Kharec/goyco
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Compare commits

base: Kharec:325cbe9c2c13634a87755c52f979108e1124ed1f
Branches Tags
Kharec:main
..
compare: Kharec:221f828860ea8790b8c3c0120ea693b1fa19346f
Branches Tags
Kharec:main

2 Commits
325cbe9c2c ... 221f828860

Author SHA1 Message Date
Kharec
221f828860 clean: remove obsoletes tests 2025-11-11 16:19:26 +01:00
Kharec
c2d4357262 feat: keep only one method to get title of a page 2025-11-11 16:19:09 +01:00
2 changed files with 8 additions and 380 deletions
Expand all files Collapse all files

126
internal/services/url_metadata_service.go
View File

@@ -262,28 +262,7 @@ func (s *URLMetadataService) FetchTitle(ctx context.Context, rawURL string) (str
} }
func (s *URLMetadataService) ExtractTitleFromHTML(html string) string { func (s *URLMetadataService) ExtractTitleFromHTML(html string) string {
return s.ExtractFromTitleTag(html)
if title := s.ExtractFromTitleTag(html); title != "" {
return title
}
if title := s.ExtractFromOpenGraph(html); title != "" {
return title
}
if title := s.ExtractFromJSONLD(html); title != "" {
return title
}
if title := s.ExtractFromTwitterCard(html); title != "" {
return title
}
if title := s.extractFromMetaTags(html); title != "" {
return title
}
return ""
} }
func (s *URLMetadataService) ExtractFromTitleTag(htmlContent string) string { func (s *URLMetadataService) ExtractFromTitleTag(htmlContent string) string {
@@ -313,109 +292,6 @@ func (s *URLMetadataService) ExtractFromTitleTag(htmlContent string) string {
} }
} }
func (s *URLMetadataService) ExtractFromOpenGraph(htmlContent string) string {
lines := strings.Split(htmlContent, "\n")
for _, line := range lines {
line = strings.TrimSpace(line)
if strings.Contains(strings.ToLower(line), `property="og:title"`) && strings.Contains(line, `content="`) {
start := strings.Index(line, `content="`)
if start != -1 {
start += 9
end := strings.Index(line[start:], `"`)
if end != -1 {
title := line[start : start+end]
cleaned := s.optimizedTitleClean(title)
if cleaned != "" {
return cleaned
}
}
}
}
}
return ""
}
func (s *URLMetadataService) ExtractFromJSONLD(htmlContent string) string {
lines := strings.Split(htmlContent, "\n")
for _, line := range lines {
line = strings.TrimSpace(line)
if strings.Contains(line, `"@type":"VideoObject"`) || strings.Contains(line, `"@type":"WebPage"`) {
if strings.Contains(line, `"name":`) {
start := strings.Index(line, `"name":`)
if start != -1 {
start += 7
for i := start; i < len(line); i++ {
if line[i] == '"' {
start = i + 1
break
}
}
end := strings.Index(line[start:], `"`)
if end != -1 {
title := line[start : start+end]
cleaned := s.optimizedTitleClean(title)
if cleaned != "" {
return cleaned
}
}
}
}
}
}
return ""
}
func (s *URLMetadataService) ExtractFromTwitterCard(htmlContent string) string {
lines := strings.Split(htmlContent, "\n")
for _, line := range lines {
line = strings.TrimSpace(line)
if strings.Contains(strings.ToLower(line), `name="twitter:title"`) && strings.Contains(line, `content="`) {
start := strings.Index(line, `content="`)
if start != -1 {
start += 9
end := strings.Index(line[start:], `"`)
if end != -1 {
title := line[start : start+end]
cleaned := s.optimizedTitleClean(title)
if cleaned != "" {
return cleaned
}
}
}
}
}
return ""
}
func (s *URLMetadataService) extractFromMetaTags(htmlContent string) string {
lines := strings.Split(htmlContent, "\n")
for _, line := range lines {
line = strings.TrimSpace(line)
if strings.Contains(strings.ToLower(line), `name="title"`) && strings.Contains(line, `content="`) {
start := strings.Index(line, `content="`)
if start != -1 {
start += 9
end := strings.Index(line[start:], `"`)
if end != -1 {
title := line[start : start+end]
cleaned := s.optimizedTitleClean(title)
if cleaned != "" {
return cleaned
}
}
}
}
}
return ""
}
func (s *URLMetadataService) optimizedTitleClean(title string) string { func (s *URLMetadataService) optimizedTitleClean(title string) string {
if title == "" { if title == "" {
return "" return ""

