commit eaa9623f8a7b6354e1dd00d878c7182c30028b5c
author Akron <nils@diewald-online.de> Thu Oct 15 17:06:15 2020 +0200
committer Marc Kupietz <kupietz@ids-mannheim.de> Tue Nov 24 14:03:43 2020 +0100
tree a87fd00411f05fc548fe6bade2ef13a40dfddf3d
parent 7c88a7bb1dbc34e4c9e6546e4274f3be202f5ecd

Switch input encoding based on XML processing instruction

Change-Id: I89e20c8af762615d37c216b0c89227fc3644fcb3

t/script.t

diff --git a/t/script.t b/t/script.t
index 88b3d2d..6bf98e4 100644
--- a/t/script.t
+++ b/t/script.t
@@ -306,11 +306,11 @@
   my (undef, $outzip) = korap_tempfile('script_out5');
 
   # because output 'textid=...' goes to STDERR (see script/tei2korapxml)
-  binmode STDERR, qw{ :encoding(UTF-8) };
+  binmode STDERR;
 
   stderr_like(
     sub { `cat '$tplfile' | perl '$script' -ti > '$outzip'` },
-    qr!tei2korapxml: .*? text_id=$text_sigle_lax!, # see above: print $fh encode_utf8($tpl);
+    qr!tei2korapxml: .*? text_id=$text_sigle_esc!, # see above: print $fh encode_utf8($tpl);
   );
 };
 
@@ -408,6 +408,32 @@
 };
 
 
+subtest 'Check input encoding' => sub {
+
+  # Load example file
+  test_tei2korapxml(
+    file => catfile($f, 'data', 'goe_sample.i5.xml'),
+    env => 'KORAPXMLTEI_INLINE=1',
+    tmp => 'script_utf8_enc'
+  )
+    ->stderr_like(qr!tei2korapxml: .*? text_id=GOE_AGA\.00000!)
+    ->unzip_xml('GOE/AGA/00000/data.xml')
+    ->content_like(qr/\Q"Kriegstheater"\E/)
+    ->content_like(qr/\QTür'\E/)
+    ;
+
+  test_tei2korapxml(
+    file => catfile($f, 'data', 'goe_sample.i5.iso.xml'),
+    env => 'KORAPXMLTEI_INLINE=1',
+    tmp => 'script_iso_enc'
+  )
+    ->stderr_like(qr!tei2korapxml: .*? text_id=GOE_AGA\.00000!)
+    ->unzip_xml('GOE/AGA/00000/data.xml')
+    ->content_like(qr/\Q"Kriegstheater"\E/)
+    ->content_like(qr/\QTür'\E/)
+    ;
+};
+
 subtest 'Test Log' => sub {
   test_tei2korapxml(
     tmp => 'script_out',