UTF CSV parsing stripping characters

[justsomeguy]

I'm still doing research, but I thought I would ask here to see if anyone has run into this before. I'm trying to parse a CSV file that is saved as UTF-8 with BOM. This is the code I'm testing:

<?php header('Content-Type: text/html; charset=UTF-8'); ini_set('display_errors', 1);error_reporting(E_ALL); $fname = 'French_Language_Upload_File_full_utf.csv'; $contents = file_get_contents($fname); echo '<pre>' . print_r(mb_detect_order(), true) . '</pre><br>';echo 'encoding: ' . mb_detect_encoding($contents) . '<br>'; echo '<pre>' . $contents . '</pre><br>'; $csv = new SplFileObject($fname);$csv->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY); echo '<pre>';foreach ($csv as $row){  print_r($row);}echo '</pre>';

What I'm seeing is that some lines that start with certain characters are having those characters stripped. One of the lines in the CSV file contains this, for example: common,inactive_students,Inactive students,Inactive students,Étudiants inactifs After CSV parsing, it looks like this: Array( [0] => common [1] => inactive_students [2] => Inactive students [3] => Inactive students [4] => tudiants inactifs) This only happens with lines that start with the certain characters, if that same accented capital E appears in the middle of the text it does not get lost. It also doesn't strip all characters, this one is correct: Array( [0] => common [1] => ug_admin_news_permission [2] => At this time User Group Admins are only able to add news items to their own user groups. [3] => At this time User Group Admins are only able to add news items to their own user groups. [4] => À l’heure actuelle, les administrateurs peuvent seulement ajouter des nouveaux items à leurs propres groupes d’utilisateurs.) This line, though: common,r_u_sure_add,Are you sure you want to add the selected users?,Are you sure you want to add the selected users?,Êtes-vous sûr(e) de vouloir ajouter les utilisateurs sélectionnés ? Ends up like this: Array( [0] => common [1] => r_u_sure_add [2] => Are you sure you want to add the selected users? [3] => Are you sure you want to add the selected users? [4] => tes-vous sûr(e) de vouloir ajouter les utilisateurs sélectionnés ?) This also happens with fgetcsv. Has anyone run into this before? Any ideas? The server is running PHP 5.2.17.

[justsomeguy]

This problem does not appear if I wrap all strings with quotes, so now I have to figure out how to force Excel to put quotes around everything.