(x)dvipdfmx: Handle UTP-16 surrogates for UCS-4 supplementary planes

git-svn-id: svn://tug.org/texlive/trunk@25568 c570f23f-e606-0410-a88d-b1316a301751

5 files changed, 97 insertions, 12 deletions

diff --git a/Build/source/texk/dvipdfmx/dvipdfmx-20110311-PATCHES/ChangeLog b/Build/source/texk/dvipdfmx/dvipdfmx-20110311-PATCHES/ChangeLog
index 5c76be80116..a0579a900f3 100644
--- a/Build/source/texk/dvipdfmx/dvipdfmx-20110311-PATCHES/ChangeLog
+++ b/Build/source/texk/dvipdfmx/dvipdfmx-20110311-PATCHES/ChangeLog
@@ -1,3 +1,8 @@
+2012-03-05  Peter Breitenlohner  <peb@mppmu.mpg.de>
+
+	* patch-07-surrogates (new): Handle UTP-16 surrogates for UCS-4
+	supplementary planes.  Based on patch from Tanaka Takuji-san.
+
 2012-01-17  Hironori Kitagawa  <h_kitagawa2001@yahoo.co.jp>
 
 	* patch-06-dtou (new): Support for (e-)(u)pTeX dtou direction.
diff --git a/Build/source/texk/dvipdfmx/dvipdfmx-20110311-PATCHES/patch-07-surrogates b/Build/source/texk/dvipdfmx/dvipdfmx-20110311-PATCHES/patch-07-surrogates
new file mode 100644
index 00000000000..22c547b19aa
--- /dev/null
+++ b/Build/source/texk/dvipdfmx/dvipdfmx-20110311-PATCHES/patch-07-surrogates
@@ -0,0 +1,44 @@
+	Handle UTP-16 surrogates for UCS-4 supplementary planes.
+
+diff -ur dvipdfmx-20110311.orig/src/pdfdev.c dvipdfmx-20110311/src/pdfdev.c
+--- dvipdfmx-20110311.orig/src/pdfdev.c	2012-01-17 19:27:40.000000000 +0100
++++ dvipdfmx-20110311/src/pdfdev.c	2012-03-05 13:44:30.000000000 +0100
+@@ -960,17 +960,32 @@
+       }
+       length *= 4;
+     } else if (ctype == 2) {
++      int len = 0;
++
+       if (length * 2 >= FORMAT_BUF_SIZE) {
+         WARN("Too long string...");
+         return -1;
+       }
+-      for (i = 0; i < length; i += 2) {
+-        sbuf1[i*2  ] = font->ucs_group;
+-        sbuf1[i*2+1] = font->ucs_plane;
+-        sbuf1[i*2+2] = p[i];
+-        sbuf1[i*2+3] = p[i+1];
++      for (i = 0; i < length; i += 2, len += 4) {
++        sbuf1[len  ] = font->ucs_group;
++        if ((p[i] & 0xf8) == 0xd8) {
++          int c;
++          /* Check for valid surrogate pair.  */ 
++          if ((p[i] & 0xfc) != 0xd8 || i + 2 >= length || (p[i+2] & 0xfc) != 0xdc) {
++            WARN("Invalid surrogate p[%d]=%02X...", i, p[i]);
++            return -1;
++          }
++          c = (((p[i] & 0x03) << 10) | (p[i+1] << 2) | (p[i+2] & 0x03)) + 0x100;
++          sbuf1[len+1] = (c >> 8) & 0xff;
++          sbuf1[len+2] = c & 0xff;
++          i += 2;
++        } else {
++          sbuf1[len+1] = font->ucs_plane;
++          sbuf1[len+2] = p[i];
++        }
++        sbuf1[len+3] = p[i+1];
+       }
+-      length *= 2;
++      length = len;
+     }
+     p = sbuf1;
+   } else if (ctype == 1 && font->mapc >= 0) {
diff --git a/Build/source/texk/dvipdfmx/dvipdfmx-20110311/src/pdfdev.c b/Build/source/texk/dvipdfmx/dvipdfmx-20110311/src/pdfdev.c
index 3e0c3e0fd62..ff7a83e728f 100644
--- a/Build/source/texk/dvipdfmx/dvipdfmx-20110311/src/pdfdev.c
+++ b/Build/source/texk/dvipdfmx/dvipdfmx-20110311/src/pdfdev.c
@@ -960,17 +960,32 @@ handle_multibyte_string (struct dev_font *font,
       }
       length *= 4;
     } else if (ctype == 2) {
+      int len = 0;
+
       if (length * 2 >= FORMAT_BUF_SIZE) {
         WARN("Too long string...");
         return -1;
       }
-      for (i = 0; i < length; i += 2) {
-        sbuf1[i*2  ] = font->ucs_group;
-        sbuf1[i*2+1] = font->ucs_plane;
-        sbuf1[i*2+2] = p[i];
-        sbuf1[i*2+3] = p[i+1];
+      for (i = 0; i < length; i += 2, len += 4) {
+        sbuf1[len  ] = font->ucs_group;
+        if ((p[i] & 0xf8) == 0xd8) {
+          int c;
+          /* Check for valid surrogate pair.  */ 
+          if ((p[i] & 0xfc) != 0xd8 || i + 2 >= length || (p[i+2] & 0xfc) != 0xdc) {
+            WARN("Invalid surrogate p[%d]=%02X...", i, p[i]);
+            return -1;
+          }
+          c = (((p[i] & 0x03) << 10) | (p[i+1] << 2) | (p[i+2] & 0x03)) + 0x100;
+          sbuf1[len+1] = (c >> 8) & 0xff;
+          sbuf1[len+2] = c & 0xff;
+          i += 2;
+        } else {
+          sbuf1[len+1] = font->ucs_plane;
+          sbuf1[len+2] = p[i];
+        }
+        sbuf1[len+3] = p[i+1];
       }
-      length *= 2;
+      length = len;
     }
     p = sbuf1;
   } else if (ctype == 1 && font->mapc >= 0) {
diff --git a/Build/source/texk/xdvipdfmx/ChangeLog.TL b/Build/source/texk/xdvipdfmx/ChangeLog.TL
index b571158605c..ea1a38b7520 100644
--- a/Build/source/texk/xdvipdfmx/ChangeLog.TL
+++ b/Build/source/texk/xdvipdfmx/ChangeLog.TL
@@ -1,6 +1,12 @@
 ChangeLog.TL: TeX Live (TL) changes for xdvipdfmx
 =================================================
 
