fastpaper/
registry.rs

1//! The one place that maps a source name to everything the commands need:
2//! its base URL, what it can do, and which functions to call.
3//!
4//! Sources stay plain free functions in their own files — this table holds
5//! function pointers rather than a trait, so no source has to implement
6//! anything. Where a source's signature does not yet match the table's, a
7//! three-line adapter bridges it; those adapters disappear as each source is
8//! migrated.
9
10use clap::ValueEnum;
11
12use crate::download;
13use crate::sources::{self, Capabilities, Direction, Paper, SearchCaps, SearchQuery};
14
15#[derive(ValueEnum, Clone, Copy, Debug, PartialEq, Eq)]
16pub enum Source {
17    Arxiv,
18    Biorxiv,
19    Medrxiv,
20    Pubmed,
21    Pmc,
22    Europepmc,
23    Scholar,
24    Xueshu,
25    Semantic,
26    Crossref,
27    Openalex,
28    Dblp,
29    Core,
30    Openaire,
31    Doaj,
32    Unpaywall,
33    Zenodo,
34    Hal,
35}
36
37/// Every source, in the order `fastpaper sources` lists them.
38pub const ALL: &[Source] = &[
39    Source::Arxiv,
40    Source::Biorxiv,
41    Source::Medrxiv,
42    Source::Pubmed,
43    Source::Pmc,
44    Source::Europepmc,
45    Source::Scholar,
46    Source::Xueshu,
47    Source::Semantic,
48    Source::Crossref,
49    Source::Openalex,
50    Source::Dblp,
51    Source::Core,
52    Source::Openaire,
53    Source::Doaj,
54    Source::Unpaywall,
55    Source::Zenodo,
56    Source::Hal,
57];
58
59pub struct SourceEntry {
60    pub name: &'static str,
61    pub caps: Capabilities,
62    /// Env var that overrides `default_base`.
63    pub env_var: &'static str,
64    pub default_base: &'static str,
65    /// Separate host for PDF downloads where the API and the files differ.
66    pub pdf_env_var: Option<&'static str>,
67    pub pdf_default_base: Option<&'static str>,
68    pub search: Option<fn(&str, &SearchQuery) -> Result<Vec<Paper>, String>>,
69    pub get: Option<fn(&str, &str) -> Result<Option<Paper>, String>>,
70    pub pdf: Option<fn(&str, &str) -> Result<Vec<u8>, String>>,
71    pub cite: Option<fn(&str, &str, Direction, u32) -> Result<Vec<Paper>, String>>,
72}
73
74impl Source {
75    pub fn name(&self) -> &'static str {
76        self.entry().name
77    }
78
79    pub fn caps(&self) -> Capabilities {
80        self.entry().caps
81    }
82
83    /// Base URL for metadata requests, overridable via this source's env var.
84    pub fn base_url(&self) -> String {
85        let e = self.entry();
86        std::env::var(e.env_var).unwrap_or_else(|_| e.default_base.to_string())
87    }
88
89    /// Base URL for PDF downloads.
90    ///
91    /// Sources whose files live on a different host than their API get their
92    /// own override; failing that an explicitly-set general override still
93    /// applies, so pointing one variable at a test server redirects both.
94    /// Only the *defaults* are kept separate.
95    pub fn pdf_base_url(&self) -> String {
96        let e = self.entry();
97        if let Some(var) = e.pdf_env_var {
98            if let Ok(value) = std::env::var(var) {
99                return value;
100            }
101        }
102        if let Ok(value) = std::env::var(e.env_var) {
103            return value;
104        }
105        e.pdf_default_base.unwrap_or(e.default_base).to_string()
106    }
107
108    /// Resolve a source from a CLI token, for the `[source] <id>` forms where
109    /// clap cannot type the argument as an enum.
