fix(push): encode push labels

Fixes #652

core/include/prometheus/labels.h

@@ -8,4 +8,7 @@ namespace prometheus {
 /// \brief Multiple labels and their value.
 using Labels = std::map<std::string, std::string>;
 
+/// \brief Single label and its value.
+using Label = Labels::value_type;
+
 }  // namespace prometheus

push/BUILD.bazel

@@ -24,6 +24,20 @@ cc_library(
     visibility = ["//visibility:public"],
     deps = [
         "//core",
+        "//util",
         "@com_github_curl//:curl",
     ],
 )
+
+cc_library(
+    name = "push_internal_headers",
+    hdrs = glob(
+        ["src/detail/*.h"],
+    ),
+    strip_include_prefix = "src",
+    visibility = ["//push/tests:__subpackages__"],
+    deps = [
+        "//core",
+        "//push",
+    ],
+)

push/CMakeLists.txt

@@ -2,9 +2,12 @@
 find_package(CURL REQUIRED)
 
 add_library(push
-  src/curl_wrapper.cc
-  src/curl_wrapper.h
   src/gateway.cc
+
+  src/detail/curl_wrapper.cc
+  src/detail/curl_wrapper.h
+  src/detail/label_encoder.cc
+  src/detail/label_encoder.h
 )
 
 add_library(${PROJECT_NAME}::push ALIAS push)
@@ -18,6 +21,7 @@ target_link_libraries(push
   PUBLIC
     ${PROJECT_NAME}::core
   PRIVATE
+    ${PROJECT_NAME}::util
     Threads::Threads
     CURL::libcurl
     $<$<AND:$<BOOL:UNIX>,$<NOT:$<BOOL:APPLE>>>:rt>
@@ -78,5 +82,10 @@ if(GENERATE_PKGCONFIG)
 endif()
 
 if(ENABLE_TESTING)
+  add_library(push_internal_headers INTERFACE)
+  add_library(${PROJECT_NAME}::push_internal_headers ALIAS push_internal_headers)
+  target_include_directories(push_internal_headers INTERFACE src)
+  target_link_libraries(push_internal_headers INTERFACE ${PROJECT_NAME}::push)
+
   add_subdirectory(tests)
 endif()

push/src/detail/label_encoder.cc

@@ -0,0 +1,42 @@
+#include "label_encoder.h"
+
+#include <algorithm>
+#include <ostream>
+
+#include "prometheus/detail/base64.h"
+
+namespace prometheus {
+namespace detail {
+
+namespace {
+// Does this character need encoding like in RFC 3986 section 2.3?
+bool needsEncoding(char c) {
+  if (c >= 'a' && c <= 'z') {
+    return false;
+  }
+  if (c >= 'A' && c <= 'Z') {
+    return false;
+  }
+  if (c >= '0' && c <= '9') {
+    return false;
+  }
+  if (c == '-' || c == '.' || c == '_' || c == '~') {
+    return false;
+  }
+  return true;
+}
+}  // namespace
+
+void encodeLabel(std::ostream& os, const Label& label) {
+  if (label.second.empty()) {
+    os << "/" << label.first << "@base64/=";
+  } else if (std::any_of(label.second.begin(), label.second.end(),
+                         needsEncoding)) {
+    os << "/" << label.first << "@base64/"
+       << detail::base64url_encode(label.second);
+  } else {
+    os << "/" << label.first << "/" << label.second;
+  }
+}
+}  // namespace detail
+}  // namespace prometheus

push/src/detail/label_encoder.h

@@ -0,0 +1,13 @@
+#pragma once
+
+#include <iosfwd>
+
+#include "prometheus/labels.h"
+
+namespace prometheus {
+namespace detail {
+
+void encodeLabel(std::ostream& os, const Label& label);
+
+}
+}  // namespace prometheus

push/src/gateway.cc

@@ -8,7 +8,8 @@
 #include <mutex>
 #include <sstream>
 
-#include "curl_wrapper.h"
+#include "detail/curl_wrapper.h"
+#include "detail/label_encoder.h"
 #include "prometheus/detail/future_std.h"
 #include "prometheus/metric_family.h"  // IWYU pragma: keep
 #include "prometheus/text_serializer.h"
@@ -24,12 +25,13 @@ Gateway::Gateway(const std::string& host, const std::string& port,
   curlWrapper_ = detail::make_unique<detail::CurlWrapper>(username, password);
 
   std::stringstream jobUriStream;
-  jobUriStream << host << ':' << port << "/metrics/job/" << jobname;
+  jobUriStream << host << ':' << port << "/metrics";
+  detail::encodeLabel(jobUriStream, {"job", jobname});
   jobUri_ = jobUriStream.str();
 
