feat: add dataset info to output json [wip]

packages_rs/nextclade-cli/src/cli/nextclade_loop.rs

@@ -124,6 +124,7 @@ pub fn nextclade_run(run_args: NextcladeRunArgs) -> Result<(), Report> {
       } = nextclade.get_initial_data();
 
       let mut output_writer = NextcladeOrderedWriter::new(
+        &inputs.dataset_info,
         &nextclade.gene_map,
         clade_node_attr_key_descs,
         phenotype_attr_descs,

packages_rs/nextclade-cli/src/cli/nextclade_ordered_writer.rs

@@ -6,6 +6,7 @@ use log::{info, warn};
 use nextclade::alphabet::nuc::from_nuc_seq;
 use nextclade::analyze::virus_properties::PhenotypeAttrDesc;
 use nextclade::gene::gene_map::GeneMap;
+use nextclade::io::dataset::DatasetInfoShort;
 use nextclade::io::fasta::{FastaPeptideWriter, FastaRecord, FastaWriter};
 use nextclade::io::ndjson::NdjsonFileWriter;
 use nextclade::io::nextclade_csv::{CsvColumnConfig, NextcladeResultsCsvFileWriter};
@@ -35,6 +36,7 @@ pub struct NextcladeOrderedWriter {
 
 impl NextcladeOrderedWriter {
   pub fn new(
+    dataset_info: &DatasetInfoShort,
     gene_map: &GeneMap,
     clade_node_attr_key_descs: &[CladeNodeAttrKeyDesc],
     phenotype_attr_key_desc: &[PhenotypeAttrDesc],
@@ -50,7 +52,12 @@ impl NextcladeOrderedWriter {
       .map_ref_fallible(|output_translations| FastaPeptideWriter::new(gene_map, output_translations))?;
 
     let output_json_writer = output_params.output_json.map_ref_fallible(|output_json| {
-      ResultsJsonWriter::new(output_json, clade_node_attr_key_descs, phenotype_attr_key_desc)
+      ResultsJsonWriter::new(
+        dataset_info,
+        output_json,
+        clade_node_attr_key_descs,
+        phenotype_attr_key_desc,
+      )
     })?;
 
     let output_ndjson_writer = output_params.output_ndjson.map_ref_fallible(NdjsonFileWriter::new)?;

packages_rs/nextclade-cli/src/dataset/dataset_download.rs

@@ -6,7 +6,9 @@ use itertools::Itertools;
 use log::LevelFilter;
 use nextclade::analyze::virus_properties::{LabelledMutationsConfig, VirusProperties};
 use nextclade::gene::gene_map::{filter_gene_map, GeneMap};
-use nextclade::io::dataset::{Dataset, DatasetFiles, DatasetMeta, DatasetsIndexJson};
+use nextclade::io::dataset::{
+  Dataset, DatasetAttributeValue, DatasetAttributes, DatasetFiles, DatasetInfoShort, DatasetsIndexJson, DatasetMeta,
+};
 use nextclade::io::fasta::{read_one_fasta, read_one_fasta_str};
 use nextclade::io::file::create_file_or_stdout;
 use nextclade::io::fs::{ensure_dir, has_extension, read_file_to_string};

packages_rs/nextclade/src/io/dataset.rs

@@ -79,6 +79,15 @@ pub struct Dataset {
 }
 
 impl Dataset {
+  pub fn short_info(&self) -> DatasetInfoShort {
+    DatasetInfoShort {
+      path: self.path.clone(),
+      version: self.version.clone(),
+      server: None,
+      other: serde_json::Value::default(),
+    }
+  }
+
   pub fn name(&self) -> Option<&str> {
     self.attributes.get("name").and_then(AnyType::as_str_maybe)
   }
@@ -331,3 +340,24 @@ pub struct MinimizerIndexVersion {
   #[serde(flatten)]
   pub other: serde_json::Value,
 }
+
+#[derive(Clone, Debug, Serialize, Deserialize, JsonSchema)]
+#[serde(rename_all = "camelCase")]
+pub struct DatasetInfoShort {
+  pub path: String,
+
+  #[serde(default, skip_serializing_if = "DatasetVersion::is_empty")]
+  pub version: DatasetVersion,
+
+  #[serde(default, skip_serializing_if = "Option::is_none")]
+  pub server: Option<String>,
+
+  #[serde(flatten)]
+  pub other: serde_json::Value,
+}
+
+impl DatasetInfoShort {
+  pub fn from_str(s: impl AsRef<str>) -> Result<Self, Report> {
+    json_parse(s)
+  }
+}

packages_rs/nextclade/src/io/results_json.rs

@@ -1,15 +1,16 @@
 use crate::analyze::virus_properties::PhenotypeAttrDesc;
+use crate::io::dataset::DatasetInfoShort;
 use crate::io::json::{json_stringify, json_write, JsonPretty};
 use crate::io::ndjson::NdjsonWriter;
 use crate::tree::tree::CladeNodeAttrKeyDesc;
 use crate::types::outputs::{
   combine_outputs_and_errors_sorted, NextcladeErrorOutputs, NextcladeOutputOrError, NextcladeOutputs,
 };
 use crate::utils::datetime::date_iso_now;
+use crate::utils::info::this_package_version_str;
 use eyre::Report;
 use serde::{Deserialize, Serialize};
 use std::path::{Path, PathBuf};
-use crate::utils::info::this_package_version_str;
 
 #[derive(Serialize, Deserialize, schemars::JsonSchema)]
 #[serde(rename_all = "camelCase")]
@@ -21,6 +22,8 @@ pub struct ResultsJson {
   #[serde(skip_serializing_if = "Option::is_none")]
   pub nextclade_web_version: Option<String>,
 
