Filtering

What are filters

With filter rules, you can manage incoming messages: decide which messages to publish and which not (for example, those containing advertising).

You can apply filters to all sources in a folder or to a specific source. You can use and combine any number of filtering rules.

Where to find

Menu → My forwards → Select source and destination → Filtering

Check out the filter debugging feature, which can potentially save you many hours.

‍

^New! AI-powered filtering

You can describe any filtering criteria in free form and let the artificial intelligence automatically filter your messages. Interestingly, this way you can filter not only by the text of the message itself, but also by its meaning and sentiment. For example, you can ask the AI to let through only positive news :)

How to set this up, see here.

‍

Allowing and blocking filters

On the filter settings page, you can choose what to do with messages that match the filter criteria.

‍

^{An allowing filter} (also known as a mustHave filter) will only let through messages that match the criterion. If you have one or more allow filters, but the message does not match any of them, the message will definitely be stopped. In other words, multiple mustHave filters will be combined in an "and" manner.

^{A deny filter} (also known as a mustNotHave filter) will stop messages that match the criteria. If you have one or more deny filters, and the message matches at least one of them, it will definitely be stopped. In other words, several mustNotHave filters will be combined according to the "or" principle.

The filter type is indicated in the first column: a green icon is an allow filter, and a red icon is a prohibit filter. When creating a new filter, check or uncheck the left checkbox to make the filter allow or deny.

‍

If you have multiple text allowing filters and multiple text denying filters, a message will be published only if it matches at least one allowing text filter and does not match any denying text filter.

‍

These combination rules work within a set of filters of the same type. That is, for example, if you have text-match allowing filters and regular expression allowing filters, then the message will be published only if it matches at least one text-match allowing filter and at least one regular expression allowing filter.

There are different types of filters: text match, regular expression, date, etc. Filtering occurs for each filter type separately. That is, first text filters are applied, then regular expression filters, etc. If, as a result of applying, for example, text filters, the message should be stopped, then it will be stopped, regardless of the result of applying other filters.

Text match filter

This is the simplest type of filtering. The criterion works on the principle of an exact text match, that is, the filter condition is triggered only if the exact text of the criterion is found in the message text.

Where to find

Menu → My forwards → Select source and destination → Filtering → Filters

To create a new text filter, select "By text match" at the bottom of the screen, click on the plus sign. In the "Value" field, enter the text and save.

‍

- Note, that the match filter is case sensitive. That is, if the criterion consists of the word "Bitcoin", it will not work if the text only contains "bitcoin".

- Remember that if the criterion text consists of several words, the message text must contain the entire criterion line, all the words and in that order and with those separators. That is, if the criterion consists of the line "Buy bitcoin", it will not work if the text only contains "Buy" or only "bitcoin" or, for example, "Buy or sell bitcoin".

- To create more flexible filtering rules, use regular expression filtering.

‍

‍Regular expression filter

In Junction Bot, you can set up complex filtering rules using regular expressions (Regular Expressions, RegEx).

Where to find

Menu → My forwards → Select source and destination → Filtering → Filters

To create a new text filter, select "By regular expression" at the bottom of the screen, click on the plus sign. In the "Value" field, enter a regular expression and save.

‍

Here are some popular examples:

1. Don't skip posts with the word word in the image caption

_{.*word.*<photo>.*}

2. Two words — bitcoin and ethereum — in a specific order

_{.*bitcoin.*ethereum.*}

3. Two words — litecoin and ethereum — in any order

_{.*(?=.*\blitecoin\b)(?=.*\bethereum\b).*}

5. Two words — bitcoin and litecoin — in a specific order, ignore case

_{(?i:.*BitCoin.*liTeCoin.*)}

6. Any of the words

_{.*bitcoin.*|.*litecoin.*|.*ethereum.*}

7. Rules can be combined. For example, any of the listed words, case is not important

_{(?i:.*BitCoin.*|.*liTeCoin.*|.*EthereUm.*)}

8. A word in Russian

_{.*([а-яА-Я]+).*}

9. Any link

_{.*(http:\/\/|https:\/\/).*}

10. Any number

_.*([0-9]+).*

11. Any @username

_{.*@([a-zA-Z0-9=&_\-]+).*}

12. Skip only images with captions (do not skip without captions)

_{\S+<photo>\s*}

13. Skip only images without captions

_{\s*<photo>\s*}

‍

IMPORTANT! ‍ -When filtering, the bot may add a space and/or media content code at the end of the text. Therefore, we recommend always putting _.* at the end of your RegEx filters.

