texk/dvipdfm-x: Better detection of inconsistent DVI files and additional test

git-svn-id: svn://tug.org/texlive/trunk@38029 c570f23f-e606-0410-a88d-b1316a301751

14 files changed, 189 insertions, 67 deletions

diff --git a/Build/source/texk/dvipdfm-x/ChangeLog b/Build/source/texk/dvipdfm-x/ChangeLog
index 6c6f24730c0..9e54da35d63 100644
--- a/Build/source/texk/dvipdfm-x/ChangeLog
+++ b/Build/source/texk/dvipdfm-x/ChangeLog
@@ -1,3 +1,13 @@
+2015-08-03  Peter Breitenlohner  <peb@mppmu.mpg.de>
+
+	* dvi.c: Better detection of bad (inconsistent) DVI files.
+	* xdvipdfm-bad.test: New shell script to test bad DVI files.
+	* tests/ids_2_6.dvi, tests/ids_3_2.dvi, tests/ids_a_b.dvi,
+	tests/opc_fe.dvi, tests/ptx_2_2.dvi, tests/ptx_6_6.dvi,
+	tests/rev_2_2.dvi, tests/rev_2_3.dvi, tests/void.dvi: New test
+	input data (bad DVI files).
+	* Makefile.am: Add the new test and input data.
+
 2015-07-31  Peter Breitenlohner  <peb@mppmu.mpg.de>
 
 	* dvipdfmx.c, xbb.c: Use getopt_long() for option parsing.
diff --git a/Build/source/texk/dvipdfm-x/Makefile.am b/Build/source/texk/dvipdfm-x/Makefile.am
index cf9cc495583..7f96ccf0a69 100644
--- a/Build/source/texk/dvipdfm-x/Makefile.am
+++ b/Build/source/texk/dvipdfm-x/Makefile.am
@@ -244,24 +244,50 @@ DISTCLEANFILES = config.force
 
