Moved mime-type configuration into a configuration file.

2 files changed, 349 insertions, 303 deletions

diff --git a/core/ext/mime_types.conf b/core/ext/mime_types.conf
new file mode 100644
index 00000000..c46efdbe
--- /dev/null
+++ b/core/ext/mime_types.conf
@@ -0,0 +1,293 @@
+% Copyright 2007-2009 Mitchell mitchell<att>caladbolg.net. See LICENSE.
+%
+% Configuration file for mime-types.
+% Each line defines a mime-type (by extension, shebang word, or Lua pattern) and
+% its associated lexer. A shebang word is one that is contained in the file's
+% shebang line (#!). Only the last space, the one separating the mime-type from
+% its lexer, in a definition is significant.
+%
+% Syntax:
+%   % Comment (must start a line)
+%   ext lexer_name
+%   #shbang_word lexer_name
+%   /lua_pattern lexer_name
+
+
+% Actionscript
+as actionscript
+
+% ADA
+ada ada
+adb ada
+ads ada
+
+% ANTLR
+g antlr
+
+% APDL
+ans apdl
+inp apdl
+mac apdl
+
+% Applescript
+applescript applescript
+
+% ASP
+asa asp
+asp asp
+
+% AWK
+awk awk
+#awk awk
+
+% Batch
+bat batch
+cmd batch
+
+% Boo
+boo boo
+
+% C/C++
+c cpp
+cc cpp
+C cpp
+cpp cpp
+cxx cpp
+c++ cpp
+h cpp
+hh cpp
+hpp cpp
+
+% C#
+cs csharp
+
+% CSS
+css css
+
+% D
+d d
+
+% diff
+diff diff
+patch diff
+
+% Eiffel
+e eiffel
+eif eiffel
+
+% Erlang
+erl erlang
+
+% Forth
+f forth
+frt forth
+fs forth
+
+% Fortran
+for fortran
+fort fortran
+f77 fortran
+f90 fortran
+f95 fortran
+
+% Gap
+g gap
+gd gap
+gi gap
+gap gap
+
+% Gettext
+po gettext
+pot gettext
+
+% GNUPlot
+dem gnuplot
+plt gnuplot
+
+% Groovy
+groovy groovy
+grv groovy
+
+% Haskell
+hs haskell
+
+% HTML
+htm html
+html html
+shtm html
+shtml html
+
+% IDL
+idl idl
+
+% ini
+ini ini
+reg ini
+
+% Io
+io io
+
+% Java
+bsh java
+java java
+
+% Javascript
+js javascript
+
+% Latex
+bbl latex
+dtx latex
+ins latex
+ltx latex
+tex latex
+sty latex
+
+% Lisp
+el lisp
+lisp lisp
+lsp lisp
+
+% Lua
+lua lua
+#lua lua
+
+% Makefile
+GNUmakefile makefile
+iface makefile
+mak makefile
+makefile makefile
+Makefile makefile
+
+% Maxima
+maxima maxima
+
+% Objective C
+m objective_c
+objc objective_c
+
+% OCAML
+ml ocaml
+mli ocaml
+mll ocaml
+mly ocaml
+
+% Pascal
+dpk pascal
+dpr pascal
+p pascal
+pas pascal
+
+% Perl
+al perl
+perl perl
+pl perl
+pm perl
+#perl perl
+
+% PHP
+inc php
+php php
+php3 php
+php4 php
+phtml php
+#php php
+
+% Pike
+pike pike
+pmod pike
+
+% Postscript
+eps postscript
+ps postscript
+
+% Properties
+props props
+properties props
+
+% Python
+sc python
+py python
+pyw python
+#python python
+
+% R
+R r
+Rout r
+Rhistory r
+Rt r
+Rout.save r
+Rout.fail r
+
+% Rebol
+r rebol
+
+% Rexx
+orx rexx
+rex rexx
+
+% RHTML
+rhtml rhtml
+
+% Ruby
+rb ruby
+rbw ruby
+#ruby ruby
+
+% Ragel
+rl ragel
+
+% Scheme
+scm scheme
+
+% Shell
+bash shell
+csh shell
+sh shell
+#sh shell
+
+% Smalltalk
+changes smalltalk
+st smalltalk
+sources smalltalk
+
+% SQL
+sql sql
+
+% TCL
+tcl tcl
+tk tcl
+
+% Vala
+vala vala
+
+% Verilog
+v verilog
+ver verlilog
+
+% VHDL
+vh vhdl
+vhd vhdl
+vhdl vhdl
+
+% Visual Basic
+asa visualbasic
+bas visualbasic
+cls visualbasic
+ctl visualbasic
+dob visualbasic
+dsm visualbasic
+dsr visualbasic
+frm visualbasic
+pag visualbasic
+pag visualbasic
+vb visualbasic
+vba visualbasic
+vbs visualbasic
+
+% XML
+xhtml xml
+xml xml
+xsd xml
+xsl xml
+xslt xml
+/^%s*<%?xml%s xml
diff --git a/core/ext/mime_types.lua b/core/ext/mime_types.lua
index e4960e41..2948ade5 100644
--- a/core/ext/mime_types.lua
+++ b/core/ext/mime_types.lua
@@ -7,337 +7,90 @@ local locale = _G.locale
 module('textadept.mime_types', package.seeall)
 
