fix: exclude known bad instances

these aren't pds or labeler, please stop inserting them!

scripts/export-dids.ts

@@ -14,7 +14,7 @@ import {
 	type SerializedState,
 } from '../src/state';
 
-import { DEFAULT_HEADERS, MAX_FAILURE_DAYS, PLC_URL, RELAY_URL } from '../src/constants';
+import { DEFAULT_HEADERS, EXCLUSIONS_RE, MAX_FAILURE_DAYS, PLC_URL, RELAY_URL } from '../src/constants';
 import { didDocument, type DidDocument } from '../src/utils/did';
 import { PromiseQueue } from '../src/utils/pqueue';
 import { LineBreakStream, TextDecoderStream } from '../src/utils/stream';
@@ -74,7 +74,12 @@ let firehoseCursor: string | undefined = state?.firehose.cursor;
 				const pds = getEndpoint(operation.services.atproto_pds?.endpoint);
 				const labeler = getEndpoint(operation.services.atproto_labeler?.endpoint);
 
-				if (pds) {
+				jump: if (pds) {
+					if (EXCLUSIONS_RE.test(pds)) {
+						console.log(`  found excluded pds: ${pds}`);
+						break jump;
+					}
+
 					const info = pdses.get(pds);
 
 					if (info === undefined) {
@@ -87,7 +92,12 @@ let firehoseCursor: string | undefined = state?.firehose.cursor;
 					}
 				}
 
-				if (labeler) {
+				jump: if (labeler) {
+					if (EXCLUSIONS_RE.test(labeler)) {
+						console.log(`  found excluded labeler: ${labeler}`);
+						break jump;
+					}
+
 					const info = labelers.get(labeler);
 
 					if (info === undefined) {
@@ -247,7 +257,12 @@ let firehoseCursor: string | undefined = state?.firehose.cursor;
 
 						console.log(`  ${did}: pass (updated)`);
 
-						if (pds) {
+						jump: if (pds) {
+							if (EXCLUSIONS_RE.test(pds)) {
+								console.log(`  found excluded pds: ${pds}`);
+								break jump;
+							}
+
 							const info = pdses.get(pds);
 
 							if (info === undefined) {
@@ -260,7 +275,12 @@ let firehoseCursor: string | undefined = state?.firehose.cursor;
 							}
 						}
 
-						if (labeler) {
+						jump: if (labeler) {
+							if (EXCLUSIONS_RE.test(labeler)) {
+								console.log(`  found excluded labeler: ${labeler}`);
+								break jump;
+							}
+
 							const info = labelers.get(labeler);
 
 							if (info === undefined) {

src/constants.ts

@@ -8,3 +8,7 @@ export const USER_AGENT = 'github:mary-ext/atproto-scraping';
 export const DEFAULT_HEADERS = {
 	'user-agent': USER_AGENT,
 };
+
+// None of these are either a personal data server or labeler instance.
+// - bsky.social is an entryway, not the actual PDS.
+export const EXCLUSIONS_RE = /^https?:\/\/(?:bsky\.social|bsky\.network|jetstream\d+\.[a-z-]+\.bsky\.network)$/;

state.json

@@ -401,14 +401,6 @@
 			"did": "did:plc:skibpmllbhxvbvwgtjxl3uao",
 			"version": "0.1.38"
 		},
-		"https://jetstream1.us-east.bsky.network/": {
-			"did": "did:plc:3hzvshto76gqe5up6gvs7fwq",
-			"errorAt": 1731892814312
-		},
-		"https://jetstream1.us-west.bsky.network/": {
-			"did": "did:plc:4j6e6lhmrjsihjtshsuh3chk",
-			"errorAt": 1731719983905
-		},
 		"https://jntestlabeler.jamienemeth.co.uk/": {
 			"did": "did:plc:x7d5qy5kmr34qd4w7ji4cbr4",
 			"version": null
@@ -3570,9 +3562,6 @@
 			"inviteCodeRequired": true,
 			"version": "0.4.67"
 		},
-		"https://bsky.network/": {
-			"errorAt": 1732246709198
-		},
 		"https://bsky.newhimazin.net/": {
 			"errorAt": 1731806551573
 		},
@@ -3860,10 +3849,6 @@
 			"inviteCodeRequired": true,
 			"version": "0.4.67"
 		},
-		"https://bsky.social/": {
-			"inviteCodeRequired": false,
-			"version": "git-7edd51a"
-		},
 		"https://bsky.soopy.moe/": {
 			"inviteCodeRequired": true,
 			"version": "0.4.67"