Merge pull request #502 from lightpanda-io/cdp_node_children

Cdp node children
This commit is contained in:
Pierre Tachoire
2025-04-08 15:45:16 +02:00
committed by GitHub
5 changed files with 356 additions and 171 deletions

View File

@@ -22,79 +22,19 @@ const Allocator = std.mem.Allocator;
pub const Id = u32; pub const Id = u32;
const log = std.log.scoped(.cdp_node);
const Node = @This(); const Node = @This();
id: Id, id: Id,
parent_id: ?Id = null,
node_type: u32,
backend_node_id: Id,
node_name: []const u8,
local_name: []const u8,
node_value: []const u8,
child_node_count: u32,
children: []const *Node,
document_url: ?[]const u8,
base_url: ?[]const u8,
xml_version: []const u8,
compatibility_mode: CompatibilityMode,
is_scrollable: bool,
_node: *parser.Node, _node: *parser.Node,
const CompatibilityMode = enum {
NoQuirksMode,
};
pub fn jsonStringify(self: *const Node, writer: anytype) !void {
try writer.beginObject();
try writer.objectField("nodeId");
try writer.write(self.id);
try writer.objectField("parentId");
try writer.write(self.parent_id);
try writer.objectField("backendNodeId");
try writer.write(self.backend_node_id);
try writer.objectField("nodeType");
try writer.write(self.node_type);
try writer.objectField("nodeName");
try writer.write(self.node_name);
try writer.objectField("localName");
try writer.write(self.local_name);
try writer.objectField("nodeValue");
try writer.write(self.node_value);
try writer.objectField("childNodeCount");
try writer.write(self.child_node_count);
try writer.objectField("children");
try writer.write(self.children);
try writer.objectField("documentURL");
try writer.write(self.document_url);
try writer.objectField("baseURL");
try writer.write(self.base_url);
try writer.objectField("xmlVersion");
try writer.write(self.xml_version);
try writer.objectField("compatibilityMode");
try writer.write(self.compatibility_mode);
try writer.objectField("isScrollable");
try writer.write(self.is_scrollable);
try writer.endObject();
}
// Whenever we send a node to the client, we register it here for future lookup. // Whenever we send a node to the client, we register it here for future lookup.
// We maintain a node -> id and id -> node lookup. // We maintain a node -> id and id -> node lookup.
pub const Registry = struct { pub const Registry = struct {
node_id: u32, node_id: u32,
allocator: Allocator, allocator: Allocator,
arena: std.heap.ArenaAllocator,
node_pool: std.heap.MemoryPool(Node), node_pool: std.heap.MemoryPool(Node),
lookup_by_id: std.AutoHashMapUnmanaged(Id, *Node), lookup_by_id: std.AutoHashMapUnmanaged(Id, *Node),
lookup_by_node: std.HashMapUnmanaged(*parser.Node, *Node, NodeContext, std.hash_map.default_max_load_percentage), lookup_by_node: std.HashMapUnmanaged(*parser.Node, *Node, NodeContext, std.hash_map.default_max_load_percentage),
@@ -102,9 +42,10 @@ pub const Registry = struct {
pub fn init(allocator: Allocator) Registry { pub fn init(allocator: Allocator) Registry {
return .{ return .{
.node_id = 0, .node_id = 0,
.allocator = allocator,
.lookup_by_id = .{}, .lookup_by_id = .{},
.lookup_by_node = .{}, .lookup_by_node = .{},
.allocator = allocator,
.arena = std.heap.ArenaAllocator.init(allocator),
.node_pool = std.heap.MemoryPool(Node).init(allocator), .node_pool = std.heap.MemoryPool(Node).init(allocator),
}; };
} }
@@ -114,12 +55,14 @@ pub const Registry = struct {
self.lookup_by_id.deinit(allocator); self.lookup_by_id.deinit(allocator);
self.lookup_by_node.deinit(allocator); self.lookup_by_node.deinit(allocator);
self.node_pool.deinit(); self.node_pool.deinit();
self.arena.deinit();
} }
pub fn reset(self: *Registry) void { pub fn reset(self: *Registry) void {
self.lookup_by_id.clearRetainingCapacity(); self.lookup_by_id.clearRetainingCapacity();
self.lookup_by_node.clearRetainingCapacity(); self.lookup_by_node.clearRetainingCapacity();
_ = self.node_pool.reset(.{ .retain_capacity = {} }); _ = self.arena.reset(.{ .retain_with_limit = 1024 });
_ = self.node_pool.reset(.{ .retain_with_limit = 1024 });
} }
pub fn register(self: *Registry, n: *parser.Node) !*Node { pub fn register(self: *Registry, n: *parser.Node) !*Node {
@@ -132,38 +75,17 @@ pub const Registry = struct {