 ---
--- [Local table] Language names with their associated lexers.
--- @class table
--- @name languages
-local languages = {
-  as = 'actionscript',
-  ada = 'ada',
-  antlr = 'antlr',
-  apdl = 'apdl',
-  applescript = 'applescript',
-  asp = 'asp',
-  awk = 'awk',
-  batch = 'batch',
-  boo = 'boo',
-  cpp = 'cpp',
-  csharp = 'csharp',
-  css = 'css',
-  d = 'd',
-  diff = 'diff',
-  django = 'django',
-  eiffel = 'eiffel',
-  erlang = 'erlang',
-  forth = 'forth',
-  fortran = 'fortran',
-  gap = 'gap',
-  gettext = 'gettext',
-  gnuplot = 'gnuplot',
-  groovy = 'groovy',
-  haskell = 'haskell',
-  html = 'html',
-  idl = 'idl',
-  ini = 'ini',
-  io = 'io',
-  java = 'java',
-  js = 'javascript',
-  latex = 'latex',
-  lisp = 'lisp',
-  lua = 'lua',
-  makefile = 'makefile',
-  maxima = 'maxima',
-  sql = 'sql',
-  objc = 'objective_c',
-  ocaml = 'ocaml',
-  pascal = 'pascal',
-  perl = 'perl',
-  php = 'php',
-  pike = 'pike',
-  postscript = 'postscript',
-  props = 'props',
-  python = 'python',
-  r = 'r',
-  ragel = 'ragel',
-  rebol = 'rebol',
-  rexx = 'rexx',
-  rhtml = 'rhtml',
-  ruby = 'ruby',
-  scheme = 'scheme',
-  shell = 'shellscript',
-  smalltalk = 'smalltalk',
-  tcl = 'tcl',
-  vala = 'vala',
-  verilog = 'verilog',
-  vhdl = 'vhdl',
-  vb = 'visualbasic',
-  xml = 'xml'
-}
-
-local l = languages
----
--- [Local table] File extensions with their associated languages.
+-- [Local table] File extensions with their associated lexers.
 -- @class table
 -- @name extensions
-local extensions = {
-  -- Actionscript
-  as = l.as,
-  -- Ada
-  ada = l.ada, adb = l.ada, ads = l.ada,
-  -- ANTLR
-  g = l.antlr,
-  -- APDL
-  ans = l.apdl,
-  inp = l.apdl,
-  mac = l.apdl,
-  -- Applescript
-  applescript = l.applescript,
-  -- ASP
-  asa = l.asp, asp = l.asp,
-  -- AWK
-  awk = l.awk,
-  -- Batch
-  bat = l.batch,
-  cmd = l.batch,
-  -- Boo
-  boo = l.boo,
-  -- C/C++
-  c = l.cpp, cc = l.cpp, C = l.cpp, cpp = l.cpp, cxx = l.cpp, ['c++'] = l.cpp,
-  h = l.cpp, hh = l.cpp, hpp = l.cpp,
-  -- C#
-  cs = l.csharp,
-  -- CSS
-  css = l.css,
-  -- D
-  d = l.d,
-  -- Diff
-  diff = l.diff,
-  patch = l.diff,
-  -- Eiffel
-  e = l.eiffel, eif = l.eiffel,
-  -- Erlang
-  erl = l.erlang,
-  -- Forth
-  f = l.forth, frt = l.forth, fs = l.forth,
-  -- Fortran
-  ['for'] = l.fortran, fort = l.fortran, f77 = l.fortran, f90 = l.fortran,
-    f95 = l.fortran,
-  -- Gap
