pkg/analyzer/content.go

@@ -27,7 +27,6 @@ import (
 	"net/http"
 	"net/url"
 	"regexp"
-	"slices"
 	"strings"
 	"time"
 	"unicode"
@@ -38,9 +37,8 @@ import (
 
 // ContentAnalyzer analyzes email content (HTML, links, images)
 type ContentAnalyzer struct {
-	Timeout             time.Duration
-	httpClient          *http.Client
-	listUnsubscribeURLs []string // URLs from List-Unsubscribe header
+	Timeout    time.Duration
+	httpClient *http.Client
 }
 
 // NewContentAnalyzer creates a new content analyzer with configurable timeout
@@ -112,9 +110,6 @@ func (c *ContentAnalyzer) AnalyzeContent(email *EmailMessage) *ContentResults {
 
 	results.IsMultipart = len(email.Parts) > 1
 
-	// Parse List-Unsubscribe header URLs for use in link detection
-	c.listUnsubscribeURLs = email.GetListUnsubscribeURLs()
-
 	// Get HTML and text parts
 	htmlParts := email.GetHTMLParts()
 	textParts := email.GetTextParts()
@@ -336,11 +331,6 @@ func (c *ContentAnalyzer) getAttr(n *html.Node, key string) string {
 
 // isUnsubscribeLink checks if a link is an unsubscribe link
 func (c *ContentAnalyzer) isUnsubscribeLink(href string, node *html.Node) bool {
-	// First check: does the href match a URL from the List-Unsubscribe header?
-	if slices.Contains(c.listUnsubscribeURLs, href) {
-		return true
-	}
-
 	// Check href for unsubscribe keywords
 	lowerHref := strings.ToLower(href)
 	unsubKeywords := []string{"unsubscribe", "opt-out", "optout", "remove", "list-unsubscribe"}
@@ -449,8 +439,7 @@ func (c *ContentAnalyzer) hasDomainMisalignment(href, linkText string) bool {
 	// Extract the actual destination domain/email based on scheme
 	var actualDomain string
 
-	switch parsedURL.Scheme {
-	case "mailto":
+	if parsedURL.Scheme == "mailto" {
 		// Extract email address from mailto: URL
 		// Format can be: mailto:user@domain.com or mailto:user@domain.com?subject=...
 		mailtoAddr := parsedURL.Opaque
@@ -468,8 +457,7 @@ func (c *ContentAnalyzer) hasDomainMisalignment(href, linkText string) bool {
 		} else {
 			return false // Invalid mailto
 		}
-	case "http":
-	case "https":
+	} else if parsedURL.Scheme == "http" || parsedURL.Scheme == "https" {
 		// Check if URL has a host
 		if parsedURL.Host == "" {
 			return false
@@ -481,7 +469,7 @@ func (c *ContentAnalyzer) hasDomainMisalignment(href, linkText string) bool {
 			actualDomain = actualDomain[:idx]
 		}
 		actualDomain = strings.ToLower(actualDomain)
-	default:
+	} else {
 		// Skip checks for other URL schemes (tel, etc.)
 		return false
 	}
@@ -504,8 +492,10 @@ func (c *ContentAnalyzer) hasDomainMisalignment(href, linkText string) bool {
 		"email us", "contact us", "send email", "get in touch", "reach out",
 		"contact", "email", "write to us",
 	}
-	if slices.Contains(genericTexts, linkText) {
-		return false
+	for _, generic := range genericTexts {
+		if linkText == generic {
+			return false
+		}
 	}
 
 	// Extract domain-like patterns from link text using regex
@@ -572,8 +562,10 @@ func (c *ContentAnalyzer) isSuspiciousURL(urlStr string, parsedURL *url.URL) boo
 		"bit.ly", "tinyurl.com", "goo.gl", "ow.ly", "t.co",
 		"buff.ly", "is.gd", "bl.ink", "short.io",
 	}
-	if slices.Contains(shorteners, strings.ToLower(parsedURL.Host)) {
-		return true
+	for _, shortener := range shorteners {
+		if strings.ToLower(parsedURL.Host) == shortener {
+			return true
+		}
 	}
 
 	// Check for excessive subdomains (possible obfuscation)

pkg/analyzer/parser.go

@@ -301,20 +301,3 @@ func (e *EmailMessage) GetHeaderValue(key string) string {
 func (e *EmailMessage) HasHeader(key string) bool {
 	return e.Header.Get(key) != ""
 }
-
-// GetListUnsubscribeURLs parses the List-Unsubscribe header and returns all URLs.
-// The header format is: <url1>, <url2>, ...
-func (e *EmailMessage) GetListUnsubscribeURLs() []string {
-	value := e.Header.Get("List-Unsubscribe")
-	if value == "" {
-		return nil
-	}
-	var urls []string
-	for _, part := range strings.Split(value, ",") {
-		part = strings.TrimSpace(part)
-		if strings.HasPrefix(part, "<") && strings.HasSuffix(part, ">") {
-			urls = append(urls, part[1:len(part)-1])
-		}
-	}
-	return urls
-}