feat: express existing thread safety (as most things are immutable) (#…

…115)

Cargo.toml

@@ -5,7 +5,7 @@ edition = "2021"
 authors = ["Manuel Holtgrewe <manuel.holtgrewe@bih-charite.de>"]
 description = "Port of biocommons/hgvs to Rust"
 license = "Apache-2.0"
-homepage = "https://github.com/bihealth/seqrepo-rs"
+homepage = "https://github.com/bihealth/hgvs-rs"
 readme = "README.md"
 rust-version = "1.64.0"
 
@@ -27,7 +27,7 @@ postgres = { version = "0.19", features = ["with-chrono-0_4"] }
 quick_cache = "0.3"
 regex = "1.7"
 rustc-hash = "1.1"
-seqrepo = { version = "0.5" }
+seqrepo = { version = "0.6" }
 serde_json = "1.0"
 serde = { version = "1.0", features = ["derive"] }
 thiserror = "1.0"

src/data/cdot/json.rs

@@ -2,21 +2,21 @@
 //!
 //! https://github.com/SACGF/cdot
 
-use std::{collections::HashMap, path::PathBuf, rc::Rc, time::Instant};
+use std::{collections::HashMap, path::PathBuf, sync::Arc, time::Instant};
 
 use crate::{
     data::error::Error,
     data::interface::{
-        GeneInfoRecord, Provider as ProviderInterface, TxExonsRecord, TxForRegionRecord,
-        TxIdentityInfo, TxInfoRecord, TxMappingOptionsRecord, TxSimilarityRecord,
+        self, GeneInfoRecord, TxExonsRecord, TxForRegionRecord, TxIdentityInfo, TxInfoRecord,
+        TxMappingOptionsRecord, TxSimilarityRecord,
     },
     static_data::{Assembly, ASSEMBLY_INFOS},
 };
 
 use bio::data_structures::interval_tree::ArrayBackedIntervalTree;
 use chrono::NaiveDateTime;
 use indexmap::IndexMap;
-use seqrepo::{Interface as SeqRepoInterface, SeqRepo};
+use seqrepo::{self, SeqRepo};
 
 /// Configurationf or the `data::cdot::Provider`.
 #[derive(Debug, PartialEq, Clone)]
@@ -42,7 +42,7 @@ pub struct Config {
 /// `exon_aln_id`.
 pub struct Provider {
     inner: TxProvider,
-    seqrepo: Rc<dyn SeqRepoInterface>,
+    seqrepo: Arc<dyn seqrepo::Interface + Sync + Send>,
 }
 
 impl Provider {
@@ -70,14 +70,14 @@ impl Provider {
                     .collect::<Vec<&str>>()
                     .as_ref(),
             )?,
-            seqrepo: Rc::new(SeqRepo::new(path, &instance)?),
+            seqrepo: Arc::new(SeqRepo::new(path, &instance)?),
         })
     }
 
     /// Create a new provider allowing to inject a seqrepo.
     pub fn with_seqrepo(
         config: Config,
-        seqrepo: Rc<dyn SeqRepoInterface>,
+        seqrepo: Arc<dyn seqrepo::Interface + Sync + Send>,
     ) -> Result<Provider, Error> {
         Ok(Self {
             inner: TxProvider::with_config(
@@ -93,7 +93,7 @@ impl Provider {
     }
 }
 
-impl ProviderInterface for Provider {
+impl interface::Provider for Provider {
     fn data_version(&self) -> &str {
         self.inner.data_version()
     }
@@ -699,7 +699,7 @@ pub static REQUIRED_VERSION: &str = "1.1";
 /// The alignment method returned for all cdot transcripts.
 pub static NCBI_ALN_METHOD: &str = "splign";
 
-/// Implementation for `ProviderInterface`.
+/// Implementation for `interface::Provider`.
 impl TxProvider {
     fn data_version(&self) -> &str {
         REQUIRED_VERSION
@@ -1024,12 +1024,12 @@ impl TxProvider {
 #[cfg(test)]
 pub mod test_helpers {
     use anyhow::Error;
-    use std::rc::Rc;
+    use std::sync::Arc;
 
     use crate::data::uta_sr::test_helpers::build_writing_sr;
 
     use super::{Config, Provider};
-    use seqrepo::{CacheReadingSeqRepo, Interface as SeqRepoInterface};
+    use seqrepo;
 
