Fix #605 - Support `vcol: 1` for multi-byte character positions

4 files changed, 63 insertions, 2 deletions

diff --git a/autoload/ale/engine.vim b/autoload/ale/engine.vim
index dd871c36..0704fd5b 100644
--- a/autoload/ale/engine.vim
+++ b/autoload/ale/engine.vim
@@ -393,7 +393,7 @@ function! ale#engine#FixLocList(buffer, linter_name, loclist) abort
         \   'text': l:old_item.text,
         \   'lnum': str2nr(l:old_item.lnum),
         \   'col': str2nr(get(l:old_item, 'col', 0)),
-        \   'vcol': get(l:old_item, 'vcol', 0),
+        \   'vcol': 0,
         \   'type': get(l:old_item, 'type', 'E'),
         \   'nr': get(l:old_item, 'nr', -1),
         \   'linter_name': a:linter_name,
@@ -453,6 +453,20 @@ function! ale#engine#FixLocList(buffer, linter_name, loclist) abort
             " When errors go beyond the end of the file, put them at the end.
             " This is only done for the current buffer.
             let l:item.lnum = l:last_line_number
+        elseif get(l:old_item, 'vcol', 0)
+            " Convert virtual column positions to byte positions.
+            " The positions will be off if the buffer has changed recently.
+            let l:line = getbufline(a:buffer, l:item.lnum)[0]
+
+            let l:item.col = ale#util#Col(l:line, l:item.col)
+
+            if has_key(l:item, 'end_col')
+                let l:end_line = get(l:item, 'end_lnum', l:line) != l:line
+                \   ? getbufline(a:buffer, l:item.end_lnum)[0]
+                \   : l:line
+
+                let l:item.end_col = ale#util#Col(l:end_line, l:item.end_col)
+            endif
         endif
 
         call add(l:new_loclist, l:item)
diff --git a/autoload/ale/util.vim b/autoload/ale/util.vim
index b94a11b7..e0b31e2c 100644
--- a/autoload/ale/util.vim
+++ b/autoload/ale/util.vim
@@ -311,3 +311,13 @@ function! ale#util#StopPartialTimer(timer_id) abort
         call remove(s:partial_timers, a:timer_id)
     endif
 endfunction
+
+" Given a possibly multi-byte string and a 1-based character position on a
+" line, return the 1-based byte position on that line.
+function! ale#util#Col(str, chr) abort
+    if a:chr < 2
+        return a:chr
+    endif
+
+    return strlen(join(split(a:str, '\zs')[0:a:chr - 2], '')) + 1
+endfunction
diff --git a/doc/ale.txt b/doc/ale.txt
index 2e5c5e47..97f1b594 100644
--- a/doc/ale.txt
+++ b/doc/ale.txt
@@ -2072,6 +2072,13 @@ ale#linter#Define(filetype, linter)                       *ale#linter#Define()*
                            key is also set with a valid number for some other
                            buffer.
                          `vcol` - Defaults to `0`.
+
+                           If set to `1`, ALE will convert virtual column
+                           positions for `col` and `end_col` to byte column
+                           positions. If the buffer is changed in-between
+                           checking it and displaying the results, the
+                           calculated byte column positions will probably be
+                           wrong.
                          `type` - Defaults to `'E'`.
                          `nr` - Defaults to `-1`.
 
diff --git a/test/test_loclist_corrections.vader b/test/test_loclist_corrections.vader
index 46c7e272..48aa1f78 100644
--- a/test/test_loclist_corrections.vader
+++ b/test/test_loclist_corrections.vader
@@ -52,7 +52,7 @@ Execute(FixLocList should use the values we supply):
   \     'lnum': 3,
   \     'col': 4,
   \     'bufnr': 10000,
-  \     'vcol': 1,
+  \     'vcol': 0,
   \     'type': 'W',
   \     'nr': 42,
   \     'linter_name': 'foobar',
@@ -348,3 +348,33 @@ Execute(The error code should be passed on):
   \   'foobar',
   \   [{'text': 'a', 'lnum': 11, 'code': 'some-code'}],
   \ )
+
+Given(A file with Japanese multi-byte text):
+  はじめまして!
+  -私はワープです。
+Execute(character positions should be converted to byte positions):
+  AssertEqual
+  \ [
+  \   {'lnum': 1, 'bufnr': bufnr(''), 'col': 0, 'linter_name': 'foobar', 'nr': -1, 'type': 'E', 'vcol': 0, 'text': 'a'},
+  \   {'lnum': 1, 'bufnr': bufnr(''), 'col': 1, 'linter_name': 'foobar', 'nr': -1, 'type': 'E', 'vcol': 0, 'text': 'a'},
+  \   {'lnum': 1, 'bufnr': bufnr(''), 'col': 4, 'linter_name': 'foobar', 'nr': -1, 'type': 'E', 'vcol': 0, 'text': 'a'},
+  \   {'lnum': 1, 'bufnr': bufnr(''), 'col': 7, 'linter_name': 'foobar', 'nr': -1, 'type': 'E', 'vcol': 0, 'text': 'a'},
+  \   {'lnum': 1, 'bufnr': bufnr(''), 'col': 7, 'end_col': 13, 'linter_name': 'foobar', 'nr': -1, 'type': 'E', 'vcol': 0, 'text': 'a'},
+  \   {'lnum': 1, 'bufnr': bufnr(''), 'col': 7, 'end_col': 13, 'end_lnum': 1, 'linter_name': 'foobar', 'nr': -1, 'type': 'E', 'vcol': 0, 'text': 'a'},
+  \   {'lnum': 1, 'bufnr': bufnr(''), 'col': 7, 'end_col': 17, 'end_lnum': 2, 'linter_name': 'foobar', 'nr': -1, 'type': 'E', 'vcol': 0, 'text': 'a'},
+  \   {'lnum': 2, 'bufnr': bufnr(''), 'col': 17, 'linter_name': 'foobar', 'nr': -1, 'type': 'E', 'vcol': 0, 'text': 'a'},
+  \],
+  \ ale#engine#FixLocList(
+  \   bufnr('%'),
+  \   'foobar',
+  \   [
+  \     {'text': 'a', 'lnum': 1, 'col': 0, 'vcol': 1},
+  \     {'text': 'a', 'lnum': 1, 'col': 1, 'vcol': 1},
+  \     {'text': 'a', 'lnum': 1, 'col': 2, 'vcol': 1},
+  \     {'text': 'a', 'lnum': 1, 'col': 3, 'vcol': 1},
+  \     {'text': 'a', 'lnum': 1, 'col': 3, 'end_col': 5, 'vcol': 1},
+  \     {'text': 'a', 'lnum': 1, 'col': 3, 'end_col': 5, 'end_lnum': 1, 'vcol': 1},
+  \     {'text': 'a', 'lnum': 1, 'col': 3, 'end_col': 7, 'end_lnum': 2, 'vcol': 1},
+  \     {'text': 'a', 'lnum': 2, 'col': 7, 'vcol': 1},
+  \   ],
+  \ )