dfmt is a tool for converting data files between formats (JSON, YAML,
TOML, etc.).
Example usage:
dfmt convert in.json out.yamlFor command line options:
dfmt --helpCurrently, the following formats are supported
| Format | Input | Output |
|---|---|---|
| JSON | supported | supported |
| YAML | supported | supported |
| TOML | supported | supported |
| INI | supported | not supported |
| strings (by line or null-separated) | supported | not supported |
| character-separated fields (CSF) | supported | not supported |
For INI and CSF files only, an attempt at converting strings consisting of only finite numbers is made if the corresponding command line option is given. This may result in slightly different output such as missing surrounding spaces, rounding, etc.
YAML multi-document files are supported but they are treated as an array and will therefore be converted to a single document for all formats, including to YAML itself.
Many thanks to the authors of the following libraries used in this tool:
- https://github.com/BurntSushi/toml
- https://github.com/jawher/mow.cli
- https://github.com/go-ini/ini
- https://github.com/go-yaml/yaml
I needed a utility for conversions – mostly to JSON – and this is the second version of it, basically a port to golang. It does mostly what I need. As such it may be totally useless/unsuitable for any other purpose.
In particular, the tool is not meant to be optimized for speed or memory consumption. It will hold the data in memory between read and write.
Additional limitations:
-
The CLI is not stable and it is not suitable for scripting at this point.
-
Format-specific limitations on outputs apply and at the moment it is not possible to configure things such as the default keys for TOML output and INI input or case-(in)sensitivity of keys.
-
If strings are converted to numbers, an attempt at converting them to signed 64-bit integers is made. If that fails, they are converted to floats with rounding. If this, in turn fails, they will be kept as strings.
-
Non-finite floating point numbers (
+Inf,-Inf,NaN) are kept as strings even in transformation cases that parse numbers in input. -
CSF configured without a field delimiter and with the record delimiters corresponding to strings (NL or NUL) may behave like strings rather than as CSF input. "May" means this is undefined behaviour at this point. Different subcommands may have different behaviour and it may change over time.
Also check the Issues section for reported bugs and limitations.
Feel free to report bugs. Please include sample input to reproduce any issues if possible.
Additional features will be limited to what I need personally. The current plan is to include some transformations, mostly for data clean-up.