From d9893579b03eb718c7620923233b0000357c5443 Mon Sep 17 00:00:00 2001
From: Ted Unangst <tedu@tedunangst.com>
Date: Mon, 23 Dec 2019 01:39:17 -0500
Subject: [PATCH] import command support the elephant in the room

---
 docs/changelog.txt |   2 +
 docs/honk.8        |   7 ++-
 honk.go            |   2 +-
 import.go          | 120 +++++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 128 insertions(+), 3 deletions(-)

diff --git a/docs/changelog.txt b/docs/changelog.txt
index db0ad93..4b5b774 100644
--- a/docs/changelog.txt
+++ b/docs/changelog.txt
@@ -2,6 +2,8 @@ changelog
 
 === next
 
+++ Import command now supports the elephant in the room.
+
 + deluser command.
 
 + Configurable avatar colors.
diff --git a/docs/honk.8 b/docs/honk.8
index 0feaaa5..1575f4d 100644
--- a/docs/honk.8
+++ b/docs/honk.8
@@ -162,9 +162,12 @@ Data may be imported and converted from other services using the
 command.
 Posts are imported and backdated to appear as old honks.
 No other information is imported.
+Currently supports Mastodon and Twitter exported data.
 .Pp
-Currently limited to Twitter import.
-This requires a Twitter data archive.
+To prepare a Mastodon data archive, extract the archive-longhash.tar.gz file.
+.Dl ./honk import username mastodon source-directory
+.Pp
+To prepare a Twitter data archive, extract the twitter-longhash.zip file.
 After unzipping the data archive, navigate to the tweet_media directory
 and unzip any zip files contained within.
 .Dl ./honk import username twitter source-directory
diff --git a/honk.go b/honk.go
index bd0357d..ef096ab 100644
--- a/honk.go
+++ b/honk.go
@@ -253,7 +253,7 @@ func main() {
 		adminscreen()
 	case "import":
 		if len(args) != 4 {
-			log.Fatal("import username twitter [srcdir]")
+			log.Fatal("import username mastodon|twitter srcdir")
 		}
 		importMain(args[1], args[2], args[3])
 	case "debug":
diff --git a/import.go b/import.go
index df7b0ac..c2e8233 100644
--- a/import.go
+++ b/import.go
@@ -22,13 +22,18 @@ import (
 	"io/ioutil"
 	"log"
 	"os"
+	"regexp"
 	"sort"
 	"strings"
 	"time"
+
+	"humungus.tedunangst.com/r/webs/htfilter"
 )
 
 func importMain(username, flavor, source string) {
 	switch flavor {
+	case "mastodon":
+		importMastodon(username, source)
 	case "twitter":
 		importTwitter(username, source)
 	default:
@@ -36,6 +41,121 @@ func importMain(username, flavor, source string) {
 	}
 }
 
+func importMastodon(username, source string) {
+	user, err := butwhatabout(username)
+	if err != nil {
+		log.Fatal(err)
+	}
+	type Toot struct {
+		Id           string
+		Type         string
+		To           []string
+		Cc           []string
+		Summary      string
+		Content      string
+		InReplyTo    string
+		Conversation string
+		Published    time.Time
+		Tag          []struct {
+			Type string
+			Name string
+		}
+		Attachment []struct {
+			Type      string
+			MediaType string
+			Url       string
+			Name      string
+		}
+	}
+	var outbox struct {
+		OrderedItems []struct {
+			Object Toot
+		}
+	}
+	fd, err := os.Open(source + "/outbox.json")
+	if err != nil {
+		log.Fatal(err)
+	}
+	dec := json.NewDecoder(fd)
+	err = dec.Decode(&outbox)
+	if err != nil {
+		log.Fatalf("error parsing json: %s", err)
+	}
+	fd.Close()
+
+	havetoot := func(xid string) bool {
+		var id int64
+		row := stmtFindXonk.QueryRow(user.ID, xid)
+		err := row.Scan(&id)
+		if err == nil {
+			return true
+		}
+		return false
+	}
+
+	re_tootid := regexp.MustCompile("[^/]+$")
+	for _, item := range outbox.OrderedItems {
+		toot := item.Object
+		tootid := re_tootid.FindString(toot.Id)
+		xid := fmt.Sprintf("%s/%s/%s", user.URL, honkSep, tootid)
+		if havetoot(xid) {
+			continue
+		}
+		honk := Honk{
+			UserID:   user.ID,
+			What:     "honk",
+			Honker:   user.URL,
+			XID:      xid,
+			RID:      toot.InReplyTo,
+			Date:     toot.Published,
+			URL:      xid,
+			Audience: append(toot.To, toot.Cc...),
+			Noise:    toot.Content,
+			Convoy:   toot.Conversation,
+			Whofore:  2,
+			Format:   "html",
+			Precis:   toot.Summary,
+		}
+		if honk.RID != "" {
+			honk.What = "tonk"
+		}
+		if !loudandproud(honk.Audience) {
+			honk.Whofore = 3
+		}
+		for _, att := range toot.Attachment {
+			switch att.Type {
+			case "Document":
+				fname := fmt.Sprintf("%s/%s", source, att.Url)
+				data, err := ioutil.ReadFile(fname)
+				if err != nil {
+					log.Printf("error reading media: %s", fname)
+					continue
+				}
+				u := xfiltrate()
+				name := att.Name
+				desc := name
+				newurl := fmt.Sprintf("https://%s/d/%s", serverName, u)
+				fileid, err := savefile(u, name, desc, newurl, att.MediaType, true, data)
+				if err != nil {
+					log.Printf("error saving media: %s", fname)
+					continue
+				}
+				donk := &Donk{
+					FileID: fileid,
+				}
+				honk.Donks = append(honk.Donks, donk)
+			}
+		}
+		for _, t := range toot.Tag {
+			switch t.Type {
+			case "Hashtag":
+				honk.Onts = append(honk.Onts, t.Name)
+			}
+		}
+		savehonk(&honk)
+	}
+}
+
 func importTwitter(username, source string) {
 	user, err := butwhatabout(username)
 	if err != nil {