rumble: fix legacy csv query and csv output to match what the chart j…

…s expects (#1831)

Drop unused fields `scanner_version` and `scanner_db_version`, append
the tag to the end of the `image` name as expected by the javascript
generating the chart, and exclude images with `latest-dev` tags since
those are never used in image comparisons (cuts `data.csv` down from
~8MB to 4MB 📉 )

Signed-off-by: Colin Douglas <colin@chainguard.dev>

tools/rumble/cmd/legacy_csv.go

@@ -79,10 +79,8 @@ func (c *rumbleBase) generateCsv() error {
 		if !strings.Contains(image, "cgr.dev") || strings.Contains(image, "cgr.dev/chainguard/") {
 			csvRow := []string{
 				strconv.Itoa(i + 1),
-				r.(*cgbigquery.LegacyScan).Image,
+				fmt.Sprintf("%s:%s", r.(*cgbigquery.LegacyScan).Image, r.(*cgbigquery.LegacyScan).Tags),
 				r.(*cgbigquery.LegacyScan).Scanner,
-				r.(*cgbigquery.LegacyScan).Scanner_version,
-				r.(*cgbigquery.LegacyScan).Scanner_db_version,
 				r.(*cgbigquery.LegacyScan).Time,
 				strconv.FormatInt(r.(*cgbigquery.LegacyScan).Low_cve_cnt, 10),
 				strconv.FormatInt(r.(*cgbigquery.LegacyScan).Med_cve_cnt, 10),

tools/rumble/pkg/bigquery/bigquery.go

@@ -26,19 +26,18 @@ type BqClient struct {
 }
 
 type LegacyScan struct {
-	Row                int64
-	Image              string
-	Scanner            string
-	Scanner_version    string
-	Scanner_db_version string
-	Time               string
-	Low_cve_cnt        int64
-	Med_cve_cnt        int64
-	High_cve_cnt       int64
-	Crit_cve_cnt       int64
-	Unknown_cve_cnt    int64
-	Tot_cve_cnt        int64
-	Digest             string
+	Row             int64
+	Image           string
+	Tags            string
+	Scanner         string
+	Time            string
+	Low_cve_cnt     int64
+	Med_cve_cnt     int64
+	High_cve_cnt    int64
+	Crit_cve_cnt    int64
+	Unknown_cve_cnt int64
+	Tot_cve_cnt     int64
+	Digest          string
 }
 
 type ImageScan struct {

tools/rumble/pkg/bigquery/queries.go

@@ -9,15 +9,15 @@ const (
 	vulnsTable   = "`cloudevents_grype_scan_results.rumble_vulns`"
 	summaryTable = "`cloudevents_grype_scan_results.rumble_summary`"
 
-	LegacyCsvHeader    = `f0_,image,scanner,scanner_version,scanner_db_version,time,low_cve_cnt,med_cve_cnt,high_cve_cnt,crit_cve_cnt,unknown_cve_cnt,tot_cve_cnt,digest`
+	LegacyCsvHeader    = `f0_,image,scanner,time,low_cve_cnt,med_cve_cnt,high_cve_cnt,crit_cve_cnt,unknown_cve_cnt,tot_cve_cnt,digest`
 	ImageScanCsvHeader = `image,package,vulnerability,version,type,s`
 
 	AllVulnsQuery = `
 SELECT DISTINCT vulnerability
 FROM ` + vulnsTable
 
 	AffectedImagesQuery = `
-SELECT summ.image, summ.time as time,
+SELECT scan.image, scan.time as time,
 FROM ` + vulnsTable + ` AS vulns
 INNER JOIN ` + summaryTable + ` AS scan
 ON scan.id = vulns.scan_id
@@ -30,9 +30,8 @@ ORDER BY scan.image, scan.time
 SELECT
 	ROW_NUMBER() OVER (ORDER BY time),
 	image,
+    tags,
 	scanner,
-	scanner_version,
-	scanner_db_version,
 	FORMAT_DATETIME("%Y-%m-%d %H:%M:%S", DATE(time)) as time,
 	low_cve_count as low_cve_cnt,
 	med_cve_count as med_cve_cnt,
@@ -41,7 +40,8 @@ SELECT
 	unknown_cve_count as unknown_cve_cnt,
 	low_cve_count + med_cve_count + high_cve_count + crit_cve_count + unknown_cve_count AS tot_cve_cnt,
 	digest
-FROM ` + summaryTable
+FROM ` + summaryTable + `
+WHERE tags NOT LIKE '%latest-dev%'`
 
 	ImageComparisonCsvQuery = `
 WITH ruuuumble AS (