Skip to content

Lightpanda: the headless browser designed for AI and automation

webapi: node length and Range text offsets are UTF-16 code units

Francis Bouvier20299e4b5c50
diff --git a/src/browser/webapi/Node.zig b/src/browser/webapi/Node.zig
index 76efe5141..4e7907139 100644
--- a/src/browser/webapi/Node.zig
+++ b/src/browser/webapi/Node.zig
@@ -915,7 +915,8 @@ pub fn getChildrenCount(self: *Node) usize {
pub fn getLength(self: *Node) u32 {
switch (self._type) {
.cdata => |cdata| {
- return @intCast(cdata.getData().len);
+ // The node length of CharacterData is in UTF-16 code units.
+ return @intCast(cdata.getLength());
},
.element, .document, .document_fragment => {
var count: u32 = 0;
diff --git a/src/browser/webapi/Range.zig b/src/browser/webapi/Range.zig
index d0f23c649..100e2a3a5 100644
--- a/src/browser/webapi/Range.zig
+++ b/src/browser/webapi/Range.zig
@@ -346,12 +346,13 @@ pub fn insertNode(self: *Range, node: *Node, frame: *Frame) !void {
_ = try parent.insertBefore(node, container, frame);
} else {
const text_data = container.getData().str();
- if (offset >= text_data.len) {
+ const byte_off = byteOffset(text_data, offset);
+ if (byte_off >= text_data.len) {
_ = try parent.insertBefore(node, container.nextSibling(), frame);
} else {
// Split the text node into before and after parts
- const before_text = text_data[0..offset];
- const after_text = text_data[offset..];
+ const before_text = text_data[0..byte_off];
+ const after_text = text_data[byte_off..];
const before = try Frame.node_factory.createTextNode(frame, before_text);
const after = try Frame.node_factory.createTextNode(frame, after_text);
@@ -374,6 +375,12 @@ pub fn insertNode(self: *Range, node: *Node, frame: *Frame) !void {
}
}
+// Range offsets in CharacterData are UTF-16 code units; convert one to a
+// byte index into the UTF-8 data, clamping out-of-range offsets to the end.
+fn byteOffset(data: []const u8, utf16_offset: u32) usize {
+ return Node.CData.utf16OffsetToUtf8(data, utf16_offset) catch data.len;
+}
+
pub fn deleteContents(self: *Range, frame: *Frame) !void {
if (self._proto.getCollapsed()) {
return;
@@ -388,7 +395,10 @@ pub fn deleteContents(self: *Range, frame: *Frame) !void {
const text_data = old_value.str();
cdata._data = try String.concat(
frame.arena,
- &.{ text_data[0..self._proto._start_offset], text_data[self._proto._end_offset..] },
+ &.{
+ text_data[0..byteOffset(text_data, self._proto._start_offset)],
+ text_data[byteOffset(text_data, self._proto._end_offset)..],
+ },
);
Frame.observers.notifyCharacterDataChange(frame, self._proto._start_container, old_value);
} else {
@@ -411,9 +421,10 @@ pub fn deleteContents(self: *Range, frame: *Frame) !void {
// Handle start container - if it's a text node, truncate it
if (self._proto._start_container.is(Node.CData)) |cdata| {
const text_data = cdata._data.str();
- if (self._proto._start_offset < text_data.len) {
+ const byte_start = byteOffset(text_data, self._proto._start_offset);
+ if (byte_start < text_data.len) {
// Keep only the part before start_offset
- const new_text = text_data[0..self._proto._start_offset];
+ const new_text = text_data[0..byte_start];
try self._proto._start_container.setData(new_text, frame);
}
}
@@ -421,11 +432,12 @@ pub fn deleteContents(self: *Range, frame: *Frame) !void {
// Handle end container - if it's a text node, truncate it
if (self._proto._end_container.is(Node.CData)) |cdata| {
const text_data = cdata._data.str();
- if (self._proto._end_offset < text_data.len) {
+ const byte_end = byteOffset(text_data, self._proto._end_offset);
+ if (byte_end < text_data.len) {
// Keep only the part from end_offset onwards
