path: root/runtime/syntax/pandoc.vim

scriptencoding utf-8
"
" Language:	Pandoc (superset of Markdown)
" Maintainer:	Felipe Morales <hel.sheep@gmail.com>
" Maintainer:	Caleb Maclennan <caleb@alerque.com>
" Upstream:	https://github.com/vim-pandoc/vim-pandoc-syntax
"
" Contributor:	David Sanson <dsanson@gmail.com>
"		Jorge Israel Peña <jorge.israel.p@gmail.com>
" Original Author:	Jeremy Schultz <taozhyn@gmail.com>
" Version: 5.0
" Last Change:	2024 Apr 08

let s:cpo_save = &cpo
set cpo&vim

" Configuration: {{{1
"
" use conceal? {{{2
if !exists('g:pandoc#syntax#conceal#use')
    let g:pandoc#syntax#conceal#use = 1
endif
"}}}2

" what groups not to use conceal in. works as a blacklist {{{2
if !exists('g:pandoc#syntax#conceal#blacklist')
    let g:pandoc#syntax#conceal#blacklist = []
endif
" }}}2

" cchars used in conceal rules {{{2
" utf-8 defaults (preferred)
if &encoding ==# 'utf-8'
    let s:cchars = {
                \'newline': '↵',
                \'image': '▨',
                \'super': 'ⁿ',
                \'sub': 'ₙ',
                \'strike': 'x̶',
                \'atx': '§',
                \'codelang': 'λ',
                \'codeend': '—',
                \'abbrev': '→',
                \'footnote': '†',
                \'definition': ' ',
                \'li': '•',
                \'html_c_s': '‹',
                \'html_c_e': '›',
                \'quote_s': '“',
                \'quote_e': '”'}
else
    " ascii defaults
    let s:cchars = {
                \'newline': ' ',
                \'image': 'i',
                \'super': '^',
                \'sub': '_',
                \'strike': '~',
                \'atx': '#',
                \'codelang': 'l',
                \'codeend': '-',
                \'abbrev': 'a',
                \'footnote': 'f',
                \'definition': ' ',
                \'li': '*',
                \'html_c_s': '+',
                \'html_c_e': '+'}
endif
" }}}2

" if the user has a dictionary with replacements for the default cchars, use those {{{2
if exists('g:pandoc#syntax#conceal#cchar_overrides')
    let s:cchars = extend(s:cchars, g:pandoc#syntax#conceal#cchar_overrides)
endif
" }}}2

"should the urls in links be concealed? {{{2
if !exists('g:pandoc#syntax#conceal#urls')
    let g:pandoc#syntax#conceal#urls = 0
endif
" should backslashes in escapes be concealed? {{{2
if !exists('g:pandoc#syntax#conceal#backslash')
    let g:pandoc#syntax#conceal#backslash = 0
endif
" }}}2

" leave specified codeblocks as Normal (i.e. 'unhighlighted') {{{2
if !exists('g:pandoc#syntax#codeblocks#ignore')
    let g:pandoc#syntax#codeblocks#ignore = []
endif
" }}}2

" use embedded highlighting for delimited codeblocks where a language is specifed. {{{2
if !exists('g:pandoc#syntax#codeblocks#embeds#use')
    let g:pandoc#syntax#codeblocks#embeds#use = 1
endif
" }}}2

" for what languages and using what vim syntax files highlight those embeds. {{{2
" defaults to None.
if !exists('g:pandoc#syntax#codeblocks#embeds#langs')
    let g:pandoc#syntax#codeblocks#embeds#langs = []
endif
" }}}2

" use italics ? {{{2
if !exists('g:pandoc#syntax#style#emphases')
    let g:pandoc#syntax#style#emphases = 1
endif
" if 0, we don't conceal the emphasis marks, otherwise there wouldn't be a way
" to tell where the styles apply.
if g:pandoc#syntax#style#emphases == 0
    call add(g:pandoc#syntax#conceal#blacklist, 'block')
endif
" }}}2

" underline subscript, superscript and strikeout? {{{2
if !exists('g:pandoc#syntax#style#underline_special')
    let g:pandoc#syntax#style#underline_special = 1
endif
" }}}2

" protect code blocks? {{{2
if !exists('g:pandoc#syntax#protect#codeblocks')
    let g:pandoc#syntax#protect#codeblocks = 1
endif
" }}}2

" use color column? {{{2
if !exists('g:pandoc#syntax#colorcolumn')
    let g:pandoc#syntax#colorcolumn = 0
endif
" }}}2

" highlight new lines? {{{2
if !exists('g:pandoc#syntax#newlines')
    let g:pandoc#syntax#newlines = 1
endif
" }}}

" detect roman-numeral list items? {{{2
if !exists('g:pandoc#syntax#roman_lists')
    let g:pandoc#syntax#roman_lists = 0
endif
" }}}2

" disable syntax highlighting for definition lists? (better performances) {{{2
if !exists('g:pandoc#syntax#use_definition_lists')
    let g:pandoc#syntax#use_definition_lists = 1
endif
" }}}2

