feat(monkeyd): add /monkeyd_tags to report a novel's tags as hashtags

Replies with a hashtag line led by #MonkeyD, then a blank line and the novel URL, sent as a code block so it can be copied in one tap. Each genre label becomes one hashtag with spaces and punctuation stripped and every word capitalised, since Telegram ends a hashtag at the first character that is not a letter, digit, or underscore. Diacritics survive; a label with no letters is dropped rather than emitted as a bare hash. The command costs one request and runs inline under a short timeout rather than in the background: handlers are dispatched one at a time, so a stalled fetch would block every other command. It shares the export page cache. Advance the submodule to the tag parser, which matches itemprop="genre" microdata so the site-wide genre navigation stays out of the result.
2026-07-31 06:26:13 +00:00 · 2026-07-30 00:52:10 +07:00
parent 0ddc6b3ccc
commit d2f6e97fb5
9 changed files with 508 additions and 22 deletions
@@ -16,7 +16,7 @@ Atlas via long polling and an in-process cron scheduler.
 | `gold` | Gold paper trading (opt-in; VNAppMob SJC buy/sell VND/luong) |
 | `coin` | Crypto paper trading in USD (Binance -> Coinbase -> CoinGecko price fallback) |
 | `stats` | `/stats` (top commands), `/stats users`, `/stats user <username>`, `/stats cmd <command_name>` |
-| `monkeyd` | `/monkeyd_crawl <url> [font_size]` — export a monkeydd.com novel as a PDF |
+| `monkeyd` | `/monkeyd_crawl <url> [font_size]` export a monkeydd.com novel as a PDF, `/monkeyd_tags <url>` list its tags as hashtags |

 Disable modules with the `MODULES` environment variable.

@@ -170,6 +170,33 @@ The PDF embeds a font covering Vietnamese diacritics. The crawler prefers a
 system font and falls back to one compiled into the binary, so the runtime
 image needs no fonts installed.

+### Novel tags
+
+`/monkeyd_tags <url>` reports a novel's tags as a hashtag line, sent as a code
+block so it can be copied in one tap:
+
+```text
+#MonkeyD #CổĐại #GiaĐình
+
+https://monkeydd.com/truong-an-gwem.html
+```
+
+Each label becomes one hashtag with spaces and punctuation removed and every
+word capitalised, because Telegram ends a hashtag at the first character that is
+not a letter, digit, or underscore. Diacritics are kept. A label with no letters
+is dropped rather than emitted as a bare `#`.
+
+The tags are the novel's own genres, identified by their `itemprop="genre"`
+microdata. The same page also links every genre on the site as navigation — 69
+of them against one novel's 6 in a sampled page — so matching the category URL
+shape instead would return the whole menu.
+
+This command costs a single request and runs inline rather than in the
+background, bounded by a short timeout: handlers are dispatched one at a time, so
+a stalled fetch would hold up every other command. It shares the export page
+cache, so a lookup for a novel that was already exported costs no request at
+all.
+
 ## Layout

 ```
@@ -68,6 +68,7 @@ func TestCommandDiscovery_AllPublicCommandsHaveSafeMetadata(t *testing.T) {
 		"lol":                  "[date]",
 		"loldle":               "[champion]",
 		"monkeyd_crawl":        "<url> [font_size]",
+		"monkeyd_tags":         "<url>",
 		"random":               "<option,...>",
 		"stats":                "[users | user <username> | cmd <command_name>]",
 		"stock_events":         "<ticker> [days]",
@@ -26,12 +26,22 @@ const testNovelURL = "https://monkeydd.com/tro-lai-nam-thang-cu.html"
 // ownerID is permitted, which /monkeyd_crawl requires: the command is
 // Protected, and the dispatcher drops unauthorized calls silently.
 func install(t *testing.T, ownerID int64) (*testutil.RecordingBot, *runner) {
+	t.Helper()
+	return installWith(t, ownerID, func(context.Context, string) ([]string, error) {
+		t.Helper()
+		t.Error("tags fetcher called by an export test")
+		return nil, nil
+	})
+}
+
+// installWith is install with an explicit tag fetcher, for the tags command.
