lambda CLI

A command line interface to Wikifunctions (or any other WikiLambda installation).

lambda is usable as a CLI tool in two ways: either calling it for a specific command or in interactive mode.

Installation

Clone the lambda CLI and set up the dependencies:

git clone https://github.com/vrandezo/lambda-cli.git
cd lambda-cli
git submodule init
git submodule update
npm install

Batch

The command line works in the following way:

src/lambda.js <parameters> <input>

Examples:

lambda --label --language Z1004 Z4
lambda --normal '"Hello"'
lambda --canonical '{ "Z1K1": "Z6", "Z6K1": "Hello" }'
lambda --evaluate 'and(true, false)'
lambda --noevaluate 'head(string to list("abc"))'

The input may need to be escaped from the shell, e.g. if it includes parentheses or quotes or other special character.

The following parameters are available:

• --[no]tokens: shows the results of the tokenizer
• --[no]delabel: shows the results of the delabeler
• --[no]ast: shows the results of the parser
• --[no]eval[uate]: performs an evaluation or not. The evaluation is done on the wiki, not locally
• --[no]raw: displays the raw result from the evaluation call
• --[no]normal: normalizes the result
• --[no]canonical: canonicalizes the result
• --[no]prettyprint: prints the result in the format used in the data directory
• --[no]label: replaces all ZIDs with labels
• --[no]format: uses a formatter to show the result in an easier to read way
• --[no]timer: displays how much time the call took
• --[no]metadata: displays the metadata from the call
• --file: followed by a path to a file to be loaded by the tool
• --config: followed by a path to a config file (see below)
• --language: followed by the language to use. Can be a ZID or a language code
• --wiki: URL to a wiki, or a path to a local directory
• --cache: path to a local directory

The input can be given in the following ways:

• in JSON: whenever it starts with ", {, or [
• as a ZID, and then it will be loaded from the configured wiki or local path
• in functional syntax, i.e. the name of a type or function, followed by comma-separated arguments in parentheses

Interactive

You can call the interactive mode by starting lambda without input data. You can use the parameters given above, and they overwrite whatever is in the config file, e.g.

src/lambda.js --language en

The command line interface per default runs:

• the standard parser. If parsing fails, respective errors are displayed and we stop.
• full validation. If validation fails, respective errors are displayed and we stop.
• evaluation, if it is a function call.
• the writer on what we received so far.

In the standard setting, you can start typing. If you start with a {, " or [, the interactive shell assumes that you are entering a JSON file. If you type in the label or ZID of an existing ZObject that object will be displayed. If you type in a function name or a type followed by a the values for the keys in a comma separated list enclosed in (), this will be understood as a function call or an instance construction.

The interactive shell has the following magic features:

• you can go up and down in a history, also across sessions
• you can type a . and then use any of the following commands
• .help: List all commands
• .break: Sometimes you get stuck, this gets you out
• .clear: Break, and also clear the local context
• .editor: Enter editor mode
• .exit: Exit the REPL
• .load: Load a local file as if you were inputting it
• .save: Save all evaluated commands in this REPL session to a file
• .version: Version number of the lambda CLI
• .language: show or set the natural language
• .tokens: tokenizes the input
• .delabel: turns the input labels into ZIDs
• .ast: shows the parsed call
• .evaluate: switches the evaluator on or off
• .raw: displays the raw result from the evaluator
• .normal: shows the normalizad version
• .canonical: shows the canonicalized version
• .prettyprint: return the exact format as used for the data directory in the WikiLambda repository
• .labelize: use a writer that shows the canonical version but labelized
• .timer: switch the timer on or off
• .metadata: switch the metadata on or off
• .wiki: URL of the wiki, or a local path to the data
• .label: if a ZID, returns the label, if a label, returns all matching ZIDs with their types.
• .reload: reloads a specific ZID. Without arguments, deletes all cache.

Many of the commands take either an argument or, if none is given, assume the result from the previous command as the argument.

Configuration

The config.json file allows to configure all the settings given above.

• language: natural language
• parser
• writer
• wiki: where to load the data from (a local directory or via http)
• which evaluator to run (local one or via http)
• cache: local file with cache

The values of the configuration parameters are either a single value, or a map from a short name to a value. In case of the latter, the first value is taken as the default value.

The following configuration keys are just booleans, either true or false.

• tokens: shows the results of the tokenizer
• delabel: shows the results of the delabeler
• ast: shows the results of the parser
• evaluate: performs an evaluation or not. The evaluation is done on the wiki, not locally
• raw: displays the raw result from the evaluation call
• normal: normalizes the result
• canonical: canonicalizes the result
• prettyprint: prints the result in the format used in the data directory
• label: replaces all ZIDs with labels
• format: uses a formatter to show the result in an easier to read way
• timer: displays how much time the call took
• metadata: displays the metadata from the function call

The configuration can be overriden through command line arguments which have the same name but are prefixed by a -- and preceded by a "no" if meant to be switched off.

lambda --labelize --language Z1002 --timer Z4

The plans of what this tool should be able to do is described in FUTURE_README.md