Spaces:
Running
Running
Tony Powell
commited on
Commit
·
5534d69
1
Parent(s):
843fee0
Move parquet table loading into custom hook
Browse files- src/App.tsx +12 -64
- src/useParquetTable.ts +76 -0
src/App.tsx
CHANGED
@@ -1,22 +1,14 @@
|
|
1 |
-
import { useCallback,
|
2 |
import { Table } from "./table";
|
3 |
-
import { AsyncDuckDBConnection } from "@duckdb/duckdb-wasm";
|
4 |
import { Button } from "~/components/ui/button";
|
5 |
import { Textarea } from "~/components/ui/textarea";
|
6 |
import { useSearchParams } from "~/useSearchParams";
|
7 |
import { useDuckDb } from "~/useDuckDb";
|
|
|
8 |
|
9 |
const DEFAULT_DATASET_URL =
|
10 |
"https://huggingface.co/datasets/openai/openai_humaneval/resolve/main/openai_humaneval/test-00000-of-00001.parquet";
|
11 |
|
12 |
-
// https://duckdb.org/docs/api/wasm/query#arrow-table-to-json
|
13 |
-
// TODO: import the arrow lib and use the correct type
|
14 |
-
// eslint-disable-next-line @typescript-eslint/no-explicit-any
|
15 |
-
const arrowResultToJson = (arrowResult: any) => {
|
16 |
-
// eslint-disable-next-line @typescript-eslint/no-explicit-any
|
17 |
-
return arrowResult.toArray().map((row: any) => row.toJSON());
|
18 |
-
};
|
19 |
-
|
20 |
const usePersistedTextfield = (fieldName: string) => {
|
21 |
const { searchParams, setSearchParams } = useSearchParams();
|
22 |
const searchParamsObj = new URLSearchParams(searchParams);
|
@@ -31,61 +23,13 @@ const usePersistedTextfield = (fieldName: string) => {
|
|
31 |
};
|
32 |
|
33 |
function App() {
|
34 |
-
const [loading, setLoading] = useState(false);
|
35 |
const [datasetUrl, setDatasetUrl] = usePersistedTextfield("datasetUrl");
|
36 |
const [nextDatasetUrl, setNextDatasetUrl] = useState(() => datasetUrl);
|
37 |
-
const [dataset, setDataset] = useState<null>(null);
|
38 |
-
const [error, setError] = useState<string | null>(null);
|
39 |
const db = useDuckDb();
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
setDataset(null);
|
45 |
-
setError(null);
|
46 |
-
console.log("Loading", datasetUrl);
|
47 |
-
const status: { conn: AsyncDuckDBConnection | null; killed: boolean } = {
|
48 |
-
conn: null,
|
49 |
-
killed: false,
|
50 |
-
};
|
51 |
-
db.connect().then((conn) => {
|
52 |
-
if (status.killed) {
|
53 |
-
conn.close();
|
54 |
-
return;
|
55 |
-
}
|
56 |
-
status.conn = conn;
|
57 |
-
conn
|
58 |
-
// we get httpfs for free with cors restrictions
|
59 |
-
// we get parquet downloaded as we load it
|
60 |
-
// https://duckdb.org/docs/api/wasm/extensions
|
61 |
-
// https://duckdb.org/docs/api/wasm/data_ingestion#parquet
|
62 |
-
.query(
|
63 |
-
`LOAD parquet;LOAD httpfs;SELECT * FROM '${datasetUrl}' LIMIT 10`
|
64 |
-
)
|
65 |
-
.then((result) => setDataset(arrowResultToJson(result)))
|
66 |
-
.catch((err) => {
|
67 |
-
console.error(err);
|
68 |
-
setError(err.message);
|
69 |
-
setDataset(null);
|
70 |
-
})
|
71 |
-
.finally(() => {
|
72 |
-
conn.close();
|
73 |
-
setLoading(false);
|
74 |
-
});
|
75 |
-
});
|
76 |
-
|
77 |
-
return () => {
|
78 |
-
status.killed = true;
|
79 |
-
if (status.conn) {
|
80 |
-
console.log("Closing connection");
|
81 |
-
status.conn.close();
|
82 |
-
}
|
83 |
-
};
|
84 |
-
} else if (db) {
|
85 |
-
console.log("Resetting db");
|
86 |
-
db.reset();
|
87 |
-
}
|
88 |
-
}, [db, datasetUrl]);
|
89 |
|
90 |
const handleSubmit = (e: React.FormEvent<HTMLFormElement>) => {
|
91 |
e.preventDefault();
|
@@ -111,7 +55,11 @@ function App() {
|
|
111 |
<Button
|
112 |
className="w-full"
|
113 |
type="submit"
|
114 |
-
disabled={
|
|
|
|
|
|
|
|
|
115 |
>
|
116 |
Load
|
117 |
</Button>
|
@@ -130,7 +78,7 @@ function App() {
|
|
130 |
variant={"destructive"}
|
131 |
onClick={() => {
|
132 |
setNextDatasetUrl("");
|
133 |
-
|
134 |
setDatasetUrl("");
|
135 |
}}
|
136 |
>
|
|
|
1 |
+
import { useCallback, useState } from "react";
|
2 |
import { Table } from "./table";
|
|
|
3 |
import { Button } from "~/components/ui/button";
|
4 |
import { Textarea } from "~/components/ui/textarea";
|
5 |
import { useSearchParams } from "~/useSearchParams";
|
6 |
import { useDuckDb } from "~/useDuckDb";
|
7 |
