pkg/email/lore: extract patch series

Refactor the code to make it more reusable.
Add a method to extract specifically the list of new patch series.

1 files changed, 163 insertions, 6 deletions

diff --git a/pkg/email/lore/parse.go b/pkg/email/lore/parse.go
index d0e4d4fe2..98152b408 100644
--- a/pkg/email/lore/parse.go
+++ b/pkg/email/lore/parse.go
@@ -4,14 +4,17 @@
 package lore
 
 import (
+	"fmt"
 	"regexp"
 	"sort"
+	"strconv"
 	"strings"
 
 	"github.com/google/syzkaller/dashboard/dashapi"
 	"github.com/google/syzkaller/pkg/email"
 )
 
+// Thread is a generic representation of a single discussion in the mailing list.
 type Thread struct {
 	Subject   string
 	MessageID string
@@ -20,9 +23,28 @@ type Thread struct {
 	Messages  []*email.Email
 }
 
+// Series represents a single patch series sent over email.
+type Series struct {
+	Subject   string
+	MessageID string
+	Version   int
+	Corrupted string // If non-empty, contains a reason why the series better be ignored.
+	Patches   []Patch
+}
+
+type Patch struct {
+	Seq int
+	*email.Email
+}
+
 // Threads extracts individual threads from a list of emails.
 func Threads(emails []*email.Email) []*Thread {
+	return listThreads(emails, 0)
+}
+
+func listThreads(emails []*email.Email, maxDepth int) []*Thread {
 	ctx := &parseCtx{
+		maxDepth: maxDepth,
 		messages: map[string]*email.Email{},
 		next:     map[*email.Email][]*email.Email{},
 	}
@@ -33,6 +55,68 @@ func Threads(emails []*email.Email) []*Thread {
 	return ctx.threads
 }
 
+// PatchSeries is similar to Threads, but returns only the patch series submitted to the mailing lists.
+func PatchSeries(emails []*email.Email) []*Series {
+	var ret []*Series
+	// Normally, all following series patches are sent in response to the first email sent.
+	// So there's no sense to look at deeper replies.
+	for _, thread := range listThreads(emails, 1) {
+		if thread.Type != dashapi.DiscussionPatch {
+			continue
+		}
+		patch, ok := parsePatchSubject(thread.Subject)
+		if !ok {
+			// It must never be happening.
+			panic("DiscussionPatch is set, but we fail to parse the thread subject")
+		}
+		total := patch.Total.ValueOr(1)
+		series := &Series{
+			Subject:   patch.Title,
+			MessageID: thread.MessageID,
+			Version:   patch.Version.ValueOr(1),
+		}
+		ret = append(ret, series)
+		if patch.Seq.IsSet() && patch.Seq.Value() > 1 {
+			series.Corrupted = "the first patch has seq>1"
+			continue
+		}
+		hasSeq := map[int]bool{}
+		for _, email := range thread.Messages {
+			patch, ok := parsePatchSubject(email.Subject)
+			if !ok {
+				continue
+			}
+			seq := patch.Seq.ValueOr(1)
+			if seq == 0 {
+				// The cover email is not of interest.
+				continue
+			}
+			if hasSeq[seq] {
+				// It's weird if that really happens, but let's skip for now.
+				continue
+			}
+			hasSeq[seq] = true
+			series.Patches = append(series.Patches, Patch{
+				Seq:   seq,
+				Email: email,
+			})
+		}
+		if len(hasSeq) != total {
+			series.Corrupted = fmt.Sprintf("the subject mentions %d patches, %d are found",
+				total, len(hasSeq))
+			continue
+		}
+		if len(series.Patches) == 0 {
+			series.Corrupted = "0 patches"
+			continue
+		}
+		sort.Slice(series.Patches, func(i, j int) bool {
+			return series.Patches[i].Seq < series.Patches[j].Seq
+		})
+	}
+	return ret
+}
+
 // DiscussionType extracts the specific discussion type from an email.
 func DiscussionType(msg *email.Email) dashapi.DiscussionType {
 	discType := dashapi.DiscussionMention
@@ -40,7 +124,7 @@ func DiscussionType(msg *email.Email) dashapi.DiscussionType {
 		discType = dashapi.DiscussionReport
 	}
 	// This is very crude, but should work for now.
-	if patchSubjectRe.MatchString(strings.ToLower(msg.Subject)) {
+	if _, ok := parsePatchSubject(msg.Subject); ok {
 		discType = dashapi.DiscussionPatch
 	} else if strings.Contains(msg.Subject, "Monthly") {
 		discType = dashapi.DiscussionReminder
@@ -48,9 +132,51 @@ func DiscussionType(msg *email.Email) dashapi.DiscussionType {
 	return discType
 }
 
