Copilot commented on code in PR #2265:
URL: https://github.com/apache/nifi-minifi-cpp/pull/2265#discussion_r4105568660


##########
extensions/opc/src/FetchOPCHistory.cpp:
##########
@@ -0,0 +1,400 @@
+/**
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "FetchOPCHistory.h"
+
+#include <optional>
+#include <string>
+#include <vector>
+
+#include "core/ProcessSession.h"
+#include "core/Resource.h"
+#include "minifi-cpp/core/ProcessContext.h"
+#include "utils/ProcessorConfigUtils.h"
+#include "utils/StringUtils.h"
+
+namespace org::apache::nifi::minifi::processors {
+
+namespace {
+
+constexpr const char* LAST_FETCHED_TIMESTAMP_KEY = "last_fetched_timestamp";
+constexpr const char* LAST_FETCHED_FINGERPRINT_KEY = 
"last_fetched_fingerprint";
+
+std::string updateTypeToString(UA_HistoryUpdateType type) {
+  switch (type) {
+    case UA_HISTORYUPDATETYPE_INSERT:
+      return "Insert";
+    case UA_HISTORYUPDATETYPE_REPLACE:
+      return "Replace";
+    case UA_HISTORYUPDATETYPE_UPDATE:
+      return "Update";
+    case UA_HISTORYUPDATETYPE_DELETE:
+      return "Delete";
+    default:
+      return "Unknown";
+  }
+}
+
+std::string uaStringToString(const UA_String& str) {
+  return {reinterpret_cast<const char*>(str.data), str.length};
+}
+
+struct HistoryEntry {
+  std::string value;
+  int64_t source_timestamp = 0;
+  const UA_ModificationInfo* modification_info = nullptr;
+
+  [[nodiscard]] int64_t modificationTime() const {
+    return modification_info ? modification_info->modificationTime : 
UA_DateTime_fromUnixTime(0);
+  }
+};
+
+struct HistoryBatch {
+  std::vector<HistoryEntry> entries;
+  bool has_modification_info = false;
+};
+
+std::string entryFingerprint(const HistoryEntry& entry, bool 
has_modification_info) {
+  // The fingerprint deduplicates entries sharing the boundary source 
timestamp across triggers. For raw value history a
+  // duplicated (value, source timestamp) pair represents no change in the 
history, so losing one to deduplication is harmless.
+  // For audit (modification) history the modification time and update type 
distinguish otherwise-identical entries, so they
+  // are included to make the fingerprint more unique and reduce the chance of 
dropping a distinct modification.
+  std::string raw = ":" + entry.value;
+  if (has_modification_info) {
+    const auto update_type = entry.modification_info ? 
updateTypeToString(entry.modification_info->updateType) : "";
+    raw = std::to_string(entry.modificationTime()) + ":" + update_type + raw;
+  }

Review Comment:
   The audit fingerprint omits `ModificationUsername`, even though it is part 
of the emitted audit record. If a batch boundary splits two entries with the 
same source timestamp, value, modification time, and update type but different 
users, the second entry matches the persisted fingerprint and is silently 
dropped on the next trigger.
   
   This issue also appears on line 373 of the same file.



