Merge pull request #13333 from adrian5/filetype-xml

runtime: Patch xml, xmllint, xmlformat filetypes

1 files changed, 126 insertions, 38 deletions

diff --git a/runtime/autoload/xmlformat.vim b/runtime/autoload/xmlformat.vim
index ea89401977..712e6f7351 100644
--- a/runtime/autoload/xmlformat.vim
+++ b/runtime/autoload/xmlformat.vim
@@ -1,6 +1,6 @@
 " Vim plugin for formatting XML
-" Last Change: Thu, 07 Dec 2018
-"     Version: 0.1
+" Last Change: 2020 Jan 06
+"     Version: 0.3
 "      Author: Christian Brabandt <cb@256bit.org>
 "  Repository: https://github.com/chrisbra/vim-xml-ftplugin
 "     License: VIM License
@@ -15,51 +15,92 @@ let s:keepcpo       = &cpo
 set cpo&vim
 
 " Main function: Format the input {{{1
-func! xmlformat#Format()
+func! xmlformat#Format() abort
   " only allow reformatting through the gq command
   " (e.g. Vim is in normal mode)
   if mode() != 'n'
     " do not fall back to internal formatting
     return 0
   endif
+  let count_orig = v:count
   let sw  = shiftwidth()
   let prev = prevnonblank(v:lnum-1)
   let s:indent = indent(prev)/sw
   let result = []
   let lastitem = prev ? getline(prev) : ''
   let is_xml_decl = 0
-  " split on `<`, but don't split on very first opening <
-  for item in split(join(getline(v:lnum, (v:lnum + v:count - 1))), '.\@<=[>]\zs')
-    if s:EndTag(item)
-      let s:indent = s:DecreaseIndent()
-      call add(result, s:Indent(item))
-    elseif s:EmptyTag(lastitem)
-      call add(result, s:Indent(item))
-    elseif s:StartTag(lastitem) && s:IsTag(item)
-      let s:indent += 1
-      call add(result, s:Indent(item))
-     else
-       if !s:IsTag(item)
-         " Simply split on '<'
-         let t=split(item, '.<\@=\zs')
-         let s:indent+=1
-         call add(result, s:Indent(t[0]))
-         let s:indent = s:DecreaseIndent()
-         call add(result, s:Indent(t[1]))
-       else
+  " go through every line, but don't join all content together and join it
+  " back. We might lose empty lines
+  let list = getline(v:lnum, (v:lnum + count_orig - 1))
+  let current = 0
+  for line in list
+    " Keep empty input lines?
+    if empty(line)
+      call add(result, '')
+      continue
+    elseif line !~# '<[/]\?[^>]*>'
+      let nextmatch = match(list, '<[/]\?[^>]*>', current)
+      if nextmatch > -1
+        let line .= ' '. join(list[(current + 1):(nextmatch-1)], " ")
+        call remove(list, current+1, nextmatch-1)
+      endif
+    endif
+    " split on `>`, but don't split on very first opening <
+    " this means, items can be like ['<tag>', 'tag content</tag>']
+    for item in split(line, '.\@<=[>]\zs')
+      if s:EndTag(item)
+        call s:DecreaseIndent()
+        call add(result, s:Indent(item))
+      elseif s:EmptyTag(lastitem)
         call add(result, s:Indent(item))
+      elseif s:StartTag(lastitem) && s:IsTag(item)
+        let s:indent += 1
+        call add(result, s:Indent(item))
+      else
+        if !s:IsTag(item)
+          " Simply split on '<', if there is one,
+          " but reformat according to &textwidth
+          let t=split(item, '.<\@=\zs')
+
+          " if the content fits well within a single line, add it there
+          " so that the output looks like this:
+          "
+          " <foobar>1</foobar>
+          if s:TagContent(lastitem) is# s:TagContent(t[1]) && strlen(result[-1]) + strlen(item) <= s:Textwidth()
+            let result[-1] .= item
+            let lastitem = t[1]
+            continue
+          endif
+          " t should only contain 2 items, but just be safe here
+          if s:IsTag(lastitem)
+            let s:indent+=1
+          endif
+          let result+=s:FormatContent([t[0]])
+          if s:EndTag(t[1])
+            call s:DecreaseIndent()
+          endif
+          "for y in t[1:]
+            let result+=s:FormatContent(t[1:])
+          "endfor
+        else
+          call add(result, s:Indent(item))
+        endif
       endif
-     endif
-     let lastitem = item
-   endfor
+      let lastitem = item
+    endfor
+    let current += 1
+  endfor
 
