abstracts

library(PubmedPlot)
library(jsonlite)
library(dplyr)
library(XML)


term <- '"Mendelian randomisation" [Title] OR "Mendelian randomization" [Title]'

search_url <- "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/esearch.fcgi"
search_params <- list(
    db = "pubmed",
    term = term,
    retmode = "json",
    usehistory = "y",
    retmax = 20000
)

search_response <- httr::GET(url = search_url, query = search_params)
search_content <- httr::content(search_response, "text")
search_result <- jsonlite::fromJSON(search_content)

pmids <- search_result$esearchresult$idlist

count <- search_result$esearchresult$count %>% as.numeric()
retmax <- search_result$esearchresult$retmax %>% as.numeric()
remainder <- count - retmax

if (remainder > 0) {
    search_params$retstart <- retmax
    search_response <- httr::GET(url = search_url, query = search_params)
    search_content <- httr::content(search_response, "text")
    search_result <- jsonlite::fromJSON(search_content)
    pmids <- c(pmids, search_result$esearchresult$idlist)
}

# length(pmids)

efetch_url <- "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi"




# Prepare the body of the POST request for XML output
efetch_params <- list(
  db = "pubmed",
  id = paste(pmids, collapse = ","),
  rettype = "abstract",
  retmode = "xml"
)

# Query how many pmids

# Make the POST request to fetch abstracts
efetch_response <- httr::POST(url = efetch_url, body = efetch_params, encode = "form")
efetch_content <- httr::content(efetch_response, "text", encoding = "UTF-8")


# Parse the XML content
doc <- XML::xmlParse(efetch_content)
xmltop <- XML::xmlRoot(doc)
# xmlSize(xmltop)
# xmlName(xmltop[[1]][[1]][[1]])
# xmlValue(xmltop[[1]][[]][["PMID"]])

pub_dates <- xpathApply(doc, '//PubmedArticle', \(x) {
    dplyr::tibble(
        pmid = xmlValue(x[[1]][["PMID"]]),
        ab = xmlValue(x[[1]][["Article"]][["Abstract"]]),
        pub_date = lubridate::ymd(
            paste(
                xmlValue(x[["PubmedData"]][["History"]][["PubMedPubDate"]][["Year"]]),
                xmlValue(x[["PubmedData"]][["History"]][["PubMedPubDate"]][["Month"]]),
                xmlValue(x[["PubmedData"]][["History"]][["PubMedPubDate"]][["Day"]])
            )
        ),
        title = xmlValue(x[[1]][["Article"]][["ArticleTitle"]]),
        journal_issn = xmlValue(x[[1]][["Article"]][["Journal"]][["ISSN"]]),
        journal = xmlValue(x[[1]][["Article"]][["Journal"]][["Title"]]),
        author_affil = xmlValue(x[[1]][["Article"]][["AuthorList"]][[1]][["AffiliationInfo"]])
    )
}) %>% bind_rows()

jsonlite::write_json(pub_dates, path="pubmed.json", pretty = TRUE)

a <- PubmedPlot::search_term_by_year('"Mendelian randomisation" [Title] OR "Mendelian randomization" [Title]', 2003:lubridate::year(Sys.Date()))
jsonlite::write_json(a, path="pubmed.json", pretty = TRUE)