@gscdump/engine 1.4.10 → 1.5.0
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/dist/entities/empty-types.d.mts +22 -0
- package/dist/entities/empty-types.mjs +58 -0
- package/dist/entities/indexing-metadata.d.mts +26 -0
- package/dist/entities/indexing-metadata.mjs +31 -0
- package/dist/entities/inspection.d.mts +240 -0
- package/dist/entities/inspection.mjs +443 -0
- package/dist/entities/io.mjs +16 -0
- package/dist/{query-dim.d.mts → entities/query-dim.d.mts} +3 -3
- package/dist/{query-dim.mjs → entities/query-dim.mjs} +4 -4
- package/dist/{sitemap-projection.mjs → entities/sitemap-projection.mjs} +1 -1
- package/dist/entities/sitemap-shared.d.mts +199 -0
- package/dist/entities/sitemap-shared.mjs +243 -0
- package/dist/entities/sitemap-write.d.mts +3 -0
- package/dist/entities/sitemap-write.mjs +528 -0
- package/dist/entities/sitemap.d.mts +3 -0
- package/dist/entities/sitemap.mjs +91 -0
- package/dist/entities.d.mts +9 -481
- package/dist/entities.mjs +7 -1380
- package/dist/rollups/canonical.d.mts +71 -0
- package/dist/rollups/canonical.mjs +335 -0
- package/dist/rollups/core.d.mts +201 -0
- package/dist/rollups/core.mjs +116 -0
- package/dist/rollups/dates.mjs +11 -0
- package/dist/rollups/defaults.d.mts +11 -0
- package/dist/rollups/defaults.mjs +17 -0
- package/dist/rollups/hourly.d.mts +38 -0
- package/dist/rollups/hourly.mjs +38 -0
- package/dist/rollups/indexing.d.mts +46 -0
- package/dist/rollups/indexing.mjs +357 -0
- package/dist/rollups/traffic.d.mts +38 -0
- package/dist/rollups/traffic.mjs +289 -0
- package/dist/rollups/windows.d.mts +90 -0
- package/dist/rollups/windows.mjs +176 -0
- package/dist/rollups.d.mts +8 -471
- package/dist/rollups.mjs +7 -1316
- package/package.json +4 -4
- /package/dist/{sitemap-projection.d.mts → entities/sitemap-projection.d.mts} +0 -0
|
@@ -0,0 +1,243 @@
|
|
|
1
|
+
import { readOptional } from "../adapters/read-optional.mjs";
|
|
2
|
+
import { decodeParquetToRows } from "../adapters/hyparquet.mjs";
|
|
3
|
+
import { mapEntityIo } from "./io.mjs";
|
|
4
|
+
const SITEMAP_URLS_EVENT_PREFIX_RE = /\/urls\/events\/(\d{4}-\d{2}-\d{2})__[0-9a-f]+__\d+__[0-9a-f]+\.parquet$/;
|
|
5
|
+
const SITEMAP_URLS_PENDING_GENERATION_RE = /\/urls\/generations\/pending\/([0-9a-f]+)\.json$/;
|
|
6
|
+
const URLS_INDEX_COLUMNS = [
|
|
7
|
+
{
|
|
8
|
+
name: "feedpath",
|
|
9
|
+
type: "VARCHAR",
|
|
10
|
+
nullable: false
|
|
11
|
+
},
|
|
12
|
+
{
|
|
13
|
+
name: "feedpath_hash",
|
|
14
|
+
type: "VARCHAR",
|
|
15
|
+
nullable: false
|
|
16
|
+
},
|
|
17
|
+
{
|
|
18
|
+
name: "url_hash",
|
|
19
|
+
type: "VARCHAR",
|
|
20
|
+
nullable: false
|
|
21
|
+
},
|
|
22
|
+
{
|
|
23
|
+
name: "loc",
|
|
24
|
+
type: "VARCHAR",
|
|
25
|
+
nullable: false
|
|
26
|
+
},
|
|
27
|
+
{
|
|
28
|
+
name: "lastmod",
|
|
29
|
+
type: "VARCHAR",
|
|
30
|
+
nullable: true
|
|
31
|
+
},
|
|
32
|
+
{
|
|
33
|
+
name: "first_seen_at",
|
|
34
|
+
type: "BIGINT",
|
|
35
|
+
nullable: false
|
|
36
|
+
},
|
|
37
|
+
{
|
|
38
|
+
name: "last_seen_at",
|
|
39
|
+
type: "BIGINT",
|
|
40
|
+
nullable: false
|
|
41
|
+
},
|
|
42
|
+
{
|
|
43
|
+
name: "removed_at",
|
|
44
|
+
