summaryrefslogtreecommitdiff
path: root/support/texlab/src/citation.rs
diff options
context:
space:
mode:
authorNorbert Preining <norbert@preining.info>2021-05-23 03:00:39 +0000
committerNorbert Preining <norbert@preining.info>2021-05-23 03:00:39 +0000
commitf1261b349e875b842745b63258c3e338cb1fe3bf (patch)
treeb5d402b3e80818cde2c079a42249f3dcb9732247 /support/texlab/src/citation.rs
parent58aa1ac09b1d9e4769d0a0661cf12e2b2db41b14 (diff)
CTAN sync 202105230300
Diffstat (limited to 'support/texlab/src/citation.rs')
-rw-r--r--support/texlab/src/citation.rs165
1 files changed, 165 insertions, 0 deletions
diff --git a/support/texlab/src/citation.rs b/support/texlab/src/citation.rs
new file mode 100644
index 0000000000..e185c14b55
--- /dev/null
+++ b/support/texlab/src/citation.rs
@@ -0,0 +1,165 @@
+mod bibutils;
+mod name;
+mod ris;
+
+use std::sync::Arc;
+
+use citeproc::{prelude::SupportedFormat, ClusterPosition, Processor};
+use citeproc_db::PredefinedLocales;
+use citeproc_io::{Cite, Cluster, Reference};
+use lsp_types::{MarkupContent, MarkupKind};
+use once_cell::sync::Lazy;
+use regex::Regex;
+
+use crate::syntax::{bibtex, CstNode};
+
+use self::{
+ bibutils::*,
+ ris::{RisLibrary, RisReference},
+};
+
+static APA_STYLE: &str = include_str!("citation/apa.csl");
+
+static DOI_URL_PATTERN: &str = r#"https://doi.org/\[.*\]\(.*\)"#;
+
+static DOI_URL_REGEX: Lazy<Regex> = Lazy::new(|| Regex::new(DOI_URL_PATTERN).unwrap());
+
+pub fn render_citation(root: &bibtex::SyntaxNode, key: &str) -> Option<MarkupContent> {
+ let ris_reference = convert_to_ris(root, key)?;
+ let doi_url = get_doi_url_markdown(&ris_reference);
+ let csl_reference: Reference = ris_reference.into();
+ let html = generate_bibliography(csl_reference)?;
+
+ let mut markdown = html2md::parse_html(&html).trim().to_owned();
+ if markdown == "" {
+ return None;
+ }
+
+ if let Some(doi_url) = doi_url {
+ markdown = DOI_URL_REGEX
+ .replace(&markdown, doi_url.as_str())
+ .into_owned();
+ }
+
+ markdown = markdown
+ .replace("..", ".")
+ .replace("\\\'", "'")
+ .replace("\\-", "-")
+ .replace("\\\\textsubscript", "")
+ .replace("\\\\textsuperscript", "");
+ let content = MarkupContent {
+ kind: MarkupKind::Markdown,
+ value: markdown,
+ };
+ Some(content)
+}
+
+fn convert_to_ris(root: &bibtex::SyntaxNode, key: &str) -> Option<RisReference> {
+ let mut bib_code = String::new();
+ for string in root
+ .children()
+ .filter_map(bibtex::String::cast)
+ .filter(|string| string.name().is_some())
+ {
+ bib_code.push_str(&string.syntax().to_string());
+ }
+
+ let entry = root
+ .children()
+ .filter_map(bibtex::Entry::cast)
+ .find(|entry| entry.key().map(|key| key.text()) == Some(key))?;
+ bib_code.push_str(&entry.syntax().to_string());
+
+ bib_code = bib_code.replace("\\hypen", "-");
+
+ let ris_code = bibutils::convert(&bib_code, InputFormat::Biblatex, OutputFormat::Ris)?;
+ let ris_lib = RisLibrary::parse(ris_code.lines());
+ ris_lib
+ .references
+ .into_iter()
+ .find(|reference| reference.id.as_ref().map(AsRef::as_ref) == Some(key))
+}
+
+fn get_doi_url_markdown(ris_reference: &RisReference) -> Option<String> {
+ ris_reference
+ .doi
+ .as_ref()
+ .map(|doi| format!("[doi:{}](https://doi.org/{})", doi, doi))
+}
+
+fn generate_bibliography(reference: Reference) -> Option<String> {
+ let locales = Arc::new(PredefinedLocales::bundled_en_us());
+ let mut processor = Processor::new(APA_STYLE, locales, false, SupportedFormat::Html).unwrap();
+ let cite = Cite::basic(&reference.id);
+ let cluster = Cluster {
+ id: 1,
+ cites: vec![cite],
+ };
+ processor.insert_reference(reference);
+ processor.init_clusters(vec![cluster]);
+ processor
+ .set_cluster_order(&[ClusterPosition {
+ id: 1,
+ note: Some(1),
+ }])
+ .unwrap();
+ processor.get_bibliography().pop()
+}
+
+#[cfg(test)]
+mod tests {
+ use super::*;
+
+ #[test]
+ fn test_simple() {
+ let document = bibtex::parse(
+ r#"
+ @article{foo,
+ author = {Foo Bar},
+ title = {Baz Qux},
+ year = {2020}
+ }
+ "#,
+ );
+
+ let actual_md = render_citation(&document.root, "foo").unwrap();
+
+ let expected_md = MarkupContent {
+ kind: MarkupKind::Markdown,
+ value: "Bar, F. (2020). *Baz Qux*.".into(),
+ };
+
+ assert_eq!(actual_md, expected_md);
+ }
+
+ #[test]
+ fn test_string() {
+ let document = bibtex::parse(
+ r#"
+ @string{author = "Foo Bar"}
+ @article{foo,
+ author = author,
+ title = {Baz Qux},
+ year = {2020}
+ }
+ "#,
+ );
+ let actual_md = render_citation(&document.root, "foo").unwrap();
+
+ let expected_md = MarkupContent {
+ kind: MarkupKind::Markdown,
+ value: "Bar, F. (2020). *Baz Qux*.".into(),
+ };
+
+ assert_eq!(actual_md, expected_md);
+ }
+
+ #[test]
+ fn test_unknown_key() {
+ let document = bibtex::parse("");
+
+ let actual_md = render_citation(&document.root, "foo");
+
+ assert_eq!(actual_md, None);
+ }
+}