+
import { useParquetTable } from "~/useParquetTable";
|
8 |
|
9 |
const DEFAULT_DATASET_URL =
|
10 |
"https://huggingface.co/datasets/openai/openai_humaneval/resolve/main/openai_humaneval/test-00000-of-00001.parquet";
|
11 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
const usePersistedTextfield = (fieldName: string) => {
|
13 |
const { searchParams, setSearchParams } = useSearchParams();
|
14 |
const searchParamsObj = new URLSearchParams(searchParams);
|
|
|
23 |
};
|
24 |
|
25 |
function App() {
|
|
|
26 |
const [datasetUrl, setDatasetUrl] = usePersistedTextfield("datasetUrl");
|
27 |
const [nextDatasetUrl, setNextDatasetUrl] = useState(() => datasetUrl);
|
|
|
|
|
28 |
const db = useDuckDb();
|
29 |
+
const { loading, dataset, error, clearDataset } = useParquetTable(
|
30 |
+
db,
|
31 |
+
datasetUrl
|
32 |
+
);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
|
34 |
const handleSubmit = (e: React.FormEvent<HTMLFormElement>) => {
|
35 |
e.preventDefault();
|
|
|
55 |
<Button
|
56 |
className="w-full"
|
57 |
type="submit"
|
58 |
+
disabled={
|
59 |
+
!nextDatasetUrl ||
|
60 |
+
(nextDatasetUrl === datasetUrl && !!dataset) ||
|
61 |
+
loading
|
62 |
+
}
|
63 |
>
|
64 |
Load
|
65 |
</Button>
|
|
|
78 |
variant={"destructive"}
|
79 |
onClick={() => {
|
80 |
setNextDatasetUrl("");
|
81 |
+
clearDataset();
|
82 |
setDatasetUrl("");
|
83 |
}}
|
84 |
>
|
src/useParquetTable.ts
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import { AsyncDuckDB, AsyncDuckDBConnection } from "@duckdb/duckdb-wasm";
|
2 |
+
import { useCallback, useEffect, useState } from "react";
|
3 |
+
|
4 |
+
// https://duckdb.org/docs/api/wasm/query#arrow-table-to-json
|
5 |
+
// TODO: import the arrow lib and use the correct type
|
6 |
+
// eslint-disable-next-line @typescript-eslint/no-explicit-any
|
7 |
+
const arrowResultToJson = (arrowResult: any) => {
|
8 |
+
// eslint-disable-next-line @typescript-eslint/no-explicit-any
|
9 |
+
return arrowResult.toArray().map((row: any) => row.toJSON());
|
10 |
+
};
|
11 |
+
|
12 |
+
export const useParquetTable = (db: AsyncDuckDB | null, datasetUrl: string) => {
|
13 |
+
const [loading, setLoading] = useState(false);
|
14 |
+
const [dataset, setDataset] = useState<null>(null);
|
15 |
+
const [error, setError] = useState<string | null>(null);
|
16 |
+
|
17 |
+
const clearDataset = useCallback(() => {
|
18 |
+
setDataset(null);
|
19 |
+
setError(null);
|
20 |
+
}, []);
|
21 |
+
|
22 |
+
const loadDataset = useCallback(() => {
|
23 |
+
if (db && datasetUrl) {
|
24 |
+
setLoading(true);
|
25 |
+
setDataset(null);
|
26 |
+
setError(null);
|
27 |
+
console.log("Loading", datasetUrl);
|
28 |
+
const status: { conn: AsyncDuckDBConnection | null; killed: boolean } = {
|
29 |
+
conn: null,
|
30 |
+
killed: false,
|
31 |
+
};
|
32 |
+
db.connect().then((conn) => {
|
33 |
+
if (status.killed) {
|
34 |
+
conn.close();
|
35 |
+
return;
|
36 |
+
}
|
37 |
+
status.conn = conn;
|
38 |
+
conn
|
39 |
+
// we get httpfs for free with cors restrictions
|
40 |
+
// we get parquet downloaded as we load it
|
41 |
+
// https://duckdb.org/docs/api/wasm/extensions
|
42 |
+
// https://duckdb.org/docs/api/wasm/data_ingestion#parquet
|
43 |
+
.query(
|
44 |
+
`LOAD parquet;LOAD httpfs;SELECT * FROM '${datasetUrl}' LIMIT 10`
|
45 |
+
)
|
46 |
+
.then((result) => setDataset(arrowResultToJson(result)))
|
47 |
+
.catch((err) => {
|
48 |
+
console.error(err);
|
49 |
+
setError(err.message);
|
50 |
+
setDataset(null);
|
51 |
+
})
|
52 |
+
.finally(() => {
|
53 |
+
conn.close();
|
54 |
+
setLoading(false);
|
55 |
+
});
|
56 |
+
});
|
57 |
+
|
58 |
+
return () => {
|
59 |
+
status.killed = true;
|
60 |
+
if (status.conn) {
|
61 |
+
console.log("Closing connection");
|
62 |
+
status.conn.close();
|
63 |
+
}
|
64 |
+
};
|
65 |
+
} else if (db) {
|
66 |
+
console.log("Resetting db");
|
67 |
+
db.reset();
|
68 |
+
}
|
69 |
+
}, [db, datasetUrl]);
|
70 |
+
|
71 |
+
useEffect(() => {
|
72 |
+
loadDataset();
|
73 |
+
}, [loadDataset]);
|
74 |
+
|
75 |
+
return { loading, dataset, error, clearDataset, loadDataset };
|
76 |
+
};
|