emdash-plugin-relink 0.2.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +32 -0
- package/CONTRIBUTING.md +21 -0
- package/LICENSE +167 -0
- package/LICENSES/MIT-legacy.txt +21 -0
- package/README.md +95 -0
- package/THIRD_PARTY_NOTICES.md +19 -0
- package/dist/admin/app.d.ts +9 -0
- package/dist/admin/app.d.ts.map +1 -0
- package/dist/admin/app.js +333 -0
- package/dist/admin/app.js.map +1 -0
- package/dist/admin/client.d.ts +9 -0
- package/dist/admin/client.d.ts.map +1 -0
- package/dist/admin/client.js +22 -0
- package/dist/admin/client.js.map +1 -0
- package/dist/admin/i18n.d.ts +8 -0
- package/dist/admin/i18n.d.ts.map +1 -0
- package/dist/admin/i18n.js +35 -0
- package/dist/admin/i18n.js.map +1 -0
- package/dist/admin/index.d.ts +18 -0
- package/dist/admin/index.d.ts.map +1 -0
- package/dist/admin/index.js +22 -0
- package/dist/admin/index.js.map +1 -0
- package/dist/admin/messages.d.ts +270 -0
- package/dist/admin/messages.d.ts.map +1 -0
- package/dist/admin/messages.js +269 -0
- package/dist/admin/messages.js.map +1 -0
- package/dist/admin/styles.css +330 -0
- package/dist/core/api.d.ts +559 -0
- package/dist/core/api.d.ts.map +1 -0
- package/dist/core/api.js +76 -0
- package/dist/core/api.js.map +1 -0
- package/dist/core/engine.d.ts +30 -0
- package/dist/core/engine.d.ts.map +1 -0
- package/dist/core/engine.js +392 -0
- package/dist/core/engine.js.map +1 -0
- package/dist/core/extract.d.ts +11 -0
- package/dist/core/extract.d.ts.map +1 -0
- package/dist/core/extract.js +99 -0
- package/dist/core/extract.js.map +1 -0
- package/dist/core/html.d.ts +12 -0
- package/dist/core/html.d.ts.map +1 -0
- package/dist/core/html.js +103 -0
- package/dist/core/html.js.map +1 -0
- package/dist/core/http.d.ts +22 -0
- package/dist/core/http.d.ts.map +1 -0
- package/dist/core/http.js +217 -0
- package/dist/core/http.js.map +1 -0
- package/dist/core/index.d.ts +10 -0
- package/dist/core/index.d.ts.map +1 -0
- package/dist/core/index.js +10 -0
- package/dist/core/index.js.map +1 -0
- package/dist/core/paths.d.ts +4 -0
- package/dist/core/paths.d.ts.map +1 -0
- package/dist/core/paths.js +38 -0
- package/dist/core/paths.js.map +1 -0
- package/dist/core/providers.d.ts +32 -0
- package/dist/core/providers.d.ts.map +1 -0
- package/dist/core/providers.js +198 -0
- package/dist/core/providers.js.map +1 -0
- package/dist/core/schema.d.ts +526 -0
- package/dist/core/schema.d.ts.map +1 -0
- package/dist/core/schema.js +238 -0
- package/dist/core/schema.js.map +1 -0
- package/dist/core/storage.d.ts +47 -0
- package/dist/core/storage.d.ts.map +1 -0
- package/dist/core/storage.js +89 -0
- package/dist/core/storage.js.map +1 -0
- package/dist/core/url.d.ts +7 -0
- package/dist/core/url.d.ts.map +1 -0
- package/dist/core/url.js +62 -0
- package/dist/core/url.js.map +1 -0
- package/dist/index.d.ts +8 -0
- package/dist/index.d.ts.map +1 -0
- package/dist/index.js +16 -0
- package/dist/index.js.map +1 -0
- package/dist/integration.d.ts +5 -0
- package/dist/integration.d.ts.map +1 -0
- package/dist/integration.js +29 -0
- package/dist/integration.js.map +1 -0
- package/dist/metadata.d.ts +14 -0
- package/dist/metadata.d.ts.map +1 -0
- package/dist/metadata.js +6 -0
- package/dist/metadata.js.map +1 -0
- package/dist/middleware.d.ts +8 -0
- package/dist/middleware.d.ts.map +1 -0
- package/dist/middleware.js +97 -0
- package/dist/middleware.js.map +1 -0
- package/dist/plugin.d.ts +13 -0
- package/dist/plugin.d.ts.map +1 -0
- package/dist/plugin.js +260 -0
- package/dist/plugin.js.map +1 -0
- package/docs/architecture.md +50 -0
- package/docs/configuration.md +37 -0
- package/docs/installation.md +71 -0
- package/docs/licensing.md +42 -0
- package/docs/recovery.md +27 -0
- package/docs/testing.md +31 -0
- package/docs/translations.md +24 -0
- package/package.json +124 -0
|
@@ -0,0 +1,392 @@
|
|
|
1
|
+
import { extractOccurrences } from "./extract.js";
|
|
2
|
+
import { LinkChecker } from "./http.js";
|
|
3
|
+
import { GrokipediaProvider, WaybackProvider } from "./providers.js";
|
|
4
|
+
import { DAY, HOUR, assertNever } from "./schema.js";
|
|
5
|
+
import { checkUrl, domainExcluded } from "./url.js";
|
|
6
|
+
export function newLink(id, url, now) {
|
|
7
|
+
return {
|
|
8
|
+
schemaVersion: 1,
|
|
9
|
+
id,
|
|
10
|
+
url,
|
|
11
|
+
host: new URL(url).hostname,
|
|
12
|
+
createdAt: now,
|
|
13
|
+
updatedAt: now,
|
|
14
|
+
nextCheckAt: now,
|
|
15
|
+
firstFailureAt: null,
|
|
16
|
+
lastFailureAt: null,
|
|
17
|
+
failures: 0,
|
|
18
|
+
recoveryAt: null,
|
|
19
|
+
lastCheck: null,
|
|
20
|
+
replacement: null,
|
|
21
|
+
review: null,
|
|
22
|
+
excluded: false,
|
|
23
|
+
automaticDisabled: false,
|
|
24
|
+
active: true,
|
|
25
|
+
archiveRetryAt: null,
|
|
26
|
+
grokipediaChecked: false,
|
|
27
|
+
};
|
|
28
|
+
}
|
|
29
|
+
/** Invoke tick only through EmDash's single, atomically claimed recurring job. */
|
|
30
|
+
export class RelinkEngine {
|
|
31
|
+
dependencies;
|
|
32
|
+
clock;
|
|
33
|
+
constructor(dependencies) {
|
|
34
|
+
this.dependencies = dependencies;
|
|
35
|
+
this.clock = dependencies.clock ?? Date.now;
|
|
36
|
+
}
|
|
37
|
+
get store() {
|
|
38
|
+
return this.dependencies.store;
|
|
39
|
+
}
|
|
40
|
+
async tick() {
|
|
41
|
+
const state = await this.store.state();
|
|
42
|
+
state.lastStartedAt = this.clock();
|
|
43
|
+
state.lastError = null;
|
|
44
|
+
await this.store.saveState(state);
|
|
45
|
+
try {
|
|
46
|
+
await this.processCommands(state);
|
|
47
|
+
const settings = await this.store.settings();
|
|
48
|
+
if (settings.enabled) {
|
|
49
|
+
await this.scanBatch(state);
|
|
50
|
+
const checker = new LinkChecker(this.dependencies.transport, settings, this.clock);
|
|
51
|
+
const due = await this.store.links.page({
|
|
52
|
+
where: { active: true, nextCheckAt: { lte: this.clock() } },
|
|
53
|
+
orderBy: { nextCheckAt: "asc" },
|
|
54
|
+
limit: 100,
|
|
55
|
+
});
|
|
56
|
+
// A bounded tick stays below EmDash's ten-minute stale-claim window.
|
|
57
|
+
const groups = new Map();
|
|
58
|
+
for (const link of due.items)
|
|
59
|
+
if (!groups.has(link.host) && groups.size < settings.batchSize)
|
|
60
|
+
groups.set(link.host, link);
|
|
61
|
+
await Promise.all([...groups.values()].map(async (link) => {
|
|
62
|
+
await this.checkLink(link, settings, checker);
|
|
63
|
+
state.processed++;
|
|
64
|
+
}));
|
|
65
|
+
}
|
|
66
|
+
state.lastCompletedAt = this.clock();
|
|
67
|
+
await this.store.saveState(state);
|
|
68
|
+
}
|
|
69
|
+
catch (error) {
|
|
70
|
+
state.lastError = "storage";
|
|
71
|
+
await this.store.saveState(state);
|
|
72
|
+
throw error;
|
|
73
|
+
}
|
|
74
|
+
}
|
|
75
|
+
async event(link, event, id) {
|
|
76
|
+
await this.store.history.put(id, {
|
|
77
|
+
schemaVersion: 1,
|
|
78
|
+
id,
|
|
79
|
+
linkId: link.id,
|
|
80
|
+
originalUrl: link.url,
|
|
81
|
+
at: this.clock(),
|
|
82
|
+
event,
|
|
83
|
+
result: link.lastCheck,
|
|
84
|
+
destination: link.replacement?.url ?? null,
|
|
85
|
+
replacement: link.replacement,
|
|
86
|
+
});
|
|
87
|
+
}
|
|
88
|
+
async processCommands(state) {
|
|
89
|
+
const commands = await this.store.commands.page({
|
|
90
|
+
orderBy: { createdAt: "asc" },
|
|
91
|
+
limit: 4,
|
|
92
|
+
});
|
|
93
|
+
for (const command of commands.items) {
|
|
94
|
+
await this.apply(command, state);
|
|
95
|
+
// Persist the scan request before acknowledging its command.
