from PIL import Image
from PIL import ImageDraw
from PIL import ImageFont
import urllib.request
import io
import binascii
data = urllib.request.urlopen('http://pastebin.ca/raw/2311595').read()
r_data = binascii.unhexlify(data)
stream = io.BytesIO(r_data)
img = Image.open(stream)
draw = ImageDraw.Draw(img)
font = ImageFont.truetype("arial.ttf",14)
draw.text((0, 220),"This is a test11",(0,255,0),font=font)
draw = ImageDraw.Draw(img)
with open(img,'rb') as in_file: #error on here invalid file:
hex_data = in_file.read()
# Unhexlify the data.
bin_data = binascii.unhexlify(bytes(hex_data))
print(bin_data)
Question
converting hex to image and draw a text on the image, after that convert image to binary hex,but having the problem at here with open(img,'rb') as in_file:, how to convert img to hex?
The img object needs to be saved again; write it to another BytesIO object:
output = io.BytesIO()
img.save(output, format='JPEG')
then get the written data with the .getvalue() method:
hex_data = output.getvalue()
The PIL-for-python-3 landscape is rather muddled at the moment. The Pillow fork looks to be the best, maintained version out there at the moment. It includes fixes that make saving to a BytesIO object work. If you run into a io.UnsupportedOperation: fileno exception using the above code, you have a version that was not yet fixed, in which case you'll have to resort to using a temporary file instead.
Related
I stored an image to StringIO from PIL. When I store it to a file from stringIO, it doesn't produce the original image.
Code:
from PIL import Image
from cStringIO import StringIO
buff=StringIO()
img = Image.open("test.jpg")
img.save(buff,format='JPEG')
#img=img.crop((1,1,100,100))
buff.seek(0)
#Produces a distorted image
with open("vv.jpg", "w") as handle:
handle.write(buff.read())
Original Image is below
Output image is below
What is wrong with the above code
You need to use BytesIO and not StringIO.
Also the destination file has to be opened in binary mode using "wb"
Here is code that works (cStringIO is replaced with io)
from PIL import Image
from io import BytesIO
buff=BytesIO()
img = Image.open('test.jpg')
img.save(buff,format='JPEG')
#img=img.crop((1,1,100,100))
buff.seek(0)
#Produces a distorted image
with open('vv.jpg', "wb") as handle:
handle.write(buff.read())
I have a remote PDF file that I need to read page by page and keep passing each to an OCR which will give me its OCR text.
import pytesseract
from pyPdf import PdfFileWriter, PdfFileReader
import cStringIO
from wand.image import Image
import urllib2
import tempfile
import pytesseract
from PIL import Image
remoteFile = urllib2.urlopen(urllib2.Request("file:///home/user/Documents/TestDocs/test.pdf")).read()
memoryFile = cStringIO.StringIO(remoteFile)
pdfFile = PdfFileReader(memoryFile)
for pageNum in xrange(pdfFile.getNumPages()):
currentPage = pdfFile.getPage(pageNum)
## somehow convert currentPage to wand type
## image and then pass to tesseract-api
##
## TEMP_IMAGE = some conversion to temp file
## pytesseract.image_to_string(Image.open(TEMP_IMAGE))
memoryFile.close()
I thought of using cStringIO or tempfile but I cannot figure out how to use them for this purpose.
How can solve this issue?
There's a couple options for doing this, the more compatible way given the code you supplied is to store the images temporarily in that directory and then delete them after reading the text using pytesseract. I create a wand type image to extract each image from the PDF individually, then convert it to a PIL type image for pytesseract. Here's the code I used for this with the detected text bring written to an array 'text' where each element is an image in the original PDF, I also updated some of your imports to make it compatible with Python3 (cStringIO->io and urllib2->urllib.request).
