// Copyright 2024 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;
option java_package = "org.chromium.components.metrics";

option java_outer_classname = "OmniboxScoringSignalsProtos";

package metrics;

// Signals for machine learning scoring (used during ML inference).
// See details: http://shortn/_B21YgmkLs9.
//
// Next tag: 38
message OmniboxScoringSignals {
  // Number of times the URL was navigated to using the Omnibox for this
  // profile on this device, and all other devices syncing with this
  // profile. Only include typed visits from the last 90 days. Discounted by
  // a time-decaying factor with a 30-day half-life based on the last visit
  // timestamp.
  optional int32 typed_count = 1;

  // Number of times the URL was visited in general for this profile on this
  // device. Note that this will soon be across syncing devices with the
  // rollout of full history sync in 2023. Only include visits from the last
  // 90 days. Discounted by a time-decaying factor with a 30-day half-life
  // based on the last visit timestamp.
  optional int32 visit_count = 2;

  // Elapsed time since last visit for this profile on this device. Uses the
  // local client-side timestamps. Can be unreliable as local times can
  // change in between different runs of Chrome.
  optional int64 elapsed_time_last_visit_secs = 3;

  // Number of times the suggestion was visited with the current input or
  // prefix of it for this profile on this device.
  // Discounted by a time-decaying factor with a 1-week half-life based on
  // the last visit timestamp.
  optional int32 shortcut_visit_count = 4;

  // Length of the shortest shortcut text.
  // Useful in comparison with |typed_length|.
  optional int32 shortest_shortcut_len = 5;

  // Elapsed time since last shortcut visit for this profile on this device.
  // Uses the local client-side timestamps. Can be unreliable as local times
  // can change in between different runs of Chrome.
  optional int64 elapsed_time_last_shortcut_visit_sec = 6;

  // URL points to the root page of a website, i.e., no query, path words,
  // or references after "/".
  optional bool is_host_only = 7;

  // Number of bookmarks for this profile with this URL.
  optional int32 num_bookmarks_of_url = 8;

  // Position of the first matched bookmark title term.
  // E.g. 4 for input 'x' and title "0123x56".
  // Not set when there is no match in the bookmark title.
  optional int32 first_bookmark_title_match_position = 9;

  // Total length of matched strings in the bookmark title. Can be larger
  // than the input text length. The input text is split by whitespaces, and
  // each input word is matched against the title separately. Their matching
  // lengths are summed. Similarly for other text matching signals below.
  // E.g. 9 for input "[abc] ijk [xyz]" and title "[abc] def [xyz] - [xyz]".
  // Set to 0 when there is no match in the title.
  optional int32 total_bookmark_title_match_length = 10;

  // Number of input terms matched by bookmark title. Take the maximum when
  // there are multiple matching bookmarks of this URL.
  // E.g., 1 for input "[a] b" and bookmark title "[a] book title".
  // Not set when the user does not have this URL bookmarked.
  optional int32 num_input_terms_matched_by_bookmark_title = 11;

  // Position of the first matched URL substring.
  // URL scheme or TLD matches are excluded, though those characters are
  // counted when assessing match position. E.g., 11 for 't' in
  // "https://hos[t].com" Not set when there is no URL match.
  optional int32 first_url_match_position = 12;

  // Total length of the matched URL strings. Can be longer than the input
  // string. E.g., Given input "ab abc" and url "abc.com/ab", total is 7 (4
  // for "ab" in "[ab]c.com/[ab]" and 3 for "abc" in
  // "[abc].com"). Set to 0 when there is no URL string match.
  optional int32 total_url_match_length = 13;

  // One word matches host at a word boundary. E.g., true for input "[h] a"
  // and "[h].com", Set to false when there are matches in the host but none
  // at word boundaries. E.g., false for input "a" and "h[a].com". Not set
  // when there is no host match.
  optional bool host_match_at_word_boundary = 14;

  // Total length of the matched host substrings.
  // Can be larger than the input text.
  // E.g., 3 for input "h a" and "[h]ost[aa].com".
  // Set to 0 when there is no host match.
  optional int32 total_host_match_length = 15;

  // Total length of the matched substrings in the path at word boundaries.
  // Can be larger than the input text. E.g. 3 for 'p' in
  // 'a.com/[p]ath_[p]ath/[p]ath'. Set to 0 when there are no such matches.
  optional int32 total_path_match_length = 16;