110    pub fn from_name(token: &str) -> Option<Source> {
111        ALL.iter().copied().find(|s| s.name() == token)
112    }
113
114    pub fn entry(&self) -> &'static SourceEntry {
115        match self {
116            Source::Arxiv => &ARXIV,
117            Source::Biorxiv => &BIORXIV,
118            Source::Medrxiv => &MEDRXIV,
119            Source::Pubmed => &PUBMED,
120            Source::Pmc => &PMC,
121            Source::Europepmc => &EUROPEPMC,
122            Source::Scholar => &SCHOLAR,
123            Source::Xueshu => &XUESHU,
124            Source::Semantic => &SEMANTIC,
125            Source::Crossref => &CROSSREF,
126            Source::Openalex => &OPENALEX,
127            Source::Dblp => &DBLP,
128            Source::Core => &CORE,
129            Source::Openaire => &OPENAIRE,
130            Source::Doaj => &DOAJ,
131            Source::Unpaywall => &UNPAYWALL,
132            Source::Zenodo => &ZENODO,
133            Source::Hal => &HAL,
134        }
135    }
136}
137
138// ── adapters ────────────────────────────────────
139
140/// Unpaywall returns a bare `Paper`; a missing DOI surfaces as an error there.
141fn g_unpaywall(base: &str, id: &str) -> Result<Option<Paper>, String> {
142    sources::unpaywall::lookup_doi(base, id).map(Some)
143}
144
145// ── entries ─────────────────────────────────────
146
147static ARXIV: SourceEntry = SourceEntry {
148    name: "arxiv",
149    caps: Capabilities {
150        search: Some(SearchCaps {
151            offset: true,
152            sort: true,
153            year: true,
154            date_range: true,
155            author: true,
156            field: true,
157            // Every arXiv paper is freely readable, so the filter is
158            // trivially satisfied rather than unsupported.
159            open_access: true,
160            patents: false,
161        }),
162        get: true,
163        download: true,
164        cite: false,
165        max_limit: Some(2000),
166        notes: "--sort citations is unavailable: arXiv publishes no citation counts",
167    },
168    env_var: "FASTPAPER_ARXIV_URL",
169    default_base: "https://export.arxiv.org",
170    pdf_env_var: Some("FASTPAPER_ARXIV_PDF_URL"),
171    pdf_default_base: Some("https://arxiv.org"),
172    search: Some(sources::arxiv::search),
173    get: Some(sources::arxiv::get_by_id),
174    pdf: Some(download::pdf_bytes_arxiv),
175    cite: None,
176};
177
178static BIORXIV: SourceEntry = SourceEntry {
179    name: "biorxiv",
180    caps: Capabilities {
181        search: Some(SearchCaps {
182            offset: true,
183            sort: false,
184            year: true,
185            date_range: true,
186            author: false,
187            field: false,
188            // Every preprint here is freely readable.
189            open_access: true,
190            patents: false,
191        }),
192        get: true,
193        download: true,
194        cite: false,
195        max_limit: None,
196        notes: "no keyword search API: browses a date window and matches the \
197                keyword locally, so --after/--before/--year decide what is searched",
198    },
199    env_var: "FASTPAPER_BIORXIV_URL",
200    default_base: "https://api.biorxiv.org",
201    pdf_env_var: Some("FASTPAPER_BIORXIV_DL_URL"),
202    pdf_default_base: Some("https://www.biorxiv.org"),
203    search: Some(sources::biorxiv::search),
204    get: Some(sources::biorxiv::get_by_id),
205    pdf: Some(download::pdf_bytes_biorxiv),
206    cite: None,
207};
208
209static MEDRXIV: SourceEntry = SourceEntry {
210    name: "medrxiv",
211    caps: Capabilities {
212        search: Some(SearchCaps {
213            offset: true,
214            sort: false,
215            year: true,
216            date_range: true,
217            author: false,
218            field: false,
219            // Every preprint here is freely readable.