262
internal/services/url_metadata_service_test.go
View File

@@ -539,239 +539,6 @@ func TestExtractFromTitleTag(t *testing.T) {
} }
} }
func TestExtractFromOpenGraph(t *testing.T) {
svc := NewURLMetadataService()
tests := []struct {
name string
html string
expected string
}{
{
name: "simple og:title",
html: `<meta property="og:title" content="Open Graph Title">`,
expected: "Open Graph Title",
},
{
name: "og:title with whitespace",
html: `<meta property="og:title" content=" Open Graph Title ">`,
expected: "Open Graph Title",
},
{
name: "empty og:title",
html: `<meta property="og:title" content="">`,
expected: "",
},
{
name: "whitespace only og:title",
html: `<meta property="og:title" content=" ">`,
expected: "",
},
{
name: "no og:title",
html: `<meta property="og:description" content="Description">`,
expected: "",
},
{
name: "case insensitive property",
html: `<meta PROPERTY="OG:TITLE" content="Case Insensitive Title">`,
expected: "Case Insensitive Title",
},
{
name: "multiple og:title (first one)",
html: `<meta property="og:title" content="First Title"><meta property="og:title" content="Second Title">`,
expected: "First Title",
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
result := svc.ExtractFromOpenGraph(tt.html)
if result != tt.expected {
t.Fatalf("expected %q, got %q", tt.expected, result)
}
})
}
}
func TestExtractFromJSONLD(t *testing.T) {
svc := NewURLMetadataService()
tests := []struct {
name string
html string
expected string
}{
{
name: "VideoObject with name",
html: `{"@type":"VideoObject","name":"Video Title"}`,
expected: "Video Title",
},
{
name: "WebPage with name",
html: `{"@type":"WebPage","name":"Page Title"}`,
expected: "Page Title",
},
{
name: "VideoObject with whitespace in name",
html: `{"@type":"VideoObject","name":" Video Title "}`,
expected: "Video Title",
},
{
name: "empty name",
html: `{"@type":"VideoObject","name":""}`,
expected: "",
},
{
name: "whitespace only name",
html: `{"@type":"VideoObject","name":" "}`,
expected: "",
},
{
name: "no name field",
html: `{"@type":"VideoObject","description":"Description"}`,
expected: "",
},
{
name: "wrong type",
html: `{"@type":"Article","name":"Article Title"}`,
expected: "",
},
{
name: "no @type",
html: `{"name":"Some Title"}`,
expected: "",
},
{
name: "multiple objects (first VideoObject)",
html: `{"@type":"VideoObject","name":"Video Title"} {"@type":"WebPage","name":"Page Title"}`,
expected: "Video Title",
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
result := svc.ExtractFromJSONLD(tt.html)
if result != tt.expected {
t.Fatalf("expected %q, got %q", tt.expected, result)
}
})
}
}
func TestExtractFromTwitterCard(t *testing.T) {
svc := NewURLMetadataService()
tests := []struct {
name string
html string
expected string
}{
{
name: "simple twitter:title",
html: `<meta name="twitter:title" content="Twitter Title">`,
expected: "Twitter Title",
},
{
name: "twitter:title with whitespace",
html: `<meta name="twitter:title" content=" Twitter Title ">`,
expected: "Twitter Title",
},
{
name: "empty twitter:title",
html: `<meta name="twitter:title" content="">`,
expected: "",
},
{
name: "whitespace only twitter:title",
html: `<meta name="twitter:title" content=" ">`,
expected: "",
},
{
name: "no twitter:title",
html: `<meta name="twitter:description" content="Description">`,
expected: "",
},
{
name: "case insensitive name",
html: `<meta NAME="TWITTER:TITLE" content="Case Insensitive Title">`,
expected: "Case Insensitive Title",
},
{
name: "multiple twitter:title (first one)",