type: "BIGINT",
|
|
45
|
+
nullable: true
|
|
46
|
+
}
|
|
47
|
+
];
|
|
48
|
+
const URLS_DELTA_COLUMNS = [
|
|
49
|
+
{
|
|
50
|
+
name: "feedpath",
|
|
51
|
+
type: "VARCHAR",
|
|
52
|
+
nullable: false
|
|
53
|
+
},
|
|
54
|
+
{
|
|
55
|
+
name: "feedpath_hash",
|
|
56
|
+
type: "VARCHAR",
|
|
57
|
+
nullable: false
|
|
58
|
+
},
|
|
59
|
+
{
|
|
60
|
+
name: "url_hash",
|
|
61
|
+
type: "VARCHAR",
|
|
62
|
+
nullable: false
|
|
63
|
+
},
|
|
64
|
+
{
|
|
65
|
+
name: "op",
|
|
66
|
+
type: "VARCHAR",
|
|
67
|
+
nullable: false
|
|
68
|
+
},
|
|
69
|
+
{
|
|
70
|
+
name: "loc",
|
|
71
|
+
type: "VARCHAR",
|
|
72
|
+
nullable: false
|
|
73
|
+
},
|
|
74
|
+
{
|
|
75
|
+
name: "lastmod",
|
|
76
|
+
type: "VARCHAR",
|
|
77
|
+
nullable: true
|
|
78
|
+
},
|
|
79
|
+
{
|
|
80
|
+
name: "at",
|
|
81
|
+
type: "BIGINT",
|
|
82
|
+
nullable: false
|
|
83
|
+
},
|
|
84
|
+
{
|
|
85
|
+
name: "generation_id",
|
|
86
|
+
type: "VARCHAR",
|
|
87
|
+
nullable: true
|
|
88
|
+
}
|
|
89
|
+
];
|
|
90
|
+
const URLS_EVENT_COLUMNS = [
|
|
91
|
+
{
|
|
92
|
+
name: "feedpath",
|
|
93
|
+
type: "VARCHAR",
|
|
94
|
+
nullable: false
|
|
95
|
+
},
|
|
96
|
+
{
|
|
97
|
+
name: "feedpath_hash",
|
|
98
|
+
type: "VARCHAR",
|
|
99
|
+
nullable: false
|
|
100
|
+
},
|
|
101
|
+
{
|
|
102
|
+
name: "url_hash",
|
|
103
|
+
type: "VARCHAR",
|
|
104
|
+
nullable: false
|
|
105
|
+
},
|
|
106
|
+
{
|
|
107
|
+
name: "op",
|
|
108
|
+
type: "VARCHAR",
|
|
109
|
+
nullable: false
|
|
110
|
+
},
|
|
111
|
+
{
|
|
112
|
+
name: "loc",
|
|
113
|
+
type: "VARCHAR",
|
|
114
|
+
nullable: false
|
|
115
|
+
},
|
|
116
|
+
{
|
|
117
|
+
name: "lastmod",
|
|
118
|
+
type: "VARCHAR",
|
|
119
|
+
nullable: true
|
|
120
|
+
},
|
|
121
|
+
{
|
|
122
|
+
name: "generation_id",
|
|
123
|
+
type: "VARCHAR",
|
|
124
|
+
nullable: false
|
|
125
|
+
},
|
|
126
|
+
{
|
|
127
|
+
name: "observed_at",
|
|
128
|
+
type: "BIGINT",
|
|
129
|
+
nullable: false
|
|
130
|
+
},
|
|
131
|
+
{
|
|
132
|
+
name: "sequence",
|
|
133
|
+
type: "INTEGER",
|
|
134
|
+
nullable: false
|
|
135
|
+
},
|
|
136
|
+
{
|
|
137
|
+
name: "projects_state",
|
|
138
|
+
type: "INTEGER",
|
|
139
|
+
nullable: false
|
|
140
|
+
},
|
|
141
|
+
{
|
|
142
|
+
name: "seed_generation",
|
|
143
|
+
type: "INTEGER",
|
|
144
|
+
nullable: false
|
|
145
|
+
},
|
|
146
|
+
{
|
|
147
|
+
name: "generation_kind",
|
|
148
|
+
type: "VARCHAR",
|
|
149
|
+
nullable: false
|
|
150
|
+
},
|
|
151
|
+
{
|
|
152
|
+
name: "content_hash",
|
|
153
|
+
type: "VARCHAR",
|
|
154
|
+
nullable: true
|
|
155
|
+
},
|
|
156
|
+
{
|
|
157
|
+
name: "input_count",
|
|
158
|
+
type: "INTEGER",
|
|
159
|
+
nullable: true
|
|
160
|
+
}
|
|
161
|
+
];
|
|
162
|
+
function rowToUrlRecord(row) {
|
|
163
|
+
return {
|
|
164
|
+
feedpath: String(row.feedpath),
|
|
165
|
+
feedpathHash: String(row.feedpath_hash),
|
|
166
|
+
urlHash: String(row.url_hash),
|
|
167
|
+
loc: String(row.loc),
|
|
168
|
+
lastmod: row.lastmod == null ? void 0 : String(row.lastmod),
|
|
169
|
+
firstSeenAt: Number(row.first_seen_at),
|
|
170
|
+
lastSeenAt: Number(row.last_seen_at),
|
|
171
|
+
removedAt: row.removed_at == null ? void 0 : Number(row.removed_at)