// but, just in case, let's try to keep things tidy. // but, just in case, let's try to keep things tidy.
errdefer _ = self.lookup_by_node.remove(n); errdefer _ = self.lookup_by_node.remove(n);
const children = try parser.nodeGetChildNodes(n);
const children_count = try parser.nodeListLength(children);
const id = self.node_id;
defer self.node_id = id + 1;
const node = try self.node_pool.create(); const node = try self.node_pool.create();
errdefer self.node_pool.destroy(node); errdefer self.node_pool.destroy(node);
const id = self.node_id;
self.node_id = id + 1;
node.* = .{ node.* = .{
._node = n, ._node = n,
.id = id, .id = id,
.parent_id = null, // TODO
.backend_node_id = id, // ??
.node_name = try parser.nodeName(n),
.local_name = try parser.nodeLocalName(n),
.node_value = try parser.nodeValue(n) orelse "",
.node_type = @intFromEnum(try parser.nodeType(n)),
.child_node_count = children_count,
.children = &.{}, // TODO
.document_url = null,
.base_url = null,
.xml_version = "",
.compatibility_mode = .NoQuirksMode,
.is_scrollable = false,
}; };
// if (try parser.nodeParentNode(n)) |pn| {
// _ = pn;
// // TODO
// }
node_lookup_gop.value_ptr.* = node; node_lookup_gop.value_ptr.* = node;
try self.lookup_by_id.putNoClobber(self.allocator, id, node); try self.lookup_by_id.putNoClobber(self.allocator, id, node);
return node; return node;
@@ -271,8 +193,107 @@ pub const Search = struct {
}; };
}; };
// Need a custom writer, because we can't just serialize the node as-is.
// Sometimes we want to serializ the node without chidren, sometimes with just
// its direct children, and sometimes the entire tree.
// (For now, we only support direct children)
pub const Writer = struct {
opts: Opts,
node: *const Node,
registry: *Registry,
pub const Opts = struct {};
pub fn jsonStringify(self: *const Writer, w: anytype) !void {
self.toJSON(w) catch |err| {
// The only error our jsonStringify method can return is
// @TypeOf(w).Error. In other words, our code can't return its own
// error, we can only return a writer error. Kinda sucks.
log.err("json stringify: {}", .{err});
return error.OutOfMemory;
};
}
fn toJSON(self: *const Writer, w: anytype) !void {
try w.beginObject();
try writeCommon(self.node, false, w);
{
var registry = self.registry;
const child_nodes = try parser.nodeGetChildNodes(self.node._node);
const child_count = try parser.nodeListLength(child_nodes);
var i: usize = 0;
try w.objectField("children");
try w.beginArray();
for (0..child_count) |_| {
const child = (try parser.nodeListItem(child_nodes, @intCast(i))) orelse break;
const child_node = try registry.register(child);
try w.beginObject();
try writeCommon(child_node, true, w);
try w.endObject();
i += 1;
}
try w.endArray();
try w.objectField("childNodeCount");
try w.write(i);
}
try w.endObject();
}
fn writeCommon(node: *const Node, include_child_count: bool, w: anytype) !void {
try w.objectField("nodeId");
try w.write(node.id);
try w.objectField("backendNodeId");
try w.write(node.id);
const n = node._node;
// TODO:
// try w.objectField("parentId");
// try w.write(pid);
try w.objectField("nodeType");
try w.write(@intFromEnum(try parser.nodeType(n)));
try w.objectField("nodeName");
try w.write(try parser.nodeName(n));
try w.objectField("localName");
try w.write(try parser.nodeLocalName(n));
try w.objectField("nodeValue");
try w.write((try parser.nodeValue(n)) orelse "");
if (include_child_count) {
try w.objectField("childNodeCount");
const child_nodes = try parser.nodeGetChildNodes(n);
try w.write(try parser.nodeListLength(child_nodes));
}
try w.objectField("documentURL");
try w.write(null);
try w.objectField("baseURL");
try w.write(null);
try w.objectField("xmlVersion");
try w.write("");
try w.objectField("compatibilityMode");
try w.write("NoQuirksMode");
try w.objectField("isScrollable");
try w.write(false);
}
};
const testing = @import("testing.zig"); const testing = @import("testing.zig");
test "CDP Node: Registry register" { test "cdp Node: Registry register" {
var registry = Registry.init(testing.allocator); var registry = Registry.init(testing.allocator);
defer registry.deinit(); defer registry.deinit();
@@ -291,19 +312,6 @@ test "CDP Node: Registry register" {