##########
extensions/opc/src/FetchOPCHistory.cpp:
##########
@@ -0,0 +1,400 @@
+/**
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "FetchOPCHistory.h"
+
+#include <optional>
+#include <string>
+#include <vector>
+
+#include "core/ProcessSession.h"
+#include "core/Resource.h"
+#include "minifi-cpp/core/ProcessContext.h"
+#include "utils/ProcessorConfigUtils.h"
+#include "utils/StringUtils.h"
+
+namespace org::apache::nifi::minifi::processors {
+
+namespace {
+
+constexpr const char* LAST_FETCHED_TIMESTAMP_KEY = "last_fetched_timestamp";
+constexpr const char* LAST_FETCHED_FINGERPRINT_KEY = 
"last_fetched_fingerprint";
+
+std::string updateTypeToString(UA_HistoryUpdateType type) {
+  switch (type) {
+    case UA_HISTORYUPDATETYPE_INSERT:
+      return "Insert";
+    case UA_HISTORYUPDATETYPE_REPLACE:
+      return "Replace";
+    case UA_HISTORYUPDATETYPE_UPDATE:
+      return "Update";
+    case UA_HISTORYUPDATETYPE_DELETE:
+      return "Delete";
+    default:
+      return "Unknown";
+  }
+}
+
+std::string uaStringToString(const UA_String& str) {
+  return {reinterpret_cast<const char*>(str.data), str.length};
+}
+
+struct HistoryEntry {
+  std::string value;
+  int64_t source_timestamp = 0;
+  const UA_ModificationInfo* modification_info = nullptr;
+
+  [[nodiscard]] int64_t modificationTime() const {
+    return modification_info ? modification_info->modificationTime : 
UA_DateTime_fromUnixTime(0);
+  }
+};
+
+struct HistoryBatch {
+  std::vector<HistoryEntry> entries;
+  bool has_modification_info = false;
+};
+
+std::string entryFingerprint(const HistoryEntry& entry, bool 
has_modification_info) {
+  // The fingerprint deduplicates entries sharing the boundary source 
timestamp across triggers. For raw value history a
+  // duplicated (value, source timestamp) pair represents no change in the 
history, so losing one to deduplication is harmless.
+  // For audit (modification) history the modification time and update type 
distinguish otherwise-identical entries, so they
+  // are included to make the fingerprint more unique and reduce the chance of 
dropping a distinct modification.
+  std::string raw = ":" + entry.value;
+  if (has_modification_info) {
+    const auto update_type = entry.modification_info ? 
updateTypeToString(entry.modification_info->updateType) : "";
+    raw = std::to_string(entry.modificationTime()) + ":" + update_type + raw;
+  }
+  return utils::string::to_hex(raw);
+}
+
+// NOLINTBEGIN(cppcoreguidelines-pro-type-union-access)
+std::optional<HistoryBatch> extractHistoryBatch(const UA_ExtensionObject* 
data, opc::BinaryEncoding binary_encoding,
+    const std::shared_ptr<core::logging::Logger>& logger) {
+  const UA_DataValue* data_values = nullptr;
+  size_t data_value_size = 0;
+  const UA_ModificationInfo* modification_infos = nullptr;
+  size_t modification_infos_size = 0;
+
+  if (data->content.decoded.type == &UA_TYPES[UA_TYPES_HISTORYDATA]) {
+    const auto* history_data = static_cast<const 
UA_HistoryData*>(data->content.decoded.data);
+    data_values = history_data->dataValues;
+    data_value_size = history_data->dataValuesSize;
+  } else if (data->content.decoded.type == 
&UA_TYPES[UA_TYPES_HISTORYMODIFIEDDATA]) {
+    const auto* modified_data = static_cast<const 
UA_HistoryModifiedData*>(data->content.decoded.data);
+    data_values = modified_data->dataValues;
+    data_value_size = modified_data->dataValuesSize;
+    modification_infos = modified_data->modificationInfos;
+    modification_infos_size = modified_data->modificationInfosSize;
+  } else {
+    logger->log_error("Unexpected data type received in the history read 
callback: {}", data->content.decoded.type->typeName);
+    return std::nullopt;
+  }
+
+  HistoryBatch batch;
+  batch.has_modification_info = modification_infos != nullptr;
+  batch.entries.reserve(data_value_size);
+  for (size_t i = 0; i < data_value_size; ++i) {
+    HistoryEntry entry;
+    try {
+      entry.value = opc::variantToString(data_values[i].value, 
binary_encoding);
+    } catch (const opc::OPCException& ex) {
+      logger->log_warn("Failed to convert value at index {} to string, 
skipping entry: {}", i, ex.what());
+      continue;
+    }
+    entry.source_timestamp = data_values[i].sourceTimestamp;
+    entry.modification_info = (modification_infos && i < 
modification_infos_size) ? &modification_infos[i] : nullptr;
+    batch.entries.push_back(std::move(entry));
+  }
+  return batch;
+}
+// NOLINTEND(cppcoreguidelines-pro-type-union-access)
+
+std::vector<HistoryEntry> selectNewEntries(std::vector<HistoryEntry> entries, 
const std::optional<FetchedState>& last_fetched,
+    bool has_modification_info, std::optional<size_t> max_entries) {
+  std::vector<HistoryEntry> new_entries;
+  new_entries.reserve(max_entries ? std::min(*max_entries, entries.size()) : 
entries.size());
+  for (auto& entry : entries) {
+    if (max_entries && new_entries.size() >= *max_entries) {
+      break;
+    }
+    if (last_fetched && entry.source_timestamp == last_fetched->timestamp &&
+        last_fetched->fingerprints.contains(entryFingerprint(entry, 
has_modification_info))) {
+      continue;
+    }
+    new_entries.push_back(std::move(entry));
+  }
+  return new_entries;
+}
+
+void addModificationInfo(core::Record& record, const UA_ModificationInfo& 
modification_info) {
+  if (modification_info.userName.length > 0) {
+    record.emplace("ModificationUsername", 
core::RecordField(uaStringToString(modification_info.userName)));
+  }
+  record.emplace("ModificationTime", 
core::RecordField(opc::OPCDateTime2String(modification_info.modificationTime)));
+  record.emplace("ModificationUpdateType", 
core::RecordField(updateTypeToString(modification_info.updateType)));
+}
+
+void addModificationInfo(core::FlowFile& flow_file, const UA_ModificationInfo& 
modification_info) {
+  if (modification_info.userName.length > 0) {
+    flow_file.addAttribute("ModificationUsername", 
uaStringToString(modification_info.userName));
+  }
+  flow_file.addAttribute("ModificationTime", 
opc::OPCDateTime2String(modification_info.modificationTime));
+  flow_file.addAttribute("ModificationUpdateType", 
updateTypeToString(modification_info.updateType));
+}
+
+core::Record toRecord(const std::string& node_id, const int32_t 
namespace_index, const HistoryEntry& entry) {
+  core::Record record;
+  record.emplace("Value", core::RecordField(entry.value));
+  record.emplace("NodeID", core::RecordField(node_id));
+  record.emplace("Namespace index", 
