Software package for comparative analysis of text message encoding methods in messenger communications

The study focuses on methods for encoding textual messages and binary data, including encrypted results, for transmission through the text channels of messaging applications. While a messenger receives a Unicode string from the user, it does not always deliver this string unchanged to the recipient: client applications may trim whitespace at the edges of messages, compress consecutive whitespace, replace character combinations with typographic symbols, interpret control characters as markup, links, or commands, and impose message length limitations. These alterations are typically unnoticed in casual correspondence, but a change in a single character renders encoded data irrecoverable. The research examines a 7-bit substitution table derived from a graduation thesis, as well as the windows-1251 and UTF-8 encodings, and Base64 and Base62 representations. These options are compared based on compactness, the completeness of the alphabet, and resilience to messenger client transformations. The approaches are described using a two-level model: character encoding and text representation of bytes. Compactness and distortion probability are assessed analytically. An experiment was conducted using a Node.js software suite with an emulator for client transformations, analyzing a corpus of 375 messages and 20,000 random messages of each length. The novelty of the study lies in formalizing the transformations of messenger clients as operations on strings, providing analytical estimates of distortion probability, and comparing encoding methods using a unified methodology. It is demonstrated that the 7-bit substitution table yields the shortest length but is irreversible, distorting 23-28% of the text characters. When encoding ciphertext with the substitution table, between 3.9% and 41% of messages are lost, depending on the length, which aligns with the analytical assessment. Base64url can lose up to 74% of long messages due to automatic replacement and markup, whereas standard Base64 and Base62 showed no failures. The combination of UTF-8 with Base64 or Base62 ensures completeness of the alphabet at the cost of increasing the length by 2.2 to 2.5 times, while windows-1251 with Base64 increases the length by a factor of 1.34. Block-based Base62 retains linear complexity.