compatibility with gnames 0.8.0 (close #89)

Author: dimus

.markdownlint.json

@@ -0,0 +1,5 @@
+{
+  "default": true,
+  "MD003": { "style": "atx" },
+  "no-hard-tabs": false
+}

CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## [v0.8.0] - 2022-02-24 Thu
+
+- Add[#89]: compatibility with gnames v0.8.0
+
 ## [v0.7.3] - 2022-02-14 Mon
 
 - Add: make gnverifier compatible with gnames v0.7.1
@@ -139,17 +143,18 @@
 
 This document follows [changelog guidelines]
 
-[v0.7.3]: https://github.com/gnames/gnverifier/compare/v0.7.2...v1.7.3
-[v0.7.2]: https://github.com/gnames/gnverifier/compare/v0.7.1...v1.7.2
-[v0.7.1]: https://github.com/gnames/gnverifier/compare/v0.7.0...v1.7.1
-[v0.7.0]: https://github.com/gnames/gnverifier/compare/v0.6.6...v1.7.0
-[v0.6.6]: https://github.com/gnames/gnverifier/compare/v0.6.5...v1.6.6
-[v0.6.5]: https://github.com/gnames/gnverifier/compare/v0.6.4...v1.6.5
-[v0.6.4]: https://github.com/gnames/gnverifier/compare/v0.6.3...v1.6.4
-[v0.6.3]: https://github.com/gnames/gnverifier/compare/v0.6.2...v1.6.3
-[v0.6.2]: https://github.com/gnames/gnverifier/compare/v0.6.1...v1.6.2
-[v0.6.1]: https://github.com/gnames/gnverifier/compare/v0.6.0...v1.6.1
-[v0.6.0]: https://github.com/gnames/gnverifier/compare/v0.5.2...v1.6.0
+[v0.8.0]: https://github.com/gnames/gnverifier/compare/v0.7.3...v0.8.0
+[v0.7.3]: https://github.com/gnames/gnverifier/compare/v0.7.2...v0.7.3
+[v0.7.2]: https://github.com/gnames/gnverifier/compare/v0.7.1...v0.7.2
+[v0.7.1]: https://github.com/gnames/gnverifier/compare/v0.7.0...v0.7.1
+[v0.7.0]: https://github.com/gnames/gnverifier/compare/v0.6.6...v0.7.0
+[v0.6.6]: https://github.com/gnames/gnverifier/compare/v0.6.5...v0.6.6
+[v0.6.5]: https://github.com/gnames/gnverifier/compare/v0.6.4...v0.6.5
+[v0.6.4]: https://github.com/gnames/gnverifier/compare/v0.6.3...v0.6.4
+[v0.6.3]: https://github.com/gnames/gnverifier/compare/v0.6.2...v0.6.3
+[v0.6.2]: https://github.com/gnames/gnverifier/compare/v0.6.1...v0.6.2
+[v0.6.1]: https://github.com/gnames/gnverifier/compare/v0.6.0...v0.6.1
+[v0.6.0]: https://github.com/gnames/gnverifier/compare/v0.5.2...v0.6.0
 [v0.5.2]: https://github.com/gnames/gnverifier/compare/v0.5.1...v0.5.2
 [v0.5.1]: https://github.com/gnames/gnverifier/compare/v0.5.0...v0.5.1
 [v0.5.0]: https://github.com/gnames/gnverifier/compare/v0.4.1...v0.5.0

README.md

@@ -30,7 +30,6 @@ search feature.
     * [capitalize](#capitalize)
     * [format](#format)
     * [jobs](#jobs)
-    * [only_preferred](#only_preferred)
     * [quiet](#quiet)
     * [sources](#sources)
   * [web-logs](#web-logs)
@@ -299,20 +298,6 @@ gnverifier -j 1 file.txt
 
 This option is ignored by advanced search.
 
