add kanjidic and radicals
This commit is contained in:
1 parent
0f993cffe3
commit
6e2c025e2a
6 files changed
+224
-10
No files matched your search
@@ -0,0 +1,98 @@
|
||||
import createConnectionPool, { ConnectionPool, sql } from '@databases/pg'
|
||||
import sqlite3 from 'better-sqlite3'
|
||||
import { PythonShell } from 'python-shell'
|
||||
|
||||
export async function main(db: ConnectionPool) {
|
||||
const s3 = sqlite3('./assets/radical.db')
|
||||
|
||||
const rows = s3
|
||||
.prepare(
|
||||
/* sql */ `
|
||||
SELECT "entry", "sub", "sup", "var" FROM "radical"
|
||||
`
|
||||
)
|
||||
.all()
|
||||
|
||||
const fPrev = new Set<string>(
|
||||
await db
|
||||
.query(
|
||||
sql`
|
||||
SELECT "entry" FROM "frequency"
|
||||
`
|
||||
)
|
||||
.then((rs) => rs.map((r) => r.entry))
|
||||
)
|
||||
|
||||
const py = new PythonShell('./scripts/freq.py', {
|
||||
mode: 'text',
|
||||
pythonOptions: ['-u'],
|
||||
})
|
||||
|
||||
const fMap = new Map<string, number>()
|
||||
py.on('message', (msg: string) => {
|
||||
const [w = '', f] = msg.split('=')
|
||||
fMap.set(w, Number(f))
|
||||
})
|
||||
;[
|
||||
...new Set(
|
||||
rows.flatMap((r) =>
|
||||
[
|
||||
...((r.entry + r.sub + r.sup + r.var) as string).matchAll(
|
||||
/\p{sc=Han}/gu
|
||||
),
|
||||
].map((m) => m[0]!)
|
||||
)
|
||||
),
|
||||
]
|
||||
.filter((v) => !fPrev.has(v))
|
||||
.map((v) => py.send(v))
|
||||
|
||||
await new Promise<void>((resolve, reject) => {
|
||||
py.end((err) => (err ? reject(err) : resolve()))
|
||||
})
|
||||
|
||||
const batchSize = 5000
|
||||
|
||||
if (fMap.size) {
|
||||
const fRow = [...fMap]
|
||||
for (let i = 0; i < fRow.length; i += batchSize) {
|
||||
console.log(i)
|
||||
|
||||
await db.query(sql`
|
||||
INSERT INTO "frequency" ("entry", "frequency")
|
||||
VALUES ${sql.join(
|
||||
fRow.slice(i, i + batchSize).map(([k, v]) => sql`(${k}, ${v})`),
|
||||
','
|
||||
)}
|
||||
ON CONFLICT ("entry")
|
||||
DO UPDATE SET "frequency" = EXCLUDED."frequency";
|
||||
`)
|
||||
}
|
||||
}
|
||||
|
||||
for (let i = 0; i < rows.length; i += batchSize) {
|
||||
console.log(i)
|
||||
const sublot = rows.slice(i, i + batchSize)
|
||||
|
||||
await db.query(sql`
|
||||
INSERT INTO "character" ("entry", "sub", "sup", "var")
|
||||
VALUES ${sql.join(
|
||||
sublot.map((r) => {
|
||||
return sql`(${r.entry}, ${[
|
||||
...(r.sub as string).matchAll(/\p{sc=Han}/gu),
|
||||
].map((m) => m[0]!)}, ${[
|
||||
...(r.sup as string).matchAll(/\p{sc=Han}/gu),
|
||||
].map((m) => m[0]!)}, ${[
|
||||
...(r.var as string).matchAll(/\p{sc=Han}/gu),
|
||||
].map((m) => m[0]!)})`
|
||||
}),
|
||||
','
|
||||
)}
|
||||
ON CONFLICT DO NOTHING;
|
||||
`)
|
||||
}
|
||||
}
|
||||
|
||||
if (require.main === module) {
|
||||
main(createConnectionPool({ bigIntMode: 'number' }))
|
||||
}
|
||||
Reference in new issue
Block a user