|
|
96
|
+
await this.store.saveState(state);
|
|
97
|
+
await this.store.commands.delete(command.id);
|
|
98
|
+
}
|
|
99
|
+
}
|
|
100
|
+
async apply(command, state) {
|
|
101
|
+
const input = command.input;
|
|
102
|
+
if (input.action === "scan") {
|
|
103
|
+
state.nextScanAt = 0;
|
|
104
|
+
return;
|
|
105
|
+
}
|
|
106
|
+
if (input.action === "settings") {
|
|
107
|
+
await this.store.saveSettings(input.settings);
|
|
108
|
+
return;
|
|
109
|
+
}
|
|
110
|
+
const link = await this.store.links.get(input.linkId);
|
|
111
|
+
if (!link)
|
|
112
|
+
return;
|
|
113
|
+
let event = null;
|
|
114
|
+
switch (input.action) {
|
|
115
|
+
case "check":
|
|
116
|
+
link.nextCheckAt = Math.min(link.nextCheckAt, command.createdAt);
|
|
117
|
+
link.grokipediaChecked = false;
|
|
118
|
+
break;
|
|
119
|
+
case "exclude":
|
|
120
|
+
link.excluded = input.value;
|
|
121
|
+
event = input.value ? "excluded" : "included";
|
|
122
|
+
link.nextCheckAt = command.createdAt;
|
|
123
|
+
break;
|
|
124
|
+
case "undo":
|
|
125
|
+
link.replacement = null;
|
|
126
|
+
link.review = null;
|
|
127
|
+
link.automaticDisabled = true;
|
|
128
|
+
event = "undo";
|
|
129
|
+
break;
|
|
130
|
+
case "reschedule":
|
|
131
|
+
link.review = { dueAt: input.dueAt, reason: "manual-review" };
|
|
132
|
+
link.nextCheckAt = input.dueAt;
|
|
133
|
+
event = "rescheduled";
|
|
134
|
+
break;
|
|
135
|
+
case "destination": {
|
|
136
|
+
const checker = new LinkChecker(this.dependencies.transport, await this.store.settings(), this.clock);
|
|
137
|
+
const result = await checker.check(input.url);
|
|
138
|
+
if (result.kind === "healthy") {
|
|
139
|
+
link.replacement = {
|
|
140
|
+
kind: "manual",
|
|
141
|
+
url: input.url,
|
|
142
|
+
verifiedAt: this.clock(),
|
|
143
|
+
reason: "editor-override",
|
|
144
|
+
};
|
|
145
|
+
link.automaticDisabled = true;
|
|
146
|
+
event = "override";
|
|
147
|
+
}
|
|
148
|
+
else {
|
|
149
|
+
link.review = { dueAt: this.clock(), reason: "replacement-failed" };
|
|
150
|
+
event = "replacement-unavailable";
|
|
151
|
+
}
|
|
152
|
+
break;
|
|
153
|
+
}
|
|
154
|
+
default:
|
|
155
|
+
assertNever(input);
|
|
156
|
+
}
|
|
157
|
+
link.updatedAt = this.clock();
|
|
158
|
+
await this.store.links.put(link.id, link);
|
|
159
|
+
if (event)
|
|
160
|
+
await this.event(link, event, command.id);
|
|
161
|
+
}
|
|
162
|
+
async scanBatch(state) {
|
|
163
|
+
if (!state.scan && state.nextScanAt <= this.clock())
|
|
164
|
+
state.scan = {
|
|
165
|
+
sourceIndex: 0,
|
|
166
|
+
cursor: null,
|
|
167
|
+
generation: crypto.randomUUID(),
|
|
168
|
+
startedAt: this.clock(),
|
|
169
|
+
};
|
|
170
|
+
const scan = state.scan;
|
|
171
|
+
if (!scan)
|
|
172
|
+
return;
|
|
173
|
+
const source = this.dependencies.options.sources[scan.sourceIndex];
|
|
174
|
+
if (!source) {
|
|
175
|
+
state.scan = null;
|
|
176
|
+
state.nextScanAt = this.clock() + DAY;
|
|
177
|
+
return;
|
|
178
|
+
}
|
|
179
|
+
const page = await this.dependencies.content.list(source.collection, {
|
|
180
|
+
where: { status: "published" },
|
|
181
|
+
limit: 10,
|
|
182
|
+
...(scan.cursor ? { cursor: scan.cursor } : {}),
|
|
183
|
+
});
|
|
184
|
+
for (const content of page.items) {
|
|
185
|
+
if (content.status !== "published")
|
|
186
|
+
continue;
|
|
187
|
+
const occurrences = await extractOccurrences({
|
|
188
|
+
id: content.id,
|
|
189
|
+
slug: content.slug ?? content.id,
|
|
190
|
+
locale: content.locale ?? "en",
|
|
191
|
+
title: typeof content.data["title"] === "string"
|
|
192
|
+
? content.data["title"]
|
|
193
|
+
: (content.slug ?? content.id),
|
|
194
|
+
data: content.data,
|
|
195
|
+
}, source, this.dependencies.options.siteUrl, scan.generation);
|
|
196
|
+
for (const occurrence of occurrences) {
|
|
197
|
+
const existing = await this.store.links.get(occurrence.linkId);
|
|
198
|
+
const link = existing ??
|
|
199
|
+
newLink(occurrence.linkId, checkUrl(occurrence.originalUrl), this.clock());
|
|
200
|
+
if (!link.active)
|
|
201
|
+
link.nextCheckAt = this.clock();
|
|
202
|
+
link.active = true;
|
|
203
|
+
await this.store.links.put(link.id, link);
|
|
204
|
+
await this.store.occurrences.put(occurrence.id, occurrence);
|
|
205
|
+
}
|
|
206
|
+
}
|
|
207
|
+
scan.cursor = page.hasMore ? (page.cursor ?? null) : null;
|
|
208
|
+
if (!scan.cursor) {
|
|
209
|
+
// Only sweep after completing a collection: a failed scan never drops valid occurrences.
|
|
210
|
+
const previous = await this.store.occurrences.all({
|
|
211
|
+
where: { collection: source.collection },
|
|
212
|
+
});
|
|
213
|
+
const touched = new Set();
|
|
214
|
+
for (const occurrence of previous) {
|
|
215
|
+
if (occurrence.generation !== scan.generation) {
|
|
216
|
+
await this.store.occurrences.delete(occurrence.id);
|
|
217
|
+
touched.add(occurrence.linkId);
|
|
218
|
+
}
|
|
219
|
+
}
|
|
220
|
+
for (const id of touched) {
|
|
221
|
+
const remaining = await this.store.occurrences.page({
|
|
222
|
+
where: { linkId: id },
|
|
223
|
+
limit: 1,
|
|
224
|
+
});
|
|
225
|
+
if (remaining.items.length === 0) {
|
|
226
|
+
const link = await this.store.links.get(id);
|
|
227
|
+
if (link) {
|
|
228
|
+
link.active = false;
|
|
229
|
+
await this.store.links.put(id, link);
|
|
230
|
+
}
|
|
231
|
+
}
|
|
232
|
+
}
|
|
233
|
+
scan.sourceIndex++;
|
|
234
|
+
if (scan.sourceIndex >= this.dependencies.options.sources.length) {
|
|
235
|
+
state.scan = null;
|
|
236
|
+
state.nextScanAt = this.clock() + DAY;
|
|
237
|
+
}
|
|
238
|
+
}
|
|
239
|
+
await this.store.saveState(state);
|
|
240
|
+
}
|
|
241
|
+
async checkLink(link, settings, checker) {
|
|
242
|
+
const now = this.clock();
|
|
243
|
+
if (link.excluded || domainExcluded(link.host, settings.excludedDomains)) {
|
|
244
|
+
link.nextCheckAt = now + settings.healthyDays * DAY;
|
|
245
|
+
await this.store.links.put(link.id, link);
|
|
246
|
+
return;
|
|
247
|
+
}
|
|
248
|
+
let result = await checker.check(link.url);
|
|
249
|
+
if (result.kind === "broken" &&
|
|
250
|
+
(result.status === null || result.status >= 500) &&
|
|
251
|
+
!(await checker.networkAvailable(this.dependencies.options.networkProbeUrls)))
|
|
252
|
+
result = {
|
|
253
|
+
kind: "unverifiable",
|
|
254
|
