Add status of taxon samples variable.

README.md

@@ -2,7 +2,7 @@
 
 A command line interface for GoaT (Genomes on a Tree). GoaT presents metadata for taxa across the tree of life. 
 
-The CLI here builds URLs to query the <b><a href="https://goat.genomehubs.org/api-docs/">Goat API</a></b>. The underlying API is very powerful, but complex. The aim of this CLI is to remove some of the complexity for the end user. TSV's are returned.
+The CLI here builds URLs to query the <b><a href="https://goat.genomehubs.org/api-docs/">Goat API</a></b>. The underlying API is very powerful, but complex. The aim of this CLI is to remove some of the complexity for the end user. Only TSV files are returned. 
 
 This CLI is actively being built so no guarantees are made for stability or bugs.
 
@@ -100,6 +100,9 @@ FLAGS:
     -P, --ploidy               Print ploidy estimates.
     -r, --raw                  Print raw values (i.e. no aggregation/summary).
     -S, --sex-determination    Print sex determination data.
+        --status               Print all data associated with how far this taxon has progressed with genomic sequencing.
+                               This includes sample collection, acquisition, progress in sequencing, and whether
+                               submitted to INSDC.
     -t, --target-lists         Print target list data associated with each taxon.
     -T, --tidy                 Print data in tidy format.
     -u, --url                  Print the underlying GoaT API URL(s). Useful for debugging.

src/main.rs

@@ -188,6 +188,11 @@ async fn main() -> Result<()> {
                         .long("include-estimates")
                         .conflicts_with("raw")
                         .help("Include ancestral estimates. Omitting this flag includes only direct estimates from a taxon. Cannot be used with --raw.")
+                )
+                .arg(
+                    Arg::with_name("status")
+                        .long("status")
+                        .help("Print all data associated with how far this taxon has progressed with genomic sequencing.\nThis includes sample collection, acquisition, progress in sequencing, and whether submitted to INSDC.")
                 ),
         )
         // copy of the above.
@@ -363,6 +368,11 @@ async fn main() -> Result<()> {
                         .long("include-estimates")
                         .conflicts_with("raw")
                         .help("Include ancestral estimates. Omitting this flag includes only direct estimates from a taxon. Cannot be used with --raw.")
+                )
+                .arg(
+                    Arg::with_name("status")
+                        .long("status")
+                        .help("Print all data associated with how far this taxon has progressed with genomic sequencing.\nThis includes sample collection, acquisition, progress in sequencing, and whether submitted to INSDC.")
                 ),
         )
         .subcommand(

src/progress/progress.rs

@@ -135,7 +135,7 @@ pub async fn progress_bar<'a>(matches: &clap::ArgMatches<'a>, api: &str) -> Resu
 
         bar.set_position(progress_x_total);
 
-        if progress_x_total == progress_total_total {
+        if progress_x_total >= progress_total_total {
             break;
         }
 

src/utils/cli_matches.rs

@@ -49,6 +49,8 @@ pub fn process_cli_args(
     let country_list = matches.is_present("country-list");
     // including estimates
     let include_estimates = matches.is_present("include-estimates");
+    // status
+    let status = matches.is_present("status");
 
     // merge the field flags
     let fields = url::FieldBuilder {
@@ -69,6 +71,7 @@ pub fn process_cli_args(
         plastid,
         ploidy,
         sex_determination,
+        status,
         target_lists,
         tidy,
     };

src/utils/url.rs

@@ -54,24 +54,25 @@ pub struct FieldBuilder {
     pub busco: bool,
     pub country_list: bool,
     pub cvalues: bool,
+    pub date: bool,
+    pub gene_count: bool,
     pub gs: bool,
     pub karyotype: bool,
+    pub legislation: bool,
     pub mitochondrion: bool,
+    pub n50: bool,
+    pub names: bool,
     pub plastid: bool,
     pub ploidy: bool,
     pub sex_determination: bool,
-    pub legislation: bool,
-    pub names: bool,
+    pub status: bool,
     pub target_lists: bool,
-    pub n50: bool,
     pub tidy: bool,
-    pub gene_count: bool,
-    pub date: bool,
 }
 
 impl FieldBuilder {
     // private fn used below in build_fields_string
-    fn as_array(&self) -> [bool; 19] {
+    fn as_array(&self) -> [bool; 20] {
         [
             self.all,
             self.assembly,
@@ -90,11 +91,13 @@ impl FieldBuilder {
             self.plastid,
             self.ploidy,
             self.sex_determination,
+            self.status,
             self.target_lists,
             self.tidy,
         ]
     }
 
+    // add
     pub fn build_fields_string(&self) -> String {
         let base = "&fields=";
         let delimiter = "%2C";
@@ -137,6 +140,16 @@ impl FieldBuilder {
         let ebp_metric_date = "ebp_metric_date";
         // country list
         let country_list = "country_list";
+        // sequencing status of the taxon
+        // lump all these together at the moment.
+        let sequencing_status = "sequencing_status";
+        let sample_collected = "sample_collected";
+        let sample_acquired = "sample_acquired";
+        let in_progress = "in_progress";
+        let insdc_submitted = "insdc_submitted";
+        let insdc_open = "insdc_open";
+        let published = "published";
+        let sample_collected_by = "sample_collected_by";
 
         let field_array = self.as_array();
         let mut field_string = String::new();
@@ -244,6 +257,24 @@ impl FieldBuilder {
             field_string += country_list;
             field_string += delimiter;
         }
+        if self.status || self.all {
+            field_string += sequencing_status;
+            field_string += delimiter;
+            field_string += sample_collected;
+            field_string += delimiter;
+            field_string += sample_acquired;
+            field_string += delimiter;
+            field_string += in_progress;
+            field_string += delimiter;
+            field_string += insdc_submitted;
+            field_string += delimiter;
+            field_string += insdc_open;
+            field_string += delimiter;
+            field_string += published;
+            field_string += delimiter;
+            field_string += sample_collected_by;
+            field_string += delimiter;
+        }
 
         // remove the last three chars == '&2C'
         field_string.drain(field_string.len() - 3..);