Words away

Words-away

What is this thing for?

A text processing tool that prevents sensitive word detection to text. Mainly because Kuan's folding word is too sweet, **For example, "appeal" will also be folded (doge)**, so this tool is available.

The design goal is Cooan, which theoretically can be used by all communities that fully support Unicode, but the effect is uncertain. Actual tests have certain effects on NetEase Yidun's text detection API.

This is a web version, and there is still an Android version in the development stage.

principle:

Here are the principles of different options

• Normal mode: Insert Unicode zero-width spacer (U+200B) between two characters apart to circumvent detection.
• Double inversion mode for each line: Invert all texts in one line in sequence (physical) and add Unicode inversion control character (U+202E) to achieve enhanced evasion detection.
• Every two-word inversion mode: another enhancement mode, no explanation here~~(actually lazy)~~
• Convert to vertical arrangement: a purely "physical" method, does not use Unicode control characters, and only converts text to vertical arrangement.
• Avoid brackets and links: No processing is done when brackets and links are encountered to be compatible with emojis and links.
• Homomorphic letter replacement: Replace some ordinary Latin letters with letters with the same shape in other languages
• Link to short link: Use the is.gd API to replace http/https link in the text to avoid domain name detection of links
• Young Girl Code

Example

The following shows the general processing method

• Zero width interval

 "ABCD" => "Au200bBu200bCu200bDu200b"

• Double reversal for each line

 "ABCD" => "u202eDCBA"

• Double reversal of every two words

 "ABCDEF" => "u200eAu202eCu200bBu202cu200eDu202eFu200bEu202c"

If you look at the detailed principle, please search the source code (don't hit me if you don't write annotation)

privacy:

JavaScript-based web page implementation only runs on the browser side and will not pass data to the server ~ You can use it with confidence

Implemented/to be implemented:

• Enhanced Mode
• Automatically judge links and detours
• Vertical arrangement mode
• A better interface
• Android APP
• Restore mode

Known issues:

• Due to the characteristics of U+202E, after using the double inversion mode for each line, when a line of text is displayed in multiple lines, the order of those lines will be reversed, which cannot be solved for the time being.
• Due to the characteristics of U+202E, after processing in every two-word double inversion mode, when a line of text is displayed in multiple lines, there is a one-third probability that the last word of the previous line will be exchanged with the first word of the next line. Not yet resolved

use:

• Online web page
• Android apps

1. Open a web page or app
2. Enter or paste text to be processed
3. Select the corresponding option as needed.
4. Process, copy
5. Paste to Cooan and can be published normally (most cases)

Disclaimer:

Overly sensitive words will be detected even if they are reversed , so this tool is not omnipotent. In addition, although the Unicode control character will not be detected by the machine, the editor of Kuan himself said that using the Unicode control character to send serious violation messages will not be solved.
Therefore, although this is a good thing, do not abuse it, otherwise one day the Kuan algorithm can recognize the control symbols and it will be good for anyone.