git.cworth.org Git - notmuch/blob - test/T300-encoding.sh

   1 #!/usr/bin/env bash
   2 test_description="encoding issues"
   3 . $(dirname "$0")/test-lib.sh || exit 1
   4
   5 test_begin_subtest "Message with text of unknown charset"
   6 add_message '[content-type]="text/plain; charset=unknown-8bit"' \
   7             "[body]=irrelevant"
   8 output=$(notmuch show id:${gen_msg_id} 2>&1 | notmuch_show_sanitize_all)
   9 test_expect_equal "$output" "\fmessage{ id:XXXXX depth:0 match:1 excluded:0 filename:XXXXX
  10 \fheader{
  11 Notmuch Test Suite <test_suite@notmuchmail.org> (2001-01-05) (inbox unread)
  12 Subject: Message with text of unknown charset
  13 From: Notmuch Test Suite <test_suite@notmuchmail.org>
  14 To: Notmuch Test Suite <test_suite@notmuchmail.org>
  15 Date: GENERATED_DATE
  16 \fheader}
  17 \fbody{
  18 \fpart{ ID: 1, Content-type: text/plain
  19 irrelevant
  20 \fpart}
  21 \fbody}
  22 \fmessage}"
  23
  24 test_begin_subtest "Search for ISO-8859-2 encoded message"
  25 add_message '[content-type]="text/plain; charset=iso-8859-2"' \
  26             '[content-transfer-encoding]=8bit' \
  27             '[subject]="ISO-8859-2 encoded message"' \
  28             "[body]=$'Czech word tu\350\362\341\350\350\355 means pinguin\'s.'" # ISO-8859-2 characters are generated by shell's escape sequences
  29 output=$(notmuch search tučňáččí 2>&1 | notmuch_show_sanitize_all)
  30 test_expect_equal "$output" "thread:0000000000000002   2001-01-05 [1/1] Notmuch Test Suite; ISO-8859-2 encoded message (inbox unread)"
  31
  32 test_begin_subtest "RFC 2047 encoded word with spaces"
  33 add_message '[subject]="=?utf-8?q?encoded word with spaces?="'
  34 output=$(notmuch search id:${gen_msg_id} 2>&1 | notmuch_show_sanitize)
  35 test_expect_equal "$output" "thread:0000000000000003   2001-01-05 [1/1] Notmuch Test Suite; encoded word with spaces (inbox unread)"
  36
  37 test_begin_subtest "RFC 2047 encoded words back to back"
  38 add_message '[subject]="=?utf-8?q?encoded-words-back?==?utf-8?q?to-back?="'
  39 output=$(notmuch search id:${gen_msg_id} 2>&1 | notmuch_show_sanitize)
  40 test_expect_equal "$output" "thread:0000000000000004   2001-01-05 [1/1] Notmuch Test Suite; encoded-words-backto-back (inbox unread)"
  41
  42 test_begin_subtest "RFC 2047 encoded words without space before or after"
  43 add_message '[subject]="=?utf-8?q?encoded?=word without=?utf-8?q?space?=" '
  44 output=$(notmuch search id:${gen_msg_id} 2>&1 | notmuch_show_sanitize)
  45 test_expect_equal "$output" "thread:0000000000000005   2001-01-05 [1/1] Notmuch Test Suite; encodedword withoutspace (inbox unread)"
  46
  47 test_begin_subtest "Mislabeled Windows-1252 encoding"
  48 add_message '[content-type]="text/plain; charset=iso-8859-1"'                           \
  49             "[body]=$'This text contains \x93Windows-1252\x94 character codes.'"
  50 cat <<EOF > EXPECTED
  51 \fmessage{ id:XXXXX depth:0 match:1 excluded:0 filename:XXXXX
  52 \fheader{
  53 Notmuch Test Suite <test_suite@notmuchmail.org> (2001-01-05) (inbox unread)
  54 Subject: Mislabeled Windows-1252 encoding
  55 From: Notmuch Test Suite <test_suite@notmuchmail.org>
  56 To: Notmuch Test Suite <test_suite@notmuchmail.org>
  57 Date: GENERATED_DATE
  58 \fheader}
  59 \fbody{
  60 \fpart{ ID: 1, Content-type: text/plain
  61 This text contains “Windows-1252” character codes.
  62 \fpart}
  63 \fbody}
  64 \fmessage}
  65 EOF
  66 notmuch show id:${gen_msg_id} 2>&1 | notmuch_show_sanitize_all > OUTPUT
  67 test_expect_equal_file EXPECTED OUTPUT
  68
  69 test_done