1mod http;
16mod wire;
17
18use std::sync::Arc;
19use std::time::Duration;
20
21use aspero_core::http::path_segment;
22use aspero_core::{
23 Capabilities, Cursor, Doi, Error, FieldSet, Filter, Hit, HttpClient, HttpRequest, HttpResponse,
24 IdKind, Lookup, Page, Provider, ProviderId, ResiliencePolicy, Response, Result, SearchMode,
25 SearchQuery, SourceRecord, WorkId,
26};
27use async_trait::async_trait;
28
29use crate::wire::{Envelope, WireList};
30
31pub const ID: ProviderId = ProviderId::from_static("crossref");
32
33pub const DEFAULT_BASE_URL: &str = "https://api.crossref.org";
34
35const MAX_BATCH: usize = 50;
37
38const MAX_ROWS: u32 = 1000;
40
41#[derive(Debug, Clone)]
42#[non_exhaustive]
43pub struct CrossrefConfig {
44 pub mailto: Option<String>,
47 pub base_url: String,
48}
49
50impl Default for CrossrefConfig {
51 fn default() -> Self {
52 CrossrefConfig {
53 mailto: None,
54 base_url: DEFAULT_BASE_URL.to_owned(),
55 }
56 }
57}
58
59pub struct Crossref {
60 config: CrossrefConfig,
61 http: Arc<dyn HttpClient>,
62 capabilities: Capabilities,
63}
64
65impl Crossref {
66 pub fn new(config: CrossrefConfig, http: Arc<dyn HttpClient>) -> Self {
67 let mut capabilities = Capabilities::default();
68 capabilities.lookup_kinds = vec![IdKind::Doi];
69 capabilities.max_batch = MAX_BATCH;
70 capabilities.fields = FieldSet::ALL;
71 capabilities.search = true;
72 capabilities.citation_search = true;
73 capabilities.search_filters = vec![Filter::Years, Filter::WorkType];
74 Crossref {
75 config,
76 http,
77 capabilities,
78 }
79 }
80
81 pub fn default_policy() -> ResiliencePolicy {
84 let mut policy = ResiliencePolicy::default();
85 policy.min_interval = Duration::from_millis(350);
86 policy
87 }
88
89 fn url(&self, rest: &str, params: &[(&str, String)]) -> Result<String> {
91 let base = format!("{}/works{rest}", self.config.base_url.trim_end_matches('/'));
92 let mut url = url::Url::parse(&base)
93 .map_err(|e| Error::BadRequest(format!("invalid Crossref base URL: {e}")))?;
94 {
95 let mut query = url.query_pairs_mut();
96 for (k, v) in params {
97 query.append_pair(k, v);
98 }
99 if let Some(mailto) = &self.config.mailto {
100 query.append_pair("mailto", mailto);
101 }
102 }
103 if url.query() == Some("") {
105 url.set_query(None);
106 }
107 Ok(url.into())
108 }
109
110 async fn get(&self, url: String) -> Result<HttpResponse> {
111 let req = HttpRequest::get(url).header("accept", "application/json");
112 let resp = self.http.send(req).await?;
113 http::check(&resp)?;
114 Ok(resp)
115 }
116
117 async fn lookup_one(&self, doi: &Doi) -> Result<Option<SourceRecord>> {
118 let url = self.url(&format!("/{}", path_segment(doi.as_str())), &[])?;
119 let resp = match self.get(url).await {
120 Ok(resp) => resp,
121 Err(Error::NotFound) => return Ok(None),
122 Err(e) => return Err(e),
123 };
124 let envelope: Envelope<serde_json::Value> =
125 serde_json::from_slice(&resp.body).map_err(Error::decode)?;
126 wire::to_record(&envelope.message, resp.received_at)
127 .map(Some)
128 .map_err(Error::decode)
129 }
130
131 async fn list(&self, params: &[(&str, String)]) -> Result<(Vec<Hit>, WireList)> {
132 let resp = self.get(self.url("", params)?).await?;
133 let envelope: Envelope<WireList> =
134 serde_json::from_slice(&resp.body).map_err(Error::decode)?;
135 let list = envelope.message;
136 let hits = list
137 .items
138 .iter()
139 .map(|v| {
140 let score = v.get("score").and_then(serde_json::Value::as_f64);
141 wire::to_record(v, resp.received_at).map(|r| Hit::new(r, score))
142 })
