Titlecase.py: Titlecase in python | Comments (21)
Posted in Code on 27th May 2008, 8:29 am by Stuart
John Gruber recently published a perl script to convert strings into title case avoiding capitalizing small words based on rules from the New York Times Manual of style as well as catering for several special cases.
Before porting the perl script I tried out Python’s inbuilt title string method to see how well it handled text according to the various rules:
>>> test="this is zed's favorite outburst"
>>> test.title()
"This Is Zed'S Favorite Outburst"As you can see it really can’t cope with anything remotely complicated.
I originally knocked up a direct port of the script, but I found it a tad unwieldy and so I decided on a fresh approach which processes the text after splitting the strings on whitespace characters.
Before I started both approaches I wrote test cases for all of the examples John gave in his post, and added a few of my own some based on John’s archive of posts later. This made coding a lot easier as it meant once the tests stopped failing I could stop coding. Of course with something like this there’s likely to be more cases that need to be catered for, if you have some ideas as to new test cases please add them in the comments.
The script is flexible from the point of view that you can either import it and use the function itself directly e.g:
>>> from titlecase import titlecase
>>> titlecase('a thing')
'A Thing'You can pass a file of text to stdin:
$ ./titlecase.py < ~/title-case-examples
Q&A With Steve Jobs: 'That's What Happens in Technology'
What Is AT&T's Problem?
Apple Deal With AT&T Falls Through
This v That
This vs That
This v. That
This vs. That
The SEC's Apple Probe: What You Need to Know
'By the Way, Small Word at the Start but Within Quotes.'
Small Word at End Is Nothing to Be Afraid of
Starting Sub-Phrase With a Small Word: A Trick, Perhaps?
Sub-Phrase With a Small Word in Quotes: 'A Trick, Perhaps?'
Sub-Phrase With a Small Word in Quotes: "A Trick, Perhaps?"
"Nothing to Be Afraid Of?"
"Nothing to Be Afraid Of?"
A Thing
2lmc Spool: 'Gruber on OmniFocus and Vapo(u)rware'Finally running the script from the command line without args results in the tests being run:
$ ./titlecase.py
Testing: a thing ... ok
Testing: Apple Deal With AT&T Falls Through ... ok
Testing: The SEC's Apple Probe: What You Need to Know ... ok
Testing: What Is AT&T's Problem? ... ok
Testing: this is just an example.com ... ok
Testing: this is something listed on an del.icio.us ... ok
Testing: Generalissimo Francisco Franco: Still Dead; Kieren McCarthy: Still a Jackass ... ok
Testing: iTunes should be unmolested ... ok
Testing: "Nothing to Be Afraid of?" ... ok
Testing: "Nothing to Be Afraid Of?" ... ok
Testing: Q&A With Steve Jobs: 'That's What Happens In ... ok
Testing: Seriously, ‘Repair Permissions’ Is Voodoo ... ok
Testing: Sub-Phrase With a Small Word in Quotes: "a Trick, ... ok
Testing: Small word at end is nothing to be afraid of ... ok
Testing: 'by the Way, small word at the start but within ... ok
Testing: Sub-Phrase With a Small Word in Quotes: 'a Trick, ... ok
Testing: Starting Sub-Phrase With a Small Word: a Trick, ... ok
Testing: this v that ... ok
Testing: this v. that ... ok
Testing: this vs that ... ok
Testing: this vs. that ... ok
Testing: Reading Between the Lines of Steve Jobs’s ‘Thoughts on ... ok
Testing: 2lmc Spool: 'Gruber on OmniFocus and Vapo(u)rware' ... ok
----------------------------------------------------------------------
Ran 23 tests in 0.019s
To get the code check it out with bzr with the command:
bzr branch lp:titlecase.py
OR:
easy_install titlecase
Or download it from here: https://launchpad.net/titlecase.py/trunk/0.2
[...] Titlecase.py: Titlecase in python by Stuart Colville John Gruber recently published a perl script to convert strings into title case avoiding capitalizing small words based on rules from the New York Times Manual of style as well as catering for several special cases. (tags: software opensource python titlecase) [...]
Very cool! Do you know of a Django filter version of this script anywhere?
I’m somewhat of a novice but I seem to observe that the tests pass because the inputs are already in title case. If i add text=text.lower() at the start of the function the test cases fail.
@James: It’s not quite that simple as lowering all of the text at input is likely to break quite a lot of tests as some of the functionality takes into account the case of the input.
However you do have a point that a few of the test cases had the same input as the expected output and in the process of addressing that I found a small bug where a special quote was missing from the punct regex.
I’ve updated the tests, cleaned up the code and released it as version 0.2.
https://launchpad.net/titlecase.py/trunk/0.2
Thanks for your comment — it was definitely a worthwile observation!
@Tom Watson: Have a look at Christian Metts’ Typogrify library. He recently added support for Titlecase.py.
Sweet! I’ve been looking for this…
Two questions: (1) If I pass in unicode strings, is any damage done to the unicode characters? (2) Is titling logic applied to the unicode characters where capitalization makes sense?
Please put this in the cheese shop.
@nerkles: Done – you can now get it from the cheese shop with:
easy_install titlecaseStuart,
You might want to add
my name is o'briento you test, since at the moment it’s a fail// Tony
Hi!
handy util! i was wondering, would it be possible to add support for cases like WASHINGTON D.C.?
the script has come very handy so far!
best,
Diederik
Seems like this script is missing things like 34Th Street (capitalizing the T)
newlines (more specifically, all characters which satisfy “\s” in regex) are replaced by spaces.
Otherwise, this is great! Thanks a ton!
@brian, I really don’t think that that T should be capitalized. At least, I’ve never seen it that way.
also, “word/word” becomes “Word/word”. I think it should be “Word/Word”.
@george: I’ve fixed the newlines issue in 0.5.1.
The biggest problem with this script is that it becomes necessary to special case lots of edge cases. As a result of that I’ll probably look at providing some way for the user to easily add their own customisations.
Further to George’s point. Songs in the titles of medleys are customarily joined with a slash, so it not a particularly esoteric case. “dance with me/let’s face the music and dance” fails. A disappointment to me, ‘cos song-titles was what I was hoping to use it for.
Titles with numbers in them fail (for example “HELLO 10 WORLD”).
@george and michael: The slashed case has now been addressed – see the version in the trunk branch.
@Troy: Based on the rule-set HELLO 10 WORLD will be untouched as it’s been explicitly capitalised.
Shouldn’t:
SMALL_LAST = re.compile(r’\b(%s)%s?$’ % (SMALL, PUNCT), re.I)
Be instead:
SMALL_LAST = re.compile(r’\b(%s)(%s)?$’ % (SMALL, PUNCT), re.I)
Oh… m.group(0).capitalize(), Nevermind! xD
Fix for “Step-by-Step”: http://codepad.org/e671lRhl