In benjaminguinaudeau/librarrry:

knitr::opts_chunk$set(echo = TRUE)

pacman::p_load(tidyverse, librarrry)
load("../librarian/data/publications.Rdata")

publications %>%
  slice(1:100) %>%
  select(bibrecord) %>%
  unnest %>%
  select(author_group) %>%
  list %>%
  View

test$authors[[4]]

test <- publications %>% 
    slice(1:100) 
.x <- test$bibrecord[[3]]

scopus_clean_publications <- function(publications){
  test <- publications %>% 
    slice(1:100) %>%
    mutate(
      authors = bibrecord %>% 
        map(~{
          if(is.null(.x$author_group)) return(NULL)
          #if(length(.x$author_group) > 1){
            tmp <- .x$author_group %>% 
              map_df(~{
                if("V1" %in% names(.x)){
                  return(
                    .x %>% 
                      select(matches("V\\d")) %>%
                      gather_group %>%
                      unnest
                      )
                } else {
                  .x %>%
                  gather_group %>% 
                  select(author) %>% 
                  unnest
                }
              })

            if(!any(colnames(tmp) == "auid"))  return(NULL)

            tmp <- tmp %>% 
              gather_group() %>% 
              drop_na(auid) %>% 
              arrange(seq) %>% 
              map_select(c("seq", "ce_given_name", "ce_surname", "auid")) %>%
              set_names(c("seq", "first_name", "last_name", "auid"))
            return(tmp)
          # } else {
          #   tmp <- .x$author_group[[1]][,length(.x$author_group[[1]])] %>%
          #     unnest 
          #   
          #   if(!any(colnames(tmp) == "auid"))  return(NULL)
          #   tmp <- tmp %>%               
          #     map_select(c("seq", "ce_given_name", "ce_surname", "auid")) %>% 
          #     set_names(c("seq", "first_name", "last_name", "auid"))
          #   return(tmp)
          # }
        })
    ) %>%
    mutate(source_list = bibrecord %>%
             map( ~ {
               .x$source[[1]] %>%
                 map_select(c("codencode", "srcid")) %>%
                 set_names(c("source_short", "source_list")) %>% 
                 unnest_pos(source_list)
             })
    ) %>%
    unnest(source_list) %>%
    mutate(classification = bibrecord %>%
             map( ~ {
               .x$enhancement %>% 
                 listify() %>%
                 gather_group()
             })
    ) %>%
    mutate(ids = bibrecord %>%
             map( ~ {
               tmp <- .x$itemidlist[[1]] %>% 
                 dplyr::select(matches("V\\d")) %>%
                 gather_group() 

               if(!(any(colnames(tmp) == "V"))) return(NULL)

               tmp %>% 
                 mutate(is_tibble = map_lgl(V, is.data.frame)) %>%
                 filter(is_tibble) %>% 
                 unnest %>%
                 select(-is_tibble) %>% 
                 map_select(c("x", "idtype")) %>% 
                 set_names(c("id_type", "id"))
             })
    ) %>%
    mutate(ref = bibrecord %>%
             map(~{
               if(!any(colnames(.x) == "bibliography")) return(NULL)

               tmp <- .x$bibliography[[1]]$reference[[1]] %>% 
                 gather_group() %>% 
                 unnest

               if(length(tmp$itemid) == 0){return(tibble(NA))}
               tmp %>% 
                 select(itemid) %>%
                 mutate(itemid = itemid %>% 
                          map(~{
                            if(is.null(.x)){
                              tibble(NA)
                            } else {
                              .x
                            }
                          })) %>%
                 unnest %>%
                 map_select(c("x")) %>%
                 unlist %>%
                 as.character()
             })
    ) %>%
    mutate(lang = xml_lang %>% map_chr(~ifelse(is.null(.x[1]), NA_character_, as.character(.x[1])))) %>%
    mutate(publisher = dc_publisher %>% map_chr(~ifelse(is.null(.x[1]), NA_character_, as.character(.x[1])))) %>%
    mutate(keywords = author_keyword %>%
             map(~{
               .x %>% 
                 gather_group %>% 
                 map_select("x") %>% 
                 set_names("keyword")
             })
    ) %>%
    mutate(isbns = prism_isbn %>% 
             map(~{
               if(is.null(.x)){
                 return(tibble(NA))
               } else {
                 .x %>%
                   map_select(c("x", "x1")) %>% 
                   set_names(c("online_isbn", "print_isbn"))
               }
             })
    ) %>% 
    mutate(scopus_id = ids %>% map_chr(~{
      if(is.null(.x)) return(NA_character_)
      .x %>% filter(id_type == "SCP") %>% 
        .$id %>% .[1] %>% as.character()
    })
    ) %>% 
    dplyr::select(
      scopus_id,
      authors,
      classification,
      source_short,
      source_id,
      ids,
      ref,
      lang,
      keywords,
      publisher,
      isbns
    ) %>%
    unnest_pos(source_id)
}

benjaminguinaudeau/librarrry documentation built on May 3, 2019, 7:39 p.m.