  // Total length of the matched substrings in the query_or_ref at word
  // boundaries. Can be larger than the input text.
  // E.g., 3 for input '[qu] [a]' and 'a.com/a?[qu]ery_[a]'.
  // Set to 0 when there are no such matches.
  optional int32 total_query_or_ref_match_length = 17;

  // Total length of the matched substrings in the page title at word
  // boundaries. Can be larger than the input text.
  // E.g., 2 for input "[a] [t] x" and page title "[a]bc [t]itle".
  // Set to 0 when there is no title match.
  optional int32 total_title_match_length = 18;

  // Has matches that are not in schemes (e.g., "https") or "www".
  optional bool has_non_scheme_www_match = 19;

  // Number of input terms matched by title.
  // E.g., 1 for input "[a] b" and title "[a] title [a]".
  // Useful in comparison with |num_typed_terms|.
  optional int32 num_input_terms_matched_by_title = 20;

  // Number of input terms matched by url.
  // Useful in comparison with |num_typed_terms|.
  // E.g., 1 for input "[a] b" and url "[a].com".
  optional int32 num_input_terms_matched_by_url = 21;

  // Length of url. E.g., 22 for "https://www.host.com/p".
  optional int32 length_of_url = 22;

  // Site engagement score for the site the URL is on. See
  // https://www.chromium.org/developers/design-documents/site-engagement/
  optional float site_engagement = 23;

  // True if url can be default match.
  // Currently, this requires single-term input, and match needs to begin
  // immediately after '', scheme, or 'http://www' in the URL, e.g., given
  // input 'w', true for "https://www.[w]sj.com" or "[w]ww.a.com", false for
  // "host[w].com". ref:
  // https://source.chromium.org/chromium/chromium/src/+/main:components/omnibox/browser/scored_history_match.cc;l=187?q=inline%20autocomplete
  optional bool allowed_to_be_default_match = 24;

  // Server-generated relevance score provided by the remote Suggest service
  // for this suggestion.
  // This signal contains the raw score received from the Suggest service
  // ("google:suggestrelevance") and does not reflect any client-side
  // adjustments (unlike the value of the `relevance` field above).
  // As such, this signal will take on the following values:
  //   - Non-zero: Search suggestion with `relevance_from_server` set to `true`
  //   - Zero:     Search suggestion with `relevance_from_server` set to `false`
  //               OR non-Search suggestion (obtained from any source)
  //   - Unset:    Otherwise (due to unexpected client-side behavior)
  optional int32 search_suggest_relevance = 25;

  // Whether this particular Search suggestion (or one of the duplicates deduped
  // under it) refers to an entity.
  // As such, this signal will take on the following values:
  //   - True:  Search suggestion (or one of its duplicates) refers to an entity
  //   - False: Neither this suggestion nor any of its duplicates refers to an
  //            entity
  //   - Unset: Otherwise (due to unexpected client-side behavior)
  optional bool is_search_suggest_entity = 26;

  // Whether this suggestion (or one of the duplicates deduped under it) is a
  // verbatim match with respect to the user's input (e.g. URL_WHAT_YOU_TYPED).
  // As such, this signal will take on the following values:
  //   - True:  Suggestion (or one of its duplicates) is a verbatim match
  //            (URL_WHAT_YOU_TYPED, SEARCH_WHAT_YOU_TYPED, etc.)
  //   - False: Neither this suggestion nor any of its duplicates is a verbatim
  //            match
  //   - Unset: Otherwise (due to unexpected client-side behavior)
  optional bool is_verbatim = 27;

  // Whether this suggestion (or one of the duplicates deduped under it) is a
  // NAVSUGGEST match.
  // As such, this signal will take on the following values:
  //   - True:  Suggestion (or one of its duplicates) is a NAVSUGGEST match
  //   - False: Neither this suggestion nor any of its duplicates is a
  //            NAVSUGGEST match
  //   - Unset: Otherwise (due to unexpected client-side behavior)
  optional bool is_navsuggest = 28;

  // Whether this suggestion (or one of the duplicates deduped under it) is a
  // SEARCH_SUGGEST_TAIL match.
  // As such, this signal will take on the following values:
  //   - True:  Suggestion (or one of its duplicates) is a SEARCH_SUGGEST_TAIL
  //            match
  //   - False: Neither this suggestion nor any of its duplicates is a
  //            SEARCH_SUGGEST_TAIL match
  //   - Unset: Otherwise (due to unexpected client-side behavior)
  optional bool is_search_suggest_tail = 29;

