LeetTransformer :one::three::three::seven::robot:

• LeetTransformer :one::three::three::seven::robot:
  • Overview
  • Installation
  • Word camouflaging
    • LeetSpeaker
      • Parameters
      • Modes
      • Basic Use
      • Define your own changes
      • Uniform substitutions
      • Get all changes

Overview

---

Word camouflage is currently used to evade content moderation in Social Media. Therefore, this tool aims to counter new misinformation that emerges in social media platforms by providing a mechanism for simulating and generating leetspeak/word camouflaging data.

pyleetspeak2 provides the LeetSpeaker word camouflaging method.

• LeetSpeaker: This module apply the canonical 'leetspeak' method of producing visually similar character strings by replacing alphabet characters with special symbols or numbers. There's many different ways you can use leet speak. Ranging from basic vowel substitutions to really advanced combinations of various punctuation marks and glyphs. Different leetspeak levels are included.

Installation

---

pip install pyleetspeak2

Word camouflaging

---

LeetSpeaker

Canonical leetspeak in which standard letters are often replaced by numerals or special characters that resemble the letters in appearance

---

Parameters

You can see an example of use in a Heroku App:

https://user-images.githubusercontent.com/56938752/147962824-c347e184-14b6-41fe-8b05-ef670ac0a5f9.mp4

The only required argument that the user has to provide is the text_in argument which represent the casual text to transform to leetspeak. Nonetheless, there are other optional arguments that control the behaviour of the transformation:`

• change_prb determines the probability of a transformation to take place (i.e, if it is equal 1 all the possible transformation will be applied).
• change_frq is affects how frequently a transformation will occur (i.e, if it is equal 1 all the letters of this transformation type will be changed).
• mode controls the level of leetspeak transformation. Currently only basic mode is available. We are working on more modes. Stay tuned.
• seed controls the reproducibility of the results. By default no seed is applied.
• verbose controls the verbosity of the proccess.
• get_all_combs to obtain all the possible leetspeak versions of a casual text
• uniform_change determines if the same substitution character should be used in all the positions where the casual text will be modified.

Minor concerns about the package behaviour: accents are deleted before the substitutions are applied. This is important for languages like Spanish, where the word "melocotón" is preprocessed as "melocoton" and finally transformed to leetspeak.

---

Modes

There are several modes available:

• basic
• intermediate
• advanced
• covid_basic
• covid_intermediate

---

Basic Use

Let's see a simple working example:

from pyleetspeak2 import LeetSpeaker

text_in = "I speak leetspeak"
leeter = LeetSpeaker(
    change_prb=0.8, change_frq=0.6, mode="basic", seed=None, verbose=False
)
leet_result = leeter.text2leet(text_in)
print(leet_result)

For the sake of reproducibility you can set a random seed:

from pyleetspeak2 import LeetSpeaker

leeter = LeetSpeaker(
    change_prb=0.8,
    change_frq=0.5,
    mode="basic",
    seed=42,  # for reproducibility purposes
    verbose=False,
)
leet_result = leeter.text2leet(text_in)
print(leet_result)
# "1 sp34k leetsp3ak"

---

Define your own changes

pyleetspeak2 is prepared to apply substitutions defined by the user. It is essential to highlight that these new user-defined changes have to follow two possible formats, dictionary or List of tuples. Here we show a toy example to add two new target characters from the original text to be replaced by two and one different characters, respectively:

• Dictionary type:

  {"target_chr_1": ["sub_chr_1", "sub_chr_1"], "target_chr_2": ["sub_chr_1"]}
  

• List[Tuple] type:

  [("target_chr_1", ["sub_chr_1", "sub_chr_1"]), (("target_chr_2", ["sub_chr_1"])]
  

You can add new user-defined substitutions:

from pyleetspeak2 import LeetSpeaker

text_in = "New changes Leetspeak"
letter = LeetSpeaker(
    change_prb=1,
    change_frq=0.8,
    mode="basic",
    seed=21,
    verbose=False,
    get_all_combs=False,
    user_changes=[("a", "#"), ("s", "$")],  # user-defined changes
)
print(letter.text2leet(text_in))
# N3w ch@ng3$ L33t$pe4k

Moreover, you can use only the user-defined substitutions:

from pyleetspeak2 import LeetSpeaker

text_in = "Only user changes: Leetspeak"
letter = LeetSpeaker(
    change_prb=1,
    change_frq=0.8,
    mode=None, # None pre-defined changes will be applied
    seed=21,
    verbose=False,
    get_all_combs=False,
    user_changes = [("a", "#"), ("s", "$")], # user-defined changes
)
print(letter.text2leet(text_in))
# Only u$er ch#nge$: Leet$pe#k

---

Uniform substitutions

Usually, the same substitution character is used in all the matches for a specific substitution type. In other words, the same target character is usually replaced by the same substitution character. In order to reproduce this situation, pyleetspeak2 includes the uniform_change parameter that determines if all the matches of a target character are jointly or independently substituted. In the following example notice how the target character "e" is always replaced by "€" when uniform_changes is se to True.

from pyleetspeak2 import LeetSpeaker

text_in = "Leetspeak"
leeter = LeetSpeaker(
    change_prb=1,  # All subs type will occur
    change_frq=1,  # All matches of target chr will be changed
    mode="basic",
    seed=41,
    user_changes=[
        ("e", ["3", "%", "€", "£"])
    ],  # Add diferent subs characters for target chr "e"
    uniform_change=True,  # Use the same substitution chr for each target chr
)
print(leeter.text2leet(text_in))
# L€€tsp€4k

---

Get all changes

You can also obtain all the possible versions of a leetspeak text using the get_all_combs parameter like this:

from pyleetspeak2 import LeetSpeaker

text_in = "leetspeak"
leeter = LeetSpeaker(
    mode="basic",
    get_all_combs=True,
    user_changes = [("e", "€"), ("s", "$")], # user-defined changes
)
leet_result = leeter.text2leet(text_in)
print(len(leet_result))
assert len(leet_result) == 162 # all possible combinations
leet_result[20]
# 162
# 'le3t$p34k'

If you are only interested in the combinations that apply the same substitution character for each target target, you can also set uniform_change to True.

from pyleetspeak2 import LeetSpeaker

text_in = "leetspeak"
leeter = LeetSpeaker(
    mode="basic",
    get_all_combs=True,
    user_changes = [("e", "€"), ("s", "$")], # user-defined changes
    uniform_change = True
)
leet_result = leeter.text2leet(text_in)
print(len(leet_result))
assert len(leet_result) == 90 # all possible combinations
leet_result[60]
# 90
# 'le3t$peak'