path: root/runtime/tools/unicode.vim

" Script to extract tables from Unicode .txt files, to be used in src/mbyte.c.
" The format of the UnicodeData.txt file is explained here:
" http://www.unicode.org/Public/5.1.0/ucd/UCD.html
" For the other files see the header.
"
" Might need to update the URL to the emoji-data.txt
" Usage: Vim -S <this-file>
"
" Author: Bram Moolenaar
" Last Update: 2020 Aug 24

" Parse lines of UnicodeData.txt.  Creates a list of lists in s:dataprops.
func! ParseDataToProps()
  let s:dataprops = []
  let lnum = 1
  while lnum <= line('$')
    let l = split(getline(lnum), '\s*;\s*', 1)
    if len(l) != 15
      echoerr 'Found ' . len(l) . ' items in line ' . lnum . ', expected 15'
      return
    endif
    call add(s:dataprops, l)
    let lnum += 1
  endwhile
endfunc

" Parse lines of CaseFolding.txt.  Creates a list of lists in s:foldprops.
func! ParseFoldProps()
  let s:foldprops = []
  let lnum = 1
  while lnum <= line('$')
    let line = getline(lnum)
    if line !~ '^#' && line !~ '^\s*$'
      let l = split(line, '\s*;\s*', 1)
      if len(l) != 4
        echoerr 'Found ' . len(l) . ' items in line ' . lnum . ', expected 4'
        return
      endif
      call add(s:foldprops, l)
    endif
    let lnum += 1
  endwhile
endfunc

" Parse lines of EastAsianWidth.txt.  Creates a list of lists in s:widthprops.
func! ParseWidthProps()
  let s:widthprops = []
  let lnum = 1
  while lnum <= line('$')
    let line = getline(lnum)
    if line !~ '^#' && line !~ '^\s*$'
      let l = split(line, '\s*;\s*', 1)
      if len(l) != 2
        echoerr 'Found ' . len(l) . ' items in line ' . lnum . ', expected 2'
        return
      endif
      call add(s:widthprops, l)
    endif
    let lnum += 1
  endwhile
endfunc

" Build the toLower or toUpper table in a new buffer.
" Uses s:dataprops.
func! BuildCaseTable(name, index)
  let start = -1
  let end = -1
  let step = 0
  let add = -1
  let ranges = []
  for p in s:dataprops
    if p[a:index] != ''
      let n = ('0x' . p[0]) + 0
      let nl = ('0x' . p[a:index]) + 0
      if start >= 0 && add == nl - n && (step == 0 || n - end == step)
        " continue with same range.
        let step = n - end
        let end = n
      else
        if start >= 0
          " produce previous range
          call Range(ranges, start, end, step, add)
        endif
        let start = n
        let end = n
        let step = 0
        let add = nl - n
      endif
    endif
  endfor
  if start >= 0
    call Range(ranges, start, end, step, add)
  endif

  " New buffer to put the result in.
  new
  exe "file to" . a:name
  call setline(1, "static convertStruct to" . a:name . "[] =")
  call setline(2, "{")
  call append('$', ranges)
  call setline('$', getline('$')[:-2])  " remove last comma
  call setline(line('$') + 1, "};")
  wincmd p
endfunc

" Build the foldCase table in a new buffer.
" Uses s:foldprops.
func! BuildFoldTable()
  let start = -1
  let end = -1
  let step = 0
  let add = -1
  let ranges = []
  for p in s:foldprops
    if p[1] == 'C' || p[1] == 'S'
      let n = ('0x' . p[0]) + 0
      let nl = ('0x' . p[2]) + 0
      if start >= 0 && add == nl - n && (step == 0 || n - end == step)
        " continue with same range.
        let step = n - end
        let end = n
      else
        if start >= 0
          " produce previous range
          call Range(ranges, start, end, step, add)
        endif
        let start = n
        let end = n
        let step = 0
        let add = nl - n
      endif
    endif
  endfor
  if start >= 0
    call Range(ranges, start, end, step, add)
  endif

  " New buffer to put the result in.
  new
  file foldCase
  call setline(1, "static convertStruct foldCase[] =")
  call setline(2, "{")
  call append('$', ranges)
  call setline('$', getline('$')[:-2])  " remove last comma
  call setline(line('$') + 1, "};")
  wincmd p
endfunc

func! Range(ranges, start, end, step, add)
  let s = printf("\t{0x%x,0x%x,%d,%d},", a:start, a:end, a:step == 0 ? -1 : a:step, a:add)
  call add(a:ranges, s)
endfunc

