syntax: basic CID (string only) syntax

atproto/syntax/cid.go

···
+package syntax
+
+import (
+	"fmt"
+	"regexp"
+	"strings"
+)
+
+// Represents a CIDv1 in string format, as would pass Lexicon syntax validation.
+//
+// You usually want to use the github.com/ipfs/go-cid package and type when working with CIDs ("Links") in atproto. This specific type (syntax.CID) is an informal/incomplete helper specifically for doing fast string verification or pass-through without parsing, re-serialization, or normalization.
+//
+// Always use [ParseCID] instead of wrapping strings directly, especially when working with network input.
+type CID string
+
+func ParseCID(raw string) (CID, error) {
+	if len(raw) > 256 {
+		return "", fmt.Errorf("CID is too long (256 chars max)")
+	}
+	if len(raw) < 8 {
+		return "", fmt.Errorf("CID is too short (8 chars min)")
+	}
+	var cidRegex = regexp.MustCompile(`^[a-zA-Z0-9+=]{8,256}$`)
+	if !cidRegex.MatchString(raw) {
+		return "", fmt.Errorf("CID syntax didn't validate via regex")
+	}
+	if strings.HasPrefix(raw, "Qmb") {
+		return "", fmt.Errorf("CIDv0 not allowed in this version of atproto")
+	}
+	return CID(raw), nil
+}
+
+func (c CID) String() string {
+	return string(c)
+}
+
+func (c CID) MarshalText() ([]byte, error) {
+	return []byte(c.String()), nil
+}
+
+func (c *CID) UnmarshalText(text []byte) error {
+	cid, err := ParseCID(string(text))
+	if err != nil {
+		return err
+	}
+	*c = cid
+	return nil
+}

atproto/syntax/cid_test.go

···
+package syntax
+
+import (
+	"bufio"
+	"fmt"
+	"os"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestInteropCIDsValid(t *testing.T) {
+	assert := assert.New(t)
+	file, err := os.Open("testdata/cid_syntax_valid.txt")
+	assert.NoError(err)
+	defer file.Close()
+	scanner := bufio.NewScanner(file)
+	for scanner.Scan() {
+		line := scanner.Text()
+		if len(line) == 0 || line[0] == '#' {
+			continue
+		}
+		_, err := ParseCID(line)
+		if err != nil {
+			fmt.Println("GOOD: " + line)
+		}
+		assert.NoError(err)
+	}
+	assert.NoError(scanner.Err())
+}
+
+func TestInteropCIDsInvalid(t *testing.T) {
+	assert := assert.New(t)
+	file, err := os.Open("testdata/cid_syntax_invalid.txt")
+	assert.NoError(err)
+	defer file.Close()
+	scanner := bufio.NewScanner(file)
+	for scanner.Scan() {
+		line := scanner.Text()
+		if len(line) == 0 || line[0] == '#' {
+			continue
+		}
+		_, err := ParseCID(line)
+		if err == nil {
+			fmt.Println("BAD: " + line)
+		}
+		assert.Error(err)
+	}
+	assert.NoError(scanner.Err())
+}

atproto/syntax/testdata/cid_syntax_invalid.txt

···
+example.com
+https://example.com
+cid:bafybeigdyrzt5sfp7udm7hu76uh7y26nf3efuylqabf3oclgtqy55fbzdi
+.
+12345
+
+# whitespace
+ bafybeigdyrzt5sfp7udm7hu76uh7y26nf3efuylqabf3oclgtqy55fbzdi
+bafybeigdyrzt5sfp7udm7hu76uh7y26nf3efuylqabf3oclgtqy55fbzdi 
+bafybe igdyrzt5sfp7udm7hu76uh7y26nf3efuylqabf3oclgtqy55fbzdi
+
+# old CIDv0 not supported
+QmbWqxBEKC3P8tqsKc98xmWNzrzDtRLMiMPL8wBuTGsMnR
+
+# https://github.com/ipfs-shipyard/is-ipfs/blob/master/test/test-cid.spec.ts
+noop

atproto/syntax/testdata/cid_syntax_valid.txt

···
+
+# examples from https://docs.ipfs.tech/concepts/content-addressing
+bafybeigdyrzt5sfp7udm7hu76uh7y26nf3efuylqabf3oclgtqy55fbzdi
+
+# https://github.com/ipfs-shipyard/is-ipfs/blob/master/test/test-cid.spec.ts
+zdj7WWeQ43G6JJvLWQWZpyHuAMq6uYWRjkBXFad11vE2LHhQ7
+bafybeie5gq4jxvzmsym6hjlwxej4rwdoxt7wadqvmmwbqi7r27fclha2va
+
+# more contrived examples
+mBcDxtdWx0aWhhc2g+
+z7x3CtScH765HvShXT
+zdj7WhuEjrB52m1BisYCtmjH1hSKa7yZ3jEZ9JcXaFRD51wVz
+7134036155352661643226414134664076
+f017012202c5f688262e0ece8569aa6f94d60aad55ca8d9d83734e4a7430d0cff6588ec2b