AG Filters Registry

This repository contains the known filters subscriptions available to AdGuard users. We re-host these filters on filters.adtidy.org. Also, these filters can be slightly modified in order to achieve better compatibility with AdGuard.

Introduction
What filters can be added to this repository
About filters
How to build filters and patches
Use cases

What filters can be added to this repository

We may add third-party filters to AdGuard Filters Registry. When making a decision about adding a third-party filter, we follow these rules:

The filter should be oriented towards browser content blockers.
The filter should be legal. If it has rules for paywall circumvention, we won't add such a filter.
The filter should have a place for receiving user complaints and holding discussions, such as a repository on github.com, or a website open to public.
The filter should be relatively popular, meaning:
- if there is a repository on GitHub, the number of stars should be at least 50
- if there is no repository on GitHub, the number of analyzed issues and discussions is estimated at 10 per month on the filter's website
- the filter should be actively supported for at least 6 months
The filter should be regularly updated with at least 10 updates per month.
The filter should be compatible with AdGuard products. You can familiarize yourself with AdGuard syntax here: https://adguard.com/kb/general/ad-filtering/create-own-filters/.
If the filter works only in some operating systems and satisfies all other criteria, it will be added but only for the supported platforms.
Previously added filters that haven't received any support for a year will be removed. We reserve the right to remove the filter earlier, depending on circumstances.
If the filter contains too many problematic rules, it will not be added. A rule is considered problematic if it causes false positives or otherwise displays unintended behavior. Decisions about filters with problematic rules are arbitrary and there may be exceptions (see items 9 and 10, for example).
- If the filter intentionally blocks or restricts access to any services for no reason other than being a reflection of the filter author's opinion, the filter will not get added, or will get removed if already added.
If the filter is popular in a specific region and there are no alternatives to it, then it can be added as is.
If the filter gets added, it receives a so-called trustLevel (Low, High, Full), based on the number of problematic rules it contains and some other factors. Filters without "Full" trust level may have part of their rules disabled.
- The trust level of a filter can be re-reviewed and raised if the author improves the filter over time.
If there are two or more similar filters that satisfy all other criteria, they all may be added if they don't duplicate each other and don't conflict with each other. If there is a large amount of conflicting or duplicate rules, the filter with more matches on such rules gets the priority.

About filters

Metadata

template.txt

Template file is used by the filters compiler to prepare the final filter version.
exclude.txt

A list of regular expressions. Rules that match these exclusions will not be included in the resulting filter.
metadata.json

Filter metadata. Includes name, description, etc.
- filterId — unique filter identifier (integer)
- name — filter name; can be localized
- description — filter description
- timeAdded — time when this filter was added to the registry; milliseconds since January 1, 1970; you can exec new Date().getTime() in the browser console to get the current time
- homepage — filter website/homepage
- expires — filter's default expiration period; used as filter update interval if "Default" is chosen for according setting in AdGuard product
- displayNumber — this number is used when AdGuard sorts available filters (GUI)
- groupId — group identifier
- subscriptionUrl — default filter subscription URL
- tags — a list of tags
- trustLevel — level of trust which describe allowed and permitted rules types; possible values:
  - low — only low-risk rule types are allowed; defaults to low if trust level is not configured at all
  - high — trusted third-party filter lists; some particular rules from there are still permitted
  - full — all types of filter rules are allowed; only AdGuard filter lists have full trust at the moment
- ignoreTrustLevel - note that this directive will allow ignoring the trustLevel for a specific internal file of a third-party filter
- platformsIncluded — the list of platforms to compile the filter for, e.g. ["mac", "windows", "android"]. If you need to compile the filter for all platforms remove this property
- platformsExcluded — the list of platforms to skip while filter compiling, e.g. ["ios", "ext_safari"]. If you need to compile the filter for all platforms remove this property
Note please that both platformsIncluded and platformsExcluded should not be set in filter's metadata simultaneously.
Metadata example
```
{
  "filterId": 2,
  "name": "AdGuard Base filter",
  "description": "EasyList + AdGuard English filter. This filter is necessary for quality ad blocking.",
  "timeAdded": 1404115015843,
  "homepage": "https://adguard.com/kb/general/ad-filtering/adguard-filters/",
  "expires": "4 days",
  "displayNumber": 1,
  "groupId": 1,
  "subscriptionUrl": "https://filters.adtidy.org/extension/chromium/filters/2.txt",
  "tags": [
    "purpose:ads",
    "reference:101",
    "recommended",
    "reference:2"
  ],
  "trustLevel": "full",
  "platformsIncluded": [
    "windows",
    "mac",
    "android",
    "ext_ublock"
  ]
}
```
revision.json

Filter version metadata, automatically filled and overwritten on each build.
filter.txt

Resulting compiled filter.
diff.txt

Build log that contains excluded and converted rules with an explanation.
trusted-rules.txt

This file contains rules that are excluded by the filters compiler based on the specified trustLevel.

Groups

/groups/metadata.json — filters groups metadata. Each filter should belong to one of the groups.

Optimization

For each filter, AdGuard compiles two versions: full and optimized. Optimized version is much more lightweight and does not contain rules which are not used at all or used rarely. Rules usage frequency comes from the collected filter rules statistics (thanks to the volunteers who enabled it in their AdGuard).

optimization_config.json - This file defines the target for the optimization process. AdGuard will attempt to compress the lists by removing the least frequently used rules until the compression goal (defined in percentages) is achieved.

NOTE: However, these changes will only take effect after being uploaded to the server, as filters-compiler does not use this file locally but retrieves it from the server.

