From 75ab12cb4d52ef8e4729d742bff645b49bfce66e Mon Sep 17 00:00:00 2001
From: Erik Verbruggen <erik@verbruggen.consulting>
Date: Fri, 29 Nov 2024 16:36:01 +0100
Subject: [PATCH] chore: log percent-encoded URL/filename (#11998)

This can be used to debug Unicode normalization/encoding issues.
---
 src/libsync/discovery.cpp  | 11 ++++++++++-
 src/libsync/httplogger.cpp |  2 +-
 2 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/src/libsync/discovery.cpp b/src/libsync/discovery.cpp
index 2953d3a83c3..f96cb363f9b 100644
--- a/src/libsync/discovery.cpp
+++ b/src/libsync/discovery.cpp
@@ -308,8 +308,15 @@ void ProcessDirectoryJob::processFile(const PathTuple &path,
     const LocalInfo &localEntry, const RemoteInfo &serverEntry,
     const SyncJournalFileRecord &dbEntry)
 {
+    // The percent-encoded file name as it would be passed in an HTTP request. This can be used to
+    // debug Unicode encoding/normalization issues.
+    auto encodedFileName = QUrl::fromLocalFile(path._original).toEncoded().sliced(std::string_view("file:").length());
     const char *hasServer = serverEntry.isValid() ? "true" : _queryServer == ParentNotChanged ? "db" : "false";
     const char *hasLocal = localEntry.isValid() ? "true" : _queryLocal == ParentNotChanged ? "db" : "false";
+
+    // The code below is formatted like this ON PURPOSE: each field is db/local/server, one field
+    // per (source code) line, so it's easy to understand.
+    // clang-format off
     qCInfo(lcDisco).nospace() << "Processing " << path._original
                               << " | valid: " << dbEntry.isValid() << "/" << hasLocal << "/" << hasServer
                               << " | mtime: " << dbEntry._modtime << "/" << localEntry.modtime << "/" << serverEntry.modtime
@@ -319,7 +326,9 @@ void ProcessDirectoryJob::processFile(const PathTuple &path,
                               << " | perm: " << dbEntry._remotePerm << "//" << serverEntry.remotePerm
                               << " | fileid: " << dbEntry._fileId << "//" << serverEntry.fileId
                               << " | inode: " << dbEntry._inode << "/" << localEntry.inode << "/"
-                              << " | type: " << dbEntry._type << "/" << localEntry.type << "/" << (serverEntry.isDirectory ? ItemTypeDirectory : ItemTypeFile);
+                              << " | type: " << dbEntry._type << "/" << localEntry.type << "/" << (serverEntry.isDirectory ? ItemTypeDirectory : ItemTypeFile)
+                              << " (" << encodedFileName << ")";
+    // clang-format on
 
     if (_discoveryData->isRenamed(path._original)) {
         qCDebug(lcDisco) << "Ignoring renamed";
diff --git a/src/libsync/httplogger.cpp b/src/libsync/httplogger.cpp
index 059dc90c5a8..d9b36a3f018 100644
--- a/src/libsync/httplogger.cpp
+++ b/src/libsync/httplogger.cpp
@@ -43,7 +43,7 @@ bool isTextBody(const QString &s)
 struct HttpContext
 {
     HttpContext(const QNetworkRequest &request)
-        : originalUrl(request.url().toString())
+        : originalUrl(QString::fromUtf8(request.url().toEncoded())) // Encoded URL as it is passed "over the wire"
         , lastUrl(request.url())
         , id(QString::fromUtf8(request.rawHeader(QByteArrayLiteral("X-Request-ID"))))
     {