leaflet.pub
a tool for shared writing and social publishing
de-duplicate publications and documents
awarm.space
+122
src/utils/deduplicateRecords.ts
+122
src/utils/deduplicateRecords.ts
···
···
1
+
/**
2
+
* Utilities for deduplicating records that may exist under both
3
+
* pub.leaflet.* and site.standard.* namespaces.
4
+
*
5
+
* After the migration to site.standard.*, records can exist in both namespaces
6
+
* with the same DID and rkey. This utility deduplicates them, preferring
7
+
* site.standard.* records when available.
8
+
*/
9
+
10
+
import { AtUri } from "@atproto/syntax";
11
+
12
+
/**
13
+
* Extracts the identity key (DID + rkey) from an AT URI.
14
+
* This key uniquely identifies a record across namespaces.
15
+
*
16
+
* @example
17
+
* getRecordIdentityKey("at://did:plc:abc/pub.leaflet.document/3abc")
18
+
* // Returns: "did:plc:abc/3abc"
19
+
*
20
+
* getRecordIdentityKey("at://did:plc:abc/site.standard.document/3abc")
21
+
* // Returns: "did:plc:abc/3abc" (same key, different namespace)
22
+
*/
23
+
function getRecordIdentityKey(uri: string): string | null {
24
+
try {
25
+
const parsed = new AtUri(uri);
26
+
return `${parsed.host}/${parsed.rkey}`;
27
+
} catch {
28
+
return null;
29
+
}
30
+
}
31
+
32
+
/**
33
+
* Checks if a URI is from the site.standard namespace.
34
+
*/
35
+
function isSiteStandardUri(uri: string): boolean {
36
+
return uri.includes("/site.standard.");
37
+
}
38
+
39
+
/**
40
+
* Deduplicates an array of records that have a `uri` property.
41
+
*
42
+
* When records exist under both pub.leaflet.* and site.standard.* namespaces
43
+
* (same DID and rkey), this function keeps only the site.standard version.
44
+
*
45
+
* @param records - Array of records with a `uri` property
46
+
* @returns Deduplicated array, preferring site.standard records
47
+
*
48
+
* @example
49
+
* const docs = [
50
+
* { uri: "at://did:plc:abc/pub.leaflet.document/3abc", data: {...} },
51
+
* { uri: "at://did:plc:abc/site.standard.document/3abc", data: {...} },
52
+
* { uri: "at://did:plc:abc/pub.leaflet.document/3def", data: {...} },
53
+
* ];
54
+
* const deduped = deduplicateByUri(docs);
55
+
* // Returns: [
56
+
* // { uri: "at://did:plc:abc/site.standard.document/3abc", data: {...} },
57
+
* // { uri: "at://did:plc:abc/pub.leaflet.document/3def", data: {...} },
58
+
* // ]
59
+
*/
60
+
export function deduplicateByUri<T extends { uri: string }>(records: T[]): T[] {
61
+
const recordsByKey = new Map<string, T>();
62
+
63
+
for (const record of records) {
64
+
const key = getRecordIdentityKey(record.uri);
65
+
if (!key) {
66
+
// Invalid URI, keep the record as-is
67
+
continue;
68
+
}
69
+
70
+
const existing = recordsByKey.get(key);
71
+
if (!existing) {
72
+
recordsByKey.set(key, record);
73
+
} else {
74
+
// Prefer site.standard records over pub.leaflet records
75
+
if (isSiteStandardUri(record.uri) && !isSiteStandardUri(existing.uri)) {
76
+
recordsByKey.set(key, record);
77
+
}
78
+
// If both are same namespace or existing is already site.standard, keep existing
79
+
}
80
+
}
81
+
82
+
return Array.from(recordsByKey.values());
83
+
}
84
+
85
+
/**
86
+
* Deduplicates records while preserving the original order based on the first
87
+
* occurrence of each unique record.
88
+
*
89
+
* Same deduplication logic as deduplicateByUri, but maintains insertion order.
90
+
*
91
+
* @param records - Array of records with a `uri` property
92
+
* @returns Deduplicated array in original order, preferring site.standard records
93
+
*/
94
+
export function deduplicateByUriOrdered<T extends { uri: string }>(
95
+
records: T[]
96
+
): T[] {
97
+
const recordsByKey = new Map<string, { record: T; index: number }>();
98
+
99
+
for (let i = 0; i < records.length; i++) {
100
+
const record = records[i];
101
+
const key = getRecordIdentityKey(record.uri);
102
+
if (!key) {
103
+
continue;
104
+
}
105
+
106
+
const existing = recordsByKey.get(key);
107
+
if (!existing) {
108
+
recordsByKey.set(key, { record, index: i });
109
+
} else {
110
+
// Prefer site.standard records over pub.leaflet records
111
+
if (isSiteStandardUri(record.uri) && !isSiteStandardUri(existing.record.uri)) {
112
+
// Replace with site.standard but keep original position
113
+
recordsByKey.set(key, { record, index: existing.index });
114
+
}
115
+
}
116
+
}
117
+
118
+
// Sort by original index to maintain order
119
+
return Array.from(recordsByKey.values())
120
+
.sort((a, b) => a.index - b.index)
121
+
.map((entry) => entry.record);
122
+
}