220            open_access: true,
221            patents: false,
222        }),
223        get: true,
224        download: false,
225        cite: false,
226        max_limit: None,
227        notes: "no keyword search API: browses a date window and matches the \
228                keyword locally, so --after/--before/--year decide what is \
229                searched; PDF downloads are blocked by medRxiv (HTTP 403)",
230    },
231    env_var: "FASTPAPER_MEDRXIV_URL",
232    default_base: "https://api.biorxiv.org",
233    pdf_env_var: Some("FASTPAPER_MEDRXIV_DL_URL"),
234    pdf_default_base: Some("https://www.medrxiv.org"),
235    search: Some(sources::medrxiv::search),
236    get: Some(sources::medrxiv::get_by_id),
237    pdf: None,
238    cite: None,
239};
240
241static PUBMED: SourceEntry = SourceEntry {
242    name: "pubmed",
243    caps: Capabilities {
244        search: Some(SearchCaps {
245            offset: true,
246            sort: true,
247            year: true,
248            date_range: true,
249            author: true,
250            // PubMed classifies by MeSH heading, which is not the same notion
251            // as a field of study and would mislead under this flag.
252            field: false,
253            open_access: false,
254            patents: false,
255        }),
256        get: true,
257        download: false,
258        cite: false,
259        max_limit: Some(10000),
260        notes: "metadata only, for full text try pmc; --sort citations unavailable",
261    },
262    env_var: "FASTPAPER_PUBMED_URL",
263    default_base: "https://eutils.ncbi.nlm.nih.gov",
264    pdf_env_var: None,
265    pdf_default_base: None,
266    search: Some(sources::pubmed::search),
267    get: Some(sources::pubmed::get_by_pmid),
268    pdf: None,
269    cite: None,
270};
271
272static PMC: SourceEntry = SourceEntry {
273    name: "pmc",
274    caps: Capabilities {
275        search: Some(SearchCaps {
276            offset: true,
277            sort: true,
278            year: true,
279            date_range: true,
280            author: true,
281            field: false,
282            open_access: true,
283            patents: false,
284        }),
285        get: true,
286        download: true,
287        cite: false,
288        max_limit: Some(10000),
289        notes: "--sort citations unavailable; PDFs come from the OA subset only",
290    },
291    env_var: "FASTPAPER_PMC_URL",
292    default_base: "https://eutils.ncbi.nlm.nih.gov",
293    pdf_env_var: Some("FASTPAPER_PMC_DL_URL"),
294    // The PMC Cloud Service on AWS Open Data; see download::pdf_bytes_pmc for
295    // why the article page's /pdf/ URL and the OA Web Service's ftp:// links
296    // are both unusable.
297    pdf_default_base: Some("https://pmc-oa-opendata.s3.amazonaws.com"),
298    search: Some(sources::pmc::search),
299    get: Some(sources::pmc::get_by_pmc_id),
300    pdf: Some(download::pdf_bytes_pmc),
301    cite: None,
302};
303
304static EUROPEPMC: SourceEntry = SourceEntry {
305    name: "europepmc",
306    caps: Capabilities {
307        search: Some(SearchCaps {
308            offset: true,
309            sort: true,
310            year: true,
311            date_range: true,
312            author: true,
313            field: false,
314            open_access: true,
315            patents: true,
316        }),
317        get: true,
318        download: true,
319        cite: false,
320        max_limit: Some(1000),
321        notes: "orders newest/most-cited first only, --order asc unavailable; \
322                --offset must be a multiple of -n; PDFs come from the open \
323                access subset",
324    },
325    env_var: "FASTPAPER_EUROPEPMC_URL",
326    default_base: "https://www.ebi.ac.uk",
327    pdf_env_var: None,
328    pdf_default_base: None,
329    search: Some(sources::europepmc::search),
330    get: Some(sources::europepmc::get_by_id),
331    pdf: Some(download::pdf_bytes_europepmc),
332    cite: None,
333};
334
335static SCHOLAR: SourceEntry = SourceEntry {
336    name: "scholar",
337    caps: Capabilities {
338        search: Some(SearchCaps {
339            offset: true,
340            ..SearchCaps::BASIC
341        }),
342        get: false,
343        download: false,
344        cite: false,
345        max_limit: None,
346        notes: "HTML scraping, no official API; brittle and rate-limited",
347    },
348    env_var: "FASTPAPER_SCHOLAR_URL",
349    default_base: "https://scholar.google.com",
350    pdf_env_var: None,
351    pdf_default_base: None,
352    search: Some(sources::scholar::search),
