Merge pull request #1266 from nextstrain/feat/cli-read-annotation

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

@@ -3,5 +3,6 @@ pub mod nextclade_dataset_get;
 pub mod nextclade_dataset_list;
 pub mod nextclade_loop;
 pub mod nextclade_ordered_writer;
+pub mod nextclade_read_annotation;
 pub mod nextclade_seq_sort;
 pub mod verbosity;

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

@@ -1,6 +1,7 @@
 use crate::cli::nextclade_dataset_get::nextclade_dataset_get;
 use crate::cli::nextclade_dataset_list::nextclade_dataset_list;
 use crate::cli::nextclade_loop::nextclade_run;
+use crate::cli::nextclade_read_annotation::nextclade_read_annotation;
 use crate::cli::nextclade_seq_sort::nextclade_seq_sort;
 use crate::cli::verbosity::{Verbosity, WarnLevel};
 use crate::io::http_client::ProxyConfig;
@@ -92,6 +93,11 @@ pub enum NextcladeCommands {
   ///
   /// For short help type: `nextclade -h`, for extended help type: `nextclade --help`. Each subcommand has its own help, for example: `nextclade sort --help`.
   Sort(Box<NextcladeSortArgs>),
+
+  /// Read genome annotation and present it in Nextclade's internal formats. This is mostly only useful for Nextclade maintainers and the most curious users. Note that these internal formats have no stability guarantees and can be changed at any time without notice.
+  ///
+  /// For short help type: `nextclade -h`, for extended help type: `nextclade --help`. Each subcommand has its own help, for example: `nextclade sort --help`.
+  ReadAnnotation(Box<NextcladeReadAnnotationArgs>),
 }
 
 #[derive(Parser, Debug)]
@@ -702,6 +708,35 @@ pub struct NextcladeSortArgs {
   pub proxy_config: ProxyConfig,
 }
 
+#[allow(clippy::struct_excessive_bools)]
+#[derive(Parser, Debug)]
+#[clap(verbatim_doc_comment)]
+pub struct NextcladeReadAnnotationArgs {
+  /// Genome annotation file in GFF3 format.
+  ///
+  /// Learn more about Generic Feature Format Version 3 (GFF3):
+  /// https://github.com/The-Sequence-Ontology/Specifications/blob/master/gff3.md
+  ///
+  #[clap(value_hint = ValueHint::FilePath)]
+  #[clap(display_order = 0)]
+  pub input_annotation: Option<PathBuf>,
+
+  /// Path to output JSON or YAML file.
+  ///
+  /// The format is chosen based on file extension: ".json" or ".yaml".
+  #[clap(long, short = 'o')]
+  #[clap(value_hint = ValueHint::DirPath)]
+  pub output: Option<PathBuf>,
+
+  /// Present features in "feature tree" format. This format is a precursor of genome annotation format - it contains all genetic features, even the ones that Nextclade does not use, but also less information about each feature.
+  #[clap(long)]
+  pub feature_tree: bool,
+
+  /// Print console output in JSON format, rather than human-readable table.
+  #[clap(long)]
+  pub json: bool,
+}
+
 fn generate_completions(shell: &str) -> Result<(), Report> {
   let mut command = NextcladeArgs::command();
 
@@ -989,5 +1024,6 @@ pub fn nextclade_parse_cli_args() -> Result<(), Report> {
       NextcladeDatasetCommands::Get(dataset_get_args) => nextclade_dataset_get(&dataset_get_args),
     },
     NextcladeCommands::Sort(seq_sort_args) => nextclade_seq_sort(&seq_sort_args),
+    NextcladeCommands::ReadAnnotation(read_annotation_args) => nextclade_read_annotation(&read_annotation_args),
   }
 }

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

@@ -0,0 +1,54 @@
+use crate::cli::nextclade_cli::NextcladeReadAnnotationArgs;
+use eyre::Report;
+use nextclade::features::feature_tree::FeatureTree;
+use nextclade::gene::gene_map::GeneMap;
+use nextclade::gene::gene_map_display::gene_map_to_table_string;
+use nextclade::io::file::open_file_or_stdin;
+use nextclade::io::json::{json_or_yaml_write, json_stringify, JsonPretty};
+use std::io::Read;
+
+pub fn nextclade_read_annotation(args: &NextcladeReadAnnotationArgs) -> Result<(), Report> {
+  let content = {
+    let mut content = String::new();
+    open_file_or_stdin(&args.input_annotation)?.read_to_string(&mut content)?;
+    content
+  };
+
+  if args.feature_tree {
+    handle_feature_tree(args, &content)
+  } else {
+    handle_genome_annotation(args, &content)
+  }
+}
+
+fn handle_genome_annotation(args: &NextcladeReadAnnotationArgs, content: &str) -> Result<(), Report> {
+  let data = GeneMap::from_str(content)?;
+
+  if args.json {
+    println!("{}\n", json_stringify(&data, JsonPretty(true))?);
+  } else {
+    println!("{}", gene_map_to_table_string(&data)?);
+  }
+
+  if let Some(output) = &args.output {
+    json_or_yaml_write(output, &data)?;
+  }
+
+  Ok(())
+}
+
+fn handle_feature_tree(args: &NextcladeReadAnnotationArgs, content: &str) -> Result<(), Report> {
+  let data = FeatureTree::from_gff3_str(content)?;
+
+  if args.json {
+    println!("{}\n", json_stringify(&data, JsonPretty(true))?);
+  } else {
+    println!("{}", data.to_pretty_string()?);
+  }
+
+  if let Some(output) = &args.output {
+    json_or_yaml_write(output, &data)?;
+  }
+
+  Ok(())
+}

packages_rs/nextclade/src/io/json.rs

@@ -1,4 +1,5 @@
 use crate::io::file::create_file_or_stdout;
+use crate::io::yaml::yaml_write;
 use eyre::{Report, WrapErr};
 use serde::{Deserialize, Serialize};
 use serde_json::{de::Read, Deserializer};
@@ -61,3 +62,13 @@ pub fn json_write<T: Serialize>(filepath: impl AsRef<Path>, obj: &T, pretty: Jso
   let file = create_file_or_stdout(filepath)?;
   json_write_impl(file, &obj, pretty).wrap_err("When writing JSON to file: {filepath:#?}")
 }
+
+pub fn json_or_yaml_write<T: Serialize>(filepath: impl AsRef<Path>, obj: &T) -> Result<(), Report> {
+  let filepath = filepath.as_ref();
+  let filepath_str = filepath.to_string_lossy();
+  if filepath_str.ends_with("yaml") || filepath_str.ends_with("yml") {
+    yaml_write(filepath, &obj)
+  } else {
+    json_write(filepath, &obj, JsonPretty(true))
+  }
+}