Adds exclusion for words that contain green (#5)

Signed-off-by: William Rizzo <[email protected]>
cncf-tags · Nov 27, 2023 · a8fc999 · a8fc999
1 parent 536e4c4
commit a8fc999
Show file tree

Hide file tree

Showing 3 changed files with 55 additions and 9 deletions.
diff --git a/greenscraper/cmd/main.go b/greenscraper/cmd/main.go
@@ -29,16 +29,16 @@ func ReadLinesFromFile(filename string) ([]string, error) {
 	return lines, scanner.Err()
 }
 
-func containsAnyOfKeywords(title string, keywords []string) bool {
-	for _, keyword := range keywords {
-		if contains := strings.Contains(title, keyword); contains {
+func containsAnyOfExclusions(title string, exclusions []string) bool {
+	for _, exclusion := range exclusions {
+		if strings.Contains(title, exclusion) {
 			return true
 		}
 	}
 	return false
 }
 
-func ProcessURL(url string, keywordRegexes []*regexp.Regexp, titleRegex *regexp.Regexp, wg *sync.WaitGroup) {
+func ProcessURL(url string, keywordRegexes []*regexp.Regexp, exclusions []string, titleRegex *regexp.Regexp, wg *sync.WaitGroup) {
 	defer wg.Done()
 
 	resp, err := http.Get(url)
@@ -66,7 +66,7 @@ func ProcessURL(url string, keywordRegexes []*regexp.Regexp, titleRegex *regexp.
 			if len(titleMatch) > 1 {
 				title := titleMatch[1]
 
-				if len(title) >= 40 && !encounteredTitles[title] && !containsBadgePickUp(title) {
+				if len(title) >= 40 && !encounteredTitles[title] && !containsAnyOfExclusions(title, exclusions) && !containsBadgePickUp(title) {
 					encounteredTitles[title] = true
 					talks = append(talks, "- "+title)
 				}

diff --git a/greenscraper/exclusions.txt b/greenscraper/exclusions.txt
@@ -0,0 +1,41 @@
+Greene
+Greenley
+Greenwald
+Greenberg
+Blue/Green
+blue/green
+Green Light
+Blue-green
+Blue green
+COVIDGreen
+Greenwood
+Greentree
+Gvergreen
+Greenfield
+Greenleaf
+Greenhill
+Greenway
+Greenland
+Greenlee
+Greenville
+Greenwood
+Greenberg
+Greenfield
+Greenbaum
+Greenberg
+Greenwald
+Greenspan
+Greene
+Greenberg
+Greenman
+Greenland
+Greenleaf
+Greenlee
+Greenspan
+Greenstein
+Greenwald
+Greenwell
+Greenblatt
+Greenberg
+Greenfield
+Greenbaum
diff --git a/greenscraper/main.go b/greenscraper/main.go
@@ -17,6 +17,11 @@ func main() {
 	if err != nil {
 		panic(fmt.Sprintf("Failed to read URLs from file: %v", err))
 	}
+	// Read exclusions from exclusions.txt
+	exclusions, err := cmd.ReadLinesFromFile("exclusions.txt")
+	if err != nil {
+		panic(fmt.Sprintf("Failed to read exclusions from file: %v", err))
+	}
 
 	titleRegex := regexp.MustCompile(".*'>(.*?)<span class=\"vs\">.*")
 	keywordRegexes := make([]*regexp.Regexp, len(keywords))
@@ -25,15 +30,15 @@ func main() {
 	}
 
 	const concurrentLimit = 5
-	sem := make(chan struct{}, concurrentLimit) // semaphore pattern for limiting concurrency
+	sem := make(chan struct{}, concurrentLimit)
 	var wg sync.WaitGroup
 
 	for _, url := range urls {
 		wg.Add(1)
-		sem <- struct{}{} // acquire a token
+		sem <- struct{}{}
 		go func(u string) {
-			cmd.ProcessURL(u, keywordRegexes, titleRegex, &wg)
-			<-sem // release a token
+			cmd.ProcessURL(u, keywordRegexes, exclusions, titleRegex, &wg)
+			<-sem
 		}(url)
 	}