Git Product home page Git Product logo

cligen's Introduction

cligen: A Native API-Inferred Command-Line Interface Generator For Nim

This approach to CLIs was inspired by Andrey Mikhaylenko's nice Python module 'argh' (in turn preceded by Plac ). The basic idea is that native language proc signatures already encode/declare almost everything needed to generate a CLI - names, types, and default values. A little reflection/introspection then suffices to generate a parser-dispatcher that translates seq[string] command input into calls to a wrapped proc. In Nim, adding a CLI can be as easy as adding one line of code:

proc fun(foo=1,bar=2.0,baz="hi",verb=false,paths: seq[string]): int=
  ## Some existing API call
  result = 1        # Of course, real code would have real logic here
import cligen; dispatch(fun) # Whoa..Just 1 line??

Compile it to fun (e.g., nim c fun.nim) and then run ./fun --help to get a minimal (but not so useless!) help message:

Usage:
  fun [optional-params] [paths: string...]
Some existing API call
Options:
  -h, --help                    print this cligen-erated help
  --help-syntax                 advanced: prepend,plurals,..
  -f=, --foo=    int     1      set foo
  -b=, --bar=    float   2.0    set bar
  --baz=         string  "hi"   set baz
  -v, --verb     bool    false  set verb

Other invocations (./fun --foo=2 --bar=2.7 ...) all work as expected. Default help tables work with automated "help to X" tools such as complete -F _longopt in bash, compdef _gnu_generic in zsh, or the GNU help2man.

cligen-erated parsers accept any unambiguous prefix for long options. In other words, long options can be as short as possible. In yet other words, hitting the TAB key to complete is unnecessary if the completion is unique.

When you want more specific help than set foo, just add parameter-keyed metadata with Nim's association-list literals:

dispatch(fun, help = { "foo" : "the beginning", "bar" : "the rate" })

That's it. No specification language or complex arg parsing APIs to learn. If you aren't immediately sold, here is some more MOTIVATION. Many CLI authors who have understood things this far can use cligen already. Enter illegal commands or --help to get help messages to exhibit the mappings.


Most commands have some trailing variable length sequence of arguments like the paths in the above example. cligen automatically treats the first non-defaulted seq[T] proc parameter as such an optional sequence. cligen applies the same basic string-to-native type converters/parsers used for option values to such parameters. If a proc parameter has no explicit default value, it becomes required input, but the syntax for input is the same as for optional values. So, in the below

proc fun(myRequired: float, mynums: seq[int], foo=1, verb=false) =
  discard           # Of course, real code would have real logic here
when isMainModule:  # Preserve ability to `import api`/call from Nim
  import cligen; dispatch(fun)

the command-line user must give --myRequired=something somewhere to avoid an error. Non-option arguments must be parsable as int with whitespace stripped, e.g. ./fun --myRequired=2.0 1 2 " -3".


dispatchMulti lets you expose two or more procs with subcommands a la git or nimble. Each [] list in dispatchMulti is the argument list for each sub-dispatch. Tune command syntax and help strings in the same way as dispatch as in:

proc foo(myRequired: int, mynums: seq[int], foo=1, verb=false) =
  ## Some API call
  discard
proc bar(yippee: int, myfloats: seq[float], verb=false) =
  ## Some other API call
  discard
when isMainModule:
  import cligen
  dispatchMulti([foo, help={"myRequired": "Need it!"}], [bar])

With the above in cmd.nim, CLI users can run ./cmd foo -m1 or ./cmd bar -y10 1.0 2.0. ./cmd or ./cmd --help print brief help messages while ./cmd help prints a comprehensive message, and ./cmd SUBCMD --help or ./cmd help SUBCMD print a message for just SUBCMD (e.g. foo|bar).

Like long option keys or enum values, any unambiguous prefix is accepted. So, in the above ./cmd f -m1 would also work. This is patterned after, e.g. Mercurial, gdb, or gnuplot. Additionally, long option keys can be spelled flexibly, e.g. --dry-run or --dryRun, much like Nim's style-insensitive identifiers, but with extra insensitivity to so-called "kebab case".


Rather than dispatching to a proc and exiting, you can also initialize the fields of an object/tuple from the command-line with initFromCL which has the same keyword parameters as the most salient features of dispatch:

type App* = object
  srcFile*: string
  show*: bool
const dfl* = App(srcFile: "junk")  # set defaults != default for type

proc logic*(a: var App) = echo "app is: ", a

when isMainModule:
  import cligen
  var app = initFromCL(dfl, help = { "srcFile": "yadda yadda" })
  app.logic()  # Only --help/--version/parse errors cause early exit

More Controls For More Subtle Cases/More Picky CLI authors

You can manually control the short option for any parameter via the short macro parameter:

dispatch(fun, short = { "bar" : 'r' }))

With that (and our first fun example), "bar" gets 'r' while "baz" gets 'b' as short options. To suppress some long option getting any short option, specify '\0' as the value for its short key. To suppress all short options, give short a key of "".


The default exit protocol is (with boolean short-circuiting) quit(int(result)) or (echo $result or discard; quit(0)). If echoResult==true, it's just echo $result; quit(0), while if noAutoEcho==true it's quit(int(result)) or (discard; quit(0)). The ors above are based on whether the wrapped proc has a return type or $ defined on the type. So,

import editdistance, cligen   # gen CLI for Nim stdlib editDistance
dispatch(editDistanceASCII, echoResult=true)

makes a program to print edit distance between two required parameters.

If these exit protocols are inadequate then you may need to call dispatchGen() and later call try: dispatchFoo(someSeqString) except: discard yourself. This is all dispatch itself does. Return types and values of the generated dispatcher match the wrapped proc. { Other parameters to generated dispatchers are for internal use in dispatchMulti and probably don't matter to you. } A dispatcher raises 3 exception types: HelpOnly, VersionOnly, ParseError. These are hopefully self-explanatory.


If you want cligen to merge parameters from other sources like a $CMD environment variable then you can redefine mergeParams() after import cligen but before dispatch/dispatchMulti:

import cligen, os, strutils
proc mergeParams(cmdNames: seq[string],
                 cmdLine=commandLineParams()): seq[string] =
  let e = os.getEnv(toUpperAscii(join(cmdNames, "_")))  # $MULTI_(FOO|_BAR)
  if e.len > 0: parseCmdLine(e) & cmdLine else: cmdLine # See os.parseCmdLine
dispatchMulti([foo, short={"verb": 'v'}], [bar])

You can also just include cligen/mergeCfgEnv between import cligen and dispatch to merge ${CMD_CONFIG:-${XDG_CONFIG_HOME:-$HOME/.config}}/CMD (with Nim stdlib's parsecfg module) and then $CMD with parseCmdLine as shown above, if that works for you.

Even More Controls and Details

After many feature requests cligen grew many knobs & levers. First there are more DETAILS on the restrictions on wrappable procs and extending the parser to new argument types. A good starting point for various advanced usages is the many examples in my automated test suite: test/. Then there is the documentation for the three main modules: parseopt3 argcvt cligen Finally, I try to keep track of possibly breaking changes and new features in RELEASE-NOTES.

cligen's People

Contributors

c-blake avatar clyybber avatar enthus1ast avatar genotrance avatar kaushalmodi avatar oskaritimperi avatar sspkrolik avatar timotheecour avatar

Watchers

 avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.