Paper not downloaded and no error is thrown

[raffaem]

Consider the following code:

works_from_dois <- oa_fetch(
  entity = "works",
  doi = c("10.3233/978-1-61499-432-9-116", "10.1016/S0168-1605(01)00734-6"),
  verbose = TRUE
)

It will return just 1 work (the second one) without giving you any error/warning why the first work was not returned

[yjunechoe]

Thanks! I can replicate this. It seems to be a problem on the API end, where a mix of valid and invalid DOIs silently drop the invalid ones from the search without throwing an informative message/warning

oa_fetch(
  entity = "works",
  doi = "foobar"
)
#> Warning in oa_request(oa_query(filter = filter_i, multiple_id = multiple_id,
#> : No records found!
#> NULL

oa_fetch(
  entity = "works",
  doi = c("foobar", "10.1016/S0168-1605(01)00734-6")
)
#> # A tibble: 1 × 39
#>   id                    title display_name authorships doi   publication_date
#>   <chr>                 <chr> <chr>        <list>      <chr> <date>          
#> 1 https://openalex.org… Anti… Antimicrobi… <tibble>    http… 2002-03-25      
#> # ℹ 33 more variables: publication_year <int>, fwci <dbl>,
#> #   cited_by_count <int>, counts_by_year <list>, cited_by_api_url <chr>,
#> #   ids <list>, type <chr>, is_oa <lgl>, is_oa_anywhere <lgl>,
#> #   oa_status <chr>, oa_url <lgl>, any_repository_has_fulltext <lgl>,
#> #   source_display_name <chr>, source_id <chr>, issn_l <chr>,
#> #   host_organization <chr>, host_organization_name <chr>,
#> #   landing_page_url <chr>, referenced_works <list>, …

You can see the query we constructed [here] - this returns a valid response, so {openalexR} happily parses it.

oa_query(
  entity = "works",
  doi = c("10.3233/978-1-61499-432-9-116", "10.1016/S0168-1605(01)00734-6")
)
#> [1] "https://api.openalex.org/works?filter=doi%3A10.3233%2F978-1-61499-432-9-116%7C10.1016%2FS0168-1605%2801%2900734-6"

I agree this is an issue but I'm hesitant to rush to implement a check on our end in case this can be fixed back over at OpenAlex.

[raffaem]

Thank you.

Where do we go from here? We report it to OpenAlex?

[yjunechoe]

@trangdata do you know what's the appropriate channel for submitting issues w.r.t. the OpenAlex API? Do you think this issue falls under their scope of interest (or should we just handle it here)?

[trangdata]

Apologies for the delayed response: I'm on vacation until Feb 10.

Thanks for submitting the issue @raffaem. You can submit a ticket to the OpenAlex team at https://openalex.org/feedback

I agree with June that we should wait to hear from the OA team before fixing this issue in openalexR. This generic OR operator works for other filters as well, so I'm not sure a warning for anything that is missing would be helpful.

[rkrug]

Please note that doi has two meanings:

1. DOI to get an individual work. (https://docs.openalex.org/api-entities/works/get-a-single-work_. Here an error is raised if the DOI does not exist.
2. Doi is a filter (https://docs.openalex.org/api-entities/works/work-object#doi) here, the works corresponding top the filter are returned. If a filter does not match anything, nothing is returned and no error is returned

openalexR::oa_query() translates the doi argument into a filter, which is correct:
``https://api.openalex.org/works?filter=doi%3Afoobar`. This results in a warning if OpenAlex returns nothing (empty results) issued by openalexR which makes sense, while the second filter for doi does not result in an empty results, therefore no warning. This behavior makes perfect sense, as doi is translated into a filter.

Hope this clarifies.