Check-in Chromium protobuf definitions for download tracking, perform…

.gitignore

@@ -14,4 +14,5 @@ pyhindsight/static/web_modules/
 .claude/
 hindsight-temp/
 data-sets/
-/temp_outputs/
+/temp_outputs/
+/test-fixtures/

pyhindsight/lib/components/download/database/proto/download_entry.proto

@@ -0,0 +1,110 @@
+// Copyright 2017 The Chromium Authors
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+syntax = "proto2";
+
+option optimize_for = LITE_RUNTIME;
+
+package download_pb;
+
+import "components/download/database/proto/download_source.proto";
+
+message HttpRequestHeader {
+  optional string key = 1;
+  optional string value = 2;
+}
+
+// Slice information for parallel downloading.
+message ReceivedSlice {
+  optional int64 offset = 1;
+  optional int64 received_bytes = 2;
+  optional bool finished = 3;
+}
+
+// Stores various in-progress metadata related to a download.
+message DownloadEntry {
+  optional string guid = 1;
+  optional string request_origin = 2;
+  optional DownloadSource download_source = 3;
+  optional int64 ukm_download_id = 4;
+  optional int64 bytes_wasted = 5;
+  optional bool fetch_error_body = 6;
+  repeated HttpRequestHeader request_headers = 7;
+}
+
+// Contains a list of entries.
+message DownloadEntries {
+  repeated DownloadEntry entries = 1;
+}
+
+// Information for ukm reporting
+message UkmInfo {
+  optional DownloadSource download_source = 1;
+  optional int64 ukm_download_id = 2;
+}
+
+// Information about when to start the download, used by download later feature.
+// Deprecated.
+message DownloadSchedule {
+  optional int64 start_time = 1;
+}
+
+// Information about an in progress download.
+message InProgressInfo {
+  repeated string url_chain = 1;
+  optional string referrer_url = 2;
+  optional string site_url = 3;  // deprecated
+  optional string tab_url = 4;
+  optional string tab_referrer_url = 5;
+  optional bool fetch_error_body = 6;
+  repeated HttpRequestHeader request_headers = 7;
+  optional string etag = 8;
+  optional string last_modified = 9;
+  optional int64 total_bytes = 10;
+  optional string mime_type = 11;
+  optional string original_mime_type = 12;
+  optional bytes current_path = 13;  // Serialized pickles to support string16
+  optional bytes target_path = 14;   // Serialized pickles to support string16
+  optional int64 received_bytes = 15;
+  optional int64 start_time = 16;
+  optional int64 end_time = 17;
+  repeated ReceivedSlice received_slices = 18;
+  optional bytes hash = 19;  // Hindsight: upstream declares this `string` but it holds the
+                             // raw binary SHA-256; `bytes` lets us extract/hex it cleanly.
+  optional bool transient = 20;
+  optional int32 state = 21;
+  optional int32 danger_type = 22;
+  optional int32 interrupt_reason = 23;
+  optional bool paused = 24;
+  optional bool metered = 25;
+  optional int64 bytes_wasted = 26;
+  optional int32 auto_resume_count = 27;
+  optional DownloadSchedule download_schedule = 28;  // // Deprecated.
+  // Removing DownloadItemRerouteInfo since FileSystem Connector will not
+  // be released, implementation deleted.
+  reserved 29;
+  optional int32 credentials_mode = 30;  // network::mojom::CredentialsMode
+  optional int64 range_request_from = 31;
+  optional int64 range_request_to = 32;
+  optional bytes serialized_embedder_download_data = 33;  // Hindsight: upstream `string`,
+                             // but holds a serialized content.proto.EmbedderDownloadData;
+                             // `bytes` lets us FromString it cleanly.
+  reserved 34;
+  optional bool fetched_via_service_worker = 35;
+}
+
+// Stores various metadata related to a download.
+// WIP and will replace DownloadEntry.
+message DownloadInfo {
+  optional string guid = 1;
+  optional int32 id = 2;
+  optional UkmInfo ukm_info = 3;
+  optional InProgressInfo in_progress_info = 4;
+}
+
+// In progress database entry for download information.
+message DownloadDBEntry {
+  // Add field for offline page download.
+  oneof entry { DownloadInfo download_info = 1; }
+}

pyhindsight/lib/components/download/database/proto/download_source.proto

@@ -0,0 +1,27 @@
+// Copyright 2017 The Chromium Authors
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+syntax = "proto2";
+
+option optimize_for = LITE_RUNTIME;
+
+package download_pb;
+
+// This should stay in sync with the download::DownloadSource enum in
+// components/download/public/common/download_source.h.
+enum DownloadSource {
+  UNKNOWN = 0;
+  NAVIGATION = 1;
+  DRAG_AND_DROP = 2;
+  FROM_RENDERER = 3;
+  EXTENSION_API = 4;
+  EXTENSION_INSTALLER = 5;
+  INTERNAL_API = 6;
+  WEB_CONTENTS_API = 7;
+  OFFLINE_PAGE = 8;
+  CONTEXT_MENU = 9;
+  RETRY = 10;
+  RETRY_FROM_BUBBLE = 11;
+  TOOLBAR_MENU = 12;
+}

pyhindsight/lib/components/performance_manager/persistence/site_data/site_data.proto

@@ -0,0 +1,49 @@
+// Copyright 2019 The Chromium Authors
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+syntax = "proto2";
+
+option optimize_for = LITE_RUNTIME;
+
+// Contains the information that we want to track about a given site feature.
+// Next Id: 3
+message SiteDataFeatureProto {
+  // The cumulative observation time for this feature in seconds, set to 0 once
+  // this feature has been observed.
+  optional int64 observation_duration = 1;
+  // The time at which this feature has been used (set to 0 if it hasn't been
+  // used), in seconds since epoch.
+  optional int64 use_timestamp = 2;
+}
+
+// Contains decaying average performance measurement estimates.
+// Next Id: 4
+message SiteDataPerformanceMeasurement {
+  // A decaying average of the CPU usage measurements. Units: microseconds.
+  optional float avg_cpu_usage_us = 1;
+  // A decaying average of the process footprint measurements. Units: kilobytes.
+  optional float avg_footprint_kb = 2;
+  // A decaying average of the duration from navigation commit to "loaded".
+  // Units: microseconds.
+  optional float avg_load_duration_us = 3;
+};
+
+// Defines the data that we want to track about a given site.
+// Next Id: 7
+message SiteDataProto {
+  // The last time this site has been in the loaded state, in seconds since
+  // epoch.
+  optional uint32 last_loaded = 1;
+
+  // List of features that we're tracking.
+  optional SiteDataFeatureProto updates_favicon_in_background = 2;
+  optional SiteDataFeatureProto updates_title_in_background = 3;
+  optional SiteDataFeatureProto uses_audio_in_background = 4;
+  optional SiteDataFeatureProto deprecated_uses_notifications_in_background = 5;
+
+  // Load time performance measurement estimates. This maintains a decaying
+  // average of the resource usage of a page until shortly after it becomes
+  // idle.
+  optional SiteDataPerformanceMeasurement load_time_estimates = 6;
+}