+  pub dataset_info: DatasetInfoShort,
+
   pub created_at: String,
 
   pub clade_node_attr_keys: Vec<CladeNodeAttrKeyDesc>,
@@ -33,11 +36,16 @@ pub struct ResultsJson {
 }
 
 impl ResultsJson {
-  pub fn new(clade_node_attrs: &[CladeNodeAttrKeyDesc], phenotype_attr_keys: &[PhenotypeAttrDesc]) -> Self {
+  pub fn new(
+    dataset_info: &DatasetInfoShort,
+    clade_node_attrs: &[CladeNodeAttrKeyDesc],
+    phenotype_attr_keys: &[PhenotypeAttrDesc],
+  ) -> Self {
     Self {
       schema_version: "3.0.0".to_owned(),
       nextclade_algo_version: this_package_version_str().to_owned(),
       nextclade_web_version: None,
+      dataset_info: dataset_info.clone(),
       created_at: date_iso_now(),
       clade_node_attr_keys: clade_node_attrs.to_vec(),
       phenotype_attr_keys: phenotype_attr_keys.to_vec(),
@@ -47,13 +55,14 @@ impl ResultsJson {
   }
 
   pub fn from_outputs(
+    dataset_info: &DatasetInfoShort,
     outputs: &[NextcladeOutputs],
     errors: &[NextcladeErrorOutputs],
     clade_node_attrs: &[CladeNodeAttrKeyDesc],
     phenotype_attr_keys: &[PhenotypeAttrDesc],
     nextclade_web_version: &Option<String>,
   ) -> Self {
-    let mut this = Self::new(clade_node_attrs, phenotype_attr_keys);
+    let mut this = Self::new(dataset_info, clade_node_attrs, phenotype_attr_keys);
     this.results = outputs.to_vec();
     this.errors = errors.to_vec();
     this.nextclade_web_version = nextclade_web_version.clone();
@@ -68,13 +77,14 @@ pub struct ResultsJsonWriter {
 
 impl ResultsJsonWriter {
   pub fn new(
+    dataset_info: &DatasetInfoShort,
     filepath: impl AsRef<Path>,
     clade_node_attrs: &[CladeNodeAttrKeyDesc],
     phenotype_attr_keys: &[PhenotypeAttrDesc],
   ) -> Result<Self, Report> {
     Ok(Self {
       filepath: filepath.as_ref().to_owned(),
-      result: ResultsJson::new(clade_node_attrs, phenotype_attr_keys),
+      result: ResultsJson::new(dataset_info, clade_node_attrs, phenotype_attr_keys),
     })
   }
 
@@ -103,13 +113,15 @@ impl Drop for ResultsJsonWriter {
 }
 
 pub fn results_to_json_string(
+  dataset_info: &DatasetInfoShort,
   outputs: &[NextcladeOutputs],
   errors: &[NextcladeErrorOutputs],
   clade_node_attrs: &[CladeNodeAttrKeyDesc],
   phenotype_attr_keys: &[PhenotypeAttrDesc],
   nextclade_web_version: &Option<String>,
 ) -> Result<String, Report> {
   let results_json = ResultsJson::from_outputs(
+    dataset_info,
     outputs,
     errors,
     clade_node_attrs,

packages_rs/nextclade/src/run/nextclade_wasm.rs

@@ -8,6 +8,7 @@ use crate::analyze::phenotype::get_phenotype_attr_descs;
 use crate::analyze::virus_properties::{AaMotifsDesc, PhenotypeAttrDesc, VirusProperties};
 use crate::gene::gene_map::GeneMap;
 use crate::graph::graph::{convert_auspice_tree_to_graph, convert_graph_to_auspice_tree};
+use crate::io::dataset::DatasetInfoShort;
 use crate::io::fasta::{read_one_fasta_str, FastaRecord};
 use crate::io::nextclade_csv::CsvColumnConfig;
 use crate::io::nwk_writer::convert_graph_to_nwk_string;
@@ -22,6 +23,7 @@ use crate::types::outputs::NextcladeOutputs;
 use eyre::{Report, WrapErr};
 use itertools::Itertools;
 use schemars::JsonSchema;
+use serde::__private::de::IdentifierDeserializer;
 use serde::{Deserialize, Serialize};
 use std::collections::BTreeMap;
 use std::str::FromStr;
@@ -34,6 +36,7 @@ pub struct NextcladeParams {
   pub gene_map: GeneMap,
   pub tree: Option<AuspiceTree>,
   pub virus_properties: VirusProperties,
+  pub dataset_info: DatasetInfoShort,
 }
 
 impl NextcladeParams {
@@ -52,11 +55,14 @@ impl NextcladeParams {
       |gene_map| GeneMap::from_str(gene_map).wrap_err("When parsing genome annotation"),
     )?;
 
+    let dataset_info = DatasetInfoShort::from_str(&raw.dataset_info).wrap_err("When parsing dataset short info")?;
+
     Ok(Self {
       ref_record,
       gene_map,
       tree,
       virus_properties,
+      dataset_info,
     })
   }
 }
@@ -69,6 +75,7 @@ pub struct NextcladeParamsRaw {
   pub gene_map: Option<String>,
   pub tree: Option<String>,
   pub virus_properties: String,
+  pub dataset_info: String,
 }
 
 #[derive(Clone, Debug, Serialize, Deserialize, schemars::JsonSchema)]
@@ -158,6 +165,7 @@ impl Nextclade {
       gene_map,
       tree,
       virus_properties,
+      dataset_info,
     } = inputs;
 
     let params = NextcladeInputParams::from_optional(params, &virus_properties)?;