+func installWith(t *testing.T, ownerID int64, fetch tagsFetcher) (*testutil.RecordingBot, *runner) {
 	t.Helper()
 	rb := testutil.NewRecordingBot(t)
 	r := newRunner()
 	// Run the export inline so assertions do not race a detached goroutine.
 	r.launch = func(job func()) { job() }
-	mod := newModule(r)
+	mod := newModule(r, fetch)

 	reg := &modules.Registry{
 		Modules:     []modules.Module{{Name: "monkeyd", Commands: mod.Commands}},
@@ -335,25 +345,40 @@ func TestCrawl_AvailableToAnySender(t *testing.T) {

 func TestRegistration(t *testing.T) {
 	mod := New(modules.Deps{Store: storage.NewMemoryProvider().Collection("monkeyd")})
-	if len(mod.Commands) != 1 {
-		t.Fatalf("expected 1 command, got %d", len(mod.Commands))
+
+	wantParameters := map[string]string{
+		commandName:     "<url> [font_size]",
+		tagsCommandName: "<url>",
 	}
-	cmd := mod.Commands[0]
-	if cmd.Name != commandName {
-		t.Errorf("Name = %q, want %q", cmd.Name, commandName)
+	if len(mod.Commands) != len(wantParameters) {
+		t.Fatalf("expected %d commands, got %d", len(wantParameters), len(mod.Commands))
 	}
-	if cmd.Visibility != modules.VisibilityPublic {
-		t.Errorf("Visibility = %v, want Public", cmd.Visibility)
+
+	for _, cmd := range mod.Commands {
+		want, known := wantParameters[cmd.Name]
+		if !known {
+			t.Errorf("unexpected command %q", cmd.Name)
+			continue
+		}
+		delete(wantParameters, cmd.Name)
+
+		if cmd.Parameters != want {
+			t.Errorf("/%s parameters = %q, want %q", cmd.Name, cmd.Parameters, want)
+		}
+		if cmd.Visibility != modules.VisibilityPublic {
+			t.Errorf("/%s visibility = %v, want Public", cmd.Name, cmd.Visibility)
+		}
+		if cmd.Description == "" {
+			t.Errorf("/%s description is empty; command discovery requires one", cmd.Name)
+		}
+		if cmd.Handler == nil {
+			t.Errorf("/%s handler is nil", cmd.Name)
+		}
 	}
-	if cmd.Parameters != "<url> [font_size]" {
-		t.Errorf("Parameters = %q, want %q", cmd.Parameters, "<url> [font_size]")
-	}
-	if cmd.Description == "" {
-		t.Error("Description is empty; command discovery requires one")
-	}
-	if cmd.Handler == nil {
-		t.Error("Handler is nil")
+	for name := range wantParameters {
+		t.Errorf("command %q was not registered", name)
 	}
+
 	if len(mod.Crons) != 0 || len(mod.Callbacks) != 0 {
 		t.Errorf("expected no crons or callbacks, got %d crons and %d callbacks",
 			len(mod.Crons), len(mod.Callbacks))
@@ -0,0 +1,77 @@
+package monkeyd
+
+import (
+	"strings"
+	"unicode"
+)
+
+// sourceHashtag leads every tag line so the source is visible when the line is
+// pasted somewhere on its own.
+const sourceHashtag = "#MonkeyD"
+
+// tagsMessage renders the tag line followed by the novel URL, separated by a
+// blank line. Plain text: the caller decides how to present it.
+func tagsMessage(tags []string, novelURL string) string {
+	return hashtagLine(tags) + "\n\n" + novelURL
+}
+
+// hashtagLine turns genre labels into a single space-separated line of
+// hashtags, led by sourceHashtag. Labels that cannot form a usable hashtag are
+// dropped, and duplicates collapse — two labels differing only in punctuation
+// or spacing produce the same hashtag.
+func hashtagLine(tags []string) string {
+	parts := []string{sourceHashtag}
+	seen := map[string]bool{sourceHashtag: true}
+	for _, tag := range tags {
+		body := hashtag(tag)
+		if body == "" {
+			continue
+		}
+		tag := "#" + body
+		if seen[tag] {
+			continue
+		}
+		seen[tag] = true
+		parts = append(parts, tag)
+	}
+	return strings.Join(parts, " ")
+}
+
+// hashtag converts a genre label into a hashtag body: "Cổ Đại" -> "CổĐại".