+2012-03-05  Peter Breitenlohner  <peb@mppmu.mpg.de>
+
+	* pdfdev.c (handle_multibyte_string): Handle UTP-16 surrogates
+	for UCS-4 supplementary planes.
+	Based on patch from Tanaka Takuji-san.
+
 2012-03-04  Peter Breitenlohner  <peb@mppmu.mpg.de>
 
 	Avoid 'the address of XXX will always evaluate as true' warnings.
diff --git a/Build/source/texk/xdvipdfmx/src/pdfdev.c b/Build/source/texk/xdvipdfmx/src/pdfdev.c
index e74a60f7296..24727d12944 100644
--- a/Build/source/texk/xdvipdfmx/src/pdfdev.c
+++ b/Build/source/texk/xdvipdfmx/src/pdfdev.c
@@ -918,17 +918,32 @@ handle_multibyte_string (struct dev_font *font,
       }
       length *= 4;
     } else if (ctype == 2) {
+      int len = 0;
+
       if (length * 2 >= FORMAT_BUF_SIZE) {
         WARN("Too long string...");
         return -1;
       }
-      for (i = 0; i < length; i += 2) {
-        sbuf1[i*2  ] = font->ucs_group;
-        sbuf1[i*2+1] = font->ucs_plane;
-        sbuf1[i*2+2] = p[i];
-        sbuf1[i*2+3] = p[i+1];
+      for (i = 0; i < length; i += 2, len += 4) {
+        sbuf1[len  ] = font->ucs_group;
+        if ((p[i] & 0xf8) == 0xd8) {
+          int c;
+          /* Check for valid surrogate pair.  */ 
+          if ((p[i] & 0xfc) != 0xd8 || i + 2 >= length || (p[i+2] & 0xfc) != 0xdc) {
+            WARN("Invalid surrogate p[%d]=%02X...", i, p[i]);
+            return -1;
+          }
+          c = (((p[i] & 0x03) << 10) | (p[i+1] << 2) | (p[i+2] & 0x03)) + 0x100;
+          sbuf1[len+1] = (c >> 8) & 0xff;
+          sbuf1[len+2] = c & 0xff;
+          i += 2;
+        } else {
+          sbuf1[len+1] = font->ucs_plane;
+          sbuf1[len+2] = p[i];
+        }
+        sbuf1[len+3] = p[i+1];
       }
-      length *= 2;
+      length = len;
     }
     p = sbuf1;
   } else if (ctype == 1 && font->mapc >= 0) {