verb

Easy make mini-languages to do python things.

To install: pip install verb

Do things like this:

from verb import mk_executer

func_of_key = {
    "plus": lambda x, y: x + y,
    "minus": lambda x, y: x - y,
}
execute = mk_executer(func_of_key)
execute("3 minus 2 plus 1")
## 2
execute("9 minus 6")
## 3

A quick intro to Command

Uses cases: In situations where you want to get some input from a user (from the web, in a command line, etc.) that specifies a computation to be carried out, you know (right) that you definitely shouldn't resort to using eval or exec. Because it's dangerous for everyone involved -- let's just not go there.

verb offers an alternative: Easily building minilanguages that will allow the user to only execute the functions you choose, through a vocabulary you choose, and everyone can go home (as) safe (as you allow).

In a nutshell, you make a key-to-func mapping (or use the default). This func_of_key mapping is what specifies your interpreter:

from verb import *
import operator as o

func_of_key = {  # Note: Order represents precedence!
    "-": o.sub,
    "+": o.add,
    "*": o.mul,
    "/": o.truediv,
}

Now you have a minilanguage! Out-of-the-box it will allow you to "speak it in string" or "speak it in json/dict", but you can extend to enable the language to be written in any container you want.

If you give it a "command string":

command_str = "1 + 2 - 3 * 4 / 8"
command = Command(command_str, func_of_key)

command_str = '1 + 2 - 3 * 4 / 8' command = Command(command_str, func_of_key)

It will use func_of_key to both parse it and replace the keys with an indication that the corresponding function should be called. command is a callable object, and when you call it, it executes it's instructions:

command()

1.5

It may be useful to see what the operation structure looks like

d = command.to_dict()
d

{'-': ({'+': (1, 2)}, {'*': (3, {'/': (4, 8)})})}

# Or if you read better with indents

from functools import partial
import json
from lined import Pipe

print_jdict = Pipe(
    partial(json.dumps, indent=2), print
)  # Note: Only works if your dict is JSON-izable.

print_jdict(d)

{
  "-": [
    {
      "+": [
        1,
        2
      ]
    },
    {
      "*": [
        3,
        {
          "/": [
            4,
            8
          ]
        }
      ]
    }
  ]
}

That same dict can be used as a parameter to make the same command

command = Command(d, func_of_key)
command()

1.5

Example: Table Selector Mini Language

import operator as o
from typing import Callable, Mapping
from functools import partial

import pandas as pd
from lined import Pipe
from verb import str_to_basic_pyobj, Command


dflt_func_of_key_for_table_selection = {  # Note: Order represents precedence!
    "&": o.__and__,
    "==": o.__eq__,
    "<=": o.__le__,
    ">=": o.__ge__,
    "<": o.__lt__,
    ">": o.__gt__,
}


def mk_table_selector(
    table: pd.DataFrame,
    func_of_key: Mapping[str, Callable] = dflt_func_of_key_for_table_selection,
):

    def leaf_processor(x):
        x = str_to_basic_pyobj(x)
        if x in table:
            return table[x]
        return x

    run_command = Pipe(
        partial(
            Command.from_string, func_of_key=func_of_key, leaf_processor=leaf_processor
        ),
        lambda f: f(),
        lambda idx: table[idx],
    )

    return run_command

import pandas as pd

df = pd.DataFrame(
    [
        {"source": "audio", "bt": 5, "tt": 7, "annot": "cat"},
        {
            "source": "audio",
            "bt": 6,
            "tt": 9,
            "annot": "dog",
            "comments": "barks and chases cat away",
        },
        {"source": "visual", "bt": 5, "tt": 8, "annot": "cat"},
        {
            "source": "visual",
            "bt": 6,
            "tt": 15,
            "annot": "dog",
            "comments": "dog remains in view after bark ceases",
        },
    ]
)
df

run_command = mk_table_selector(df)

run_command("source == audio")

run_command("tt <= 8")

run_command("source == audio & tt <= 8")