Add tests for pbmtext -wchar

git-svn-id: http://svn.code.sf.net/p/netpbm/code/trunk@3189 9d0c8265-081b-0410-96cb-a4ca84ce46f8

10 files changed, 262 insertions, 100 deletions

diff --git a/test/Execute-Tests b/test/Execute-Tests
index e69c84e7..7a65d51f 100755
--- a/test/Execute-Tests
+++ b/test/Execute-Tests
@@ -180,7 +180,7 @@ elif [ $VALGRIND_TESTS = "on" ]
   vg_command_base="valgrind --trace-children=yes";
 
   for i in awk cat cksum cmp cp cut date dirname egrep fgrep file grep gs \
-    head mkdir mktemp perl rm sed seq sh tee testrandom tr uniq \
+    head iconv mkdir mktemp perl rm sed seq sh tee testrandom tr uniq \
     Available-Testprog
 
     # Tell valgrind not to probe execution of the above programs.
diff --git a/test/Test-Order b/test/Test-Order
index 9b280d12..6abc2287 100644
--- a/test/Test-Order
+++ b/test/Test-Order
@@ -12,6 +12,9 @@ pamseq.test
 
 pbmpage.test
 pbmtext.test
+pbmtext-bdf.test
+pbmtext-iso88591.test
+pbmtext-utf8.test
 pbmupc.test
 pgmramp.test
 pamgauss.test
diff --git a/test/pbmtext-bdf.ok b/test/pbmtext-bdf.ok
new file mode 100644
index 00000000..5ab8b4af
--- /dev/null
+++ b/test/pbmtext-bdf.ok
@@ -0,0 +1,8 @@
+386826492 35
+1
+1
+1
+1
+1
+1
+1
diff --git a/test/pbmtext-bdf.test b/test/pbmtext-bdf.test
new file mode 100755
index 00000000..1bd7c52c
--- /dev/null
+++ b/test/pbmtext-bdf.test
@@ -0,0 +1,99 @@
+#! /bin/bash
+# This script tests: pbmtext
+# Also requires:
+
+tmpdir=${tmpdir:-/tmp}
+
+font_bdf=${tmpdir}/font.bdf
+font_corrupt_bdf=${tmpdir}/fontcorrupt.bdf
+
+# Though this BDF font file defines only three letters, it is valid.
+
+cat > ${font_bdf} << EOF
+STARTFONT 2.1
+COMMENT simple font for pbmtext test
+COMMENT derived from: $XFree86: xc/fonts/bdf/misc/micro.bdf,v 1.1 1999/09/25 14:36:34 dawes Exp $
+FONT test
+SIZE 4 75 75
+FONTBOUNDINGBOX 4 5 0 0
+STARTPROPERTIES 5
+FONT_DESCENT 0
+FONT_ASCENT 5
+CHARSET_REGISTRY "ISO88591"
+CHARSET_ENCODING "1"
+COPYRIGHT "Public domain font.  Share and enjoy."
+ENDPROPERTIES
+CHARS 3
+STARTCHAR A
+ENCODING 65
+SWIDTH 1000 0
+DWIDTH 4 0
+BBX 4 5 0 0
+BITMAP
+e0
+a0
+e0
+a0
+a0
+ENDCHAR
+STARTCHAR B
+ENCODING 66
+SWIDTH 1000 0
+DWIDTH 4 0
+BBX 4 5 0 0
+BITMAP
+e0
+a0
+c0
+a0
+e0
+ENDCHAR
+STARTCHAR C
+ENCODING 67
+SWIDTH 1000 0
+DWIDTH 4 0
+BBX 4 5 0 0
+BITMAP
+e0
+80
+80
+80
+e0
+ENDCHAR
+ENDFONT
+EOF
+
+# Test 1
+# This should succeed and produce 386826492 35
+pbmtext -font ${font_bdf} ABC | cksum
+
+
+# Test 2
+# The rest should all fail.  Writes 1 seven times.
+
+echo "Test whether corrupted BDF font files are properly handled." 1>&2
+echo "Error messages will appear." 1>&2
+echo 1>&2
+
+pbmtext -font ${font_bdf} BCD
+echo $?
+
+sed 's/FONTBOUNDINGBOX 4 5 0 0/FONTBOUNDINGBOX 4 4 0 0/' \
+  ${font_bdf} >  ${font_corrupt_bdf}
+pbmtext -font ${font_corrupt_bdf} ABC > /dev/null
+echo $?
+rm ${font_corrupt_bdf}
+
+sed 's/BBX 4 5 0 0/BBX 4 6 0 0/' \
+  ${font_bdf} >  ${font_corrupt_bdf}
+pbmtext -font ${font_corrupt_bdf} ABC > /dev/null
+echo $?
+rm ${font_corrupt_bdf}
+
+for delete_line in 14 16 18 20
+  do
+  sed "${delete_line}"d ${font_bdf} >  ${font_corrupt_bdf}
+  pbmtext -font ${font_corrupt_bdf} ABC > /dev/null
+  echo $?
+  rm ${font_corrupt_bdf}
+  done
diff --git a/test/pbmtext-iso88591.ok b/test/pbmtext-iso88591.ok
new file mode 100644
index 00000000..d1516357
--- /dev/null
+++ b/test/pbmtext-iso88591.ok
@@ -0,0 +1,4 @@
+3491766365 5110
+3491766365 5110
+259944121 191
+259944121 191
diff --git a/test/pbmtext-iso88591.test b/test/pbmtext-iso88591.test
new file mode 100755
index 00000000..34346f5c
--- /dev/null
+++ b/test/pbmtext-iso88591.test
@@ -0,0 +1,46 @@
+#! /bin/bash
+# This script tests: pbmtext
+# Also requires:
+
+# This test requires the following locale:
+#   LC_ALL en_US.iso88591
+# Skip this test if it is not available
+
+iconv /dev/null
+if [ $? -ne 0  ]
+  then echo "iconv command not available." 1>&2
+       echo "Skipping." 1>&2
+  exit 80;
+fi
+
+echo "A" | LC_ALL=en_US.iso88591 pbmtext -wchar > /dev/null
+if [ $? -ne 0  ]
+  then echo "LC_ALL could not be set to en_US.iso88591" 1>&2
+       echo "Skipping." 1>&2
+  exit 80;
+fi
+
+# Two rows
+# Should print 3491766365 5110 twice
+LC_ALL=C \
+awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i); print ""; \
