Merged in rrt/pygments-main (pull request #608)

Use hexdump for binary files (plus a tweak to keep MIME types sorted)

1 files changed, 28 insertions, 11 deletions

diff --git a/external/autopygmentize b/external/autopygmentize
index f18cac09..d2d05970 100755
--- a/external/autopygmentize
+++ b/external/autopygmentize
@@ -1,6 +1,6 @@
 #!/bin/bash
 # Best effort auto-pygmentization with transparent decompression
-# by Reuben Thomas 2008-2015
+# by Reuben Thomas 2008-2016
 # This program is in the public domain.
 
 # Strategy: first see if pygmentize can find a lexer; if not, ask file; if that finds nothing, fail
@@ -25,6 +25,7 @@ if [[ "$lexer" == text ]]; then
         text/x-awk) lexer=awk;;
         text/x-c) lexer=c;;
         text/x-c++) lexer=cpp;;
+        text/x-crystal) lexer=crystal;;
         text/x-diff) lexer=diff;;
         text/x-fortran) lexer=fortran;;
         text/x-gawk) lexer=gawk;;
@@ -40,7 +41,6 @@ if [[ "$lexer" == text ]]; then
         text/x-po) lexer=po;;
         text/x-python) lexer=python;;
         text/x-ruby) lexer=ruby;;
-        text/x-crystal) lexer=crystal;;
         text/x-shellscript) lexer=sh;;
         text/x-tcl) lexer=tcl;;
         text/x-tex|text/x-texinfo) lexer=latex;; # FIXME: texinfo really needs its own lexer
@@ -66,19 +66,36 @@ if [[ "$lexer" == text ]]; then
     esac
 fi
 
+# Find a preprocessor for compressed files
+concat=cat
+case $(file $file_common_opts --mime-type "$file") in
+    application/x-gzip)  concat=zcat;;
+    application/x-bzip2) concat=bzcat;;
+    application/x-xz)    concat=xzcat;;
+esac
+
+# Find a suitable lexer, preceded by a hex dump for binary files
+prereader=""
 encoding=$(file --mime-encoding --uncompress $file_common_opts "$file")
 if [[ $encoding == "binary" ]]; then
-    encoding="latin1"
+    prereader="od -x" # POSIX fallback
+    if [[ -n $(which hd) ]]; then
+        prereader="hd" # preferred
+    fi
+    lexer=hexdump
+    encoding=latin1
 fi
-
 if [[ -n "$lexer" ]]; then
-    concat=cat
-    case $(file $file_common_opts --mime-type "$file") in
-        application/x-gzip)  concat=zcat;;
-        application/x-bzip2) concat=bzcat;;
-        application/x-xz)    concat=xzcat;;
-    esac
-    exec $concat "$file" | pygmentize -O inencoding=$encoding $PYGMENTIZE_OPTS $options -l $lexer
+    reader="pygmentize -O inencoding=$encoding $PYGMENTIZE_OPTS $options -l $lexer"
+fi
+
+# If we found a reader, run it
+if [[ -n "$reader" ]]; then
+    if [[ -n "$prereader" ]]; then
+        exec $concat "$file" | $prereader | $reader
+    else
+        exec $concat "$file" | $reader
+    fi
 fi
 
 exit 1