-#### only_preferred
-
-Sometimes a users wants to map a list of names to a DataSource. They
-are not interested if name matched anywhere else. In such case you can use
-the ``only_preferred`` flag.
-
-```bash
-gnverifier -o -s '12' file.txt
-# or
-gnverifier --only_preferred --sources='1,12' file.tsv
-```
-
-In case of advanced search use `all:t` together with this flag.
-
 #### quiet
 
 Removes log messages from the output. Note that results of verification go
@@ -398,7 +383,6 @@ possible to use environment variables for configuration.
 | Env. Var.               | Configuration      |
 | :---------------------- | :----------------- |
 | GNV_FORMAT              | Format             |
-| GNV_PREFERRED_ONLY      | PreferredOnly      |
 | GNV_DATA_SOURCES        | DataSources        |
 | GNV_WITH_ALL_MATCHES    | WithAllMatches     |
 | GNV_WITH_CAPITALIZATION | WithCapitalization |

config/config.go

@@ -50,10 +50,6 @@ type Config struct {
 	// Example: `api\/v(0|1)`
 	NsqdRegexFilter *regexp.Regexp
 
-	// PreferredOnly hides BestResult if the user wants to see only
-	// preferred results.
-	PreferredOnly bool
-
 	// VerifierURL URL for gnames verification service. It only needs to
 	// be changed if user sets local version of gnames.
 	VerifierURL string
@@ -126,14 +122,6 @@ func OptNsqdTCPAddress(s string) Option {
 	}
 }
 
-// OptPreferredOnly sets PreferredOnly field. If it is true output only
-// contains results from preferred data-sources.
-func OptPreferredOnly(b bool) Option {
-	return func(cnf *Config) {
-		cnf.PreferredOnly = b
-	}
-}
-
 // OptVerifierURL sets URL of the verification resource.
 func OptVerifierURL(s string) Option {
 	return func(cnf *Config) {

config/config_test.go

@@ -16,23 +16,21 @@ func TestConfigDefault(t *testing.T) {
 		Format:      gnfmt.CSV,
 		VerifierURL: "https://verifier.globalnames.org/api/v0/",
 	}
-	assert.Equal(t, cnf.Format, deflt.Format)
-	assert.Equal(t, cnf.VerifierURL, deflt.VerifierURL)
+	assert.Equal(t, deflt.Format, cnf.Format)
+	assert.Equal(t, deflt.VerifierURL, cnf.VerifierURL)
 }
 
 func TestConfigOpts(t *testing.T) {
 	opts := opts()
 	cnf := config.New(opts...)
 	updt := config.Config{
-		Format:        gnfmt.PrettyJSON,
-		PreferredOnly: true,
-		DataSources:   []int{1, 2, 3},
-		VerifierURL:   url,
+		Format:      gnfmt.PrettyJSON,
+		DataSources: []int{1, 2, 3},
+		VerifierURL: url,
 	}
-	assert.Equal(t, cnf.Format, updt.Format)
-	assert.Equal(t, cnf.PreferredOnly, updt.PreferredOnly)
-	assert.Equal(t, cnf.DataSources, updt.DataSources)
-	assert.Equal(t, cnf.VerifierURL, updt.VerifierURL)
+	assert.Equal(t, updt.Format, cnf.Format)
+	assert.Equal(t, updt.DataSources, cnf.DataSources)
+	assert.Equal(t, updt.VerifierURL, cnf.VerifierURL)
 }
 
 type formatTest struct {
@@ -43,7 +41,6 @@ type formatTest struct {
 func opts() []config.Option {
 	return []config.Option{
 		config.OptFormat(gnfmt.PrettyJSON),
-		config.OptPreferredOnly(true),
 		config.OptDataSources([]int{1, 2, 3}),
 		config.OptVerifierURL(url),
 	}

ent/output/output.go

@@ -1,6 +1,7 @@
 package output
 
 import (
+	"fmt"
 	"strconv"
 	"strings"
 
@@ -12,6 +13,7 @@ type csvField int
 
 const (
 	kind csvField = iota
+	sortScore
 	matchType
 	editDistance
 	input
@@ -26,27 +28,27 @@ const (
 	error
 )
 
-const prefMatch = "PreferredMatch"
+const sortedMatch = "SortedMatch"
 
