Merge pull request #216 from noamteyssier/215-slim-features-for-clean…

…er-install

215 slim features for cleaner install

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "ggetrs"
-version = "0.1.86"
+version = "0.1.87"
 edition = "2021"
 license = "MIT"
 description = "Efficient querying of biological databases from the command line"
@@ -21,31 +21,36 @@ default = ["pyo3/extension-module"]
 [dependencies]
 anyhow = "1.0.64"
 bitvec = "1.0.1"
-bon = "2.3.0"
+bon = "3.0.2"
 chrono = "0.4.22"
 clap = { version = "4.0.18", features = ["derive"] }
 clap_complete = "4.0.3"
 ftp = "3.0.1"
 futures = "0.3.24"
 indicatif = "0.17.5"
 mysql = { version = "25.0.1", default-features = false, features = [
-  "minimal",
-  "rustls-tls",
+    "minimal",
+    "rustls-tls",
 ] }
-polars = { version = "0.43.1", default-features = false, features = ["json"] }
-pyo3 = { version = "0.22.2", features = ["extension-module", "anyhow"] }
+# polars = { version = "0.43.1", default-features = false, features = ["json"] }
+polars-core = { version = "0.44.2", default-features = false }
+polars-io = { version = "0.44.2", default-features = false, features = [
+    "json",
+    "csv",
+] }
+pyo3 = { version = "0.23.1", features = ["extension-module", "anyhow"] }
 regex = "1.6.0"
 reqwest = { version = "0.12.5", default-features = false, features = [
-  "json",
-  "multipart",
-  "blocking",
-  "stream",
-  "rustls-tls",
+    "json",
+    "multipart",
+    "blocking",
+    "stream",
+    "rustls-tls",
 ] }
 serde = { version = "1.0.144", features = ["derive"] }
 serde-xml-rs = "0.6.0"
 serde_json = "1.0.85"
-tokio = { version = "1.21.0", features = ["full"] }
+tokio = { version = "1.21.0", default-features = false }
 
 [lints.clippy]
 pedantic = { level = "warn", priority = -1 }

src/archs4/python.rs

@@ -31,7 +31,7 @@ pub fn python_archs4_tissue<'py>(
 
 /// Wraps `ARCHS4` specific functions into a python submodule
 pub fn python_archs4(py: Python<'_>, module: &Bound<'_, PyModule>) -> PyResult<()> {
-    let submodule = PyModule::new_bound(py, "archs4")?;
+    let submodule = PyModule::new(py, "archs4")?;
     submodule.add_function(wrap_pyfunction!(python_archs4_tissue, module)?)?;
     submodule.add_function(wrap_pyfunction!(python_archs4_correlate, module)?)?;
     module.add_submodule(&submodule)?;

src/archs4/types/correlation.rs

@@ -52,7 +52,7 @@ impl fmt::Display for Correlations {
 }
 impl Correlations {
     pub fn as_pydict<'py>(&self, py: Python<'py>) -> PyResult<Bound<'py, PyDict>> {
-        let dict = PyDict::new_bound(py);
+        let dict = PyDict::new(py);
         dict.set_item(
             "correlations",
             self.correlations
@@ -90,7 +90,7 @@ impl fmt::Display for Correlation {
 }
 impl Correlation {
     pub fn as_pydict<'py>(&self, py: Python<'py>) -> PyResult<Bound<'py, PyDict>> {
-        let dict = PyDict::new_bound(py);
+        let dict = PyDict::new(py);
         dict.set_item("gene_symbol", &self.gene_symbol)?;
         dict.set_item("pearson_correlation", self.pearson_correlation)?;
         Ok(dict)

src/archs4/types/tissue.rs

@@ -53,7 +53,7 @@ impl ResponseTissue {
             .collect()
     }
     pub fn as_pydict<'py>(&self, py: Python<'py>) -> PyResult<Bound<'py, PyDict>> {
-        let dict = PyDict::new_bound(py);
+        let dict = PyDict::new(py);
         dict.set_item(
             "tissues",
             self.results
@@ -124,7 +124,7 @@ impl ResultTissue {
     }
 
     pub fn as_pydict<'py>(&self, py: Python<'py>) -> PyResult<Bound<'py, PyDict>> {
-        let dict = PyDict::new_bound(py);
+        let dict = PyDict::new(py);
         dict.set_item("id", &self.id)?;
         dict.set_item("min", self.min)?;
         dict.set_item("q1", self.q1)?;