353    get: None,
354    pdf: None,
355    cite: None,
356};
357
358static XUESHU: SourceEntry = SourceEntry {
359    name: "xueshu",
360    caps: Capabilities {
361        search: Some(SearchCaps {
362            offset: true,
363            patents: true,
364            ..SearchCaps::BASIC
365        }),
366        get: false,
367        download: false,
368        cite: false,
369        max_limit: None,
370        notes: "unofficial endpoint; subject to bot detection; --patents filters \
371                locally, so it pages further to fill -n",
372    },
373    env_var: "FASTPAPER_XUESHU_URL",
374    default_base: "https://xueshu.baidu.com",
375    pdf_env_var: None,
376    pdf_default_base: None,
377    search: Some(sources::xueshu::search),
378    get: None,
379    pdf: None,
380    cite: None,
381};
382
383static SEMANTIC: SourceEntry = SourceEntry {
384    name: "semantic",
385    caps: Capabilities {
386        search: Some(SearchCaps {
387            offset: true,
388            sort: true,
389            year: true,
390            date_range: true,
391            // Paper search exposes no author parameter.
392            author: false,
393            field: true,
394            open_access: true,
395            patents: false,
396        }),
397        get: true,
398        download: true,
399        cite: true,
400        max_limit: Some(100),
401        notes: "--sort switches to the bulk endpoint, which pages by token, so it \
402                cannot be combined with --offset; set SEMANTIC_SCHOLAR_API_KEY to \
403                avoid heavy throttling",
404    },
405    env_var: "FASTPAPER_SEMANTIC_URL",
406    default_base: "https://api.semanticscholar.org",
407    pdf_env_var: None,
408    pdf_default_base: None,
409    search: Some(sources::semantic::search),
410    get: Some(sources::semantic::get_by_id),
411    pdf: Some(download::pdf_bytes_semantic),
412    cite: Some(sources::semantic::cite),
413};
414
415static CROSSREF: SourceEntry = SourceEntry {
416    name: "crossref",
417    caps: Capabilities {
418        search: Some(SearchCaps {
419            offset: true,
420            sort: true,
421            year: true,
422            date_range: true,
423            author: true,
424            // Crossref indexes registered metadata; it classifies neither
425            // subject area nor open access status.
426            field: false,
427            open_access: false,
428            patents: false,
429        }),
430        get: true,
431        download: false,
432        cite: false,
433        max_limit: Some(1000),
434        notes: "metadata only, no PDF links; --offset caps at 10000",
435    },
436    env_var: "FASTPAPER_CROSSREF_URL",
437    default_base: "https://api.crossref.org",
438    pdf_env_var: None,
439    pdf_default_base: None,
440    search: Some(sources::crossref::search),
441    get: Some(sources::crossref::get_by_doi),
442    pdf: None,
443    cite: None,
444};
445
446static OPENALEX: SourceEntry = SourceEntry {
447    name: "openalex",
448    caps: Capabilities {
449        search: Some(SearchCaps {
450            offset: true,
451            sort: true,
452            year: true,
453            date_range: true,
454            author: true,
455            field: true,
456            open_access: true,
457            patents: false,
458        }),
459        get: true,
460        download: false,
461        cite: true,
462        max_limit: Some(100),
463        notes: "usage-metered since 2026-02, set OPENALEX_API_KEY for the larger free tier; \
464                --field takes a concept ID (e.g. C154945302), not a name; \
465                --offset must be a multiple of -n",
466    },
467    env_var: "FASTPAPER_OPENALEX_URL",
468    default_base: "https://api.openalex.org",
469    pdf_env_var: None,
470    pdf_default_base: None,
471    search: Some(sources::openalex::search),
472    get: Some(sources::openalex::get_by_id),
473    pdf: None,
474    cite: Some(sources::openalex::cite),
475};
476
477static DBLP: SourceEntry = SourceEntry {
478    name: "dblp",
479    caps: Capabilities {
480        search: Some(SearchCaps {
481            offset: true,
482            ..SearchCaps::BASIC
483        }),
484        get: false,
485        download: false,
486        cite: false,
487        max_limit: Some(1000),
488        notes: "computer science only, metadata only; the API takes a query and \
489                paging, nothing else",
490    },
491    env_var: "FASTPAPER_DBLP_URL",
492    // dblp.org's search API has been returning HTTP 500 for every query.