-var patchSubjectRe = regexp.MustCompile(`\[(?:(?:rfc|resend)\s+)*patch`)
+type PatchSubject struct {
+	Title   string
+	Tags    []string // Sometimes there's e.g. "net" or "next-next" in the subject.
+	Version Optional[int]
+	Seq     Optional[int] // The "Seq/Total" part.
+	Total   Optional[int]
+}
+
+// nolint: lll
+var patchSubjectRe = regexp.MustCompile(`(?mi)^\[(?:([\w\s-]+)\s)?PATCH(?:\s([\w\s-]+))??(?:\s0*(\d+)\/(\d+))?\]\s*(.+)`)
+
+func parsePatchSubject(subject string) (PatchSubject, bool) {
+	var ret PatchSubject
+	groups := patchSubjectRe.FindStringSubmatch(subject)
+	if len(groups) == 0 {
+		return ret, false
+	}
+	tags := strings.Fields(groups[1])
+	for _, tag := range append(tags, strings.Fields(groups[2])...) {
+		if strings.HasPrefix(tag, "v") {
+			val, err := strconv.Atoi(strings.TrimPrefix(tag, "v"))
+			if err == nil {
+				ret.Version.Set(val)
+				continue
+			}
+		}
+		ret.Tags = append(ret.Tags, tag)
+	}
+	sort.Strings(ret.Tags)
+	if groups[3] != "" {
+		if val, err := strconv.Atoi(groups[3]); err == nil {
+			ret.Seq.Set(val)
+		}
+	}
+	if groups[4] != "" {
+		if val, err := strconv.Atoi(groups[4]); err == nil {
+			ret.Total.Set(val)
+		}
+	}
+	ret.Title = groups[5]
+	return ret, true
+}
 
 type parseCtx struct {
+	maxDepth int
 	threads  []*Thread
 	messages map[string]*email.Email
 	next     map[*email.Email][]*email.Email
@@ -73,7 +199,7 @@ func (c *parseCtx) process() {
 	}
 	// Iterate starting from these tree nodes.
 	for _, node := range nodes {
-		c.visit(node, nil)
+		c.visit(node, nil, 0)
 	}
 	// Collect BugIDs.
 	for _, thread := range c.threads {
@@ -92,7 +218,7 @@ func (c *parseCtx) process() {
 	}
 }
 
-func (c *parseCtx) visit(msg *email.Email, thread *Thread) {
+func (c *parseCtx) visit(msg *email.Email, thread *Thread, depth int) {
 	var oldInfo *email.OldThreadInfo
 	if thread != nil {
 		oldInfo = &email.OldThreadInfo{
@@ -114,7 +240,38 @@ func (c *parseCtx) visit(msg *email.Email, thread *Thread) {
 		}
 		c.threads = append(c.threads, thread)
 	}
-	for _, nextMsg := range c.next[msg] {
-		c.visit(nextMsg, thread)
+	if c.maxDepth == 0 || depth < c.maxDepth {
+		for _, nextMsg := range c.next[msg] {
+			c.visit(nextMsg, thread, depth+1)
+		}
 	}
 }
+
+type Optional[T any] struct {
+	val T
+	set bool
+}
+
+func value[T any](val T) Optional[T] {
+	return Optional[T]{val: val, set: true}
+}
+
+func (o Optional[T]) IsSet() bool {
+	return o.set
+}
+
+func (o Optional[T]) Value() T {
+	return o.val
+}
+
+func (o Optional[T]) ValueOr(def T) T {
+	if o.set {
+		return o.val
+	}
+	return def
+}
+
+func (o *Optional[T]) Set(val T) {
+	o.val = val
+	o.set = true
+}