src/blast/python.rs

@@ -15,7 +15,6 @@ use pyo3::{
     signature = (query, program = None, database = None, limit = None, expect = None, low_comp_filter = None, megablast = None)
 )]
 #[allow(clippy::too_many_arguments)]
-
 pub fn python_blast<'py>(
     py: Python<'py>,
     query: &str,
@@ -61,5 +60,5 @@ pub fn python_blast<'py>(
         low_comp_filter,
         megablast,
     )?;
-    Ok(response.into_py_dict_bound(py))
+    Ok(response.into_py_dict(py)?)
 }

src/blast/types/result.rs

@@ -1,6 +1,6 @@
 use pyo3::{
     types::{IntoPyDict, PyDict, PyDictMethods},
-    Bound,
+    Bound, PyResult,
 };
 use serde::{Deserialize, Serialize};
 use std::fmt;
@@ -19,20 +19,20 @@ impl fmt::Display for BlastResult {
         )
     }
 }
-impl IntoPyDict for BlastResult {
-    fn into_py_dict_bound(self, py: pyo3::Python<'_>) -> Bound<'_, PyDict> {
-        let map = PyDict::new_bound(py);
+impl<'py> IntoPyDict<'py> for BlastResult {
+    fn into_py_dict(self, py: pyo3::Python<'py>) -> PyResult<Bound<'py, PyDict>> {
+        let map = PyDict::new(py);
         map.set_item("query", self.query).unwrap();
         map.set_item(
             "results",
             self.results
                 .iter()
                 .cloned()
-                .map(|x| x.into_py_dict_bound(py))
-                .collect::<Vec<Bound<'_, PyDict>>>(),
+                .map(|x| x.into_py_dict(py).unwrap())
+                .collect::<Vec<Bound<'py, PyDict>>>(),
         )
         .unwrap();
-        map
+        Ok(map)
     }
 }
 impl BlastResult {
@@ -97,9 +97,9 @@ impl BlastHit {
         }
     }
 }
-impl IntoPyDict for BlastHit {
-    fn into_py_dict_bound(self, py: pyo3::Python<'_>) -> Bound<'_, PyDict> {
-        let map = PyDict::new_bound(py);
+impl<'py> IntoPyDict<'py> for BlastHit {
+    fn into_py_dict(self, py: pyo3::Python<'py>) -> PyResult<Bound<'py, PyDict>> {
+        let map = PyDict::new(py);
         map.set_item("num", self.num).unwrap();
         map.set_item("id", self.num).unwrap();
         map.set_item("definition", self.num).unwrap();
@@ -114,7 +114,7 @@ impl IntoPyDict for BlastHit {
         map.set_item("query_end", self.num).unwrap();
         map.set_item("subject_start", self.num).unwrap();
         map.set_item("subject_end", self.num).unwrap();
-        map
+        Ok(map)
     }
 }
 

src/cli/cli.rs

@@ -116,7 +116,7 @@ pub enum Commands {
         #[clap(required = true)]
         search_terms: Vec<String>,
 
-        /// Species name to use: currently this MUST match the taxon_id
+        /// Species name to use: currently this MUST match the `taxon_id`
         #[clap(short, long, default_value = "homo_sapiens")]
         species: String,
 
@@ -129,7 +129,7 @@ pub enum Commands {
         output: Option<String>,
     },
 
-    /// Queries sequences from ensembl and UniProt
+    /// Queries sequences from ensembl and `UniProt`
     Seq {
         /// Search terms to query (can be Ensembl IDs or Gene Symbols)
         #[clap(value_parser, required = true)]

src/cli/enrichr.rs

@@ -4,7 +4,7 @@ use clap::Subcommand;
 pub enum ModEnrichr {
     /// Perform the Enrichr gene set enrichment analysis
     Enrichr {
-        /// any database listed at: https://maayanlab.cloud/Enrichr/#libraries
+        /// any database listed at: <https://maayanlab.cloud/Enrichr/#libraries>
         /// some shorthands include: pathway, transcription, ontology, diseases_drugs, celltypes,
         /// and kinase_interactions.
         #[clap(short, long)]

src/enrichr/types/enrich.rs

@@ -23,12 +23,15 @@ impl fmt::Display for ResponseEnrich {
 }
 impl ResponseEnrich {
     pub fn as_pydict<'py>(&self, py: Python<'py>) -> PyResult<Bound<'py, PyDict>> {
-        let dict = PyDict::new_bound(py);
+        let dict = PyDict::new(py);
         for (key, results) in &self.0 {
-            let all_results: Vec<Bound<'py, PyDict>> = results
-                .iter()
-                .map(|x| x.as_pydict(py).expect("could not create dictionary"))
-                .collect();
+            let all_results =
+                results
+                    .iter()
+                    .try_fold(Vec::new(), |mut acc, x| -> PyResult<Vec<_>> {
+                        acc.push(x.as_pydict(py)?);
+                        Ok(acc)
+                    })?;
             dict.set_item(key, all_results)?;
         }
         Ok(dict)
@@ -72,7 +75,7 @@ impl fmt::Display for ResultEnrichr {
 }
 impl ResultEnrichr {
     pub fn as_pydict<'py>(&self, py: Python<'py>) -> PyResult<Bound<'py, PyDict>> {
-        let dict = PyDict::new_bound(py);
+        let dict = PyDict::new(py);
         dict.set_item("rank", self.rank)?;
         dict.set_item("term_name", &self.term_name)?;
         dict.set_item("pvalue", self.pvalue)?;