     pub fn build_provider() -> Result<Provider, Error> {
         let sr_cache_mode = std::env::var("TEST_SEQREPO_CACHE_MODE")
@@ -1040,8 +1040,8 @@ pub mod test_helpers {
         log::debug!("building provider...");
         let seqrepo = if sr_cache_mode == "read" {
             log::debug!("reading provider...");
-            let seqrepo: Rc<dyn SeqRepoInterface> =
-                Rc::new(CacheReadingSeqRepo::new(sr_cache_path)?);
+            let seqrepo: Arc<dyn seqrepo::Interface + Send + Sync> =
+                Arc::new(seqrepo::CacheReadingSeqRepo::new(sr_cache_path)?);
             log::debug!("construction done...");
             seqrepo
         } else if sr_cache_mode == "write" {
@@ -1066,8 +1066,9 @@ pub mod test_helpers {
 #[cfg(test)]
 pub mod tests {
     use anyhow::Error;
-    use std::rc::Rc;
+
     use std::str::FromStr;
+    use std::sync::Arc;
 
     use chrono::NaiveDateTime;
     use pretty_assertions::assert_eq;
@@ -1076,13 +1077,19 @@ pub mod tests {
     use super::models::{gap_to_cigar, Container};
     use super::test_helpers::build_provider;
     use crate::data::interface::{
-        GeneInfoRecord, Provider as ProviderInterface, TxExonsRecord, TxForRegionRecord,
-        TxInfoRecord, TxMappingOptionsRecord, TxSimilarityRecord,
+        GeneInfoRecord, Provider, TxExonsRecord, TxForRegionRecord, TxInfoRecord,
+        TxMappingOptionsRecord, TxSimilarityRecord,
     };
-    use crate::mapper::assembly::{Config as AssemblyMapperConfig, Mapper};
+    use crate::mapper::assembly::{self, Mapper};
     use crate::parser::HgvsVariant;
     use crate::static_data::Assembly;
 
+    #[test]
+    fn test_sync() {
+        fn is_sync<T: Sync>() {}
+        is_sync::<super::Provider>();
+    }
+
     #[test]
     fn deserialize_brca1() -> Result<(), Error> {
         let json = std::fs::read_to_string(
@@ -2152,8 +2159,8 @@ pub mod tests {
     }
 
     fn build_mapper_37(normalize: bool) -> Result<Mapper, Error> {
-        let provider = Rc::new(build_provider()?);
-        let config = AssemblyMapperConfig {
+        let provider = Arc::new(build_provider()?);
+        let config = assembly::Config {
             assembly: Assembly::Grch37,
             normalize,
             ..Default::default()

src/data/uta.rs

@@ -12,9 +12,9 @@ use crate::sequences::seq_md5;
 use crate::static_data::{Assembly, ASSEMBLY_INFOS};
 
 use crate::data::{
-    error::Error, interface::GeneInfoRecord, interface::Provider as ProviderInterface,
-    interface::TxExonsRecord, interface::TxForRegionRecord, interface::TxIdentityInfo,
-    interface::TxInfoRecord, interface::TxMappingOptionsRecord, interface::TxSimilarityRecord,
+    error::Error, interface, interface::GeneInfoRecord, interface::TxExonsRecord,
+    interface::TxForRegionRecord, interface::TxIdentityInfo, interface::TxInfoRecord,
+    interface::TxMappingOptionsRecord, interface::TxSimilarityRecord,
 };
 
 /// Configuration for the `data::uta::Provider`.
@@ -234,7 +234,7 @@ impl Provider {
     }
 }
 
-impl ProviderInterface for Provider {
+impl interface::Provider for Provider {
     fn data_version(&self) -> &str {
         &self.config.db_schema
     }
@@ -604,11 +604,17 @@ impl ProviderInterface for Provider {
 
 #[cfg(test)]
 mod test {
-    use crate::{data::interface::Provider as ProviderInterface, static_data::Assembly};
+    use crate::{data::interface::Provider as InterfaceProvider, static_data::Assembly};
     use anyhow::Error;
 
     use super::{Config, Provider};
 
+    #[test]
+    fn test_sync() {
+        fn is_sync<T: Sync>() {}
+        is_sync::<super::Provider>();
+    }
+
     fn get_config() -> Config {
         Config {
             db_url: std::env::var("TEST_UTA_DATABASE_URL")

src/data/uta_sr.rs

@@ -5,15 +5,15 @@
 //! * https://github.com/bihealth/seqrepo-rs
 
 use std::path::PathBuf;
-use std::rc::Rc;
+use std::sync::Arc;
 