html: `<meta name="twitter:title" content="First Title"><meta name="twitter:title" content="Second Title">`,
expected: "First Title",
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
result := svc.ExtractFromTwitterCard(tt.html)
if result != tt.expected {
t.Fatalf("expected %q, got %q", tt.expected, result)
}
})
}
}
func TestExtractFromMetaTags(t *testing.T) {
svc := NewURLMetadataService()
tests := []struct {
name string
html string
expected string
}{
{
name: "simple meta title",
html: `<meta name="title" content="Meta Title">`,
expected: "Meta Title",
},
{
name: "meta title with whitespace",
html: `<meta name="title" content=" Meta Title ">`,
expected: "Meta Title",
},
{
name: "empty meta title",
html: `<meta name="title" content="">`,
expected: "",
},
{
name: "whitespace only meta title",
html: `<meta name="title" content=" ">`,
expected: "",
},
{
name: "no meta title",
html: `<meta name="description" content="Description">`,
expected: "",
},
{
name: "case insensitive name",
html: `<meta NAME="TITLE" content="Case Insensitive Title">`,
expected: "Case Insensitive Title",
},
{
name: "multiple meta title (first one)",
html: `<meta name="title" content="First Title"><meta name="title" content="Second Title">`,
expected: "First Title",
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
result := svc.extractFromMetaTags(tt.html)
if result != tt.expected {
t.Fatalf("expected %q, got %q", tt.expected, result)
}
})
}
}
func TestExtractTitleFromHTML(t *testing.T) { func TestExtractTitleFromHTML(t *testing.T) {
svc := NewURLMetadataService() svc := NewURLMetadataService()
@@ -781,39 +548,24 @@ func TestExtractTitleFromHTML(t *testing.T) {
expected string expected string
}{ }{
{ {
name: "title tag takes precedence", name: "title tag extracted",
html: `<html><head><title>Title Tag</title><meta property="og:title" content="OG Title"></head></html>`, html: `<html><head><title>Title Tag</title><meta property="og:title" content="OG Title"></head></html>`,
expected: "Title Tag", expected: "Title Tag",
}, },
{ {
name: "og:title fallback when no title tag", name: "no title tag returns empty",
html: `<html><head><meta property="og:title" content="OG Title"></head></html>`, html: `<html><head><meta property="og:title" content="OG Title"></head></html>`,
expected: "OG Title", expected: "",
}, },
{ {
name: "JSON-LD fallback when no title or og", name: "empty title tag returns empty",
html: `<html><head><script type="application/ld+json">{"@type":"VideoObject","name":"JSON Title"}</script></head></html>`,
expected: "JSON Title",
},
{
name: "twitter fallback when no title, og, or json",
html: `<html><head><meta name="twitter:title" content="Twitter Title"></head></html>`,
expected: "Twitter Title",
},
{
name: "meta title fallback when no other methods work",
html: `<html><head><meta name="title" content="Meta Title"></head></html>`,
expected: "Meta Title",
},
{
name: "empty title tag falls back to og:title",
html: `<html><head><title></title><meta property="og:title" content="OG Title"></head></html>`, html: `<html><head><title></title><meta property="og:title" content="OG Title"></head></html>`,
expected: "OG Title", expected: "",
}, },
{ {
name: "whitespace title tag falls back to og:title", name: "whitespace title tag returns empty",
html: `<html><head><title> </title><meta property="og:title" content="OG Title"></head></html>`, html: `<html><head><title> </title><meta property="og:title" content="OG Title"></head></html>`,
expected: "OG Title", expected: "",
}, },
{ {
name: "no title found", name: "no title found",