import PyPDF2
import os
import pytesseract
from wand.image import Image
from PIL import Image as PILImage
import urllib.request
import io
with urllib.request.urlopen('file:///home/user/Documents/TestDocs/test.pdf') as response:
pdf_read = response.read()
pdf_im = PyPDF2.PdfFileReader(io.BytesIO(pdf_read))
text = []
for p in range(pdf_im.getNumPages()):
with Image(filename='file:///home/user/Documents/TestDocs/test.pdf' + '[' + str(p) + ']') as img:
with Image(image = img) as converted: #Need second with to convert SingleImage object from wand to Image
converted.save(filename=tempFile_Location)
text.append(pytesseract.image_to_string(PILImage.open(tempFile_Location)))
os.remove(tempFile_Location)
Alternatively, if you want to avoid creating and deleting a temporary file for each image you can use numpy and OpenCV to extract the image as a blob, convert it to a numpy array and then turn it into a PIL image for pytesseract to perform OCR on (reference)
import PyPDF2
import os
import pytesseract
from wand.image import Image
from PIL import Image as PILImage
import urllib.request
import io
import numpy as np
import cv2
with urllib.request.urlopen('file:///home/user/Documents/TestDocs/test.pdf') as response:
pdf_read = response.read()
pdf_im = PyPDF2.PdfFileReader(io.BytesIO(pdf_read))
text = []
for p in range(pdf_im.getNumPages()):
with Image(filename=('file:///home/user/Documents/TestDocs/test.pdf') + '[' + str(p) + ']') as img:
img_buffer=np.asarray(bytearray(img.make_blob()), dtype=np.uint8)
retval = cv2.imdecode(img_buffer, cv2.IMREAD_GRAYSCALE)
text.append(pytesseract.image_to_string(PILImage.fromarray(retval)))
I am writing an application that uses images intensively. It is composed of two parts. The client part is written in Python. It does some preprocessing on images and sends them over TCP to a Node.js server.
After preprocessing, the Image object looks like this:
window = img.crop((x,y,width+x,height+y))
window = window.resize((48,48),Image.ANTIALIAS)
To send that over socket, I have to have it in binary format. What I am doing now is:
window.save("window.jpg")
infile = open("window.jpg","rb")
encodedWindow = base64.b64encode(infile.read())
#Then send encodedWindow
This is a huge overhead, though, since I am saving the image to the hard disk first, then loading it again to obtain the binary format. This is causing my application to be extremely slow.
I read the documentation of PIL Image, but found nothing useful there.
According to the documentation, (at effbot.org):
"You can use a file object instead of a filename. In this case, you must always specify the format. The file object must implement the seek, tell, and write methods, and be opened in binary mode."
This means you can pass a StringIO object. Write to it and get the size without ever hitting the disk.
Like this:
s = StringIO.StringIO()
window.save(s, "jpg")
encodedWindow = base64.b64encode(s.getvalue())
use BytesIO
from io import BytesIO
from PIL import Image
photo=Image.open('photo.jpg')
s=BytesIO()
photo.save(s,'jpeg')
data = s.getvalue()
with open('photo2.jpg', mode='wb') as f:
f.write(data)
It's about the difference between in-memory file-like object and BufferedReader object.
Here is my experiment in Jupyter(Python 3.8.10):
from PIL import Image as PILImage, ImageOps as PILImageOps
from IPython.display import display, Image
from io import BytesIO
import base64
url = "https://learn.microsoft.com/en-us/archive/msdn-magazine/2018/april/images/mt846470.0418_mccaffreytrun_figure2_hires(en-us,msdn.10).png"
print("get computer-readable bytes from the url")
img_bytes = requests.get(url).content
print(type(img_bytes))
display(Image(img_bytes))
print("convert to in-memory file-like object")
in_memory_file_like_object = BytesIO(img_bytes)
print(type(in_memory_file_like_object))
print("convert to an PIL Image object for manipulating")
pil_img = PILImage.open(in_memory_file_like_object)
print("let's rotate it, and it remains a PIL Image object")
pil_img.show()
rotated_img = pil_img.rotate(45)
print(type(rotated_img))
print("let's create an in-memory file-like object and save the PIL Image object into it")
in_memory_file_like_object = BytesIO()
rotated_img.save(in_memory_file_like_object, 'png')
print(type(in_memory_file_like_object))
print("get computer-readable bytes")
img_bytes = in_memory_file_like_object.getvalue()
print(type(img_bytes))
display(Image(img_bytes))
print('convert to base64 to be transmitted over channels that do not preserve all 8-bits of data, such as email')
# https://stackoverflow.com/a/8909233/3552975
base_64 = base64.b64encode(img_bytes)
print(type(base_64))
# https://stackoverflow.com/a/45928164/3552975
assert base64.b64encode(base64.b64decode(base_64)) == base_64
In short you can save a PIL Image object into an in-memory file-like object by rotated_img.save(in_memory_file_like_object, 'png') as shown above, and then conver the in-memory file-like object into base64.