+             for (i=160;i<=255;++i) printf("%c",i); }' | \
+    pbmtext -builtin bdf | cksum
+
+
+LC_ALL=C \
+awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i);  print ""; \
+             for (i=160;i<=255;++i) printf("%c",i); }' | \
+    LC_ALL=en_US.iso88591 pbmtext -builtin bdf -wchar | cksum
+
+
+# Two rows
+# Should print 259944121 191 twice
+LC_ALL=C \
+awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i); print ""; \
+             for (i=161;i<=255;++i) printf("%c",i); print "" }' | cksum
+
+LC_ALL=C \
+awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i); print ""; \
+             for (i=161;i<=255;++i) printf("%c",i); print ""}' | \
+    LC_ALL=en_US.iso88591 pbmtext -builtin bdf -wchar -text-dump | cksum
\ No newline at end of file
diff --git a/test/pbmtext-utf8.ok b/test/pbmtext-utf8.ok
new file mode 100644
index 00000000..864c530a
--- /dev/null
+++ b/test/pbmtext-utf8.ok
@@ -0,0 +1,8 @@
+1240895458 5110
+1240895458 5110
+898975479 2272
+898975479 2272
+0
+ !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}
+ !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}
+1
diff --git a/test/pbmtext-utf8.test b/test/pbmtext-utf8.test
new file mode 100755
index 00000000..ca1f45a2
--- /dev/null
+++ b/test/pbmtext-utf8.test
@@ -0,0 +1,91 @@
+#! /bin/bash
+# This script tests: pbmtext
+# Also requires:
+
+# This test requires the en_US.utf8 locale
+# Skip this test if it is not available
+
+iconv /dev/null
+if [ $? -ne 0  ]
+  then echo "iconv command not available." 1>&2
+       echo "Skipping." 1>&2
+  exit 80;
+fi
+
+echo "A" | LC_ALL=en_US.utf8 pbmtext -wchar > /dev/null
+if [ $? -ne 0  ]
+  then echo "LC_ALL could not be set to en_US.utf8." 1>&2
+       echo "Skipping." 1>&2
+  exit 80;
+fi
+
+# Test 1.
+# Two rows
+# Should print 1240895458 5110 twice
+LC_ALL=C \
+awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i); print ""; \
+             for (i=161;i<=255;++i) printf("%c",i); }' | \
+    pbmtext -builtin bdf | cksum
+
+LC_ALL=C \
+awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i); print ""; \
+             for (i=161;i<=255;++i) printf("%c",i);  }' | \
+    iconv -f iso8859-1 -t utf-8 | \
+    LC_ALL=en_US.utf8 pbmtext -builtin bdf -wchar | cksum
+
+
+# Test 2.
+# One row
+# Should print 898975479 2272 twice
+LC_ALL=C \
+awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i); print "" }' | \
+    pbmtext -builtin bdf | cksum
+
+LC_ALL=C \
+awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i);  print ""}' | \
+    LC_ALL=en_US.utf8 pbmtext -builtin bdf -wchar | cksum
+
+
+tmpdir=${tmpdir:-/tmp}
+output=${tmpdir}/output
+
+
+# Test 3.
+# Two rows
+# Output may be affected by locale.  Compare with cmp.
+# Should print 0
+LC_ALL=C \
+awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i); print ""; \
+             for (i=161;i<=255;++i) printf("%c",i); print "" }' | \
+    iconv -f iso88591 -t utf8 > ${output}
+
+LC_ALL=C \
+awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i); print ""; \
+             for (i=161;i<=255;++i) printf("%c",i); print "" }' | \
+    iconv -f iso8859-1 -t utf-8 | \
+    LC_ALL=en_US.utf8 pbmtext -builtin bdf -wchar -text-dump | \
+    cmp --quiet - ${output}
+
+echo $?
+rm ${output}
+
+
+# Test 4.
+# One row
+# Should print the following twice:
+# !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}
+LC_ALL=C \
+awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i); print "" } '
+
+LC_ALL=C \
+awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i);  print ""}' | \
+        LC_ALL=en_US.utf8 pbmtext -builtin bdf -wchar -text-dump
+
+
+# Test 5.
+# Invalid utf-8 sequence
+# Should print 1
+LC_ALL=C \
+awk 'BEGIN { for (i=128; i<=129;++i) printf("%c",i);  print ""}' | \
+        LC_ALL=en_US.utf8 pbmtext -builtin bdf -wchar -text-dump
+echo $?
diff --git a/test/pbmtext.ok b/test/pbmtext.ok
index bd12fb94..000c0897 100644
--- a/test/pbmtext.ok
+++ b/test/pbmtext.ok
@@ -14,11 +14,3 @@
 1647614653 2027
 3233136020 4535
 1216262214 5711
-386826492 35
-1
-1
-1
-1
-1
-1
-1
diff --git a/test/pbmtext.test b/test/pbmtext.test
index eeb8bf50..c92ed599 100755
--- a/test/pbmtext.test
+++ b/test/pbmtext.test
@@ -81,103 +81,14 @@ rm ${fontRectangle_txt} ${font_pbm}
 
