How can I convert a small CSV file to JSON without changing the source?

Read the CSV into memory, then write the result to a separate JSON path. Python’s csv.DictReader uses the header row as dictionary keys; its CSV fields are text values, not automatically inferred numbers or other types. [S1] The example below assumes your input is named input.csv, its first row contains headers, and it is UTF-8 encoded. Choose an output path that does not refer to the source file.

import csv
import json
from pathlib import Path

input_path = Path("input.csv")
output_path = Path("output.json")

if input_path.resolve() == output_path.resolve():
    raise ValueError("Choose a separate output path")

with input_path.open("r", newline="", encoding="utf-8") as csv_file:
    records = list(csv.DictReader(csv_file))

with output_path.open("w", encoding="utf-8") as json_file:
    json.dump(records, json_file, ensure_ascii=False, indent=2)

The input is opened for reading, and JSON is written to the separate output file. The Python tutorial documents json.dump for serializing a Python data structure to a text file and specifies UTF-8 for JSON files. [S2] This example assumes the CSV uses UTF-8; it does not detect the file’s encoding or determine whether its headers and values match your intended schema.

Keep them as strings unless you deliberately define and validate a conversion rule. Since CSV fields are read as text, a value such as 007 remains text rather than being silently treated as a number. [S1] JSON represents strings with quotation marks, so "007" and "12.50" show that the original text form was retained.

For example, with this hypothetical CSV:

item,code,price
notebook,007,12.50

the corresponding JSON is:

[
  {
    "item": "notebook",
    "code": "007",
    "price": "12.50"
  }
]

That example demonstrates text preservation, not a rule about what the fields mean. If you need price to be numeric, decide what formats and values are valid first, then convert only that field using an explicit rule. Do not infer a type just because a value looks numeric.

Use separate input and output paths, then reopen the output as JSON and compare the parsed records with the in-memory records read from the CSV. This checks that the file can be parsed and that its loaded data matches what the script was about to write; it does not prove that the CSV’s contents are correct for your purposes.

Add this after the writing step:

with output_path.open("r", encoding="utf-8") as json_file:
    saved_records = json.load(json_file)

assert saved_records == records
print(f"Verified {len(saved_records)} records")

The earlier path check stops the script if the resolved input and output paths match. Keep the output path separate from the source file, and inspect the output as well, especially if the CSV has unusual formatting or empty fields. The sources establish the CSV reading and JSON writing approach, but not that this workflow will interpret every CSV layout or schema as intended. [S1] [S2]

Replace input.csv and output.json with your chosen paths, keeping them separate. If your CSV uses a different encoding, use the encoding you know applies rather than assuming UTF-8; the example’s encoding is an explicit assumption. Before converting a field, write down the rule—for example, which values are valid and what should happen to blanks or unexpected text—then check each value against that rule. The CSV reader does not perform type conversion automatically. [S1]

A practical order is to first produce and verify a text-preserving JSON file, then, if needed, add a narrow conversion for a named field and check it on representative records. Keep the original CSV unchanged and review the converted output for the cases your rule allows. The example does not infer missing-value meaning, validate a schema, or decide whether a numeric-looking value is an identifier.