+//
+// Telegram ends a hashtag at the first character that is not a letter, digit or
+// underscore, so spaces and punctuation are removed rather than replaced, and
+// each word is capitalised to keep the boundaries readable. Diacritics survive:
+// Telegram accepts non-ASCII letters in hashtags.
+//
+// Returns "" for a label with no letters. A digits-only hashtag is not linkified
+// by Telegram, so emitting one would only add noise.
+func hashtag(tag string) string {
+	var b strings.Builder
+	b.Grow(len(tag))
+
+	hasLetter := false
+	wordStart := true
+	for _, r := range tag {
+		switch {
+		case unicode.IsLetter(r):
+			hasLetter = true
+			if wordStart {
+				r = unicode.ToUpper(r)
+			}
+			b.WriteRune(r)
+			wordStart = false
+		case unicode.IsDigit(r):
+			b.WriteRune(r)
+			wordStart = false
+		default:
+			// Dropped, but it still ends the word, so the next letter is
+			// capitalised: "sci-fi" -> "SciFi".
+			wordStart = true
+		}
+	}
+	if !hasLetter {
+		return ""
+	}
+	return b.String()
+}
@@ -0,0 +1,81 @@
+package monkeyd
+
+import "testing"
+
+func TestHashtag(t *testing.T) {
+	tests := []struct {
+		tag  string
+		want string
+	}{
+		{"Cổ Đại", "CổĐại"},
+		{"Gia Đình", "GiaĐình"},
+		{"Trọng Sinh", "TrọngSinh"},
+		{"Nữ Cường", "NữCường"},
+		{"Vả Mặt", "VảMặt"},
+		// Single word passes through untouched.
+		{"Ngôn", "Ngôn"},
+		// A lower-case label is capitalised per word.
+		{"ngôn tình", "NgônTình"},
+		// Punctuation is removed but still ends the word.
+		{"sci-fi", "SciFi"},
+		{"Đông/Tây", "ĐôngTây"},
+		{"Truyện (Chữ)", "TruyệnChữ"},
+		// Collapsed whitespace and surrounding space.
+		{"  Cổ   Đại  ", "CổĐại"},
+		// Digits are kept.
+		{"3D", "3D"},
+		{"18+ Tuổi", "18Tuổi"},
+		// Nothing usable: Telegram will not linkify a digits-only hashtag.
+		{"18+", ""},
+		{"", ""},
+		{"---", ""},
+		{"+", ""},
+	}
+	for _, tt := range tests {
+		if got := hashtag(tt.tag); got != tt.want {
+			t.Errorf("hashtag(%q) = %q, want %q", tt.tag, got, tt.want)
+		}
+	}
+}
+
+func TestHashtagLine(t *testing.T) {
+	got := hashtagLine([]string{"Cổ Đại", "Gia Đình"})
+	want := "#MonkeyD #CổĐại #GiaĐình"
+	if got != want {
+		t.Errorf("hashtagLine() = %q, want %q", got, want)
+	}
+}
+
+func TestHashtagLineWithoutTags(t *testing.T) {
+	if got := hashtagLine(nil); got != sourceHashtag {
+		t.Errorf("hashtagLine(nil) = %q, want %q", got, sourceHashtag)
+	}
+}
+
+// Labels differing only in punctuation collapse to one hashtag, and an unusable
+// label is dropped rather than emitted as a bare "#".
+func TestHashtagLineDropsUnusableAndDuplicates(t *testing.T) {
+	got := hashtagLine([]string{"Cổ Đại", "Cổ-Đại", "18+", "Gia Đình"})
+	want := "#MonkeyD #CổĐại #GiaĐình"
+	if got != want {
+		t.Errorf("hashtagLine() = %q, want %q", got, want)
+	}
+}
+
+func TestTagsMessage(t *testing.T) {
+	got := tagsMessage([]string{"Cổ Đại", "Gia Đình"}, "https://monkeydd.com/truong-an-gwem.html")
+	want := "#MonkeyD #CổĐại #GiaĐình\n\nhttps://monkeydd.com/truong-an-gwem.html"
+	if got != want {
+		t.Errorf("tagsMessage() =\n%q\nwant\n%q", got, want)
+	}
+}
+
+// The block is what makes the line copyable in one tap, and its contents must be
+// escaped so a label containing markup cannot break out of it.
