|
|
@ -36,11 +36,12 @@ actor_fetcher <- function(out, sent_dict = NULL, cores = 1, localhost = NULL, va
|
|
|
|
p_ids <- c(str_c(pid,'_f'),str_c(pid,'_s'))
|
|
|
|
p_ids <- c(str_c(pid,'_f'),str_c(pid,'_s'))
|
|
|
|
### Party ids including actors
|
|
|
|
### Party ids including actors
|
|
|
|
p_ids_a <- c(p_ids,str_c(pid,'_a'))
|
|
|
|
p_ids_a <- c(p_ids,str_c(pid,'_a'))
|
|
|
|
summarizer <- function (p_ids, out_row, merged_id) {
|
|
|
|
summarizer <- function (p_ids, dupe_df, merged_id) {
|
|
|
|
|
|
|
|
dupe_df <- dupe_df %>%
|
|
|
|
|
|
|
|
filter(ids %in% p_ids)
|
|
|
|
|
|
|
|
if (nrow(dupe_df) > 0) {
|
|
|
|
return(
|
|
|
|
return(
|
|
|
|
out_row %>%
|
|
|
|
dupe_df %>% summarise(
|
|
|
|
filter(ids %in% p_ids) %>%
|
|
|
|
|
|
|
|
summarise(
|
|
|
|
|
|
|
|
`_id` = first(`_id`),
|
|
|
|
`_id` = first(`_id`),
|
|
|
|
`_source.doctype` = first(`_source.doctype`),
|
|
|
|
`_source.doctype` = first(`_source.doctype`),
|
|
|
|
`_source.publication_date` = first(`_source.publication_date`),
|
|
|
|
`_source.publication_date` = first(`_source.publication_date`),
|
|
|
@ -56,6 +57,10 @@ actor_fetcher <- function(out, sent_dict = NULL, cores = 1, localhost = NULL, va
|
|
|
|
sentence_end = list(sort(unique(unlist(sentence_end))))
|
|
|
|
sentence_end = list(sort(unique(unlist(sentence_end))))
|
|
|
|
)
|
|
|
|
)
|
|
|
|
)
|
|
|
|
)
|
|
|
|
|
|
|
|
} else {
|
|
|
|
|
|
|
|
print(dupe_df$`_id`[[1]])
|
|
|
|
|
|
|
|
return(NULL)
|
|
|
|
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
party <- summarizer(p_ids, dupe_df, str_c(pid,'_mfs'))
|
|
|
|
party <- summarizer(p_ids, dupe_df, str_c(pid,'_mfs'))
|
|
|
|
party_actor <- summarizer(p_ids_a, dupe_df, str_c(pid,'_mfsa'))
|
|
|
|
party_actor <- summarizer(p_ids_a, dupe_df, str_c(pid,'_mfsa'))
|
|
|
|