fix: JUnit processing for large files

cmd/testworkflow-toolkit/artifacts/junit_post_processor.go

@@ -1,8 +1,6 @@
 package artifacts
 
 import (
-	"bufio"
-	"bytes"
 	"context"
 	"fmt"
 	"io"
@@ -39,6 +37,7 @@ func (p *JUnitPostProcessor) Start() error {
 
 // Add checks if the file is a JUnit report and sends it to the cloud.
 func (p *JUnitPostProcessor) Add(path string) error {
+	fmt.Printf("Checking file: %s\n", ui.LightCyan(path))
 	uploadPath := path
 	if p.pathPrefix != "" {
 		uploadPath = filepath.Join(p.pathPrefix, uploadPath)
@@ -63,10 +62,7 @@ func (p *JUnitPostProcessor) Add(path string) error {
 	if err != nil {
 		return errors.Wrapf(err, "failed to read %s", path)
 	}
-	ok, err := isJUnitReport(xmlData)
-	if err != nil {
-		return errors.Wrapf(err, "failed to check if %s is a JUnit report", path)
-	}
+	ok := isJUnitReport(xmlData)
 	if !ok {
 		return nil
 	}
@@ -85,6 +81,7 @@ func (p *JUnitPostProcessor) sendJUnitReport(path string, report []byte) error {
 	}
 	ctx, cancel := context.WithTimeout(context.Background(), 15*time.Second)
 	defer cancel()
+	fmt.Printf("Sending JUnit report to the cloud: %s\n", ui.LightCyan(path))
 	_, err := p.client.Execute(ctx, testworkflow.CmdTestWorkflowExecutionAddReport, &testworkflow.ExecutionsAddReportRequest{
 		ID:           env.ExecutionId(),
 		WorkflowName: env.WorkflowName(),
@@ -104,27 +101,28 @@ func isXMLFile(stat fs.FileInfo) bool {
 	return strings.HasSuffix(stat.Name(), ".xml")
 }
 
-// isJUnitReport checks if the file starts with a JUnit XML tag
-func isJUnitReport(xmlData []byte) (bool, error) {
-	scanner := bufio.NewScanner(bytes.NewReader(xmlData))
-	// Read only the first few lines of the file
-	for scanner.Scan() {
-		line := scanner.Text()
-		line = strings.TrimSpace(line) // Remove leading and trailing whitespace
-
-		// Skip comments and declarations
-		if strings.HasPrefix(line, "<!--") || strings.HasPrefix(line, "<?xml") {
-			continue
-		}
-		if strings.Contains(line, "<testsuite") || strings.Contains(line, "<testsuites") {
-			return true, nil
-		}
-		if strings.Contains(line, "<") { // Stop if any non-JUnit tag is found
-			break
+// isJUnitReport checks if the XML data is a JUnit report.
+func isJUnitReport(xmlData []byte) bool {
+	const BYTE_SIZE_8KB = 8 * 1024
+
+	tags := []string{
+		"<testsuite",
+		"<testsuites",
+	}
+
+	if len(xmlData) > BYTE_SIZE_8KB {
+		xmlData = xmlData[:BYTE_SIZE_8KB]
+	}
+
+	content := string(xmlData)
+
+	for _, tag := range tags {
+		if strings.Contains(content, tag) {
+			return true
 		}
 	}
 
-	return false, scanner.Err()
+	return false
 }
 
 func (p *JUnitPostProcessor) End() error {

cmd/testworkflow-toolkit/artifacts/junit_post_processor_test.go

@@ -146,6 +146,21 @@ func TestIsJUnitReport(t *testing.T) {
 			file: filesystem.NewMockFile("invalid.xml", []byte(testdata.InvalidJUnit)),
 			want: false,
 		},
+		{
+			name: "one-line junit",
+			file: filesystem.NewMockFile("oneline.xml", []byte(testdata.OneLineJUnit)),
+			want: true,
+		},
+		{
+			name: "testsuites only junit",
+			file: filesystem.NewMockFile("testsuites.xml", []byte(testdata.TestsuitesOnlyJUnit)),
+			want: true,
+		},
+		{
+			name: "testsuite only junit",
+			file: filesystem.NewMockFile("testsuite.xml", []byte(testdata.TestsuiteOnlyJUnit)),
+			want: true,
+		},
 	}
 