core::RecordField(std::to_string(namespace_index)));
+  record.emplace("Sourcetimestamp", 
core::RecordField(opc::OPCDateTime2String(entry.source_timestamp)));
+  if (entry.modification_info) {
+    addModificationInfo(record, *entry.modification_info);
+  }
+  return record;
+}
+
+void writeAsRecordSet(FetchOPCHistoryContext& context, const 
std::vector<HistoryEntry>& entries) {
+  core::RecordSet record_set;
+  for (const auto& entry : entries) {
+    record_set.push_back(toRecord(context.node_id, context.namespace_index, 
entry));
+  }
+
+  auto flow_file = context.session.create();
+  context.record_set_writer->write(record_set, flow_file, context.session);
+  context.session.transfer(flow_file, FetchOPCHistory::Success);
+  context.entries_transferred += entries.size();
+}
+
+void writeAsFlowFiles(FetchOPCHistoryContext& context, const 
std::vector<HistoryEntry>& entries) {
+  for (const auto& entry : entries) {
+    auto flow_file = context.session.create();
+    context.session.write(flow_file, [&entry](const 
std::shared_ptr<io::OutputStream>& output_stream) -> io::IoResult {
+      output_stream->write(reinterpret_cast<const 
uint8_t*>(entry.value.data()), entry.value.size());
+      return io::IoResult::from(entry.value.size());
+    });
+    flow_file->addAttribute("NodeID", context.node_id);
+    flow_file->addAttribute("Namespace index", 
std::to_string(context.namespace_index));
+    flow_file->addAttribute("Sourcetimestamp", 
opc::OPCDateTime2String(entry.source_timestamp));
+    if (entry.modification_info) {
+      addModificationInfo(*flow_file, *entry.modification_info);
+    }
+    context.session.transfer(flow_file, FetchOPCHistory::Success);
+    ++context.entries_transferred;
+  }
+}
+
+void updateState(std::unordered_map<std::string, std::string>& state_map, 
const std::vector<HistoryEntry>& new_entries, bool has_modification_info) {
+  const int64_t new_timestamp = new_entries.back().source_timestamp;
+  const auto new_timestamp_str = std::to_string(new_timestamp);
+
+  auto& stored_timestamp = state_map[LAST_FETCHED_TIMESTAMP_KEY];
+  auto& fingerprints = state_map[LAST_FETCHED_FINGERPRINT_KEY];
+  if (stored_timestamp != new_timestamp_str) {
+    stored_timestamp = new_timestamp_str;
+    fingerprints.clear();
+  }
+
+  for (const auto& entry : new_entries) {
+    if (entry.source_timestamp == new_timestamp) {
+      if (!fingerprints.empty()) {
+        fingerprints += ",";
+      }
+      fingerprints += entryFingerprint(entry, has_modification_info);
+    }
+  }
+}
+
+UA_Boolean historyReadCallback(UA_Client* /*client*/, const UA_NodeId* 
/*node_id*/, UA_Boolean more_data_available, const UA_ExtensionObject* data,
+    void* ctx) {
+  auto* opc_history_context = static_cast<FetchOPCHistoryContext*>(ctx);
+
+  const auto binary_encoding = opc_history_context->record_set_writer ? 
opc::BinaryEncoding::Base64 : opc::BinaryEncoding::Raw;
+  auto batch = extractHistoryBatch(data, binary_encoding, 
opc_history_context->logger);
+  if (batch && !batch->entries.empty()) {
+    const std::optional<size_t> remaining = opc_history_context->batch_size != 0
+        ? std::optional<size_t>(opc_history_context->batch_size - 
opc_history_context->entries_transferred)
+        : std::nullopt;
+    auto new_entries = selectNewEntries(std::move(batch->entries), 
opc_history_context->fetched_state, batch->has_modification_info, remaining);
+
+    if (!new_entries.empty()) {
+      if (opc_history_context->record_set_writer) {
+        writeAsRecordSet(*opc_history_context, new_entries);
+      } else {
+        writeAsFlowFiles(*opc_history_context, new_entries);
+      }
+      updateState(opc_history_context->state_map, new_entries, 
batch->has_modification_info);
+    }
+  }
+
+  const bool batch_limit_reached = opc_history_context->batch_size != 0 &&
+      opc_history_context->entries_transferred >= 
opc_history_context->batch_size;
+  return more_data_available && !batch_limit_reached;
+}
+
+UA_DateTime toUaDateTime(std::chrono::system_clock::time_point tp) {
+  // UA_DateTime counts 100 ns ticks since 1601; UA_DATETIME_USEC ticks make 
up one microsecond. Converting at microsecond
+  // resolution (rather than truncating to whole seconds) preserves the 
sub-second precision of user-provided timestamps.
+  const auto usec_since_epoch = 
std::chrono::duration_cast<std::chrono::microseconds>(tp.time_since_epoch()).count();
+  return UA_DATETIME_UNIX_EPOCH + usec_since_epoch * UA_DATETIME_USEC;
+}
+
+UA_DateTime calculateStartTime(const std::optional<FetchedState>& 
fetched_state,
+    const std::optional<std::chrono::system_clock::time_point>& 
start_timestamp) {
+  if (fetched_state && fetched_state->timestamp != 0) {
+    return fetched_state->timestamp;
+  } else if (start_timestamp.has_value()) {
+    return toUaDateTime(*start_timestamp);
+  }
+  return UA_DateTime_fromUnixTime(0);
+}
+
+UA_DateTime calculateEndTime(const 
std::optional<std::chrono::system_clock::time_point>& end_timestamp) {
+  if (end_timestamp.has_value()) {
+    return toUaDateTime(*end_timestamp);
+  }
+  return UA_DateTime_now();
+}
+
+}  // namespace
+
+void FetchOPCHistory::initialize() {
+  setSupportedProperties(Properties);
+  setSupportedRelationships(Relationships);
+}
+
+void FetchOPCHistory::onSchedule(core::ProcessContext& context, 
core::ProcessSessionFactory& factory) {
+  logger_->log_trace("FetchOPCHistory::onSchedule");
+  BaseOPCProcessor::onSchedule(context, factory);
+  node_id_ = utils::parseProperty(context, NodeID);
+  parseIdType(context, NodeIDType);
+  namespace_idx_ = gsl::narrow<int32_t>(utils::parseI64Property(context, 
NameSpaceIndex));
+
+  switch (id_type_) {
+    case opc::OPCNodeIDType::String:
+      node_ = opc::NodeId{UA_NODEID_STRING_ALLOC(namespace_idx_, 
node_id_.c_str())};
+      break;
+    case opc::OPCNodeIDType::Int:
+      node_ = opc::NodeId{UA_NODEID_NUMERIC(namespace_idx_, 
std::stoi(node_id_))};
+      break;
+    case opc::OPCNodeIDType::Guid: {
+      UA_Guid guid;
+      if (UA_Guid_parse(&guid, UA_STRING(const_cast<char*>(node_id_.c_str()))) 
!= UA_STATUSCODE_GOOD) {
+        throw Exception(PROCESS_SCHEDULE_EXCEPTION, fmt::format("{} cannot be 
used as a GUID type node ID", node_id_));
+      }
+      node_ = opc::NodeId{UA_NODEID_GUID(namespace_idx_, guid)};
+      break;
+    }
+    case opc::OPCNodeIDType::Path:
+      readPathReferenceTypes(context, node_id_);
+      path_node_id_resolved_ = false;
+      break;