- RegEx for filters usually has _.* at the beginning and at the end, because you want the entire message to match the expression. If you want to use such an expression for replacements, then most likely you want to remove these _.* (otherwise the replacement will work on the entire message text).

‍

Regular expression help

For more information on debugging regular expressions, see here.

‍

Filtering by all word forms

Sometimes you need to skip messages containing any form of a word. However, if this word is found in other words, then such matches do not need to be counted.

For example, we need to skip all messages with word forms of the word “way”, but not skip messages with the word “subway”.

Let's create a list of word forms of the word “way” and put each form in the regular expression \bword\b, which will skip only words that completely match these forms (and will not skip the word "subway"): _\bway\b, _\bways\b.

Add these 4 regular filters via Menu → My forwards → Select source and destination → Filters.

Done! Now only messages with the word “way” in any form will be published. However, messages with words like “subway” will not be published.

If you have a lot of filters, it is more convenient to collect them first in a table and then import them all at once into the system using the bulk import button.

‍

Filter by publication date/time

This filter type allows you to set rules for skipping messages depending on the time of their arrival.

The filter criterion is a time interval. It specifies:

The period when messages should arrive for allow filters (mustHave).
The period when messages should not arrive for deny filters (mustNotHave).

Where to find

Menu → My forwards → Select source and destination → Filtering → Filters → Create new: By date/time

‍

The time interval can contain the day of the week, time and time zone:

^{Day of the week} It always starts with the letter _w and then there are numbers corresponding to the days of the week.
For example:

_w135 - corresponds to the 1st, 3rd and 5th days of the week (Mon, Wed, Fri)

_w67 - corresponds to the 6th, 7th days of the week - days off.

^Time Time is always specified in 24-hour format, with an interval from - to, inclusive, in the format - hours:minutes.
For example:
_9:30-17:30
_0:00-23:59

^{Time zone} See the list of available time zones here. If no time zone is specified, Central European Time (Amsterdam/Berlin/Prague/Vienna) is used.

EXAMPLES

_0:00-7:00 - Messages will not arrive between midnight and 7am Central European Time.

_{w12345 8:00-20:00 Europe/Moscow} - Messages will only arrive on weekdays from 8am to 8pm Moscow time.

Alternative setup method via commands

_{/filtertime source_link mustHave w12345 8:00-20:00 Europe/Moscow}

‍

To add a date/time filter to all your sources at once, use the _{/filtertimeall} command.
‍
For example: _{/filtertimeall add mustNotHave 0:00-7:00}‍To remove: _{/filtertimeall remove mustNotHave 0:00-7:00}

‍

Delayed filter

_{Advanced filtering extension}

With this setting, you can:

— Delay the publication of a message for a fixed or random time,

— Specify the condition under which the message should be published (minimum number of ❤️ reactions, 👀 views, ✈️ forwards or 🗨️ replies).

Guide Delayed filters

How to set up delayed publishing

1. If you want to delay a message for a fixed time (for example, 3 hours), specify 3 hours in the "From" field. Let's say the message appeared in the source chat at 14:00. Then the message will be forwarded exactly 3 hours later, that is, at 17:00 (picture on the left).

2. If you set “From” and “To”, the message will be sent at a random time in the specified range. Let’s say the message in the source chat appeared at 14:00. With the setting “From 3 to 6 hours”, the message will be sent at a random time in the range from 17:00 to 20:00 (picture on the right).

How to set up a publishing condition

This condition is needed in cases when you want to receive in the destination chat only messages with a sufficient number of ❤️ reactions, 👀 views, ✈️ forwards or 🗨️ replies.

For example, you want to receive in the chat destination only messages that have received at least 2 reactions. To do this, you need to create a filter "Minimum reactions" with the value "2". We will leave the verification time interval the same - 3-6 hours.

‍

Let's say the message in the source chat appeared at 14:00. In this case, the bot will check the message as follows:

1st check - 3 hours after publication (at 17:00). If the message has 2 or more reactions, the message will be published. If the message has less than 2 reactions, the bot will schedule the next check.
2nd check - 6 hours after publication (at 20:00). If the message has 2 or more reactions by that time, the message will be published. If the message still has less than 2 reactions, the message will not be published at all.

