improve string.split

3 files changed, 37 insertions, 14 deletions

diff --git a/xmake/core/base/colors.lua b/xmake/core/base/colors.lua
index 8c5b2c287..2223c3f85 100644
--- a/xmake/core/base/colors.lua
+++ b/xmake/core/base/colors.lua
@@ -324,7 +324,7 @@ function colors.translate(str)
         end
 
         -- split words
-        local blocks_raw = word:split("%s+")
+        local blocks_raw = word:split("%s")
 
         -- translate theme color first, e.g ${color.error}
         local blocks = {}
@@ -332,7 +332,7 @@ function colors.translate(str)
             if theme then
                 local theme_block = theme:get(block)
                 if theme_block then
-                    for _, theme_block_sub in ipairs(theme_block:split("%s+")) do
+                    for _, theme_block_sub in ipairs(theme_block:split("%s")) do
                         table.insert(blocks, theme_block_sub)
                     end
                 else 
diff --git a/xmake/core/base/interpreter.lua b/xmake/core/base/interpreter.lua
index 029fe3e60..01d76973e 100644
--- a/xmake/core/base/interpreter.lua
+++ b/xmake/core/base/interpreter.lua
@@ -119,7 +119,7 @@ function interpreter._fetch_root_scope(root)
     for scope_kind_and_name, _ in pairs(root or {}) do
         
         -- is scope_kind@@scope_name?
-        scope_kind_and_name = scope_kind_and_name:split("@@")
+        scope_kind_and_name = scope_kind_and_name:split("@@", {plain = true})
         if #scope_kind_and_name == 2 then
             local scope_kind = scope_kind_and_name[1] 
             local scope_name = scope_kind_and_name[2]
@@ -1413,7 +1413,7 @@ function interpreter:api_define(apis)
             -- get api function 
             local apiscope = nil
             local funcname = nil
-            apifunc = apifunc:split('.')
+            apifunc = apifunc:split('.', {plain = true})
             assert(apifunc)
             if #apifunc == 2 then
                 apiscope = apifunc[1]
@@ -1470,7 +1470,7 @@ function interpreter:api_builtin_set_xmakever(minver)
     end
 
     -- parse minimum version
-    local minvers = minver:split('.')
+    local minvers = minver:split('.', {plain = true})
     if not minvers or #minvers ~= 3 then
         os.raise("[nobacktrace]: set_xmakever(\"%s\"): invalid version format!", minver)
     end
@@ -1479,7 +1479,7 @@ function interpreter:api_builtin_set_xmakever(minver)
     local minvers_num = minvers[1] * 100 + minvers[2] * 10 + minvers[3]
 
     -- parse current version
-    local curvers = xmake._VERSION_SHORT:split('.')
+    local curvers = xmake._VERSION_SHORT:split('.', {plain = true})
 
     -- make current numerical version
     local curvers_num = curvers[1] * 100 + curvers[2] * 10 + curvers[3]
diff --git a/xmake/core/base/string.lua b/xmake/core/base/string.lua
index b27d076d5..878a2f683 100644
--- a/xmake/core/base/string.lua
+++ b/xmake/core/base/string.lua
@@ -143,19 +143,42 @@ function string:find_last(pattern, plain)
     end
 end
 
--- split string with the given characters
+-- split string with the given substring/characters
 --
+-- pattern match and ignore empty string
 -- ("1\n\n2\n3"):split('\n') => 1, 2, 3
--- ("1\n\n2\n3"):split('\n', true) => 1, , 2, 3
+-- ("abc123123xyz123abc"):split('123') => abc, xyz, abc
+-- ("abc123123xyz123abc"):split('[123]+') => abc, xyz, abc
 --
-function string:split(delimiter, strict)
+-- plain match and ignore empty string
+-- ("1\n\n2\n3"):split('\n', {plain = true}) => 1, 2, 3
+-- ("abc123123xyz123abc"):split('123', {plain = true}) => abc, xyz, abc
+--
+-- pattern match and contains empty string
+-- ("1\n\n2\n3"):split('\n', {strict = true}) => 1, , 2, 3
+-- ("abc123123xyz123abc"):split('123', {strict = true}) => abc, , xyz, abc
+-- ("abc123123xyz123abc"):split('[123]+', {strict = true}) => abc, xyz, abc
+--
+-- plain match and contains empty string
+-- ("1\n\n2\n3"):split('\n', {plain = true, strict = true}) => 1, , 2, 3
+-- ("abc123123xyz123abc"):split('123', {plain = true, strict = true}) => abc, , xyz, abc
+--
+function string:split(delimiter, opt)
     local result = {}
-    if strict then
-        for match in (self .. delimiter):gmatch("(.-)" .. delimiter) do
-            table.insert(result, match)
+    local start = 1
+    local pos, epos = self:find(delimiter, start, opt and opt.plain) 
+    while pos do
+        local substr = self:sub(start, pos - 1)
+        if #substr > 0 then
+            table.insert(result, substr)
+        elseif opt and opt.strict then
+            table.insert(result, substr)
         end
-    else
-        self:gsub("[^" .. delimiter .."]+", function(v) table.insert(result, v) end)
+        start = epos + 1
+        pos, epos = self:find(delimiter, start, opt and opt.plain) 
+    end
+    if start <= #self then
+        table.insert(result, self:sub(start))
     end
     return result
 end