wip pipermail threading

1 files changed, 77 insertions, 56 deletions

diff --git a/cmd/generate/forge_pipermail.go b/cmd/generate/forge_pipermail.go
index 2c5cf01..e015bb5 100644
--- a/cmd/generate/forge_pipermail.go
+++ b/cmd/generate/forge_pipermail.go
@@ -2,14 +2,17 @@ package main
 
 import (
 	"fmt"
+	"net/url"
 	"regexp"
+	"strconv"
 	"strings"
 	"time"
 )
 
 var (
-	rePiperMailDate = regexp.MustCompile(`^\s*<I>([^<]+)</I>\s*$`)
-	reGitHubCommit  = regexp.MustCompile(`^https://github\.com/([^/?#]+)/([^/?#]+)/commit/([0-9a-f]+)(?:\?[^#]*)?(?:#.*)?$`)
+	rePiperMailMessage = regexp.MustCompile(`^(https?://.*/pipermail/.*/)([0-4]{4}-(?:January|February|March|April|May|June|July|August|September|October|November|December))/([0-9]+)\.html$`)
+	rePiperMailDate    = regexp.MustCompile(`^\s*<I>([^<]+)</I>\s*$`)
+	rePiperMailReply   = regexp.MustCompile(`^\s*<LINK REL="made" HREF="(.*)">\s$`)
 )
 
 type PiperMail struct{}
@@ -17,35 +20,12 @@ type PiperMail struct{}
 var _ Forge = PiperMail{}
 
 func (PiperMail) FetchStatus(urls []string) (string, error) {
-	var gitURL string
-	var gitCommits []string
-	for _, u := range urls {
-		if m := reGitHubCommit.FindStringSubmatch(u); m != nil {
-			user := m[1]
-			repo := m[2]
-			hash := m[3]
-
-			gitURL = "https://github.com/" + user + "/" + repo
-			gitCommits = append(gitCommits, hash)
-		}
-	}
-	if len(gitCommits) == 0 {
-		return "", nil
-	}
-	ret := statusMerged
-	tag, err := getGitTagThatContainsAll(gitURL, gitCommits...)
-	if err != nil {
-		return "", err
-	}
-	if tag != "" {
-		ret = fmt.Sprintf(statusReleasedFmt, tag)
-	}
-	return ret, nil
+	return "", nil
 }
 
 func (PiperMail) FetchSubmittedAt(urls []string) (time.Time, error) {
 	for _, u := range urls {
-		if !strings.Contains(u, "/pipermail/") {
+		if !rePiperMailMessage.MatchString(u) {
 			continue
 		}
 		htmlStr, err := httpGet(u, nil)
@@ -61,38 +41,79 @@ func (PiperMail) FetchSubmittedAt(urls []string) (time.Time, error) {
 	return time.Time{}, nil
 }
 
-func (PiperMail) FetchLastUpdated(urls []string) (time.Time, User, error) {
-	var ret time.Time
+func (PiperMail) nextMonth(ym string) string {
+	yStr, mStr, ok := strings.Cut(ym, "-")
+	if !ok {
+		panic(fmt.Errorf("invalid year-month: %q", ym))
+	}
+	switch mStr {
+	case "January":
+		return yStr + "-February"
+	case "February":
+		return yStr + "-March"
+	case "March":
+		return yStr + "-April"
+	case "April":
+		return yStr + "-May"
+	case "May":
+		return yStr + "-June"
+	case "June":
+		return yStr + "-July"
+	case "July":
+		return yStr + "-August"
+	case "August":
+		return yStr + "-September"
+	case "September":
+		return yStr + "-October"
+	case "October":
+		return yStr + "-November"
+	case "November":
+		return yStr + "-December"
+	case "December":
+		y, _ := strconv.Atoi(yStr)
+		return fmt.Sprintf("%d-January", y+1)
+	default:
+		panic(fmt.Errorf("invalid year-month: %q", ym))
+	}
+}
+
+func (PiperMail) messageID(u string) (string, error) {
+}
+
+
+func (p PiperMail) FetchLastUpdated(urls []string) (time.Time, User, error) {
 	for _, u := range urls {
-		if m := reGitHubCommit.FindStringSubmatch(u); m != nil {
-			user := m[1]
-			repo := m[2]
-			hash := m[3]
+		m := rePiperMailMessage.FindStringSubmatch(u)
+		if m == nil {
+			continue
+		}
+		uBase := m[1]
+		uYM := m[2]
+		//uInt := m[3]
 
-			urlStr := "https://api.github.com/repos/" + user + "/" + repo + "/commits/" + hash
-			var obj struct {
-				Commit struct {
-					Author struct {
-						Date time.Time `json:"date"`
-					} `json:"author"`
-					Committer struct {
-						Date time.Time `json:"date"`
-					} `json:"committer"`
-				} `json:"commit"`
-			}
-			if err := httpGetJSON(urlStr, nil, &obj); err != nil {
-				return time.Time{}, User{}, err
-			}
-			if obj.Commit.Author.Date.After(ret) {
-				ret = obj.Commit.Author.Date
-			}
-			if obj.Commit.Committer.Date.After(ret) {
-				ret = obj.Commit.Committer.Date
+		htmlStr, err := httpGet(u, nil)
+		if err != nil {
+			return time.Time{}, User{}, err
+		}
+		var msgid string
+		for _, line := range strings.Split(htmlStr, "\n") {
+			if m := rePiperMailReply.FindStringSubmatch(line); m != nil {
+				ru, err := url.Parse(m[1])
+				if err != nil {
+					continue
+				}
+				if msgid = ru.Query().Get("In-Reply-To"); msgid != "" {
+					break
+				}
 			}
 		}
+		if msgid == "" {
+			continue
+		}
+		mboxStr, err := httpGet(uBase+uYM+".txt.gz", nil)
+		if err != nil {
+			return time.Time{}, User{}, err
+		}
 	}
-	if ret.IsZero() {
-		return time.Time{}, User{}, nil
-	}
-	return ret, User{}, nil
+	return time.Time{}, User{}, nil
 }