+
checkedAt: now,
|
|
255
|
+
finalUrl: link.url,
|
|
256
|
+
status: null,
|
|
257
|
+
reason: "network-outage",
|
|
258
|
+
retryAt: now + HOUR,
|
|
259
|
+
};
|
|
260
|
+
link.lastCheck = result;
|
|
261
|
+
link.updatedAt = now;
|
|
262
|
+
let event = "checked";
|
|
263
|
+
const occurrences = await this.store.occurrences.all({
|
|
264
|
+
where: { linkId: link.id },
|
|
265
|
+
});
|
|
266
|
+
const fragments = [
|
|
267
|
+
...new Set(occurrences.map((occurrence) => new URL(occurrence.originalUrl).hash)),
|
|
268
|
+
];
|
|
269
|
+
switch (result.kind) {
|
|
270
|
+
case "unverifiable":
|
|
271
|
+
link.nextCheckAt = result.retryAt ?? now + DAY;
|
|
272
|
+
link.review = { dueAt: link.nextCheckAt, reason: "unverifiable" };
|
|
273
|
+
if (link.replacement) {
|
|
274
|
+
const replacement = await checker.check(link.replacement.url);
|
|
275
|
+
if (replacement.kind === "broken") {
|
|
276
|
+
if (link.replacement.kind !== "manual")
|
|
277
|
+
link.replacement = null;
|
|
278
|
+
link.review = { dueAt: now, reason: "replacement-failed" };
|
|
279
|
+
event = "replacement-unavailable";
|
|
280
|
+
}
|
|
281
|
+
}
|
|
282
|
+
break;
|
|
283
|
+
case "healthy": {
|
|
284
|
+
link.failures = 0;
|
|
285
|
+
link.firstFailureAt = null;
|
|
286
|
+
link.lastFailureAt = null;
|
|
287
|
+
link.archiveRetryAt = null;
|
|
288
|
+
link.nextCheckAt = now + settings.healthyDays * DAY;
|
|
289
|
+
if (link.replacement?.kind === "archive") {
|
|
290
|
+
if (link.recoveryAt !== null && now - link.recoveryAt >= DAY) {
|
|
291
|
+
link.replacement = null;
|
|
292
|
+
link.review = null;
|
|
293
|
+
link.recoveryAt = null;
|
|
294
|
+
event = "original-restored";
|
|
295
|
+
}
|
|
296
|
+
else {
|
|
297
|
+
link.recoveryAt ??= now;
|
|
298
|
+
link.nextCheckAt = link.recoveryAt + DAY;
|
|
299
|
+
}
|
|
300
|
+
}
|
|
301
|
+
else if (link.replacement?.kind === "grokipedia") {
|
|
302
|
+
const replacement = await checker.check(link.replacement.url);
|
|
303
|
+
if (replacement.kind !== "healthy") {
|
|
304
|
+
link.replacement = null;
|
|
305
|
+
link.review = null;
|
|
306
|
+
link.grokipediaChecked = true;
|
|
307
|
+
event = "original-restored";
|
|
308
|
+
}
|
|
309
|
+
}
|
|
310
|
+
else if (link.replacement?.kind === "manual") {
|
|
311
|
+
const replacement = await checker.check(link.replacement.url);
|
|
312
|
+
if (replacement.kind !== "healthy")
|
|
313
|
+
link.review = { dueAt: now, reason: "replacement-failed" };
|
|
314
|
+
}
|
|
315
|
+
else if (settings.grokipediaEnabled &&
|
|
316
|
+
!link.automaticDisabled &&
|
|
317
|
+
!link.grokipediaChecked) {
|
|
318
|
+
const provider = this.dependencies.encyclopedia ??
|
|
319
|
+
new GrokipediaProvider(checker, this.clock);
|
|
320
|
+
const discovery = await provider.discover(link, fragments);
|
|
321
|
+
link.grokipediaChecked = discovery.kind !== "unavailable";
|
|
322
|
+
if (discovery.kind === "found") {
|
|
323
|
+
link.replacement = discovery.replacement;
|
|
324
|
+
link.review = null;
|
|
325
|
+
event = "grokipedia-applied";
|
|
326
|
+
}
|
|
327
|
+
else if (discovery.kind === "ambiguous")
|
|
328
|
+
link.review = { dueAt: now, reason: "grokipedia-ambiguous" };
|
|
329
|
+
}
|
|
330
|
+
else if (link.review?.reason === "unverifiable")
|
|
331
|
+
link.review = null;
|
|
332
|
+
break;
|
|
333
|
+
}
|
|
334
|
+
case "broken": {
|
|
335
|
+
link.recoveryAt = null;
|
|
336
|
+
link.firstFailureAt ??= now;
|
|
337
|
+
if (link.lastFailureAt === null || now - link.lastFailureAt >= DAY) {
|
|
338
|
+
link.failures++;
|
|
339
|
+
link.lastFailureAt = now;
|
|
340
|
+
}
|
|
341
|
+
link.nextCheckAt =
|
|
342
|
+
link.failures < 2
|
|
343
|
+
? link.lastFailureAt + DAY
|
|
344
|
+
: link.failures < 3
|
|
345
|
+
? Math.max(now + DAY, link.firstFailureAt + 3 * DAY)
|
|
346
|
+
: now + settings.reviewDays * DAY;
|
|
347
|
+
if (link.replacement) {
|
|
348
|
+
const replacement = await checker.check(link.replacement.url);
|
|
349
|
+
if (replacement.kind === "broken") {
|
|
350
|
+
if (link.replacement.kind !== "manual")
|
|
351
|
+
link.replacement = null;
|
|
352
|
+
link.review = { dueAt: now, reason: "replacement-failed" };
|
|
353
|
+
event = "replacement-unavailable";
|
|
354
|
+
}
|
|
355
|
+
}
|
|
356
|
+
if (link.failures >= 3 &&
|
|
357
|
+
!link.replacement &&
|
|
358
|
+
!link.automaticDisabled &&
|
|
359
|
+
settings.archiveEnabled &&
|
|
360
|
+
(link.archiveRetryAt === null || link.archiveRetryAt <= now)) {
|
|
361
|
+
const provider = this.dependencies.archive ??
|
|
362
|
+
new WaybackProvider(checker, this.clock);
|
|
363
|
+
const discovery = await provider.discover(link, fragments);
|
|
364
|
+
if (discovery.kind === "found") {
|
|
365
|
+
link.replacement = discovery.replacement;
|
|
366
|
+
link.review = {
|
|
367
|
+
dueAt: now + settings.reviewDays * DAY,
|
|
368
|
+
reason: "archive-active",
|
|
369
|
+
};
|
|
370
|
+
event = "archive-applied";
|
|
371
|
+
}
|
|
372
|
+
else {
|
|
373
|
+
const unavailable = discovery.kind === "unavailable";
|
|
374
|
+
link.archiveRetryAt =
|
|
375
|
+
now + (unavailable ? DAY : settings.archiveRetryDays * DAY);
|
|
376
|
+
link.review = {
|
|
377
|
+
dueAt: link.archiveRetryAt,
|
|
378
|
+
reason: unavailable ? "archive-unavailable" : "archive-missing",
|
|
379
|
+
};
|
|
380
|
+
link.nextCheckAt = Math.min(link.nextCheckAt, link.archiveRetryAt);
|
|
381
|
+
}
|
|
382
|
+
}
|
|
383
|
+
break;
|
|
384
|
+
}
|
|
385
|
+
default:
|
|
386
|
+
assertNever(result);
|
|
387
|
+
}
|
|
388
|
+
await this.store.links.put(link.id, link);
|
|
389
|
+
await this.event(link, event, `${link.id}:${now}:${event}`);
|
|
390
|
+
}
|
|
391
|
+
}
|
|
392
|
+