-use crate::data::uta::{Config as UtaConfig, Provider as UtaProvider};
+use crate::data::uta;
 use crate::data::{
-    error::Error, interface::GeneInfoRecord, interface::Provider as ProviderInterface,
-    interface::TxExonsRecord, interface::TxForRegionRecord, interface::TxIdentityInfo,
-    interface::TxInfoRecord, interface::TxMappingOptionsRecord, interface::TxSimilarityRecord,
+    error::Error, interface, interface::GeneInfoRecord, interface::TxExonsRecord,
+    interface::TxForRegionRecord, interface::TxIdentityInfo, interface::TxInfoRecord,
+    interface::TxMappingOptionsRecord, interface::TxSimilarityRecord,
 };
-use seqrepo::{AliasOrSeqId, Interface as SeqRepoInterface, SeqRepo};
+use seqrepo::{self, AliasOrSeqId, SeqRepo};
 
 /// Configuration for the `data::uta_sr::Provider`.
 #[derive(Debug, PartialEq, Clone)]
@@ -33,8 +33,8 @@ pub struct Config {
 /// Transcripts from a UTA Postgres database, sequences comes from a SeqRepo.  This makes
 /// genome contig information available in contrast to `data::uta::Provider`.
 pub struct Provider {
-    inner: UtaProvider,
-    seqrepo: Rc<dyn SeqRepoInterface>,
+    inner: uta::Provider,
+    seqrepo: Arc<dyn seqrepo::Interface + Send + Sync>,
 }
 
 impl Provider {
@@ -58,21 +58,21 @@ impl Provider {
             .to_string();
 
         Ok(Self {
-            inner: UtaProvider::with_config(&UtaConfig {
+            inner: uta::Provider::with_config(&uta::Config {
                 db_url: config.db_url.clone(),
                 db_schema: config.db_schema,
             })?,
-            seqrepo: Rc::new(SeqRepo::new(path, &instance)?),
+            seqrepo: Arc::new(SeqRepo::new(path, &instance)?),
         })
     }
 
     /// Create a new provider allowing to inject a seqrepo.
     pub fn with_seqrepo(
         config: Config,
-        seqrepo: Rc<dyn SeqRepoInterface>,
+        seqrepo: Arc<dyn seqrepo::Interface + Send + Sync>,
     ) -> Result<Provider, Error> {
         Ok(Self {
-            inner: UtaProvider::with_config(&UtaConfig {
+            inner: uta::Provider::with_config(&uta::Config {
                 db_url: config.db_url.clone(),
                 db_schema: config.db_schema,
             })?,
@@ -81,7 +81,7 @@ impl Provider {
     }
 }
 
-impl ProviderInterface for Provider {
+impl interface::Provider for Provider {
     fn data_version(&self) -> &str {
         self.inner.data_version()
     }
@@ -174,18 +174,23 @@ impl ProviderInterface for Provider {
 #[cfg(test)]
 pub mod test_helpers {
     use anyhow::Error;
-    use std::{path::PathBuf, rc::Rc};
+    use seqrepo::{CacheReadingSeqRepo, CacheWritingSeqRepo, SeqRepo};
+    use std::{path::PathBuf, sync::Arc};
 
-    use seqrepo::{
-        CacheReadingSeqRepo, CacheWritingSeqRepo, Interface as SeqRepoInterface, SeqRepo,
-    };
+    use crate::data::interface;
 
-    use super::{Config, Provider, ProviderInterface};
+    use super::{Config, Provider};
+
+    #[test]
+    fn test_sync() {
+        fn is_sync<T: Sync>() {}
+        is_sync::<super::Provider>();
+    }
 