" Build the combining table.
" Uses s:dataprops.
func! BuildCombiningTable()
  let start = -1
  let end = -1
  let ranges = []
  for p in s:dataprops
    " The 'Mc' property was removed, it does take up space.
    if p[2] == 'Mn' || p[2] == 'Me'
      let n = ('0x' . p[0]) + 0
      if start >= 0 && end + 1 == n
        " continue with same range.
        let end = n
      else
        if start >= 0
          " produce previous range
          call add(ranges, printf("\t{0x%04x, 0x%04x},", start, end))
        endif
        let start = n
        let end = n
      endif
    endif
  endfor
  if start >= 0
    call add(ranges, printf("\t{0x%04x, 0x%04x},", start, end))
  endif

  " New buffer to put the result in.
  new
  file combining
  call setline(1, "    static struct interval combining[] =")
  call setline(2, "    {")
  call append('$', ranges)
  call setline('$', getline('$')[:-2])  " remove last comma
  call setline(line('$') + 1, "    };")
  wincmd p
endfunc

" Build the double width or ambiguous width table in a new buffer.
" Uses s:widthprops and s:dataprops.
func! BuildWidthTable(pattern, tableName)
  let start = -1
  let end = -1
  let ranges = []
  let dataidx = 0
  " Account for indentation differences between ambiguous and doublewidth
  " table in mbyte.c
  if a:pattern == 'A'
    let spc = '    '
  else
    let spc = "\t"
  endif
  for p in s:widthprops
    if p[1][0] =~ a:pattern
      if p[0] =~ '\.\.'
        " It is a range.  we don't check for composing char then.
        let rng = split(p[0], '\.\.')
        if len(rng) != 2
          echoerr "Cannot parse range: '" . p[0] . "' in width table"
        endif
        let n = ('0x' . rng[0]) + 0
        let n_last =  ('0x' . rng[1]) + 0
      else
        let n = ('0x' . p[0]) + 0
        let n_last = n
      endif
      " Find this char in the data table.
      while 1
        let dn = ('0x' . s:dataprops[dataidx][0]) + 0
        if dn >= n
          break
        endif
        let dataidx += 1
      endwhile
      if dn != n && n_last == n
        echoerr "Cannot find character " . n . " in data table"
      endif
      " Only use the char when it's not a composing char.
      " But use all chars from a range.
      let dp = s:dataprops[dataidx]
      if n_last > n || (dp[2] != 'Mn' && dp[2] != 'Mc' && dp[2] != 'Me')
        if start >= 0 && end + 1 == n
          " continue with same range.
        else
          if start >= 0
            " produce previous range
            call add(ranges, printf("%s{0x%04x, 0x%04x},", spc, start, end))
	    if a:pattern == 'A'
	      call add(s:ambitable, [start, end])
	    else
	      call add(s:doubletable, [start, end])
	    endif
          endif
          let start = n
        endif
        let end = n_last
      endif
    endif
  endfor
  if start >= 0
    call add(ranges, printf("%s{0x%04x, 0x%04x},", spc, start, end))
    if a:pattern == 'A'
      call add(s:ambitable, [start, end])
    else
      call add(s:doubletable, [start, end])
    endif
  endif

  " New buffer to put the result in.
  new
  exe "file " . a:tableName
  if a:pattern == 'A'
    call setline(1, "static struct interval " . a:tableName . "[] =")
    call setline(2, "{")
  else
    call setline(1, "    static struct interval " . a:tableName . "[] =")
    call setline(2, "    {")
  endif
  call append('$', ranges)
  call setline('$', getline('$')[:-2])  " remove last comma
  if a:pattern == 'A'
    call setline(line('$') + 1, "};")
  else
    call setline(line('$') + 1, "    };")
  endif
  wincmd p
endfunc


" Get characters from a list of lines in form "12ab .." or "12ab..56cd ..."
" and put them in dictionary "chardict"
func AddLinesToCharDict(lines, chardict)
  for line in a:lines
    let tokens = split(line, '\.\.')
    let first = str2nr(tokens[0], 16)
    if len(tokens) == 1
      let last = first
    else
      let last = str2nr(tokens[1], 16)
    endif
    for nr in range(first, last)
      let a:chardict[nr] = 1
    endfor
  endfor
endfunc

func Test_AddLinesToCharDict()
  let dict = {}
  call AddLinesToCharDict([
	\ '1234 blah blah',
	\ '1235 blah blah',
	\ '12a0..12a2 blah blah',
	\ '12a1 blah blah',
	\ ], dict)
  call assert_equal({0x1234: 1, 0x1235: 1,
	\ 0x12a0: 1, 0x12a1: 1, 0x12a2: 1,
	\ }, dict)
  if v:errors != []
    echoerr 'AddLinesToCharDict' v:errors
    return 1
  endif
  return 0
endfunc