 ## Tests
 ##
-TESTS = xdvipdfmx.test xdvipdfm-ann.test xdvipdfm-bb.test xdvipdfm-bkm.test xdvipdfm-psz.test
-TESTS += xdvipdfm-ptx.test xdvipdfm-res.test xdvipdfm-rev.test xdvipdfm-ttc.test
-xdvipdfmx.log xdvipdfm-ann.log xdvipdfm-bb.log xdvipdfm-bkm.log xdvipdfm-psz.log \
-	xdvipdfm-ptx.log xdvipdfm-res.log xdvipdfm-rev.log xdvipdfm-ttc.log: xdvipdfmx$(EXEEXT)
+TESTS = xdvipdfmx.test xdvipdfm-ann.test xdvipdfm-bad.test xdvipdfm-bb.test
+TESTS += xdvipdfm-bkm.test xdvipdfm-psz.test xdvipdfm-ptx.test xdvipdfm-res.test
+TESTS += xdvipdfm-rev.test xdvipdfm-ttc.test
+xdvipdfmx.log xdvipdfm-ann.log xdvipdfm-bad.log xdvipdfm-bb.log \
+	xdvipdfm-bkm.log xdvipdfm-psz.log xdvipdfm-ptx.log xdvipdfm-res.log \
+	xdvipdfm-rev.log xdvipdfm-ttc.log: xdvipdfmx$(EXEEXT)
 EXTRA_DIST = $(TESTS)
-## Test Data
+## xdvipdfmx.test
 EXTRA_DIST += tests/dvipdfmx.cfg tests/psfonts.map
-EXTRA_DIST += tests/fullmap.dvi tests/fullmap.tex
+EXTRA_DIST += tests/cmr10.pfb tests/cmr10.tfm
+EXTRA_DIST += tests/image.dvi tests/image.tex
+EXTRA_DIST += tests/xbmc.dvi tests/xbmc.tex tests/xbmc10.600pk tests/xbmc10.tfm
+DISTCLEANFILES += image*.pdf xbmc*.pdf
+## xdvipdfm-ann.test
+EXTRA_DIST += tests/annot.dvi tests/annot.tex
+DISTCLEANFILES += annot*.pdf
+## xdvipdfm-bad.test
+EXTRA_DIST += tests/ids_2_6.dvi tests/ids_3_2.dvi tests/ids_a_b.dvi tests/opc_fe.dvi
+EXTRA_DIST += tests/ptx_2_2.dvi tests/ptx_6_6.dvi tests/rev_2_2.dvi tests/rev_2_3.dvi
+EXTRA_DIST += tests/void.dvi
+## xdvipdfm-bb.test
 EXTRA_DIST += tests/image.bmp tests/picbmp.bb tests/picbmp.xbb
 EXTRA_DIST += tests/image.jp2 tests/picjp2.bb tests/picjp2.xbb
 EXTRA_DIST += tests/image.jpeg tests/picjpeg.bb tests/picjpeg.xbb
 EXTRA_DIST += tests/image.png tests/picpng.bb tests/picpng.xbb
 EXTRA_DIST += tests/image.pdf tests/picpdf.bb tests/picpdf.xbb
-EXTRA_DIST += tests/image.dvi tests/image.tex
-EXTRA_DIST += tests/cmr10.pfb tests/cmr10.tfm tests/annot.dvi tests/annot.tex
-EXTRA_DIST += tests/bookm.dvi tests/bookm.tex tests/paper.dvi tests/paper.tex
-EXTRA_DIST += tests/resrc.dvi tests/resrc.tex tests/ptex.dvi tests/reverse.dvi
-EXTRA_DIST += tests/xbmc.dvi tests/xbmc.tex tests/xbmc10.600pk tests/xbmc10.tfm
+DISTCLEANFILES += pic*.*
+## xdvipdfm-bkm.test
+EXTRA_DIST += tests/bookm.dvi tests/bookm.tex
+DISTCLEANFILES += bookm*.pdf
+## xdvipdfm-psz.test
+EXTRA_DIST += tests/paper.dvi tests/paper.tex
+DISTCLEANFILES += paper*.pdf
+## xdvipdfm-ptx.test
+EXTRA_DIST += tests/ptex.dvi
+DISTCLEANFILES += ptex*.pdf
+## xdvipdfm-res.test
+EXTRA_DIST += tests/resrc.dvi tests/resrc.tex
+DISTCLEANFILES += resrc*.pdf
+## xdvipdfm-rev.test
+EXTRA_DIST += tests/reverse.dvi
+DISTCLEANFILES += reverse.pdf
+## xdvipdfm-ttc.test
 EXTRA_DIST += tests/ttc.dvi tests/ttc.tex tests/test.ttc
-DISTCLEANFILES += pic*.* image*.pdf annot*.pdf bookm*.pdf paper*.pdf resrc*.pdf
-DISTCLEANFILES +=  xbmc*.pdf ttc*.pdf ptex*.pdf reverse.pdf
+DISTCLEANFILES += ttc*.pdf
+##
+EXTRA_DIST += tests/fullmap.dvi tests/fullmap.tex
diff --git a/Build/source/texk/dvipdfm-x/Makefile.in b/Build/source/texk/dvipdfm-x/Makefile.in
index 1d4fac84af0..b38db3defa1 100644
--- a/Build/source/texk/dvipdfm-x/Makefile.in
+++ b/Build/source/texk/dvipdfm-x/Makefile.in
@@ -779,25 +779,29 @@ mapdatadir = $(datarootdir)/texmf-dist/fonts/map/dvipdfmx
 dist_mapdata_DATA = data/cid-x.map
 cmapdatadir = $(datarootdir)/texmf-dist/fonts/cmap/dvipdfmx
 dist_cmapdata_DATA = data/EUC-UCS2
-DISTCLEANFILES = config.force pic*.* image*.pdf annot*.pdf bookm*.pdf \
-	paper*.pdf resrc*.pdf xbmc*.pdf ttc*.pdf ptex*.pdf reverse.pdf
-TESTS = xdvipdfmx.test xdvipdfm-ann.test xdvipdfm-bb.test \
-	xdvipdfm-bkm.test xdvipdfm-psz.test xdvipdfm-ptx.test \
-	xdvipdfm-res.test xdvipdfm-rev.test xdvipdfm-ttc.test
+DISTCLEANFILES = config.force image*.pdf xbmc*.pdf annot*.pdf pic*.* \
+	bookm*.pdf paper*.pdf ptex*.pdf resrc*.pdf reverse.pdf \
+	ttc*.pdf
+TESTS = xdvipdfmx.test xdvipdfm-ann.test xdvipdfm-bad.test \
+	xdvipdfm-bb.test xdvipdfm-bkm.test xdvipdfm-psz.test \
+	xdvipdfm-ptx.test xdvipdfm-res.test xdvipdfm-rev.test \
+	xdvipdfm-ttc.test
 EXTRA_DIST = $(TESTS) tests/dvipdfmx.cfg tests/psfonts.map \
-	tests/fullmap.dvi tests/fullmap.tex tests/image.bmp \
-	tests/picbmp.bb tests/picbmp.xbb tests/image.jp2 \
-	tests/picjp2.bb tests/picjp2.xbb tests/image.jpeg \
-	tests/picjpeg.bb tests/picjpeg.xbb tests/image.png \
-	tests/picpng.bb tests/picpng.xbb tests/image.pdf \
-	tests/picpdf.bb tests/picpdf.xbb tests/image.dvi \
-	tests/image.tex tests/cmr10.pfb tests/cmr10.tfm \
-	tests/annot.dvi tests/annot.tex tests/bookm.dvi \
-	tests/bookm.tex tests/paper.dvi tests/paper.tex \
-	tests/resrc.dvi tests/resrc.tex tests/ptex.dvi \
-	tests/reverse.dvi tests/xbmc.dvi tests/xbmc.tex \
-	tests/xbmc10.600pk tests/xbmc10.tfm tests/ttc.dvi \
-	tests/ttc.tex tests/test.ttc
+	tests/cmr10.pfb tests/cmr10.tfm tests/image.dvi \
+	tests/image.tex tests/xbmc.dvi tests/xbmc.tex \
+	tests/xbmc10.600pk tests/xbmc10.tfm tests/annot.dvi \
+	tests/annot.tex tests/ids_2_6.dvi tests/ids_3_2.dvi \
+	tests/ids_a_b.dvi tests/opc_fe.dvi tests/ptx_2_2.dvi \
+	tests/ptx_6_6.dvi tests/rev_2_2.dvi tests/rev_2_3.dvi \
+	tests/void.dvi tests/image.bmp tests/picbmp.bb \
+	tests/picbmp.xbb tests/image.jp2 tests/picjp2.bb \
+	tests/picjp2.xbb tests/image.jpeg tests/picjpeg.bb \
+	tests/picjpeg.xbb tests/image.png tests/picpng.bb \
+	tests/picpng.xbb tests/image.pdf tests/picpdf.bb \
+	tests/picpdf.xbb tests/bookm.dvi tests/bookm.tex \
+	tests/paper.dvi tests/paper.tex tests/ptex.dvi tests/resrc.dvi \
+	tests/resrc.tex tests/reverse.dvi tests/ttc.dvi tests/ttc.tex \
+	tests/test.ttc tests/fullmap.dvi tests/fullmap.tex
 all: config.h
 	$(MAKE) $(AM_MAKEFLAGS) all-am
 
