If you know what recursion is and like a challege, here's a puzzle
to keep you up nights. If not, you'll probably just want to mutter to
yourself "what a poor, unlucky bastard" and pass on by.
I'm parsing some data files into a PHP array and am stumped. I'm at
a loss for how to do this without grinding through character by character.
That would work, but my subconcious is nagging at me that there's got to be
a more elegant way to do it that I'm just not seeing, so I'm going to
describe the problem and ask for help before I start grinding.
The app I'm getting this from has 4 data types: int, string, array
and mapping (associative array).
Ints and strings are pretty straightforward, but there's no way to
tell 0 from null int or null string. This is an annoying limitation of the
app that just has to be ignored and dealt with by whatever gets this data
from us. This (among other reasons) make me glad PHP is weakly-typed.
Arrays are indexed from 0 and values can mix ints and strings
freely.
To start, mappings are arrays indexed by ints or strings. Mappings
aren't just arrays, though, they have a "width" (which is really a nested
array that I'm pretty certain is an ugly historical artifact.) Width allows
multiple values for one key and must be the same for all values in a
mapping, though the values (both of keys and their values) don't have to be
of the same type. Mappings can also mix ints and strings.
The tough part is that arrays and mappings can nest inside of each
other and the only characters quoted in strings are \, " and \n. This means
recursion must be used, but I just can't figure out a way to find the
boundaries of each element. Anyone with a clean way to do this (probably
with some kind of crazy regexp) will recieve my awe and gratitude.
Here's an example file[1]:
null_string 0
some_string "Fourscore and seven years ago..."
unset_int 0
an_int 42
negative_int -12
null_array ({ })
null_mapping ([ ])
easy_array ({9,22,"test",})
easy_mapping (["string":3,"foo":"bar",[12]:"I am not a crook!",])
medium_array ({"a string, containing a comma and a \"",23,})
medium_mapping (["str\"ing":3;5;7, 9:"Read my lips.";11;13;,])
hard_array ({"comma, string",({3,4,5,}),({"'\"str'",4,({3,4,({ }),}),}),})
hard_mapping ([17:"str";15,"foo":([ ]);17,"b'l\\a\nh":19;([21:23,]),"tour de
force":({29,31});({([ ])}),])
You may notice the last one is pathological[2]. Yes, PHP will really
let you use " and \n in array keys. The real data do sometimes get about
this complex; consider this a compressed version. As a fun fact, I've
learned vim's % command doesn't work when there's an odd number of double
quotes between your parens/braces.
Calling print_r on the array this generates would return:
Array
(
[null_string] => 0
[some_string] => "Fourscore and seven years ago..."
[unset_int] => 0
[an_int] => 42
[negative_int] => -12
[null_array] => Array
(
)
[null_mapping] => Array
(
)
[easy_array] => Array
(
[0] => 9
[1] => 22
[0] => test
)
[easy_mapping] => Array
(
[string] => 3
[foo] => "bar"
[12] => "I am not a crook!"
)
[medium_array] => Array
(
[0] => a string, containing a comma and a "
[1] => 23
)
[medium_mapping] => Array
(
[string] => Array
(
[0] => 3
[1] => 5
[2] => 7
)
[9] => Array
(
[0] => Read my lips.
[1] => 11
[2] => 13
)
)
[hard_array] => Array
(
[0] => comma, string
[1] => Array
(
[0] => 3
[1] => 4
[2] => 5
)
[2] => Array
(
[0] => '"str'
[1] => 4
[2] => Array
(
[0] => 3
[1] => 4
[2] => Array
(
)
)
)
)
[hard_mapping] => Array
(
[17] => Array
(
[0] => str
[1] => 15
)
[foo] => Array
(
[0] => Array
(
)
[1] => 17
)
[b'l\a\nh] => Array
(
[0] => 19
[1] => Array
(
[21] => 23
)
)
[tour de force] => Array
(
[0] => Array
(
[29] => 31
)
[1] => Array
(
[0] => Array
(
)
)
)
)
)
The truly eagle-eyed will note that print_r doesn't really escape \n
when outputting, which is ugly and should probably have a bug report filed
on it. I've ignored this behavior of print_r to produce more readable
output.
Anyone who gets the 'tour de force' can contact me for the 'victory
lap' puzzle if it didn't occur to them while solving this one. It's equally
frustrating.
[1] Brownie points to anyone who recognizes the app that outputs this data
format. Yep, I'm doing web integration. Feel free to contact me off-list if
you've any ideas/code to share or would like to see what I'm doing. Yes,
technically, mappings can be indexed by an array or mapping. I've ignored
this because 1. PHP can't do this and 2. Anyone doing this will be dragged
out into the street and shot.
[2] http://tuxedo.org/~esr/jargon/html/entry/pathological.html
--
PHP General Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php