 	for _, tc := range tests {
@@ -154,10 +169,7 @@ func TestIsJUnitReport(t *testing.T) {
 			if err != nil {
 				t.Fatalf("unexpected error: %v", err)
 			}
-			ok, err := isJUnitReport(data)
-			if err != nil {
-				t.Fatalf("unexpected error: %v", err)
-			}
+			ok := isJUnitReport(data)
 			assert.Equal(t, tc.want, ok)
 		})
 	}

cmd/testworkflow-toolkit/artifacts/walker.go

@@ -197,6 +197,8 @@ func (w *walker) walk(fsys fs.FS, path string, walker WalkerFn) error {
 		sanitizedPath = "."
 	}
 
+	fmt.Printf("Walking: %s\n", sanitizedPath)
+
 	return fs.WalkDir(fsys, sanitizedPath, func(filePath string, d fs.DirEntry, err error) error {
 		resolvedPath := "/" + filepath.ToSlash(filePath)
 		if !w.matches(resolvedPath) {
@@ -219,6 +221,7 @@ func (w *walker) walk(fsys fs.FS, path string, walker WalkerFn) error {
 
 		// Pass the data to final walker
 		relativeFilePath := strings.TrimLeft(resolvedPath[len(w.root):], "/")
+		fmt.Printf("Relative path: %s\n", relativeFilePath)
 		file, err := fsys.Open(filePath)
 		return walker(relativeFilePath, file, stat, err)
 	})

cmd/testworkflow-toolkit/commands/artifacts.go

@@ -180,6 +180,8 @@ func run(handler artifacts.Handler, walker artifacts.Walker, dirFS fs.FS) {
 
 	started := time.Now()
 	err = walker.Walk(dirFS, func(path string, file fs.File, _ fs.FileInfo, err error) error {
+		fmt.Printf("Processing: %s\n", ui.LightCyan(path))
+
 		if err != nil {
 			fmt.Printf("Warning: '%s' has been ignored, as there was a problem reading it: %s\n", path, err.Error())
 			return nil

cmd/testworkflow-toolkit/common/testdata/junit.go

@@ -176,3 +176,16 @@ var InvalidJUnit = `<?xml version="1.0" encoding="UTF-8"?>
 <foo>
 	<bar>
 </foo>`
+
+var OneLineJUnit = `<?xml version="1.0" encoding="UTF-8"?><testsuites><testsuite name="TestSuite" tests="2" errors="0" failures="1" skipped="0"><testcase name="Test1" classname="TestClass"><failure message="Test failed">Failure details</failure></testcase><testcase name="Test2" classname="TestClass"/></testsuite></testsuites>`
+
+var TestsuitesOnlyJUnit = `<testsuites id="" name="" tests="2" failures="0" skipped="0" errors="0" time="14.511833">
+    <testsuite name="smoke.spec.js" timestamp="2024-10-01T12:48:47.332Z" hostname="chromium" tests="1" failures="0" skipped="0" time="6.259" errors="0">
+        <testcase name="Smoke 1 - has title" classname="smoke.spec.js" time="6.259"></testcase>
+    </testsuite>
+    <testsuite name="smoke2.spec.js" timestamp="2024-10-01T12:48:47.332Z" hostname="chromium" tests="1" failures="0" skipped="0" time="6.657" errors="0">
+        <testcase name="Smoke 2 - has title" classname="smoke2.spec.js" time="6.657"></testcase>
+    </testsuite>
+</testsuites>`
+
+var TestsuiteOnlyJUnit = `<testsuite name="smoke.spec.js" timestamp="2024-10-01T12:48:47.332Z" hostname="chromium" tests="1" failures="0" skipped="0" time="6.259" errors="0"><testcase name="Smoke 1 - has title" classname="smoke.spec.js" time="6.259"></testcase></testsuite>`