try testing.expectEqual(node, n1c); try testing.expectEqual(node, n1c);
try testing.expectEqual(0, node.id); try testing.expectEqual(0, node.id);
try testing.expectEqual(null, node.parent_id);
try testing.expectEqual(1, node.node_type);
try testing.expectEqual(0, node.backend_node_id);
try testing.expectEqual("A", node.node_name);
try testing.expectEqual("a", node.local_name);
try testing.expectEqual("", node.node_value);
try testing.expectEqual(1, node.child_node_count);
try testing.expectEqual(0, node.children.len);
try testing.expectEqual(null, node.document_url);
try testing.expectEqual(null, node.base_url);
try testing.expectEqual("", node.xml_version);
try testing.expectEqual(.NoQuirksMode, node.compatibility_mode);
try testing.expectEqual(false, node.is_scrollable);
try testing.expectEqual(n, node._node); try testing.expectEqual(n, node._node);
} }
@@ -316,24 +324,11 @@ test "CDP Node: Registry register" {
try testing.expectEqual(node, n1c); try testing.expectEqual(node, n1c);
try testing.expectEqual(1, node.id); try testing.expectEqual(1, node.id);
try testing.expectEqual(null, node.parent_id);
try testing.expectEqual(1, node.node_type);
try testing.expectEqual(1, node.backend_node_id);
try testing.expectEqual("P", node.node_name);
try testing.expectEqual("p", node.local_name);
try testing.expectEqual("", node.node_value);
try testing.expectEqual(1, node.child_node_count);
try testing.expectEqual(0, node.children.len);
try testing.expectEqual(null, node.document_url);
try testing.expectEqual(null, node.base_url);
try testing.expectEqual("", node.xml_version);
try testing.expectEqual(.NoQuirksMode, node.compatibility_mode);
try testing.expectEqual(false, node.is_scrollable);
try testing.expectEqual(n, node._node); try testing.expectEqual(n, node._node);
} }
} }
test "CDP Node: search list" { test "cdp Node: search list" {
var registry = Registry.init(testing.allocator); var registry = Registry.init(testing.allocator);
defer registry.deinit(); defer registry.deinit();
@@ -383,3 +378,102 @@ test "CDP Node: search list" {
try testing.expectEqual(2, registry.lookup_by_node.count()); try testing.expectEqual(2, registry.lookup_by_node.count());
} }
} }
test "cdp Node: Writer" {
var registry = Registry.init(testing.allocator);
defer registry.deinit();
var doc = try testing.Document.init("<a id=a1></a><a id=a2></a>");
defer doc.deinit();
{
const node = try registry.register(doc.asNode());
const json = try std.json.stringifyAlloc(testing.allocator, Writer{
.node = node,
.opts = .{},
.registry = &registry,
}, .{});
defer testing.allocator.free(json);
try testing.expectJson(.{
.nodeId = 0,
.backendNodeId = 0,
.nodeType = 9,
.nodeName = "#document",
.localName = "",
.nodeValue = "",
.documentURL = null,
.baseURL = null,
.xmlVersion = "",
.isScrollable = false,
.compatibilityMode = "NoQuirksMode",
.childNodeCount = 1,
.children = &.{.{
.nodeId = 1,
.backendNodeId = 1,
.nodeType = 1,
.nodeName = "HTML",
.localName = "html",
.nodeValue = "",
.childNodeCount = 2,
.documentURL = null,
.baseURL = null,
.xmlVersion = "",
.compatibilityMode = "NoQuirksMode",
.isScrollable = false,
}},
}, json);
}
{
const node = registry.lookup_by_id.get(1).?;
const json = try std.json.stringifyAlloc(testing.allocator, Writer{
.node = node,
.opts = .{},
.registry = &registry,
}, .{});
defer testing.allocator.free(json);
try testing.expectJson(.{
.nodeId = 1,
.backendNodeId = 1,
.nodeType = 1,
.nodeName = "HTML",
.localName = "html",
.nodeValue = "",
.childNodeCount = 2,
.documentURL = null,
.baseURL = null,
.xmlVersion = "",
.compatibilityMode = "NoQuirksMode",
.isScrollable = false,
.children = &.{ .{
.nodeId = 2,
.backendNodeId = 2,
.nodeType = 1,
.nodeName = "HEAD",
.localName = "head",
.nodeValue = "",
.childNodeCount = 0,
.documentURL = null,
.baseURL = null,
.xmlVersion = "",
.compatibilityMode = "NoQuirksMode",
.isScrollable = false,
}, .{
.nodeId = 3,
.backendNodeId = 3,
.nodeType = 1,
.nodeName = "BODY",
.localName = "body",
.nodeValue = "",
.childNodeCount = 2,
.documentURL = null,
.baseURL = null,
.xmlVersion = "",
.compatibilityMode = "NoQuirksMode",
.isScrollable = false,
} },
}, json);
}
}