493    default_base: "https://dblp.uni-trier.de",
494    pdf_env_var: None,
495    pdf_default_base: None,
496    search: Some(sources::dblp::search),
497    get: None,
498    pdf: None,
499    cite: None,
500};
501
502static CORE: SourceEntry = SourceEntry {
503    name: "core",
504    caps: Capabilities {
505        search: Some(SearchCaps {
506            offset: true,
507            year: true,
508            author: true,
509            ..SearchCaps::BASIC
510        }),
511        get: true,
512        download: true,
513        cite: false,
514        max_limit: Some(100),
515        notes: "CORE_API_KEY is effectively required -- anonymous requests are \
516                throttled to 429; --author matches authors.name",
517    },
518    env_var: "FASTPAPER_CORE_URL",
519    default_base: "https://api.core.ac.uk",
520    pdf_env_var: None,
521    pdf_default_base: None,
522    search: Some(sources::core::search),
523    get: Some(sources::core::get_by_id),
524    pdf: Some(download::pdf_bytes_core),
525    cite: None,
526};
527
528static OPENAIRE: SourceEntry = SourceEntry {
529    name: "openaire",
530    caps: Capabilities {
531        search: Some(SearchCaps {
532            offset: true,
533            sort: true,
534            year: true,
535            date_range: true,
536            author: true,
537            field: true,
538            open_access: true,
539            patents: false,
540        }),
541        get: true,
542        download: false,
543        cite: false,
544        max_limit: None,
545        notes: "--field takes an OpenAIRE field-of-science value; an unrecognised \
546                one is answered with the allowed list",
547    },
548    env_var: "FASTPAPER_OPENAIRE_URL",
549    default_base: "https://api.openaire.eu",
550    pdf_env_var: None,
551    pdf_default_base: None,
552    search: Some(sources::openaire::search),
553    get: Some(sources::openaire::get_by_id),
554    pdf: None,
555    cite: None,
556};
557
558static DOAJ: SourceEntry = SourceEntry {
559    name: "doaj",
560    caps: Capabilities {
561        search: Some(SearchCaps {
562            offset: true,
563            // bibjson.year is not a sortable field and created_date is the
564            // indexing date, not the publication date.
565            sort: false,
566            year: true,
567            // DOAJ records carry a year, not a date, so a day-granular range
568            // would have to be silently widened.
569            date_range: false,
570            author: true,
571            field: false,
572            // Everything in the Directory of Open Access Journals is open
573            // access, so the filter is trivially satisfied.
574            open_access: true,
575            patents: false,
576        }),
577        get: true,
578        download: false,
579        cite: false,
580        max_limit: Some(100),
581        notes: "year granularity only, use --year rather than --after/--before; \
582                no PDFs -- its fulltext links point at publisher landing pages, \
583                not files",
584    },
585    env_var: "FASTPAPER_DOAJ_URL",
586    default_base: "https://doaj.org",
587    pdf_env_var: None,
588    pdf_default_base: None,
589    search: Some(sources::doaj::search),
590    get: Some(sources::doaj::get_by_id),
591    pdf: None,
592    cite: None,
593};
594
595static UNPAYWALL: SourceEntry = SourceEntry {
596    name: "unpaywall",
597    caps: Capabilities {
598        // Unpaywall is a DOI lookup service, not a search engine.