@@ -1801,8 +1805,9 @@ config.force: $(ZLIB_DEPEND) $(LIBPNG_DEPEND) $(KPATHSEA_DEPEND) $(LIBPAPER_DEPE
 @LIBPNG_RULE@
 @ZLIB_RULE@
 @LIBPAPER_RULE@
-xdvipdfmx.log xdvipdfm-ann.log xdvipdfm-bb.log xdvipdfm-bkm.log xdvipdfm-psz.log \
-	xdvipdfm-ptx.log xdvipdfm-res.log xdvipdfm-rev.log xdvipdfm-ttc.log: xdvipdfmx$(EXEEXT)
+xdvipdfmx.log xdvipdfm-ann.log xdvipdfm-bad.log xdvipdfm-bb.log \
+	xdvipdfm-bkm.log xdvipdfm-psz.log xdvipdfm-ptx.log xdvipdfm-res.log \
+	xdvipdfm-rev.log xdvipdfm-ttc.log: xdvipdfmx$(EXEEXT)
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
 # Otherwise a system limit (for SysV at least) may be exceeded.
diff --git a/Build/source/texk/dvipdfm-x/dvi.c b/Build/source/texk/dvipdfm-x/dvi.c
index 65824ee2aaf..27ddf4fc78c 100644
--- a/Build/source/texk/dvipdfm-x/dvi.c
+++ b/Build/source/texk/dvipdfm-x/dvi.c
@@ -319,7 +319,7 @@ static int pre_id_byte, post_id_byte, is_ptex = 0, has_ptex = 0;
 static void
 check_id_bytes (void) {
   if (pre_id_byte != post_id_byte && (pre_id_byte != DVI_ID || post_id_byte != DVIV_ID))
-    ERROR ("Inconsistent DVI id_bytes in preamble (%i) and postamble (%i)");
+    ERROR ("Inconsistent DVI id_bytes %d (pre) and %d (post)", pre_id_byte, post_id_byte);
 }
 
