Link parsing in VSCode breaks on non-ascii characters in titles

[zkbpkp]

Example:

I figured this is because regex returns offsets in bytes, not runes. Below is my quick fix for that, seems to work in VSCode, didn't test in Neovim.

diff --git a/internal/adapter/lsp/document.go b/internal/adapter/lsp/document.go
index bf32d4f..954d93e 100644
--- a/internal/adapter/lsp/document.go
+++ b/internal/adapter/lsp/document.go
@@ -176,6 +176,15 @@ func (d *document) DocumentLinkAt(pos protocol.Position) (*documentLink, error)
 	return nil, nil
 }
 
+func toRuneMatch(s string, i int) int {
+	res := 0
+	for j, _ := range s {
+	  if j >= i { break }
+	  res += 1
+	}
+	return res
+  }
+
 // DocumentLinks returns all the internal and external links found in the
 // document.
 func (d *document) DocumentLinks() ([]documentLink, error) {
@@ -217,13 +226,13 @@ func (d *document) DocumentLinks() ([]documentLink, error) {
 			if decodedHref, err := url.PathUnescape(href); err == nil {
 				href = decodedHref
 			}
-			appendLink(href, match[0], match[1], false, false)
+			appendLink(href, toRuneMatch(line, match[0]), toRuneMatch(line, match[1]), false, false)
 		}
 
 		for _, match := range wikiLinkRegex.FindAllStringSubmatchIndex(line, -1) {
 			href := line[match[2]:match[3]]
 			hasTitle := match[4] != -1
-			appendLink(href, match[0], match[1], hasTitle, true)
+			appendLink(href, toRuneMatch(line, match[0]), toRuneMatch(line, match[1]), hasTitle, true)
 		}
 	}
 

[mickael-menu]

@zkbpkp Thank you for spotting this issue! I opened the PR #238 (branch issue-235), would you mind double-checking before I merge it? Thanks

[zkbpkp]

@mickael-menu thank you for responding so quickly! Just checked #238, this fixes the issue for me 👍