Support UTF-8 Lua pattern matching.

Makes use of an external luautf8 library, but only a subset of it.

2 files changed, 46 insertions, 3 deletions

diff --git a/modules/lua/lua.luadoc b/modules/lua/lua.luadoc
index f92adc48..8902482d 100644
--- a/modules/lua/lua.luadoc
+++ b/modules/lua/lua.luadoc
@@ -2046,3 +2046,39 @@ function lfs.touch(filepath [, atime [, mtime]]) end
 -- Returns true if the operation was successful; in case of error, it returns
 -- nil plus an error string.
 function lfs.unlock(filehandle[, start[, length]]) end
+
+---
+-- UTF-8 version of `string.byte`.
+function utf8.byte(s [, i [, j]]) end
+
+---
+-- UTF-8 version of `string.find`.
+function utf8.find(s, pattern [, init [, plain]]) end
+
+---
+-- UTF-8 version of `string.gmatch`.
+function utf8.gmatch(s, pattern) end
+
+---
+-- UTF-8 version of `string.gsub`.
+function utf8.gsub(s, pattern, repl [, n]) end
+
+---
+-- UTF-8 version of `string.lower`.
+function utf8.lower(s) end
+
+---
+-- UTF-8 version of `string.match`.
+function utf8.match(s, pattern [, init]) end
+
+---
+-- UTF-8 version of `string.reverse`.
+function utf8.reverse(s) end
+
+---
+-- UTF-8 version of `string.sub`.
+function utf8.sub(s, i [, j]) end
+
+---
+-- UTF-8 version of `string.upper`.
+function utf8.upper(s) end
diff --git a/modules/textadept/find.lua b/modules/textadept/find.lua
index ea6a287e..a9769cb8 100644
--- a/modules/textadept/find.lua
+++ b/modules/textadept/find.lua
@@ -154,14 +154,21 @@ local function find_(text, next, flags, no_wrap, wrapped)
     M.captures = nil -- clear captures from any previous Lua pattern searches
   elseif flags < 16 then
     -- Lua pattern search.
+    -- Note: I do not trust utf8.find completely, so only use it if there are
+    -- UTF-8 characters in patt. Otherwise default to string.find.
     local patt = text:gsub('\\[abfnrtv\\]', escapes)
     local s = next and buffer.current_pos or 0
     local e = next and buffer.length or buffer.current_pos
-    local caps = {buffer:text_range(s, e):find(next and patt or '^.*()'..patt)}
+    local find = not patt:find('[\xC2-\xF4]') and string.find or utf8.find
+    local caps = {find(buffer:text_range(s, e), next and patt or '^.*()'..patt)}
     M.captures = {table.unpack(caps, next and 3 or 4)}
     if #caps > 0 and caps[2] >= caps[1] then
-      pos = buffer:position_relative(s, caps[next and 1 or 3] - 1)
-      e = buffer:position_relative(s, caps[2])
+      if find == string.find then
+        pos, e = s + caps[next and 1 or 3] - 1, s + caps[2]
+      else
+        pos = buffer:position_relative(s, caps[next and 1 or 3] - 1)
+        e = buffer:position_relative(s, caps[2])
+      end
       M.captures[0] = buffer:text_range(pos, e)
       buffer:set_sel(e, pos)
     end