feat: ignore seen links

internal/linkvalidator/linkvalidator.go

@@ -22,6 +22,7 @@ package linkvalidator
 import (
 	"net/url"
 	"strings"
+	"sync"
 
 	"git.weirdcat.su/weirdcat/auto-attendance/internal/config"
 	"git.weirdcat.su/weirdcat/auto-attendance/internal/logger"
@@ -29,11 +30,15 @@ import (
 
 type LinkValidator interface {
 	ValidateLink(string) (token string, ok bool)
+	ResetSeenLinks()
+	GetSeenCount() int
 }
 
 type linkValidatorImpl struct {
-	config *config.Config
+	config  *config.Config
-	log    *logger.Logger
+	log     *logger.Logger
+	seenMu  sync.RWMutex
+	seenLinks    map[string]bool
 }
 
 // ValidateLink implements LinkValidator.
@@ -76,13 +81,46 @@ func (v *linkValidatorImpl) ValidateLink(rawURL string) (token string, ok bool)
 		return "", false
 	}
 
-	v.log.Debug("URL validation successful", "url", rawURL)
+	// Check if we've already seen this token
+	v.seenMu.RLock()
+	alreadySeen := v.seenLinks[token]
+	v.seenMu.RUnlock()
+
+	if alreadySeen {
+		v.log.Debug("URL token already processed, skipping", "token", token, "url", rawURL)
+		return token, false
+	}
+
+	// Mark this token as seen
+	v.seenMu.Lock()
+	v.seenLinks[token] = true
+	v.seenMu.Unlock()
+
+	v.log.Debug("URL validation successful", "url", rawURL, "token", token)
 	return token, true
 }
 
+// ResetSeenLinks clears the cache of seen links
+func (v *linkValidatorImpl) ResetSeenLinks() {
+	v.seenMu.Lock()
+	defer v.seenMu.Unlock()
+	
+	oldCount := len(v.seenLinks)
+	v.seenLinks = make(map[string]bool)
+	v.log.Debug("Reset seen links cache", "previous_count", oldCount)
+}
+
+// GetSeenCount returns the number of unique links seen
+func (v *linkValidatorImpl) GetSeenCount() int {
+	v.seenMu.RLock()
+	defer v.seenMu.RUnlock()
+	return len(v.seenLinks)
+}
+
 func NewLinkValidator(config *config.Config, log *logger.Logger) LinkValidator {
 	return &linkValidatorImpl{
 		config: config,
 		log:    log,
+		seenLinks:   make(map[string]bool),
 	}
 }