-  g = l.gap, gd = l.gap, gi = l.gap, gap = l.gap,
-  -- Gettext
-  po = l.gettext, pot = l.gettext,
-  -- GNUPlot
-  dem = l.gnuplot,
-  plt = l.gnuplot,
-  -- Goovy
-  groovy = l.groovy, grv = l.groovy,
-  -- Haskell
-  hs = l.haskell,
-  -- HTML
-  htm = l.html, html = l.html,
-  shtm = l.html, shtml = l.html,
-  -- IDL
-  idl = l.idl,
-  -- ini
-  ini = l.ini,
-  reg = l.ini,
-  -- Io
-  io = l.io,
-  -- Java
-  bsh = l.java,
-  java = l.java,
-  -- Javascript
-  js = l.js,
-  -- Latex
-  bbl = l.latex,
-  dtx = l.latex,
-  ins = l.latex,
-  ltx = l.latex,
-  tex = l.latex,
-  sty = l.latex,
-  -- Lisp
-  el = l.lisp,
-  lisp = l.lisp,
-  lsp = l.lisp,
-  -- Lua
-  lua = l.lua,
-  -- Makefile
-  GNUmakefile = l.makefile,
-  iface = l.makefile,
-  mak = l.makefile, makefile = l.makefile, Makefile = l.makefile,
-  -- Maxima
-  maxima = l.maxima,
-  -- Objective C
-  m = l.objc,
-  objc = l.objc,
-  -- OCAML,
-  ml = l.ocaml, mli = l.ocaml, mll = l.ocaml, mly = l.ocaml,
-  -- Pascal
-  dpk = l.pascal, dpr = l.pascal,
-  p = l.pascal, pas = l.pascal,
-  -- Perl
-  al = l.perl,
-  perl = l.perl, pl = l.perl, pm = l.perl,
-  -- PHP
-  inc = l.php,
-  php = l.php, php3 = l.php, php4 = l.php, phtml = l.php,
-  -- Pike
-  pike = l.pike, pmod = l.pike,
-  -- Postscript
-  eps = l.postscript,
-  ps = l.postscript,
-  -- Properties
-  props = l.props, properties = l.props,
-  -- Python
-  sc = l.python,
-  py = l.python, pyw = l.python,
-  -- R
-  R = l.r, Rout = l.r, Rhistory = l.r, Rt = l.r, ['Rout.save'] = l.r,
-    ['Rout.fail'] = l.r,
-  -- Rebol
-  r = l.rebol,
-  -- Rexx
-  orx = l.rexx,
-  rex = l.rexx,
-  -- RHTML
-  rhtml = l.rhtml,
-  -- Ruby
-  rb = l.ruby, rbw = l.ruby,
-  -- Ragel
-  rl = l.ragel,
-  -- Scheme
-  scm = l.scheme,
-  -- Shell
-  bash = l.shell,
-  csh = l.shell,
-  sh = l.shell,
-  -- Smalltalk
-  changes = l.smalltalk,
-  st = l.smalltalk, sources = l.smalltalk,
-  -- SQL
-  sql = l.sql,
-  -- TCL
-  tcl = l.tcl, tk = l.tk,
-  -- Vala
-  vala = l.vala,
-  -- Verilog
-  v = l.verilog, ver = l.verilog,
-  -- VHDL
-  vh = l.vhdl, vhd = l.vhdl, vhdl = l.vhdl,
-  -- Visual Basic
-  asa = l.vb,
-  bas = l.vb,
-  cls = l.vb, ctl = l.vb,
-  dob = l.vb, dsm = l.vb, dsr = l.vb,
-  frm = l.vb,
-  pag = l.vb,
-  vb = l.vb, vba = l.vb, vbs = l.vb,
-  -- XML
-  xhtml = l.xml, xml = l.xml, xsd = l.xml, xsl = l.xml, xslt = l.xml
-}
+local extensions = {}
 