func CharDictToPairList(chardict)
  let result = []
  let keys = keys(a:chardict)->map('str2nr(v:val)')->sort('N')
  let low = keys[0]
  let high = keys[0]
  for key in keys
    if key > high + 1
      call add(result, [low, high])
      let low = key
      let high = key
    else
      let high = key
    endif
  endfor
  call add(result, [low, high])
  return result
endfunc

func Test_CharDictToPairList()
  let dict = {0x1020: 1, 0x1021: 1, 0x1022: 1,
	\ 0x1024: 1,
	\ 0x2022: 1,
	\ 0x2024: 1, 0x2025: 1}
  call assert_equal([
	\ [0x1020, 0x1022],
	\ [0x1024, 0x1024],
	\ [0x2022, 0x2022],
	\ [0x2024, 0x2025],
	\ ], CharDictToPairList(dict))
  if v:errors != []
    echoerr 'CharDictToPairList' v:errors
    return 1
  endif
  return 0
endfunc


" Build the amoji width table in a new buffer.
func BuildEmojiTable()
  " First make the table for all emojis.
  let pattern = '; Emoji\s\+#\s'
  let lines = map(filter(filter(getline(1, '$'), 'v:val=~"^[1-9]"'), 'v:val=~pattern'), 'matchstr(v:val,"^\\S\\+")')

  " Make a dictionary with an entry for each character.
  let chardict = {}
  call AddLinesToCharDict(lines, chardict)
  let pairlist = CharDictToPairList(chardict)
  let allranges = map(pairlist, 'printf("    {0x%04x, 0x%04x},", v:val[0], v:val[1])')

  " New buffer to put the result in.
  new
  exe 'file emoji_all'
  call setline(1, "static struct interval emoji_all[] =")
  call setline(2, "{")
  call append('$', allranges)
  call setline('$', getline('$')[:-2])  " remove last comma
  call setline(line('$') + 1, "};")
  wincmd p

  " Make the table for wide emojis.
  let pattern = '; Emoji_\(Presentation\|Modifier_Base\)\s\+#\s'
  let lines = map(filter(filter(getline(1, '$'), 'v:val=~"^[1-9]"'), 'v:val=~pattern'), 'matchstr(v:val,"^\\S\\+")')

  " Make a dictionary with an entry for each character.
  let chardict = {}
  call AddLinesToCharDict(lines, chardict)

  " exclude characters that are in the "ambiguous" or "doublewidth" table
  for ambi in s:ambitable
    for nr in range(ambi[0], ambi[1])
      if has_key(chardict, nr)
	call remove(chardict, nr)
      endif
    endfor
  endfor

  for wide in s:doubletable
    for nr in range(wide[0], wide[1])
      if has_key(chardict, nr)
	call remove(chardict, nr)
      endif
    endfor
  endfor

  let pairlist = CharDictToPairList(chardict)
  let wide_ranges = map(pairlist, 'printf("\t{0x%04x, 0x%04x},", v:val[0], v:val[1])')

  " New buffer to put the result in.
  new
  exe 'file emoji_wide'
  call setline(1, "    static struct interval emoji_wide[] =")
  call setline(2, "    {")
  call append('$', wide_ranges)
  call setline('$', getline('$')[:-2])  " remove last comma
  call setline(line('$') + 1, "    };")
  wincmd p
endfunc

" First test a few things
let v:errors = []
if Test_AddLinesToCharDict() || Test_CharDictToPairList()
  finish
endif


" Try to avoid hitting E36
set equalalways

" Edit the Unicode text file.  Requires the netrw plugin.
edit http://unicode.org/Public/UNIDATA/UnicodeData.txt

" Parse each line, create a list of lists.
call ParseDataToProps()

" Build the toLower table.
call BuildCaseTable("Lower", 13)

" Build the toUpper table.
call BuildCaseTable("Upper", 12)

" Build the ranges of composing chars.
call BuildCombiningTable()

" Edit the case folding text file.  Requires the netrw plugin.
edit http://www.unicode.org/Public/UNIDATA/CaseFolding.txt

" Parse each line, create a list of lists.
call ParseFoldProps()

" Build the foldCase table.
call BuildFoldTable()

" Edit the width text file.  Requires the netrw plugin.
edit http://www.unicode.org/Public/UNIDATA/EastAsianWidth.txt

" Parse each line, create a list of lists.
call ParseWidthProps()

" Build the double width table.
let s:doubletable = []
call BuildWidthTable('[WF]', 'doublewidth')

" Build the ambiguous width table.
let s:ambitable = []
call BuildWidthTable('A', 'ambiguous')

" Edit the emoji text file.  Requires the netrw plugin.
" commented out, because it drops too many characters
"edit https://unicode.org/Public/15.0.0/ucd/emoji/emoji-data.txt
"
"" Build the emoji table. Ver. 1.0 - 6.0
"" Must come after the "ambiguous" and "doublewidth" tables
"call BuildEmojiTable()