+func TestMonospaceBlockEscapes(t *testing.T) {
+	got := monospaceBlock("#MonkeyD #A<b>&")
+	want := "<pre>#MonkeyD #A&lt;b&gt;&amp;</pre>"
+	if got != want {
+		t.Errorf("monospaceBlock() = %q, want %q", got, want)
+	}
+}
@@ -36,12 +36,13 @@ var usage = fmt.Sprintf(
 // New is the module Factory. The module keeps no persistent state — an export
 // is a one-shot job — so deps.Store is unused.
 func New(_ modules.Deps) modules.Module {
-	return newModule(newRunner())
+	return newModule(newRunner(), fetchTags)
 }

-// newModule builds the module around a given runner, which is how tests supply
-// one with a stubbed exporter and a synchronous launch.
-func newModule(r *runner) modules.Module {
+// newModule builds the module around a given runner and tag fetcher, which is
+// how tests supply a stubbed exporter, a synchronous launch, and tags without
+// network access.
+func newModule(r *runner, fetch tagsFetcher) modules.Module {
 	return modules.Module{
 		Commands: []modules.Command{
 			{
@@ -55,6 +56,7 @@ func newModule(r *runner) modules.Module {
 				Parameters:  parameters,
 				Handler:     r.handle,
 			},
+			tagsCommand(fetch),
 		},
 	}
 }
@@ -0,0 +1,122 @@
+package monkeyd
+
+import (
+	"context"
+	"fmt"
+	"html"
+	"os"
+	"path/filepath"
+	"strings"
+	"time"
+
+	"github.com/go-telegram/bot"
+	"github.com/go-telegram/bot/models"
+
+	"github.com/tiennm99/monkeyd-crawler/export"
+	crawler "github.com/tiennm99/monkeyd-crawler/monkeyd"
+
+	"github.com/tiennm99/miti99bot/internal/log"
+	"github.com/tiennm99/miti99bot/internal/modules"
+	"github.com/tiennm99/miti99bot/internal/modules/util/chathelper"
+)
+
+// tagsCommandName is the command that reports a novel's tags.
+const tagsCommandName = "monkeyd_tags"
+
+// tagsParameters is the display syntax shared by the command menu, /help, and
+// the usage text.
+const tagsParameters = "<url>"
+
+const tagsUsage = "Usage: /" + tagsCommandName + " " + tagsParameters
+
+const (
+	// tagsFetchTimeout bounds the whole command. Unlike an export this is one
+	// request and runs inline, and handlers are dispatched one at a time, so a
+	// stalled fetch would hold up every other command until it gives up.
+	tagsFetchTimeout = 30 * time.Second
+
+	// tagsRetries is deliberately lower than the export's: the caller is
+	// waiting on this reply, so failing early beats a long retry chain.
+	tagsRetries = 1
+)
+
+// tagsFetcher returns a novel's tags. A field on the module rather than a
+// direct call so tests can exercise the command without network access.
+type tagsFetcher func(ctx context.Context, novelURL string) ([]string, error)
+
+func tagsCommand(fetch tagsFetcher) modules.Command {
+	return modules.Command{
+		Name:        tagsCommandName,
+		Visibility:  modules.VisibilityPublic,
+		Description: "Show a " + AllowedHostsHint + " novel's tags as hashtags",
+		Parameters:  tagsParameters,
+		Handler:     tagsHandler(fetch),
+	}
+}
+
+func tagsHandler(fetch tagsFetcher) modules.CommandHandler {
+	return func(ctx context.Context, b *bot.Bot, update *models.Update) error {
+		msg := update.Message
+		if msg == nil {
+			return nil
+		}
+
+		arg := chathelper.ArgAfterCommand(msg.Text)
+		if arg == "" {
+			return chathelper.Reply(ctx, b, msg, tagsUsage)
+		}
+		fields := strings.Fields(arg)
+		if len(fields) > 1 {
+			return chathelper.Reply(ctx, b, msg,
+				fmt.Sprintf("%s.\n%s", capitalize(errTooManyArgs.Error()), tagsUsage))
+		}
+		novelURL, err := normalizeNovelURL(fields[0])
+		if err != nil {
+			return chathelper.Reply(ctx, b, msg,
+				fmt.Sprintf("%s.\n%s", capitalize(err.Error()), tagsUsage))
+		}
+
+		// Reserve part of the handler's budget for the reply itself, then cap
+		// the fetch so it cannot outlive the command.