Review Comment:
   `readPathReferenceTypes` appends to the inherited vector, but this 
scheduling path never clears it. After the processor is stopped and scheduled 
again, references from the previous configuration remain (and the same 
references are duplicated), so path translation can fail or resolve using stale 
configuration.



##########
extensions/opc/src/FetchOPCHistory.cpp:
##########
@@ -0,0 +1,400 @@
+/**
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "FetchOPCHistory.h"
+
+#include <optional>
+#include <string>
+#include <vector>
+
+#include "core/ProcessSession.h"
+#include "core/Resource.h"
+#include "minifi-cpp/core/ProcessContext.h"
+#include "utils/ProcessorConfigUtils.h"
+#include "utils/StringUtils.h"
+
+namespace org::apache::nifi::minifi::processors {
+
+namespace {
+
+constexpr const char* LAST_FETCHED_TIMESTAMP_KEY = "last_fetched_timestamp";
+constexpr const char* LAST_FETCHED_FINGERPRINT_KEY = 
"last_fetched_fingerprint";
+
+std::string updateTypeToString(UA_HistoryUpdateType type) {
+  switch (type) {
+    case UA_HISTORYUPDATETYPE_INSERT:
+      return "Insert";
+    case UA_HISTORYUPDATETYPE_REPLACE:
+      return "Replace";
+    case UA_HISTORYUPDATETYPE_UPDATE:
+      return "Update";
+    case UA_HISTORYUPDATETYPE_DELETE:
+      return "Delete";
+    default:
+      return "Unknown";
+  }
+}
+
+std::string uaStringToString(const UA_String& str) {
+  return {reinterpret_cast<const char*>(str.data), str.length};
+}
+
+struct HistoryEntry {
+  std::string value;
+  int64_t source_timestamp = 0;
+  const UA_ModificationInfo* modification_info = nullptr;
+
+  [[nodiscard]] int64_t modificationTime() const {
+    return modification_info ? modification_info->modificationTime : 
UA_DateTime_fromUnixTime(0);
+  }
+};
+
+struct HistoryBatch {
+  std::vector<HistoryEntry> entries;
+  bool has_modification_info = false;
+};
+
+std::string entryFingerprint(const HistoryEntry& entry, bool 
has_modification_info) {
+  // The fingerprint deduplicates entries sharing the boundary source 
timestamp across triggers. For raw value history a
+  // duplicated (value, source timestamp) pair represents no change in the 
history, so losing one to deduplication is harmless.
+  // For audit (modification) history the modification time and update type 
distinguish otherwise-identical entries, so they
+  // are included to make the fingerprint more unique and reduce the chance of 
dropping a distinct modification.
+  std::string raw = ":" + entry.value;
+  if (has_modification_info) {
+    const auto update_type = entry.modification_info ? 
updateTypeToString(entry.modification_info->updateType) : "";
+    raw = std::to_string(entry.modificationTime()) + ":" + update_type + raw;
+  }
+  return utils::string::to_hex(raw);
+}
+
+// NOLINTBEGIN(cppcoreguidelines-pro-type-union-access)
+std::optional<HistoryBatch> extractHistoryBatch(const UA_ExtensionObject* 
data, opc::BinaryEncoding binary_encoding,
+    const std::shared_ptr<core::logging::Logger>& logger) {
+  const UA_DataValue* data_values = nullptr;
+  size_t data_value_size = 0;
+  const UA_ModificationInfo* modification_infos = nullptr;
+  size_t modification_infos_size = 0;
+
+  if (data->content.decoded.type == &UA_TYPES[UA_TYPES_HISTORYDATA]) {
+    const auto* history_data = static_cast<const 
UA_HistoryData*>(data->content.decoded.data);
+    data_values = history_data->dataValues;
+    data_value_size = history_data->dataValuesSize;
+  } else if (data->content.decoded.type == 
&UA_TYPES[UA_TYPES_HISTORYMODIFIEDDATA]) {
+    const auto* modified_data = static_cast<const 
UA_HistoryModifiedData*>(data->content.decoded.data);
+    data_values = modified_data->dataValues;
+    data_value_size = modified_data->dataValuesSize;
+    modification_infos = modified_data->modificationInfos;
+    modification_infos_size = modified_data->modificationInfosSize;
+  } else {
+    logger->log_error("Unexpected data type received in the history read 
callback: {}", data->content.decoded.type->typeName);
+    return std::nullopt;
+  }
+
+  HistoryBatch batch;
+  batch.has_modification_info = modification_infos != nullptr;
+  batch.entries.reserve(data_value_size);
+  for (size_t i = 0; i < data_value_size; ++i) {
+    HistoryEntry entry;
+    try {
+      entry.value = opc::variantToString(data_values[i].value, 
binary_encoding);
+    } catch (const opc::OPCException& ex) {
+      logger->log_warn("Failed to convert value at index {} to string, 
skipping entry: {}", i, ex.what());
+      continue;
+    }
+    entry.source_timestamp = data_values[i].sourceTimestamp;
+    entry.modification_info = (modification_infos && i < 
modification_infos_size) ? &modification_infos[i] : nullptr;
+    batch.entries.push_back(std::move(entry));
+  }
+  return batch;
+}
+// NOLINTEND(cppcoreguidelines-pro-type-union-access)
+
+std::vector<HistoryEntry> selectNewEntries(std::vector<HistoryEntry> entries, 
const std::optional<FetchedState>& last_fetched,
+    bool has_modification_info, std::optional<size_t> max_entries) {
+  std::vector<HistoryEntry> new_entries;
+  new_entries.reserve(max_entries ? std::min(*max_entries, entries.size()) : 
entries.size());
+  for (auto& entry : entries) {
+    if (max_entries && new_entries.size() >= *max_entries) {
+      break;
+    }
+    if (last_fetched && entry.source_timestamp == last_fetched->timestamp &&
+        last_fetched->fingerprints.contains(entryFingerprint(entry, 
has_modification_info))) {
+      continue;
+    }
+    new_entries.push_back(std::move(entry));
+  }
+  return new_entries;
+}
+
+void addModificationInfo(core::Record& record, const UA_ModificationInfo& 
modification_info) {
+  if (modification_info.userName.length > 0) {
+    record.emplace("ModificationUsername", 
core::RecordField(uaStringToString(modification_info.userName)));
+  }
+  record.emplace("ModificationTime", 
core::RecordField(opc::OPCDateTime2String(modification_info.modificationTime)));
+  record.emplace("ModificationUpdateType", 
core::RecordField(updateTypeToString(modification_info.updateType)));
+}
+
+void addModificationInfo(core::FlowFile& flow_file, const UA_ModificationInfo& 
modification_info) {
+  if (modification_info.userName.length > 0) {
+    flow_file.addAttribute("ModificationUsername", 
uaStringToString(modification_info.userName));
+  }
+  flow_file.addAttribute("ModificationTime", 
opc::OPCDateTime2String(modification_info.modificationTime));
+  flow_file.addAttribute("ModificationUpdateType", 
updateTypeToString(modification_info.updateType));
+}
+
+core::Record toRecord(const std::string& node_id, const int32_t 
namespace_index, const HistoryEntry& entry) {
+  core::Record record;
+  record.emplace("Value", core::RecordField(entry.value));
+  record.emplace("NodeID", core::RecordField(node_id));
+  record.emplace("Namespace index", 
core::RecordField(std::to_string(namespace_index)));
+  record.emplace("Sourcetimestamp", 
core::RecordField(opc::OPCDateTime2String(entry.source_timestamp)));
+  if (entry.modification_info) {
+    addModificationInfo(record, *entry.modification_info);
+  }
+  return record;
+}
+
+void writeAsRecordSet(FetchOPCHistoryContext& context, const 
std::vector<HistoryEntry>& entries) {
+  core::RecordSet record_set;
+  for (const auto& entry : entries) {
+    record_set.push_back(toRecord(context.node_id, context.namespace_index, 
entry));
+  }
+
+  auto flow_file = context.session.create();
+  context.record_set_writer->write(record_set, flow_file, context.session);
+  context.session.transfer(flow_file, FetchOPCHistory::Success);
+  context.entries_transferred += entries.size();
+}
+
+void writeAsFlowFiles(FetchOPCHistoryContext& context, const 
std::vector<HistoryEntry>& entries) {
+  for (const auto& entry : entries) {
+    auto flow_file = context.session.create();
+    context.session.write(flow_file, [&entry](const 
std::shared_ptr<io::OutputStream>& output_stream) -> io::IoResult {
+      output_stream->write(reinterpret_cast<const 
uint8_t*>(entry.value.data()), entry.value.size());
+      return io::IoResult::from(entry.value.size());
+    });
+    flow_file->addAttribute("NodeID", context.node_id);
+    flow_file->addAttribute("Namespace index", 
std::to_string(context.namespace_index));
+    flow_file->addAttribute("Sourcetimestamp", 
opc::OPCDateTime2String(entry.source_timestamp));
+    if (entry.modification_info) {
+      addModificationInfo(*flow_file, *entry.modification_info);
+    }
+    context.session.transfer(flow_file, FetchOPCHistory::Success);
+    ++context.entries_transferred;
