Bug #55200 [Opn->Ver]: str_getcsv parses lines incorrectly

Edit report at https://bugs.php.net/bug.php?id=55200&edit=1 ID: 55200 Updated by: cmb@php.net Reported by: dmitry dot dulepov at gmail dot com Summary: str_getcsv parses lines incorrectly -Status: Open +Status: Verified Type: Bug -Package: Unknown/Other Function +Package: Strings related PHP Version: 5.3.6 Block user comment: N Private report: N New Comment: The given test input is not valid according to RFC 4180; whitespace is not allowed between 'COMMA' and 'escaped'. However, the current behavior of str_getcsv() doesn't make sense to me, see <http://3v4l.org/roJnp> and <http://3v4l.org/XcrLG>. Returning FALSE or at least giving a warning or notice for such input seems to be more appropriate. By the way, fgetcsv() shows the same behavior. Previous Comments: ------------------------------------------------------------------------ [2011-07-29 05:42:34] dmitry dot dulepov at gmail dot com It is not about input validation :) Your example is clearly invalid input. The function should fail and return FALSE. My example perfectly fits into formal CSV grammar, thus it is valid input. It is just parsed incorrectly. I would not send you a bug about invalid input :) ------------------------------------------------------------------------ [2011-07-29 01:44:30] lonnyk at gmail dot com str_getcsv is a conversion tool and not an input validator. Any invalid input is surely not going to work correctly and I do not think it is good practice to account for bad input b/c it would cause extra processing time when it is not needed. If the input were: var_dump(str_getcsv('"123" a , "456" ', ',', '"')) what would you expect str_getcsv to do? ------------------------------------------------------------------------ [2011-07-13 09:10:16] dmitry dot dulepov at gmail dot com Description: ------------ Putting a space around the separator *and* using quotes around fields adds spaces to the field. The following line: "123" , "456" should produce: "123" and "456" but it makes: "123 " and " 456". In the RFC4180 the specification suggests that if the field contains quotes, only the text inside quotes is the content of the field. Here is the formal gramma: record = field *(COMMA field) field = (escaped / non-escaped) escaped = DQUOTE *(TEXTDATA / COMMA / CR / LF / 2DQUOTE) DQUOTE non-escaped = *TEXTDATA Thus spaces should appear in field only if the field is not quoted. Test script: --------------- print_r(str_getcsv('"123" , "456" ', ',', '"')) Expected result: ---------------- array( 0 => "123", 1 => "456", ) Actual result: -------------- array( 0 => "123 ", 1 => "456 ", ) ------------------------------------------------------------------------ -- Edit this bug report at https://bugs.php.net/bug.php?id=55200&edit=1