If you check the box “Repeat check once an hour during the entire period”, the bot will check the message additionally at 18:00 and 19:00.

If the time range is less than 1 hour and is set in minutes or seconds, then the check will be performed only 2 times (at the beginning and at the end of the range).

These delayed filters let you forward only the most engaging messages — based on interaction-to-view ratios:

❤️/👁 Min. reactions to views
➡️/👁 Min. forwards to views
💬/👁 Min. replies to views

How it works: Enter the filter as a ratio (interactions ÷ views). Examples:

To copy posts with 1 reaction per 10 views → enter 0.1
To copy posts with 50 replies per 1,000 views → enter 0.05
To copy posts with 20 forwards per 100 views → enter 0.2

^New! Filtering order

This flexible setting allows you to determine the sequence in which filtering, replacements and AI processing are applied to your messages. The filtering order is applied to each forward separately.

Where to find

Menu → My forwards → Select source and destination → Filtering → Filtering order

Options

^{after reception} - just after the reception of the message, before applying replacement rules

Processing order: Filters → Replacement → AI processing
‍Filter by original text: the bot first filters, then applies replacements. This can be useful if you have complex replacement rules, so that replacements do not affect filters. Set the setting to _{after reception}.

‍

^{after replacements} - after replacement rules, but before AI processing

Processing order: Replacement → Filters → AI processing
Filtering after replacements: the bot first replaces the source text, and then applies filters. This is useful when you need to pass only posts with replaced links and then modify the text using AI.

‍

^{after ai} - after AI processing, just before publishing

Processing order: Replacement → AI processing → Filters
Filtering after AI: AI marks non-target content with a keyword, and the filter cuts it out.
How to set up:
‍1. Configure AI processing: in the forwarding rule settings, find AI processing and select the prompt “Ad Check” from Library. Edit it so that the AI replaces the text of irrelevant messages with a specific word, such as “SKIP.”
2. Add a blocking text filter for the word “SKIP.”
3. Set the Filtering order setting to _{after ai}.
‍4. Done! Now all ads from the source are filtered out, and only relevant content without spam reaches the destination.

Filter by author

Where to find

Menu → My forwards → Select source and destination → Filtering → Filters → Create new: By author

^{Filter by author ID} - This type of filter allows you to set rules for skipping messages depending on the number (ID) of the message author. For anonymous messages, the author number is the number of the corresponding group or channel. Please note that the criterion is the number, not the author's name.

Filtering by author number is the most effective and convenient method, we recommend using it. But sometimes you need to filter by author name.

To filter by ^{the name of the message author}, and not by its number, the procedure is a little more complicated:

Step 1

Menu → My forwards → Select source and destination → About source → Author

Enable the display of the author's name in the forwarding settings.

Step 2

Menu → My forwards → Select source and destination → Filtering → Filters → New filter "By text match"

Add a text filter containing the name as a criterion author.

With this approach, the resulting messages will contain the name of the message author. Sometimes it is necessary for the author's name not to be there. This will require a slightly more complex process - through an intermediate channel.

Suppose we need to redirect messages from A to B. We create channel C and set up redirection with filtering from A (source) to C (intermediate). We apply the settings from steps 1 and 2 to it. ‍Then, we set up redirection from C (intermediate) to B (destination). For this forwarding, we create a replacement that will remove the authors' names:

Menu → My forwards → Select source and destination → Modification → New replacement

Source: _{@user (Name)} - where @user is the author's username, and (Name) is the name displayed in Telegram.

New value: (empty)

Done! If you apply the replacement to the first forwarding (from A to C), it will not affect the author's name. The author's name, like the source channel name, is a service text, is not replaced with Replacements and is configured in other ways.

‍

Filter for your own messages

_{Direct or Expert tariffs}

This setting allows you to show or hide messages sent on behalf of the connected account.

Where to find

Menu → My forwards → Select source and destination → Filtering → My messages

Options

^pass- disabled, show personal messages with your authorship

^filter - filter (do not show)

‍

Filter by media type

Guide How to filter photos/videos (or any other media)

To filter by media message type, use one of the following text filters:

_<photo> - pictures and photos.