View File

@@ -325,6 +325,14 @@ pub fn BrowserContext(comptime CDP_T: type) type {
self.node_search_list.reset(); self.node_search_list.reset();
} }
pub fn nodeWriter(self: *Self, node: *const Node, opts: Node.Writer.Opts) Node.Writer {
return .{
.node = node,
.opts = opts,
.registry = &self.node_registry,
};
}
pub fn onInspectorResponse(ctx: *anyopaque, _: u32, msg: []const u8) void { pub fn onInspectorResponse(ctx: *anyopaque, _: u32, msg: []const u8) void {
if (std.log.defaultLogEnabled(.debug)) { if (std.log.defaultLogEnabled(.debug)) {
// msg should be {"id":<id>,... // msg should be {"id":<id>,...

View File

@@ -51,9 +51,7 @@ fn getDocument(cmd: anytype) !void {
const doc = page.doc orelse return error.DocumentNotLoaded; const doc = page.doc orelse return error.DocumentNotLoaded;
const node = try bc.node_registry.register(parser.documentToNode(doc)); const node = try bc.node_registry.register(parser.documentToNode(doc));
return cmd.sendResult(.{ return cmd.sendResult(.{ .root = bc.nodeWriter(node, .{}) }, .{});
.root = node,
}, .{});
} }
// https://chromedevtools.github.io/devtools-protocol/tot/DOM/#method-performSearch // https://chromedevtools.github.io/devtools-protocol/tot/DOM/#method-performSearch
@@ -118,6 +116,7 @@ fn getSearchResults(cmd: anytype) !void {
} }
const testing = @import("../testing.zig"); const testing = @import("../testing.zig");
test "cdp.dom: getSearchResults unknown search id" { test "cdp.dom: getSearchResults unknown search id" {
var ctx = testing.context(); var ctx = testing.context();
defer ctx.deinit(); defer ctx.deinit();

