WWWSnippetResponse

Was odp used in the snippets? DEPRECATED - this is no longer populated as of June 2017.

How tokens are rendered in generating title. Note: In rendering a title, the page title part and the site/host/domain title part can be flipped after initial rendering. The flip, if happened, may not be reflected in this field. That is, this field may contain the tokens in the original, pre-flip, order.

Character counts of snippet prefix, if any. E.g. section heading, list summary, byline date.

These fields were previously optional messages, but CL 2388905 moved them into the MessageSet. However, at this time, old Mustang binaries are still deployed in production and probably will be around for awhile. So, servers which need to talk to old binaries and need to use these fields need to check both the obsolete versions and the MessageSet version. Sitemap

If requested, the extra snippet info

Bitmap representing matches to leaf query terms within document (body section and url). It gets populated if query_matches_info = true.

A hash for duplicate detection. Two results with the same content can return different snippets if, for example, one has an ODP entry and the other does not. Gws can use this value reliably to filter duplicates. It is a hash of body only tidbits.

Meta tags

Document keywords

Will only be set when title_use_num_of_chars is false.

Tidbits chosen from the document body. Consists of repeated [begin, end) half-open ranges in token offsets from the beginning of the document.

Additional available data (message type ids)

Bitfield of snippet events and the various events. See SnippetEvents in ./snippets/defines.h for details on the contents.

Similar to num_tokens_skipped_by_in_doc_restrictions_in_scoring, but this number is for tokens skipped during printing, since printer owns its own token info manager which populates tokens.

Byline date for time sensitive snippets. Most of the time it originates from quality_timebased::SyntacticDate and it is floored to PT midnight.

List summary phrase for list pages.

How tokens are rendered in generating snippet.

If requested, the snippet generator may take note of query items present in an entry in an on-page table-of-contents (i.e. a series of on-page links to named anchors.) If so, these two fields contain the formatted and highlighted entry and the name of the on-page anchor it links to, respectively. This may be used by GWS to show a direct link to that named anchor on the page.

ManyboxData

DEPRECATED If requested, the orion entities

DEPRECATED Sentiment snippets

A bitvector of the tidbits in the snippet that are appropriate for the Quick Scroll (Findy) Chrome extension. Typically these contain "extra body matches", i.e., important query items not in the title.

A list of answers that had at least one hit in the document. Answers are identified by their index into the QRewriteQueryParams_AnswerSnippetInfo array (see //query/proto/query-params.proto).

Abbreviated bibliographic data from Google Scholar.

True if the title length is already adjusted for the browser width. If it is true, GWS needs not truncate the title.

Records features to analyze titles/snippets in ranklab.

True if the document represents a login page.

DocPreviewRestrictions for amp result.

The domain-level display name of the website, such as "Google" for google.com. See go/site-display-name for more details. As of Aug 2023, this field is being deprecated in favor of info.[AlternativeTitlesResponse].site_display_name_response field, which also contains host-level site display names with additional information.

DocPreviewRestrictions for canonical url.

Only for desktop web search. Please refer to Title.keep_original_title_and_populate_truncated_one for more details.

If requested the page number on which the snippet begins. (Only for documents such as PDFs where page numbers are well-defined.)

Did a negative query term match the meta description?

Number of tokens that is skipped because of in doc restrictions during scoring. This is an estimate, as the list of tokens is cached in TokenInfoManager. We should only monitor the cases where this number is too big or non-zero.

List of bitmaps representing matches to leaf query terms within each of the highlighted snippet text fragments. Consecutive bitmaps correspond to consecutive text fragments. It gets populated iff return_query_snippet_highlight_matches = true. Example: document body section: "This cafe has pet friendly patio." squery: (a (o dog :o pet :syn:general) friendly :o (o restaurant :o cafe :syn:general)) Let's assume the returned snippet text contains the whole document body section where two fragments get highlighted as follows: "This cafe has pet friendly patio." Then, the returned snippet_highlight_matches_bitmap[] list will have two bitmaps: bitmap[0] = <"cafe" -> leaf term with index 4> = {encoded: DenseEncode("00001"), size: 5} bitmap[1] = <"pet" and "friendly" -> leaf terms with indexes 1 and 2> = {encoded: DenseEncode("011"), size: 3}

LocalWWWInfo

Returned if want_long_structured_snippets. If present, caller should ignore the normal snippet.

List snippet data.

This field is never set.

Assume this is true unless we discover that the result doesn't match the the query, in which case this result is invalid despite having returned docinfo.

Additional data. Currently, this is used for sitelinks, localinfo, manybox, discussion metadata, richsnippets, similarpages and breadcrumbs.