//# sourceMappingURL=engine.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"engine.js","sourceRoot":"","sources":["../../src/core/engine.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,kBAAkB,EAAE,MAAM,cAAc,CAAC;AAClD,OAAO,EAAE,WAAW,EAAE,MAAM,WAAW,CAAC;AAExC,OAAO,EAAE,kBAAkB,EAAE,eAAe,EAAE,MAAM,gBAAgB,CAAC;AAErE,OAAO,EAAE,GAAG,EAAE,IAAI,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AAUrD,OAAO,EAAE,QAAQ,EAAE,cAAc,EAAE,MAAM,UAAU,CAAC;AAEpD,MAAM,UAAU,OAAO,CAAC,EAAU,EAAE,GAAW,EAAE,GAAW;IAC1D,OAAO;QACL,aAAa,EAAE,CAAC;QAChB,EAAE;QACF,GAAG;QACH,IAAI,EAAE,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC,QAAQ;QAC3B,SAAS,EAAE,GAAG;QACd,SAAS,EAAE,GAAG;QACd,WAAW,EAAE,GAAG;QAChB,cAAc,EAAE,IAAI;QACpB,aAAa,EAAE,IAAI;QACnB,QAAQ,EAAE,CAAC;QACX,UAAU,EAAE,IAAI;QAChB,SAAS,EAAE,IAAI;QACf,WAAW,EAAE,IAAI;QACjB,MAAM,EAAE,IAAI;QACZ,QAAQ,EAAE,KAAK;QACf,iBAAiB,EAAE,KAAK;QACxB,MAAM,EAAE,IAAI;QACZ,cAAc,EAAE,IAAI;QACpB,iBAAiB,EAAE,KAAK;KACzB,CAAC;AACJ,CAAC;AAWD,kFAAkF;AAClF,MAAM,OAAO,YAAY;IAEM;IADZ,KAAK,CAAe;IACrC,YAA6B,YAAgC;QAAhC,iBAAY,GAAZ,YAAY,CAAoB;QAC3D,IAAI,CAAC,KAAK,GAAG,YAAY,CAAC,KAAK,IAAI,IAAI,CAAC,GAAG,CAAC;IAC9C,CAAC;IACD,IAAY,KAAK;QACf,OAAO,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC;IACjC,CAAC;IAED,KAAK,CAAC,IAAI;QACR,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;QACvC,KAAK,CAAC,aAAa,GAAG,IAAI,CAAC,KAAK,EAAE,CAAC;QACnC,KAAK,CAAC,SAAS,GAAG,IAAI,CAAC;QACvB,MAAM,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;QAClC,IAAI,CAAC;YACH,MAAM,IAAI,CAAC,eAAe,CAAC,KAAK,CAAC,CAAC;YAClC,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,KAAK,CAAC,QAAQ,EAAE,CAAC;YAC7C,IAAI,QAAQ,CAAC,OAAO,EAAE,CAAC;gBACrB,MAAM,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;gBAC5B,MAAM,OAAO,GAAG,IAAI,WAAW,CAC7B,IAAI,CAAC,YAAY,CAAC,SAAS,EAC3B,QAAQ,EACR,IAAI,CAAC,KAAK,CACX,CAAC;gBACF,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC;oBACtC,KAAK,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,WAAW,EAAE,EAAE,GAAG,EAAE,IAAI,CAAC,KAAK,EAAE,EAAE,EAAE;oBAC3D,OAAO,EAAE,EAAE,WAAW,EAAE,KAAK,EAAE;oBAC/B,KAAK,EAAE,GAAG;iBACX,CAAC,CAAC;gBACH,qEAAqE;gBACrE,MAAM,MAAM,GAAG,IAAI,GAAG,EAAsB,CAAC;gBAC7C,KAAK,MAAM,IAAI,IAAI,GAAG,CAAC,KAAK;oBAC1B,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,MAAM,CAAC,IAAI,GAAG,QAAQ,CAAC,SAAS;wBAC5D,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;gBAChC,MAAM,OAAO,CAAC,GAAG,CACf,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC,GAAG,CAAC,KAAK,EAAE,IAAI,EAAE,EAAE;oBACtC,MAAM,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,QAAQ,EAAE,OAAO,CAAC,CAAC;oBAC9C,KAAK,CAAC,SAAS,EAAE,CAAC;gBACpB,CAAC,CAAC,CACH,CAAC;YACJ,CAAC;YACD,KAAK,CAAC,eAAe,GAAG,IAAI,CAAC,KAAK,EAAE,CAAC;YACrC,MAAM,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;QACpC,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,KAAK,CAAC,SAAS,GAAG,SAAS,CAAC;YAC5B,MAAM,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;YAClC,MAAM,KAAK,CAAC;QACd,CAAC;IACH,CAAC;IAEO,KAAK,CAAC,KAAK,CACjB,IAAgB,EAChB,KAA4B,EAC5B,EAAU;QAEV,MAAM,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,EAAE;YAC/B,aAAa,EAAE,CAAC;YAChB,EAAE;YACF,MAAM,EAAE,IAAI,CAAC,EAAE;YACf,WAAW,EAAE,IAAI,CAAC,GAAG;YACrB,EAAE,EAAE,IAAI,CAAC,KAAK,EAAE;YAChB,KAAK;YACL,MAAM,EAAE,IAAI,CAAC,SAAS;YACtB,WAAW,EAAE,IAAI,CAAC,WAAW,EAAE,GAAG,IAAI,IAAI;YAC1C,WAAW,EAAE,IAAI,CAAC,WAAW;SAC9B,CAAC,CAAC;IACL,CAAC;IAEO,KAAK,CAAC,eAAe,CAAC,KAAkB;QAC9C,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC;YAC9C,OAAO,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE;YAC7B,KAAK,EAAE,CAAC;SACT,CAAC,CAAC;QACH,KAAK,MAAM,OAAO,IAAI,QAAQ,CAAC,KAAK,EAAE,CAAC;YACrC,MAAM,IAAI,CAAC,KAAK,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;YACjC,6DAA6D;YAC7D,MAAM,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;YAClC,MAAM,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,MAAM,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC;QAC/C,CAAC;IACH,CAAC;IAEO,KAAK,CAAC,KAAK,CAAC,OAAgB,EAAE,KAAkB;QACtD,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC;QAC5B,IAAI,KAAK,CAAC,MAAM,KAAK,MAAM,EAAE,CAAC;YAC5B,KAAK,CAAC,UAAU,GAAG,CAAC,CAAC;YACrB,OAAO;QACT,CAAC;QACD,IAAI,KAAK,CAAC,MAAM,KAAK,UAAU,EAAE,CAAC;YAChC,MAAM,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC;YAC9C,OAAO;QACT,CAAC;QACD,MAAM,IAAI,GAAG,MAAM,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;QACtD,IAAI,CAAC,IAAI;YAAE,OAAO;QAClB,IAAI,KAAK,GAAiC,IAAI,CAAC;QAC/C,QAAQ,KAAK,CAAC,MAAM,EAAE,CAAC;YACrB,KAAK,OAAO;gBACV,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,EAAE,OAAO,CAAC,SAAS,CAAC,CAAC;gBACjE,IAAI,CAAC,iBAAiB,GAAG,KAAK,CAAC;gBAC/B,MAAM;YACR,KAAK,SAAS;gBACZ,IAAI,CAAC,QAAQ,GAAG,KAAK,CAAC,KAAK,CAAC;gBAC5B,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,UAAU,CAAC;gBAC9C,IAAI,CAAC,WAAW,GAAG,OAAO,CAAC,SAAS,CAAC;gBACrC,MAAM;YACR,KAAK,MAAM;gBACT,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC;gBACxB,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;gBACnB,IAAI,CAAC,iBAAiB,GAAG,IAAI,CAAC;gBAC9B,KAAK,GAAG,MAAM,CAAC;gBACf,MAAM;YACR,KAAK,YAAY;gBACf,IAAI,CAAC,MAAM,GAAG,EAAE,KAAK,EAAE,KAAK,CAAC,KAAK,EAAE,MAAM,EAAE,eAAe,EAAE,CAAC;gBAC9D,IAAI,CAAC,WAAW,GAAG,KAAK,CAAC,KAAK,CAAC;gBAC/B,KAAK,GAAG,aAAa,CAAC;gBACtB,MAAM;YACR,KAAK,aAAa,CAAC,CAAC,CAAC;gBACnB,MAAM,OAAO,GAAG,IAAI,WAAW,CAC7B,IAAI,CAAC,YAAY,CAAC,SAAS,EAC3B,MAAM,IAAI,CAAC,KAAK,CAAC,QAAQ,EAAE,EAC3B,IAAI,CAAC,KAAK,CACX,CAAC;gBACF,MAAM,MAAM,GAAG,MAAM,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;gBAC9C,IAAI,MAAM,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;oBAC9B,IAAI,CAAC,WAAW,GAAG;wBACjB,IAAI,EAAE,QAAQ;wBACd,GAAG,EAAE,KAAK,CAAC,GAAG;wBACd,UAAU,EAAE,IAAI,CAAC,KAAK,EAAE;wBACxB,MAAM,EAAE,iBAAiB;qBAC1B,CAAC;oBACF,IAAI,CAAC,iBAAiB,GAAG,IAAI,CAAC;oBAC9B,KAAK,GAAG,UAAU,CAAC;gBACrB,CAAC;qBAAM,CAAC;oBACN,IAAI,CAAC,MAAM,GAAG,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,EAAE,EAAE,MAAM,EAAE,oBAAoB,EAAE,CAAC;oBACpE,KAAK,GAAG,yBAAyB,CAAC;gBACpC,CAAC;gBACD,MAAM;YACR,CAAC;YACD;gBACE,WAAW,CAAC,KAAK,CAAC,CAAC;QACvB,CAAC;QACD,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC,KAAK,EAAE,CAAC;QAC9B,MAAM,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;QAC1C,IAAI,KAAK;YAAE,MAAM,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,EAAE,OAAO,CAAC,EAAE,CAAC,CAAC;IACvD,CAAC;IAEO,KAAK,CAAC,SAAS,CAAC,KAAkB;QACxC,IAAI,CAAC,KAAK,CAAC,IAAI,IAAI,KAAK,CAAC,UAAU,IAAI,IAAI,CAAC,KAAK,EAAE;YACjD,KAAK,CAAC,IAAI,GAAG;gBACX,WAAW,EAAE,CAAC;gBACd,MAAM,EAAE,IAAI;gBACZ,UAAU,EAAE,MAAM,CAAC,UAAU,EAAE;gBAC/B,SAAS,EAAE,IAAI,CAAC,KAAK,EAAE;aACxB,CAAC;QACJ,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC;QACxB,IAAI,CAAC,IAAI;YAAE,OAAO;QAClB,MAAM,MAAM,GAAG,IAAI,CAAC,YAAY,CAAC,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;QACnE,IAAI,CAAC,MAAM,EAAE,CAAC;YACZ,KAAK,CAAC,IAAI,GAAG,IAAI,CAAC;YAClB,KAAK,CAAC,UAAU,GAAG,IAAI,CAAC,KAAK,EAAE,GAAG,GAAG,CAAC;YACtC,OAAO;QACT,CAAC;QACD,MAAM,IAAI,