 ---
--- [Local table] Shebang words and their associated languages.
+-- [Local table] Shebang words and their associated lexers.
 -- @class table
 -- @name shebangs
-local shebangs = {
-  awk = l.awk,
-  lua = l.lua,
-  perl = l.perl,
-  php = l.php,
-  python = l.python,
-  ruby = l.ruby,
-  sh = l.shell,
-}
+local shebangs = {}
 
 ---
--- [Local table] First-line patterns and their associated languages.
+-- [Local table] First-line patterns and their associated lexers.
 -- @class table
 -- @name patterns
-local patterns = {
-  ['^%s*<%?xml%s'] = l.xml
-}
+local patterns = {}
 
----
--- [Local function] Sets the buffer's lexer language based on a filename.
--- @param filename The filename used to set the lexer language.
--- @return boolean indicating whether or not a lexer language was set.
-local function set_lexer_from_filename(filename)
-  local lexer
-  if filename then
-    local ext = filename:match('[^/\\]+$'):match('[^.]+$')
-    lexer = extensions[ext]
-  end
-  buffer:set_lexer_language(lexer or 'container')
-  return lexer
-end
-
----
--- [Local function] Sets the buffer's lexer language based on a shebang line.
-local function set_lexer_from_sh_bang()
-  local line = buffer:get_line(0)
-  if line:find('^#!') then
-    line = line:gsub('[\\/]', ' ')
-    for word in line:gmatch('%S+') do
-      if shebangs[word] then
-        buffer:set_lexer_language(shebangs[word])
-        return true
+-- Load mime-types from mime_types.conf
+local f = io.open(_HOME..'/core/ext/mime_types.conf', 'rb')
+if f then
+  for line in f:lines() do
+    if not line:find('^%s*%%') then
+      if line:find('^%s*[^#/]') then -- extension definition
+        local ext, lexer_name = line:match('^%s*(.+)%s+(%S+)$')
+        if ext and lexer_name then extensions[ext] = lexer_name end
+      else -- shebang or pattern
+        local ch, text, lexer_name = line:match('^%s*([#/])(.+)%s+(%S+)$')
+        if ch and text and lexer_name then
+          (ch == '#' and shebangs or patterns)[text] = lexer_name
+        end
       end
     end
   end
 end
 
 ---
--- [Local function] Sets the buffer's lexer language based on a pattern that
--- matches its first line.
-local function set_lexer_from_pattern()
-  local line = buffer:get_line(0)
-  for patt, lexer in pairs(patterns) do
-    if line:find(patt) then
-      buffer:set_lexer_language(lexer)
-      return true
-    end
+-- [Local function] Sets the buffer's lexer based on filename, shebang words, or
+-- first line pattern.
+local function set_lexer()
+  local lexer
+  if buffer.filename then
+    lexer = extensions[buffer.filename:match('[^/\\.]+$')]
   end
-end
-
----
--- [Local function] Loads a language module based on a filename (if it hasn't
--- been loaded already).
--- @param filename The filename used to load a language module from.
-local function load_language_module_from_filename(filename)
-  if not filename then return end
-  local ext = filename:match('[^/\\]+$'):match('[^.]+$')
-  local lang = extensions[ext]
-  if lang then
-    local ret, err = pcall(require, lang)
-    if ret then
-      _m[lang].set_buffer_properties()
-    elseif not ret and not err:find("^module '"..lang.."' not found:") then
-      textadept.events.error(err)
+  if not lexer then
+    local line = buffer:get_line(0)
+    if line:find('^#!') then
+      for word in line:gsub('[/\\]', ' '):gmatch('%S+') do
+        if shebangs[word] then
+          lexer = shebangs[word]
+          break
+        end
+      end
+    end
+    if not lexer then
+      for patt, lex in pairs(patterns) do
+        if line:find(patt) then
+          lexer = lex
+          break
+        end
+      end
     end
   end
+  buffer:set_lexer_language(lexer or 'container')
 end
 
 ---
 -- [Local function] Performs actions suitable for a new buffer.
 -- Sets the buffer's lexer language and loads the language module.
 local function handle_new()
-  local buffer = buffer
-  if not set_lexer_from_filename(buffer.filename) then
-    set_lexer_from_sh_bang()
-  end
-  load_language_module_from_filename(buffer.filename)
-end
-
----
--- [Local function] Performs actions suitable for when buffers are switched.
--- Sets the buffer's lexer language.
-local function handle_switch()
-  if not set_lexer_from_filename(buffer.filename) then
-    if not set_lexer_from_sh_bang() then set_lexer_from_pattern() end
+  set_lexer()
+  if buffer.filename then
+    local lang = extensions[buffer.filename:match('[^/\\.]+$')]
+    if lang then
+      local ret, err = pcall(require, lang)
+      if ret then
+        _m[lang].set_buffer_properties()
+      elseif not ret and not err:find("^module '"..lang.."' not found:") then
+        textadept.events.error(err)
+      end
+    end
   end
 end
 
 textadept.events.add_handler('file_opened', handle_new)
 textadept.events.add_handler('file_saved_as', handle_new)
-textadept.events.add_handler('buffer_switch', handle_switch)
-textadept.events.add_handler('view_new', handle_switch)
+textadept.events.add_handler('buffer_switch', set_lexer)
+textadept.events.add_handler('view_new', set_lexer)