 # One long row
 # Should print 3233136020 4535
-LC_CTYPE=C \
+LC_ALL=C \
 awk 'BEGIN { for (i=32; i<=125;++i) printf("%c",i);
              for (i=160;i<=255;++i) printf("%c",i); }' | \
     pbmtext -builtin bdf | cksum
 
 # One tall column
 # Should print 1216262214 5711
-LC_CTYPE=C \
+LC_ALL=C \
 awk 'BEGIN { for (i=32; i<=125;++i) printf("%c\n",i);
              for (i=160;i<=255;++i) printf("%c\n",i); }' | \
     pbmtext -nomargins -builtin bdf | cksum
-
-
-# Test 6:
-# Test corruptions in BDF font file.
-
-font_bdf=${tmpdir}/font.bdf
-font_corrupt_bdf=${tmpdir}/fontcorrupt.bdf
-
-# Though this BDF font file defines only three letters, it is valid.
-
-cat > ${font_bdf} << EOF
-STARTFONT 2.1
-COMMENT $XFree86: xc/fonts/bdf/misc/micro.bdf,v 1.1 1999/09/25 14:36:34 dawes Exp $  - Modified for test
-FONT micro
-SIZE 4 75 75
-FONTBOUNDINGBOX 4 5 0 0
-STARTPROPERTIES 3
-FONT_DESCENT 0
-FONT_ASCENT 5
-CHARSET_REGISTRY "ISO88591"
-CHARSET_ENCODING "1"
-COPYRIGHT "Public domain font.  Share and enjoy."
-ENDPROPERTIES
-CHARS 3
-STARTCHAR A
-ENCODING 65
-SWIDTH 1000 0
-DWIDTH 4 0
-BBX 4 5 0 0
-BITMAP
-e0
-a0
-e0
-a0
-a0
-ENDCHAR
-STARTCHAR B
-ENCODING 66
-SWIDTH 1000 0
-DWIDTH 4 0
-BBX 4 5 0 0
-BITMAP
-e0
-a0
-c0
-a0
-e0
-ENDCHAR
-STARTCHAR C
-ENCODING 67
-SWIDTH 1000 0
-DWIDTH 4 0
-BBX 4 5 0 0
-BITMAP
-e0
-80
-80
-80
-e0
-ENDCHAR
-ENDFONT
-EOF
-
-# This should succeed and produce 386826492 35
-pbmtext -font ${font_bdf} ABC | cksum
-
-# The rest should all fail.  Writes 1 seven times.
-pbmtext -font ${font_bdf} BCD
-echo $?
-
-sed 's/FONTBOUNDINGBOX 4 5 0 0/FONTBOUNDINGBOX 4 4 0 0/' \
-  ${font_bdf} >  ${font_corrupt_bdf}
-pbmtext -font ${font_corrupt_bdf} ABC > /dev/null
-echo $?
-rm ${font_corrupt_bdf}
-
-sed 's/BBX 4 5 0 0/BBX 4 6 0 0/' \
-  ${font_bdf} >  ${font_corrupt_bdf}
-pbmtext -font ${font_corrupt_bdf} ABC > /dev/null
-echo $?
-rm ${font_corrupt_bdf}
-
-for delete_line in 14 16 18 20
-  do
-  sed "${delete_line}"d ${font_bdf} >  ${font_corrupt_bdf}
-  pbmtext -font ${font_corrupt_bdf} ABC > /dev/null
-  echo $?
-  rm ${font_corrupt_bdf}
-  done