Write heuristics for perl, closes #550v4.10.2

2 files changed, 102 insertions, 9 deletions

diff --git a/autoload/polyglot.vim b/autoload/polyglot.vim
index 29287ece..5d4bfe9a 100644
--- a/autoload/polyglot.vim
+++ b/autoload/polyglot.vim
@@ -10,11 +10,11 @@ func! s:Setf(ft)
   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
 
@@ -52,6 +52,8 @@ let s:interpreters = {
   \ 'cperl': 'perl',
   \ 'perl': 'perl',
   \ 'php': 'php',
+  \ 'swipl': 'prolog',
+  \ 'yap': 'prolog',
   \ 'pwsh': 'ps1',
   \ 'python': 'python',
   \ 'python2': 'python',
@@ -95,7 +97,7 @@ let s:r_hashbang = '^#!\s*\(\S\+\)\s*\(.*\)\s*'
 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 !~# "^#!"
@@ -133,7 +135,7 @@ func! polyglot#Shebang()
 endfunc
 
 func! polyglot#DetectInpFiletype()
-  let line = getline(1)
+  let line = getline(nextnonblank(1))
   if line =~# '^\*'
     call s:Setf('abaqus') | return
   endif
@@ -269,10 +271,10 @@ func! polyglot#DetectIdrFiletype()
     if line =~# '^%access .*'
       call s:Setf('idris') | return
     endif
-    if exists("g:filetype_idr")
-      call s:Setf(g:filetype_idr) | return
-    endif
   endfor
+  if exists("g:filetype_idr")
+    call s:Setf(g:filetype_idr) | return
+  endif
   call s:Setf('idris2') | return
 endfunc
 
@@ -296,6 +298,89 @@ func! polyglot#DetectBasFiletype()
   call s:Setf('basic') | return
 endfunc
 
