fixed autopygmentize to work with nonencoded files, passed options, and .zshrc.

1 files changed, 65 insertions, 48 deletions

diff --git a/external/autopygmentize b/external/autopygmentize
index 85c8dfd2..964c138f 100755
--- a/external/autopygmentize
+++ b/external/autopygmentize
@@ -1,64 +1,81 @@
-#!/bin/sh
+#!/bin/bash
 # Best effort auto-pygmentization with transparent decompression
 # (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
-# Set the environment variable PYGMENTIZE_OPTS to configure pygments.
+# Set the environment variable PYGMENTIZE_OPTS or pass options before the file path to configure pygments.
 
 # This program can be used as a .lessfilter for the less pager to auto-color less's output
 
-lexer=`pygmentize -N "$1"`
-if [ "$lexer" = "text" ]; then
-  file_common_opts="--brief --dereference --uncompress"
+file="${!#}"              # last argument
+options=${@:1:$(($#-1))}  # handle others args as options to pass to pygmentize
 
-  unset lexer
-  case `file --mime-type $file_common_opts "$1"` in
-      application/xml|image/svg+xml) lexer=xml;;
-      text/html) lexer=html;;
-      text/troff) lexer=nroff;;
-      text/x-asm) lexer=nasm;;
-      text/x-awk) lexer=awk;;
-      text/x-c) lexer=c;;
-      text/x-c++) lexer=cpp;;
-      text/x-diff) lexer=diff;;
-      text/x-fortran) lexer=fortran;;
-      text/x-gawk) lexer=gawk;;
-      text/x-java) lexer=java;;
-      text/x-lisp) lexer=common-lisp;;
-      text/x-lua) lexer=lua;;
-      text/x-makefile) lexer=make;;
-      text/x-msdos-batch) lexer=bat;;
-      text/x-nawk) lexer=nawk;;
-      text/x-pascal) lexer=pascal;;
-      text/x-perl) lexer=perl;;
-      text/x-php) lexer=php;;
-      text/x-po) lexer=po;;
-      text/x-python) lexer=python;;
-      text/x-ruby) lexer=ruby;;
-      text/x-shellscript) lexer=sh;;
-      text/x-tcl) lexer=tcl;;
-      text/x-tex|text/x-texinfo) lexer=latex;; # FIXME: texinfo really needs its own lexer
-      
-      # Types that file outputs which pygmentize didn't support as of file 5.11, pygments 1.6rc1
-      # text/calendar
-      # text/PGP
-      # text/rtf
-      # text/texmacs
-      # text/x-bcpl
-      # text/x-info
-      # text/x-m4
-      # text/x-vcard
-      # text/x-xmcd
-  esac
+file_common_opts="--brief --dereference"
+
+lexer=$(pygmentize -N "$file")
+if [[ "$lexer" == text ]]; then
+    unset lexer
+    case $(file --mime-type --uncompress $file_common_opts "$file") 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 1.6rc1
+        # text/calendar
+        # text/PGP
+        # text/rtf
+        # text/texmacs
+        # text/x-bcpl
+        # text/x-info
+        # text/x-m4
+        # text/x-vcard
+        # text/x-xmcd
+        
+        text/plain)  # special filenames. TODO: insert more
+            case $(basename "$file") in
+                .zshrc) lexer=sh;;
+            esac
+            ;;
+    esac
 fi
 
-encoding=`file --brief --mime-encoding $file_common_opts "$1"`
+encoding=$(file --mime-encoding --uncompress $file_common_opts "$file")
+if [[ $encoding == "us-asciibinarybinary" ]]; then
+	encoding="us-ascii"
+fi
 
-if [ -n "$lexer" ]; then
+if [[ -n "$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
     # FIXME: Specify input encoding rather than output encoding https://bitbucket.org/birkenfeld/pygments-main/issue/800
-    zcat "$1" | pygmentize -O encoding=$encoding,outencoding=UTF-8 $PYGMENTIZE_OPTS -l $lexer
-    exit 0
+    exec $concat "$file" | pygmentize -f terminal256 -O style=native,encoding=$encoding,outencoding=UTF-8 $PYGMENTIZE_OPTS $options -l $lexer
 fi
 
 exit 1