store.rs

  1use std::path::PathBuf;
  2use std::sync::atomic::AtomicBool;
  3use std::sync::Arc;
  4
  5use anyhow::{anyhow, Result};
  6use async_trait::async_trait;
  7use collections::HashMap;
  8use derive_more::{Deref, Display};
  9use futures::future::{self, BoxFuture, Shared};
 10use futures::FutureExt;
 11use fuzzy::StringMatchCandidate;
 12use gpui::{AppContext, BackgroundExecutor, Task};
 13use heed::types::SerdeBincode;
 14use heed::Database;
 15use parking_lot::RwLock;
 16use serde::{Deserialize, Serialize};
 17use util::ResultExt;
 18
 19use crate::IndexedDocsRegistry;
 20
 21#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Deref, Display)]
 22pub struct ProviderId(pub Arc<str>);
 23
 24impl ProviderId {
 25    pub fn rustdoc() -> Self {
 26        Self("rustdoc".into())
 27    }
 28}
 29
 30/// The name of a package.
 31#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Deref, Display)]
 32pub struct PackageName(Arc<str>);
 33
 34impl From<&str> for PackageName {
 35    fn from(value: &str) -> Self {
 36        Self(value.into())
 37    }
 38}
 39
 40#[async_trait]
 41pub trait IndexedDocsProvider {
 42    /// Returns the ID of this provider.
 43    fn id(&self) -> ProviderId;
 44
 45    /// Returns the path to the database for this provider.
 46    fn database_path(&self) -> PathBuf;
 47
 48    /// Indexes the package with the given name.
 49    async fn index(&self, package: PackageName, database: Arc<IndexedDocsDatabase>) -> Result<()>;
 50}
 51
 52/// A store for indexed docs.
 53pub struct IndexedDocsStore {
 54    executor: BackgroundExecutor,
 55    database_future:
 56        Shared<BoxFuture<'static, Result<Arc<IndexedDocsDatabase>, Arc<anyhow::Error>>>>,
 57    provider: Box<dyn IndexedDocsProvider + Send + Sync + 'static>,
 58    indexing_tasks_by_package:
 59        RwLock<HashMap<PackageName, Shared<Task<Result<(), Arc<anyhow::Error>>>>>>,
 60}
 61
 62impl IndexedDocsStore {
 63    pub fn try_global(provider: ProviderId, cx: &AppContext) -> Result<Arc<Self>> {
 64        let registry = IndexedDocsRegistry::global(cx);
 65        registry
 66            .get_provider_store(provider.clone())
 67            .ok_or_else(|| anyhow!("no indexed docs store found for {provider}"))
 68    }
 69
 70    pub fn new(
 71        provider: Box<dyn IndexedDocsProvider + Send + Sync + 'static>,
 72        executor: BackgroundExecutor,
 73    ) -> Self {
 74        let database_future = executor
 75            .spawn({
 76                let executor = executor.clone();
 77                let database_path = provider.database_path();
 78                async move { IndexedDocsDatabase::new(database_path, executor) }
 79            })
 80            .then(|result| future::ready(result.map(Arc::new).map_err(Arc::new)))
 81            .boxed()
 82            .shared();
 83
 84        Self {
 85            executor,
 86            database_future,
 87            provider,
 88            indexing_tasks_by_package: RwLock::new(HashMap::default()),
 89        }
 90    }
 91
 92    /// Returns whether the package with the given name is currently being indexed.
 93    pub fn is_indexing(&self, package: &PackageName) -> bool {
 94        self.indexing_tasks_by_package.read().contains_key(package)
 95    }
 96
 97    pub async fn load(&self, key: String) -> Result<MarkdownDocs> {
 98        self.database_future
 99            .clone()
100            .await
101            .map_err(|err| anyhow!(err))?
