Use prometheus text format as LabelsKey representation

src/metric.cpp

@@ -134,15 +134,16 @@ static std::string_view arena_copy_string(std::string_view str,
 }
 
 // Arena-based labels key for second level of map
-// Uses string_view to point to arena-allocated strings
+// Uses string_view containing labels in Prometheus text format
 struct LabelsKey {
-  ArenaVector<std::pair<std::string_view, std::string_view>> labels;
+  std::string_view prometheus_format;
 
-  // Arena-owning constructor (copies strings into arena)
+  // Arena-owning constructor (copies strings into arena and formats as
+  // Prometheus text)
   LabelsKey(std::span<const std::pair<std::string_view, std::string_view>> l,
-            ArenaAllocator &arena)
-      : labels(&arena) {
-    // Copy and validate all label keys and values into arena
+            ArenaAllocator &arena) {
+    // Copy and validate all label keys and values, sort by key
+    ArenaVector<std::pair<std::string_view, std::string_view>> labels(&arena);
     for (const auto &[key, value] : l) {
       validate_or_abort(is_valid_label_key(key), "invalid label key", key);
       validate_or_abort(is_valid_label_value(value), "invalid label value",
@@ -156,33 +157,67 @@ struct LabelsKey {
     // Sort labels by key for Prometheus compatibility
     std::sort(labels.data(), labels.data() + labels.size(),
               [](const auto &a, const auto &b) { return a.first < b.first; });
+
+    // Generate Prometheus text format: {key1="value1",key2="value2"}
+    if (labels.empty()) {
+      prometheus_format = "";
+    } else {
+      // Calculate required size for formatted string
+      size_t required_size = 2; // {}
+      for (const auto &[key, value] : labels) {
+        required_size += key.length() + 3 + value.length(); // key="value"
+        for (char c : value) {
+          if (c == '\\' || c == '"' || c == '\n') {
+            required_size++;
+          }
+        }
+      }
+      required_size += labels.size() - 1; // commas
+
+      // Generate formatted string in arena
+      char *buf = arena.allocate<char>(required_size);
+      char *p = buf;
+
+      *p++ = '{';
+      for (size_t i = 0; i < labels.size(); ++i) {
+        if (i > 0)
+          *p++ = ',';
+        std::memcpy(p, labels[i].first.data(), labels[i].first.length());
+        p += labels[i].first.length();
+        *p++ = '=';
+        *p++ = '"';
+        for (char c : labels[i].second) {
+          switch (c) {
+          case '\\':
+            *p++ = '\\';
+            *p++ = '\\';
+            break;
+          case '"':
+            *p++ = '\\';
+            *p++ = '"';
+            break;
+          case '\n':
+            *p++ = '\\';
+            *p++ = 'n';
+            break;
+          default:
+            *p++ = c;
+            break;
+          }
+        }
+        *p++ = '"';
+      }
+      *p++ = '}';
+      prometheus_format = std::string_view(buf, p - buf);
+    }
   }
 
   bool operator==(const LabelsKey &other) const {
-    if (labels.size() != other.labels.size())
-      return false;
-    for (size_t i = 0; i < labels.size(); ++i) {
-      if (labels[i].first != other.labels[i].first ||
-          labels[i].second != other.labels[i].second) {
-        return false;
-      }
-    }
-    return true;
+    return prometheus_format == other.prometheus_format;
   }
 
   bool operator<(const LabelsKey &other) const {
-    if (labels.size() != other.labels.size()) {
-      return labels.size() < other.labels.size();
-    }
-    for (size_t i = 0; i < labels.size(); ++i) {
-      if (labels[i].first != other.labels[i].first) {
-        return labels[i].first < other.labels[i].first;
-      }
-      if (labels[i].second != other.labels[i].second) {
-        return labels[i].second < other.labels[i].second;
-      }
-    }
-    return false; // They are equal
+    return prometheus_format < other.prometheus_format;
   }
 };
 
@@ -191,16 +226,7 @@ struct LabelsKey {
 namespace std {
 template <> struct hash<metric::LabelsKey> {
   std::size_t operator()(const metric::LabelsKey &k) const {
-    std::size_t hash_value = 0;
-    for (size_t i = 0; i < k.labels.size(); ++i) {
-      const auto &[key, value] = k.labels[i];
-      // Combine hashes using a simple but effective method
-      hash_value ^= std::hash<std::string_view>{}(key) + 0x9e3779b9 +
-                    (hash_value << 6) + (hash_value >> 2);
-      hash_value ^= std::hash<std::string_view>{}(value) + 0x9e3779b9 +
-                    (hash_value << 6) + (hash_value >> 2);
-    }
-    return hash_value;
+    return std::hash<std::string_view>{}(k.prometheus_format);
   }
 };
 } // namespace std
@@ -486,10 +512,10 @@ struct Metric {
     const LabelsKey &key = intern_labels(labels);
 
