Attention per source over time

Splits the matching stories by source and by time bucket, which is what the "attention over time by source" view of the web app shows.

Usage

mc_count_by_source(
  query,
  start_date,
  end_date,
  collection_id = NULL,
  source_id = NULL,
  interval = c("week", "day", "month", "year"),
  platform = mc_platforms,
  .token = NULL
)

Arguments

query

the search string, in Elasticsearch/Lucene syntax. "*" matches everything.

start_date, end_date

the period to search, as Date objects or "YYYY-MM-DD" strings. Both are required.

collection_id, source_id

ids to search within, as returned by mc_collections() and mc_sources(). At least one of the two is required. Vectors are allowed.

interval

size of the time buckets: "day", "week", "month" or "year".

platform

which archive to query. "onlinenews-mediacloud" is the Media Cloud index; "onlinenews-waybackmachine" searches the Wayback Machine instead.

.token

a token, or path to a saved token file, to use instead of the stored one.

Value

A tibble with one row per source and bucket: media_name, interval, bucket, matching_stories, total_stories and ratio.

Details

The server refuses queries whose number of sources times number of buckets is too large, so narrow the period or use a coarser interval when searching a big collection. This endpoint costs four quota hits.

Examples

if (FALSE) { # \dontrun{
mc_count_by_source("klima", "2026-08-01", "2026-08-31",
                   collection_id = 34412409, interval = "week")
} # }