 static void
@@ -366,12 +366,6 @@ find_post (void)
   post_id_byte = ch;
   is_xdv = ch == XDV_ID;
   is_ptex = ch == DVIV_ID;
-  if (pre_id_byte) {
-    check_id_bytes();
-    if (has_ptex && !is_ptex) {
-      ERROR ("DVI opcode %i only valid for Ascii pTeX", PTEXDIR);
-    }
-  }
 
   /* Make sure post_post is really there */
   current = current - 5;
@@ -387,6 +381,21 @@ find_post (void)
     ERROR(invalid_signature);
   }
 
+  /* Finally check the ID byte in the preamble */
+  /* An Ascii pTeX DVI file has id_byte DVI_ID in the preamble but DVIV_ID in the postamble. */
+  xseek_absolute (dvi_file, 0, "DVI");
+  if ((ch = get_unsigned_byte(dvi_file)) != PRE) {
+    MESG("Found %d where PRE was expected\n", ch);
+    ERROR(invalid_signature);
+  }
+  ch = get_unsigned_byte(dvi_file);
+  if (!(ch == DVI_ID || ch == XDV_ID)) {
+    MESG("DVI ID = %d\n", ch);
+    ERROR(invalid_signature);
+  }
+  pre_id_byte = ch;
+  check_id_bytes();
+
   return current;
 }
 
@@ -474,7 +483,7 @@ get_preamble_dvi_info (void)
     ERROR(invalid_signature);
   }
   
-  /* A DVI file Ascii pTeX has id_byte DVI_ID in the preamble but DVIV_ID in the postamble. */
+  /* An Ascii pTeX DVI file has id_byte DVI_ID in the preamble but DVIV_ID in the postamble. */
   ch = get_unsigned_byte(dvi_file);
   if (!(ch == DVI_ID || ch == XDV_ID)) {
     MESG("DVI ID = %d\n", ch);
@@ -483,10 +492,7 @@ get_preamble_dvi_info (void)
 
   pre_id_byte = ch;
   is_xdv = ch == XDV_ID;
-  if (post_id_byte)
-    check_id_bytes();
-  else
-    is_ptex = ch == DVI_ID; /* maybe */
+  is_ptex = ch == DVI_ID; /* maybe */
   
   dvi_info.unit_num = get_positive_quad(dvi_file, "DVI", "unit_num");
   dvi_info.unit_den = get_positive_quad(dvi_file, "DVI", "unit_den");
@@ -1598,24 +1604,26 @@ dvi_end_reflect (void)
 }
 
 static void
+skip_native_font_def (void)
+{
+  unsigned int flags;
+  int name_length;
+
+  skip_bytes(4, dvi_file); /* skip point size */
+  flags = get_unsigned_pair(dvi_file);
+  name_length = get_unsigned_byte(dvi_file);
+  skip_bytes(name_length + 4, dvi_file);
+  if (flags & XDV_FLAG_COLORED)
+    skip_bytes(4, dvi_file);
+ }
+
+static void
 do_native_font_def (int32_t tex_id)
 {
-  if (linear) {
+  if (linear)
     read_native_font_record(tex_id);
-  } else {
-    unsigned int flags;
-    int name_length, i;
-
-    get_unsigned_quad(dvi_file); /* skip point size */
-    flags = get_unsigned_pair(dvi_file);
-    name_length = (int) get_unsigned_byte(dvi_file);
-    for (i = 0; i < name_length; ++i)
-      get_unsigned_byte(dvi_file);
-    get_unsigned_quad(dvi_file);
-    if (flags & XDV_FLAG_COLORED) {
-      get_unsigned_quad(dvi_file);
-    }
-  }
+  else
+    skip_native_font_def();
   --dvi_page_buf_index; /* don't buffer the opcode */
 }
 