src/ensembl/functions/database.rs

@@ -6,7 +6,7 @@ use mysql::{prelude::Queryable, Conn, OptsBuilder};
 pub fn database(filter: &Option<String>) -> anyhow::Result<ResponseDatabases> {
     let opts = get_mysql_options();
     let mut conn = Conn::new(opts)?;
-    let query = build_search_query(filter);
+    let query = build_search_query(filter.as_ref());
     let results: Vec<Database> = conn.query_map(query, Database)?;
     if results.is_empty() {
         match filter {
@@ -28,7 +28,7 @@ fn get_mysql_options() -> OptsBuilder {
 /// Generates the search query.
 ///
 /// Searches through databases for a related token
-fn build_search_query(search_term: &Option<String>) -> String {
+fn build_search_query(search_term: Option<&String>) -> String {
     if let Some(token) = search_term {
         format!("SHOW databases LIKE '%{token}%'")
     } else {

src/ensembl/python.rs

@@ -107,7 +107,7 @@ pub fn python_ensembl_species(
 }
 
 pub fn python_ensembl(py: Python<'_>, module: &Bound<'_, PyModule>) -> PyResult<()> {
-    let submodule = PyModule::new_bound(py, "ensembl")?;
+    let submodule = PyModule::new(py, "ensembl")?;
     submodule.add_function(wrap_pyfunction!(python_ensembl_search, module)?)?;
     submodule.add_function(wrap_pyfunction!(python_ensembl_database, module)?)?;
     submodule.add_function(wrap_pyfunction!(python_ensembl_release, module)?)?;

src/ensembl/types/ftpfile.rs

@@ -41,7 +41,7 @@ impl FtpFile {
     }
 
     pub fn as_pydict<'py>(&self, py: Python<'py>) -> PyResult<Bound<'py, PyDict>> {
-        let dict = PyDict::new_bound(py);
+        let dict = PyDict::new(py);
         dict.set_item("url", &self.url)?;
         dict.set_item("ensembl_release", self.ensembl_release)?;
         dict.set_item("release_date", &self.release_date)?;

src/ensembl/types/search_results.rs

@@ -21,7 +21,7 @@ impl fmt::Display for SearchResults {
 }
 impl SearchResults {
     pub fn as_pydict<'py>(&self, py: Python<'py>) -> Result<Bound<'py, PyDict>> {
-        let dict = PyDict::new_bound(py);
+        let dict = PyDict::new(py);
         dict.set_item(
             "results",
             self.0
@@ -82,7 +82,7 @@ impl SearchResult {
     }
 
     pub fn as_pydict<'py>(&self, py: Python<'py>) -> PyResult<Bound<'py, PyDict>> {
-        let dict = PyDict::new_bound(py);
+        let dict = PyDict::new(py);
         dict.set_item("stable_id", &self.stable_id)?;
         dict.set_item("display_label", &self.display_label)?;
         dict.set_item("ensembl_description", &self.ensembl_description)?;

src/info/python.rs

@@ -23,5 +23,5 @@ pub fn python_info(
     let species = species.unwrap_or("homo_sapiens".to_string());
     let taxon_id = taxon_id.unwrap_or(9606);
     let results = info(&search_terms, &species, taxon_id)?;
-    Ok(results.into_py_dict_bound(py))
+    Ok(results.into_py_dict(py)?)
 }

