Copilot commented on code in PR #3743:
URL: https://github.com/apache/brpc/pull/3743#discussion_r4571669369
##########
src/brpc/builtin/common.cpp:
##########
@@ -128,21 +129,19 @@ std::ostream& operator<<(std::ostream& os, const
PrintedAsDateTime& d) {
}
std::ostream& operator<<(std::ostream& os, const Path& link) {
- if (link.html_addr) {
- if (link.html_addr != Path::LOCAL) {
- os << "<a href=\"http://" << *link.html_addr << link.uri << "\">";
- } else {
- os << "<a href=\"" << link.uri << "\">";
- }
+ const char* const text = (link.text ? link.text : link.uri);
+ if (!link.html_addr) {
+ os << text;
+ return os;
}
- if (link.text) {
- os << link.text;
+
+ const std::string escaped_uri = WebEscape(link.uri);
+ if (link.html_addr != Path::LOCAL) {
+ os << "<a href=\"http://" << *link.html_addr << escaped_uri << "\">";
} else {
- os << link.uri;
- }
- if (link.html_addr) {
- os << "</a>";
+ os << "<a href=\"" << escaped_uri << "\">";
Review Comment:
`WebEscape` only performs HTML entity escaping; it does not encode URI path
syntax. `/dir` passes raw filesystem names here, so a valid entry such as `a?x`
or `a#x` is rendered as `/dir/.../a?x` or `/dir/.../a#x`, causing the browser
to treat the suffix as a query or fragment instead of part of the filename.
Encode path components (while preserving the surrounding URI structure) before
HTML-escaping the href, and cover reserved filename characters in the
regression test.
##########
src/brpc/builtin/hotspots_service.cpp:
##########
@@ -1095,11 +1097,11 @@ static void StartProfiling(ProfilingType type,
"<select id='view_prof' onchange='onSelectProf()'>";
os << "<option value=''><new profile></option>";
for (size_t i = 0; i < past_profs.size(); ++i) {
- os << "<option value='" << past_profs[i] << "' ";
+ os << "<option value='" << WebEscape(past_profs[i]) << "' ";
if (view != nullptr && past_profs[i] == *view) {
os << "selected";
}
- os << '>' << GetBaseName(&past_profs[i]);
+ os << '>' << WebEscape(GetBaseName(&past_profs[i]));
Review Comment:
Escaping these option values only protects the HTML; the browser decodes the
entities back to the original path, while `generateURL()` and the AJAX URL
below still concatenate `view`/`base` without percent-encoding. The handlers
likewise read the query values without decoding before `ValidProfilePath`, so
the values produced by these options cannot be selected reliably (and reserved
characters can be parsed as query syntax). Percent-encode both query values
when constructing the URLs and `PercentDecode` them before validation/use in
both handlers; the raw-path regression test does not exercise the
browser-generated form.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]