improve xml module

3 files changed, 110 insertions, 15 deletions

diff --git a/tests/modules/xml/test.lua b/tests/modules/xml/test.lua
index eb01aad86..80ce486f6 100644
--- a/tests/modules/xml/test.lua
+++ b/tests/modules/xml/test.lua
@@ -7,13 +7,18 @@ function test_parse_basic(t)
     t:are_equal(#doc.children, 2)
     t:are_equal(doc.children[1].name, "item")
     t:are_equal(xml.text_of(doc.children[1]), "foo")
+    t:are_equal(doc.children[1].attrs, nil)
     t:are_equal(doc.children[2].attrs.id, "2")
 end
 
 function test_encode(t)
-    local doc = xml.new("root", {id = "1"}, {
-        xml.new("item", {}, {xml.text("foo")}),
-        xml.new("item", {id = "2"})
+    local doc = xml.new({
+        name = "root",
+        attrs = {id = "1"},
+        children = {
+            xml.new({name = "item", children = {xml.text("foo")}}),
+            xml.new({name = "item", attrs = {id = "2"}})
+        }
     })
     local compact = xml.encode(doc)
     t:are_equal(compact, '<root id="1"><item>foo</item><item id="2"/></root>')
@@ -27,3 +32,22 @@ function test_encode(t)
     t:are_equal(pretty, expected)
 end
 
+function test_special_nodes(t)
+    t:are_equal(xml.encode(xml.comment("note")), "<!--note-->")
+    t:are_equal(xml.encode(xml.cdata("a < b")), "<![CDATA[a < b]]>")
+    t:are_equal(xml.encode(xml.doctype("note SYSTEM \"note.dtd\"")), "<!DOCTYPE note SYSTEM \"note.dtd\">")
+    t:are_equal(xml.encode(xml.empty("br")), "<br/>")
+end
+
+function test_parse_special_nodes(t)
+    local doc = xml.decode([[<root><!--note--><![CDATA[a < b]]><child/></root>]])
+    t:are_equal(doc.children[1].name, "__comment__")
+    t:are_equal(doc.children[1].children[1], "note")
+    t:are_equal(doc.children[2].name, "__cdata__")
+    t:are_equal(doc.children[2].children[1], "a < b")
+    t:are_equal(doc.children[3].name, "child")
+    local nodes = xml.decode([[<!DOCTYPE note SYSTEM "note.dtd"><root/>]])
+    t:are_equal(nodes[1].name, "__doctype__")
+    t:are_equal(nodes[2].name, "root")
+end
+
diff --git a/xmake/core/base/xml.lua b/xmake/core/base/xml.lua
index 7ac12f133..a3f2c96f0 100644
--- a/xmake/core/base/xml.lua
+++ b/xmake/core/base/xml.lua
@@ -45,25 +45,55 @@ function xml._encode_attr(str)
 end
 
 function xml._parse_attrs(attrstr)
-    local attrs = {}
+    local attrs
     attrstr:gsub("([%w_:%-%.]+)%s*=%s*([\"'])(.-)%2", function(key, quote, value)
+        attrs = attrs or {}
         attrs[key] = xml._decode_entities(value)
     end)
     return attrs
 end
 
 -- create an xml element node
-function xml.new(name, attrs, children)
+-- e.g. `local node = xml.new({name = "item", attrs = {id = "1"}, children = {xml.text("value")}})`
+function xml.new(opt)
+    opt = opt or {}
     return {
-        name = name,
-        attrs = attrs or {},
-        children = children or {}
+        name = opt.name,
+        attrs = opt.attrs,
+        kind = opt.kind or "element",
+        text = opt.text,
+        children = opt.children or {}
     }
 end
 
 -- create a text node
+-- e.g. `local textnode = xml.text("hello")`
 function xml.text(value)
-    return {type = "text", text = value or ""}
+    return xml.new({kind = "text", text = value or ""})
+end
+
+-- create an empty element node
+-- e.g. `local br = xml.empty("br", {class = "line"})`
+function xml.empty(name, attrs)
+    return xml.new({name = name, attrs = attrs})
+end
+
+-- create a comment node
+-- e.g. `local comment = xml.comment("generated by xmake")`
+function xml.comment(value)
+    return xml.new({kind = "comment", text = value or ""})
+end
+
+-- create a CDATA node
+-- e.g. `local cdata = xml.cdata("if (a < b) { ... }")`
+function xml.cdata(value)
+    return xml.new({kind = "cdata", text = value or ""})
+end
+
+-- create a doctype node
+-- e.g. `local doc = xml.doctype('html')`
+function xml.doctype(value)
+    return xml.new({kind = "doctype", text = value or ""})
 end
 
 function xml._append_text(stack, text, opt)
@@ -74,7 +104,7 @@ function xml._append_text(stack, text, opt)
     if text ~= "" then
         local top = stack[#stack]
         top.children = top.children or {}
-        table.insert(top.children, {type = "text", text = xml._decode_entities(text)})
+        table.insert(top.children, xml.text(xml._decode_entities(text)))
     end
 end
 