from io import BytesIO
b = BytesIO()
img.save(b, format="png")
b.seek(0)
data = b.read()
del b
I wrote this code :
with Image.open(objective.picture.read()) as image:
image_file = BytesIO()
exifdata = image.info['exif']
image.save(image_file, 'JPEG', quality=50, exif=exifdata)
zf.writestr(zipped_filename, image_file)
Which is supposed to open the image stored in my model (this is in a Django application). I want to reduce the quality of the image file before adding it to the zipfile (zf). So I decided to work with BytesIO to prevent writing useless file on the disk. Though I'm getting an error here. It says :
embedded NUL character
Could someone help me out with this ? I don't understand what's going on.
Well I was kind of dumb. objective.picture.read() returns a byte string (really long byte string...) so I shouldn't have used Image but ImageFile.Parser() and feed that byte string to the parser so it can return an Image that I can work with. Here is the code :
from PIL import ImageFile
from io import BytesIO
p = ImageFile.Parser()
p.feed(objective.picture.read())
image = p.close()
image_file = BytesIO()
exifdata = image.info['exif']
image.save(image_file, 'JPEG', quality=50, exif=exifdata)
# Here zf is a zipfile writer
zf.writestr(zipped_filename, image_file.getvalue())
The close() actually returns the image parsed from the bytestring.
Here is the doc : The ImageFile Documentation
I am looking to create base64 inline encoded data of images for display in a table using canvases. Python generates and creates the web page dynamically. As it stands python uses the Image module to create thumbnails. After all of the thumbnails are created Python then generates base64 data of each thumbnail and puts the b64 data into hidden spans on the user's webpage. A user then clicks check marks by each thumbnail relative to their interest. They then create a pdf file containing their selected images by clicking a generate pdf button. The JavaScript using jsPDF generates the hidden span b64 data to create the image files in the pdf file and then ultimately the pdf file.
I am looking to hopefully shave down Python script execution time and minimize some disk I/O operations by generating the base64 thumbnail data in memory while the script executes.
Here is an example of what I would like to accomplish.
import os, sys
import Image
size = 128, 128
im = Image.open("/original/image/1.jpeg")
im.thumbnail(size)
thumb = base64.b64encode(im)
This doesn't work sadly, get a TypeErorr -
TypeError: must be string or buffer, not instance
Any thoughts on how to accomplish this?
You first need to save the image again in JPEG format; using the im.tostring() method would otherwise return raw image data that no browser would recognize:
from io import BytesIO
output = BytesIO()
im.save(output, format='JPEG')
im_data = output.getvalue()
This you can then encode to base64:
image_data = base64.b64encode(im_data)
if not isinstance(image_data, str):
# Python 3, decode from bytes to string
image_data = image_data.decode()
data_url = 'data:image/jpg;base64,' + image_data
Here is one I made with this method:
data:image/jpg;base64,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
Unfortunately the Markdown parser doesn't let me use this as an actual image, but you can see it in action in a snippet instead:
<img src="data:image/jpg;base64,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"/>
In Python 3, you may need to use BytesIO:
from io import BytesIO
...
outputBuffer = BytesIO()
bg.save(outputBuffer, format='JPEG')
bgBase64Data = outputBuffer.getvalue()
# http://stackoverflow.com/q/16748083/2603230
return 'data:image/jpeg;base64,' + base64.b64encode(bgBase64Data).decode()
thumb = base64.b64encode(im.tostring())
I think would work
I use PNG when I save to the buffer. With JPEG the numpy arrays are a bit different.
import base64
import io
import numpy as np
from PIL import Image
image_path = 'dog.jpg'
img2 = np.array(Image.open(image_path))
# Numpy -> b64
buffered = io.BytesIO()
Image.fromarray(img2).save(buffered, format="PNG")
b64image = base64.b64encode(buffered.getvalue())
# b64 -> Numpy
img = np.array(Image.open(io.BytesIO(base64.b64decode(b64image))))
print(img.shape)
np.testing.assert_almost_equal(img, img2)
Note that it will be slower.