     // Validate that labels aren't already registered as callback
-    validate_or_abort(family->p->callbacks.find(key) ==
-                          family->p->callbacks.end(),
+    validate_or_abort(
+        family->p->callbacks.find(key) == family->p->callbacks.end(),
         "labels already registered as callback",
-                      key.labels.empty() ? "(no labels)" : key.labels[0].first);
+        key.prometheus_format.empty() ? "(no labels)" : key.prometheus_format);
 
     // Ensure thread state exists
     auto thread_id = std::this_thread::get_id();
@@ -519,10 +545,10 @@ struct Metric {
     const LabelsKey &key = intern_labels(labels);
 
     // Validate that labels aren't already registered as callback
-    validate_or_abort(family->p->callbacks.find(key) ==
-                          family->p->callbacks.end(),
+    validate_or_abort(
+        family->p->callbacks.find(key) == family->p->callbacks.end(),
         "labels already registered as callback",
-                      key.labels.empty() ? "(no labels)" : key.labels[0].first);
+        key.prometheus_format.empty() ? "(no labels)" : key.prometheus_format);
 
     auto &ptr = family->p->instances[key];
     if (!ptr) {
@@ -1210,38 +1236,29 @@ std::span<std::string_view> render(ArenaAllocator &arena) {
 
   ArenaVector<std::string_view> output(&arena);
 
-  auto format_labels =
-      [&](const ArenaVector<std::pair<std::string_view, std::string_view>>
-              &labels) -> std::string_view {
-    if (labels.empty()) {
-      return "";
-    }
-
-    size_t required_size = 2; // {}
-    for (const auto &[key, value] : labels) {
-      required_size += key.length() + 3 + value.length(); // key="value"
+  // Helper function to append an additional label to existing Prometheus format
+  auto append_label_to_format =
+      [&](std::string_view base_format, std::string_view key,
+          std::string_view value) -> std::string_view {
+    // Calculate size for key="value" with escaping
+    size_t key_value_size = key.length() + 3 + value.length(); // key="value"
     for (char c : value) {
       if (c == '\\' || c == '"' || c == '\n') {
-          required_size++;
+        key_value_size++;
       }
     }
-    }
-    if (!labels.empty()) {
-      required_size += labels.size() - 1; // commas
-    }
 
+    if (base_format.empty()) {
+      // Create new format: {key="value"}
+      size_t required_size = 2 + key_value_size; // {}
       char *buf = arena.allocate<char>(required_size);
       char *p = buf;
-
       *p++ = '{';
-    for (size_t i = 0; i < labels.size(); ++i) {
-      if (i > 0)
-        *p++ = ',';
-      std::memcpy(p, labels[i].first.data(), labels[i].first.length());
-      p += labels[i].first.length();
+      std::memcpy(p, key.data(), key.length());
+      p += key.length();
       *p++ = '=';
       *p++ = '"';
-      for (char c : labels[i].second) {
+      for (char c : value) {
         switch (c) {
         case '\\':
           *p++ = '\\';
@@ -1261,9 +1278,45 @@ std::span<std::string_view> render(ArenaAllocator &arena) {
         }
       }
       *p++ = '"';
-    }
       *p++ = '}';
       return std::string_view(buf, p - buf);
+    } else {
+      // Append to existing format: {existing,key="value"}
+      size_t required_size = base_format.length() + 1 +
+                             key_value_size; // comma + key="value", replace }
+      char *buf = arena.allocate<char>(required_size);
+      char *p = buf;
+      // Copy everything except the closing }
+      std::memcpy(p, base_format.data(), base_format.length() - 1);
+      p += base_format.length() - 1;
+      *p++ = ',';
+      std::memcpy(p, key.data(), key.length());
+      p += key.length();
+      *p++ = '=';
+      *p++ = '"';
+      for (char c : value) {
+        switch (c) {
+        case '\\':
+          *p++ = '\\';
+          *p++ = '\\';
+          break;
+        case '"':
+          *p++ = '\\';
+          *p++ = '"';
+          break;
+        case '\n':
+          *p++ = '\\';
+          *p++ = 'n';
+          break;
+        default:
+          *p++ = c;
+          break;
+        }
+      }
+      *p++ = '"';
+      *p++ = '}';
+      return std::string_view(buf, p - buf);
+    }
   };
 
   // Format counters - ITERATION ORDER MUST MATCH COMPUTE PHASE
@@ -1276,21 +1329,13 @@ std::span<std::string_view> render(ArenaAllocator &arena) {
     output.push_back(format(arena, "# TYPE %.*s counter\n",
                             static_cast<int>(name.length()), name.data()));
 
