ghandic’s gists

ghandic / fastapi_query_utils.py

Last active April 22, 2024 04:02

fastapi sortby/filter

	from enum import auto
	from typing import List, Optional
	from collections import Counter

	from fastapi_utils.enums import StrEnum
	from fastapi_utils.inferring_router import InferringRouter
	from pydantic import BaseModel
	from fastapi import HTTPException, Query, Depends

	router = InferringRouter()

ghandic / r_rename_ext.sh

Created February 27, 2019 01:57

Recursive rename of file extension

find . -regex '.*.py' -type f | sed "s/\.py$//" | xargs -I% mv -iv %.py %.txt

ghandic / s3_pdf.py

Created July 27, 2018 12:04

Load pdf from S3 directly into memory as list of PIL images

	"""This module requires:
	- poppler-utils [apt-get upde && apt-get install poppler-utils]

	- pdf2image [pip install pdf2image]
	- pillow [pip install pillow]
	- boto3 [pip install boto3]
	"""

	import boto3
	from pdf2image import convert_from_bytes

ghandic / pil_s3.py

Last active March 15, 2024 14:16

Load image from S3 directly into memory as PIL image and write to S3 directly from memory from PIL image

	import boto3
	from PIL import Image
	from io import BytesIO
	import os

	class S3ImagesInvalidExtension(Exception):
	pass

	class S3ImagesUploadFailed(Exception):
	pass

ghandic / contour_field_detection.py

Created July 26, 2018 11:38

Function to detect fields from a form, may need fine tuning!

	from PIL import Image
	import numpy as np
	import cv2
	import time

	# Helper function for jupyter notebooks
	def showarray(a):
	return Image.fromarray(a)

	def detect_boxes(img, min_percent_of_area=0.02, max_percent_of_area=2):

ghandic / pandas_s3.py

Last active June 5, 2023 11:40

Load csv from S3 directly into memory and write to S3 directly from memory by extending pd.DataFrame class

	import boto3
	import pandas as pd
	from io import StringIO

	class S3DataFrame(pd.DataFrame):

	"""
	# Make a dataframe and upload it as csv
	s3df = S3DataFrame({'h1':[1], 'h2':[2]})
	s3df.to_s3(Bucket='bucket-name',

ghandic / faster_tess.py

Created July 17, 2018 16:29

Running Tesseract with lots of small images is much slower than multipage TIFF, so make a multipage tiff (to be implemented with multiprocessing for the major speed gains)

	## Tesseract4
	# 2 images into tiff: 843 msec 796
	# 10 images into tiff: 1003 msec
	# 53 images (same as page 1) into tiff: 1740 msec

	## Tesseract3
	# 2 images into tiff: 727 msec
	# 10 images into tiff: 816 msec
	# 53 images (same as page 1) into tiff: 1620 msec

ghandic / png_with_meta.py

Created July 12, 2018 08:08

Save image as png with metadata

	from PIL import Image
	from PIL.PngImagePlugin import PngInfo

	im = Image.open('example.png')

	im_info = PngInfo()
	im_info.add_text('Key Information', 'Value Information')

	im.save('example.png', pnginfo=im_info)

ghandic / labeled_gif.py

Last active July 6, 2018 09:16

Creates a labeled gif using opencv imageio and numpy

	import imageio
	from typing import NamedTuple

	import cv2
	import numpy as np

	LabeledGIFState = NamedTuple('LabeledGIFState', [('image', np.ndarray), ('label', str)])

	class GifSettings(object):
	red = [255, 0, 0]

ghandic / imessage.sh

Created June 18, 2018 16:40

Send an automated iMessage

	targetBuddyPhone='+44765979706'
	targetMessage='This is a automated message'

	osascript <<EOF
	tell application "Messages"
	set targetService to 1st service whose service type = iMessage
	set targetBuddy to buddy "$targetBuddyPhone" of targetService
	send "$targetMessage" to targetBuddy
	end tell
	EOF

Andy Challis ghandic