pkg/report: support alternative bug titles

Update #1575

4 files changed, 49 insertions, 13 deletions

diff --git a/pkg/report/freebsd.go b/pkg/report/freebsd.go
index e5a1fc92e..3bf0b8eb4 100644
--- a/pkg/report/freebsd.go
+++ b/pkg/report/freebsd.go
@@ -56,8 +56,9 @@ func (ctx *freebsd) Parse(output []byte) *Report {
 	if oops == nil {
 		return nil
 	}
-	title, corrupted, _ := extractDescription(output[rep.StartPos:], oops, freebsdStackParams)
+	title, corrupted, altTitles, _ := extractDescription(output[rep.StartPos:], oops, freebsdStackParams)
 	rep.Title = title
+	rep.AltTitles = altTitles
 	rep.Corrupted = corrupted != ""
 	rep.CorruptedReason = corrupted
 	return rep
diff --git a/pkg/report/linux.go b/pkg/report/linux.go
index 8aed7d4f4..dd781ae89 100644
--- a/pkg/report/linux.go
+++ b/pkg/report/linux.go
@@ -145,17 +145,18 @@ func (ctx *linux) Parse(output []byte) *Report {
 		}
 		endPos, reportEnd, report, prefix := ctx.findReport(output, oops, startPos, context, questionable)
 		rep.EndPos = endPos
-		title, corrupted, format := extractDescription(report[:reportEnd], oops, linuxStackParams)
+		title, corrupted, altTitles, format := extractDescription(report[:reportEnd], oops, linuxStackParams)
 		if title == "" {
 			prefix = nil
 			report = output[rep.StartPos:rep.EndPos]
-			title, corrupted, format = extractDescription(report, oops, linuxStackParams)
+			title, corrupted, altTitles, format = extractDescription(report, oops, linuxStackParams)
 			if title == "" {
 				panic(fmt.Sprintf("non matching oops for %q context=%q in:\n%s\n",
 					oops.header, context, report))
 			}
 		}
 		rep.Title = title
+		rep.AltTitles = altTitles
 		rep.Corrupted = corrupted != ""
 		rep.CorruptedReason = corrupted
 		for _, line := range prefix {
diff --git a/pkg/report/report.go b/pkg/report/report.go
index 397684569..72d52bbf0 100644
--- a/pkg/report/report.go
+++ b/pkg/report/report.go
@@ -32,6 +32,9 @@ type Reporter interface {
 type Report struct {
 	// Title contains a representative description of the first oops.
 	Title string
+	// Alternative titles, used for better deduplication.
+	// If two crashes have a non-empty intersection of Title/AltTitles, they are considered the same bug.
+	AltTitles []string
 	// Bug type (e.g. hang, memory leak, etc).
 	Type Type
 	// The indicative function name.
@@ -170,6 +173,9 @@ func (wrap *reporterWrapper) Parse(output []byte) *Report {
 		return nil
 	}
 	rep.Title = sanitizeTitle(replaceTable(dynamicTitleReplacement, rep.Title))
+	for i, title := range rep.AltTitles {
+		rep.AltTitles[i] = sanitizeTitle(replaceTable(dynamicTitleReplacement, title))
+	}
 	rep.Suppressed = matchesAny(rep.Output, wrap.suppressions)
 	if bytes.Contains(rep.Output, gceConsoleHangup) {
 		rep.Corrupted = true
@@ -329,6 +335,9 @@ type oopsFormat struct {
 	// Strings captured by title (or by report if present) are passed as input.
 	// If stack is not nil, extracted function name is passed as an additional last argument.
 	fmt string
+	// Alternative titles used for better crash deduplication.
+	// Format is the same as for fmt.
+	alt []string
 	// If not nil, a function name is extracted from the report and passed to fmt.
 	// If not nil but frame extraction fails, the report is considered corrupted.
 	stack        *stackFmt
@@ -398,7 +407,8 @@ func matchOops(line []byte, oops *oops, ignores []*regexp.Regexp) bool {
 	return true
 }
 
-func extractDescription(output []byte, oops *oops, params *stackParams) (desc, corrupted string, format oopsFormat) {
+func extractDescription(output []byte, oops *oops, params *stackParams) (
+	desc, corrupted string, altTitles []string, format oopsFormat) {
 	startPos := len(output)
 	matchedTitle := false
 	for _, f := range oops.formats {
@@ -438,6 +448,9 @@ func extractDescription(output []byte, oops *oops, params *stackParams) (desc, c
 			args = append(args, frame)
 		}
 		desc = fmt.Sprintf(f.fmt, args...)
+		for _, alt := range f.alt {
+			altTitles = append(altTitles, fmt.Sprintf(alt, args...))
+		}
 		format = f
 	}
 	if desc == "" {
@@ -591,8 +604,9 @@ func simpleLineParser(output []byte, oopses []*oops, params *stackParams, ignore
 	if oops == nil {
 		return nil
 	}
-	title, corrupted, _ := extractDescription(output[rep.StartPos:], oops, params)
+	title, corrupted, altTitles, _ := extractDescription(output[rep.StartPos:], oops, params)
 	rep.Title = title
+	rep.AltTitles = altTitles
 	rep.Report = output[rep.StartPos:]
 	rep.Corrupted = corrupted != ""
 	rep.CorruptedReason = corrupted
diff --git a/pkg/report/report_test.go b/pkg/report/report_test.go
index 05883a21f..355edc764 100644
--- a/pkg/report/report_test.go
+++ b/pkg/report/report_test.go
@@ -10,7 +10,9 @@ import (
 	"fmt"
 	"io/ioutil"
 	"path/filepath"
+	"reflect"
 	"regexp"
+	"sort"
 	"strings"
 	"testing"
 