View File

@@ -26,11 +26,12 @@ const main = @import("cdp.zig");
const parser = @import("netsurf"); const parser = @import("netsurf");
const App = @import("../app.zig").App; const App = @import("../app.zig").App;
pub const allocator = @import("../testing.zig").allocator; const base = @import("../testing.zig");
pub const allocator = base.allocator;
pub const expectEqual = @import("../testing.zig").expectEqual; pub const expectJson = base.expectJson;
pub const expectError = @import("../testing.zig").expectError; pub const expectEqual = base.expectEqual;
pub const expectEqualSlices = @import("../testing.zig").expectEqualSlices; pub const expectError = base.expectError;
pub const expectEqualSlices = base.expectEqualSlices;
pub const Document = @import("../testing.zig").Document; pub const Document = @import("../testing.zig").Document;
@@ -284,6 +285,7 @@ const TestContext = struct {
_ = self.client.?.serialized.orderedRemove(i); _ = self.client.?.serialized.orderedRemove(i);
return; return;
} }
std.debug.print("Error not found. Expecting:\n{s}\n\nGot:\n", .{serialized}); std.debug.print("Error not found. Expecting:\n{s}\n\nGot:\n", .{serialized});
for (self.client.?.serialized.items, 0..) |sent, i| { for (self.client.?.serialized.items, 0..) |sent, i| {
std.debug.print("#{d}\n{s}\n\n", .{ i, sent }); std.debug.print("#{d}\n{s}\n\n", .{ i, sent });
@@ -310,47 +312,5 @@ pub fn context() TestContext {
fn compareExpectedToSent(expected: []const u8, actual: json.Value) !bool { fn compareExpectedToSent(expected: []const u8, actual: json.Value) !bool {
const expected_value = try std.json.parseFromSlice(json.Value, std.testing.allocator, expected, .{}); const expected_value = try std.json.parseFromSlice(json.Value, std.testing.allocator, expected, .{});
defer expected_value.deinit(); defer expected_value.deinit();
return compareJsonValues(expected_value.value, actual); return base.isEqualJson(expected_value.value, actual);
}
fn compareJsonValues(a: std.json.Value, b: std.json.Value) bool {
if (!std.mem.eql(u8, @tagName(a), @tagName(b))) {
return false;
}
switch (a) {
.null => return true,
.bool => return a.bool == b.bool,
.integer => return a.integer == b.integer,
.float => return a.float == b.float,
.number_string => return std.mem.eql(u8, a.number_string, b.number_string),
.string => return std.mem.eql(u8, a.string, b.string),
.array => {
const a_len = a.array.items.len;
const b_len = b.array.items.len;
if (a_len != b_len) {
return false;
}
for (a.array.items, b.array.items) |a_item, b_item| {
if (compareJsonValues(a_item, b_item) == false) {
return false;
}
}
return true;
},
.object => {
var it = a.object.iterator();
while (it.next()) |entry| {
const key = entry.key_ptr.*;
if (b.object.get(key)) |b_item| {
if (compareJsonValues(entry.value_ptr.*, b_item) == false) {
return false;
}
} else {
return false;
}
}
return true;
},
}
} }

