autopygmentize: various improvements

Fix encoding detection using file (add --brief flag)

Remove workaround for non-issue #799

Only call pygmentize once, and thereby always get the benefit of
encoding detection.

1 files changed, 46 insertions, 50 deletions

diff --git a/external/autopygmentize b/external/autopygmentize
index ad9faaeb..ee32a301 100644..100755
--- a/external/autopygmentize
+++ b/external/autopygmentize
@@ -1,6 +1,6 @@
 #!/bin/sh
 # Best effort auto-pygmentization with transparent decompression
-# (c) Reuben Thomas 2012
+# (c) Reuben Thomas 2012-2013
 # 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
@@ -8,60 +8,56 @@
 
 # This program can be used as a .lessfilter for the less pager to auto-color less's output
 
-if [ `pygmentize -N $1` != "text" ]; then
-    pygmentize $PYGMENTIZE_OPTS "$1"
-    exit 0
-fi
+lexer=`pygmentize -N "$1"`
+if [ "$lexer" = "text" ]; then
+  file_common_opts="--brief --dereference --uncompress"
 
-file_common_opts="--brief --dereference --uncompress"
-
-unset lexer
-case `file --mime-type $file_common_opts "$1"` in
-    application/xml|image/svg+xml) lexer=xml;;
-    text/html) lexer=html;;
-    text/troff) lexer=nroff;;
-    text/x-asm) lexer=nasm;;
-    text/x-awk) lexer=awk;;
-    text/x-c) lexer=c;;
-    text/x-c++) lexer=cpp;;
-    text/x-diff) lexer=diff;;
-    text/x-fortran) lexer=fortran;;
-    text/x-gawk) lexer=gawk;;
-    text/x-java) lexer=java;;
-    text/x-lisp) lexer=common-lisp;;
-    text/x-lua) lexer=lua;;
-    text/x-makefile) lexer=make;;
-    text/x-msdos-batch) lexer=bat;;
-    text/x-nawk) lexer=nawk;;
-    text/x-pascal) lexer=pascal;;
-    text/x-perl) lexer=perl;;
-    text/x-php) lexer=php;;
-    text/x-po) lexer=po;;
-    text/x-python) lexer=python;;
-    text/x-ruby) lexer=ruby;;
-    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
-    
-    # Types that file outputs which pygmentize didn't support as of file 5.11, pygments 1.5
-    # text/calendar
-    # text/PGP
-    # text/rtf
-    # text/texmacs
-    # text/x-bcpl
-    # text/x-info
-    # text/x-m4
-    # text/x-vcard
-    # text/x-xmcd
-esac
+  unset lexer
+  case `file --mime-type $file_common_opts "$1"` in
+      application/xml|image/svg+xml) lexer=xml;;
+      text/html) lexer=html;;
+      text/troff) lexer=nroff;;
+      text/x-asm) lexer=nasm;;
+      text/x-awk) lexer=awk;;
+      text/x-c) lexer=c;;
+      text/x-c++) lexer=cpp;;
+      text/x-diff) lexer=diff;;
+      text/x-fortran) lexer=fortran;;
+      text/x-gawk) lexer=gawk;;
+      text/x-java) lexer=java;;
+      text/x-lisp) lexer=common-lisp;;
+      text/x-lua) lexer=lua;;
+      text/x-makefile) lexer=make;;
+      text/x-msdos-batch) lexer=bat;;
+      text/x-nawk) lexer=nawk;;
+      text/x-pascal) lexer=pascal;;
+      text/x-perl) lexer=perl;;
+      text/x-php) lexer=php;;
+      text/x-po) lexer=po;;
+      text/x-python) lexer=python;;
+      text/x-ruby) lexer=ruby;;
+      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
+      
+      # Types that file outputs which pygmentize didn't support as of file 5.11, pygments 1.5
+      # text/calendar
+      # text/PGP
+      # text/rtf
+      # text/texmacs
+      # text/x-bcpl
+      # text/x-info
+      # text/x-m4
+      # text/x-vcard
+      # text/x-xmcd
+  esac
+fi
 
-encoding=`file --mime-encoding $file_common_opts "$1"`
+encoding=`file --brief --mime-encoding $file_common_opts "$1"`
 
 if [ -n "$lexer" ]; then
     # FIXME: Specify input encoding rather than output encoding https://bitbucket.org/birkenfeld/pygments-main/issue/800
-    # FIXME: Encoding argument ignored on stdin https://bitbucket.org/birkenfeld/pygments-main/issue/799
-    #zcat "$1" | pygmentize -O encoding=$encoding,outencoding=UTF-8 $PYGMENTIZE_OPTS -l $lexer
-    pygmentize -O encoding=$encoding,outencoding=UTF-8 $PYGMENTIZE_OPTS -l $lexer $1
+    zcat "$1" | pygmentize -O encoding=$encoding,outencoding=UTF-8 $PYGMENTIZE_OPTS -l $lexer
     exit 0
 fi