use aspero_core::{Claim, WorkId};
use serde::Deserialize;
use super::{Failure, read_stdin};
#[derive(Debug, Default, Deserialize)]
#[serde(deny_unknown_fields)]
pub struct ClaimInput {
pub text: Option<String>,
pub title: Option<String>,
#[serde(default)]
pub authors: Vec<String>,
pub year: Option<i16>,
pub doi: Option<String>,
pub arxiv: Option<String>,
pub id: Option<String>,
#[serde(default)]
pub ids: Vec<String>,
}
impl ClaimInput {
pub fn into_claim(self) -> Result<Claim, String> {
let mut claim = match &self.text {
Some(text) => Claim::from_input(text),
None => Claim::default(),
};
let prefixed = [
self.doi.map(|d| format!("doi:{d}")),
self.arxiv.map(|a| format!("arxiv:{a}")),
self.id,
];
for raw in prefixed.into_iter().flatten().chain(self.ids) {
let id: WorkId = raw.parse().map_err(|e| format!("{e}"))?;
if !claim.ids.contains(&id) {
claim.ids.push(id);
}
}
claim.title = self.title;
claim.authors = self.authors;
claim.year = self.year;
if claim.is_empty() {
return Err("claim has no text, title, authors or identifiers".into());
}
Ok(claim)
}
}
pub fn line_claim(line: &str) -> Result<Claim, String> {
if line.trim_start().starts_with('{') {
let input: ClaimInput =
serde_json::from_str(line).map_err(|e| format!("invalid claim JSON: {e}"))?;
input.into_claim()
} else {
Ok(Claim::from_input(line))
}
}
pub type InputClaim = (String, Result<Claim, String>);
pub fn claims(text: Option<String>, fields: ClaimInput) -> Result<Vec<InputClaim>, Failure> {
if text.as_deref() == Some("-") {
let input = read_stdin()?;
return Ok(input
.lines()
.map(str::trim)
.filter(|l| !l.is_empty() && !l.starts_with('#'))
.map(|l| (l.to_owned(), line_claim(l)))
.collect());
}
let label = text.clone().unwrap_or_else(|| describe_fields(&fields));
let input = ClaimInput { text, ..fields };
Ok(vec![(label, input.into_claim())])
}
fn describe_fields(f: &ClaimInput) -> String {
let parts: Vec<String> = [
(!f.authors.is_empty()).then(|| f.authors.join(", ")),
f.title.clone(),
f.year.map(|y| y.to_string()),
f.doi.clone().or(f.arxiv.clone()).or(f.id.clone()),
]
.into_iter()
.flatten()
.collect();
parts.join(". ")
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn json_lines_become_structured_claims() {
let claim = line_claim(
r#"{"title": "Attention is all you need", "authors": ["Vaswani, A."], "year": 2017, "arxiv": "1706.03762"}"#,
)
.unwrap();
assert_eq!(claim.year, Some(2017));
assert_eq!(claim.ids[0].to_string(), "arxiv:1706.03762");
assert!(
line_claim(r#"{"titel": "typo"}"#)
.unwrap_err()
.contains("unknown field")
);
assert!(line_claim(r#"{"doi": "not a doi"}"#).is_err());
}
}