@@ -88,6 +118,7 @@ function xml._handle_closing(stack, tagname)
 end
 
 -- decode xml string to tree node(s)
+-- e.g. `local doc, err = xml.decode("<root><item>foo</item></root>")`
 function xml.decode(data, opt)
     opt = opt or {}
     local root = {name = "__root__", attrs = {}, children = {}}
@@ -110,7 +141,31 @@ function xml.decode(data, opt)
             if not close then
                 return nil, "unterminated xml comment"
             end
+            local value = data:sub(lt + 4, close - 1)
+            local top = stack[#stack]
+            top.children = top.children or {}
+            table.insert(top.children, xml.comment(value))
+            i = close + 3
+        elseif data:sub(lt + 1, lt + 8) == "![CDATA[" then
+            local close = data:find("]]>", lt + 9, true)
+            if not close then
+                return nil, "unterminated cdata section"
+            end
+            local value = data:sub(lt + 9, close - 1)
+            local top = stack[#stack]
+            top.children = top.children or {}
+            table.insert(top.children, xml.cdata(value))
             i = close + 3
+        elseif data:sub(lt + 1, lt + 9):upper() == "!DOCTYPE" then
+            local close = data:find(">", lt + 9)
+            if not close then
+                return nil, "unterminated doctype declaration"
+            end
+            local value = data:sub(lt + 10, close - 1)
+            local top = stack[#stack]
+            top.children = top.children or {}
+            table.insert(top.children, xml.doctype(value))
+            i = close + 1
         elseif data:sub(lt + 1, lt + 1) == "?" then
             local close = data:find("?>", lt + 2, true)
             if not close then
@@ -146,7 +201,7 @@ function xml.decode(data, opt)
             end
             local tagname, attrstr = inside:match("^%s*([^%s>]+)%s*(.-)%s*$")
             local attrs = xml._parse_attrs(attrstr or "")
-            local node = {name = tagname, attrs = attrs, children = {}}
+            local node = xml.empty(tagname, attrs)
             local top = stack[#stack]
             top.children = top.children or {}
             table.insert(top.children, node)
@@ -180,13 +235,19 @@ end
 function xml._encode_node(node, opt, level)
     opt = opt or {}
     level = level or 0
-    if node.type == "text" then
+    if node.kind == "text" then
         local indent = xml._indent(opt, level)
         local text = xml._encode_text(tostring(node.text or ""))
         if opt.pretty then
             return indent .. text
         end
         return text
+    elseif node.kind == "comment" then
+        return xml._indent(opt, level) .. string.format("<!--%s-->", tostring(node.text or ""))
+    elseif node.kind == "cdata" then
+        return xml._indent(opt, level) .. string.format("<![CDATA[%s]]>", tostring(node.text or ""))
+    elseif node.kind == "doctype" then
+        return xml._indent(opt, level) .. string.format("<!DOCTYPE %s>", tostring(node.text or ""))
     end
     local attrs = {}
     for k, v in pairs(node.attrs or {}) do
@@ -201,7 +262,7 @@ function xml._encode_node(node, opt, level)
         return xml._indent(opt, level) .. open .. "/>"
     end
     local newline = opt.pretty and "\n" or ""
-    if #node.children == 1 and node.children[1].type == "text" then
+    if #node.children == 1 and node.children[1].kind == "text" then
         local text = xml._encode_text(tostring(node.children[1].text or ""))
         return string.format("%s%s>%s</%s>", xml._indent(opt, level), open, text, node.name)
     end
@@ -215,12 +276,14 @@ function xml._encode_node(node, opt, level)
 end
 
 -- encode xml node to string
+-- e.g. `local xmlstr = xml.encode(node, {pretty = true, indent = 2})`
 function xml.encode(node, opt)
     opt = opt or {}
     return xml._encode_node(node, opt, 0)
 end
 
 -- load xml file
+-- e.g. `local doc, err = xml.load("foo.xml")`
 function xml.load(filepath, opt)
     local data, err = io.readfile(filepath, opt)
     if not data then
@@ -230,6 +293,7 @@ function xml.load(filepath, opt)
 end
 
 -- save xml node to file
+-- e.g. `assert(xml.save("foo.xml", node, {pretty = true}))`
 function xml.save(filepath, node, opt)
     local data = xml.encode(node, opt)
     if not data then
@@ -239,6 +303,8 @@ function xml.save(filepath, node, opt)
 end
 
 -- find the first child node with the given name
+-- e.g. `local doc = xml.decode("<root><item id='1'/></root>")`
+--      `local item = xml.find(doc, "item")`
 function xml.find(node, name)
     if not node or not node.children then
         return nil
@@ -251,14 +317,15 @@ function xml.find(node, name)
 end
 
 -- get concatenated text from child nodes
+-- e.g. `local text = xml.text_of(xml.decode("<item>foo</item>"))`
 function xml.text_of(node)
     if not node or not node.children then
         return ""
     end
     local buffer = {}
     for _, child in ipairs(node.children) do
-        if child.type == "text" then
-            table.insert(buffer, child.text)
+        if child.kind == "text" then
+            table.insert(buffer, child.text or "")
         end
     end
     return table.concat(buffer, "")
diff --git a/xmake/core/sandbox/modules/import/core/base/xml.lua b/xmake/core/sandbox/modules/import/core/base/xml.lua
index 7867c3295..6af763e47 100644
--- a/xmake/core/sandbox/modules/import/core/base/xml.lua
+++ b/xmake/core/sandbox/modules/import/core/base/xml.lua
@@ -31,6 +31,10 @@ sandbox_core_base_xml.find     = xml.find
 sandbox_core_base_xml.text_of  = xml.text_of
 sandbox_core_base_xml.text     = xml.text
 sandbox_core_base_xml.new      = xml.new
+sandbox_core_base_xml.empty    = xml.empty
+sandbox_core_base_xml.comment  = xml.comment
+sandbox_core_base_xml.cdata    = xml.cdata
+sandbox_core_base_xml.doctype  = xml.doctype
 
 -- decode xml data
 function sandbox_core_base_xml.decode(data, opt)