wip: select elements from html document and css selector

css.lua

@@ -0,0 +1,176 @@
+local M = {}
+
+local function trim(str)
+	return str:match("^%s*(.-)%s*$")
+end
+
+
+local COMBINATORS = {
+	DESCENDANT = {},
+	DIRECT_DESCENDANT = {},
+	NEXT_SIBLING = {},
+	SUBSEQUENT_SIBLING = {},
+}
+M.COMBINATORS = COMBINATORS
+
+local COMBINATOR_CHARS = {
+	[">"] = COMBINATORS.DIRECT_DESCENDANT,
+	["+"] = COMBINATORS.NEXT_SIBLING,
+	["~"] = COMBINATORS.SUBSEQUENT_SIBLING
+}
+
+
+
+
+
+
+local function create_tokeniser(input)
+	local pos = 1
+	local len = #input
+
+	local function peek()
+		if pos > len then return nil end
+		return input:sub(pos, pos)
+	end
+
+	local function next()
+		local char = peek()
+		if char then pos = pos + 1 end
+		return char
+	end
+
+	local function read_identifier()
+		local result = ""
+		while pos <= len do
+			local char = peek()
+			if char and char:match("[%w-]") then
+				result = result .. next()
+			else
+				break
+			end
+		end
+		return result
+	end
+
+	return {
+		peek = peek,
+		next = next,
+		read_identifier = read_identifier,
+		pos = function() return pos end
+	}
+end
+
+
+local function parse_compound_selector( tokeniser )
+	local selector = {
+		tag_name = nil,
+		id = nil,
+		class = {},
+		attributes = {},
+	}
+
+	--local selectors = {}
+
+	-- Parse first part (type or universal)
+	local char = tokeniser.peek()
+	if char == "*" then
+		tokeniser.next()
+		--table.insert(selectors, {type = "universal"})
+		selector.tag_name = "*"
+	elseif char and char:match("[%w-]") then
+		local name = tokeniser.read_identifier()
+		if name ~= "" then
+			--table.insert(selectors, {type = "type", value = name})
+			selector.tag_name = name
+		end
+	end
+
+	-- Parse additional class or ID selectors
+	while true do
+		char = tokeniser.peek()
+		if not char then break end
+
+		if char == "." then
+			tokeniser.next() -- consume '.'
+			local name = tokeniser.read_identifier()
+			if name == "" then
+				error("Expected class name at position " .. tokeniser.pos())
+			end
+			--table.insert(selectors, {type = "class", value = name})
+			table.insert( selector.class, name )
+		elseif char == "#" then
+			tokeniser.next() -- consume '#'
+			local name = tokeniser.read_identifier()
+			if name == "" then
+				error("Expected id at position " .. tokeniser.pos())
+			end
+			--table.insert(selectors, {type = "id", value = name})
+			selector.id = name
+		else
+			break
+		end
+	end
+
+	return selector
+end
+
+
+local function parse_combinator( tokeniser )
+	-- Skip leading whitespace
+	while tokeniser.peek() and tokeniser.peek():match("%s") do
+		tokeniser.next()
+	end
+
+	local char = tokeniser.peek()
+	if not char then return nil end
+
+	if char == ">" or char == "+" or char == "~" then
+		tokeniser.next()
+		-- Skip trailing whitespace
+		while tokeniser.peek() and tokeniser.peek():match("%s") do
+			tokeniser.next()
+		end
+		return COMBINATOR_CHARS[char]
+	else
+		-- Make sure next character isn't an explicit combinator
+		char = tokeniser.peek()
+		if char and not (char == ">" or char == "+" or char == "~") then
+			return COMBINATORS.DESCENDANT
+		end
+	end
+
+	return nil
+end
+
+
+
+
+function M.parse( input )
+	input = trim( input )
+
+	local tokeniser = create_tokeniser( input )
+
+	local output = { selector = parse_compound_selector( tokeniser ) }
+	local current = output
+
+	-- Parse combinations of combinators and compound selectors
+	while true do
+		local combinator = parse_combinator( tokeniser )
+		if not combinator then
+			current.combinator = nil
+			current.next = nil
+			break
+		end
+
+		local next_selector = parse_compound_selector( tokeniser )
+		current.combinator = combinator
+		current.next = { selector = next_selector }
+		current = current.next
+	end
+
+	return output
+end
+
+
+return M
+

html.lua

@@ -3,6 +3,13 @@ local function trim(str)
 	return str:match("^%s*(.-)%s*$")
 end
 
