fix(wip): duplicate leaderboard entries due to inconsistent language spelling

config.default.yml

@@ -40,6 +40,21 @@ app:
     ipynb: Python
     svelte: Svelte
     astro: Astro
+  canonical_language_names:
+    'java': 'Java'
+    'ini': 'INI'
+    'xml': 'XML'
+    'jsx': 'JSX'
+    'tsx': 'TSX'
+    'php': 'PHP'
+    'yaml': 'YAML'
+    'toml': 'TOML'
+    'sql': 'SQL'
+    'css': 'CSS'
+    'scss': 'SCSS'
+    'jsp': 'JSP'
+    'svg': 'SVG'
+    'csv': 'CSV'
 
   # url template for user avatar images (to be used with services like gravatar or dicebear)
   # available variable placeholders are: username, username_hash, email, email_hash

config/config.go

@@ -103,6 +103,7 @@ type appConfig struct {
 	DateFormat                string                       `yaml:"date_format" default:"Mon, 02 Jan 2006" env:"WAKAPI_DATE_FORMAT"`
 	DateTimeFormat            string                       `yaml:"datetime_format" default:"Mon, 02 Jan 2006 15:04" env:"WAKAPI_DATETIME_FORMAT"`
 	CustomLanguages           map[string]string            `yaml:"custom_languages"`
+	CanonicalLanguageNames    map[string]string            `yaml:"canonical_language_names"` // lower case, compacted representation -> canonical name
 	Colors                    map[string]map[string]string `yaml:"-"`
 }
 
@@ -239,6 +240,13 @@ func (c *appConfig) GetCustomLanguages() map[string]string {
 	return utils.CloneStringMap(c.CustomLanguages, false)
 }
 
+func (c *appConfig) GetCanonicalLanguageNames() map[string]string {
+	if c.CanonicalLanguageNames == nil {
+		return make(map[string]string)
+	}
+	return utils.CloneStringMap(c.CanonicalLanguageNames, false)
+}
+
 func (c *appConfig) GetLanguageColors() map[string]string {
 	return utils.CloneStringMap(c.Colors["languages"], true)
 }

models/canonical_names.go

@@ -0,0 +1,40 @@
+package models
+
+import (
+	"github.com/duke-git/lancet/v2/strutil"
+	"github.com/muety/wakapi/config"
+	"regexp"
+	"strings"
+)
+
+// special treatment for system-wide entities (language, editors, os) that are known to commonly cause confusion
+// due to being sent by different plugins in different use of capital and small letters in their spelling, e.g. "JAVA"
+
+var nonAlphanumericRegex = regexp.MustCompile(`[^a-zA-Z0-9]+`)
+var canonicalNames map[uint8]map[string]string
+
+func initLookup() {
+	cfg := config.Get()
+	canonicalNames = map[uint8]map[string]string{
+		SummaryLanguage: cfg.App.GetCanonicalLanguageNames(),
+		SummaryEditor:   {},
+		SummaryOS:       {},
+	}
+}
+
+func CanonicalName(value string, entityType uint8) string {
+	if canonicalNames == nil {
+		initLookup()
+	}
+
+	if _, ok := canonicalNames[entityType]; !ok {
+		return value
+	}
+
+	lookupKey := nonAlphanumericRegex.ReplaceAllString(strings.ToLower(value), "")
+	if canonical, ok := canonicalNames[entityType][lookupKey]; ok {
+		return canonical
+	}
+
+	return strutil.Capitalize(value) // even if no specific canonical name is provided, still always capitalize languages, editors and os for consistency
+}

models/heartbeat.go

@@ -8,7 +8,6 @@ import (
 
 	"log/slog"
 
-	"github.com/duke-git/lancet/v2/strutil"
 	"github.com/gohugoio/hashstructure"
 )
 
@@ -50,8 +49,9 @@ func (h *Heartbeat) Timely(maxAge time.Duration) bool {
 }
 
 func (h *Heartbeat) Sanitize() *Heartbeat {
-	h.OperatingSystem = strutil.Capitalize(h.OperatingSystem)
-	h.Editor = strutil.Capitalize(h.Editor)
+	h.OperatingSystem = CanonicalName(h.OperatingSystem, SummaryOS)
+	h.Editor = CanonicalName(h.Editor, SummaryEditor)
+	h.Language = CanonicalName(h.Language, SummaryLanguage)
 	if h.Category == "" && (h.Type == "domain" || h.Type == "url") {
 		h.Category = "browsing"
 	}

services/alias.go

@@ -76,6 +76,8 @@ func (srv *AliasService) GetAliasOrDefault(userId string, summaryType uint8, val
 		srv.MayInitializeUser(userId)
 	}
 
+	value = models.CanonicalName(value, summaryType) // currently only implemented for languages
+
 	match := func(aliasValue string, itemKey string) bool {
 		return wildmatch.NewWildMatch(aliasValue).IsMatch(itemKey)
 	}

services/alias_test.go

@@ -1,6 +1,7 @@
 package services
 
 import (
+	"github.com/muety/wakapi/config"
 	"github.com/muety/wakapi/mocks"
 	"github.com/muety/wakapi/models"
 	"github.com/stretchr/testify/assert"
@@ -16,6 +17,8 @@ type AliasServiceTestSuite struct {
 }
 
 func (suite *AliasServiceTestSuite) SetupSuite() {
+	config.Set(config.Empty())
+
 	suite.TestUserId = "johndoe@example.org"
 
 	aliases := []*models.Alias{
@@ -51,7 +54,8 @@ func (suite *AliasServiceTestSuite) TestAliasService_GetAliasOrDefault() {
 	result2, err2 := sut.GetAliasOrDefault(suite.TestUserId, models.SummaryProject, "wakapi")
 	result3, err3 := sut.GetAliasOrDefault(suite.TestUserId, models.SummaryProject, "anchr")
 	result4, err4 := sut.GetAliasOrDefault(suite.TestUserId, models.SummaryProject, "telepush-mobile")
-	result5, err5 := sut.GetAliasOrDefault(suite.TestUserId, models.SummaryLanguage, "telepush-mobile")
+	result5, err5 := sut.GetAliasOrDefault(suite.TestUserId, models.SummaryEntity, "telepush-mobile")
+	result6, err6 := sut.GetAliasOrDefault(suite.TestUserId, models.SummaryLanguage, "telepush-mobile")
 
 	assert.Equal(suite.T(), "wakapi", result1)
 	assert.Nil(suite.T(), err1)
@@ -63,4 +67,6 @@ func (suite *AliasServiceTestSuite) TestAliasService_GetAliasOrDefault() {
 	assert.Nil(suite.T(), err4)
 	assert.Equal(suite.T(), "telepush-mobile", result5)
 	assert.Nil(suite.T(), err5)
+	assert.Equal(suite.T(), "Telepush-mobile", result6) // not really scope of this test, but nevertheless: language shall always be capitaliized
+	assert.Nil(suite.T(), err6)
 }