‍_<document> - file attachment

_<audio> - audio and music

_<video> - video

‍_<sticker> - stickers

_<voice> - voice messages

_<location> - geolocation tags

_<animation> - GIF animation

_<videonote> - video circles

_<poll> - voting

Where to find

Menu → My forwards → Select source and destination → Filtering → Filters → Create new: By text match

Filter forwarded messages

Guide How to filter forwarded messages

Using Junction Bot, you can filter Telegram messages that have been forwarded to the source channel from other channels or chats.

Where to find

Menu → My forwards → Select source and destination → Filtering → Filter forwards

Options

^all- allows you to send all messages, both regular and forwarded, without filtering

^{direct only} - allows you to send only regular messages to the destination

^{forward only} - allows you to send only forwarded messages to the destination

The setting to send only direct messages can be useful if a channel in your subscriptions contains a lot of information from different sources, but you want to take only the author's opinion from it without involving other authors. Thus, you select the _{direct only} option and filter out those messages that are forwarded to the channel on someone else's behalf.

‍

Filter every n-th message

You can set the frequency of forwarded messages, that is, forward only every n-th message.

You can do this by sending the bot the following command:_{/filter source_link every 2} (to forward only every second message from the source).

‍

Filter duplicates

Sometimes the same message can reach you multiple times (for example, if you are subscribed to two channels that forwarded the same post). To filter out duplicate messages, use the Duplicates setting.

Where to find

Menu → My forwards → Select source and destination → Filtering → Duplicates

Options

^pass- disabled, duplicates don't matter

^filter - don't publish duplicates

^{only_duplicates}- publish only duplicates

‍

The publish only duplicates setting can be useful if you have two sources and want to publish only posts that are the same in both of them. To do this, configure two forwards from the two sources to the intermediate channel, and from that channel another forwarding to the final channel with only duplicates published. In this way, only messages that have appeared in the intermediate channel at least twice will pass through.

The duplicate filter does not take into account buttons. That is, two messages with the same text but different buttons will be considered the same.

The duplicate filter remembers the last 300 messages in the source. That is, if 300 other messages have passed between identical messages, the bot will no longer remember the first of them and, therefore, will not filter the second. If there are no messages for 24 hours, the filter memory is reset.

‍

Please note that the duplicate filter compares the source messages. This means that if you want to avoid duplicates in the destination channel resulting from your replacement rules, you need to use a slightly more complex approach - through an intermediate channel. See the example below for how to do this:

Let's say we need to redirect messages from A to B.

Step 1

Create channel C and configure redirection from A (source) to C (intermediate). Apply all your replacement rules to channel A. You don't need to set a duplicate filter on channel A.

Step 2

Then set up a redirect from C (intermediate) to B (destination channel). Set a filter rule for this redirect (via the menu or with the command _{/settingchannel C duplicate filter}). It will remove all duplicates, even those that appeared as a result of replacement rules.

Step 3

Add our bot @junction_bot to the list of admins of the destination channel. You can give the bot any rights

Done! If you had set up a duplicate filter for the first redirect (from A to C), it would not have worked on duplicates resulting from replacement rules.

Note that the duplicate filter does not work with the forward_client message delivery mode setting.

‍

Creating a filter for all sources

To add a filter to all current sources, use the _/filterall or _/filterrxall command (for regular expressions).

For example: _{/filterrxall add mustNotHave .*(http:\/\/|https:\/\/).*}

This command will add to all sources filter that does not pass messages with links. To remove the same filter from all channels, replace _add with _remove.

Note that new redirects are always created with an empty filter list. You can copy filters from an existing source to them.

‍

Editing and deleting filters

The most convenient way to edit and delete filters is through the menu here:

Where to find

Menu → My forwards → Select source and destination → Filtering

‍

If you need to delete filters from all sources at once, use the _{/filterall DELETE ALL}

‍

Alternative way to configure via commands

To see all filters applied to a channel, use the command:
_{/filter source_link}

To remove a filter from one source, simply run the same command again. For example, if you added a filter with the command:
_{/filter source_link mustHave some_text}
then you can remove it with the command:
_{/filter source_link mustHave some_text}

If a filter is added with the command:
_{/filterall add mustHave some_text}
then you can remove it from only one source with the command:
_{/filter source_link mustHave some_text}

To remove all filters from all channels, use the command:
_{/filterall DELETE ALL}

‍