SQL INSERT to Avro schema API

Code examples

These examples send the sample as multipart/form-data and save the returned bytes unchanged. Set TABLECONVERT_API_KEY in your environment.

curl --fail-with-body --max-time 40 \
  'https://api.tableconvert.com/v2/convert/sql-to-avro' \
  --header "Authorization: Bearer $TABLECONVERT_API_KEY" \
  --form-string 'data=CREATE TABLE records (id TEXT, event TEXT, country TEXT);
INSERT INTO records VALUES ('\''101'\'', '\''signup'\'', '\''DE'\'');
INSERT INTO records VALUES ('\''102'\'', '\''purchase'\'', '\''FR'\'');' \
  --output 'sql-to-avro.avro.json'

import os
from pathlib import Path
import requests

endpoint = "https://api.tableconvert.com/v2/convert/sql-to-avro"
headers = {"Authorization": "Bearer " + os.environ["TABLECONVERT_API_KEY"]}
fields = {
    "data": (None, "CREATE TABLE records (id TEXT, event TEXT, country TEXT);\nINSERT INTO records VALUES ('101', 'signup', 'DE');\nINSERT INTO records VALUES ('102', 'purchase', 'FR');"),
}

response = requests.post(endpoint, headers=headers, files=fields, timeout=40)
response.raise_for_status()
Path("sql-to-avro.avro.json").write_bytes(response.content)

import { writeFile } from "node:fs/promises";

const form = new FormData();
form.set("data", "CREATE TABLE records (id TEXT, event TEXT, country TEXT);\nINSERT INTO records VALUES ('101', 'signup', 'DE');\nINSERT INTO records VALUES ('102', 'purchase', 'FR');");

const response = await fetch("https://api.tableconvert.com/v2/convert/sql-to-avro", {
  method: "POST",
  headers: { Authorization: "Bearer " + process.env.TABLECONVERT_API_KEY },
  body: form,
  signal: AbortSignal.timeout(40_000),
});
if (!response.ok) {
  throw new Error("HTTP " + response.status + ": " + await response.text());
}
await writeFile("sql-to-avro.avro.json", Buffer.from(await response.arrayBuffer()));

Request parameters

Supported INSERT statements and their column lists are parsed as table text. The API does not execute SQL against a database.

Send one source: data, file, or url. The examples below use the source mode appropriate for this format.

Output options

Generate JSON containing a record schema and row data. This output is not an Avro object container file or binary Avro encoding.

This output format does not expose additional options.

Conversion rules

How the input is read

• Use explicit INSERT column lists and supported value tuples for predictable column mapping.
• This is text parsing, not arbitrary SQL query evaluation, schema migration or database access.
• Prepare computed expressions, dialect-specific literals and complex dumps before importing.
• Quoted identifier text and source literal semantics may need cleanup after extraction; verify nulls and escaping.

How the output is written

• Header names become record field names, and generated field types are string.
• The JSON has separate schema and data members; schema evolution and logical types require application review.
• Validate Avro naming rules, then use an Avro library if a real binary container is required.

When to use this API

Responses and errors

Check the response before retrying. A new HTTP request is a new conversion and may use credits again.

Limits, billing and versions

Local previews accept up to 200 KB. The current API implementation allows one uploaded file up to 50 MiB, limits each multipart text field to 1 MiB, and gives v2 conversion tasks a 30-second deadline. Deployment limits may be stricter. Read the returned error before retrying; some parser or renderer failures return HTTP 500.

Live calls follow the API credit policy. Failed conversions are refunded; a completed conversion is not undone if the download connection later fails.

References and format specifications

Format references explain the underlying syntax. The conversion rules and samples above describe TableConvert’s implementation.

• www.postgresql.org
• avro.apache.org