@@ -1728,13 +1736,38 @@ do_glyphs (void)
   return;
 }
 
-/* Note to be absolutely certain that the string escape buffer doesn't
- * hit its limit, FORMAT_BUF_SIZE should set to 4 times S_BUFFER_SIZE
- * in pdfobj.c.  Is there any application that genenerate words with
- * 1k characters?
- */
+static void
+check_postamble (void)
+{
+  int code;
+
+  skip_bytes(28, dvi_file);
+  while ((code = get_unsigned_byte(dvi_file)) != POST_POST) {
+    switch (code) {
+    case FNT_DEF1: case FNT_DEF2: case FNT_DEF3: case FNT_DEF4:
+      skip_bytes(code + 1 - FNT_DEF1, dvi_file);
+      skip_fntdef();
+      break;
+    case XDV_NATIVE_FONT_DEF:
+      skip_bytes(4, dvi_file);
+      skip_native_font_def();
+      break;
+    default:
+      ERROR("Unexpected op code (%u) in postamble", code);
+    }
+  }
+  skip_bytes(4, dvi_file);
+  post_id_byte = get_unsigned_byte(dvi_file);
+  if (!(post_id_byte == DVI_ID || post_id_byte == DVIV_ID || post_id_byte == XDV_ID)) {
+    MESG("DVI ID = %d\n", post_id_byte);
+    ERROR(invalid_signature);
+  }
+  check_id_bytes();
+  if (has_ptex && post_id_byte != DVIV_ID)
+    ERROR ("Saw opcode %i in DVI file not for Ascii pTeX", PTEXDIR);
 
-#define SBUF_SIZE 1024
+  num_pages = 0; /* force loop to terminate */
+}
 
 /* Most of the work of actually interpreting
  * the dvi file is here.
@@ -1798,6 +1831,12 @@ dvi_do_page (double page_paper_height, double hmargin, double vmargin)
       break;
     case EOP:
       do_eop();
+      if (linear) {
+        if ((opcode = get_unsigned_byte(dvi_file)) == POST)
+          check_postamble();
+        else
+          ungetc(opcode, dvi_file);
+      }
       return;
 
     case PUSH:
@@ -1908,6 +1947,7 @@ dvi_init (char *dvi_filename, double mag)
   int32_t post_location;
 
   if (!dvi_filename) { /* no filename: reading from stdin, probably a pipe */
+    int ch;
 #ifdef WIN32
         setmode(fileno(stdin), _O_BINARY);
 #endif
@@ -1916,6 +1956,10 @@ dvi_init (char *dvi_filename, double mag)
 
     get_preamble_dvi_info();
     do_scales(mag);
+    if ((ch = get_unsigned_byte(dvi_file)) == POST)
+      check_postamble();
+    else
+      ungetc(ch, dvi_file);
   } else {
     dvi_file = MFOPEN(dvi_filename, FOPEN_RBIN_MODE);
     if (!dvi_file) {
@@ -2315,7 +2359,7 @@ dvi_scan_specials (int page_no,
   static int     buffered_page = -1;
   unsigned int len;
 
-  if (page_no == buffered_page)
+  if (page_no == buffered_page || num_pages == 0)
     return; /* because dvipdfmx wants to scan first page twice! */
   buffered_page = page_no;
 
