From 0c7927e2d3611290d8ccf06932af9127f546b3d6 Mon Sep 17 00:00:00 2001
From: Dave Dykstra <2129743+DrDaveD@users.noreply.github.com>
Date: Wed, 23 Oct 2024 17:44:41 -0500
Subject: [PATCH] process the LINK: header

---
 cvmfs/network/download.cc | 94 +++++++++++++++++++++++++++++++++++----
 cvmfs/network/download.h  |  1 +
 2 files changed, 87 insertions(+), 8 deletions(-)

diff --git a/cvmfs/network/download.cc b/cvmfs/network/download.cc
index 00a4bf8637..9ef3f139c8 100644
--- a/cvmfs/network/download.cc
+++ b/cvmfs/network/download.cc
@@ -226,7 +226,12 @@ static size_t CallbackCurlHeader(void *ptr, size_t size, size_t nmemb,
     // This is metalink info
     LogCvmfs(kLogDownload, kLogDebug, "(id %" PRId64 ") %s",
                                       info->id(), header_line.c_str());
-    info->SetLink(header_line.substr(5));
+    std:string link = info->link();
+    if (link.size() != 0)
+      link = link + ", " + header_line.substr(5);
+    else
+      link = header_line.substr(5);
+    info->SetLink(link);
   } else if (HasPrefix(header_line, "X-SQUID-ERROR:", true)) {
     // Reinterpret host error as proxy error
     if (info->error_code() == kFailHostHttp) {
@@ -1399,6 +1404,70 @@ void DownloadManager::ReleaseCredential(JobInfo *info) {
 }
 
 
+/* Sort links based on the "pri=" parameter */
+static bool sortlinks(std::string s1, std::string s2) {
+  int pri1, pri2;
+  if ((sscanf(s1.c_str(), "%*s; pri=%d", &pri1) == 1) &&
+      (sscanf(s2.c_str(), "%*s; pri=%d", &pri2) == 1))
+    return pri1 < pri2;
+  return true;
+}
+
+/**
+ * Parses Link header and uses it to set a new host chain.
+ * See rfc6249.
+ */
+void DownloadManager::ProcessLink(JobInfo *info) {
+
+  std::vector<std::string> links = SplitString(info->link(), ",");
+  std::sort(links.begin(), links.end(), sortlinks);
+
+  std::vector<std:string> host_list;
+
+  std::vector<std::string>::const_iterator il = links.begin();
+  for (; il != links.end(); ++il) {
+    std::string link = *il;
+    if ((link.find("; rel=duplicate") == std::string::npos) &&
+        (link.find("; rel=\"duplicate\"") == std::string::npos)) {
+      LogCvmfs(kLogDownload, kLogDebug,
+        "skipping link '%s' because it does not contain rel=duplicate",
+        link.c_str());
+      continue;
+    }
+    // ignore depth= field since there's nothing useful we can do with it
+
+    int leftbracket = link.find('<');
+    if (leftbracket == std::string::npos) {
+      LogCvmfs(kLogDownload, kLogDebug,
+        "skipping link '%s' because it does not have a left angle bracket",
+        link.c_str());
+      continue;
+    }
+    int rightbracket = link.find('>');
+    if (rightbracket == std::string::npos) {
+      LogCvmfs(kLogDownload, kLogDebug,
+        "skipping link '%s' because it does not have a right angle bracket",
+        link.c_str());
+      continue;
+    }
+    if (leftbracket > rightbracket) {
+      LogCvmfs(kLogDownload, kLogDebug,
+        "skipping link '%s' because it angle brackets are out of order",
+        link.c_str());
+      continue;
+    }
+    std::string host = link.substr(leftbracket+1, rightbracket-leftbracket-1);
+    LogCvmfs(kLogDownload, kLogDebug, "adding linked host '%s'", host.c_str());
+    host_list.push_back(host);
+  }
+
+  if (host_list.size() > 0) {
+    SetHostChain(host_list);
+    opt_metalink_timestamp_link_ = time(NULL);
+  }
+}
+
+
 /**
  * Checks the result of a curl download and implements the failure logic, such
  * as changing the proxy server.  Takes care of cleanup.
@@ -1412,6 +1481,21 @@ bool DownloadManager::VerifyAndFinalize(const int curl_error, JobInfo *info) {
                            info->proxy().c_str(), curl_error);
   UpdateStatistics(info->curl_handle());
 
+  bool was_metalink;
+  std::string typ;
+  if (info->current_metalink_chain_index() >= 0) {
+    was_metalink = true;
+    typ = "metalink";
+    if (info->link() != "") {
+      // process Link header whether or not the redirected URL got an error
+      ProcessLink(info);
+    }
+  } else {
+    was_metalink = false;
+    typ = "host";
+  }
+
+
   // Verification and error classification
   switch (curl_error) {
     case CURLE_OK:
@@ -1512,18 +1596,12 @@ bool DownloadManager::VerifyAndFinalize(const int curl_error, JobInfo *info) {
       break;
   }
 
-  bool was_metalink;
-  std::string typ;
   std::vector<std::string> *host_chain;
   unsigned char num_used_hosts;
-  if (info->current_metalink_chain_index() >= 0) {
-    was_metalink = true;
-    typ = "metalink";
+  if (was_metalink) {
     host_chain = opt_metalink_.chain;
     num_used_hosts = info->num_used_metalinks();
   } else {
-    was_metalink = false;
-    typ = "host";
     host_chain = opt_host_.chain;
     num_used_hosts = info->num_used_hosts();
   }
diff --git a/cvmfs/network/download.h b/cvmfs/network/download.h
index f83b2019bb..826851a71d 100644
--- a/cvmfs/network/download.h
+++ b/cvmfs/network/download.h
@@ -287,6 +287,7 @@ class DownloadManager {  // NOLINT(clang-analyzer-optin.performance.Padding)
   void Backoff(JobInfo *info);
   void SetNocache(JobInfo *info);
   void SetRegularCache(JobInfo *info);
+  void ProcessLink(JobInfo *info);
   bool VerifyAndFinalize(const int curl_error, JobInfo *info);
   void InitHeaders();
   void CloneProxyConfig(DownloadManager *clone);