     /// Setup a UTA Provider with data source depending on environment variables.
     ///
     /// See README.md for information on environment variable setup.
-    pub fn build_provider() -> Result<Rc<dyn ProviderInterface>, Error> {
+    pub fn build_provider() -> Result<Arc<dyn interface::Provider + Send + Sync>, Error> {
         log::debug!("building provider...");
         let db_url = std::env::var("TEST_UTA_DATABASE_URL")
             .expect("Environment variable TEST_UTA_DATABASE_URL undefined!");
@@ -198,8 +203,8 @@ pub mod test_helpers {
 
         let (seqrepo, seqrepo_path) = if sr_cache_mode == "read" {
             log::debug!("reading provider...");
-            let seqrepo: Rc<dyn SeqRepoInterface> =
-                Rc::new(CacheReadingSeqRepo::new(sr_cache_path)?);
+            let seqrepo: Arc<dyn seqrepo::Interface + Send + Sync> =
+                Arc::new(CacheReadingSeqRepo::new(sr_cache_path)?);
             log::debug!("construction done...");
             (seqrepo, "".to_string())
         } else if sr_cache_mode == "write" {
@@ -210,7 +215,7 @@ pub mod test_helpers {
         };
         log::debug!("now returning provider...");
 
-        Ok(Rc::new(Provider::with_seqrepo(
+        Ok(Arc::new(Provider::with_seqrepo(
             Config {
                 db_url,
                 db_schema,
@@ -223,7 +228,7 @@ pub mod test_helpers {
     /// Helper that builds the cache writing SeqRepo with inner stock SeqRepo.
     pub fn build_writing_sr(
         sr_cache_path: String,
-    ) -> Result<(Rc<dyn SeqRepoInterface>, String), Error> {
+    ) -> Result<(Arc<dyn seqrepo::Interface + Send + Sync>, String), Error> {
         let seqrepo_path = std::env::var("TEST_SEQREPO_PATH")
             .expect("Environment variable TEST_SEQREPO_PATH undefined!");
         let path_buf = PathBuf::from(seqrepo_path.clone());
@@ -245,10 +250,9 @@ pub mod test_helpers {
             .to_str()
             .expect("problem with path to string conversion")
             .to_string();
-        let seqrepo: Rc<dyn SeqRepoInterface> = Rc::new(CacheWritingSeqRepo::new(
-            SeqRepo::new(path, &instance)?,
-            sr_cache_path,
-        )?);
+        let seqrepo: Arc<dyn seqrepo::Interface + Send + Sync> = Arc::new(
+            CacheWritingSeqRepo::new(SeqRepo::new(path, &instance)?, sr_cache_path)?,
+        );
         Ok((seqrepo, seqrepo_path))
     }
 }

src/mapper/alignment.rs

@@ -22,7 +22,7 @@
 //    n.        -2    -1  !  1     2     3     4     5     6     7     8     9
 //    g.   ... 123   124   125   126   127   128   129   130   131   132   133 ...
 
-use std::rc::Rc;
+use std::sync::Arc;
 
 use crate::{
     data::interface::{Provider, TxExonsRecord},
@@ -119,7 +119,7 @@ pub struct Mapper {
     /// Configuration for alignment mapping.
     pub config: Config,
     /// Data provider to use for the mapping.
-    pub provider: Rc<dyn Provider>,
+    pub provider: Arc<dyn Provider + Sync + Send>,
 
     /// The transcript accession.
     pub tx_ac: String,
@@ -138,7 +138,7 @@ pub struct Mapper {
 impl Mapper {
     pub fn new(
         config: &Config,
-        provider: Rc<dyn Provider>,
+        provider: Arc<dyn Provider + Send + Sync>,
         tx_ac: &str,
         alt_ac: &str,
         alt_aln_method: &str,
@@ -499,6 +499,12 @@ mod test {
 
     use super::{build_tx_cigar, none_if_default, Mapper};
 
+    #[test]
+    fn test_sync() {
+        fn is_sync<T: Sync>() {}
+        is_sync::<super::Mapper>();
+    }
+
     #[test]
     fn build_tx_cigar_empty() {
         assert!(build_tx_cigar(&Vec::new(), 1).is_err());

src/mapper/altseq.rs

@@ -1,6 +1,6 @@
 //! Code for building alternative sequence and convertion to HGVS.p.
 
-use std::{cmp::Ordering, rc::Rc};
+use std::{cmp::Ordering, sync::Arc};
 
 use crate::{
     data::interface::Provider,
@@ -35,7 +35,7 @@ impl RefTranscriptData {
     /// * `tx_ac` -- Transcript accession.
     /// * `pro_ac` -- Protein accession.
     pub fn new(
-        provider: Rc<dyn Provider>,
+        provider: Arc<dyn Provider + Send + Sync>,
         tx_ac: &str,
         pro_ac: Option<&str>,
     ) -> Result<Self, Error> {