+  }
+}
+
+void updateState(std::unordered_map<std::string, std::string>& state_map, 
const std::vector<HistoryEntry>& new_entries, bool has_modification_info) {
+  const int64_t new_timestamp = new_entries.back().source_timestamp;
+  const auto new_timestamp_str = std::to_string(new_timestamp);
+
+  auto& stored_timestamp = state_map[LAST_FETCHED_TIMESTAMP_KEY];
+  auto& fingerprints = state_map[LAST_FETCHED_FINGERPRINT_KEY];
+  if (stored_timestamp != new_timestamp_str) {
+    stored_timestamp = new_timestamp_str;
+    fingerprints.clear();
+  }
+
+  for (const auto& entry : new_entries) {
+    if (entry.source_timestamp == new_timestamp) {
+      if (!fingerprints.empty()) {
+        fingerprints += ",";
+      }
+      fingerprints += entryFingerprint(entry, has_modification_info);
+    }
+  }
+}
+
+UA_Boolean historyReadCallback(UA_Client* /*client*/, const UA_NodeId* 
/*node_id*/, UA_Boolean more_data_available, const UA_ExtensionObject* data,
+    void* ctx) {
+  auto* opc_history_context = static_cast<FetchOPCHistoryContext*>(ctx);
+
+  const auto binary_encoding = opc_history_context->record_set_writer ? 
opc::BinaryEncoding::Base64 : opc::BinaryEncoding::Raw;
+  auto batch = extractHistoryBatch(data, binary_encoding, 
opc_history_context->logger);
+  if (batch && !batch->entries.empty()) {
+    const std::optional<size_t> remaining = opc_history_context->batch_size != 0
+        ? std::optional<size_t>(opc_history_context->batch_size - 
opc_history_context->entries_transferred)
+        : std::nullopt;
+    auto new_entries = selectNewEntries(std::move(batch->entries), 
opc_history_context->fetched_state, batch->has_modification_info, remaining);
+
+    if (!new_entries.empty()) {
+      if (opc_history_context->record_set_writer) {
+        writeAsRecordSet(*opc_history_context, new_entries);
+      } else {
+        writeAsFlowFiles(*opc_history_context, new_entries);
+      }
+      updateState(opc_history_context->state_map, new_entries, 
batch->has_modification_info);
+    }
+  }
+
+  const bool batch_limit_reached = opc_history_context->batch_size != 0 &&
+      opc_history_context->entries_transferred >= 
opc_history_context->batch_size;
+  return more_data_available && !batch_limit_reached;
+}
+
+UA_DateTime toUaDateTime(std::chrono::system_clock::time_point tp) {
+  // UA_DateTime counts 100 ns ticks since 1601; UA_DATETIME_USEC ticks make 
up one microsecond. Converting at microsecond
+  // resolution (rather than truncating to whole seconds) preserves the 
sub-second precision of user-provided timestamps.
+  const auto usec_since_epoch = 
std::chrono::duration_cast<std::chrono::microseconds>(tp.time_since_epoch()).count();
+  return UA_DATETIME_UNIX_EPOCH + usec_since_epoch * UA_DATETIME_USEC;
+}
+
+UA_DateTime calculateStartTime(const std::optional<FetchedState>& 
fetched_state,
+    const std::optional<std::chrono::system_clock::time_point>& 
start_timestamp) {
+  if (fetched_state && fetched_state->timestamp != 0) {
+    return fetched_state->timestamp;
+  } else if (start_timestamp.has_value()) {
+    return toUaDateTime(*start_timestamp);
+  }
+  return UA_DateTime_fromUnixTime(0);
+}
+
+UA_DateTime calculateEndTime(const 
std::optional<std::chrono::system_clock::time_point>& end_timestamp) {
+  if (end_timestamp.has_value()) {
+    return toUaDateTime(*end_timestamp);
+  }
+  return UA_DateTime_now();
+}
+
+}  // namespace
+
+void FetchOPCHistory::initialize() {
+  setSupportedProperties(Properties);
+  setSupportedRelationships(Relationships);
+}
+
+void FetchOPCHistory::onSchedule(core::ProcessContext& context, 
core::ProcessSessionFactory& factory) {
+  logger_->log_trace("FetchOPCHistory::onSchedule");
+  BaseOPCProcessor::onSchedule(context, factory);
+  node_id_ = utils::parseProperty(context, NodeID);
+  parseIdType(context, NodeIDType);
+  namespace_idx_ = gsl::narrow<int32_t>(utils::parseI64Property(context, 
NameSpaceIndex));
+
+  switch (id_type_) {
+    case opc::OPCNodeIDType::String:
+      node_ = opc::NodeId{UA_NODEID_STRING_ALLOC(namespace_idx_, 
node_id_.c_str())};
+      break;
+    case opc::OPCNodeIDType::Int:
+      node_ = opc::NodeId{UA_NODEID_NUMERIC(namespace_idx_, 
std::stoi(node_id_))};
+      break;
+    case opc::OPCNodeIDType::Guid: {
+      UA_Guid guid;
+      if (UA_Guid_parse(&guid, UA_STRING(const_cast<char*>(node_id_.c_str()))) 
!= UA_STATUSCODE_GOOD) {
+        throw Exception(PROCESS_SCHEDULE_EXCEPTION, fmt::format("{} cannot be 
used as a GUID type node ID", node_id_));
+      }
+      node_ = opc::NodeId{UA_NODEID_GUID(namespace_idx_, guid)};
+      break;
+    }
+    case opc::OPCNodeIDType::Path:
+      readPathReferenceTypes(context, node_id_);
+      path_node_id_resolved_ = false;
+      break;
+    default:
+      throw Exception(PROCESS_SCHEDULE_EXCEPTION, fmt::format("Unsupported 
Node ID type: {}", magic_enum::enum_name(id_type_)));
+  }
+
+  history_type_ = 
utils::parseEnumProperty<opc::HistoryReadTypeOption>(context, HistoryReadType);
+  start_timestamp_ = utils::parseOptionalProperty(context, StartTimestamp) | 
utils::andThen(utils::timeutils::parseRfc3339);
+  end_timestamp_ = utils::parseOptionalProperty(context, EndTimestamp) | 
utils::andThen(utils::timeutils::parseRfc3339);
+  batch_size_ = utils::parseOptionalU64Property(context, 
BatchSize).value_or(0);
+  const auto record_set_writer_name = 
context.getProperty(RecordSetWriter).value_or("");
+  auto controller_service = 
context.getControllerService(record_set_writer_name, getUUID());
+  if (!record_set_writer_name.empty() && !controller_service) {
+    throw Exception(PROCESS_SCHEDULE_EXCEPTION, fmt::format("Controller 
service '{}' not found", record_set_writer_name));
+  }
+  record_set_writer_ = 
std::dynamic_pointer_cast<core::RecordSetWriter>(controller_service);
+}
+
+std::optional<FetchedState> FetchOPCHistory::parseFetchedState(const 
std::unordered_map<std::string, std::string>& state_map) {
+  const auto timestamp_it = state_map.find(LAST_FETCHED_TIMESTAMP_KEY);
+  const auto fingerprints_it = state_map.find(LAST_FETCHED_FINGERPRINT_KEY);
+  if (timestamp_it == state_map.end() || fingerprints_it == state_map.end()) {
+    return std::nullopt;
+  }
+
+  FetchedState state;
+  try {
+    state.timestamp = std::stoll(timestamp_it->second);
+  } catch (const std::exception&) {
+    logger_->log_error("Failed to parse timestamp from state map: {}", 
timestamp_it->second);
+    return std::nullopt;
+  }
+
+  for (auto& fingerprint : utils::string::split(fingerprints_it->second, ",")) 
{
+    if (!fingerprint.empty()) {
+      state.fingerprints.insert(std::move(fingerprint));
+    }
+  }
+  return state;
+}
+
+void FetchOPCHistory::onTrigger(core::ProcessContext& context, 
core::ProcessSession& session) {
+  logger_->log_trace("FetchOPCHistory::onTrigger");
+
+  if (!reconnect()) {
+    context.yield();
+    return;
+  }
+
+  if (id_type_ == opc::OPCNodeIDType::Path && !path_node_id_resolved_) {
+    std::vector<opc::NodeId> translated_node_ids;
+    auto sc = connection_->translateBrowsePathsToNodeIdsRequest(node_id_, 
translated_node_ids, namespace_idx_, path_reference_types_, logger_);
+    if (sc != UA_STATUSCODE_GOOD) {
+      logger_->log_error("Failed to translate path '{}' to a node id: {}", 
node_id_, UA_StatusCode_name(sc));
+      context.yield();
+      return;
+    }
+    if (translated_node_ids.size() != 1) {
+      logger_->log_error("Path '{}' resolved to {} node ids; exactly one is 
required to fetch history", node_id_, translated_node_ids.size());
+      context.yield();
+      return;
+    }
+    node_ = std::move(translated_node_ids[0]);
+    path_node_id_resolved_ = true;
+  }
+
+  auto* state_manager = context.getStateManager();
+  std::unordered_map<std::string, std::string> state_map;
+
+  state_manager->get(state_map);
+  const auto fetched_state = parseFetchedState(state_map);
+
+  size_t entries_transferred = 0;
+  FetchOPCHistoryContext
+      history_context{session, record_set_writer_, state_map, 
entries_transferred, batch_size_, node_id_, namespace_idx_, fetched_state, 
logger_};
+
+  auto retval = connection_->readHistory(history_type_,
+      node_,
+      &historyReadCallback,
+      calculateStartTime(fetched_state, start_timestamp_),
+      calculateEndTime(end_timestamp_),
+      static_cast<void*>(&history_context));
+
+  if (retval != UA_STATUSCODE_GOOD) {
+    logger_->log_error("Failed to read OPC UA node history, status code: {}", 
UA_StatusCode_name(retval));
+    context.yield();
+    return;
+  }

