Bug 702785: Add UTF-8 encoding for strings in Info dict to pdf_info.ps

Thanks to Peter Cherepanov for this change.

1 files changed, 84 insertions, 9 deletions

diff --git a/lib/pdf_info.ps b/lib/pdf_info.ps
index 0b4654ab0..9c940ed07 100644
--- a/lib/pdf_info.ps
+++ b/lib/pdf_info.ps
@@ -69,6 +69,81 @@ cleartomark		% discard the dict from --where--
 % ---- No more executable code on the top level after this line -----
 % ---- except 2 lines at the very end                           -----
 
+% Write a character to the standard output.
+/putchar [ % int -> -
+  (%stdout) (w) file
+  /exch cvx /write cvx
+] cvx bind def
+
+% Write U+xxxx to the standard output as UTF-8.
+/put-ucode { % int -> -
+  dup 16#80 ge {
+    dup 16#800 ge {
+      dup 16#10000 ge {
+        dup -18 bitshift 16#f0 or putchar
+        dup -12 bitshift 16#3f and 16#80 or putchar
+      } {
+        dup -12 bitshift 16#e0 or putchar
+      } ifelse
+      dup -6  bitshift 16#3f and 16#80 or putchar
+    } {
+      dup -6 bitshift 16#C0 or putchar
+    } ifelse
+    16#3f and 16#80 or
+  } if
+  putchar
+} bind def
+
+% PDFDocEncoding to U+xxxx decoding table.
+/doc-to-ucode [
+  0 1 23 {} for
+  16#2d8 16#2c7 16#2c6 16#2d9 16#2dd 16#2db 16#2da 16#2dc
+  32 1 127 {} for
+  16#2022 16#2020 16#2021 16#2026 16#2014 16#2013 16#192
+  16#2044 16#2039 16#203a 16#2212 16#2030 16#201e 16#201c 
+  16#201d 16#2018 16#2019 16#201a 16#2122 16#fb01 16#fb02
+  16#141 16#152 16#160 16#178 16#17d 16#131 16#142 16#153
+  16#161 16#17e 0 16#20ac
+  161 1 255 {} for
+] readonly def
+
+% Convert a doc string from PDFDocEncoding or UTF-16BE to UTF-8
+% and write it to standard output. 
+/write-doc-string { % (string) -> -
+  <feff> anchorsearch {
+    pop
+    0 exch                 % hi16 (str) 
+    0 2 2 index length 2 sub {
+      2 copy 2 copy        % hi16 (str) i (str) i (str) i
+      get 256 mul 3 1 roll % hi16 (str) i hi*256 (str) i
+      1 add get add        % hi16 (str) i 256*hi+lo
+      dup 16#fc00 and dup  % hi16 (str) i 256*hi+lo tag tag
+      16#d800 eq { % High surrogate
+        pop
+        16#3ff and
+        10 bitshift
+        16#10000 add       % hi16  (str) i hi16'
+        4 1 roll           % hi16' hi16 (str) i
+        pop exch pop       % hi16' (str)
+      } {
+        16#dc00 eq { % Low surrogate
+          16#3ff and       % hi16 (str) i (256*hi+lo)&0x3ff
+          4 -1 roll add    % (str) i (256*hi+lo)&0x3ff+hi16
+          put-ucode        % (str) i
+          pop 0 exch       % 0 (str)
+        } { % BMP plane
+          put-ucode        % hi16 (str) i
+          pop              % hi16 (str)
+        } ifelse
+      } ifelse
+    } for
+    pop pop                % -
+  } {
+    { //doc-to-ucode exch get put-ucode 
+    } forall
+  } ifelse
+} bind def
+
 %% When checking Resources for Font objects, we must ensure that we don't end
 %% up following a circular reference. Circular references in Resources should
 %% not, of course, be present but we've seen a number of cases. If we do detect
@@ -129,15 +204,15 @@ cleartomark		% discard the dict from --where--
 
   % Print out the "Info" dictionary if present
   Trailer /Info knownoget {
-     dup /Title knownoget { (Title: ) print = flush } if
-     dup /Author knownoget { (Author: ) print = flush } if
-     dup /Subject knownoget { (Subject: ) print = flush } if
-     dup /Keywords knownoget { (Keywords: ) print = flush } if
-     dup /Creator knownoget { (Creator: ) print = flush } if
-     dup /Producer knownoget { (Producer: ) print = flush } if
-     dup /CreationDate knownoget { (CreationDate: ) print = flush } if
-     dup /ModDate knownoget { (ModDate: ) print = flush } if
-     dup /Trapped knownoget { (Trapped: ) print = flush } if
+     dup /Title knownoget { (Title: ) print write-doc-string () = flush } if
+     dup /Author knownoget { (Author: ) print write-doc-string () = flush } if
+     dup /Subject knownoget { (Subject: ) print write-doc-string () = flush } if
+     dup /Keywords knownoget { (Keywords: ) print write-doc-string () = flush } if
+     dup /Creator knownoget { (Creator: ) print write-doc-string () = flush } if
+     dup /Producer knownoget { (Producer: ) print write-doc-string () = flush } if
+     dup /CreationDate knownoget { (CreationDate: ) print write-doc-string () = flush } if
+     dup /ModDate knownoget { (ModDate: ) print write-doc-string () = flush } if
+     dup /Trapped knownoget { (Trapped: ) print write-doc-string () = flush } if
      pop
   } if