Updated PCRE to use utf8, and created sample plugin.

data/core/regex.lua

@@ -4,26 +4,58 @@
 regex.__index = function(table, key) return regex[key]; end
 
 regex.match = function(pattern_string, string)
-  local pattern = type(pattern_string) == "userdata" and
+  local pattern = type(pattern_string) == "table" and
     pattern_string or regex.compile(pattern_string)
   return regex.cmatch(pattern, string)
 end
 
--- Build off matching. For now, only support basic replacements, but capture
+-- Will iterate back through any UTF-8 bytes so that we don't replace bits mid character.
--- groupings should be doable. We can even have custom group replacements and
+local function previous_character(str, index)
--- transformations and stuff in lua.
+  local byte
-regex.gsub = function(pattern_string, string, replacement)
-  local pattern = type(pattern_string) == "userdata" and 
-    pattern_string or regex.compile(pattern_string)
-  local offset, result, str, indices = 0, "", string
   repeat
-    str = str:sub(offset)
+    index = index - 1
-    indices = { regex.cmatch(pattern, str) }
+    byte = string.byte(str, index)
-    if #indices > 0 then
+  until byte < 128 or byte >= 192
-      result = result .. str:sub(offset, indices[1] - 1) .. replacement
+  return index
-      offset = indices[2]
+end
-    end
+
-  until #indices == 0 or indices[1] == indices[2]
+-- Moves to the end of the identified character.
-  return result .. str:sub(offset - 1)
+local function end_character(str, index)
+  local byte = string.byte(str, index + 1)
+  while byte >= 128 and byte < 192 do
+    index = index + 1
+    byte = string.byte(str, index + 1)
+  end
+  return index
+end
+
+-- Build off matching. For now, only support basic replacements, but capture
+-- groupings should be doable. We can even have custom group replacements and
+-- transformations and stuff in lua. Currently, this takes group replacements 
+-- as \1 - \9.
+-- Should work on UTF-8 text.
+regex.gsub = function(pattern_string, string, replacement)
+  local pattern = type(pattern_string) == "table" and 
+    pattern_string or regex.compile(pattern_string)
+  local result, str, indices = "", string
+  repeat
+    indices = { regex.cmatch(pattern, str) }
+    if #indices > 0 then
+      local currentReplacement = replacement
+      if #indices > 2 then
+        for i = 1, (#indices/2 - 1) do
+          currentReplacement = string.gsub(currentReplacement, "\\" .. i, str:sub(indices[i*2+1], end_character(str,indices[i*2+2]-1)))
+        end
+      end
+      currentReplacement = string.gsub(currentReplacement, "\\%d", "")
+      if indices[1] > 1 then
+        result = result .. str:sub(1, previous_character(str, indices[1])) .. currentReplacement
+      else
+        result = result .. currentReplacement      
+      end
+      str = str:sub(indices[2])
+    end
+  until #indices == 0 or indices[1] == indices[2]
+  return result .. str
 end
 

data/plugins/regex_find_replace.lua

@@ -0,0 +1,53 @@
+-- lite-xl 1.16
+
+local core = require "core"
+local command = require "core.command"
+
+-- Takes the following pattern: /pattern/replace/
+-- Capture groupings can be replaced using \1 through \9
+local function regex_replace_file(pattern)
+  local doc = core.active_view.doc
+  
+  local start_pattern = 2;
+  local end_pattern = 2
+  repeat 
+    end_pattern = string.find(pattern, "/", end_pattern)
+  until end_pattern == nil or pattern[end_pattern-1] ~= "\\"
+  if end_pattern == nil then
+    core.log("Can't find end to pattern.")
+    return
+  end
+  end_pattern = end_pattern - 1
+  local start_replacement = end_pattern+2;
+  local end_replacement = end_pattern+2;
+  repeat
+    end_replacement = string.find(pattern, "/", end_replacement)
+  until end_replacement == nil or pattern[end_replacement-1] ~= "\\"
+  if end_replacement == nil then
+    core.log("Can't find end to replacement.")
+    return
+  end
+  end_replacement = end_replacement - 1
+  
+  local re = regex.compile(pattern:sub(start_pattern, end_pattern))
+  local replacement = pattern:sub(start_replacement, end_replacement)
+  for i=1,#doc.lines do
+      local old_length = #doc.lines[i]
+      local old_text = doc:get_text(i, 1, i, old_length)
+      local new_text = regex.gsub(re, old_text, replacement)
+      if old_text ~= new_text then
+        doc:insert(i, old_length, new_text)
+        doc:remove(i, 1, i, old_length)
+      end
+  end
+end
+
+local initial_regex_replace = "/"
+command.add("core.docview", {
+  ["regex:find-replace"] = function()
+    core.command_view:set_text(initial_regex_replace)
+    core.command_view:enter("Regex Replace (enter pattern as /old/new/)", function(pattern)
+    regex_replace_file(pattern)
+    initial_regex_replace = pattern
+  end) end
+})

src/api/regex.c

@@ -17,11 +17,21 @@ static int f_pcre_compile(lua_State *L) {
   size_t len;
   PCRE2_SIZE errorOffset;
   int errorNumber;
-  const char* str = luaL_checklstring(L, -1, &len);
+  int pattern = PCRE2_UTF;
+  const char* str = luaL_checklstring(L, 1, &len);
+  if (lua_gettop(L) > 1) {
+    const char* options = luaL_checkstring(L, 2);
+    if (strstr(options,"i"))
+      pattern |= PCRE2_CASELESS;
+    if (strstr(options,"m"))
+      pattern |= PCRE2_MULTILINE;
+    if (strstr(options,"s"))
+      pattern |= PCRE2_DOTALL;
+  }
   pcre2_code* re = pcre2_compile(
     (PCRE2_SPTR)str,
     len,
-    0,
+    PCRE2_UTF,
     &errorNumber,
     &errorOffset,
     NULL