add ranks and ids to classification dump (close #53)

gnames · Sep 12, 2022 · 48d819e · 48d819e
1 parent ff29566
commit 48d819e
Show file tree

Hide file tree

Showing 6 changed files with 28 additions and 10 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -2,6 +2,10 @@
 
 ## Unreleased
 
+## [v0.13.2] - 2022-09-12 Mon
+
+- Add [#53] - classification ranks and ids in dump files.
+
 ## [v0.13.1] - 2022-09-08 Thu
 
 - Add [#52]: dump pages information

diff --git a/ent/output/name-format.go b/ent/output/name-format.go
@@ -15,7 +15,8 @@ func (on OutputName) header() []string {
 		"NameID", "DetectedName", "Cardinality", "OccurrencesNumber", "OddsLog10",
 		"MatchType", "EditDistance", "StemEditDistance", "MatchedCanonical",
 		"MatchedFullName", "MatchedCardinality", "CurrentCanonical",
-		"CurrentFullName", "CurrentCardinality", "Classification", "RecordID",
+		"CurrentFullName", "CurrentCardinality", "Classification",
+		"ClassificationRanks", "ClassificationIDs", "RecordID",
 		"DataSourceID", "DataSource", "DataSourcesNumber", "Curation", "Error",
 	}
 }
@@ -37,10 +38,11 @@ func (on OutputName) csvOutput(sep rune) string {
 	dsNum := strconv.Itoa(on.DataSourcesNumber)
 
 	s := []string{
-		on.NameID, on.DetectedName, card, occNum, odds, on.MatchType,
-		eDist, stEDist, on.MatchedCanonical, on.MatchedFullName, matchCard,
+		on.NameID, on.DetectedName, card, occNum, odds, on.MatchType, eDist,
+		stEDist, on.MatchedCanonical, on.MatchedFullName, matchCard,
 		on.CurrentCanonical, on.CurrentFullName, currCard, on.Classification,
-		on.RecordID, dsID, on.DataSource, dsNum, on.Curation, on.VerifError,
+		on.ClassificationRanks, on.ClassificationIDs, on.RecordID, dsID,
+		on.DataSource, dsNum, on.Curation, on.VerifError,
 	}
 
 	return gnfmt.ToCSV(s, sep)

diff --git a/ent/output/output.go b/ent/output/output.go
@@ -80,6 +80,13 @@ type OutputName struct {
 	// DataSource.
 	Classification string `json:"classification"`
 
+	// ClassificationRanks provide data about ranks used in the classificaiton.
+	ClassificationRanks string `json:"classificationRanks"`
+
+	// ClassificationIDs provides data about IDs a DataSource assigns to
+	// taxons in the classification.
+	ClassificationIDs string `json:"classificationIDs"`
+
 	// RecordID is the ID assigned by the DataSource to the name.
 	RecordID string `json:"recordID"`
 
@@ -150,6 +157,8 @@ type OutputOccurrence struct {
 	Annotation string `json:"annotNomen"`
 }
 
+// CSVHeader takes any object that implements Output interface
+// and generates TSV or CSV header.
 func CSVHeader[O Output](o O, f gnfmt.Format) string {
 	sep := rune(',')
 	if f == gnfmt.TSV {

diff --git a/io/dumpio/db-dump.go b/io/dumpio/db-dump.go
@@ -102,9 +102,10 @@ func (d *dumpio) outputNames(id, limit int, ds []int) ([]output.OutputName, erro
 	q := fmt.Sprintf(`
 SELECT
   name, cardinality, occurrences, odds_log10, match_type, edit_distance,
-  stem_edit_distance, matched_canonical, matched_name, matched_cardinality,
-  current_canonical, current_name, current_cardinality, classification,
-  record_id, data_source_id, data_source_title, data_sources_number,
+    stem_edit_distance, matched_canonical, matched_name, matched_cardinality,
+    current_canonical, current_name, current_cardinality, classification,
+    classification_ranks, classification_ids, record_id, data_source_id,
+    data_source_title, data_sources_number,
   curation, error
   FROM verified_names
   WHERE name_id >= $1 and name_id < $2
@@ -125,8 +126,9 @@ ORDER by name_id
 			&o.MatchType, &o.EditDistance, &o.StemEditDistance, &o.MatchedCanonical,
 			&o.MatchedFullName, &o.MatchedCardinality, &o.CurrentCanonical,
 			&o.CurrentFullName, &o.CurrentCardinality, &o.Classification,
-			&o.RecordID, &o.DataSourceID, &o.DataSource, &o.DataSourcesNumber,
-			&o.Curation, &o.VerifError,
+			&o.ClassificationRanks, &o.ClassificationIDs, &o.RecordID,
+			&o.DataSourceID, &o.DataSource, &o.DataSourcesNumber, &o.Curation,
+			&o.VerifError,
 		)
 		if err != nil {
 			return nil, fmt.Errorf("outputNames: %w", err)

diff --git a/io/dumpio/dumpio.go b/io/dumpio/dumpio.go
@@ -63,6 +63,7 @@ func (d *dumpio) DumpPages(ctx context.Context, ch chan<- []output.OutputPage) e
 	return nil
 }
 
+// DumpNames outputs data about verified names.
 func (d *dumpio) DumpNames(ctx context.Context, ch chan<- []output.OutputName, ds []int) error {
 	err := d.checkForVerifiedNames()
 	if err != nil {

diff --git a/version.go b/version.go
@@ -1,6 +1,6 @@
 package bhlindex
 
 var (
-	Version = "v0.13.1+"
+	Version = "v0.13.2+"
 	Build   string
 )