+		fetchCtx, cancel := chathelper.FetchContext(ctx)
+		defer cancel()
+		fetchCtx, cancelTimeout := context.WithTimeout(fetchCtx, tagsFetchTimeout)
+		defer cancelTimeout()
+
+		tags, err := fetch(fetchCtx, novelURL)
+		if err != nil {
+			log.Error("monkeyd tags fetch failed",
+				"command", tagsCommandName, "url", novelURL, "err", err)
+			return chathelper.Reply(ctx, b, msg, "Could not read that novel page: "+err.Error())
+		}
+		if len(tags) == 0 {
+			return chathelper.Reply(ctx, b, msg, "That page lists no tags.")
+		}
+
+		// Sent as a code block so the line can be copied in one tap.
+		return chathelper.ReplyHTML(ctx, b, msg, monospaceBlock(tagsMessage(tags, novelURL)))
+	}
+}
+
+// monospaceBlock wraps text in Telegram's <pre> block, which renders as a
+// copyable code box. The content is escaped, so a tag containing < or & cannot
+// break the markup.
+func monospaceBlock(text string) string {
+	return "<pre>" + html.EscapeString(text) + "</pre>"
+}
+
+// fetchTags reads a novel's tags from the site.
+//
+// It shares the export cache directory, so tags requested for a novel that was
+// already exported — or an export following a tags lookup — cost no request.
+func fetchTags(ctx context.Context, novelURL string) ([]string, error) {
+	c := &crawler.Crawler{
+		Client:   crawler.NewClient(export.DefaultDelay, tagsRetries),
+		CacheDir: filepath.Join(os.TempDir(), cacheDirName),
+	}
+	novel, err := c.NovelInfo(ctx, novelURL)
+	if err != nil {
+		return nil, err
+	}
+	return novel.Tags, nil
+}
@@ -0,0 +1,151 @@
+package monkeyd
+
+import (
+	"context"
+	"errors"
+	"strings"
+	"testing"
+
+	"github.com/tiennm99/miti99bot/internal/testutil"
+)
+
+// stubTags returns a fetcher yielding fixed tags and records the URL it saw.
+func stubTags(tags []string, gotURL *string) tagsFetcher {
+	return func(_ context.Context, novelURL string) ([]string, error) {
+		if gotURL != nil {
+			*gotURL = novelURL
+		}
+		return tags, nil
+	}
+}
+
+func TestTags_RepliesWithHashtagBlock(t *testing.T) {
+	var gotURL string
+	rb, _ := installWith(t, 999, stubTags([]string{"Cổ Đại", "Gia Đình"}, &gotURL))
+
+	rb.Bot.ProcessUpdate(context.Background(),
+		testutil.NewPrivateMessage(999, "/monkeyd_tags https://monkeydd.com/truong-an-gwem.html"))
+
+	if want := "https://monkeydd.com/truong-an-gwem.html"; gotURL != want {
+		t.Errorf("fetcher got %q, want %q", gotURL, want)
+	}
+
+	sent := rb.LastSent()
+	want := "<pre>#MonkeyD #CổĐại #GiaĐình\n\nhttps://monkeydd.com/truong-an-gwem.html</pre>"
+	if got := sent.Text(); got != want {
+		t.Errorf("reply =\n%q\nwant\n%q", got, want)
+	}
+	// Without HTML parse mode Telegram would show the literal <pre> tags.