diff --git a/Build/source/texk/dvipdfm-x/tests/ids_2_6.dvi b/Build/source/texk/dvipdfm-x/tests/ids_2_6.dvi
new file mode 100644
index 00000000000..952754daef7
--- /dev/null
+++ b/Build/source/texk/dvipdfm-x/tests/ids_2_6.dvi
diff --git a/Build/source/texk/dvipdfm-x/tests/ids_3_2.dvi b/Build/source/texk/dvipdfm-x/tests/ids_3_2.dvi
new file mode 100644
index 00000000000..4f1c937c492
--- /dev/null
+++ b/Build/source/texk/dvipdfm-x/tests/ids_3_2.dvi
diff --git a/Build/source/texk/dvipdfm-x/tests/ids_a_b.dvi b/Build/source/texk/dvipdfm-x/tests/ids_a_b.dvi
new file mode 100644
index 00000000000..1fb1523d2c4
--- /dev/null
+++ b/Build/source/texk/dvipdfm-x/tests/ids_a_b.dvi
diff --git a/Build/source/texk/dvipdfm-x/tests/opc_fe.dvi b/Build/source/texk/dvipdfm-x/tests/opc_fe.dvi
new file mode 100644
index 00000000000..c3df8bf4517
--- /dev/null
+++ b/Build/source/texk/dvipdfm-x/tests/opc_fe.dvi
diff --git a/Build/source/texk/dvipdfm-x/tests/ptx_2_2.dvi b/Build/source/texk/dvipdfm-x/tests/ptx_2_2.dvi
new file mode 100644
index 00000000000..d34a5070f83
--- /dev/null
+++ b/Build/source/texk/dvipdfm-x/tests/ptx_2_2.dvi
diff --git a/Build/source/texk/dvipdfm-x/tests/ptx_6_6.dvi b/Build/source/texk/dvipdfm-x/tests/ptx_6_6.dvi
new file mode 100644
index 00000000000..19ed6ddb628
--- /dev/null
+++ b/Build/source/texk/dvipdfm-x/tests/ptx_6_6.dvi
diff --git a/Build/source/texk/dvipdfm-x/tests/rev_2_2.dvi b/Build/source/texk/dvipdfm-x/tests/rev_2_2.dvi
new file mode 100644
index 00000000000..814322571bd
--- /dev/null
+++ b/Build/source/texk/dvipdfm-x/tests/rev_2_2.dvi
diff --git a/Build/source/texk/dvipdfm-x/tests/rev_2_3.dvi b/Build/source/texk/dvipdfm-x/tests/rev_2_3.dvi
new file mode 100644
index 00000000000..35dd637304f
--- /dev/null
+++ b/Build/source/texk/dvipdfm-x/tests/rev_2_3.dvi
diff --git a/Build/source/texk/dvipdfm-x/tests/void.dvi b/Build/source/texk/dvipdfm-x/tests/void.dvi
new file mode 100644
index 00000000000..7ec6d6c1be6
--- /dev/null
+++ b/Build/source/texk/dvipdfm-x/tests/void.dvi
diff --git a/Build/source/texk/dvipdfm-x/xdvipdfm-bad.test b/Build/source/texk/dvipdfm-x/xdvipdfm-bad.test
new file mode 100755
index 00000000000..7edba2d9c79
--- /dev/null
+++ b/Build/source/texk/dvipdfm-x/xdvipdfm-bad.test
@@ -0,0 +1,37 @@
+#! /bin/sh
+
+# Copyright (C) 2015 Peter Breitenlohner <tex-live@tug.org>
+# You may freely use, modify and/or distribute this file.
+
+TEXMFCNF=$srcdir/../kpathsea
+TFMFONTS="$srcdir/tests;$srcdir/data"
+T1FONTS="$srcdir/tests;$srcdir/data"
+TEXFONTMAPS="$srcdir/tests;$srcdir/data"
+DVIPDFMXINPUTS="$srcdir/tests;$srcdir/data"
+TEXPICTS=$srcdir/tests
+export TEXMFCNF TFMFONTS T1FONTS TEXFONTMAPS DVIPDFMXINPUTS TEXPICTS
+
+failed=
+
+list="ids_2_6 ids_3_2 ids_a_b opc_fe ptx_2_2 ptx_6_6 rev_2_2 rev_2_3 void"
+
+for f in $list
+do
+
+echo "*** xdvipdfmx -v -z0 -o bad.pdf $f" && echo \
+	&& ./xdvipdfmx -v -z0 -o bad.pdf $srcdir/tests/$f \
+	&& failed="$failed xdvipdfmx-$f" \
+	|| { echo && echo "xdvipdfmx-$f tests OK" && echo; }
+
+echo "*** xdvipdfmx -v -z0 -o bad.pdf <$f.dvi" && echo \
+	&& ./xdvipdfmx -v -z0 -o bad.pdf <$srcdir/tests/$f.dvi \
+	&& failed="$failed xdvipdfmx-$f (stdin)" \
+	|| { echo && echo "xdvipdfmx-$f (stdin) tests OK" && echo; }
+
+done
+
+test -z "$failed" && exit 0
+echo
+echo "failed tests:$failed"
+exit 1
+