Ambiglyph Server - the implementation of the Ambiglyph project backend. it is fully written in Java with Spring technology thats provide powerful API.
It can detect homoglyphs inside a text and give suggestions to recover it using Levenshtein algorithm. The main approach is to use the algorithm as it used for misspells fixing. An user will be provided with suggestions how with which word obfuscated one can be replaced.
The server uses open database of words and homoglyps: http://homoglyphs.net/
and database of words: http://www.mieliestronk.com/corncob_lowercase.txt
Requirements
-
Java 17
-
MySQL Server 8
-
Maven 3.8.+
Run in the project root directory
mvn spring-boot:runRequirements
- Docker
Just run in the project folder:
docker build -t ambiglyph-server -f Dockerfile
docker run -p 8081:8081 -t ambiglyph-serverThis is simple backend application that uses REST API. Hence, it supports only REST requests. To try it out without any front application application you can use Inomina or Postman. All necessary documentation provided in Swagger (OpenApi v3). You can access it like http://api-address:8081/ambiglyph-server/swagger-ui/index.html.
Overall, the application supports CRUD operations on database of words and users. It is required to be authenticated and use given JWT-token in headers Some basic operations:
POST /authenticate/
Authentication. In case of successful authentication a JWT token will be provided as a response. Example:
Default passwords:
| Username | Password |
|---|---|
| ADMIN | ADMIN |
| test | test123BHOS |
| client ***** | client |
* This login can be used for unauthorized users to check using only the general word database. Recommended for client applications.*
POST /users/
Register a new user. The operation accessible only by users with role ADMIN and APP.
POST /words/
Add a new word to database
GET /words/
Get all words added by current user
GET /words/{id}
Get word with {id}
POST /check/
Send text to check.
All words that are detected to be spoofed will be replaced with <%%ambiglyph-detected>num<ambiglyph-detected%%>, where num - the number of a detection.
Also, if the number of suggested word is higher than given limit and there is a letter that is in the homoglyph database, the letter will be replaced with <%%ambiglyph-warning>num<ambiglyph-warning%%> , where num - the number of a warning.
In case of any detection or warning corresponding values in the response will be marked as true.
Example:
For more requests and their structure as well as their supposed output, please take into consideration controllers and DTO in source code.
The key point of separating database users is to add an ability to add words that they use regularly, but this words specific and it is impossible to add them to the shared database. Example:
There is no detection since the word wasn't found in database at all but it its obfuscate it. However, by adding the original word to database will help to address the problem.