599        search: None,
600        get: true,
601        download: false,
602        cite: false,
603        max_limit: None,
604        notes: "DOI lookup only; requires a real address in FASTPAPER_EMAIL",
605    },
606    env_var: "FASTPAPER_UNPAYWALL_URL",
607    default_base: "https://api.unpaywall.org",
608    pdf_env_var: None,
609    pdf_default_base: None,
610    search: None,
611    get: Some(g_unpaywall),
612    pdf: None,
613    cite: None,
614};
615
616static ZENODO: SourceEntry = SourceEntry {
617    name: "zenodo",
618    caps: Capabilities {
619        search: Some(SearchCaps {
620            offset: true,
621            sort: true,
622            year: true,
623            date_range: true,
624            author: true,
625            field: false,
626            open_access: true,
627            patents: false,
628        }),
629        get: true,
630        download: true,
631        cite: false,
632        max_limit: Some(25),
633        notes: "anonymous callers get 25 results per request; --sort citations \
634                unavailable; --offset must be a multiple of -n",
635    },
636    env_var: "FASTPAPER_ZENODO_URL",
637    default_base: "https://zenodo.org",
638    pdf_env_var: None,
639    pdf_default_base: None,
640    search: Some(sources::zenodo::search),
641    get: Some(sources::zenodo::get_by_id),
642    pdf: Some(download::pdf_bytes_zenodo),
643    cite: None,
644};
645
646static HAL: SourceEntry = SourceEntry {
647    name: "hal",
648    caps: Capabilities {
649        search: Some(SearchCaps {
650            offset: true,
651            sort: true,
652            year: true,
653            // publicationDateY_i has year granularity only, so a day-granular
654            // range could only be honoured by widening it.
655            date_range: false,
656            author: true,
657            field: true,
658            open_access: true,
659            patents: false,
660        }),
661        get: true,
662        download: true,
663        cite: false,
664        max_limit: Some(10000),
665        notes: "year granularity only, use --year rather than --after/--before; \
666                --field takes a HAL domain code such as sdv or info",
667    },
668    env_var: "FASTPAPER_HAL_URL",
669    default_base: "https://api.archives-ouvertes.fr",
670    pdf_env_var: None,
671    pdf_default_base: None,
672    search: Some(sources::hal::search),
673    get: Some(sources::hal::get_by_id),
674    pdf: Some(download::pdf_bytes_hal),
675    cite: None,
676};
677
678#[cfg(test)]
679mod tests {
680    use super::*;
681
682    #[test]
683    fn every_source_has_a_matching_entry() {
684        for s in ALL {
685            assert_eq!(
686                s.name(),
687                s.entry().name,
688                "entry for {:?} is wired to the wrong record",
689                s
690            );
691        }
692    }
693
694    #[test]
695    fn source_names_are_unique() {
696        let mut names: Vec<_> = ALL.iter().map(|s| s.name()).collect();
697        names.sort_unstable();
698        let before = names.len();
699        names.dedup();
700        assert_eq!(before, names.len(), "duplicate source name in ALL");
701    }
702
703    #[test]
704    fn from_name_round_trips() {
705        for s in ALL {
706            assert_eq!(Source::from_name(s.name()), Some(*s));
707        }
708    }
709
710    #[test]
711    fn from_name_rejects_unknown() {
712        assert_eq!(Source::from_name("nope"), None);
713        assert_eq!(Source::from_name("10.1038/nature12373"), None);
714    }
715
716    #[test]
717    fn download_capable_sources_have_a_pdf_fn() {
718        for s in ALL {
719            let e = s.entry();
720            assert_eq!(
721                e.caps.download,
722                e.pdf.is_some(),
723                "{} declares download={} but pdf fn present={}",
724                e.name,
725                e.caps.download,
726                e.pdf.is_some()
727            );
728        }
729    }
730
731    #[test]
732    fn get_capable_sources_have_a_get_fn() {
733        for s in ALL {
734            let e = s.entry();
735            assert_eq!(
736                e.caps.get,
737                e.get.is_some(),
738                "{} declares get={} but get fn present={}",
739                e.name,
740                e.caps.get,
741                e.get.is_some()
742            );
743        }
744    }
745
746    // --patents means "patents only". Just two sources can honour that:
747    // europepmc filters natively with SRC:PAT, xueshu filters locally on its
748    // type code. Google Scholar can only *widen* to include patents, never
749    // narrow to them, so it must not claim the flag.