  // Whether this suggestion (or one of the duplicates deduped under it)
  // contains an answer with respect to the user's input (i.e.
  // `match.answer_type` != UNSPECIFIED in the client code).
  // As such, this signal will take on the following values:
  //   - True:  Suggestion (or one of its duplicates) contains an answer
  //   - False: Neither this suggestion nor any of its duplicates contains an
  //            answer
  //   - Unset: Otherwise (due to unexpected client-side behavior)
  optional bool is_answer_suggest = 30;

  // Whether this suggestion (or one of the duplicates deduped under it) is a
  // CALCULATOR match.
  // As such, this signal will take on the following values:
  //   - True:  Suggestion (or one of its duplicates) is a CALCULATOR match
  //   - False: Neither this suggestion nor any of its duplicates is a
  //            CALCULATOR match
  //   - Unset: Otherwise (due to unexpected client-side behavior)
  optional bool is_calculator_suggest = 31;
  // Whether this suggestion was shown in the NTP page context.
  // This signal is inclusive of situations in which an extension replaces
  // the NTP or the user has selected a default search engine other than Google.
  // More specifically, this signal will take on the following values:
  //   - True:  Suggestion was surfaced with one of the following page contexts:
  //              - NTP
  //              - NTP_REALBOX
  //              - INSTANT_NTP_WITH_OMNIBOX_AS_STARTING_FOCUS
  //   - False: Suggestion was surfaced with a different page context
  //   - Unset: Otherwise (due to unexpected client-side behavior)
  // NOTE: The "page contexts" listed above are sourced from the
  //       `PageClassification` enum in omnibox_event.proto.
  optional bool is_ntp_page_context = 32;
  // Whether this suggestion was shown when the user interacted with the Omnibox
  // on the Search Results Page (SRP) of their default search engine.
  // More specifically, this signal will take on the following values:
  //   - True:  Suggestion was surfaced with one of the following page contexts:
  //              - SEARCH_RESULT_PAGE_NO_SEARCH_TERM_REPLACEMENT
  //   - False: Suggestion was surfaced with a different page context
  //   - Unset: Otherwise (due to unexpected client-side behavior)
  // NOTE: The "page contexts" listed above are sourced from the
  //       `PageClassification` enum in omnibox_event.proto.
  optional bool is_srp_page_context = 33;
  // Whether this suggestion was shown in the Web page context.
  // More specifically, this signal will take on the following values:
  //   - True:  Suggestion was surfaced with one of the following page contexts:
  //              - OTHER
  //   - False: Suggestion was surfaced with a different page context
  //   - Unset: Otherwise (due to unexpected client-side behavior)
  // NOTE: The "page contexts" listed above are sourced from the
  //       `PageClassification` enum in omnibox_event.proto.
  optional bool is_web_page_context = 34;
  // Whether the host part of the `destination_url` associated with this
  // suggestion matches the host part of the URL for the page the user is
  // currently on.
  // As such, this signal will take on the following values:
  //   - True:  `destination_url` for this suggestion has the same host as
  //            the current page.
  //   - False: Current page is the NTP (any default search engine)
  //            OR `destination_url` for this suggestion is empty
  //            OR `destination_url` does not have the same host as the
  //            current page.
  //   - Unset: Otherwise (due to unexpected client-side behavior).
  optional bool destination_host_match_for_current_page = 35;
  // Whether selecting this suggestion would navigate to a page that's already
  // open in another tab in the current browser session (across all browser
  // windows on the current device). Cross-device tab sync does NOT have any
  // impact on the value of this signal.
  // As such, this signal will take on the following values:
  //   - True:  The destination associated with this suggestion is already open
  //            in another tab.
  //   - False: The destination associated with this suggestion is not open in
  //            another tab.
  //   - Unset: Otherwise (due to unexpected client-side behavior)
  optional bool has_open_tab_match = 36;
  // Whether this suggestion (or one of the duplicates deduped under it) is a
  // SEARCH_HISTORY match.
  // As such, this signal will take on the following values:
  //   - True:  Suggestion (or one of its duplicates) is a SEARCH_HISTORY match
  //   - False: Neither this suggestion nor any of its duplicates is a
  //            SEARCH_HISTORY match
  //   - Unset: Otherwise (due to unexpected client-side behavior)
  optional bool is_search_history = 37;
}