Xcode Template for Pythonista
TheRealBret last edited by TheRealBret
Looks like the texinfo hack needs to be modified. The code signing process uses file(1) to check the type of file in the package, and file(1) identifies executable code by checking against different file types and scoring each file type. Apple modified the matching "magic" for python to look for python keywords like "def", "import", "try", "except", etc and each use gives a higher score to python. So for the tex info hack to work, it needs to occur more frequently than the python keywords. Some files just need a single texinfo line; other files need 400+ copies of the texinfo line (some of the python checks only examine the first 4096 characters of the file). Its easy to modify the script above to accomplish this, but hardly elegant.
Hopefully somebody will find this useful and find a simple fix.... like perhaps there's another file type that short-circuits the checking magic? Also, the Apple developer documentation say that script files could go into a specially named subfolder, which would be much simpler, but I've haven't gotten this to work.
JonB last edited by
out of curiosity, is
fileused locally, or on the server?
Couldn't you use th older file version for identifying text files?
jumpbeen last edited by
TheRealBret last edited by
Thanks. Will give it a try and test an upload to TestFlight too.
sreyemnayr last edited by sreyemnayr
Here's the re-worked script, which throws as many of those dummies at the top of each file as it needs to stop the madness.
EDIT: There are some problems with this method as well, as it bounces the lines indicating the encoding style, which breaks a bunch of stuff... I ended up having to regex replace them back to the top, but will work this script out to do it better later. Just be aware that this will happen!
# Change these: pylib_path1 = '/path/to/PythonistaKit.framework/pylib' pylib_path2 = '/path/to/PythonistaKit.framework/pylib_ext' import shutil import os import subprocess def check_is_executable(file_path): file_output = subprocess.check_output(['file', file_path]) if 'executable' in file_output: return True, file_output return False, file_output def fix_executable(file_path): with open(file_path, 'r') as f: source = f.read() with open(file_path, 'w') as f: f.write('#\input texinfo\n' + source) is_executable, out = check_is_executable(file_path) return not is_executable def fix_pylib(pylib_path, dry_run=False): for path, dirs, files in os.walk(pylib_path): for filename in files: full_path = full_path = os.path.join(path, filename) is_executable, file_output = check_is_executable(full_path) if is_executable: extension = os.path.splitext(full_path).lower() if extension == '.py' or extension == '.pym' or filename == 'command_template': if dry_run: print '### Executable found: %s' % (filename,) else: print 'Fixing %s...' % (filename,) fixed = False counter = 0 while not fixed: fixed = fix_executable(full_path) counter = counter + 1 print 'Fixed with '+str(counter)+' dummy lines.' else: print '### Executable found, but does not seem to be Python code: %s' % (full_path,) if __name__ == '__main__': fix_pylib(pylib_path1) fix_pylib(pylib_path2)```
Hi @sreyemnayr, just tried your script, but still getting an error on upload to app store. Any other ideas or refinements to your script?
I'm making a bit of progress:
- My original error related to the future.py file in pylib, after running the script the Xcode error changed.
- I noticed that after running the above script the specific file that I was getting the 'Invalid Signature' error on had change to pylib/compiler/init.py. I had a look at that file and it didn't have a #\input texinfo up the top so I added it manually.
- I then reran the script and the error changed to pylib/compiler/ast.py This did have lots of #\input texinfo, so I copy and pasted all of them to double the amount of #\input texinfo lines at the top of the file. This seemed to satisfy Xcode.
- I then reran the script and the error changed one of the files in pylib/ctypes/test. I just deleted the whole pylib/ctypes/test folder.
- I than reran the script and the error changed to pylib/dateutil/READEME. I deleted the README file and the NEWS file while I was at it.
- I than reran the script and the error changes to pylib/site-packages/dropbox/init.py. As I'm not using dropbox in my app I just deleted the dropbox folder.
- I then had errors on pylib/site-packages/twitter.py, webbrowser.py, xmltodict.py and ui.py. They all had just one #\input texinfo line so I inserted an extra 1000 #\input texinfo lines at the top of each.
- I then had remaining errors in the files in pylib_ext/ so I added about #\input texinfo lines at the top of each.
- Finally my app submitted successfully!!!!
I must confess I really don't understand all of this. Feels like there's got to be a better way....
Thanks to @sreyemnayr for posting the script as I would not have solved this without it.
@peterdougstuart any chance you would be willing to share your updated template?
ccc last edited by
@sreyemnayr could you please update your script to work in Python 3?
ccc last edited by
if extension == '.py' or extension == '.pym' # can be shortened to: if extension in ('.py', '.pym')
Sorry for delay responding....
Happy to share @RomSpy, but please note that my templates have been minimized to my project i.e. deleted lots of modules I'm not using. Therefore it may not work for your project. You might be better off following the principles of steps I outlined above.
If you are still interested in my templates can you send me a PM on twitter (same handle as here) and I'll send you a zip or something. Alternatively point me to somewhere I can upload the files and I'll do so.
@peterdougstuart thank you for the offer. I ended up doing what you did and removing everything I wasn't using and then running the script above. I was able to successfully update my app in the store using that method in addition to your added steps. Thanks again.
xino386 last edited by
Hi, i am interested in your templates which is submitted to itunesConnect successfully, can you send me a zip ?
Hi @xino386 can you PM on twitter and I’ll send you a zip of my scripts (same handle as here). Note that my scripts have been trimmed down to match the subset of the libraries used in my app so you might be better off following the steps above to get it to work.
stepper last edited by stepper
I have a succes storey with your fix script (June 19/18). Here’s what I did. Pasted approx 400 #\input texinfo ‘s into a text file. Pasted the contents (all one line, no line breaks) into your script as a string (replacing the existing ‘#\input texinfo’. Changed the folder permissions for the two lib folders to read/write (temporarily).ran your script. It fixed all but a half dozen of the files. Went into the folders manually. The unchanged files stood out cause they didn’t have today’s date of modification. Manually pasted in the approx 400 from the original text file into those few.
Successfully uploaded to app connect where my app is now in Beta on test flight.
Thank You very much!
omz last edited by
@stepper That's interesting, though I've found a better solution recently: You can add something like
#import pythonistaas the first line, and it'll be recognized as Objective-C source code (which isn't directly executable, so it doesn't cause issues with code signing).
Any chance we will see an updated template that uses all the great new features in version 3?
omz last edited by
@RomSpy Yes, that's something I'd really like to do soon, and I've already done some of the work required, but it's still kind of a mess.
bg2b last edited by
Is there an up-to-date summary of the idea for packaging up a Pythonista project to make an app? I see that the template linked from
doesn't exist anymore, and I'm not sure where to begin learning.
yueqiw last edited by
@sreyemnayr I tried your re-worked script and it worked well.
#import pythonista\nrather than
To make it run faster I prepend 10 strings in each iteration, e.g.
f.write("#import pythonista\n" * 10 + source)and set counter to increment by 10.