+local function shallow_copy(t)
+  local t2 = {}
+  for k,v in pairs(t) do
+    t2[k] = v
+  end
+  return t2
+end
 
 
 local M = {}
@@ -35,10 +42,50 @@ local VOID_TAGS = {
 function M.make_dom_element( tag_name, parent_elem )
 	local o = {
 		tag_name = tag_name,
-		parent = parent_elem,
-		children = {},
 		attributes = {},
-		content = ""
+		content = "",
+
+		children = {},
+		parent = parent_elem,
+
+		get_child_index = function( self )
+			if not self.parent then
+				return -1
+			end
+
+			for i, child in ipairs(self.parent.children) do
+				if child == self then return i end
+			end
+		end,
+
+		get_next_sibling = function( self )
+			if not self.parent then return nil end
+
+			local found_self = false
+			for _, child in ipairs(self.parent.children) do
+				if found_self then
+					return child
+				end
+
+				if child == self then
+					found_self = true
+				end
+			end
+
+			return nil
+		end,
+
+		check_simple_selector = function( self, selector )
+			return M.check_simple_selector( self, selector )
+		end,
+
+		foreach = function( self, fn )
+			fn( self )
+
+			for _, child in ipairs(self.children or {}) do
+				child:foreach( fn )
+			end
+		end
 	}
 
 	if parent_elem then
@@ -260,6 +307,64 @@ function M.tokenise( content )
 end
 
 
+function M.check_simple_selector(element, selector)
+		-- Skip text nodes
+		if element.tag_name == ":text" then
+			return false
+		end
+
+		-- Check tag name if specified
+		if selector.tag_name and element.tag_name ~= selector.tag_name then
+			return false
+		end
+
+		-- Check ID if specified
+		if selector.id and element.attributes.id ~= selector.id then
+			return false
+		end
+
+		-- Check classes if specified
+		if selector.class and #selector.class > 0 then
+			local element_classes = element.attributes.class
+			if not element_classes then
+				return false
+			end
+
+			for _, class in ipairs(selector.class) do
+				local found = false
+				for _, elem_class in ipairs(element_classes) do
+					if elem_class == class then
+						found = true
+						break
+					end
+				end
+				if not found then
+					return false
+				end
+			end
+		end
+
+		return true
+	end
+
+function M.query_simple_selector(document, selector)
+	local matches = {}
+
+	local function traverse(node)
+		if M.check_simple_selector(node, selector) then
+			table.insert(matches, node)
+		end
+
+		for _, child in ipairs(node.children) do
+			traverse(child)
+		end
+	end
+
+	traverse(document)
+	return matches
+end
+
+
 function M.parse_tokens_into_document( TOKENS )
 	local DOCUMENT = M.make_dom_element(nil, nil)
 	local current_doc_element = DOCUMENT
@@ -270,7 +375,7 @@ function M.parse_tokens_into_document( TOKENS )
 		local token = TOKENS[i]
 
 		if token.type == "WORD" then
-			if current_doc_element.tag_name == "#text" then
+			if current_doc_element.tag_name == ":text" then
 				current_doc_element = current_doc_element.parent
 			end
 
@@ -322,18 +427,21 @@ function M.parse_tokens_into_document( TOKENS )
 				end
 
 
-
 				local value = nil
 				if raw_value == "" or raw_value == nil then
 					value = nil
-					--elseif raw_value:find("%S+%s+%S+") then
-					--	value = {}
-					--	print(raw_value)
-					--	for word in raw_value:gmatch("%S+") do
-					--		table.insert( value, word )
-					--	end
 				else
 					value = trim(raw_value)
+
+					if name == "class" then
+						local classes = {}
+
+						for class in value:gmatch("%S+") do
+							table.insert( classes, class )
+						end
+
+						value = classes
+					end
 				end
 
 				current_doc_element.attributes[name] = value
@@ -364,7 +472,7 @@ function M.parse_tokens_into_document( TOKENS )
 
 
 		if token.type == "TEXT" then
-			local new_elem = M.make_dom_element( "#text", current_doc_element )
+			local new_elem = M.make_dom_element( ":text", current_doc_element )
 			new_elem.content = token.value
 			current_doc_element = new_elem
 
@@ -376,18 +484,20 @@ function M.parse_tokens_into_document( TOKENS )
 		i = i+1
 	end
 
+	M.clean_text_nodes( DOCUMENT )
+
 	return DOCUMENT
 end
 
 
 function M.clean_text_nodes(node)
-	if node.tag_name ~= "#text" then
+	if node.tag_name ~= ":text" then
 		-- Don't clean anything in raw text tags
 		if RAW_TEXT_TAGS[node.tag_name] then
 			return
 		end
 