@@ -29,6 +31,7 @@ type ParseTest struct {
 	FileName   string
 	Log        []byte
 	Title      string
+	AltTitles  []string
 	Type       Type
 	Frame      string
 	StartLine  string
@@ -92,6 +95,7 @@ func testParseFile(t *testing.T, reporter Reporter, fn string) {
 func parseHeaderLine(t *testing.T, test *ParseTest, ln string) {
 	const (
 		titlePrefix      = "TITLE: "
+		altTitlePrefix   = "ALT: "
 		typePrefix       = "TYPE: "
 		framePrefix      = "FRAME: "
 		startPrefix      = "START: "
@@ -103,6 +107,8 @@ func parseHeaderLine(t *testing.T, test *ParseTest, ln string) {
 	case strings.HasPrefix(ln, "#"):
 	case strings.HasPrefix(ln, titlePrefix):
 		test.Title = ln[len(titlePrefix):]
+	case strings.HasPrefix(ln, altTitlePrefix):
+		test.AltTitles = append(test.AltTitles, ln[len(altTitlePrefix):])
 	case strings.HasPrefix(ln, typePrefix):
 		switch v := ln[len(typePrefix):]; v {
 		case Hang.String():
@@ -159,23 +165,34 @@ func testParseImpl(t *testing.T, reporter Reporter, test *ParseTest) {
 		t.Fatalf("found crash, but title is empty")
 	}
 	title, corrupted, corruptedReason, suppressed, typ, frame := "", false, "", false, Unknown, ""
+	var altTitles []string
 	if rep != nil {
 		title = rep.Title
+		altTitles = rep.AltTitles
 		corrupted = rep.Corrupted
 		corruptedReason = rep.CorruptedReason
 		suppressed = rep.Suppressed
 		typ = rep.Type
 		frame = rep.Frame
 	}
-	if title != test.Title || corrupted != test.Corrupted || suppressed != test.Suppressed ||
-		typ != test.Type || test.Frame != "" && frame != test.Frame {
+	sort.Strings(altTitles)
+	sort.Strings(test.AltTitles)
+	if title != test.Title || !reflect.DeepEqual(altTitles, test.AltTitles) || corrupted != test.Corrupted ||
+		suppressed != test.Suppressed || typ != test.Type || test.Frame != "" && frame != test.Frame {
 		if *flagUpdate && test.StartLine+test.EndLine == "" {
-			updateReportTest(t, test, title, corrupted, suppressed, typ, frame)
+			updateReportTest(t, test, title, altTitles, corrupted, suppressed, typ, frame)
 		}
-		t.Fatalf("want:\nTITLE: %s\nTYPE: %v\nFRAME: %v\nCORRUPTED: %v\nSUPPRESSED: %v\n"+
-			"got:\nTITLE: %s\nTYPE: %v\nFRAME: %v\nCORRUPTED: %v (%v)\nSUPPRESSED: %v\n",
-			test.Title, test.Type, test.Frame, test.Corrupted, test.Suppressed,
-			title, typ, frame, corrupted, corruptedReason, suppressed)
+		gotAltTitles, wantAltTitles := "", ""
+		for _, t := range altTitles {
+			gotAltTitles += "ALT: " + t + "\n"
+		}
+		for _, t := range test.AltTitles {
+			wantAltTitles += "ALT: " + t + "\n"
+		}
+		t.Fatalf("want:\nTITLE: %s\n%sTYPE: %v\nFRAME: %v\nCORRUPTED: %v\nSUPPRESSED: %v\n"+
+			"got:\nTITLE: %s\n%sTYPE: %v\nFRAME: %v\nCORRUPTED: %v (%v)\nSUPPRESSED: %v\n",
+			test.Title, wantAltTitles, test.Type, test.Frame, test.Corrupted, test.Suppressed,
+			title, gotAltTitles, typ, frame, corrupted, corruptedReason, suppressed)
 	}
 	if title != "" && len(rep.Report) == 0 {
 		t.Fatalf("found crash message but report is empty")
@@ -228,10 +245,13 @@ func checkReport(t *testing.T, reporter Reporter, rep *Report, test *ParseTest)
 	}
 }
 
-func updateReportTest(t *testing.T, test *ParseTest, title string, corrupted, suppressed bool,
+func updateReportTest(t *testing.T, test *ParseTest, title string, altTitles []string, corrupted, suppressed bool,
 	typ Type, frame string) {
 	buf := new(bytes.Buffer)
 	fmt.Fprintf(buf, "TITLE: %v\n", title)
+	for _, t := range altTitles {
+		fmt.Fprintf(buf, "ALT: %v\n", t)
+	}
 	if typ != Unknown {
 		fmt.Fprintf(buf, "TYPE: %v\n", typ)
 	}