750    #[test]
751    fn only_europepmc_and_xueshu_take_patents() {
752        for s in ALL {
753            let declared = s
754                .caps()
755                .search
756                .is_some_and(|caps| caps.supports("--patents"));
757            let expected = matches!(s, Source::Europepmc | Source::Xueshu);
758            assert_eq!(
759                declared,
760                expected,
761                "{} declares --patents support = {}",
762                s.name(),
763                declared
764            );
765        }
766    }
767
768    #[test]
769    fn search_capable_sources_have_a_search_fn() {
770        for s in ALL {
771            let e = s.entry();
772            assert_eq!(
773                e.caps.search.is_some(),
774                e.search.is_some(),
775                "{} declares search={} but search fn present={}",
776                e.name,
777                e.caps.search.is_some(),
778                e.search.is_some()
779            );
780        }
781    }
782
783    #[test]
784    #[serial_test::serial]
785    fn base_url_prefers_env_override() {
786        unsafe { std::env::set_var("FASTPAPER_ARXIV_URL", "http://localhost:1234") };
787        assert_eq!(Source::Arxiv.base_url(), "http://localhost:1234");
788        unsafe { std::env::remove_var("FASTPAPER_ARXIV_URL") };
789        assert_eq!(Source::Arxiv.base_url(), "https://export.arxiv.org");
790    }
791
792    #[test]
793    fn pdf_base_url_falls_back_to_base_url() {
794        // semantic serves PDFs off the same host it serves metadata from
795        assert_eq!(Source::Semantic.pdf_base_url(), Source::Semantic.base_url());
796    }
797
798    #[test]
799    #[serial_test::serial]
800    fn pdf_base_url_defaults_to_the_file_host_not_the_api_host() {
801        unsafe { std::env::remove_var("FASTPAPER_ARXIV_URL") };
802        unsafe { std::env::remove_var("FASTPAPER_ARXIV_PDF_URL") };
803        assert_eq!(Source::Arxiv.base_url(), "https://export.arxiv.org");
804        assert_eq!(Source::Arxiv.pdf_base_url(), "https://arxiv.org");
805    }
806
807    // Pointing the general override at a test server has always redirected
808    // downloads too; only the defaults differ per host.
809    #[test]
810    #[serial_test::serial]
811    fn general_override_still_redirects_pdf_downloads() {
812        unsafe { std::env::remove_var("FASTPAPER_ARXIV_PDF_URL") };
813        unsafe { std::env::set_var("FASTPAPER_ARXIV_URL", "http://localhost:9999") };
814        assert_eq!(Source::Arxiv.pdf_base_url(), "http://localhost:9999");
815        unsafe { std::env::remove_var("FASTPAPER_ARXIV_URL") };
816    }
817
818    #[test]
819    #[serial_test::serial]
820    fn pdf_override_wins_over_the_general_one() {
821        unsafe { std::env::set_var("FASTPAPER_ARXIV_URL", "http://api.test") };
822        unsafe { std::env::set_var("FASTPAPER_ARXIV_PDF_URL", "http://files.test") };
823        assert_eq!(Source::Arxiv.pdf_base_url(), "http://files.test");
824        unsafe { std::env::remove_var("FASTPAPER_ARXIV_URL") };
825        unsafe { std::env::remove_var("FASTPAPER_ARXIV_PDF_URL") };
826    }
827}
fastpaper/registry.rs

fastpaper/
registry.rs