102            .load(key)
103            .await
104    }
105
106    pub fn index(
107        self: Arc<Self>,
108        package: PackageName,
109    ) -> Shared<Task<Result<(), Arc<anyhow::Error>>>> {
110        if let Some(existing_task) = self.indexing_tasks_by_package.read().get(&package) {
111            return existing_task.clone();
112        }
113
114        let indexing_task = self
115            .executor
116            .spawn({
117                let this = self.clone();
118                let package = package.clone();
119                async move {
120                    let _finally = util::defer({
121                        let this = this.clone();
122                        let package = package.clone();
123                        move || {
124                            this.indexing_tasks_by_package.write().remove(&package);
125                        }
126                    });
127
128                    let index_task = async {
129                        let database = this
130                            .database_future
131                            .clone()
132                            .await
133                            .map_err(|err| anyhow!(err))?;
134                        this.provider.index(package, database).await
135                    };
136
137                    index_task.await.map_err(Arc::new)
138                }
139            })
140            .shared();
141
142        self.indexing_tasks_by_package
143            .write()
144            .insert(package, indexing_task.clone());
145
146        indexing_task
147    }
148
149    pub fn search(&self, query: String) -> Task<Vec<String>> {
150        let executor = self.executor.clone();
151        let database_future = self.database_future.clone();
152        self.executor.spawn(async move {
153            let Some(database) = database_future.await.map_err(|err| anyhow!(err)).log_err() else {
154                return Vec::new();
155            };
156
157            let Some(items) = database.keys().await.log_err() else {
158                return Vec::new();
159            };
160
161            let candidates = items
162                .iter()
163                .enumerate()
164                .map(|(ix, item_path)| StringMatchCandidate::new(ix, item_path.clone()))
165                .collect::<Vec<_>>();
166
167            let matches = fuzzy::match_strings(
168                &candidates,
169                &query,
170                false,
171                100,
172                &AtomicBool::default(),
173                executor,
174            )
175            .await;
176
177            matches
178                .into_iter()
179                .map(|mat| items[mat.candidate_id].clone())
180                .collect()
181        })
182    }
183}
184
185#[derive(Debug, PartialEq, Eq, Clone, Display, Serialize, Deserialize)]
186pub struct MarkdownDocs(pub String);
187
188pub struct IndexedDocsDatabase {
189    executor: BackgroundExecutor,
190    env: heed::Env,
191    entries: Database<SerdeBincode<String>, SerdeBincode<MarkdownDocs>>,
192}
193
194impl IndexedDocsDatabase {
195    pub fn new(path: PathBuf, executor: BackgroundExecutor) -> Result<Self> {
196        std::fs::create_dir_all(&path)?;
197
198        const ONE_GB_IN_BYTES: usize = 1024 * 1024 * 1024;
199        let env = unsafe {
200            heed::EnvOpenOptions::new()
201                .map_size(ONE_GB_IN_BYTES)
202                .max_dbs(1)
203                .open(path)?
204        };
205
206        let mut txn = env.write_txn()?;
207        let entries = env.create_database(&mut txn, Some("rustdoc_entries"))?;
208        txn.commit()?;
209
210        Ok(Self {
211            executor,
212            env,
213            entries,
214        })
215    }
216
217    pub fn keys(&self) -> Task<Result<Vec<String>>> {
218        let env = self.env.clone();
219        let entries = self.entries;
220
221        self.executor.spawn(async move {
222            let txn = env.read_txn()?;
223            let mut iter = entries.iter(&txn)?;
224            let mut keys = Vec::new();
225            while let Some((key, _value)) = iter.next().transpose()? {
226                keys.push(key);
227            }
228
229            Ok(keys)
230        })
231    }
232
233    pub fn load(&self, key: String) -> Task<Result<MarkdownDocs>> {
234        let env = self.env.clone();
235        let entries = self.entries;
236
237        self.executor.spawn(async move {
238            let txn = env.read_txn()?;
239            entries
240                .get(&txn, &key)?
241                .ok_or_else(|| anyhow!("no docs found for {key}"))
242        })
243    }
244
245    pub fn insert(&self, key: String, docs: String) -> Task<Result<()>> {
246        let env = self.env.clone();
247        let entries = self.entries;
248
249        self.executor.spawn(async move {
250            let mut txn = env.write_txn()?;
251            entries.put(&mut txn, &key, &MarkdownDocs(docs))?;
252            txn.commit()?;
253            Ok(())
254        })
255    }
256}