|
|
172
|
+
};
|
|
173
|
+
}
|
|
174
|
+
function urlRecordToRow(r) {
|
|
175
|
+
return {
|
|
176
|
+
feedpath: r.feedpath,
|
|
177
|
+
feedpath_hash: r.feedpathHash,
|
|
178
|
+
url_hash: r.urlHash,
|
|
179
|
+
loc: r.loc,
|
|
180
|
+
lastmod: r.lastmod ?? null,
|
|
181
|
+
first_seen_at: r.firstSeenAt,
|
|
182
|
+
last_seen_at: r.lastSeenAt,
|
|
183
|
+
removed_at: r.removedAt ?? null
|
|
184
|
+
};
|
|
185
|
+
}
|
|
186
|
+
function createSitemapUrlState(indexRows) {
|
|
187
|
+
const state = {
|
|
188
|
+
live: /* @__PURE__ */ new Map(),
|
|
189
|
+
removed: /* @__PURE__ */ new Map()
|
|
190
|
+
};
|
|
191
|
+
for (const row of indexRows) {
|
|
192
|
+
const record = rowToUrlRecord(row);
|
|
193
|
+
if (record.removedAt == null) state.live.set(record.urlHash, record);
|
|
194
|
+
else state.removed.set(record.urlHash, record);
|
|
195
|
+
}
|
|
196
|
+
return state;
|
|
197
|
+
}
|
|
198
|
+
function applySitemapDeltaRows(state, rows) {
|
|
199
|
+
for (const row of rows) {
|
|
200
|
+
const urlHash = String(row.url_hash);
|
|
201
|
+
const observedAt = Number(row.at);
|
|
202
|
+
const op = String(row.op);
|
|
203
|
+
if (op === "added") {
|
|
204
|
+
const previous = state.live.get(urlHash) ?? state.removed.get(urlHash);
|
|
205
|
+
state.removed.delete(urlHash);
|
|
206
|
+
state.live.set(urlHash, {
|
|
207
|
+
feedpath: String(row.feedpath),
|
|
208
|
+
feedpathHash: String(row.feedpath_hash),
|
|
209
|
+
urlHash,
|
|
210
|
+
loc: String(row.loc),
|
|
211
|
+
lastmod: row.lastmod == null ? void 0 : String(row.lastmod),
|
|
212
|
+
firstSeenAt: previous?.firstSeenAt ?? observedAt,
|
|
213
|
+
lastSeenAt: observedAt
|
|
214
|
+
});
|
|
215
|
+
} else if (op === "removed") {
|
|
216
|
+
const previous = state.live.get(urlHash);
|
|
217
|
+
state.live.delete(urlHash);
|
|
218
|
+
if (previous) state.removed.set(urlHash, {
|
|
219
|
+
...previous,
|
|
220
|
+
removedAt: observedAt
|
|
221
|
+
});
|
|
222
|
+
}
|
|
223
|
+
}
|
|
224
|
+
}
|
|
225
|
+
function applySitemapDeltaFiles(state, files) {
|
|
226
|
+
for (const file of files) applySitemapDeltaRows(state, file.rows);
|
|
227
|
+
}
|
|
228
|
+
function sortSitemapDeltaFiles(files) {
|
|
229
|
+
return files.sort((a, b) => a.key.localeCompare(b.key));
|
|
230
|
+
}
|
|
231
|
+
async function readSitemapDeltaFiles(ds, keys) {
|
|
232
|
+
return sortSitemapDeltaFiles((await mapEntityIo(keys, async (key) => {
|
|
233
|
+
const bytes = await readOptional(ds, key);
|
|
234
|
+
return bytes ? {
|
|
235
|
+
key,
|
|
236
|
+
rows: await decodeParquetToRows(bytes)
|
|
237
|
+
} : void 0;
|
|
238
|
+
})).filter((file) => file !== void 0));
|
|
239
|
+
}
|
|
240
|
+
function dateInRange(date, range) {
|
|
241
|
+
return (!range?.from || date >= range.from) && (!range?.to || date <= range.to);
|
|
242
|
+
}
|
|
243
|
+
export { SITEMAP_URLS_EVENT_PREFIX_RE, SITEMAP_URLS_PENDING_GENERATION_RE, URLS_DELTA_COLUMNS, URLS_EVENT_COLUMNS, URLS_INDEX_COLUMNS, applySitemapDeltaFiles, applySitemapDeltaRows, createSitemapUrlState, dateInRange, readSitemapDeltaFiles, rowToUrlRecord, sortSitemapDeltaFiles, urlRecordToRow };
|