so deep a rabbit hole

Ok, prefacing this here with the reminder that I'm not even to the level of script kiddie here. I'm hanging on to this with a thread.

I'm using curl to save an image to my Indigo public directory. That's working fine. But the native image is too big for the reflector or something. I get an Indigo error trying to display it.

So I've been trying various methods to resize the image automatically, and I'm now trying Python, which seems capable. Found this:

Code: Select all

import Image, ImageFile 
pic = "/Library/Application Support/Perceptive Automation/Indigo 7/IndigoWebServer/public/whatev.jpg"
im = Image.open(pic)
width, height = im.size
if width > 1200 or height > 1200:
    width = width*.7
    height = height*.7
im = im.resize((int(math.floor(width)), int(math.floor(height))), Image.ANTIALIAS)
try:
    im.save(pic,optimize=True,quality=70)
except IOError:
    ImageFile.MAXBLOCK = width * height
    im.save(pic,optimize=True,quality=70)


But when I try running this inside Indigo I get "No module named Image."

I DO have pillow installed, but it's installed via Python 2.7 apparently. Indigo uses 2.6, right?

I tried pip install pillow, but all I got was "requirement already satisfied."

Trying to follow instructions at https://stackoverflow.com/questions/553 ... on#5538353 got me more errors, and at that point I'm lost down this hole.

Alternately, if you think it's easier, I've also tried using ffmpeg inside a shell script to do this, but I keep running into trouble there too, as it appears that even though I try retrieving the image via curl then ffmpeg resizing, the shell script multi threads or something and ffmpeg is resizing the previous image before the curl retrieves the new image. And adding "sleep" commands didn't help.

First, Indigo 7 uses Python 2.7, not 2.6.

Second, your import line should be "from PIL import Image".

Third, this code makes no sense:

Code: Select all

except IOError:
    ImageFile.MAXBLOCK = width * height
    im.save(pic,optimize=True,quality=70)


Where did you get that from?

Thanks FD!

Got it from https://stackoverflow.com/questions/207 ... -in-python

I'll give it a try asap, depending on how my remote access is working. I planned to tune the image modification from the original code, but you're right I can't figure out what that is supposed to do. The ironic thing is all I apparently need to do is the slightest of compression and the file goes from 922 kb to ~200 kb. It's almost as if the camera is saving it with every possible quality setting at max.

Edit: I debugged all by my own self!!!

Here's what the script looks like now. I'm so proud. Still don't know what I'm doing, but I'm proud.

Code: Select all

from PIL import Image
import math
pic = "/Library/Application Support/Perceptive Automation/Indigo 7/IndigoWebServer/public/whatev.jpg"
im = Image.open(pic)
width, height = im.size
if width > 1200 or height > 1200:
    width = width*.99
    height = height*.99
im = im.resize((int(math.floor(width)), int(math.floor(height))), Image.ANTIALIAS)
im.save(pic,optimize=True,quality=90)

Improvement suggestions welcome.

[jay (support)]

Cool! Congrats on your first Python debug!

Another thought: I think that there's an Automator action in macOS for resizing images - you could dump the image into a temporary directory and have a folder action attached to the folder that watches for a new image, does the resize, then moves it to the Indigo directory.

I like your solution (var) better, but if someone runs across this thread and wants a solution that doesn't require installing a Python module/script, that's an alternative (albeit with more moving parts).

That's hilarious. It's exactly what I tried for my second go at fixing this--once I was unable to make ffmpeg do what I wanted.

The problem with this approach was that somehow all the parts of it would not happen automatically in the right order. A shell script sent a curl to the camera and saved the result to a watched folder, which the Automator action should then have reduced. This all worked GREAT when I ran each step manually, but when I tried to make each bit of it trigger the next bit it all fell apart, to the point that the curl would make Automator reduce the file size of the file that was about to be overwritten.

I could not fix this with sleep in the shell script, or pause in Automator. And in any case, that's a kludge because really, I want it all to happen as fast as possible--as long as it happens in the right order!

All of this is a sour reminder of how much acquired knowledge is built in to any how-to instructions. Three or four times I hit barricades because what I needed to do with ffmpeg or curl or import had examples, but they didn't exactly match up with what I needed to do, and I couldn't figure out how to piece it all together. (For example, how to make curl do authentication, or how to make ffmpeg change the quality of a static image.)