   std::stringstream labelStream;
   for (auto& label : labels) {
-    labelStream << "/" << label.first << "/" << label.second;
+    detail::encodeLabel(labelStream, label);
   }
   labels_ = labelStream.str();
 }

push/tests/CMakeLists.txt

@@ -1 +1,2 @@
 add_subdirectory(integration)
+add_subdirectory(internal)

push/tests/internal/BUILD.bazel

@@ -0,0 +1,13 @@
+cc_test(
+    name = "internal",
+    srcs = glob([
+        "*.cc",
+        "*.h",
+    ]),
+    copts = ["-Iexternal/googletest/include"],
+    linkstatic = True,
+    deps = [
+        "//push:push_internal_headers",
+        "@com_google_googletest//:gtest_main",
+    ],
+)

push/tests/internal/CMakeLists.txt

@@ -0,0 +1,14 @@
+add_executable(prometheus_push_internal_test
+  label_encoder_test.cc
+)
+
+target_link_libraries(prometheus_push_internal_test
+  PRIVATE
+    ${PROJECT_NAME}::push_internal_headers
+    GTest::gmock_main
+)
+
+add_test(
+  NAME prometheus_push_internal_test
+  COMMAND prometheus_push_internal_test
+)

push/tests/internal/label_encoder_test.cc

@@ -0,0 +1,43 @@
+#include "detail/label_encoder.h"
+
+#include <gtest/gtest.h>
+
+#include <string>
+#include <sstream>
+
+namespace prometheus {
+namespace {
+
+class LabelEncoderTest : public testing::Test {
+ protected:
+  std::string Encode(const Label& label) {
+    std::stringstream ss;
+    detail::encodeLabel(ss, label);
+    return ss.str();
+  }
+};
+
+// test cases taken from https://github.com/prometheus/pushgateway#url
+
+TEST_F(LabelEncoderTest, regular) {
+  EXPECT_EQ("/foo/bar", Encode(Label{"foo", "bar"}));
+}
+
+TEST_F(LabelEncoderTest, empty) {
+  EXPECT_EQ("/first_label@base64/=", Encode(Label{"first_label", ""}));
+}
+
+TEST_F(LabelEncoderTest, path) {
+  EXPECT_EQ("/path@base64/L3Zhci90bXA=", Encode(Label{"path", "/var/tmp"}));
+}
+
+TEST_F(LabelEncoderTest, unicode) {
+  const char unicodeText[] =
+      "\xce\xa0\xcf\x81\xce\xbf\xce\xbc\xce\xb7\xce\xb8\xce\xb5\xcf\x8d\xcf"
+      "\x82";  // Προμηθεύς
+  EXPECT_EQ("/name@base64/zqDPgc6_zrzOt864zrXPjc-C",
+            Encode(Label{"name", unicodeText}));
+}
+
+}  // namespace
+}  // namespace prometheus

util/include/prometheus/detail/base64.h

@@ -1,5 +1,6 @@
 #pragma once
 
+#include <algorithm>
 #include <cstddef>
 #include <cstdint>
 #include <stdexcept>
@@ -66,6 +67,17 @@ inline std::string base64_encode(const std::string& input) {
   return encoded;
 }
 
+// https://tools.ietf.org/html/rfc4648#section-5
+inline std::string base64url_encode(const std::string& input) {
+  std::string s = base64_encode(input);
+  std::transform(begin(s), end(s), begin(s), [](char c) {
+    if (c == '+') return '-';
+    if (c == '/') return '_';
+    return c;
+  });
+  return s;
+}
+
 inline std::string base64_decode(const std::string& input) {
   if (input.length() % 4) {
     throw std::runtime_error("Invalid base64 length!");

util/tests/unit/base64_test.cc

@@ -47,6 +47,14 @@ TEST(Base64Test, encodeTest) {
   }
 }
 
+TEST(Base64Test, encodeUrlTest) {
+  const char unicodeText[] =
+      "\xce\xa0\xcf\x81\xce\xbf\xce\xbc\xce\xb7\xce\xb8\xce\xb5\xcf\x8d\xcf"
+      "\x82";  // Προμηθεύς
+  std::string encoded = detail::base64url_encode(unicodeText);
+  EXPECT_EQ("zqDPgc6_zrzOt864zrXPjc-C", encoded);
+}
+
 TEST(Base64Test, decodeTest) {
   for (const auto& test_case : testVector) {
     std::string decoded = detail::base64_decode(test_case.encoded);