Write heuristics for perl, closes #550v4.10.2

3 files changed, 97 insertions, 7 deletions

diff --git a/scripts/build b/scripts/build
index 07bc301b..c44e8cc8 100755
--- a/scripts/build
+++ b/scripts/build
@@ -263,7 +263,7 @@ def rule_to_code(rule)
   if rule.has_key?("lines")
     if rule["lines"] == 1
       return <<~EOS
-        let line = getline(1)
+        let line = getline(nextnonblank(1))
 
         #{indent(rule_to_code(except(rule, "lines")), 0)}
       EOS
@@ -423,6 +423,7 @@ def generate_ftdetect(packages, heuristics)
   expected_filetypes = expected_filetypes.select { |e| filetype_names.include?(e["name"]) }
 
   native_extensions = Set.new(native_filetypes.flat_map { |f| f["extensions"] || [] })
+  native_filenames = Set.new(native_filetypes.flat_map { |f| f["filenames"] || [] })
 
   for package in packages
     name = package.fetch("name")
@@ -435,6 +436,12 @@ def generate_ftdetect(packages, heuristics)
           to_disable << "*." + extension
         end
       end
+
+      for filename in filetype["filenames"]
+        if native_filenames.include?(filename)
+          to_disable << filename
+        end
+      end
     end
 
     if to_disable.size > 0
@@ -457,7 +464,7 @@ def generate_ftdetect(packages, heuristics)
       set_command = "call s:Setf('#{name}')"
 
       if filetype["syntax"]
-        set_command = "if !did_filetype() | set ft=#{name} syntax=#{filetype["syntax"]} | endif"
+        set_command = "set ft=#{name} syntax=#{filetype["syntax"]}"
       end
 
       if filetype["custom_set"]
@@ -544,11 +551,11 @@ def generate_ftdetect(packages, heuristics)
       endif
     endfunc
 
-    func! polyglot#Heuristics()
+    func! polyglot#Shebang()
       " Try to detect filetype from shebang
-      let l:filetype = polyglot#Shebang()
-      if l:filetype != ""
-        exec "setf " . l:filetype
+      let ft = polyglot#ShebangFiletype()
+      if ft != ""
+        call s:Setf(ft)
         return 1
       endif
 
@@ -571,7 +578,7 @@ def generate_ftdetect(packages, heuristics)
     let s:r_envflag = '%(\\S\\+=\\S\\+\\|-[iS]\\|--ignore-environment\\|--split-string\\)'
     let s:r_env = '^\\%(\\' . s:r_envflag . '\\s\\+\\)*\\(\\S\\+\\)'
 
-    func! polyglot#Shebang()
+    func! polyglot#ShebangFiletype()
       let l:line1 = getline(1)
 
       if l:line1 !~# "^#!"
diff --git a/scripts/test_extensions.vim b/scripts/test_extensions.vim
index f2aaf66c..467ac86a 100644
--- a/scripts/test_extensions.vim
+++ b/scripts/test_extensions.vim
@@ -15,6 +15,12 @@ function! TestExtension(filetype, filename, content)
   endtry
 endfunction
 
