pocketbase-import/README.md
michal-kapala 78f3dfc7bd add batch request chunking (#1)
+ multiple batch requests of size configurable via max_batch
2024-11-27 23:29:47 +01:00

119 lines
4.4 KiB
Markdown

# Overview
PocketBase data import tools for CSV and JSON files made using
[PocketBase JS SDK](https://github.com/pocketbase/js-sdk).
Automatically creates typed PocketBase collection and populates it with data.
Columns conflicting with PocketBase's autogenerated system fields (`id`,
`created`, `updated`; case-insensitive check, target column name's case is not
affected) are prefixed with `_`. Collection conflict will cause the import to
fail without any changes to the database.
No rules, options or constraints are set for the new collection (see the import
log for a full structure). You can modify them after the import from
PocketBase's dashboard.
## Types
`pocketbase-import` detects types using regular expressions. Currently supported
PocketBase types are:
- `Bool`
- `Number`
- `Plain text`
- `Email`
- `DateTime`
- `JSON`
- `Url`
# Configuration
Install the latest [Deno runtime](https://deno.com/) to run the scripts.
In the root directory create `.env` file with the following environment
variables:
- `ADMIN_EMAIL` (required) - superadmin email
- `ADMIN_PASSWORD` (required) - superadmin password
- `POCKETBASE_URL` (optional) - PocketBase app URL, defaults to local instance
Place your import files inside of `input` directory.
Make sure the target PocketBase instance is running and pointed to by
`POCKETBASE_URL`.
For versions `>=0.23.2`, enable and configure [batch API](https://pocketbase.io/docs/api-records/#batch-createupdateupsertdelete-records) access - go to *Settings > Application* in PocketBase dashboard (see below). Adjust the parameters to your performance requirements or use the defaults for small/middle-sized datasets.
![Batch API settings](https://github.com/user-attachments/assets/3205bf36-1e86-471b-a1b9-c28c6b118065)
# Options
You can change the default import options to your needs:
| Name | Files | Required | Description | Example use |
| --------- | -------- | -------- | ------------------------------------------------------------------------------------------------------------ | ------------------- |
| input | CSV/JSON | Yes | The name of the input file (with extension) | --input=example.csv |
| id | CSV/JSON | No | Indicates that `_id` column should be typed as plain text, the type is detected by default | --id |
| max_batch | CSV/JSON | No | Max batch request size in rows, should not exceed PocketBase's `Max allowed batch requests`. Defaults to 50. | --max_batch=100 |
| lf | CSV | No | LF (`\n`) EOL character will be used instead of default CRLF (`\r\n`) | --lf |
| delimiter | CSV | No | Column value separator, defaults to `,` | --delimiter=";" |
| quote | CSV | No | Value quote character, defaults to `'` | --quote="~" |
# CSV
The import is **not** multiline-safe, so if you have a file with strings
spanning across multiple lines the best option for you is to convert the input
file to JSON with tools like
[DB Browser for SQLite](https://sqlitebrowser.org/).
## Examples
Basic import (root directory):
```
deno run csv.ts --input=example.csv
```
Import without permission prompts and with max batch request size of 1 row:
```
deno run --allow-read --allow-env --allow-net csv.ts --input=example.csv --max_batch=1
```
Import without permission prompts and with `_id` column as text:
```
deno run --allow-read --allow-env --allow-net csv.ts --input=example.csv --id
```
Import with custom parser options (you need to adjust `example.csv`):
```
deno run csv.ts --input=example.csv --delimiter=";" --quote="~" --lf
```
# JSON
The required data format is an array of row objects.
## Examples
Basic import (root directory):
```
deno run json.ts --input=example.json
```
Import without permission prompts and with max batch request size of 1 row:
```
deno run --allow-read --allow-env --allow-net json.ts --input=example.json --max_batch=1
```
Import without permission prompts and with `_id` column as text:
```
deno run --allow-read --allow-env --allow-net json.ts --input=example.json --id
```