elasticlunr/
pipeline.rs

1//! Defines the pipeline which processes text for inclusion in the index. Most users do not need
2//! to use this module directly.
3
4use serde::ser::{Serialize, SerializeSeq, Serializer};
5
6pub trait PipelineFn {
7    fn name(&self) -> String;
8
9    fn filter(&self, token: String) -> Option<String>;
10}
11
12#[derive(Clone)]
13pub struct FnWrapper(pub String, pub fn(String) -> Option<String>);
14
15impl PipelineFn for FnWrapper {
16    fn name(&self) -> String {
17        self.0.clone()
18    }
19
20    fn filter(&self, token: String) -> Option<String> {
21        (self.1)(token)
22    }
23}
24
25/// A sequence of `PipelineFn`s which are run on tokens to prepare them for searching.
26#[derive(Deserialize)]
27pub struct Pipeline {
28    #[serde(skip_deserializing)]
29    pub queue: Vec<Box<dyn PipelineFn>>,
30}
31
32impl Serialize for Pipeline {
33    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
34    where
35        S: Serializer,
36    {
37        let mut seq = serializer.serialize_seq(Some(self.queue.len()))?;
38        for elem in &self.queue {
39            seq.serialize_element(&elem.name())?;
40        }
41        seq.end()
42    }
43}
44
45impl Pipeline {
46    /// Run the Pipeline against the given vector of tokens. The returned vector may be shorter
47    /// than the input if a pipeline function returns `None` for a token.
48    pub fn run(&self, tokens: Vec<String>) -> Vec<String> {
49        let mut ret = vec![];
50        for token in tokens {
51            let mut token = Some(token);
52            for func in &self.queue {
53                if let Some(t) = token {
54                    token = func.filter(t);
55                } else {
56                    break;
57                }
58            }
59            if let Some(t) = token {
60                ret.push(t);
61            }
62        }
63        ret
64    }
65}