-		for _, child in ipairs(node.children) do
+		for _, child in ipairs( shallow_copy(node.children) ) do
 			M.clean_text_nodes( child )
 		end
 		return
@@ -399,7 +509,7 @@ function M.clean_text_nodes(node)
 			error("Text node without a parent; should be impossible !")
 		end
 
-		for i, child in ipairs(node.parent.children) do
+		for i, child in ipairs( shallow_copy(node.parent.children) ) do
 			if child == node then
 				table.remove( node.parent.children, i )
 				break
@@ -413,47 +523,91 @@ function M.clean_text_nodes(node)
 end
 
 
-function M.print_document(node, indent)
+function M.tostring(node, indent, include_internal_pseudoelements)
 	-- Default indentation is 0 (root level)
 	indent = indent or 0
+	include_internal_pseudoelements = include_internal_pseudoelements or false
+
+	local is_pseudo_element = (node.tag_name or ":root"):sub(1,1) == ":"
+
 
 	local indent_level_str = "  "
 	-- Create the indentation string (e.g., "  " for each level)
 	local indent_str = string.rep(indent_level_str, indent)
 
-	if node.tag_name == "#text" then
+	if node.tag_name == ":text" then
-		print(indent_str .. "<#text>\n" .. node.content .. "\n" .. indent_str .. "</#text>")
+		local str = ""
-		return
+
+		if include_internal_pseudoelements then
+			str = str .. "<:text>"
+		end
+
+		str = str .. node.content
+
+		if include_internal_pseudoelements then
+			str = str .. "</:text>"
+		end
+
+		return str
 	end
 
 	local node_name = ""
 
+	if not is_pseudo_element or include_internal_pseudoelements then
 	-- Print the current node's tag name
-	node_name = node_name .. indent_str .. "<" .. (node.tag_name or "#root")
+		node_name = node_name .. "\n" .. indent_str .. "<" .. (node.tag_name or ":root")
+	end
 
 	-- Print attributes if any
 	if next(node.attributes) ~= nil then
 		for attr, value in pairs(node.attributes) do
-			--print(indent_str .. "  " .. attr .. " = " .. tostring(value))
+			if type(value) == "table" then
-			node_name = node_name .. " " .. attr .. "=\"" .. tostring(value) .. "\""
+				node_name = node_name .. " " .. attr .. "=\""
+				for i, val in ipairs( value ) do
+					if i > 1 then node_name = node_name .. " " end
+					node_name = node_name .. tostring(val)
+				end
+				node_name = node_name .. "\""
+			else
+				node_name = node_name .. " " .. attr .. "=\"" .. tostring(value) .. "\""
+			end
 		end
 	end
 
-	node_name = node_name .. ">"
+	if not is_pseudo_element or include_internal_pseudoelements then
+		node_name = node_name .. ">"
+	end
 
-	print( node_name )
+	--print( node_name )
 
+	local next_indent = indent + 1
+	if is_pseudo_element and not include_internal_pseudoelements then
+		next_indent = indent
+	end
 
 	-- Recursively print children
 	for _, child in ipairs(node.children) do
-		M.print_document(child, indent + 1)
+		node_name = node_name .. M.tostring(child, next_indent, include_internal_pseudoelements)
 	end
 
-	-- Print the closing tag
+	if not VOID_TAGS[node.tag_name] and ( not is_pseudo_element or include_internal_pseudoelements ) then
-	print(indent_str .. "</" .. (node.tag_name or "#root") .. ">")
+		-- Print the closing tag
+		local end_indent = ""
+		local closing_text_tag = "</:text>"
+		if node_name:sub(#node_name, #node_name) == ">" and node_name:sub(#node_name - #closing_text_tag + 1, #node_name) ~= closing_text_tag then
+			end_indent = "\n" .. indent_str
+		end
+		node_name = node_name .. end_indent .. "</" .. (node.tag_name or ":root") .. ">"
+	end
+
+	return node_name
 end
 
 
+
+
+
+
 function M.parse( html_string )
 	local clean_html = M.preprocess( html_string )
 
@@ -461,9 +615,7 @@ function M.parse( html_string )
 
 	local document = M.parse_tokens_into_document( tokens )
 
-	local cleaned_doc = M.clean_text_nodes( document )
+	return document
-
-	return cleaned_doc
 end
 
 return M

main.lua

@@ -1,9 +1,10 @@
 #!/bin/env lua
 
 local html = require(".html")
+local css = require(".css")
 