 // NameOutput takes result of verification for one string and converts it into
 // required format (CSV or JSON).
-func NameOutput(ver vlib.Name, f gnfmt.Format, prefOnly bool) string {
+func NameOutput(ver vlib.Name, f gnfmt.Format) string {
 	switch f {
 	case gnfmt.CSV:
-		return csvOutput(ver, prefOnly, ',')
+		return csvOutput(ver, ',')
 	case gnfmt.TSV:
-		return csvOutput(ver, prefOnly, '\t')
+		return csvOutput(ver, '\t')
 	case gnfmt.CompactJSON:
-		return jsonOutput(ver, prefOnly, false)
+		return jsonOutput(ver, false)
 	case gnfmt.PrettyJSON:
-		return jsonOutput(ver, prefOnly, true)
+		return jsonOutput(ver, true)
 	}
 	return "N/A"
 }
 
 // CSVHeader returns the header string for CSV output format.
 func CSVHeader(f gnfmt.Format) string {
-	header := []string{"Kind", "MatchType", "EditDistance", "ScientificName",
+	header := []string{"Kind", "SortScore", "MatchType", "EditDistance", "ScientificName",
 		"MatchedName", "MatchedCanonical", "TaxonId", "CurrentName", "Synonym",
 		"DataSourceId", "DataSourceTitle", "ClassificationPath", "Error"}
 	switch f {
@@ -59,14 +61,13 @@ func CSVHeader(f gnfmt.Format) string {
 	}
 }
 
-func csvOutput(ver vlib.Name, prefOnly bool, sep rune) string {
+func csvOutput(ver vlib.Name, sep rune) string {
 	var res []string
-	if !prefOnly {
+	if ver.BestResult != nil {
 		best := csvRow(ver, -1, sep)
 		res = append(res, best)
-	}
-	if prefOnly && len(ver.Results) == 0 {
-		res = append(res, csvNoPrefRow(ver, sep))
+	} else if len(ver.Results) == 0 {
+		res = append(res, csvEmptyRow(ver, sep))
 	}
 	for i := range ver.Results {
 		pref := csvRow(ver, i, sep)
@@ -76,9 +77,9 @@ func csvOutput(ver vlib.Name, prefOnly bool, sep rune) string {
 	return strings.Join(res, "\n")
 }
 
-func csvNoPrefRow(ver vlib.Name, sep rune) string {
+func csvEmptyRow(ver vlib.Name, sep rune) string {
 	s := []string{
-		prefMatch, vlib.NoMatch.String(), "", ver.Name,
+		sortedMatch, "0.0", vlib.NoMatch.String(), "", ver.Name,
 		"", "", "", "", "", "", "", "", ver.Error,
 	}
 	return gnfmt.ToCSV(s, sep)
@@ -89,17 +90,21 @@ func csvRow(ver vlib.Name, prefIndex int, sep rune) string {
 	res := ver.BestResult
 
 	if prefIndex > -1 {
-		kind = prefMatch
+		if prefIndex > 0 {
+			kind = sortedMatch
+		}
 		res = ver.Results[prefIndex]
 	}
 
 	s := []string{
-		kind, ver.MatchType.String(), "", ver.Name,
+		kind, "0.0", vlib.NoMatch.String(), "", ver.Name,
 		"", "", "", "", "", "", "", "", ver.Error,
 	}
 
 	if res != nil {
 		s[editDistance] = strconv.Itoa(res.EditDistance)
+		s[sortScore] = fmt.Sprintf("%0.5f", res.SortScore)
+		s[matchType] = res.MatchType.String()
 		s[matchedName] = res.MatchedName
 		s[matchedCanonical] = res.MatchedCanonicalFull
 		s[taxonID] = res.RecordID
@@ -113,11 +118,8 @@ func csvRow(ver vlib.Name, prefIndex int, sep rune) string {
 	return gnfmt.ToCSV(s, sep)
 }
 
