autopygmentize: support more file types (#2219)

Includes new languages and new compression types.

Also, check that the decompression program exists before trying to use it.

1 files changed, 46 insertions, 8 deletions

diff --git a/external/autopygmentize b/external/autopygmentize
index 3a4aed7f..85d23668 100755
--- a/external/autopygmentize
+++ b/external/autopygmentize
@@ -23,37 +23,60 @@ case $(file --mime-type --uncompress $file_common_opts "$file") in
     text/x-awk) lexer=awk;;
     text/x-c) lexer=c;;
     text/x-c++) lexer=cpp;;
+    text/x-clojure) lexer=clojure;;
     text/x-crystal) lexer=crystal;;
     text/x-diff) lexer=diff;;
+    text/x-execline) lexer=execline;;
+    text/x-forth) lexer=forth;;
     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-lua|text/x-luatex) lexer=lua;;
     text/x-makefile) lexer=make;;
     text/x-msdos-batch) lexer=bat;;
     text/x-nawk) lexer=nawk;;
+    text/x-objective-c) lexer=objective-c;;
     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-script.python) lexer=python;;
     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
+    text/xml) lexer=xml;;
+    text/vnd.graphviz) lexer=graphviz;;
 
-    # Types that file outputs which pygmentize didn't support as of file 5.20, pygments 2.0
+    # Types that file outputs which pygmentize didn't support as of file 5.41, pygments 2.11.2
+    # text/binary
     # text/calendar
-    # text/inf
     # text/PGP
+    # text/prs.lines.tag
     # text/rtf
+    # text/spreadsheet
     # text/texmacs
-    # text/vnd.graphviz
+    # text/vcard
+    # text/vnd.sosi
+    # text/x-Algol68
     # text/x-bcpl
+    # text/x-dmtf-mif
+    # text/x-gimp-curve
+    # text/x-gimp-ggr
+    # text/x-gimp-gpl
     # text/x-info
+    # text/x-installshield-lid
     # text/x-m4
+    # text/x-modulefile
+    # text/x-ms-adm
+    # text/x-ms-cpx
+    # text/x-ms-regedirt
+    # text/x-ms-tag
+    # text/x-systemtap
     # text/x-vcard
+    # text/x-wine-extension-reg
     # text/x-xmcd
 
     text/plain)  # special filenames. TODO: insert more
@@ -67,12 +90,27 @@ case $(file --mime-type --uncompress $file_common_opts "$file") in
 esac
 
 # Find a concatenator for compressed files
-concat=cat
+concat=
+concat_opts=
 case $(file $file_common_opts --mime-type "$file") in
-    application/gzip)    concat=zcat;;
-    application/x-bzip2) concat=bzcat;;
-    application/x-xz)    concat=xzcat;;
+    # TODO: add support
+    # application/x-rzip (does not decompress to stdout)
+    # application/x-dzip (Windows only)
+    application/gzip|application/x-gzip)  concat=zcat;;
+    application/x-bzip)                   concat=bzip; concat_opts=-dc;;
+    application/x-bzip2)                  concat=bzcat;;
+    application/x-lz4)                    concat=lz4; concat_opts=-dc;;
+    application/x-lzh-compressed)         concat=p7zip; concat_opts=-dc;;
+    application/x-lzma)                   concat=lzcat;;
+    application/x-lzip)                   concat=lzip; concat_opts=-dc;;
+    application/x-xz)                     concat=xzcat;;
+    application/x-zoo)                    concat=zoo; concat_opts=fu;;
 esac
+# If concat is unset or doesn't exist, use cat instead
+if [[ "$concat" == "" ]] || ! command -v "$concat"; then
+    concat=cat
+    concat_opts=
+fi
 
 # Find a suitable reader, preceded by a hex dump for binary files,
 # or fmt for text with very long lines