 
-local file = io.open("test.html", "r")
+local file = io.open("small.html", "r")
 
 if file == nil then
 	error("File doesn't exist")
@@ -11,4 +12,94 @@ end
 
 local content = file:read("a")
 
-html.print_document( html.parse( content ) )
+local doc = html.parse( content )
+
+
+print("Write a css selector:")
+local whole_selector = css.parse( io.read() )
+local current_selector = whole_selector
+
+
+local elements = {}
+-- start with all elements matching the first selector
+doc:foreach(function( el )
+	if el:check_simple_selector( current_selector.selector ) then
+		table.insert( elements, el )
+	end
+end)
+
+while current_selector.combinator ~= nil do
+	local next_selector = current_selector.next
+
+	local new_elements = {}
+
+	if current_selector.combinator == css.COMBINATORS.DESCENDANT then
+		for _, element in ipairs( elements ) do
+			element:foreach(function( el )
+				if el:check_simple_selector( next_selector.selector ) then
+					table.insert( new_elements, el )
+				end
+			end)
+		end
+
+		goto continue
+	end
+
+	if current_selector.combinator == css.COMBINATORS.DIRECT_DESCENDANT then
+		for _, element in ipairs( elements ) do
+			for _, child in ipairs( element.children ) do
+				if child:check_simple_selector( next_selector.selector ) then
+					table.insert( new_elements, child )
+				end
+			end
+		end
+
+		goto continue
+	end
+
+	if current_selector.combinator == css.COMBINATORS.NEXT_SIBLING then
+		for _, element in ipairs( elements ) do
+			local next_sibling = element:get_next_sibling()
+			while next_sibling and next_sibling.tag_name == ":text" do
+			 next_sibling = next_sibling:get_next_sibling()
+		 end
+
+			if next_sibling and next_sibling:check_simple_selector( next_selector.selector ) then
+				table.insert( new_elements, next_sibling )
+			end
+		end
+
+		goto continue
+	end
+
+	if current_selector.combinator == css.COMBINATORS.SUBSEQUENT_SIBLING then
+		for _, element in ipairs( elements ) do
+			local sibling = element:get_next_sibling()
+			while sibling ~= nil do
+				if sibling:check_simple_selector( next_selector.selector ) then
+					table.insert( new_elements, sibling )
+				end
+
+				sibling = sibling:get_next_sibling()
+			end
+		end
+
+		goto continue
+	end
+
+	::continue::
+	elements = new_elements
+	current_selector = next_selector
+end
+
+
+
+
+
+
+for _, el in ipairs(elements) do
+	print( html.tostring( el ) )
+end
+
+
+

small.html

@@ -0,0 +1,60 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>HTMLQ Test Document</title>
+</head>
+<body>
+    <div id="main" class="container wrapper">
+        <header class="header nav-bar">
+            <h1>HTMLQ Test Document</h1>
+            <nav class="nav" data-nav-type="main">
+                <ul>
+                    <li><a href="#" data-link-type="internal">Home</a></li>
+                    <li><a href="#" data-link-type="external">About</a></li>
+                    <li><a href="#" data-link-type="internal">Contact</a></li>
+                </ul>
+            </nav>
+        </header>
+        <main class="content main-content">
+            <section class="section featured" id="featured-section">
+                <h2>Featured Section</h2>
+                <p>This is the featured section.</p>
+                <div class="card featured-card" data-card-type="featured">
+                    <h3>Featured Card</h3>
+                    <p>This is the featured card.</p>
+                    <ul>
+                        <li>Item 1</li>
+                        <li>Item 2</li>
+                        <li>Item 3</li>
+                    </ul>
+                </div>
+            </section>
+            <section class="section regular" id="regular-section">
+                <h2>Regular Section</h2>
+                <p>This is the regular section.</p>
+                <div class="card regular-card" data-card-type="regular">
+                    <h3>Regular Card</h3>
+                    <p>This is the regular card.</p>
+                    <ol>
+                        <li>Item 1</li>
+                        <li>Item 2</li>
+                        <li>Item 3</li>
+                    </ol>
+                </div>
+            </section>
+        </main>
+        <footer class="footer nav-bar" data-footer-type="main">
+            <p>&copy; 2024 HTMLQ Test Document</p>
+            <nav class="nav" data-nav-type="footer">
+                <ul>
+                    <li><a href="#" data-link-type="internal">Home</a></li>
+                    <li><a href="#" data-link-type="external">About</a></li>
+                    <li><a href="#" data-link-type="internal">Contact</a></li>
+                </ul>
+            </nav>
+        </footer>
+    </div>
+</body>
+</html>