Review Comment:
   The open62541 helper can invoke the callback for successful pages and then 
return an error from a later continuation request. At that point this session 
already contains transferred FlowFiles, and returning normally causes 
`Processor::triggerAndCommit` to commit them while the checkpoint is not 
updated, so they are duplicated on retry. Roll back the session (or throw) on 
this path.



##########
extensions/opc/src/OPCCommon.cpp:
##########
@@ -619,4 +584,12 @@ std::optional<UA_UInt32> mapOpcReferenceType(const 
std::string& ref_type) {
   return std::nullopt;
 }
 
+UA_StatusCode Client::readHistory(HistoryReadTypeOption history_type, const 
UA_NodeId& node_id, const HistoryCallback callback, UA_DateTime start_time, 
UA_DateTime end_time,
+    void* callback_context) {
+  if (history_type == HistoryReadTypeOption::Audit) {
+    return UA_Client_HistoryRead_modified(client_, &node_id, callback, 
start_time, end_time, UA_STRING_NULL, false, 0, UA_TIMESTAMPSTORETURN_SOURCE, 
callback_context);
+  }
+  return UA_Client_HistoryRead_raw(client_, &node_id, callback, start_time, 
end_time, UA_STRING_NULL, false, 0, UA_TIMESTAMPSTORETURN_SOURCE, 
callback_context);

Review Comment:
   Both history calls pass `0` for `numValuesPerNode`, which asks the server 
for all values (or its own maximum). Consequently, the processor's `Batch size` 
only truncates data after a potentially unbounded response has already been 
transferred and allocated. Thread the configured batch size into `readHistory` 
and pass it as this argument (preserving zero as unlimited).



##########
extensions/opc/src/OPCCommon.cpp:
##########
@@ -493,107 +494,71 @@ template UA_StatusCode Client::add_node<const char 
*>(const UA_NodeId parent_nod
 template UA_StatusCode Client::add_node<std::string>(const UA_NodeId 
parent_node_id, const UA_NodeId target_node_id, const UA_UInt32 ref_type_id, 
std::string_view browse_name,
   std::string value, UA_NodeId *received_node_id);
 
-std::string nodeValue2String(const NodeData& nd) {
-  std::string ret_val;
-  switch (nd.data_type_id) {
+std::string variantToString(const UA_Variant& variant, BinaryEncoding 
binary_encoding) {
+  if (variant.type == nullptr || variant.data == nullptr) {
+    throw OPCException(GENERAL_EXCEPTION, "Cannot convert an empty variant to 
string");
+  }

Review Comment:
   `UA_Variant` can hold arrays as well as scalars. For an array, `data` is 
non-null, so the switch below casts the array buffer to one scalar and silently 
emits only its first element (and an empty-array sentinel may be dereferenced). 
Either serialize every element or reject non-scalar variants before dispatching.



##########
extensions/opc/src/OPCCommon.cpp:
##########
@@ -493,107 +494,71 @@ template UA_StatusCode Client::add_node<const char 
*>(const UA_NodeId parent_nod
 template UA_StatusCode Client::add_node<std::string>(const UA_NodeId 
parent_node_id, const UA_NodeId target_node_id, const UA_UInt32 ref_type_id, 
std::string_view browse_name,
   std::string value, UA_NodeId *received_node_id);
 
-std::string nodeValue2String(const NodeData& nd) {
-  std::string ret_val;
-  switch (nd.data_type_id) {
+std::string variantToString(const UA_Variant& variant, BinaryEncoding 
binary_encoding) {
+  if (variant.type == nullptr || variant.data == nullptr) {
+    throw OPCException(GENERAL_EXCEPTION, "Cannot convert an empty variant to 
string");
+  }
+  switch (variant.type->typeKind) {
     case UA_DATATYPEKIND_STRING:
-    case UA_DATATYPEKIND_LOCALIZEDTEXT:
-    case UA_DATATYPEKIND_BYTESTRING: {
-      UA_String value = *reinterpret_cast<UA_String *>(nd.var_->data);
-      ret_val = std::string(reinterpret_cast<const char *>(value.data), 
value.length);
-      break;
-    }
-    case UA_DATATYPEKIND_BOOLEAN: {
-      bool b = false;
-      memcpy(&b, nd.data.data(), sizeof(bool));
-      ret_val = b ? "True" : "False";
-      break;
-    }
-    case UA_DATATYPEKIND_SBYTE: {
-      int8_t i8t = 0;
-      memcpy(&i8t, nd.data.data(), sizeof(i8t));
-      ret_val = std::to_string(i8t);
-      break;
-    }
-    case UA_DATATYPEKIND_BYTE: {
-      uint8_t ui8t = 0;
-      memcpy(&ui8t, nd.data.data(), sizeof(ui8t));
-      ret_val = std::to_string(ui8t);
-      break;
-    }
-    case UA_DATATYPEKIND_INT16: {
-      int16_t i16t = 0;
-      memcpy(&i16t, nd.data.data(), sizeof(i16t));
-      ret_val = std::to_string(i16t);
-      break;
+    case UA_DATATYPEKIND_LOCALIZEDTEXT: {
+      const auto *value = static_cast<const UA_String *>(variant.data);
+      return {reinterpret_cast<const char *>(value->data), value->length};

Review Comment:
   `UA_LocalizedText` is not layout-compatible with `UA_String`; it contains 
separate `locale` and `text` members. Casting it to `UA_String` returns the 
locale (or empty data) instead of the localized text value.