-func jsonOutput(ver vlib.Name, prefOnly bool, pretty bool) string {
+func jsonOutput(ver vlib.Name, pretty bool) string {
 	enc := gnfmt.GNjson{Pretty: pretty}
-	if prefOnly {
-		ver.BestResult = nil
-	}
 	res, _ := enc.Encode(ver)
 	return string(res)
 }

ent/output/output_test.go

@@ -19,53 +19,46 @@ func TestOutput(t *testing.T) {
 		msg      string
 		input    vlib.Name
 		format   gnfmt.Format
-		prefOnly bool
 		test     func(*testing.T, string)
 		linesNum int
 	}{
 		{
-			msg:      "csv_prefOnly_false",
-			input:    verifs[0],
-			format:   gnfmt.CSV,
-			prefOnly: false,
+			msg:    "csv",
+			input:  verifs[0],
+			format: gnfmt.CSV,
 			test: func(t *testing.T, res string) {
 				assert.NotContains(t, res, "inputID", "csv,false 1")
-				assert.Contains(t, res, "BestMatch", "csv, false 2")
-				assert.Contains(t, res, "PreferredMatch", "csv, false 3")
+				assert.Contains(t, res, "SortedMatch", "csv, false 3")
 				assert.True(t, strings.HasPrefix(res, "BestMatch"), "csv, false 4")
 			},
 			linesNum: 3,
 		},
 		{
-			msg:      "not_name_csv_prefOnly_true",
-			input:    verifs[2],
-			format:   gnfmt.CSV,
-			prefOnly: true,
+			msg:    "not_name_csv",
+			input:  verifs[2],
+			format: gnfmt.CSV,
 			test: func(t *testing.T, res string) {
 				assert.NotContains(t, res, "inputID", "noname, csv, true 1")
-				assert.NotContains(t, res, "BestMatch", "noname, csv, true 2")
-				assert.Contains(t, res, "PreferredMatch", "noname, csv, true 3")
+				assert.Contains(t, res, "SortedMatch", "noname, csv, true 3")
 				assert.Contains(t, res, "NoMatch", "noname, csv, true 4")
 			},
 			linesNum: 1,
 		},
 		{
-			msg:      "pretty",
-			input:    verifs[0],
-			format:   gnfmt.PrettyJSON,
-			prefOnly: false,
+			msg:    "pretty",
+			input:  verifs[0],
+			format: gnfmt.PrettyJSON,
 			test: func(t *testing.T, res string) {
 				assert.Contains(t, res, "id", "pretty 1")
 				assert.Contains(t, res, "bestResult", "pretty 2")
 				assert.Contains(t, res, "results", "pretty 3")
 			},
-			linesNum: 104,
+			linesNum: 107,
 		},
 		{
-			msg:      "compact",
-			input:    verifs[0],
-			format:   gnfmt.CompactJSON,
-			prefOnly: false,
+			msg:    "compact",
+			input:  verifs[0],
+			format: gnfmt.CompactJSON,
 			test: func(t *testing.T, res string) {
 				assert.Contains(t, res, "id", "compact 1")
 				assert.Contains(t, res, "bestResult", "compact 2")
@@ -77,9 +70,9 @@ func TestOutput(t *testing.T) {
 
 	for i := range tests {
 		t.Run(tests[i].msg, func(t *testing.T) {
-			res := output.NameOutput(tests[i].input, tests[i].format, tests[i].prefOnly)
+			res := output.NameOutput(tests[i].input, tests[i].format)
 			lines := strings.Split(res, "\n")
-			assert.Equal(t, len(lines), tests[i].linesNum)
+			assert.Equal(t, tests[i].linesNum, len(lines))
 			tests[i].test(t, res)
 		})
 	}

gnverifier/cmd/gnverifier.yaml

@@ -2,11 +2,6 @@
 #
 # Format: csv
 
-# PreferredOnly if true, do not show BestResult, only Preferred Results.
-# Its valid values are 'true' and 'false'.
-#
-# PreferredOnly: false
-
 # DataSources is a list of data-source IDs that should always return
 # matched records if they are found.
 # You can find list of all data-sources at