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,UAAU,EAAE;YACnE,KAAK,EAAE,EAAE,MAAM,EAAE,WAAW,EAAE;YAC9B,KAAK,EAAE,EAAE;YACT,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;SAChD,CAAC,CAAC;QACH,KAAK,MAAM,OAAO,IAAI,IAAI,CAAC,KAAK,EAAE,CAAC;YACjC,IAAI,OAAO,CAAC,MAAM,KAAK,WAAW;gBAAE,SAAS;YAC7C,MAAM,WAAW,GAAG,MAAM,kBAAkB,CAC1C;gBACE,EAAE,EAAE,OAAO,CAAC,EAAE;gBACd,IAAI,EAAE,OAAO,CAAC,IAAI,IAAI,OAAO,CAAC,EAAE;gBAChC,MAAM,EAAE,OAAO,CAAC,MAAM,IAAI,IAAI;gBAC9B,KAAK,EACH,OAAO,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,KAAK,QAAQ;oBACvC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC;oBACvB,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,IAAI,OAAO,CAAC,EAAE,CAAC;gBAClC,IAAI,EAAE,OAAO,CAAC,IAAI;aACnB,EACD,MAAM,EACN,IAAI,CAAC,YAAY,CAAC,OAAO,CAAC,OAAO,EACjC,IAAI,CAAC,UAAU,CAChB,CAAC;YACF,KAAK,MAAM,UAAU,IAAI,WAAW,EAAE,CAAC;gBACrC,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;gBAC/D,MAAM,IAAI,GACR,QAAQ;oBACR,OAAO,CACL,UAAU,CAAC,MAAM,EACjB,QAAQ,CAAC,UAAU,CAAC,WAAW,CAAC,EAChC,IAAI,CAAC,KAAK,EAAE,CACb,CAAC;gBACJ,IAAI,CAAC,IAAI,CAAC,MAAM;oBAAE,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,KAAK,EAAE,CAAC;gBAClD,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;gBACnB,MAAM,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;gBAC1C,MAAM,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,GAAG,CAAC,UAAU,CAAC,EAAE,EAAE,UAAU,CAAC,CAAC;YAC9D,CAAC;QACH,CAAC;QACD,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;QAC1D,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC;YACjB,yFAAyF;YACzF,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,GAAG,CAAC;gBAChD,KAAK,EAAE,EAAE,UAAU,EAAE,MAAM,CAAC,UAAU,EAAE;aACzC,CAAC,CAAC;YACH,MAAM,OAAO,GAAG,IAAI,GAAG,EAAU,CAAC;YAClC,KAAK,MAAM,UAAU,IAAI,QAAQ,EAAE,CAAC;gBAClC,IAAI,UAAU,CAAC,UAAU,KAAK,IAAI,CAAC,UAAU,EAAE,CAAC;oBAC9C,MAAM,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,MAAM,CAAC,UAAU,CAAC,EAAE,CAAC,CAAC;oBACnD,OAAO,CAAC,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;gBACjC,CAAC;YACH,CAAC;YACD,KAAK,MAAM,EAAE,IAAI,OAAO,EAAE,CAAC;gBACzB,MAAM,SAAS,GAAG,MAAM,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,IAAI,CAAC;oBAClD,KAAK,EAAE,EAAE,MAAM,EAAE,EAAE,EAAE;oBACrB,KAAK,EAAE,CAAC;iBACT,CAAC,CAAC;gBACH,IAAI,SAAS,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;oBACjC,MAAM,IAAI,GAAG,MAAM,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;oBAC5C,IAAI,IAAI,EAAE,CAAC;wBACT,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC;wBACpB,MAAM,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;oBACvC,CAAC;gBACH,CAAC;YACH,CAAC;YACD,IAAI,CAAC,WAAW,EAAE,CAAC;YACnB,IAAI,IAAI,CAAC,WAAW,IAAI,IAAI,CAAC,YAAY,CAAC,OAAO,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC;gBACjE,KAAK,CAAC,IAAI,GAAG,IAAI,CAAC;gBAClB,KAAK,CAAC,UAAU,GAAG,IAAI,CAAC,KAAK,EAAE,GAAG,GAAG,CAAC;YACxC,CAAC;QACH,CAAC;QACD,MAAM,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;IACpC,CAAC;IAED,KAAK,CAAC,SAAS,CACb,IAAgB,EAChB,QAAkB,EAClB,OAAoB;QAEpB,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,EAAE,CAAC;QACzB,IAAI,IAAI,CAAC,QAAQ,IAAI,cAAc,CAAC,IAAI,CAAC,IAAI,EAAE,QAAQ,CAAC,eAAe,CAAC,EAAE,CAAC;YACzE,IAAI,CAAC,WAAW,GAAG,GAAG,GAAG,QAAQ,CAAC,WAAW,GAAG,GAAG,CAAC;YACpD,MAAM,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;YAC1C,OAAO;QACT,CAAC;QACD,IAAI,MAAM,GAAG,MAAM,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QAC3C,IACE,MAAM,CAAC,IAAI,KAAK,QAAQ;YACxB,CAAC,MAAM,CAAC,MAAM,KAAK,IAAI,IAAI,MAAM,CAAC,MAAM,IAAI,GAAG,CAAC;YAChD,CAAC,CAAC,MAAM,OAAO,CAAC,gBAAgB,CAC9B,IAAI,CAAC,YAAY,CAAC,OAAO,CAAC,gBAAgB,CAC3C,CAAC;YAEF,MAAM,GAAG;gBACP,IAAI,EAAE,cAAc;gBACpB,SAAS,EAAE,GAAG;gBACd,QAAQ,EAAE,IAAI,CAAC,GAAG;gBAClB,MAAM,EAAE,IAAI;gBACZ,MAAM,EAAE,gBAAgB;gBACxB,OAAO,EAAE,GAAG,GAAG,IAAI;aACpB,CAAC;QACJ,IAAI,CAAC,SAAS,GAAG,MAAM,CAAC;QACxB,IAAI,CAAC,SAAS,GAAG,GAAG,CAAC;QACrB,IAAI,KAAK,GAA0B,SAAS,CAAC;QAC7C,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,GAAG,CAAC;YACnD,KAAK,EAAE,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,EAAE;SAC3B,CAAC,CAAC;QACH,MAAM,SAAS,GAAG;YAChB,GAAG,IAAI,GAAG,CACR,WAAW,CAAC,GAAG,CAAC,CAAC,UAAU,EAAE,EAAE,CAAC,IAAI,GAAG,CAAC,UAAU,CAAC,WAAW,CAAC,CAAC,IAAI,CAAC,CACtE;SACF,CAAC;QACF,QAAQ,MAAM,CAAC,IAAI,EAAE,CAAC;YACpB,KAAK,cAAc;gBACjB,IAAI,CAAC,WAAW,GAAG,MAAM,CAAC,OAAO,IAAI,GAAG,GAAG,GAAG,CAAC;gBAC/C,IAAI,CAAC,MAAM,GAAG,EAAE,KAAK,EAAE,IAAI,CAAC,WAAW,EAAE,MAAM,EAAE,cAAc,EAAE,CAAC;gBAClE,IAAI,IAAI,CAAC,WAAW,EAAE,CAAC;oBACrB,MAAM,WAAW,GAAG,MAAM,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC;oBAC9D,IAAI,WAAW,CAAC,IAAI,KAAK,QAAQ,EAAE,CAAC;wBAClC,IAAI,IAAI,CAAC,WAAW,CAAC,IAAI,KAAK,QAAQ;4BAAE,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC;wBAChE,IAAI,CAAC,MAAM,GAAG,EAAE,KAAK,EAAE,GAAG,EAAE,MAAM,EAAE,oBAAoB,EAAE,CAAC;wBAC3D,KAAK,GAAG,yBAAyB,CAAC;oBACpC,CAAC;gBACH,CAAC;gBACD,MAAM;YACR,KAAK,SAAS,CAAC,CAAC,CAAC;gBACf,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC;gBAClB,IAAI,CAAC,cAAc,GAAG,IAAI,CAAC;gBAC3B,IAAI,CAAC,aAAa,GAAG,IAAI,CAAC;gBAC1B,IAAI,CAAC,cAAc,GAAG,IAAI,CAAC;gBAC3B,IAAI,CAAC,WAAW,GAAG,GAAG,GAAG,QAAQ,CAAC,WAAW,GAAG,GAAG,CAAC;gBACpD,IAAI,IAAI,CAAC,WAAW,EAAE,IAAI,KAAK,SAAS,EAAE,CAAC;oBACzC,IAAI,IAAI,CAAC,UAAU,KAAK,IAAI,IAAI,GAAG,GAAG,IAAI,CAAC,UAAU,IAAI,GAAG,EAAE,CAAC;wBAC7D,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC;wBACxB,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;wBACnB,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC;wBACvB,KAAK,GAAG,mBAAmB,CAAC;oBAC9B,CAAC;yBAAM,CAAC;wBACN,IAAI,CAAC,UAAU,KAAK,GAAG,CAAC;wBACxB,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,UAAU,GAAG,GAAG,CAAC;oBAC3C,CAAC;gBACH,CAAC;qBAAM,IAAI,IAAI,CAAC,WAAW,EAAE,IAAI,KAAK,YAAY,EAAE,CAAC;oBACnD,MAAM,WAAW,GAAG,MAAM,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC;oBAC9D,IAAI,WAAW,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;wBACnC,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC;wBACxB,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;wBACnB,IAAI,CAAC,iBAAiB,GAAG,IAAI,CAAC;wBAC9B,KAAK,GAAG,mBAAmB,CAAC;oBAC9B,CAAC;gBACH,CAAC;qBAAM,IAAI,IAAI,CAAC,WAAW,EAAE,IAAI,KAAK,QAAQ,EAAE,CAAC;oBAC/C,MAAM,WAAW,GAAG,MAAM,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC;oBAC9D,IAAI,WAAW,CAAC,IAAI,KAAK,SAAS;wBAChC,IAAI,CAAC,MAAM,GAAG,EAAE,KAAK,EAAE,GAAG,EAAE,MAAM,EAAE,oBAAoB,EAAE,CAAC;gBAC/D,CAAC;qBAAM,IACL,QAAQ,