View File

@@ -17,14 +17,15 @@
// along with this program. If not, see <https://www.gnu.org/licenses/>. // along with this program. If not, see <https://www.gnu.org/licenses/>.
const std = @import("std"); const std = @import("std");
const parser = @import("netsurf");
const parser = @import("netsurf");
pub const allocator = std.testing.allocator; pub const allocator = std.testing.allocator;
pub const expectError = std.testing.expectError; pub const expectError = std.testing.expectError;
pub const expectString = std.testing.expectEqualStrings; pub const expectString = std.testing.expectEqualStrings;
pub const expectEqualSlices = std.testing.expectEqualSlices; pub const expectEqualSlices = std.testing.expectEqualSlices;
const App = @import("app.zig").App; const App = @import("app.zig").App;
const Allocator = std.mem.Allocator;
// Merged std.testing.expectEqual and std.testing.expectString // Merged std.testing.expectEqual and std.testing.expectString
// can be useful when testing fields of an anytype an you don't know // can be useful when testing fields of an anytype an you don't know
@@ -217,12 +218,135 @@ pub const Document = struct {
pub fn querySelectorAll(self: *Document, selector: []const u8) ![]const *parser.Node { pub fn querySelectorAll(self: *Document, selector: []const u8) ![]const *parser.Node {
const css = @import("dom/css.zig"); const css = @import("dom/css.zig");
const node_list = try css.querySelectorAll(self.arena.allocator(), parser.documentToNode(self.doc), selector); const node_list = try css.querySelectorAll(self.arena.allocator(), self.asNode(), selector);
return node_list.nodes.items; return node_list.nodes.items;
} }
pub fn querySelector(self: *Document, selector: []const u8) !?*parser.Node { pub fn querySelector(self: *Document, selector: []const u8) !?*parser.Node {
const css = @import("dom/css.zig"); const css = @import("dom/css.zig");
return css.querySelector(self.arena.allocator(), parser.documentToNode(self.doc), selector); return css.querySelector(self.arena.allocator(), self.asNode(), selector);
}
pub fn asNode(self: *const Document) *parser.Node {
return parser.documentToNode(self.doc);
} }
}; };
pub fn expectJson(a: anytype, b: anytype) !void {
var arena = std.heap.ArenaAllocator.init(allocator);
defer arena.deinit();
const aa = arena.allocator();
const a_value = try convertToJson(aa, a);
const b_value = try convertToJson(aa, b);
errdefer {
const a_json = std.json.stringifyAlloc(aa, a_value, .{ .whitespace = .indent_2 }) catch unreachable;
const b_json = std.json.stringifyAlloc(aa, b_value, .{ .whitespace = .indent_2 }) catch unreachable;
std.debug.print("== Expected ==\n{s}\n\n== Actual ==\n{s}", .{ a_json, b_json });
}
try expectJsonValue(a_value, b_value);
}
pub fn isEqualJson(a: anytype, b: anytype) !bool {
var arena = std.heap.ArenaAllocator.init(allocator);
defer arena.deinit();
const aa = arena.allocator();
const a_value = try convertToJson(aa, a);
const b_value = try convertToJson(aa, b);
return isJsonValue(a_value, b_value);
}
fn convertToJson(arena: Allocator, value: anytype) !std.json.Value {
const T = @TypeOf(value);
if (T == std.json.Value) {
return value;
}
var str: []const u8 = undefined;
if (T == []u8 or T == []const u8 or comptime isStringArray(T)) {
str = value;
} else {
str = try std.json.stringifyAlloc(arena, value, .{});
}
return std.json.parseFromSliceLeaky(std.json.Value, arena, str, .{});
}
fn expectJsonValue(a: std.json.Value, b: std.json.Value) !void {
try expectEqual(@tagName(a), @tagName(b));
// at this point, we know that if a is an int, b must also be an int
switch (a) {
.null => return,
.bool => try expectEqual(a.bool, b.bool),
.integer => try expectEqual(a.integer, b.integer),
.float => try expectEqual(a.float, b.float),
.number_string => try expectEqual(a.number_string, b.number_string),
.string => try expectEqual(a.string, b.string),
.array => {
const a_len = a.array.items.len;
const b_len = b.array.items.len;
try expectEqual(a_len, b_len);
for (a.array.items, b.array.items) |a_item, b_item| {
try expectJsonValue(a_item, b_item);
}
},
.object => {
var it = a.object.iterator();
while (it.next()) |entry| {
const key = entry.key_ptr.*;
if (b.object.get(key)) |b_item| {
try expectJsonValue(entry.value_ptr.*, b_item);
} else {
return error.MissingKey;
}
}
},
}
}
fn isJsonValue(a: std.json.Value, b: std.json.Value) bool {
if (std.mem.eql(u8, @tagName(a), @tagName(b)) == false) {
return false;
}
// at this point, we know that if a is an int, b must also be an int
switch (a) {
.null => return true,
.bool => return a.bool == b.bool,
.integer => return a.integer == b.integer,
.float => return a.float == b.float,
.number_string => return std.mem.eql(u8, a.number_string, b.number_string),
.string => return std.mem.eql(u8, a.string, b.string),
.array => {
const a_len = a.array.items.len;
const b_len = b.array.items.len;
if (a_len != b_len) {
return false;
}
for (a.array.items, b.array.items) |a_item, b_item| {
if (isJsonValue(a_item, b_item) == false) {
return false;
}
}
return true;
},
.object => {
var it = a.object.iterator();
while (it.next()) |entry| {
const key = entry.key_ptr.*;
if (b.object.get(key)) |b_item| {
if (isJsonValue(entry.value_ptr.*, b_item) == false) {
return false;
}
} else {
return false;
}
}
return true;
},
}
}