+	if got := sent.Form["parse_mode"]; got != "HTML" {
+		t.Errorf("parse_mode = %q, want HTML", got)
+	}
+}
+
+func TestTags_NoArgumentRepliesUsage(t *testing.T) {
+	rb, _ := installWith(t, 999, stubTags(nil, nil))
+	rb.Bot.ProcessUpdate(context.Background(), testutil.NewPrivateMessage(999, "/monkeyd_tags"))
+
+	if got := rb.LastSent().Text(); !strings.Contains(got, tagsUsage) {
+		t.Errorf("reply = %q, want it to contain %q", got, tagsUsage)
+	}
+}
+
+func TestTags_RejectsDisallowedHost(t *testing.T) {
+	called := false
+	rb, _ := installWith(t, 999, func(context.Context, string) ([]string, error) {
+		called = true
+		return nil, nil
+	})
+
+	rb.Bot.ProcessUpdate(context.Background(),
+		testutil.NewPrivateMessage(999, "/monkeyd_tags https://example.com/novel.html"))
+
+	if called {
+		t.Error("fetcher ran for a disallowed host")
+	}
+	if got := rb.LastSent().Text(); !strings.Contains(got, AllowedHostsHint) {
+		t.Errorf("reply = %q, want it to name %q", got, AllowedHostsHint)
+	}
+}
+
+// A bare hostname is normalised before the fetch, matching /monkeyd_crawl, and
+// the reply quotes the normalised form rather than what was typed.
+func TestTags_NormalizesURL(t *testing.T) {
+	const wantURL = "https://monkeydd.com/truong-an-gwem.html"
+
+	var gotURL string
+	rb, _ := installWith(t, 999, stubTags([]string{"Cổ Đại"}, &gotURL))
+
+	rb.Bot.ProcessUpdate(context.Background(),
+		testutil.NewPrivateMessage(999, "/monkeyd_tags monkeydd.com/truong-an-gwem.html"))
+
+	if gotURL != wantURL {
+		t.Errorf("fetcher got %q, want %q", gotURL, wantURL)
+	}
+	if got := rb.LastSent().Text(); !strings.Contains(got, wantURL) {
+		t.Errorf("reply = %q, want it to carry the normalised URL", got)
+	}
+}
+
+func TestTags_ReportsFetchFailure(t *testing.T) {
+	rb, _ := installWith(t, 999, func(context.Context, string) ([]string, error) {
+		return nil, errors.New("fetch failed: 404")
+	})
+
+	rb.Bot.ProcessUpdate(context.Background(),
+		testutil.NewPrivateMessage(999, "/monkeyd_tags "+testNovelURL))
+
+	got := rb.LastSent().Text()
+	if !strings.Contains(got, "404") {
+		t.Errorf("reply = %q, want it to carry the underlying error", got)
+	}
+	if strings.Contains(got, sourceHashtag) {
+		t.Errorf("reply = %q, want no hashtag line on failure", got)
+	}
+}
+
+// An empty tag list must say so rather than send a lone "#MonkeyD".
+func TestTags_ReportsEmptyTagList(t *testing.T) {
+	rb, _ := installWith(t, 999, stubTags(nil, nil))
+
+	rb.Bot.ProcessUpdate(context.Background(),
+		testutil.NewPrivateMessage(999, "/monkeyd_tags "+testNovelURL))
+
+	got := rb.LastSent().Text()
+	if strings.Contains(got, sourceHashtag) {
+		t.Errorf("reply = %q, want a plain explanation rather than a bare source hashtag", got)
+	}
+	if !strings.Contains(strings.ToLower(got), "no tags") {
+		t.Errorf("reply = %q, want it to say no tags were listed", got)
+	}
+}
+
+func TestTags_RejectsExtraArguments(t *testing.T) {
+	called := false
+	rb, _ := installWith(t, 999, func(context.Context, string) ([]string, error) {
+		called = true
+		return nil, nil
+	})
+
+	rb.Bot.ProcessUpdate(context.Background(),
+		testutil.NewPrivateMessage(999, "/monkeyd_tags "+testNovelURL+" extra"))
+
+	if called {
+		t.Error("fetcher ran despite extra arguments")
+	}
+	if got := rb.LastSent().Text(); !strings.Contains(got, tagsUsage) {
+		t.Errorf("reply = %q, want usage text", got)
+	}
+}
+
+// The command is public, so a non-admin in a group must get a reply.
+func TestTags_AvailableToAnySender(t *testing.T) {
+	rb, _ := installWith(t, 999, stubTags([]string{"Cổ Đại"}, nil))
+
+	rb.Bot.ProcessUpdate(context.Background(),
+		testutil.NewGroupMessage(-100, 12345, "/monkeyd_tags "+testNovelURL))
+
+	if got := rb.LastSent().Text(); !strings.Contains(got, "#CổĐại") {
+		t.Errorf("reply = %q, want the hashtag line", got)
+	}
+}