-   if !empty(result)
-    exe v:lnum. ",". (v:lnum + v:count - 1). 'd'
+  if !empty(result)
+    let lastprevline = getline(v:lnum + count_orig)
+    let delete_lastline = v:lnum + count_orig - 1 == line('$')
+    exe v:lnum. ",". (v:lnum + count_orig - 1). 'd'
     call append(v:lnum - 1, result)
     " Might need to remove the last line, if it became empty because of the
     " append() call
     let last = v:lnum + len(result)
-    if getline(last) is ''
+    " do not use empty(), it returns true for `empty(0)`
+    if getline(last) is '' && lastprevline is '' && delete_lastline
       exe last. 'd'
     endif
   endif
@@ -68,15 +109,15 @@ func! xmlformat#Format()
   return 0
 endfunc
 " Check if given tag is XML Declaration header {{{1
-func! s:IsXMLDecl(tag)
+func! s:IsXMLDecl(tag) abort
   return a:tag =~? '^\s*<?xml\s\?\%(version="[^"]*"\)\?\s\?\%(encoding="[^"]*"\)\? ?>\s*$'
 endfunc
 " Return tag indented by current level {{{1
-func! s:Indent(item)
+func! s:Indent(item) abort
   return repeat(' ', shiftwidth()*s:indent). s:Trim(a:item)
 endfu
 " Return item trimmed from leading whitespace {{{1
-func! s:Trim(item)
+func! s:Trim(item) abort
   if exists('*trim')
     return trim(a:item)
   else
@@ -84,30 +125,77 @@ func! s:Trim(item)
   endif
 endfunc
 " Check if tag is a new opening tag <tag> {{{1
-func! s:StartTag(tag)
+func! s:StartTag(tag) abort
   let is_comment = s:IsComment(a:tag)
   return a:tag =~? '^\s*<[^/?]' && !is_comment
 endfunc
-func! s:IsComment(tag)
+" Check if tag is a Comment start {{{1
+func! s:IsComment(tag) abort
   return a:tag =~? '<!--'
 endfunc
 " Remove one level of indentation {{{1
-func! s:DecreaseIndent()
-  return (s:indent > 0 ? s:indent - 1 : 0)
+func! s:DecreaseIndent() abort
+  let s:indent = (s:indent > 0 ? s:indent - 1 : 0)
 endfunc
 " Check if tag is a closing tag </tag> {{{1
-func! s:EndTag(tag)
+func! s:EndTag(tag) abort
   return a:tag =~? '^\s*</'
 endfunc
 " Check that the tag is actually a tag and not {{{1
 " something like "foobar</foobar>"
-func! s:IsTag(tag)
+func! s:IsTag(tag) abort
   return s:Trim(a:tag)[0] == '<'
 endfunc
 " Check if tag is empty <tag/> {{{1
-func! s:EmptyTag(tag)
+func! s:EmptyTag(tag) abort
   return a:tag =~ '/>\s*$'
 endfunc
+func! s:TagContent(tag) abort "{{{1
+  " Return content of a tag
+  return substitute(a:tag, '^\s*<[/]\?\([^>]*\)>\s*$', '\1', '')
+endfunc
+func! s:Textwidth() abort "{{{1
+  " return textwidth (or 80 if not set)
+  return &textwidth == 0 ? 80 : &textwidth
+endfunc
+" Format input line according to textwidth {{{1
+func! s:FormatContent(list) abort
+  let result=[]
+  let limit = s:Textwidth()
+  let column=0
+  let idx = -1
+  let add_indent = 0
+  let cnt = 0
+  for item in a:list
+    for word in split(item, '\s\+\S\+\zs')
+      if match(word, '^\s\+$') > -1
+        " skip empty words
+        continue
+      endif
+      let column += strdisplaywidth(word, column)
+      if match(word, "^\\s*\n\\+\\s*$") > -1
+        call add(result, '')
+        let idx += 1
+        let column = 0
+        let add_indent = 1
+      elseif column > limit || cnt == 0
+        let add = s:Indent(s:Trim(word))
+        call add(result, add)
+        let column = strdisplaywidth(add)
+        let idx += 1
+      else
+        if add_indent
+          let result[idx] = s:Indent(s:Trim(word))
+        else
+          let result[idx] .= ' '. s:Trim(word)
+        endif
+        let add_indent = 0
+      endif
+      let cnt += 1
+    endfor
+  endfor
+  return result
+endfunc
 " Restoration And Modelines: {{{1
 let &cpo= s:keepcpo
 unlet s:keepcpo