improved search

internal/cid/card.go

@@ -0,0 +1,92 @@
+package cid
+
+import (
+	"encoding/base32"
+	"encoding/binary"
+	"fmt"
+	"strings"
+	"unicode/utf8"
+)
+
+// DecodeCard recovers the (source, id) pair a card locator (CodeCard, 0x1007)
+// wraps. Wire form:
+//
+//	0x01 ‖ varint(0x1007) ‖ 0x00 (identity mh) ‖ varint(len(digest)) ‖ digest
+//	digest = varint(len(source)) ‖ source ‖ id        (both UTF-8)
+//
+// Decode stops at the multihash code because ranking needs nothing past it;
+// reading *which work* a card addresses needs the digest too, so this is the
+// one place in Go that walks the whole thing.
+//
+// ⚠ Mirror of meta-read `card_cid::decode` and of meta-feeder-sdk
+// `compute_card_cid`. The vectors in card_test.go are pinned literals copied
+// from meta-read, not round-trips, so a drift on either side fails a test
+// instead of silently reading a different work out of the same CID.
+//
+// Identity multihash only: a 0x1007 CID carrying any other multihash is not
+// something anyone mints, and reading its digest as (source, id) would be
+// reading hash bytes as text. Bytes past the declared digest length are
+// ignored, as meta-read does.
+func DecodeCard(cidStr string) (source, id string, err error) {
+	if len(cidStr) < 2 || cidStr[0] != 'b' {
+		return "", "", fmt.Errorf("cid %q: not a multibase base32 CIDv1", cidStr)
+	}
+	raw, err := base32.StdEncoding.WithPadding(base32.NoPadding).DecodeString(strings.ToUpper(cidStr[1:]))
+	if err != nil {
+		return "", "", fmt.Errorf("cid %q: base32 decode: %w", cidStr, err)
+	}
+
+	next := func(buf []byte, what string) (uint64, []byte, error) {
+		v, n := binary.Uvarint(buf)
+		if n <= 0 {
+			return 0, nil, fmt.Errorf("cid %q: bad %s varint", cidStr, what)
+		}
+		return v, buf[n:], nil
+	}
+
+	version, raw, err := next(raw, "version")
+	if err != nil {
+		return "", "", err
+	}
+	if version != 1 {
+		return "", "", fmt.Errorf("cid %q: unsupported CID version %d", cidStr, version)
+	}
+	codec, raw, err := next(raw, "codec")
+	if err != nil {
+		return "", "", err
+	}
+	if codec != CodeCard {
+		return "", "", fmt.Errorf("cid %q: codec 0x%x is not a card locator", cidStr, codec)
+	}
+	mh, raw, err := next(raw, "multihash-code")
+	if err != nil {
+		return "", "", err
+	}
+	if mh != 0 {
+		return "", "", fmt.Errorf("cid %q: card locator with non-identity multihash 0x%x", cidStr, mh)
+	}
+	digestLen, raw, err := next(raw, "digest-length")
+	if err != nil {
+		return "", "", err
+	}
+	if digestLen > uint64(len(raw)) {
+		return "", "", fmt.Errorf("cid %q: digest truncated", cidStr)
+	}
+	digest := raw[:digestLen]
+
+	sourceLen, rest, err := next(digest, "source-length")
+	if err != nil {
+		return "", "", err
+	}
+	if sourceLen > uint64(len(rest)) {
+		return "", "", fmt.Errorf("cid %q: source truncated", cidStr)
+	}
+	source, id = string(rest[:sourceLen]), string(rest[sourceLen:])
+	if source == "" || id == "" {
+		return "", "", fmt.Errorf("cid %q: empty source or id", cidStr)
+	}
+	if !utf8.ValidString(source) || !utf8.ValidString(id) {
+		return "", "", fmt.Errorf("cid %q: source or id is not UTF-8", cidStr)
+	}
+	return source, id, nil
+}

internal/cid/card_test.go

@@ -0,0 +1,71 @@
+package cid
+
+import (
+	"encoding/base32"
+	"encoding/binary"
+	"strings"
+	"testing"
+)
+
+// Pinned from meta-read `src/card_cid.rs` / meta-feeder-sdk `compute_card_cid`
+// — literals, not round-trips (see DecodeCard).
+const (
+	goldenAnilistCard  = "bagdsaaaua5qw42lmnfzxi3lbnztwcorrga2tonzy"
+	goldenMangadexCard = "bagdsaabybrwwc3thmfsgk6bon5zgol3nmfxgoyjpmiygenzsgftgmlldgm4dqljugq4dmllbmeygmlldgjrdaytcgmzdcnjrgi"
+)
+
+// wireCID builds a CIDv1 from raw parts so the negative cases below can vary
+// one field at a time.
+func wireCID(codec, mh uint64, digest []byte) string {
+	buf := []byte{0x01}
+	buf = binary.AppendUvarint(buf, codec)
+	buf = binary.AppendUvarint(buf, mh)
+	buf = binary.AppendUvarint(buf, uint64(len(digest)))
+	buf = append(buf, digest...)
+	return "b" + strings.ToLower(base32.StdEncoding.WithPadding(base32.NoPadding).EncodeToString(buf))
+}
+
+func cardDigest(source, id string) []byte {
+	d := binary.AppendUvarint(nil, uint64(len(source)))
+	d = append(d, source...)
+	return append(d, id...)
+}
+
+func TestDecodeCard_GoldenVectors(t *testing.T) {
+	for _, tc := range []struct {
+		cid, source, id string
+	}{
+		{goldenAnilistCard, "anilist", "manga:105778"},
+		{goldenMangadexCard, "mangadex.org", "/manga/b0b721ff-c388-4486-aa0f-c2b0bb321512"},
+	} {
+		source, id, err := DecodeCard(tc.cid)
+		if err != nil {
+			t.Fatalf("DecodeCard(%s): %v", tc.cid, err)
+		}
+		if source != tc.source || id != tc.id {
+			t.Errorf("DecodeCard(%s) = (%q, %q), want (%q, %q)", tc.cid, source, id, tc.source, tc.id)
+		}
+		// The test encoder must agree with the pinned literal, or the negative
+		// cases below are testing a different wire format.
+		if got := wireCID(CodeCard, 0, cardDigest(tc.source, tc.id)); got != tc.cid {
+			t.Errorf("wireCID(%q, %q) = %s, want %s", tc.source, tc.id, got, tc.cid)
+		}
+	}
+}
+
+func TestDecodeCard_RejectsEverythingElse(t *testing.T) {
+	for name, s := range map[string]string{
+		"url locator codec":      wireCID(CodeURL, 0, cardDigest("anilist", "1")),
+		"non-identity multihash": wireCID(CodeCard, CodeSha256, cardDigest("anilist", "1")),
+		"empty id":               wireCID(CodeCard, 0, cardDigest("anilist", "")),
+		"source past digest":     wireCID(CodeCard, 0, []byte{0x20, 'a'}),
+		"not a CID":              "title:berserk",
+		"CIDv0":                  "QmYwAPJzv5CZsnA625s3Xf2nemtYgPpHdWEz79ojWnPbdG",
+		"truncated":              goldenAnilistCard[:12],
+		"empty":                  "",
+	} {
+		if source, id, err := DecodeCard(s); err == nil {
+			t.Errorf("%s: DecodeCard(%q) = (%q, %q), want error", name, s, source, id)
+		}
+	}
+}