = {
59
+
'monday': next((p for p in parts if '1/' in p), ''),
60
+
'tuesday': next((p for p in parts[1:] if '1/' in p), ''),
61
+
'wednesday': next((p for p in parts[2:] if '1/' in p), ''),
62
+
'thursday': next((p for p in parts[3:] if '1/' in p), ''),
63
+
'friday': next((p for p in parts[4:] if '1/' in p), ''),
64
+
'saturday': next((p for p in parts[5:] if '1/' in p), ''),
65
+
'sunday': next((p for p in parts[6:] if '1/' in p), '')
66
+
}
67
+
continue
68
+
69
+
# Check if this line starts with a name (no numbers or 'OFF')
70
+
if len(parts) >= 1 and not parts[0].replace('-', '').isdigit() and parts[0] != 'OFF' and ':' not in parts[0]:
71
+
# Ignore "CASHIERS" header
72
+
if parts[0] == "CASHIERS":
73
+
continue
74
+
current_person = ' '.join(parts)
75
+
schedule[current_person] = {
76
+
'monday': {'date': dates.get('monday', ''), 'shift': ''},
77
+
'tuesday': {'date': dates.get('tuesday', ''), 'shift': ''},
78
+
'wednesday': {'date': dates.get('wednesday', ''), 'shift': ''},
79
+
'thursday': {'date': dates.get('thursday', ''), 'shift': ''},
80
+
'friday': {'date': dates.get('friday', ''), 'shift': ''},
81
+
'saturday': {'date': dates.get('saturday', ''), 'shift': ''},
82
+
'sunday': {'date': dates.get('sunday', ''), 'shift': ''}
83
+
}
84
+
# If we have shifts data and a current person
85
+
elif current_person and len(parts) >= 7:
86
+
schedule[current_person] = {
87
+
'monday': {'date': dates.get('monday', ''), 'shift': parts[0]},
88
+
'tuesday': {'date': dates.get('tuesday', ''), 'shift': parts[1]},
89
+
'wednesday': {'date': dates.get('wednesday', ''), 'shift': parts[2]},
90
+
'thursday': {'date': dates.get('thursday', ''), 'shift': parts[3]},
91
+
'friday': {'date': dates.get('friday', ''), 'shift': parts[4]},
92
+
'saturday': {'date': dates.get('saturday', ''), 'shift': parts[5]},
93
+
'sunday': {'date': dates.get('sunday', ''), 'shift': parts[6]}
94
+
}
95
+
96
+
return schedule
97
+
98
+
@app.post("/ocr")
99
+
async def ocr_endpoint(file: UploadFile = File(...)):
100
+
# Read the image file
101
+
image_bytes = await file.read()
102
+
103
+
# Process the image and get text
104
+
text = process_image(image_bytes)
105
+
106
+
# Parse the schedule
107
+
schedule = parse_schedule(text)
108
+
109
+
return {
110
+
"raw_text": text,
111
+
"schedule": schedule
112
+
}
113
+
114
+
if __name__ == "__main__":
115
+
uvicorn.run(app, host="0.0.0.0", port=8000)