+func! polyglot#DetectPmFiletype()
+  let line = getline(nextnonblank(1))
+  if line =~# 'XPM2'
+    call s:Setf('xpm2') | return
+  endif
+  if line =~# 'XPM'
+    call s:Setf('xpm') | return
+  endif
+  for lnum in range(1, min([line("$"), 50]))
+    let line = getline(lnum)
+    if line =~# '^\s*\%(use\s\+v6\(\<\|\>\)\|\(\<\|\>\)module\(\<\|\>\)\|\(\<\|\>\)\%(my\s\+\)\=class\(\<\|\>\)\)'
+      call s:Setf('raku') | return
+    endif
+    if line =~# '\(\<\|\>\)use\s\+\%(strict\(\<\|\>\)\|v\=5\.\)'
+      call s:Setf('perl') | return
+    endif
+  endfor
+  if exists("g:filetype_pm")
+    call s:Setf(g:filetype_pm) | return
+  endif
+  call s:Setf('perl') | return
+endfunc
+
+func! polyglot#DetectPlFiletype()
+  let line = getline(nextnonblank(1))
+  if line =~# '^[^#]*:-' || line =~# '^\s*\%(%\|/\*\)' || line =~# '\.\s*$'
+    call s:Setf('prolog') | return
+  endif
+  for lnum in range(1, min([line("$"), 50]))
+    let line = getline(lnum)
+    if line =~# '^\s*\%(use\s\+v6\(\<\|\>\)\|\(\<\|\>\)module\(\<\|\>\)\|\(\<\|\>\)\%(my\s\+\)\=class\(\<\|\>\)\)'
+      call s:Setf('raku') | return
+    endif
+    if line =~# '\(\<\|\>\)use\s\+\%(strict\(\<\|\>\)\|v\=5\.\)'
+      call s:Setf('perl') | return
+    endif
+  endfor
+  if exists("g:filetype_pl")
+    call s:Setf(g:filetype_pl) | return
+  endif
+  call s:Setf('perl') | return
+endfunc
+
+func! polyglot#DetectTFiletype()
+  for lnum in range(1, min([line("$"), 5]))
+    let line = getline(lnum)
+    if line =~# '^\.'
+      call s:Setf('nroff') | return
+    endif
+  endfor
+  for lnum in range(1, min([line("$"), 50]))
+    let line = getline(lnum)
+    if line =~# '^\s*\%(use\s\+v6\(\<\|\>\)\|\(\<\|\>\)module\(\<\|\>\)\|\(\<\|\>\)\%(my\s\+\)\=class\(\<\|\>\)\)'
+      call s:Setf('raku') | return
+    endif
+    if line =~# '\(\<\|\>\)use\s\+\%(strict\(\<\|\>\)\|v\=5\.\)'
+      call s:Setf('perl') | return
+    endif
+  endfor
+  if exists("g:filetype_t")
+    call s:Setf(g:filetype_t) | return
+  endif
+  call s:Setf('perl') | return
+endfunc
+
+func! polyglot#DetectTt2Filetype()
+  for lnum in range(1, min([line("$"), 3]))
+    let line = getline(lnum)
+    if line =~? '<\%(!DOCTYPE HTML\|[%?]\|html\)'
+      call s:Setf('tt2html') | return
+    endif
+  endfor
+  call s:Setf('tt2') | return
+endfunc
+
+func! polyglot#DetectHtmlFiletype()
+  let line = getline(nextnonblank(1))
+  if line =~# '^\(%\|<[%&].*>\)'
+    call s:Setf('mason') | return
+  endif
+  call s:Setf('html') | return
+endfunc
+
 " Restore 'cpoptions'
 let &cpo = s:cpo_save
 unlet s:cpo_save
diff --git a/autoload/sleuth.vim b/autoload/sleuth.vim
index c1de82cd..ae101289 100644
--- a/autoload/sleuth.vim
+++ b/autoload/sleuth.vim
@@ -38,6 +38,7 @@ let s:globs = {
   \ 'basic': '*.basic',
   \ 'blade': '*.blade,*.blade.php',
   \ 'brewfile': 'Brewfile',
+  \ 'bzl': '*.bzl,BUCK,BUILD,BUILD.bazel,Tiltfile,WORKSPACE',
   \ 'c': '*.c,*.cats,*.h,*.idc,*.qc',
   \ 'caddyfile': 'Caddyfile',
   \ 'carp': '*.carp',
@@ -129,6 +130,7 @@ let s:globs = {
   \ 'mako': '*.mako,*.mao',
   \ 'markdown': '*.md,*.markdown,*.mdown,*.mdwn,*.mkd,*.mkdn,*.mkdown,*.ronn,*.workbook,contents.lr',
   \ 'markdown.mdx': '*.mdx',
+  \ 'mason': '*.mason,*.mhtml,*.comp',
   \ 'meson': 'meson.build,meson_options.txt',
   \ 'mma': '*.mathematica,*.cdf,*.m,*.ma,*.mt,*.nb,*.nbp,*.wl,*.wlt,*.wls,*.mma',
   \ 'moon': '*.moon',
@@ -151,7 +153,9 @@ let s:globs = {
   \ 'perl': '*.pl,*.al,*.cgi,*.fcgi,*.perl,*.ph,*.plx,*.pm,*.psgi,*.t,Makefile.PL,Rexfile,ack,cpanfile',
   \ 'php': '*.php,*.aw,*.ctp,*.fcgi,*.inc,*.php3,*.php4,*.php5,*.phps,*.phpt,Phakefile',
   \ 'plantuml': '*.puml,*.iuml,*.plantuml,*.uml,*.pu',
+  \ 'pod': '*.pod',
   \ 'pony': '*.pony',
+  \ 'prolog': '*.pl,*.pro,*.prolog,*.yap',
   \ 'proto': '*.proto',
   \ 'ps1': '*.ps1,*.psd1,*.psm1,*.pssc',
   \ 'ps1xml': '*.ps1xml',
@@ -190,6 +194,7 @@ let s:globs = {
   \ 'sxhkdrc': '*.sxhkdrc,sxhkdrc',
   \ 'systemd': '*.automount,*.mount,*.path,*.service,*.socket,*.swap,*.target,*.timer',
   \ 'tablegen': '*.td',
+  \ 'tads': '*.t',
   \ 'terraform': '*.hcl,*.nomad,*.tf,*.tfvars,*.workflow',
   \ 'textile': '*.textile',
   \ 'thrift': '*.thrift',
@@ -197,6 +202,8 @@ let s:globs = {
   \ 'toml': '*.toml,Cargo.lock,Gopkg.lock,poetry.lock,Pipfile',
   \ 'tptp': '*.p,*.tptp,*.ax',
   \ 'trasys': '*.inp',
+  \ 'tt2': '*.tt2',
+  \ 'tt2html': '*.tt2',
   \ 'typescript': '*.ts',
   \ 'typescriptreact': '*.tsx',
   \ 'unison': '*.u,*.uu',
@@ -211,6 +218,7 @@ let s:globs = {
   \ 'xdc': '*.xdc',
   \ 'xml': '*.xml,*.adml,*.admx,*.ant,*.axml,*.builds,*.ccproj,*.ccxml,*.clixml,*.cproject,*.cscfg,*.csdef,*.csl,*.csproj,*.ct,*.depproj,*.dita,*.ditamap,*.ditaval,*.dll.config,*.dotsettings,*.filters,*.fsproj,*.fxml,*.glade,*.gml,*.gmx,*.grxml,*.gst,*.iml,*.ivy,*.jelly,*.jsproj,*.kml,*.launch,*.mdpolicy,*.mjml,*.mm,*.mod,*.mxml,*.natvis,*.ncl,*.ndproj,*.nproj,*.nuspec,*.odd,*.osm,*.pkgproj,*.pluginspec,*.proj,*.props,*.ps1xml,*.psc1,*.pt,*.rdf,*.resx,*.rss,*.sch,*.scxml,*.sfproj,*.shproj,*.srdf,*.storyboard,*.sublime-snippet,*.targets,*.tml,*.ui,*.urdf,*.ux,*.vbproj,*.vcxproj,*.vsixmanifest,*.vssettings,*.vstemplate,*.vxml,*.wixproj,*.workflow,*.wsdl,*.wsf,*.wxi,*.wxl,*.wxs,*.x3d,*.xacro,*.xaml,*.xib,*.xlf,*.xliff,*.xmi,*.xml.dist,*.xproj,*.xsd,*.xspec,*.xul,*.zcml,*.cdxml,App.config,NuGet.config,Settings.StyleCop,Web.Debug.config,Web.Release.config,Web.config,packages.config',
   \ 'xml.twig': '*.xml.twig',
+  \ 'xs': '*.xs',
   \ 'xsl': '*.xslt,*.xsl',
   \ 'yaml': '*.yml,*.mir,*.reek,*.rviz,*.sublime-syntax,*.syntax,*.yaml,*.yaml-tmlanguage,*.yaml.sed,*.yml.mysql,glide.lock,yarn.lock,fish_history,fish_read_history',
   \ 'yaml.ansible': 'playbook.y{a,}ml,site.y{a,}ml,main.y{a,}ml,local.y{a,}ml,requirements.y{a,}ml,tasks.*.y{a,}ml,roles.*.y{a,}ml,handlers.*.y{a,}ml',