##########
extensions/opc/src/FetchOPCHistory.cpp:
##########
@@ -0,0 +1,400 @@
+/**
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "FetchOPCHistory.h"
+
+#include <optional>
+#include <string>
+#include <vector>
+
+#include "core/ProcessSession.h"
+#include "core/Resource.h"
+#include "minifi-cpp/core/ProcessContext.h"
+#include "utils/ProcessorConfigUtils.h"
+#include "utils/StringUtils.h"
+
+namespace org::apache::nifi::minifi::processors {
+
+namespace {
+
+constexpr const char* LAST_FETCHED_TIMESTAMP_KEY = "last_fetched_timestamp";
+constexpr const char* LAST_FETCHED_FINGERPRINT_KEY = 
"last_fetched_fingerprint";
+
+std::string updateTypeToString(UA_HistoryUpdateType type) {
+  switch (type) {
+    case UA_HISTORYUPDATETYPE_INSERT:
+      return "Insert";
+    case UA_HISTORYUPDATETYPE_REPLACE:
+      return "Replace";
+    case UA_HISTORYUPDATETYPE_UPDATE:
+      return "Update";
+    case UA_HISTORYUPDATETYPE_DELETE:
+      return "Delete";
+    default:
+      return "Unknown";
+  }
+}
+
+std::string uaStringToString(const UA_String& str) {
+  return {reinterpret_cast<const char*>(str.data), str.length};
+}
+
+struct HistoryEntry {
+  std::string value;
+  int64_t source_timestamp = 0;
+  const UA_ModificationInfo* modification_info = nullptr;
+
+  [[nodiscard]] int64_t modificationTime() const {
+    return modification_info ? modification_info->modificationTime : 
UA_DateTime_fromUnixTime(0);
+  }
+};
+
+struct HistoryBatch {
+  std::vector<HistoryEntry> entries;
+  bool has_modification_info = false;
+};
+
+std::string entryFingerprint(const HistoryEntry& entry, bool 
has_modification_info) {
+  // The fingerprint deduplicates entries sharing the boundary source 
timestamp across triggers. For raw value history a
+  // duplicated (value, source timestamp) pair represents no change in the 
history, so losing one to deduplication is harmless.
+  // For audit (modification) history the modification time and update type 
distinguish otherwise-identical entries, so they
+  // are included to make the fingerprint more unique and reduce the chance of 
dropping a distinct modification.
+  std::string raw = ":" + entry.value;
+  if (has_modification_info) {
+    const auto update_type = entry.modification_info ? 
updateTypeToString(entry.modification_info->updateType) : "";
+    raw = std::to_string(entry.modificationTime()) + ":" + update_type + raw;
+  }
+  return utils::string::to_hex(raw);
+}
+
+// NOLINTBEGIN(cppcoreguidelines-pro-type-union-access)
+std::optional<HistoryBatch> extractHistoryBatch(const UA_ExtensionObject* 
data, opc::BinaryEncoding binary_encoding,
+    const std::shared_ptr<core::logging::Logger>& logger) {
+  const UA_DataValue* data_values = nullptr;
+  size_t data_value_size = 0;
+  const UA_ModificationInfo* modification_infos = nullptr;
+  size_t modification_infos_size = 0;
+
+  if (data->content.decoded.type == &UA_TYPES[UA_TYPES_HISTORYDATA]) {
+    const auto* history_data = static_cast<const 
UA_HistoryData*>(data->content.decoded.data);
+    data_values = history_data->dataValues;
+    data_value_size = history_data->dataValuesSize;
+  } else if (data->content.decoded.type == 
&UA_TYPES[UA_TYPES_HISTORYMODIFIEDDATA]) {
+    const auto* modified_data = static_cast<const 
UA_HistoryModifiedData*>(data->content.decoded.data);
+    data_values = modified_data->dataValues;
+    data_value_size = modified_data->dataValuesSize;
+    modification_infos = modified_data->modificationInfos;
+    modification_infos_size = modified_data->modificationInfosSize;
+  } else {
+    logger->log_error("Unexpected data type received in the history read 
callback: {}", data->content.decoded.type->typeName);
+    return std::nullopt;
+  }
+
+  HistoryBatch batch;
+  batch.has_modification_info = modification_infos != nullptr;
+  batch.entries.reserve(data_value_size);
+  for (size_t i = 0; i < data_value_size; ++i) {
+    HistoryEntry entry;
+    try {
+      entry.value = opc::variantToString(data_values[i].value, 
binary_encoding);
+    } catch (const opc::OPCException& ex) {
+      logger->log_warn("Failed to convert value at index {} to string, 
skipping entry: {}", i, ex.what());
+      continue;
+    }
+    entry.source_timestamp = data_values[i].sourceTimestamp;
+    entry.modification_info = (modification_infos && i < 
modification_infos_size) ? &modification_infos[i] : nullptr;
+    batch.entries.push_back(std::move(entry));
+  }
+  return batch;
+}
+// NOLINTEND(cppcoreguidelines-pro-type-union-access)
+
+std::vector<HistoryEntry> selectNewEntries(std::vector<HistoryEntry> entries, 
const std::optional<FetchedState>& last_fetched,
+    bool has_modification_info, std::optional<size_t> max_entries) {
+  std::vector<HistoryEntry> new_entries;
+  new_entries.reserve(max_entries ? std::min(*max_entries, entries.size()) : 
entries.size());
+  for (auto& entry : entries) {
+    if (max_entries && new_entries.size() >= *max_entries) {
+      break;
+    }
+    if (last_fetched && entry.source_timestamp == last_fetched->timestamp &&
+        last_fetched->fingerprints.contains(entryFingerprint(entry, 
has_modification_info))) {
+      continue;
+    }
+    new_entries.push_back(std::move(entry));
+  }
+  return new_entries;
+}
+
+void addModificationInfo(core::Record& record, const UA_ModificationInfo& 
modification_info) {
+  if (modification_info.userName.length > 0) {
+    record.emplace("ModificationUsername", 
core::RecordField(uaStringToString(modification_info.userName)));
+  }
+  record.emplace("ModificationTime", 
core::RecordField(opc::OPCDateTime2String(modification_info.modificationTime)));
+  record.emplace("ModificationUpdateType", 
core::RecordField(updateTypeToString(modification_info.updateType)));
+}
+
+void addModificationInfo(core::FlowFile& flow_file, const UA_ModificationInfo& 
modification_info) {
+  if (modification_info.userName.length > 0) {
+    flow_file.addAttribute("ModificationUsername", 
uaStringToString(modification_info.userName));
+  }
+  flow_file.addAttribute("ModificationTime", 
opc::OPCDateTime2String(modification_info.modificationTime));
+  flow_file.addAttribute("ModificationUpdateType", 
updateTypeToString(modification_info.updateType));
+}
+
+core::Record toRecord(const std::string& node_id, const int32_t 
namespace_index, const HistoryEntry& entry) {
+  core::Record record;
+  record.emplace("Value", core::RecordField(entry.value));
+  record.emplace("NodeID", core::RecordField(node_id));
+  record.emplace("Namespace index", 
core::RecordField(std::to_string(namespace_index)));
+  record.emplace("Sourcetimestamp", 
core::RecordField(opc::OPCDateTime2String(entry.source_timestamp)));
+  if (entry.modification_info) {
+    addModificationInfo(record, *entry.modification_info);
+  }
+  return record;
+}
+
+void writeAsRecordSet(FetchOPCHistoryContext& context, const 
std::vector<HistoryEntry>& entries) {
+  core::RecordSet record_set;
+  for (const auto& entry : entries) {