CAAC,iBAAiB;oBAC1B,CAAC,IAAI,CAAC,iBAAiB;oBACvB,CAAC,IAAI,CAAC,iBAAiB,EACvB,CAAC;oBACD,MAAM,QAAQ,GACZ,IAAI,CAAC,YAAY,CAAC,YAAY;wBAC9B,IAAI,kBAAkB,CAAC,OAAO,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC;oBAC9C,MAAM,SAAS,GAAG,MAAM,QAAQ,CAAC,QAAQ,CAAC,IAAI,EAAE,SAAS,CAAC,CAAC;oBAC3D,IAAI,CAAC,iBAAiB,GAAG,SAAS,CAAC,IAAI,KAAK,aAAa,CAAC;oBAC1D,IAAI,SAAS,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;wBAC/B,IAAI,CAAC,WAAW,GAAG,SAAS,CAAC,WAAW,CAAC;wBACzC,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;wBACnB,KAAK,GAAG,oBAAoB,CAAC;oBAC/B,CAAC;yBAAM,IAAI,SAAS,CAAC,IAAI,KAAK,WAAW;wBACvC,IAAI,CAAC,MAAM,GAAG,EAAE,KAAK,EAAE,GAAG,EAAE,MAAM,EAAE,sBAAsB,EAAE,CAAC;gBACjE,CAAC;qBAAM,IAAI,IAAI,CAAC,MAAM,EAAE,MAAM,KAAK,cAAc;oBAAE,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;gBACtE,MAAM;YACR,CAAC;YACD,KAAK,QAAQ,CAAC,CAAC,CAAC;gBACd,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC;gBACvB,IAAI,CAAC,cAAc,KAAK,GAAG,CAAC;gBAC5B,IAAI,IAAI,CAAC,aAAa,KAAK,IAAI,IAAI,GAAG,GAAG,IAAI,CAAC,aAAa,IAAI,GAAG,EAAE,CAAC;oBACnE,IAAI,CAAC,QAAQ,EAAE,CAAC;oBAChB,IAAI,CAAC,aAAa,GAAG,GAAG,CAAC;gBAC3B,CAAC;gBACD,IAAI,CAAC,WAAW;oBACd,IAAI,CAAC,QAAQ,GAAG,CAAC;wBACf,CAAC,CAAC,IAAI,CAAC,aAAa,GAAG,GAAG;wBAC1B,CAAC,CAAC,IAAI,CAAC,QAAQ,GAAG,CAAC;4BACjB,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,GAAG,EAAE,IAAI,CAAC,cAAc,GAAG,CAAC,GAAG,GAAG,CAAC;4BACpD,CAAC,CAAC,GAAG,GAAG,QAAQ,CAAC,UAAU,GAAG,GAAG,CAAC;gBACxC,IAAI,IAAI,CAAC,WAAW,EAAE,CAAC;oBACrB,MAAM,WAAW,GAAG,MAAM,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC;oBAC9D,IAAI,WAAW,CAAC,IAAI,KAAK,QAAQ,EAAE,CAAC;wBAClC,IAAI,IAAI,CAAC,WAAW,CAAC,IAAI,KAAK,QAAQ;4BAAE,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC;wBAChE,IAAI,CAAC,MAAM,GAAG,EAAE,KAAK,EAAE,GAAG,EAAE,MAAM,EAAE,oBAAoB,EAAE,CAAC;wBAC3D,KAAK,GAAG,yBAAyB,CAAC;oBACpC,CAAC;gBACH,CAAC;gBACD,IACE,IAAI,CAAC,QAAQ,IAAI,CAAC;oBAClB,CAAC,IAAI,CAAC,WAAW;oBACjB,CAAC,IAAI,CAAC,iBAAiB;oBACvB,QAAQ,CAAC,cAAc;oBACvB,CAAC,IAAI,CAAC,cAAc,KAAK,IAAI,IAAI,IAAI,CAAC,cAAc,IAAI,GAAG,CAAC,EAC5D,CAAC;oBACD,MAAM,QAAQ,GACZ,IAAI,CAAC,YAAY,CAAC,OAAO;wBACzB,IAAI,eAAe,CAAC,OAAO,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC;oBAC3C,MAAM,SAAS,GAAG,MAAM,QAAQ,CAAC,QAAQ,CAAC,IAAI,EAAE,SAAS,CAAC,CAAC;oBAC3D,IAAI,SAAS,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;wBAC/B,IAAI,CAAC,WAAW,GAAG,SAAS,CAAC,WAAW,CAAC;wBACzC,IAAI,CAAC,MAAM,GAAG;4BACZ,KAAK,EAAE,GAAG,GAAG,QAAQ,CAAC,UAAU,GAAG,GAAG;4BACtC,MAAM,EAAE,gBAAgB;yBACzB,CAAC;wBACF,KAAK,GAAG,iBAAiB,CAAC;oBAC5B,CAAC;yBAAM,CAAC;wBACN,MAAM,WAAW,GAAG,SAAS,CAAC,IAAI,KAAK,aAAa,CAAC;wBACrD,IAAI,CAAC,cAAc;4BACjB,GAAG,GAAG,CAAC,WAAW,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,CAAC,gBAAgB,GAAG,GAAG,CAAC,CAAC;wBAC9D,IAAI,CAAC,MAAM,GAAG;4BACZ,KAAK,EAAE,IAAI,CAAC,cAAc;4BAC1B,MAAM,EAAE,WAAW,CAAC,CAAC,CAAC,qBAAqB,CAAC,CAAC,CAAC,iBAAiB;yBAChE,CAAC;wBACF,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,EAAE,IAAI,CAAC,cAAc,CAAC,CAAC;oBACrE,CAAC;gBACH,CAAC;gBACD,MAAM;YACR,CAAC;YACD;gBACE,WAAW,CAAC,MAAM,CAAC,CAAC;QACxB,CAAC;QACD,MAAM,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;QAC1C,MAAM,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,EAAE,GAAG,IAAI,CAAC,EAAE,IAAI,GAAG,IAAI,KAAK,EAAE,CAAC,CAAC;IAC9D,CAAC;CACF"}
|
|
@@ -0,0 +1,11 @@
|
|
|
1
|
+
import type { CollectionSource, LinkOccurrence } from "./schema.js";
|
|
2
|
+
export declare function object(value: unknown): Record<string, unknown> | null;
|
|
3
|
+
export interface ContentDocument {
|
|
4
|
+
id: string;
|
|
5
|
+
slug: string;
|
|
6
|
+
locale: string;
|
|
7
|
+
title: string;
|
|
8
|
+
data: unknown;
|
|
9
|
+
}
|
|
10
|
+
export declare function extractOccurrences(document: ContentDocument, source: CollectionSource, siteUrl: string, generation: string): Promise<LinkOccurrence[]>;
|
|
11
|
+
//# sourceMappingURL=extract.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"extract.d.ts","sourceRoot":"","sources":["../../src/core/extract.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,gBAAgB,EAAE,cAAc,EAAE,MAAM,aAAa,CAAC;AAGpE,wBAAgB,MAAM,CAAC,KAAK,EAAE,OAAO,GAAG,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,IAAI,CAGrE;AACD,MAAM,WAAW,eAAe;IAC9B,EAAE,EAAE,MAAM,CAAC;IACX,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,MAAM,CAAC;IACf,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,OAAO,CAAC;CACf;AAwED,wBAAsB,kBAAkB,CACtC,QAAQ,EAAE,eAAe,EACzB,MAAM,EAAE,gBAAgB,EACxB,OAAO,EAAE,MAAM,EACf,UAAU,EAAE,MAAM,GACjB,OAAO,CAAC,cAAc,EAAE,CAAC,CAmC3B"}
|
|
@@ -0,0 +1,99 @@
|
|
|
1
|
+
import { z } from "zod";
|
|
2
|
+
import { checkUrl, normalizeUrl, stableId } from "./url.js";
|
|
3
|
+
const objectSchema = z.record(z.string(), z.unknown());
|
|
4
|
+
export function object(value) {
|
|
5
|
+
const result = objectSchema.safeParse(value);
|
|
6
|
+
return result.success ? result.data : null;
|
|
7
|
+
}
|
|
8
|
+
function visit(value, segments, path, found) {
|
|
9
|
+
if (segments.length === 0) {
|
|
10
|
+
if (typeof value === "string")
|
|
11
|
+
found.push({ url: value, path, text: "" });
|
|
12
|
+
return;
|
|
13
|
+
}
|
|
14
|
+
const [segment, ...rest] = segments;
|
|
15
|
+
if (segment === "*" && Array.isArray(value)) {
|
|
16
|
+
value.forEach((child, index) => visit(child, rest, `${path}[${index}]`, found));
|
|
17
|
+
}
|
|
18
|
+
else if (segment) {
|
|
19
|
+
const item = object(value);
|
|
20
|
+
if (item)
|
|
21
|
+
visit(item[segment], rest, path ? `${path}.${segment}` : segment, found);
|
|
22
|
+
}
|
|
23
|
+
}
|
|
24
|
+
function portableText(value, path, found) {
|
|
25
|
+
if (!Array.isArray(value))
|
|
26
|
+
return;
|
|
27
|
+
for (const [index, rawBlock] of value.entries()) {
|
|
28
|
+
const block = object(rawBlock);
|
|
29
|
+
if (!block ||
|
|
30
|
+
block["_type"] !== "block" ||
|
|
31
|
+
!Array.isArray(block["markDefs"]))
|
|
32
|
+
continue;
|
|
33
|
+
const children = Array.isArray(block["children"])
|
|
34
|
+
? block["children"]
|
|
35
|
+
: [];
|
|
36
|
+
for (const rawMark of block["markDefs"]) {
|
|
37
|
+
const mark = object(rawMark);
|
|
38
|
+
if (!mark || mark["_type"] !== "link" || typeof mark["href"] !== "string")
|
|
39
|
+
continue;
|
|
40
|
+
let run = null;
|
|
41
|
+
for (const [childIndex, child] of children.entries()) {
|
|
42
|
+
const span = object(child);