Compiler customization

Script located in scripts/build/custom_platforms.js customizes the way filters are compiled for certain platforms. We should use it if we need to temporary change rules for a platform. In all other cases, we should prefer the default configuration. Below is a example of the configuration for the platform AdGuard for Chrome with comments:

"EXTENSION_CHROMIUM": {
    // Defines the platform for which the settings are specified.
    "platform": "ext_chromium",
    // Defines the path that can be used to access the settings or resources associated with this platform.
    "path": "extension/chromium",
    // Overrides the expires value set in the filter metadata (for this platform).
    "expires": "12 hours",
    "configuration": {
        // Sets an array of regular expressions that will be used to remove certain rules.
        "removeRulePatterns": [
            "^((?!#%#).)*\\$\\$|\\$\\@\\$",
            "\\$(.*,)?replace=",
            "important,replace=",
            "\\$(.*,)?app",
            "\\$network",
            "\\$protobuf",
            "important,protobuf",
            "\\$extension",
            ",extension"
        ],
        // Sets an array of objects that will be used to replace certain values.
        "replacements": [
            {
                "from": ":has\\(",
                "to": ":-abp-has("
            }
        ],
        // Specifies whether to ignore hints for rules. A value of "false" means that hints will not be ignored.
        "ignoreRuleHints": false
    },
    "defines": {
        "adguard": true,
        "adguard_ext_chromium": true
    }
},

Localization

If you want to help with filters translations, you can join us on Crowdin: https://crowdin.com/project/adguard-applications/en#/miscellaneous/filters-registry

Please learn more about translating our products: https://adguard.com/kb/miscellaneous/contribute/translate/program/

Templates

@include directive allows to include the content of specified file into the filter.

More information about the @include directive and its options can be found here: https://github.com/AdguardTeam/FiltersCompiler/#include-directive.

How to Build Filters and Patches

Install Dependencies
```
yarn install
```
Build Filters and Patches:

To build all filters, run:
```
yarn build
```
The yarn build command accepts two parameters:
- -i: Filters to build.
- -s: Filters to skip.
For example, to build only AdGuard filters:
```
yarn build -i=1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,224
```
It takes filter templates from the /filters path, loads their content, compiles filters based on the provided whitelist and blacklist, generates log and report files, and creates a copy of platform files for future patch generation in temp/platforms to compare old and new filters.

After every yarn build, yarn will execute the postbuild task, which checks all generated filters in the platforms/ folder. If there are no changes in the filter except for deleting Diff-Path (because the compiler doesn't retrieve this tag from metadata and deletes it on every compile run), the script will undo these unnecessary changes.
Build Patches

To build patches with different TTL (Time To Live) and selectively include or exclude specific filter IDs, use the following commands:
- For a TTL of 10 hours:
```
yarn build:patches --time=10 --resolution=h
```
- For a TTL of 60 minutes:
```
yarn build:patches --time=60 --resolution=m
```
- For a TTL of 3000 seconds:
```
yarn build:patches --time=3000 --resolution=s
```
- To include specific filter IDs:
```
yarn build:patches --include=1,2,3
```
  This includes only filters with IDs 1, 2, and 3 in the patch generation process.
- To exclude specific filter IDs:
```
yarn build:patches --skip=4,5
```
  This excludes filters with IDs 4 and 5 from the patch generation process.
This script recursively finds new filter files in the platforms/ folder, generates patches by comparing them with corresponding old filter files (from temp/platforms), and saves these patches in a designated directory. The --include and --skip parameters allow for selective processing of filters based on their IDs. After generating the patches, the script copies any existing old patches to the new filter directory and cleans up temporary files, facilitating the maintenance and updates of filters.
Validate Platforms

To validate filters.json and filters_i18n.json for platforms, use the following command:
```
yarn validate:platforms ./platforms
```
Validate Locales

To validate locales, use the following command:
```
yarn validate:locales
```
(Optional) Combined Validation

Steps 4 and 5 can be combined and run with a single command:
```
yarn validate
```

Use cases

Building AdGuard Filters and Updating Filters and Patches in the Repository

Install Dependencies
```
yarn install
```
Run the Build Process
```
yarn auto-build --mode adguard
```

Building All Filters and Updating Filters and Patches in the Repository

Install Dependencies
```
yarn install
```
Run the Build Process
```
yarn auto-build --mode all
```

Working with Locales

For information on working with locales, please refer to the Translations README.

Repository compression

About

Once a year, we will compress the repository to reduce its size. We will delete all remote branches and overwrite the master branch with a squashed history. The compression script will retain the first N commits in their original order in the history. All other commits (except the first one) will be squashed into a single commit.

How to

Squash all old commits
```
yarn install
yarn compress [commits_to_keep]
```
It will retain the first [commits_to_keep] (default is 10,000, which is approximately one year of history) commits, starting from now, in their original order in the history. All other older commits (except the very first one) will be squashed into a single commit.

Overwrite master branch:

git push --set-upstream origin --force master

List all remote branches:
```
git ls-remote --heads origin
```
Remove Remote Branches: Remove remote branches that are no longer needed locally and push the removal to the remote repository:
```
git push origin --delete branchName
```
Replace `branchName`` with the name of the branch you want to delete.
Prune Remote Branches: Use git remote prune origin to remove references to remote branches that have been deleted on the remote repository. This keeps your local repository in sync with the remote.
```
git remote prune origin
```
Clean the Reflog: Over time, Git can accumulate references in the reflog that are no longer needed. You can clean the reflog using the following command:
```
git reflog expire --expire=now --all
git gc --aggressive --prune=now
```
This will remove unnecessary entries from the reflog and perform garbage collection.

After this procedure git repository will reduce it's size.

DandelionSprout/FiltersRegistry