richxerox 1.0.1

Usage

from richxerox import *

print available() # what kind of data is on the clipboard?

print paste()     # get data in the default format ('text')
print paste(format='text')  # get text (Unicode)
print paste(format='rtf')   # get RTF
print paste(format='html')  # get HTML

print "ALL CONTENTS:\n", pasteall()

clear()
print "ALL CONTENTS AFTER CLEAR:\n", pasteall()

r = "{\\rtf1\\ansi\\ansicpg1252\\cocoartf1187\\cocoasubrtf390\n" \
    "{\\fonttbl\\f0\\froman\\fcharset0 Times-Roman;}\n{\\colortbl;" \
    "\\red255\\green255\\blue255;}\n\\deftab720\n\\pard\\pardeftab720" \
    "\n\n\\f0\\fs24 \\cf0 This is \n\\b good\n\\b0 !}"
h = "this is <strong>good</strong>!"
copy(text="this is good!", html=h, rtf=r)

print "ALL CONTENTS AFTER COPY:\n", pasteall()

The API is modeled on that of xerox, with simple copy() and paste() operations.

Think of paste() as pasting into your program and copy() as copying from your program.

The main difference in the API is that, given the different formats used in rich text, one must specify the format provided or needed if it is not plain text. This is done through keyword-style arguments.

Alternative API

If you prefer an object-oriented API:

from richxerox import *

print pasteboard.get_contents(format='html')    # paste
pasteboard.clear()                              # clear
pasteboard.set_contents(text="this is good!",   # copy
                        html=h, rtf=r)

print pasteboard.get_all_contents()       # pasteall

Background

I searched long and hard but couldn’t find a simple Python module that made copy and paste operations on macOS easy and straightforward. xerox works well, but it only supports plain text. What about browsers and word processors that export rich text with hyperlinks, styles, and so on? How can you access that data?

After banging my head against this a few times, I eventually found code samples I could adapt and make work without understanding the entirety of Apple’s Foundation and AppKit. This module is the result.

Descent Into RTF

Even in this HTML-everywhere age, Apple and macOS apps are unfortunately RTF-centric.

• In my experience, RTF is often not robustly passed between applications. Different apps interpret or render the same RTF differently, so font sizes and other characteristics change.

• RTF is extremely verbose. Microsoft Word, for instance, emits 29,807 characters as the copy/cut representation of “This is good!” Microsoft is known for verbose exports, and RTF itself attempts to represent whole documents rather than individual snippets. Still, that’s roughly 1,000x (a.k.a. three decimal orders of magnitude) as verbose as HTML. Try copying existing text in some application, then running pasteall() to get your own taste of this madness.

• If you put multiple forms of text on the clipboard, you don’t have much if any control which one an application will use when you ask it to “paste” data. If you want a single format, better to just put that one format on the clipboard.

While Mac apps occasionally put HTML contents on the pasteboard, RTF seems to be the most common lingua franca. I’ve not found any particularly good, robust, or up-to-date Python tools for parsing and transforming RTF. The handy textutil tool will, however, convert an RTF file into quite clean HTML, like so:

textutil -convert html temp.rtf

yielding temp.html. This can be parsed and manipulated with lxml or your favorite HTML/XML library.

Notes

• Version 1.0.1 withdraws the wheel packaging from distribution. Wheels are in general great, but it was reported to be causing installability breakage depending on the local install status of Mac foundation libraries, given that wheels are a preferred and somewhat canned install format. Withdrawing wheels should force a more local-system-centered set of dependencies to be installed. If you still have trouble, try installing pyobjc manually before installing richxerox.

• Version 1.0.0 updates the testing matrix. Latest versions of 2.7, 3.3, 3.4, 3.5, and 3.6 are confirmed working. Old, pre-[SemVer](http://semver.org/) versions have been removed from PyPI; they were causing some install problems. Python 3.2 has been withdrawn from support as both ancient and no longer being properly supported in my local test rig.)

• As of version 0.6, much more robust handling of Unicode characters. Better auto-install, including installing foundation pyobjc module if necessary. (pyobjc auto-install only works reliably on Python 2.7 and above, so official support for Python 2.6 has been withdrawn.)

• If the underlying pyobjc library needs to be installed, the process will take a long time. For example, 4 hours 7 minutes. Don’ty just get coffee while it’s installing. Take lunch. A long, languorous lunch. And then maybe have a nap.

• Version 0.5 had a mistake in Unicode handling. Even though it passed all tests, it over-quoted Unicode coming from real apps. Fixed.

• Code inspired by and/or based on Genba’s Reading URLs from OS X clipboard with PyObjC and Carl M. Johnson’s copy_paste.py

• See also NSPasteboard docs, a discussion on UTIs, and John Siracusa’s discussion of the evolution of Mac OS types

• The author, Jonathan Eunice or @jeunice on Twitter welcomes your comments and suggestions.

Installation

To install the latest version:

pip install -U richxerox

In some cases with multiple Python installations, pip and the Python ininstallation can become disconnected. If the above does not work, try installing through Python directly. You can also specify a particular version of Python to install under.:

python -m pip install -U richxerox

(You may need to prefix these with sudo to authorize installation.)