|
|
43
|
+
if (!span ||
|
|
44
|
+
!Array.isArray(span["marks"]) ||
|
|
45
|
+
!span["marks"].includes(mark["_key"]) ||
|
|
46
|
+
typeof span["text"] !== "string") {
|
|
47
|
+
run = null;
|
|
48
|
+
continue;
|
|
49
|
+
}
|
|
50
|
+
if (run)
|
|
51
|
+
run.text = (run.text + span["text"]).slice(0, 500);
|
|
52
|
+
else {
|
|
53
|
+
run = {
|
|
54
|
+
url: mark["href"],
|
|
55
|
+
text: span["text"].slice(0, 500),
|
|
56
|
+
path: `${path}[${index}].children[${childIndex}]`,
|
|
57
|
+
};
|
|
58
|
+
found.push(run);
|
|
59
|
+
}
|
|
60
|
+
}
|
|
61
|
+
}
|
|
62
|
+
}
|
|
63
|
+
}
|
|
64
|
+
export async function extractOccurrences(document, source, siteUrl, generation) {
|
|
65
|
+
const found = [];
|
|
66
|
+
const data = object(document.data);
|
|
67
|
+
if (!data)
|
|
68
|
+
return [];
|
|
69
|
+
for (const field of source.portableTextFields)
|
|
70
|
+
portableText(data[field], field, found);
|
|
71
|
+
for (const field of source.urlFields)
|
|
72
|
+
visit(data, field.replace(/\[\]/gu, ".*").split("."), "", found);
|
|
73
|
+
const occurrences = [];
|
|
74
|
+
for (const item of found) {
|
|
75
|
+
const originalUrl = normalizeUrl(item.url, siteUrl);
|
|
76
|
+
if (!originalUrl || new URL(originalUrl).origin === new URL(siteUrl).origin)
|
|
77
|
+
continue;
|
|
78
|
+
const id = await stableId(`${source.collection}/${document.id}/${item.path}/${originalUrl}`);
|
|
79
|
+
occurrences.push({
|
|
80
|
+
schemaVersion: 1,
|
|
81
|
+
id,
|
|
82
|
+
linkId: await stableId(checkUrl(originalUrl)),
|
|
83
|
+
originalUrl,
|
|
84
|
+
collection: source.collection,
|
|
85
|
+
contentId: document.id,
|
|
86
|
+
slug: document.slug,
|
|
87
|
+
locale: document.locale,
|
|
88
|
+
title: document.title,
|
|
89
|
+
path: item.path,
|
|
90
|
+
text: item.text,
|
|
91
|
+
pagePath: source.path
|
|
92
|
+
.replaceAll("{slug}", encodeURIComponent(document.slug))
|
|
93
|
+
.replaceAll("{locale}", encodeURIComponent(document.locale)),
|
|
94
|
+
generation,
|
|
95
|
+
});
|
|
96
|
+
}
|
|
97
|
+
return occurrences;
|
|
98
|
+
}
|
|
99
|
+
//# sourceMappingURL=extract.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"extract.js","sourceRoot":"","sources":["../../src/core/extract.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AACxB,OAAO,EAAE,QAAQ,EAAE,YAAY,EAAE,QAAQ,EAAE,MAAM,UAAU,CAAC;AAG5D,MAAM,YAAY,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,CAAC;AACvD,MAAM,UAAU,MAAM,CAAC,KAAc;IACnC,MAAM,MAAM,GAAG,YAAY,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;IAC7C,OAAO,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC;AAC7C,CAAC;AAcD,SAAS,KAAK,CACZ,KAAc,EACd,QAAkB,EAClB,IAAY,EACZ,KAAc;IAEd,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC1B,IAAI,OAAO,KAAK,KAAK,QAAQ;YAAE,KAAK,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,EAAE,CAAC,CAAC;QAC1E,OAAO;IACT,CAAC;IACD,MAAM,CAAC,OAAO,EAAE,GAAG,IAAI,CAAC,GAAG,QAAQ,CAAC;IACpC,IAAI,OAAO,KAAK,GAAG,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,EAAE,CAAC;QAC5C,KAAK,CAAC,OAAO,CAAC,CAAC,KAAc,EAAE,KAAa,EAAE,EAAE,CAC9C,KAAK,CAAC,KAAK,EAAE,IAAI,EAAE,GAAG,IAAI,IAAI,KAAK,GAAG,EAAE,KAAK,CAAC,CAC/C,CAAC;IACJ,CAAC;SAAM,IAAI,OAAO,EAAE,CAAC;QACnB,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC;QAC3B,IAAI,IAAI;YACN,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC,GAAG,IAAI,IAAI,OAAO,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;IAC7E,CAAC;AACH,CAAC;AAED,SAAS,YAAY,CAAC,KAAc,EAAE,IAAY,EAAE,KAAc;IAChE,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC;QAAE,OAAO;IAClC,KAAK,MAAM,CAAC,KAAK,EAAE,QAAQ,CAAC,IAAI,KAAK,CAAC,OAAO,EAAE,EAAE,CAAC;QAChD,MAAM,KAAK,GAAG,MAAM,CAAC,QAAQ,CAAC,CAAC;QAC/B,IACE,CAAC,KAAK;YACN,KAAK,CAAC,OAAO,CAAC,KAAK,OAAO;YAC1B,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;YAEjC,SAAS;QACX,MAAM,QAAQ,GAAc,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;YAC1D,CAAC,CAAC,KAAK,CAAC,UAAU,CAAC;YACnB,CAAC,CAAC,EAAE,CAAC;QACP,KAAK,MAAM,OAAO,IAAI,KAAK,CAAC,UAAU,CAAC,EAAE,CAAC;YACxC,MAAM,IAAI,GAAG,MAAM,CAAC,OAAO,CAAC,CAAC;YAC7B,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,OAAO,CAAC,KAAK,MAAM,IAAI,OAAO,IAAI,CAAC,MAAM,CAAC,KAAK,QAAQ;gBACvE,SAAS;YACX,IAAI,GAAG,GAAiB,IAAI,CAAC;YAC7B,KAAK,MAAM,CAAC,UAAU,EAAE,KAAK,CAAC,IAAI,QAAQ,CAAC,OAAO,EAAE,EAAE,CAAC;gBACrD,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC;gBAC3B,IACE,CAAC,IAAI;oBACL,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;oBAC7B,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;oBACrC,OAAO,IAAI,CAAC,MAAM,CAAC,KAAK,QAAQ,EAChC,CAAC;oBACD,GAAG,GAAG,IAAI,CAAC;oBACX,SAAS;gBACX,CAAC;gBACD,IAAI,GAAG;oBAAE,GAAG,CAAC,IAAI,GAAG,CAAC,GAAG,CAAC,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;qBACvD,CAAC;oBACJ,GAAG,GAAG;wBACJ,GAAG,EAAE,IAAI,CAAC,MAAM,CAAC;wBACjB,IAAI,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC;wBAChC,IAAI,EAAE,GAAG,IAAI,IAAI,KAAK,cAAc,UAAU,GAAG;qBAClD,CAAC;oBACF,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBAClB,CAAC;YACH,CAAC;QACH,CAAC;IACH,CAAC;AACH,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,kBAAkB,CACtC,QAAyB,EACzB,MAAwB,EACxB,OAAe,EACf,UAAkB;IAElB,MAAM,KAAK,GAAY,EAAE,CAAC;IAC1B,MAAM,IAAI,GAAG,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;IACnC,IAAI,CAAC,IAAI;QAAE,OAAO,EAAE,CAAC;IACrB,KAAK,MAAM,KAAK,IAAI,MAAM,CAAC,kBAAkB;QAC3C,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC;IAC1C,KAAK,MAAM,KAAK,IAAI,MAAM,CAAC,SAAS;QAClC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,OAAO,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,KAAK,CAAC,CAAC;IACnE,MAAM,WAAW,GAAqB,EAAE,CAAC;IACzC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,WAAW,GAAG,YAAY,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;QACpD,IAAI,CAAC,WAAW,IAAI,IAAI,GAAG,CAAC,WAAW,CAAC,CAAC,MAAM,KAAK,IAAI,GAAG,CAAC,OAAO,CAAC,CAAC,MAAM;YACzE,SAAS;QACX,MAAM,EAAE,GAAG,MAAM,QAAQ,CACvB,GAAG,MAAM,CAAC,UAAU,IAAI,QAAQ,CAAC,EAAE,IAAI,IAAI,CAAC,IAAI,IAAI,WAAW,EAAE,CAClE,CAAC;QACF,WAAW,CAAC,IAAI,CAAC;YACf,aAAa,EAAE,CAAC;YAChB,EAAE;YACF,MAAM,EAAE,MAAM,QAAQ,CAAC,QAAQ,CAAC,WAAW,CAAC,CAAC;YAC7C,WAAW;YACX,UAAU,EAAE,MAAM,CAAC,UAAU;YAC7B,SAAS,EAAE,QAAQ,CAAC,EAAE;YACtB,IAAI,EAAE,QAAQ,CAAC,IAAI;YACnB,MAAM,EAAE,QAAQ,CAAC,MAAM;YACvB,KAAK,EAAE,QAAQ,CAAC,KAAK;YACrB,IAAI,EAAE,IAAI,CAAC,IAAI;YACf,IAAI,EAAE,IAAI,CAAC,IAAI;YACf,QAAQ,EAAE,MAAM,CAAC,IAAI;iBAClB,UAAU,CAAC,QAAQ,EAAE,kBAAkB,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;iBACvD,UAAU,CAAC,UAAU,EAAE,kBAAkB,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAC9D,UAAU;SACX,CAAC,CAAC;IACL,CAAC;IACD,OAAO,WAAW,CAAC;AACrB,CAAC"}