143 .collect::<serde_json::Result<_>>()
144 .map_err(Error::decode)?;
145 Ok((hits, list))
146 }
147}
148
149#[async_trait]
150impl Provider for Crossref {
151 fn id(&self) -> ProviderId {
152 ID
153 }
154
155 fn capabilities(&self) -> &Capabilities {
156 &self.capabilities
157 }
158
159 async fn lookup(&self, ids: &[WorkId], _fields: FieldSet) -> Result<Response<Vec<Lookup>>> {
160 let dois = ids
161 .iter()
162 .map(|id| match id {
163 WorkId::Doi(doi) => Ok(doi),
164 _ => Err(Error::Unsupported("Crossref lookup of non-DOI identifiers")),
165 })
166 .collect::<Result<Vec<_>>>()?;
167
168 if let [doi] = dois.as_slice() {
169 let record = self.lookup_one(doi).await?;
170 return Ok(Response::new(vec![Lookup {
171 id: ids[0].clone(),
172 record,
173 }]));
174 }
175
176 let (batchable, singles): (Vec<&Doi>, Vec<&Doi>) =
178 dois.iter().partition(|d| !d.as_str().contains(','));
179 let mut found = Vec::new();
180 for chunk in batchable.chunks(MAX_BATCH) {
181 let filter = chunk
182 .iter()
183 .map(|d| format!("doi:{}", d.as_str()))
184 .collect::<Vec<_>>()
185 .join(",");
186 let params = [("filter", filter), ("rows", chunk.len().to_string())];
187 let (hits, _) = self.list(¶ms).await?;
188 found.extend(hits.into_iter().map(|h| h.record));
189 }
190 for doi in singles {
191 found.extend(self.lookup_one(doi).await?);
192 }
193
194 let lookups = ids
195 .iter()
196 .map(|id| Lookup {
197 id: id.clone(),
198 record: found.iter().find(|r| r.ids.contains(id)).cloned(),
199 })
200 .collect();
201 Ok(Response::new(lookups))
202 }
203
204 async fn search(&self, query: &SearchQuery) -> Result<Response<Page>> {
205 let mut filters = Vec::new();
206 if let Some(years) = query.years {
207 if let Some(from) = years.from {
208 filters.push(format!("from-pub-date:{from:04}"));
209 }
210 if let Some(to) = years.to {
211 filters.push(format!("until-pub-date:{to:04}"));
212 }
213 }
214 if let Some(t) = &query.work_type {
215 let crossref = wire::crossref_type(t).ok_or_else(|| {
216 Error::BadRequest(format!(
217 "Crossref has no work type matching {:?}",
218 t.as_csl()
219 ))
220 })?;
221 filters.push(format!("type:{crossref}"));
222 }
223
224 let mut params: Vec<(&str, String)> = Vec::new();
225 for (key, value) in &query.extra {
226 match key.as_str() {
227 "filter" => filters.push(value.clone()),
228 "sort" | "order" => params.push((key.as_str(), value.clone())),
229 k if k.starts_with("query.") => params.push((k, value.clone())),
230 other => {
231 return Err(Error::BadRequest(format!(
232 "unknown Crossref parameter {other:?} (supported: filter, sort, order, query.*)"
233 )));
234 }
235 }
236 }
237 let text = query.text.trim();
238 if !text.is_empty() {
239 let field = match query.mode {
240 SearchMode::Citation => "query.bibliographic",
241 _ => "query",
242 };
243 params.push((field, text.to_owned()));
244 }
245 if !filters.is_empty() {
246 params.push(("filter", filters.join(",")));
247 }
248 let has_query = params.iter().any(|(k, _)| k.starts_with("query"));
252 if has_query && !params.iter().any(|(k, _)| *k == "sort") {
253 params.push(("sort", "score".into()));
254 params.push(("order", "desc".into()));
255 }
256 params.push(("rows", query.page.limit.clamp(1, MAX_ROWS).to_string()));
257 let cursor = query.page.cursor.as_ref().map_or("*", |c| c.0.as_str());
258 params.push(("cursor", cursor.to_owned()));
259
260 let (hits, list) = self.list(¶ms).await?;
261 let next = list.next_cursor.filter(|_| !hits.is_empty()).map(Cursor);
264 Ok(Response::new(Page::new(hits, list.total_results, next)))
265 }
266}