ADPS - Amateur Digital Post Service

Due to the simple design of this software there is no complex components for the indexing of the messages and the attachments. ADPS uses special file naming for quick searching of the attachments. So every file should contain the first 10 letters of the sha512’s hex output. Here is an example of the tree output:

You can see 3 attachment files with the partial collision. Technically it’s possible but the probability of this case is pretty low. If the attachment filename contains 10 hex letters, the folder contains up to 16^10 or 2^40 or about 10^12 files with unique partial hashsums. Files with different hashsums but mutual partial collisions slow down the speed of working, so maybe you need to refuse the message from another node if you suspect it tries flooding by forging these files with the same partial collision.

This filter search messages whose name field are equal to the query. Python implementation usage:

Examples:

This filter checks if the query is a substring of the message’s additional notes. The case is not important. Python implementation usage:

Examples:

This filter works like the previous one but for the inline message value of the message, so I’ll just copypase examples here =). Python implementation usage:

Examples:

It’s the main filter supposed to use. You should specify the central point and the radius. All messages in this "circle" will pass the filter. Since a coordinate field is a list, the filter is applied to every coordinate until the first match. Python implementation usage:

Examples:

This filter might be usefull when you take messages to the node associated with a big city. The idea is more populated city means higher probability that it has a courier that can go to the city you need. If in the previous city the input parameter are central point and radius, this filter depends on two other parameters: central point and base distance. By these parameters it calculates the probaility by the following formula: p = 2^(-distance/base_distance). The number range is from 0 to 1. After this the program calls random() which returns random number from 0 to 1. If the random() value is less than calculated function value the message considered as passed this filter. On C# implementation you can’t specify the base distance directly, instead it calculates by an empiric formula using population data from the simplemaps.com dataset. The empiric formula is: base_distance (meters) = population / 10. The filter has 5% threshold, so if the calculated probability is less than 0.05 the message automatically fails the filter. Let’s see the examples of the filter results:

In order to activate the filter in C# you need to fill the data of the Simple coordinate filter, after that enable Damping Distance Filter, click right mouse button on population input and click Find nearest city.

On the Python implementation you can specify the base distance directly or fill it empty, then it’ll search in the dataset the population like in C# implementation:

or if you want to use the dataset, just omit the --damping-distance-base-distance-meters parameter:

This filter is available only in C# implementation. The input data is maximum attachment size in bytes. It filters message when all attachments are less than input maximum attachment size.

Examples:

This filter lets see the messages associated with the attachment of the hashsum (or its part). This filter checks if the message hashsum starts with the query string. Python CLI usage:

This filter is enabled by default and filters all mails with the date_created field from 30 days ago to 3 days ahead. You can set other dates to the filter. Python CLI usage:

Python implementation:

C# implementation:

After the filtering files click Files → Save filtered mails to existing repository…​

If you delete some attachments and they were not unique by the partial sum, you should correct filenames in the following way:

For example if there are 6 files after removing operation with the same partial hashsum (eg 0123456789), the file list should contain the following filenames:

I need to introduce this rule in order to search attachments (a lot) more quickly. When we have a lot of files in the adps_attachments folder the OS operation get file list with the mask "0123456789*" from the adps_attachments directory takes a lot of time especially if we call this operation for every of thousands messages.

Python implementation:

C# implementation:

Todo: correct After the filtering files click Files → Delete filtered messages from the repository…​

Python implementation:

C# implementation:

Open the message (double click on message on filtered list), click the Export to folder button.

On the left-bottom corner you can choose one of 3 filtering modes:

On the page with the filtered mails you can remove the selected mails from the filtered list. Also you can invert this operation clicking Leave selected mails in the filtered mails list. Both selection modes are available on Selection menu or by right click on selected mails (context menu).

I’ve embedded 2 hard-coded translations to the program:

If you want see this program on another language you can add external translation file. To do it click Settings → Language → Save current translation to external file…​. After you saved the current embedded translation to an .xml file open with a text editor. Change the content of the tags, save and relaunch the program and you should see translated interface.

After this you can share you translation .xml file to other users who want to use this program on another language. In order to import external translation click Settings → Load existing translation file…​. This file should be accessble by the program every time so avoid importing from flash-media or a temporary folder.

You can change the language by clicking Settings → Language → Choose → <translation>

Every software has bugs. Not every user will use the Debug build with attached breakpoints on the Visual Studio. So the logging makes possible to diagnose the errors. It works by following scheme: Start capturing logs to the RAM → Using the SharpADPS and collecting logs → Dumping the captured logs from the RAM to a .txt file. The first and the last stages are controlled by Settings → Logging menu.

DotNet calculates sha512 very slowly. So I’ve attached openssl library to the program. Thanks to the openssl-net project I could integrate this library to the program. But this library was compiled for x86 and currently modern OS Windows 10 x64 can run this build with x86 openssl but it adds the requirement to platform to be x86-compatible. So I’ve added feature of switching between the internal .NET sha512 calculator and the external openssl one. You can choose .NET engine if you want to avoid the execution of non-dot-net code on the Settings → Hashsum Engine menu.

For most of the SharpADPS features described above you don’t need to set up every time you launch the application because it loads the last settings from the settings file. It can be located on different direcories (depends on OS). On Windows XP the file is accessible by C:\Documents and Settings\user\Application Data\WpfAdpsConfig.xml.