|
|
@@ -0,0 +1,12 @@
|
|
|
1
|
+
import type { DefaultTreeAdapterMap } from "parse5";
|
|
2
|
+
import type { PageEvidence } from "./schema.js";
|
|
3
|
+
type Node = DefaultTreeAdapterMap["node"];
|
|
4
|
+
type Element = DefaultTreeAdapterMap["element"];
|
|
5
|
+
export declare function elements(root: Node): Element[];
|
|
6
|
+
export declare function attribute(element: Element, name: string): string;
|
|
7
|
+
export declare function pageEvidence(html: string, url: string): PageEvidence;
|
|
8
|
+
export declare function escapeAttribute(value: string): string;
|
|
9
|
+
/** Edit only href attribute byte ranges, preserving all other HTML verbatim. */
|
|
10
|
+
export declare function rewriteLinks(html: string, replacements: ReadonlyMap<string, string>, siteUrl: string): string;
|
|
11
|
+
export {};
|
|
12
|
+
//# sourceMappingURL=html.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"html.d.ts","sourceRoot":"","sources":["../../src/core/html.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,qBAAqB,EAAE,MAAM,QAAQ,CAAC;AACpD,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,aAAa,CAAC;AAEhD,KAAK,IAAI,GAAG,qBAAqB,CAAC,MAAM,CAAC,CAAC;AAC1C,KAAK,OAAO,GAAG,qBAAqB,CAAC,SAAS,CAAC,CAAC;AAChD,wBAAgB,QAAQ,CAAC,IAAI,EAAE,IAAI,GAAG,OAAO,EAAE,CAQ9C;AACD,wBAAgB,SAAS,CAAC,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,CAEhE;AAOD,wBAAgB,YAAY,CAAC,IAAI,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,YAAY,CAmDpE;AAED,wBAAgB,eAAe,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAMrD;AAED,gFAAgF;AAChF,wBAAgB,YAAY,CAC1B,IAAI,EAAE,MAAM,EACZ,YAAY,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,EACzC,OAAO,EAAE,MAAM,GACd,MAAM,CAyBR"}
|
|
@@ -0,0 +1,103 @@
|
|
|
1
|
+
import { parse } from "parse5";
|
|
2
|
+
export function elements(root) {
|
|
3
|
+
const result = [];
|
|
4
|
+
function walk(node) {
|
|
5
|
+
if ("tagName" in node)
|
|
6
|
+
result.push(node);
|
|
7
|
+
if ("childNodes" in node)
|
|
8
|
+
node.childNodes.forEach(walk);
|
|
9
|
+
}
|
|
10
|
+
walk(root);
|
|
11
|
+
return result;
|
|
12
|
+
}
|
|
13
|
+
export function attribute(element, name) {
|
|
14
|
+
return element.attrs.find((attr) => attr.name === name)?.value ?? "";
|
|
15
|
+
}
|
|
16
|
+
function textOf(node) {
|
|
17
|
+
if (node.nodeName === "#text" && "value" in node)
|
|
18
|
+
return node.value;
|
|
19
|
+
if ("tagName" in node && ["script", "style", "nav"].includes(node.tagName))
|
|
20
|
+
return "";
|
|
21
|
+
return "childNodes" in node ? node.childNodes.map(textOf).join(" ") : "";
|
|
22
|
+
}
|
|
23
|
+
export function pageEvidence(html, url) {
|
|
24
|
+
const nodes = elements(parse(html));
|
|
25
|
+
const h1 = nodes.find((node) => node.tagName === "h1");
|
|
26
|
+
const title = nodes.find((node) => node.tagName === "title");
|
|
27
|
+
const main = nodes.find((node) => ["main", "article"].includes(node.tagName)) ??
|
|
28
|
+
nodes.find((node) => node.tagName === "body");
|
|
29
|
+
const paragraphs = main
|
|
30
|
+
? elements(main).filter((node) => node.tagName === "p")
|
|
31
|
+
: [];
|
|
32
|
+
const introduction = paragraphs
|
|
33
|
+
.map(textOf)
|
|
34
|
+
.map((text) => text.replace(/\s+/gu, " ").trim())
|
|
35
|
+
.filter((text) => text.length > 80)
|
|
36
|
+
.slice(0, 3)
|
|
37
|
+
.join(" ")
|
|
38
|
+
.slice(0, 2000);
|
|
39
|
+
const canonical = nodes.find((node) => node.tagName === "link" &&
|
|
40
|
+
attribute(node, "rel").split(" ").includes("canonical"));
|
|
41
|
+
const htmlNode = nodes.find((node) => node.tagName === "html");
|
|
42
|
+
const pageTitle = textOf(h1 ?? title ?? parse(""))
|
|
43
|
+
.replace(/\s+/gu, " ")
|
|
44
|
+
.trim();
|
|
45
|
+
return {
|
|
46
|
+
title: pageTitle,
|
|
47
|
+
language: htmlNode ? attribute(htmlNode, "lang") : "",
|
|
48
|
+
canonical: canonical
|
|
49
|
+
? new URL(attribute(canonical, "href"), url).href
|
|
50
|
+
: url,
|
|
51
|
+
introduction,
|
|
52
|
+
anchors: nodes
|
|
53
|
+
.flatMap((node) => {
|
|
54
|
+
const id = attribute(node, "id");
|
|
55
|
+
return id ? [id] : [];
|
|
56
|
+
})
|
|
57
|
+
.slice(0, 2000),
|
|
58
|
+
sourceUrls: nodes
|
|
59
|
+
.filter((node) => node.tagName === "a")
|
|
60
|
+
.map((node) => attribute(node, "href"))
|
|
61
|
+
.filter((href) => /^https:\/\/[a-z-]+\.wikipedia\.org\/wiki\//u.test(href))
|
|
62
|
+
.slice(0, 200),
|
|
63
|
+
disambiguation: /\bdisambiguation\b|may refer to|begriffsklärung|bezeichnet unter anderem/iu.test(`${pageTitle} ${introduction.slice(0, 400)}`),
|
|
64
|
+
};
|
|
65
|
+
}
|
|
66
|
+
export function escapeAttribute(value) {
|
|
67
|
+
return value
|
|
68
|
+
.replaceAll("&", "&")
|
|
69
|
+
.replaceAll('"', """)
|
|
70
|
+
.replaceAll("<", "<")
|
|
71
|
+
.replaceAll(">", ">");
|
|
72
|
+
}
|
|
73
|
+
/** Edit only href attribute byte ranges, preserving all other HTML verbatim. */
|
|
74
|
+
export function rewriteLinks(html, replacements, siteUrl) {
|
|
75
|
+
const nodes = elements(parse(html, { sourceCodeLocationInfo: true }));
|
|
76
|
+
const edits = [];
|
|
77
|
+
for (const node of nodes) {
|
|
78
|
+
if (node.tagName !== "a" || !node.sourceCodeLocation?.attrs)
|
|
79
|
+
continue;
|
|
80
|
+
const href = attribute(node, "href");
|
|
81
|
+
let original;
|
|
82
|
+
try {
|
|
83
|
+
original = new URL(href, siteUrl).href;
|
|
84
|
+
}
|
|
85
|
+
catch {
|
|
86
|
+
continue;
|
|
87
|
+
}
|
|
88
|
+
const destination = replacements.get(original);
|
|
89
|
+
const location = node.sourceCodeLocation.attrs["href"];
|
|
90
|
+
if (!destination || !location)
|
|
91
|
+
continue;
|
|
92
|
+
edits.push({
|
|
93
|
+
start: location.startOffset,
|
|
94
|
+
end: location.endOffset,
|
|
95
|
+
value: `href="${escapeAttribute(destination)}"`,
|
|
96
|
+
});
|
|
97
|
+
}
|
|
98
|
+
let output = html;
|
|
99
|
+
for (const edit of edits.sort((a, b) => b.start - a.start))
|
|
100
|
+
output = output.slice(0, edit.start) + edit.value + output.slice(edit.end);
|
|
101
|
+
return output;
|
|
102
|
+
}
|
|
103
|
+
//# sourceMappingURL=html.js.map
|