" }}}1

" Functions: {{{1
" EnableEmbedsforCodeblocksWithLang {{{2
function! EnableEmbedsforCodeblocksWithLang(entry)
    " prevent embedded language syntaxes from changing 'foldmethod' 
    if has('folding')
        let s:foldmethod = &l:foldmethod
        let s:foldtext = &l:foldtext     
    endif

    try
        let s:langname = matchstr(a:entry, '^[^=]*')
        let s:langsyntaxfile = matchstr(a:entry, '[^=]*$')
        unlet! b:current_syntax
        exe 'syn include @'.toupper(s:langname).' syntax/'.s:langsyntaxfile.'.vim'
        " We might have just turned off spellchecking by including the file,
        " so we turn it back on here.
        exe 'syntax spell toplevel'
        exe 'syn region pandocDelimitedCodeBlock_' . s:langname . ' start=/\(\_^\( \+\|\t\)\=\(`\{3,}`*\|\~\{3,}\~*\)\s*\%({[^.]*\.\)\=' . s:langname . '\>.*\n\)\@<=\_^/' .
                    \' end=/\_$\n\(\( \+\|\t\)\=\(`\{3,}`*\|\~\{3,}\~*\)\_$\n\_$\)\@=/ contained containedin=pandocDelimitedCodeBlock' .
                    \' contains=@' . toupper(s:langname)
        exe 'syn region pandocDelimitedCodeBlockinBlockQuote_' . s:langname . ' start=/>\s\(`\{3,}`*\|\~\{3,}\~*\)\s*\%({[^.]*\.\)\=' . s:langname . '\>/' .
                    \ ' end=/\(`\{3,}`*\|\~\{3,}\~*\)/ contained containedin=pandocDelimitedCodeBlock' .
                    \' contains=@' . toupper(s:langname) .
                    \',pandocDelimitedCodeBlockStart,pandocDelimitedCodeBlockEnd,pandodDelimitedCodeblockLang,pandocBlockQuoteinDelimitedCodeBlock'
    catch /E484/
      echo "No syntax file found for '" . s:langsyntaxfile . "'"
    endtry

    if exists('s:foldmethod') && s:foldmethod !=# &l:foldmethod
        let &l:foldmethod = s:foldmethod
    endif
    if exists('s:foldtext') && s:foldtext !=# &l:foldtext
        let &l:foldtext = s:foldtext
    endif
endfunction
" }}}2

" DisableEmbedsforCodeblocksWithLang {{{2
function! DisableEmbedsforCodeblocksWithLang(langname)
    try
      exe 'syn clear pandocDelimitedCodeBlock_'.a:langname
      exe 'syn clear pandocDelimitedCodeBlockinBlockQuote_'.a:langname
    catch /E28/
      echo "No existing highlight definitions found for '" . a:langname . "'"
    endtry
endfunction
" }}}2

" WithConceal {{{2
function! s:WithConceal(rule_group, rule, conceal_rule)
    let l:rule_tail = ''
    if g:pandoc#syntax#conceal#use != 0
        if index(g:pandoc#syntax#conceal#blacklist, a:rule_group) == -1
            let l:rule_tail = ' ' . a:conceal_rule
        endif
    endif
    execute a:rule . l:rule_tail
endfunction
" }}}2

" }}}1

" Commands: {{{1
command! -buffer -nargs=1 -complete=syntax PandocHighlight call EnableEmbedsforCodeblocksWithLang(<f-args>)
command! -buffer -nargs=1 -complete=syntax PandocUnhighlight call DisableEmbedsforCodeblocksWithLang(<f-args>)
" }}}1

" BASE:
syntax clear
syntax spell toplevel
" }}}1

" Syntax Rules: {{{1

" Embeds: {{{2

" prevent embedded language syntaxes from changing 'foldmethod' 
if has('folding')
    let s:foldmethod = &l:foldmethod
endif

" HTML: {{{3
" Set embedded HTML highlighting
syn include @HTML syntax/html.vim
syn match pandocHTML /<\/\?\a\_.\{-}>/ contains=@HTML
" Support HTML multi line comments
syn region pandocHTMLComment start=/<!--\s\=/ end=/\s\=-->/ keepend contains=pandocHTMLCommentStart,pandocHTMLCommentEnd
call s:WithConceal('html_c_s', 'syn match pandocHTMLCommentStart /<!--/ contained', 'conceal cchar='.s:cchars['html_c_s'])
call s:WithConceal('html_c_e', 'syn match pandocHTMLCommentEnd /-->/ contained', 'conceal cchar='.s:cchars['html_c_e'])
" }}}3