-    ArenaVector<std::pair<std::string_view, std::string_view>> labels_sv(
-        &arena);
-
     // Format callback values
     for (const auto &[labels_key, callback] : family->callbacks) {
       auto value = next_value++->as_double;
-      labels_sv.clear();
-      for (size_t i = 0; i < labels_key.labels.size(); ++i) {
-        labels_sv.push_back(labels_key.labels[i]);
-      }
-      auto labels = format_labels(labels_sv);
-      output.push_back(format(arena, "%.*s%.*s %.17g\n",
-                              static_cast<int>(name.length()), name.data(),
-                              static_cast<int>(labels.length()), labels.data(),
-                              value));
+      output.push_back(format(
+          arena, "%.*s%.*s %.17g\n", static_cast<int>(name.length()),
+          name.data(), static_cast<int>(labels_key.prometheus_format.length()),
+          labels_key.prometheus_format.data(), value));
     }
 
     // Use pre-computed data (same as compute phase)
@@ -1299,15 +1344,11 @@ std::span<std::string_view> render(ArenaAllocator &arena) {
     // Format counter values using pre-computed values
     for (const auto &data : family_data) {
       auto total_value = next_value++->as_double;
-      labels_sv.clear();
-      for (size_t i = 0; i < data.labels_key.labels.size(); ++i) {
-        labels_sv.push_back(data.labels_key.labels[i]);
-      }
-      auto labels = format_labels(labels_sv);
-      output.push_back(format(arena, "%.*s%.*s %.17g\n",
-                              static_cast<int>(name.length()), name.data(),
-                              static_cast<int>(labels.length()), labels.data(),
-                              total_value));
+      output.push_back(
+          format(arena, "%.*s%.*s %.17g\n", static_cast<int>(name.length()),
+                 name.data(),
+                 static_cast<int>(data.labels_key.prometheus_format.length()),
+                 data.labels_key.prometheus_format.data(), total_value));
     }
   }
 
@@ -1321,36 +1362,24 @@ std::span<std::string_view> render(ArenaAllocator &arena) {
     output.push_back(format(arena, "# TYPE %.*s gauge\n",
                             static_cast<int>(name.length()), name.data()));
 
-    ArenaVector<std::pair<std::string_view, std::string_view>> labels_sv(
-        &arena);
-
     // Format callback values
     for (const auto &[labels_key, callback] : family->callbacks) {
       auto value = next_value++->as_double;
-      labels_sv.clear();
-      for (size_t i = 0; i < labels_key.labels.size(); ++i) {
-        labels_sv.push_back(labels_key.labels[i]);
-      }
-      auto labels = format_labels(labels_sv);
-      output.push_back(format(arena, "%.*s%.*s %.17g\n",
-                              static_cast<int>(name.length()), name.data(),
-                              static_cast<int>(labels.length()), labels.data(),
-                              value));
+      output.push_back(format(
+          arena, "%.*s%.*s %.17g\n", static_cast<int>(name.length()),
+          name.data(), static_cast<int>(labels_key.prometheus_format.length()),
+          labels_key.prometheus_format.data(), value));
     }
 
     // Use pre-computed data (same as compute phase)
     const auto &family_data = label_sets.gauge_data[gauge_family_idx++];
     for (const auto &data : family_data) {
       auto value = next_value++->as_double;
-      labels_sv.clear();
-      for (size_t i = 0; i < data.labels_key.labels.size(); ++i) {
-        labels_sv.push_back(data.labels_key.labels[i]);
-      }
-      auto labels = format_labels(labels_sv);
-      output.push_back(format(arena, "%.*s%.*s %.17g\n",
-                              static_cast<int>(name.length()), name.data(),
-                              static_cast<int>(labels.length()), labels.data(),
-                              value));
+      output.push_back(
+          format(arena, "%.*s%.*s %.17g\n", static_cast<int>(name.length()),
+                 name.data(),
+                 static_cast<int>(data.labels_key.prometheus_format.length()),
+                 data.labels_key.prometheus_format.data(), value));
     }
   }
 
@@ -1367,9 +1396,6 @@ std::span<std::string_view> render(ArenaAllocator &arena) {
     // Use pre-computed data (same as compute phase)
     const auto &family_data = label_sets.histogram_data[histogram_family_idx++];
 