- const new_text = text_data[self._proto._end_offset..];
+ const new_text = text_data[byte_end..];
try self._proto._end_container.setData(new_text, frame);
- } else if (self._proto._end_offset == text_data.len) {
+ } else if (byte_end == text_data.len) {
// If we're at the end, set to empty (will be removed if needed)
try self._proto._end_container.setData("", frame);
}
@@ -457,8 +469,10 @@ pub fn cloneContents(self: *const Range, frame: *Frame) !*DocumentFragment {
if (self._proto._start_container.is(Node.CData)) |_| {
// Clone part of text node
const text_data = self._proto._start_container.getData().str();
- if (self._proto._start_offset < text_data.len and self._proto._end_offset <= text_data.len) {
- const cloned_text = text_data[self._proto._start_offset..self._proto._end_offset];
+ const byte_start = byteOffset(text_data, self._proto._start_offset);
+ const byte_end = byteOffset(text_data, self._proto._end_offset);
+ if (byte_start < text_data.len and byte_end <= text_data.len) {
+ const cloned_text = text_data[byte_start..byte_end];
const text_node = try Frame.node_factory.createTextNode(frame, cloned_text);
_ = try fragment.asNode().appendChild(text_node, frame);
}
@@ -478,9 +492,10 @@ pub fn cloneContents(self: *const Range, frame: *Frame) !*DocumentFragment {
// Clone partial start container
if (self._proto._start_container.is(Node.CData)) |_| {
const text_data = self._proto._start_container.getData().str();
- if (self._proto._start_offset < text_data.len) {
+ const byte_start = byteOffset(text_data, self._proto._start_offset);
+ if (byte_start < text_data.len) {
// Clone from start_offset to end of text
- const cloned_text = text_data[self._proto._start_offset..];
+ const cloned_text = text_data[byte_start..];
const text_node = try Frame.node_factory.createTextNode(frame, cloned_text);
_ = try fragment.asNode().appendChild(text_node, frame);
}
@@ -501,9 +516,10 @@ pub fn cloneContents(self: *const Range, frame: *Frame) !*DocumentFragment {
// Clone partial end container
if (self._proto._end_container.is(Node.CData)) |_| {
const text_data = self._proto._end_container.getData().str();
- if (self._proto._end_offset > 0 and self._proto._end_offset <= text_data.len) {
+ const byte_end = byteOffset(text_data, self._proto._end_offset);
+ if (byte_end > 0 and byte_end <= text_data.len) {
// Clone from start to end_offset
- const cloned_text = text_data[0..self._proto._end_offset];
+ const cloned_text = text_data[0..byte_end];
const text_node = try Frame.node_factory.createTextNode(frame, cloned_text);
_ = try fragment.asNode().appendChild(text_node, frame);
}
@@ -587,8 +603,8 @@ fn writeTextContent(self: *const Range, writer: *std.Io.Writer) !void {
if (start_node.is(Node.CData)) |cdata| {
if (!isCommentOrPI(cdata)) {
const data = cdata.getData().str();
- const s = @min(start_offset, data.len);
- const e = @min(end_offset, data.len);
+ const s = byteOffset(data, start_offset);
+ const e = byteOffset(data, end_offset);
try writer.writeAll(data[s..e]);
}
return;
@@ -601,7 +617,7 @@ fn writeTextContent(self: *const Range, writer: *std.Io.Writer) !void {
if (start_node.is(Node.CData)) |cdata| {
if (!isCommentOrPI(cdata)) {
const data = cdata.getData().str();
- const s = @min(start_offset, data.len);
+ const s = byteOffset(data, start_offset);
try writer.writeAll(data[s..]);
}
}
@@ -639,7 +655,7 @@ fn writeTextContent(self: *const Range, writer: *std.Io.Writer) !void {
if (end_node.is(Node.CData)) |cdata| {
if (!isCommentOrPI(cdata)) {
const data = cdata.getData().str();
- const e = @min(end_offset, data.len);
+ const e = byteOffset(data, end_offset);
try writer.writeAll(data[0..e]);
}
}