+    record_set.push_back(toRecord(context.node_id, context.namespace_index, 
entry));
+  }
+
+  auto flow_file = context.session.create();
+  context.record_set_writer->write(record_set, flow_file, context.session);
+  context.session.transfer(flow_file, FetchOPCHistory::Success);
+  context.entries_transferred += entries.size();
+}
+
+void writeAsFlowFiles(FetchOPCHistoryContext& context, const 
std::vector<HistoryEntry>& entries) {
+  for (const auto& entry : entries) {
+    auto flow_file = context.session.create();
+    context.session.write(flow_file, [&entry](const 
std::shared_ptr<io::OutputStream>& output_stream) -> io::IoResult {
+      output_stream->write(reinterpret_cast<const 
uint8_t*>(entry.value.data()), entry.value.size());
+      return io::IoResult::from(entry.value.size());
+    });
+    flow_file->addAttribute("NodeID", context.node_id);
+    flow_file->addAttribute("Namespace index", 
std::to_string(context.namespace_index));
+    flow_file->addAttribute("Sourcetimestamp", 
opc::OPCDateTime2String(entry.source_timestamp));
+    if (entry.modification_info) {
+      addModificationInfo(*flow_file, *entry.modification_info);
+    }
+    context.session.transfer(flow_file, FetchOPCHistory::Success);
+    ++context.entries_transferred;
+  }
+}
+
+void updateState(std::unordered_map<std::string, std::string>& state_map, 
const std::vector<HistoryEntry>& new_entries, bool has_modification_info) {
+  const int64_t new_timestamp = new_entries.back().source_timestamp;
+  const auto new_timestamp_str = std::to_string(new_timestamp);
+
+  auto& stored_timestamp = state_map[LAST_FETCHED_TIMESTAMP_KEY];
+  auto& fingerprints = state_map[LAST_FETCHED_FINGERPRINT_KEY];
+  if (stored_timestamp != new_timestamp_str) {
+    stored_timestamp = new_timestamp_str;
+    fingerprints.clear();
+  }
+
+  for (const auto& entry : new_entries) {
+    if (entry.source_timestamp == new_timestamp) {
+      if (!fingerprints.empty()) {
+        fingerprints += ",";
+      }
+      fingerprints += entryFingerprint(entry, has_modification_info);
+    }
+  }
+}
+
+UA_Boolean historyReadCallback(UA_Client* /*client*/, const UA_NodeId* 
/*node_id*/, UA_Boolean more_data_available, const UA_ExtensionObject* data,
+    void* ctx) {
+  auto* opc_history_context = static_cast<FetchOPCHistoryContext*>(ctx);
+
+  const auto binary_encoding = opc_history_context->record_set_writer ? 
opc::BinaryEncoding::Base64 : opc::BinaryEncoding::Raw;
+  auto batch = extractHistoryBatch(data, binary_encoding, 
opc_history_context->logger);
+  if (batch && !batch->entries.empty()) {
+    const std::optional<size_t> remaining = opc_history_context->batch_size != 0
+        ? std::optional<size_t>(opc_history_context->batch_size - 
opc_history_context->entries_transferred)
+        : std::nullopt;
+    auto new_entries = selectNewEntries(std::move(batch->entries), 
opc_history_context->fetched_state, batch->has_modification_info, remaining);
+
+    if (!new_entries.empty()) {
+      if (opc_history_context->record_set_writer) {
+        writeAsRecordSet(*opc_history_context, new_entries);
+      } else {
+        writeAsFlowFiles(*opc_history_context, new_entries);
+      }
+      updateState(opc_history_context->state_map, new_entries, 
batch->has_modification_info);
+    }
+  }
+
+  const bool batch_limit_reached = opc_history_context->batch_size != 0 &&
+      opc_history_context->entries_transferred >= 
opc_history_context->batch_size;
+  return more_data_available && !batch_limit_reached;
+}
+
+UA_DateTime toUaDateTime(std::chrono::system_clock::time_point tp) {
+  // UA_DateTime counts 100 ns ticks since 1601; UA_DATETIME_USEC ticks make 
up one microsecond. Converting at microsecond
+  // resolution (rather than truncating to whole seconds) preserves the 
sub-second precision of user-provided timestamps.
+  const auto usec_since_epoch = 
std::chrono::duration_cast<std::chrono::microseconds>(tp.time_since_epoch()).count();
+  return UA_DATETIME_UNIX_EPOCH + usec_since_epoch * UA_DATETIME_USEC;
+}
+
+UA_DateTime calculateStartTime(const std::optional<FetchedState>& 
fetched_state,
+    const std::optional<std::chrono::system_clock::time_point>& 
start_timestamp) {
+  if (fetched_state && fetched_state->timestamp != 0) {
+    return fetched_state->timestamp;
+  } else if (start_timestamp.has_value()) {
+    return toUaDateTime(*start_timestamp);
+  }
+  return UA_DateTime_fromUnixTime(0);
+}
+
+UA_DateTime calculateEndTime(const 
std::optional<std::chrono::system_clock::time_point>& end_timestamp) {
+  if (end_timestamp.has_value()) {
+    return toUaDateTime(*end_timestamp);
+  }
+  return UA_DateTime_now();
+}
+
+}  // namespace
+
+void FetchOPCHistory::initialize() {
+  setSupportedProperties(Properties);
+  setSupportedRelationships(Relationships);
+}
+
+void FetchOPCHistory::onSchedule(core::ProcessContext& context, 
core::ProcessSessionFactory& factory) {
+  logger_->log_trace("FetchOPCHistory::onSchedule");
+  BaseOPCProcessor::onSchedule(context, factory);
+  node_id_ = utils::parseProperty(context, NodeID);
+  parseIdType(context, NodeIDType);
+  namespace_idx_ = gsl::narrow<int32_t>(utils::parseI64Property(context, 
NameSpaceIndex));
+
+  switch (id_type_) {
+    case opc::OPCNodeIDType::String:
+      node_ = opc::NodeId{UA_NODEID_STRING_ALLOC(namespace_idx_, 
node_id_.c_str())};
+      break;
+    case opc::OPCNodeIDType::Int:
+      node_ = opc::NodeId{UA_NODEID_NUMERIC(namespace_idx_, 
std::stoi(node_id_))};
+      break;
+    case opc::OPCNodeIDType::Guid: {
+      UA_Guid guid;
+      if (UA_Guid_parse(&guid, UA_STRING(const_cast<char*>(node_id_.c_str()))) 
!= UA_STATUSCODE_GOOD) {
+        throw Exception(PROCESS_SCHEDULE_EXCEPTION, fmt::format("{} cannot be 
used as a GUID type node ID", node_id_));
+      }
+      node_ = opc::NodeId{UA_NODEID_GUID(namespace_idx_, guid)};
+      break;
+    }
+    case opc::OPCNodeIDType::Path:
+      readPathReferenceTypes(context, node_id_);
+      path_node_id_resolved_ = false;
+      break;
+    default:
+      throw Exception(PROCESS_SCHEDULE_EXCEPTION, fmt::format("Unsupported 
Node ID type: {}", magic_enum::enum_name(id_type_)));
+  }
+
+  history_type_ = 
utils::parseEnumProperty<opc::HistoryReadTypeOption>(context, HistoryReadType);
+  start_timestamp_ = utils::parseOptionalProperty(context, StartTimestamp) | 
utils::andThen(utils::timeutils::parseRfc3339);
+  end_timestamp_ = utils::parseOptionalProperty(context, EndTimestamp) | 
utils::andThen(utils::timeutils::parseRfc3339);

Review Comment:
   A configured timestamp that is not valid RFC 3339 is converted to `nullopt`, 
making an invalid start mean “beginning of history” and an invalid end mean 
“now.” This can unexpectedly fetch the entire history instead of rejecting bad 
configuration; detect parse failure and throw a scheduling exception that 
identifies the property.



-- 
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]

Reply via email to