src/info/types.rs

@@ -2,7 +2,7 @@ use crate::{
     ensembl::types::LookupResponse, ncbi::types::NcbiResults, uniprot::UniprotInfoContainer,
 };
 use pyo3::types::{IntoPyDict, PyDict, PyDictMethods};
-use pyo3::Bound;
+use pyo3::{Bound, PyResult};
 use serde::{Deserialize, Serialize};
 use std::collections::HashMap;
 use std::fmt;
@@ -19,13 +19,14 @@ impl fmt::Display for InfoContainer {
         )
     }
 }
-impl IntoPyDict for InfoContainer {
-    fn into_py_dict_bound(self, py: pyo3::Python<'_>) -> Bound<'_, PyDict> {
-        let map = PyDict::new_bound(py);
+impl<'py> IntoPyDict<'py> for InfoContainer {
+    fn into_py_dict(self, py: pyo3::Python<'py>) -> PyResult<Bound<'py, PyDict>> {
+        let map = PyDict::new(py);
         self.0.iter().for_each(|(k, v)| {
-            map.set_item(k, v.clone().into_py_dict_bound(py)).unwrap();
+            map.set_item(k, v.clone().into_py_dict(py).unwrap())
+                .unwrap();
         });
-        map
+        Ok(map)
     }
 }
 impl InfoContainer {
@@ -69,9 +70,9 @@ impl fmt::Display for Info {
         )
     }
 }
-impl IntoPyDict for Info {
-    fn into_py_dict_bound(self, py: pyo3::Python<'_>) -> Bound<'_, PyDict> {
-        let map = PyDict::new_bound(py);
+impl<'py> IntoPyDict<'py> for Info {
+    fn into_py_dict(self, py: pyo3::Python<'py>) -> PyResult<Bound<'py, PyDict>> {
+        let map = PyDict::new(py);
         map.set_item("ensembl_id", &self.ensembl_id).unwrap();
         map.set_item("uniprot_id", &self.uniprot_id).unwrap();
         map.set_item("ncbi_id", &self.ncbi_id).unwrap();
@@ -85,7 +86,7 @@ impl IntoPyDict for Info {
             .unwrap();
         map.set_item("species", &self.species).unwrap();
         map.set_item("assembly_name", &self.assembly_name).unwrap();
-        map
+        Ok(map)
     }
 }
 impl Info {

src/lib.rs

@@ -53,18 +53,18 @@ pub mod seq;
 pub type RequestError = Box<dyn std::error::Error + Send + Sync>;
 
 #[pymodule]
-fn ggetrs(py: Python<'_>, module: Bound<'_, PyModule>) -> PyResult<()> {
-    module.add_function(wrap_pyfunction!(enrichr::python_enrichr, &module)?)?;
+fn ggetrs(py: Python<'_>, module: &Bound<'_, PyModule>) -> PyResult<()> {
+    module.add_function(wrap_pyfunction!(enrichr::python_enrichr, module)?)?;
     module.add_function(wrap_pyfunction!(
         enrichr::python_enrichr_background,
-        &module
+        module
     )?)?;
-    archs4::python_archs4(py, &module)?;
-    ensembl::python_ensembl(py, &module)?;
-    ucsc::python_ucsc(py, &module)?;
-    module.add_function(wrap_pyfunction!(ensembl::python_ensembl_search, &module)?)?;
-    module.add_function(wrap_pyfunction!(seq::python_seq, &module)?)?;
-    module.add_function(wrap_pyfunction!(info::python_info, &module)?)?;
-    module.add_function(wrap_pyfunction!(blast::python_blast, &module)?)?;
+    archs4::python_archs4(py, module)?;
+    ensembl::python_ensembl(py, module)?;
+    ucsc::python_ucsc(py, module)?;
+    module.add_function(wrap_pyfunction!(ensembl::python_ensembl_search, module)?)?;
+    module.add_function(wrap_pyfunction!(seq::python_seq, module)?)?;
+    module.add_function(wrap_pyfunction!(info::python_info, module)?)?;
+    module.add_function(wrap_pyfunction!(blast::python_blast, module)?)?;
     Ok(())
 }

src/main.rs

@@ -17,7 +17,11 @@ use ggetrs::{
     ncbi::{launch_ncbi_query_ids, launch_ncbi_query_symbols, launch_ncbi_taxons},
     pdb::{launch_pdb_resource, launch_pdb_structure},
     seq::launch_seq,
-    string::*,
+    string::{
+        launch_string_annotations, launch_string_enrichment, launch_string_homology,
+        launch_string_interactions, launch_string_mapping, launch_string_network,
+        launch_string_ppi_enrichment,
+    },
     ucsc::launch_ucsc_blat,
     uniprot::launch_uniprot_query,
     utils::autocomplete::print_completions,
@@ -250,7 +254,7 @@ fn main() -> Result<(), RequestError> {
             ModString::Enrichment { args, output } => launch_string_enrichment(args, output)?,
             ModString::Annotations { args, output } => launch_string_annotations(args, output)?,
             ModString::PpiEnrichment { args, output } => {
-                launch_string_ppi_enrichment(args, output)?
+                launch_string_ppi_enrichment(args, output)?;
             }
         },
         Commands::Autocomplete { shell } => print_completions(*shell, &mut Cli::command()),