+" make sure native vim scripts.vim is respected
+call TestExtension("rib", "renderman", "##RenderMan")
+
+" make sure case of file does matter when recognizing file
+call TestExtension("ruby", "scripts/build", "#!/usr/bin/env ruby")
+
 call TestExtension("sh", "bash1", "#!/bin/bash")
 call TestExtension("sh", "bash2", "#! /bin/bash")
 call TestExtension("sh", "bash3", "#! /bin/bash2.3")
@@ -279,3 +285,72 @@ call TestExtension("c", "foo.h", "")
 unlet g:c_syntax_for_h
 let g:ch_syntax_for_h = 1
 call TestExtension("ch", "foo.h", "")
+
+" perl
+call TestExtension("perl", "empty.plx", "")
+call TestExtension("perl", "empty.al", "")
+call TestExtension("perl", "empty.psgi", "")
+call TestExtension("pod", "empty.pod", "")
+
+" raku
+call TestExtension("raku", "empty.p6", "")
+call TestExtension("raku", "empty.pm6", "")
+call TestExtension("raku", "empty.pl6", "")
+call TestExtension("raku", "empty.raku", "")
+call TestExtension("raku", "empty.rakumod", "")
+call TestExtension("raku", "empty.pod6", "")
+call TestExtension("raku", "empty.rakudoc", "")
+call TestExtension("raku", "empty.rakutest", "")
+call TestExtension("raku", "empty.t6", "")
+
+
+" .pm extension
+call TestExtension("perl", "empty.pm", "")
+call TestExtension("perl", "strict.pm", " use strict hello;")
+call TestExtension("perl", "use5.pm", " use 5;")
+call TestExtension("perl", "usev5.pm", " use v5;")
+call TestExtension("raku", "script.pm", "#!/usr/bin/env perl6\nprint('Hello world')")
+call TestExtension("raku", "class.pm", " class Class {}")
+call TestExtension("raku", "module.pm", " module foobar")
+call TestExtension("xpm", "xpm.pm", "/* XPM */")
+call TestExtension("xpm2", "xpm2.pm", "/* XPM2 */")
+let g:filetype_pm = "fizfuz"
+call TestExtension("fizfuz", "fizfuz.pm", "")
+
+" .pl extension
+call TestExtension("perl", "empty.pl", "")
+call TestExtension("prolog", "comment.pl", "% hello world")
+call TestExtension("prolog", "comment2.pl", "/* hello world */")
+call TestExtension("prolog", "statement.pl", "happy(vincent). ")
+call TestExtension("prolog", "statement2.pl", "nearbychk(X,Y) :- Y is X-1.")
+call TestExtension("perl", "strict.pl", " use strict hello;")
+call TestExtension("perl", "use5.pl", " use 5;")
+call TestExtension("perl", "usev5.pl", " use v5;")
+call TestExtension("raku", "script.pl", "#!/usr/bin/env perl6\nprint('Hello world')")
+call TestExtension("raku", "class.pl", " class Class {}")
+call TestExtension("raku", "module.pl", " module foobar")
+let g:filetype_pl = "fizfuz"
+call TestExtension("fizfuz", "fizfuz.pl", "")
+
+" .t extension
+call TestExtension("perl", "empty.t", "")
+call TestExtension("perl", "strict.t", " use strict hello;")
+call TestExtension("perl", "use5.t", " use 5;")
+call TestExtension("perl", "usev5.t", " use v5;")
+call TestExtension("raku", "script.t", "#!/usr/bin/env perl6\nprint('Hello world')")
+call TestExtension("raku", "class.t", " class Class {}")
+call TestExtension("raku", "module.t", " module foobar")
+call TestExtension("nroff", "module.t", ".nf\n101 Main Street")
+let g:filetype_t = "fizfuz"
+call TestExtension("fizfuz", "fizfuz.t", "")
+
+" .tt2 extension
+call TestExtension("tt2", "empty.tt2", "")
+call TestExtension("tt2html", "doctype.tt2", "<!DOCTYPE HTML>")
+call TestExtension("tt2html", "percent.tt2", "<%filter>")
+call TestExtension("tt2html", "html.tt2", "<html>")
+
+" .html extension
+call TestExtension("html", "empty.html", "")
+call TestExtension("mason", "mason1.html", "% my $planet = 42;")
+call TestExtension("mason", "mason2.html", "<%filter></%filter>")
diff --git a/scripts/test_filetypes.vim b/scripts/test_filetypes.vim
index a84b39e3..02d44ca4 100644
--- a/scripts/test_filetypes.vim
+++ b/scripts/test_filetypes.vim
@@ -154,6 +154,11 @@ call TestFiletype('sexplib')
 call TestFiletype('octave')
 call TestFiletype('opencl')
 call TestFiletype('perl')
+call TestFiletype('pod')
+call TestFiletype('mason')
+call TestFiletype('tt2')
+call TestFiletype('tt2html')
+call TestFiletype('xs')
 call TestFiletype('sql')
 call TestFiletype('cql')
 call TestFiletype('php')
@@ -232,3 +237,6 @@ call TestFiletype('basic')
 call TestFiletype('vb')
 call TestFiletype('dosini')
 call TestFiletype('odin')
+call TestFiletype('bzl')
+call TestFiletype('prolog')
+call TestFiletype('tads')