Mostly, I'm inordinately and unreasonably proud of realizing I needed to add "import math" up at the top of that script.

Maybe this can be simplified even more. Right now it's a shell script for getting the image, and python to make it smaller.

is this python equivalent to a curl command?

Code: Select all

import requests
res = requests.get('user:pwd@http://192.168.0.999')

seems like it doesn't say where to put the file that is retrieved. I think I'm supposed to do something like

Code: Select all

with open(image_name, 'webcampic') as outfile: outfile.write(r.content)

Am I on the right track?

[jay (support)]

The URL in your post is incorrect - if it is supposed to be:

Code: Select all

http://user@pass:IP/path/to/stuff

Then you'd need to specify the user/pass a bit differently:

Code: Select all

import requests
from requests.auth import HTTPBasicAuth
import StringIO
pic_destination = "/Library/Application Support/Perceptive Automation/Indigo 7/IndigoWebServer/public/whatev.jpg"
reply = requests.get("http://IP/path/to/stuff", auth=HTTPBasicAuth(user, pass))
# now, reply.content is the image, so you can use that to create the Image object:
im = Image.open(StringIO.StringIO(reply.content))
# then this part should work:
width, height = im.size
if width > 1200 or height > 1200:
    width = width*.99
    height = height*.99
im = im.resize((int(math.floor(width)), int(math.floor(height))), Image.ANTIALIAS)
im.save(pic_destination,optimize=True,quality=90)


I think. I just pasted together the parts without actually trying it - that's your next debug lesson!!

FWIW, just for posterity, you can also use the "sips" command on OS X to resize an image pretty easily:

Code: Select all

sips -z <imageheight> <imagewidth> <outputfile>
sips -Z <imagemaxsize> <outputfile>

The former resizes to an exact dimension, the latter resamples it so that neither dimension is greater than the max size.

Adam

@jay I almost think you used double quotes on the URL to test me. The other trick was realizing that

Code: Select all

auth=HTTPBasicAuth(user, pass)

needs to be

Code: Select all

auth=HTTPBasicAuth('user', 'pass')

Got it working and it's much, much simpler than any of the other methods I was trying. I've commented out the size stuff, because I don't need it currently.

Code: Select all

from PIL import Image
import math
import requests
from requests.auth import HTTPBasicAuth
import StringIO
pic_destination = "/Library/Application Support/Perceptive Automation/Indigo 7/IndigoWebServer/public/whatev.jpg"
reply = requests.get('http://192.168.0.999/cgi-bin/currentpic.cgi', auth=HTTPBasicAuth('user', 'pass'))
# now, reply.content is the image, so you can use that to create the Image object:
im = Image.open(StringIO.StringIO(reply.content))
# then this part should work:
width, height = im.size
#if width > 1200 or height > 1200:
 #   width = width*.99
  #  height = height*.99
im = im.resize((int(math.floor(width)), int(math.floor(height))), Image.ANTIALIAS)
im.save(pic_destination,optimize=True,quality=80)

And things I think I've learned also:
1. In the above "pic_destination" and "reply" are variables, right? Containers?
2. code snippet examples from the internet rarely include all the "Include PythonThingYouNeed" at the top
3. Just 'cause it compiles doesn't mean it runs.

@RogueProeliator thanks for that tip, I bet it will be handy for other situations, but all I really needed to do with this image was reduce its quality rather than its dimensions.

[jay (support)]

@jay I almost think you used double quotes on the URL to test me.

Only if user and auth aren't variables that contain the data (see below for more details)...

In the above "pic_destination" and "reply" are variables, right?

Those are local variables, yes. You could put the actual username/password in variables rather than directly into the call (thus the quotes you point out) if the script were long enough that you needed to make multiple HTTP requests (then you'd only need to change the variables at the top if the username and/or password changed in the future).

code snippet examples from the internet rarely include all the "Include PythonThingYouNeed" at the top

It's definitely hit or miss - I always try to make sure any code snippits I post have any imports necessary at the top. Out on the net it's less consistent.

Just 'cause it compiles doesn't mean it runs.

In the case of an embedded Python script in Indigo, we only do a syntax check when you hit the Compile button (the exact same thing would happen when you hit the Run button but it would also actually execute the script as well). Python scripts are interpreted at runtime so there actually isn't really a "compile" phase as with other languages like C/C++/Objective-C.

Great progress! You'll be a Python programmer before you know it!