" LaTeX: {{{3
" Set embedded LaTex (pandoc extension) highlighting
" Unset current_syntax so the 2nd include will work
unlet b:current_syntax
syn include @LATEX syntax/tex.vim
if index(g:pandoc#syntax#conceal#blacklist, 'inlinemath') == -1
    " Can't use WithConceal here because it will mess up all other conceals
    " when dollar signs are used normally. It must be skipped entirely if
    " inlinemath is blacklisted
    syn region pandocLaTeXInlineMath start=/\v\\@<!\$\S@=/ end=/\v\\@<!\$\d@!/ keepend contains=@LATEX
    syn region pandocLaTeXInlineMath start=/\\\@<!\\(/ end=/\\\@<!\\)/ keepend contains=@LATEX
endif
syn match pandocEscapedDollar /\\\$/ conceal cchar=$
syn match pandocProtectedFromInlineLaTeX /\\\@<!\${.*}\(\(\s\|[[:punct:]]\)\([^$]*\|.*\(\\\$.*\)\{2}\)\n\n\|$\)\@=/ display
" contains=@LATEX
syn region pandocLaTeXMathBlock start=/\$\$/ end=/\$\$/ keepend contains=@LATEX
syn region pandocLaTeXMathBlock start=/\\\@<!\\\[/ end=/\\\@<!\\\]/ keepend contains=@LATEX
syn match pandocLaTeXCommand /\\[[:alpha:]]\+\(\({.\{-}}\)\=\(\[.\{-}\]\)\=\)*/ contains=@LATEX
syn region pandocLaTeXRegion start=/\\begin{\z(.\{-}\)}/ end=/\\end{\z1}/ keepend contains=@LATEX
" we rehighlight sectioning commands, because otherwise tex.vim captures all text until EOF or a new sectioning command
syn region pandocLaTexSection start=/\\\(part\|chapter\|\(sub\)\{,2}section\|\(sub\)\=paragraph\)\*\=\(\[.*\]\)\={/ end=/\}/ keepend
syn match pandocLaTexSectionCmd /\\\(part\|chapter\|\(sub\)\{,2}section\|\(sub\)\=paragraph\)/ contained containedin=pandocLaTexSection
syn match pandocLaTeXDelimiter /[[\]{}]/ contained containedin=pandocLaTexSection
" }}}3

if exists('s:foldmethod') && s:foldmethod !=# &l:foldmethod
    let &l:foldmethod = s:foldmethod
endif

" }}}2

" Titleblock: {{{2
syn region pandocTitleBlock start=/\%^%/ end=/\n\n/ contains=pandocReferenceLabel,pandocReferenceURL,pandocNewLine
call s:WithConceal('titleblock', 'syn match pandocTitleBlockMark /%\ / contained containedin=pandocTitleBlock,pandocTitleBlockTitle', 'conceal')
syn match pandocTitleBlockTitle /\%^%.*\n/ contained containedin=pandocTitleBlock
" }}}2

" Blockquotes: {{{2
syn match pandocBlockQuote /^\s\{,3}>.*\n\(.*\n\@1<!\n\)*/ contains=@Spell,pandocEmphasis,pandocStrong,pandocPCite,pandocSuperscript,pandocSubscript,pandocStrikeout,pandocUListItem,pandocNoFormatted,pandocAmpersandEscape,pandocLaTeXInlineMath,pandocEscapedDollar,pandocLaTeXCommand,pandocLaTeXMathBlock,pandocLaTeXRegion skipnl
syn match pandocBlockQuoteMark /\_^\s\{,3}>/ contained containedin=pandocEmphasis,pandocStrong,pandocPCite,pandocSuperscript,pandocSubscript,pandocStrikeout,pandocUListItem,pandocNoFormatted
" }}}2

" Code Blocks: {{{2
if g:pandoc#syntax#protect#codeblocks == 1
    syn match pandocCodeblock /\([ ]\{4}\|\t\).*$/
endif
syn region pandocCodeBlockInsideIndent   start=/\(\(\d\|\a\|*\).*\n\)\@<!\(^\(\s\{8,}\|\t\+\)\).*\n/ end=/.\(\n^\s*\n\)\@=/ contained
" }}}2

" Links: {{{2

" Base: {{{3
syn region pandocReferenceLabel matchgroup=pandocOperator start=/!\{,1}\\\@<!\^\@<!\[/ skip=/\(\\\@<!\]\]\@=\|`.*\\\@<!].*`\)/ end=/\\\@<!\]/ keepend display
if g:pandoc#syntax#conceal#urls == 1
    syn region pandocReferenceURL matchgroup=pandocOperator start=/\]\@1<=(/ end=/)/ keepend conceal
else
    syn region pandocReferenceURL matchgroup=pandocOperator start=/\]\@1<=(/ end=/)/ keepend
endif
" let's not consider "a [label] a" as a label, remove formatting - Note: breaks implicit links
syn match pandocNoLabel /\]\@1<!\(\s\{,3}\|^\)\[[^\[\]]\{-}\]\(\s\+\|$\)[\[(]\@!/ contains=pandocPCite
syn match pandocLinkTip /\s*".\{-}"/ contained containedin=pandocReferenceURL contains=@Spell,pandocAmpersandEscape display
call s:WithConceal('image', 'syn match pandocImageIcon /!\[\@=/ display', 'conceal cchar='. s:cchars['image'])
" }}}3

" Definitions: {{{3
syn region pandocReferenceDefinition start=/\[.\{-}\]:/ end=/\(\n\s*".*"$\|$\)/ keepend
syn match pandocReferenceDefinitionLabel /\[\zs.\{-}\ze\]:/ contained containedin=pandocReferenceDefinition display
syn