-    ArenaVector<std::pair<std::string_view, std::string_view>> bucket_labels_sv(
-        &arena);
-
     // Format histogram data using pre-computed values
     for (const auto &data : family_data) {
       // Get bucket count from pre-computed data
@@ -1378,13 +1404,9 @@ std::span<std::string_view> render(ArenaAllocator &arena) {
       // Format explicit bucket counts
       for (size_t i = 0; i < bucket_count; ++i) {
         auto count = next_value++->as_uint64;
-        bucket_labels_sv.clear();
-        for (size_t j = 0; j < data.labels_key.labels.size(); ++j) {
-          bucket_labels_sv.push_back(data.labels_key.labels[j]);
-        }
-        bucket_labels_sv.push_back(
-            {"le", static_format(arena, family->buckets[i])});
-        auto labels = format_labels(bucket_labels_sv);
+        auto bucket_value = static_format(arena, family->buckets[i]);
+        auto labels = append_label_to_format(data.labels_key.prometheus_format,
+                                             "le", bucket_value);
         output.push_back(format(
             arena, "%.*s_bucket%.*s %llu\n", static_cast<int>(name.length()),
             name.data(), static_cast<int>(labels.length()), labels.data(),
@@ -1393,12 +1415,8 @@ std::span<std::string_view> render(ArenaAllocator &arena) {
 
       // Format +Inf bucket
       auto observations = next_value++->as_uint64;
-      bucket_labels_sv.clear();
-      for (size_t j = 0; j < data.labels_key.labels.size(); ++j) {
-        bucket_labels_sv.push_back(data.labels_key.labels[j]);
-      }
-      bucket_labels_sv.push_back({"le", "+Inf"});
-      auto inf_labels = format_labels(bucket_labels_sv);
+      auto inf_labels = append_label_to_format(
+          data.labels_key.prometheus_format, "le", "+Inf");
       output.push_back(format(
           arena, "%.*s_bucket%.*s %llu\n", static_cast<int>(name.length()),
           name.data(), static_cast<int>(inf_labels.length()), inf_labels.data(),
@@ -1406,20 +1424,19 @@ std::span<std::string_view> render(ArenaAllocator &arena) {
 
       // Format sum
       auto sum = next_value++->as_double;
-      bucket_labels_sv.clear();
-      for (size_t j = 0; j < data.labels_key.labels.size(); ++j) {
-        bucket_labels_sv.push_back(data.labels_key.labels[j]);
-      }
-      auto labels = format_labels(bucket_labels_sv);
-      output.push_back(format(
-          arena, "%.*s_sum%.*s %.17g\n", static_cast<int>(name.length()),
-          name.data(), static_cast<int>(labels.length()), labels.data(), sum));
+      output.push_back(
+          format(arena, "%.*s_sum%.*s %.17g\n", static_cast<int>(name.length()),
+                 name.data(),
+                 static_cast<int>(data.labels_key.prometheus_format.length()),
+                 data.labels_key.prometheus_format.data(), sum));
 
       // Format count
       auto count = next_value++->as_uint64;
-      output.push_back(format(arena, "%.*s_count%.*s %llu\n",
+      output.push_back(
+          format(arena, "%.*s_count%.*s %llu\n",
                  static_cast<int>(name.length()), name.data(),
-                              static_cast<int>(labels.length()), labels.data(),
+                 static_cast<int>(data.labels_key.prometheus_format.length()),
+                 data.labels_key.prometheus_format.data(),
                  static_cast<unsigned long long>(count)));
     }
   }
@@ -1437,16 +1454,17 @@ void Family<Counter>::register_callback(
 
   // Validate that labels aren't already in use by create() calls
   for (const auto &[thread_id, per_thread] : p->per_thread_state) {
-    validate_or_abort(per_thread.instances.find(key) ==
-                          per_thread.instances.end(),
+    validate_or_abort(
+        per_thread.instances.find(key) == per_thread.instances.end(),
         "labels already registered as static instance",
-                      key.labels.empty() ? "(no labels)" : key.labels[0].first);
+        key.prometheus_format.empty() ? "(no labels)" : key.prometheus_format);
   }
 
   // Validate that callback isn't already registered for these labels
   validate_or_abort(p->callbacks.find(key) == p->callbacks.end(),
                     "callback already registered for labels",
-                    key.labels.empty() ? "(no labels)" : key.labels[0].first);
+                    key.prometheus_format.empty() ? "(no labels)"
+                                                  : key.prometheus_format);
 
   p->callbacks[std::move(key)] = std::move(callback);
 }
@@ -1461,12 +1479,14 @@ void Family<Gauge>::register_callback(
   // Validate that labels aren't already in use by create() calls
   validate_or_abort(p->instances.find(key) == p->instances.end(),
                     "labels already registered as static instance",
-                    key.labels.empty() ? "(no labels)" : key.labels[0].first);
+                    key.prometheus_format.empty() ? "(no labels)"
+                                                  : key.prometheus_format);
 
   // Validate that callback isn't already registered for these labels
   validate_or_abort(p->callbacks.find(key) == p->callbacks.end(),
                     "callback already registered for labels",
-                    key.labels.empty() ? "(no labels)" : key.labels[0].first);
+                    key.prometheus_format.empty() ? "(no labels)"
+                                                  : key.prometheus_format);
 
   p->callbacks[std::move(key)] = std::move(callback);
 }