skip to navigation
skip to content

Not Logged In

say 0.407

Simple printing with templates. E.g.: say("Hello, {whoever}!", indent=1)

Latest Version: 1.2.1

print, format, and %, evolved.

Q: It's been forty years since C introduced printf() and the basic formatted printing of positional parameters. Isn't it time for an upgrade?

A: Yes! ZOMG, yes!

say supplements or replaces Python's print statement/function, format function/method, and % string interpolation operator with higher-level facilities:

  • Straightforward string formatting with DRY, Pythonic templates that piggyback the built in format() method, formatting syntax, and well-proven underlying engine.
  • A single output mechanism compatible with both Python 2.x and Python 3.x.
  • Indentation (to help stucture output)
  • Convenience printing functions for horizontal rules (lines), titles, and vertical whitespace.

Usage

from say import say, fmt

x = 12
nums = list(range(4))

say("There are {x} things.")
say("Nums has {len(nums)} items: {nums}")

yields:

There are 12 things.
Nums has 4 items: [1, 2, 3, 4]

say is basically a simpler, nicer recasting of:

print "There are {} things.".format(x)
print "Nums has {} items: {}".format(len(nums), nums)

(NB in Python 2.6 one must number each of the {} placeholders--e.g. "Nums has {0} items: {1}"-- in order to avoid a ValueError: zero length field name in format error. Python 2.7 and later assume the placeholders are sequential.)

The more items that are being printed, and the complicated the format invocation, the more valuable having it stated in-line becomes. Note that full expressions are are supported. They are evaluated in the context of the caller.

Printing Where You Like

say() writes to a list of files--by default just sys.stdout. But with it simple configuration call, it will write to different--even multiple--files:

from say import say, stdout

say.setfiles([stdout, "report.txt"])
say(...)   # now prints to both stdout and report.txt

This has the advantage of allowing you to both capture and see program output, without changing any code (other than the config statement). You can also define your own targeted Say instances:

from say import say, Say, stderr

err = say.clone().setfiles([stderr, 'error.txt'])
err("Failed with error {errcode}")  # writes to stderr, error.txt

Note that stdout and stderr are just convenience aliases to the respective sys equivalents.

Printing When You Like

If you want to stop printing for a while:

say.set(silent=True)  # no printing until set to False

Or transiently:

say(...stuff..., silent=not verbose) # prints iff bool(verbose) is True

Of course, you don't have to print to any file. There's a predefined sayer fmt() that works exactly like say() and inherits most of its options, but doesn't print. (The C analogy: say : fmt :: printf : sprintf.)

Indentation

Indentation is a common way to display data hierarchically. say will help you manage it. For example:

say('TITLE')
for item in items:
    say(item, indent=1)

will indent the items by one indentation level (by default, each indent level is four spaces, but you can change that with the indent_str option).

If you want to change the default indentation level:

say.set(indent=1)      # to an absolute level
say.set(indent='+1')   # strings => set relative to current level

...

say.set(indent=0)      # to get back to the default, no indent

Or you can use a with construct:

with say.settings(indent='+1'):
    say(...)

    # anything say() emits here will be auto-indented +1 levels

While it's easy enough for any print statement or function to have a few space characters added to its format string, it's easy to mistakenly type too many or too few spaces, or to forget to type them in some format strings. And if you're indenting strings that themselves may contain multiple lines, the simple print approach breaks because won't take multi-line strings into account.

say, however, simply handles the indent level, and it properly handles the multi-line string case. Their subsequent lines will be just as nicely and correctly indented as the first one--something not otherwise easily accomplished without adding gunky, complexifying string manipulation code to every place in your program that prints strings.

This starts to illustrate the "do the right thing" philosophy behind say. So many languages' printing and formatting functions a restricted to "outputting values" at a low level. They may format basic data types, but they don't provide straightforward ways to do neat text transformations like indentation that let programmers rapidly provide correct, highly-formatted ouput. Over time, say will provide higher-level formatting options. For now: indentation.

Encodings

say() and fmt() try to work with Unicode strings, for example providing them as return values. But character encodings remain a fractious and often exasperating part of IT. When writing formatted strings, say handles this by encoding into utf-8.

If you are using strings containing utf-8 rather than Unicode characters, say may complain. But it complains in the same places the built-in format() does, so no harm done. (Python 3 doesn't generally allow utf-8 in strings, so it's cleaner on this front.)

You can get creative with the encoding:

say('I am a truck!', encoding='base64')  # SSBhbSBhIHRydWNrIQo=

Or change the default:

say.set(encoding='rot-13')

Knock yourself out with all the exciting opportunites! If you really want the formatted text returned just as it is written to files, use the encoded option. Set to True and it returns text in the output encoding. Or set to an actual encoding name, and that will be the return encoding.

say() returns the formatted text with one small tweak: it removes the final newline if a newline is the very last character. Though odd, this is exactly what you need if you're going to print or say the resulting text without a gratuitous "extra" newline.

Titles and Horizontal Rules

say defines a few convenience formatting functions:

say.title('Errors', sep='-')
for i,e in enumerate(errors, start=1):
    say("{i:3}: {e['name'].upper()}")

might yield:

--------------- Errors ---------------
  1: I/O ERROR
  2: COMPUTE ERROR

A similar method hr produces just a horizontal line, like the HTML <hr> element. For either, one can optionally specify the width (width), character repeated to make the line (sep), and vertical separation/whitespace above and below the item (vsep). Good options for the separator might be be '-', '=', or parts of the Unicode box drawing character set.

Python 3

Say works virtually the same way in Python 2 and Python 3. This can simplify software that should work across the versions, without all the from __future__ import print_function hassle.

say attempts to mask some of the quirky compexities of the 2-to-3 divide, such as string encodings and codec use.

Alternatives

  • ScopeFormatter provides variable interpolation into strings. It is amazingly compact and elegant. Sadly, it only interpolates Python names, not full expressions. say has full expressions, as well as a framework for higher-level printing features beyond ScopeFormatter's...um...scope.

  • Even simpler are invocations of % or format() using locals(). E.g.:

    name = "Joe"
    print "Hello, %(name)!" % locals()
    # or
    print "Hello, {name}!".format(**locals())
    

    Unfortunately this has even more limitations than ScopeFormatter: it only supports local variables, not globals or expressions. And the interpolation code seems gratuitous. Simpler:

    say("Hello, {name}!")
    

Notes

  • The say name was inspired by Perl's say, but the similarity stops there.
  • Automated multi-version testing with the wonderful pytest and tox modules has commenced. say is now successfully packaged for, and tested against, all late-model verions of Python: 2.6, 2.7, 3.2, and 3.3.
  • say has greater ambitions than just simple template printing. It's part of a larger rethinking of how output should be formatted. Stay tuned.
  • In addition to being a practical module in its own right, say is testbed for options, a package that provides high-flexibility option, configuration, and parameter management.
  • The author, Jonathan Eunice or @jeunice on Twitter welcomes your comments and suggestions.

Installation

pip install say

To easy_install under a specific Python version (3.3 in this example):

python3.3 -m easy_install say

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

 
File Type Py Version Uploaded on Size
say-0.407.tar.gz (md5) Source 2012-10-18 11KB
say-0.407.zip (md5) Source 2012-10-18 21KB
  • Downloads (All Versions):
  • 282 downloads in the last day
  • 920 downloads in the last week
  • 5646 downloads in the last month