Microsoft word - m2ffaq3.doc

Opera’s M2 Filter Guide
Document Created: 12/15/03
Topics
• What is a filter?
• Advanced filtering with Regular Expressions What is a filter?
Opera offers several ways of finding messages that share the same characteristics. The access points for
example automatically group e-mails "To and From" your contacts, or e-mails with the same type of
attachment.
You can also create custom views, which have similar functionality but are user defined. Note that the
basic idea of Opera’s e-mail client M2 is that the access points and views do not contain the messages.
Instead, they provide shortcuts to the message. This makes it possible for a single message to appear in
multiple views.
For views that you create in M2 you can define filters so that particular messages are shown in that view.
Filters make it possible for M2 to sort and display your mail for you, using your predefined search criteria.
They are applied to new incoming e-mail as well as e-mail that has already been received.
The views are a useful addition to the existing access points, as the latter are somewhat limited. Also, it is
not recommended or very practical to look for certain messages in the Received folder. In general, one
will only need a few views to suit his or her needs. In some cases a user may require subviews to further
filter or breakdown messages.
In the following we will discuss how to setup a filter. We will furthermore mention some filtering rules and
show several examples of how this can be applied.
How does a filter work?
Each filter rule works as a true or false expression that is verified in any of the following portions of a
message (depending on your choice):
• Subject
• Any header (all of the aforementioned combined) • Entire message (which includes all headers and the message body). To make a filter you only need to know which combination of characters (or words) distinguishes the messages you want to see from all others. This search string may be part of the messages you are looking for or it may be found in the messages you don’t want to see. Hence there are two available options for matching messages: contains and does not contain. As soon as you have defined the search string, the filter may look like this: Match messages where subject contains important or Match messages where To header does not contain boss@mycompany.com. Setting up your first filter
Setting up filters for your views is not difficult. Once you have created a view by right-clicking on the Mail
Panel of Hotlist and have opened the view’s properties dialog (<ALT>+<ENTER>), you can see all available
options in the Filter tab.
The learning filter is not available in the filter dialog as of Opera v.7.23. Choose Add filter to add a new line to
the list of filters. (You can undo this by
clicking on the Remove filter button.)
The two dropdown boxes and the edit
field immediately become usable and let
you select a variable (message portion),
a matching option and a search string.
Changes are applied when you confirm
with OK. Before closing the dialog you
should consider using one or more of
the four available options which can be
turned on or off with checkboxes:
Hide messages from other views
Matching messages will not be shown in
any other view. This option makes it
possible to emulate the folder system of
other e-mail clients, but it will prevent
you from finding the matching messages
in any other way (including searches
and Unread). If a message matches the
filters of two views that have this option
enabled, it will be shown in both views.
Mark matching messages as read
All incoming messages that match the filter(s) will be automatically be marked as read. Apart from the e-
mail notification pop-up you will not any sign of received messages that you haven’t handled yet.
Inherit filter from parent
Only applicable to subviews. The filter(s) will only be applied to the messages that are already shown in
the parent view.
Filter existing messages
Filters are applied to new messages by default. If you also want to verify which old messages match the
search string, you should enable this option.
Notes:
a) Search strings are case insensitive. If you look for report, all messages containing report, Report,
REPORT or rEpOrT will be found. You will find the same messages if you look for RepoRT. b) Multiple filters for a single view should be combined with logical operators (AND, OR), of which AND will be processed first: filter 1 AND filter 2 OR filter 3 is logically grouped as (filter 1 AND filter 2) OR filter 3. If you need filter 1 AND (filter 2 OR filter 3) you should use filter 1 AND filter 2 OR filter 1 AND filter 3. Examples of simple filters
Here we will show some examples of useful filters that you should now be able to create.
Match messages where
subject contains important
This is the most simple type
of filter. All messages with a
subject containing the word
important are found.
Match messages where To
header contains @yahoo.com
All messages you sent to
people who have an Yahoo
account are found.
Match messages where entire
message contains Marie
All messages from and to anyone called Marie or with an e-mail address containing that name are found,
as well as e-mails in which she is mentioned in the subject line or in the message body and messages
with an attachment that has marie in its name.
Match messages where From header contains Fred@msn.com
OR where From header contains John@operamail.com
All messages you received from Fred or John are found. Match messages where To header contains Fred@msn.com AND where To header contains John@operamail.com All messages that were sent to both Fred and John are found (i.e. messages you sent to Fred and John, but also messages that someone else sent to Fred, John and you). Match messages where subject contains Re: AND where From header does not contain Anna@freemail.com OR where subject contains RE: AND where From header does not contain Anna@freemail.com This filter finds all replies in your Received folder, which are characterized by a subject line starting with RE: or re:, except those sent by Anna (you could exclude your own replies by using your e-mail address in the second and fourth line). Note that in this case it is necessary to repeat the second line due to the order in which the operators are dealt with. Match messages where From header contains service@ OR where From header contains support@ OR where From header contains admin@ OR where From header contains administrator@ OR where From header contains sysadmin@ OR where From header contains help@ OR where From header contains helpdesk@ This filter is useful if you want to group messages from service desks. Match messages where entire message contains password AND where entire message contains login Finds all e-mails containing login and password information. Match messages where any header contains Disposition-Notification-To All messages with a read receipt request are found. Advanced filtering with Regular Expressions
Until now we have only discussed filters that use either the including or the excluding filtering option
(contains or does not contain, respectively). There is one more option, though: matches regexp. A typical
filter of this type looks like this: Match messages where subject matches regular expression \w*\s(’02|’03).
A regexp (or Regular Expression) is a way of describing a class of texts. In the example above all messages with a subject containing a string of letters and/or numbers followed by a space and ’02 or ’03 will match the filter. Subjects that fall into this category are July ’03, xyz5 ’03 and ABC ‘0294738. Below is a list of characters you can use to make a regexp. Only the most frequently used functions are mentioned. Note that letters with a special function are preceded by a backslash; if you omit the backslash the special function will not be triggered. On the other hand there are characters that have a use in regexp and can only be interpreted as the character itself if the special function is turned off by the backslash. repeats the previous item 0 or more times repeats the previous item 1 or more times repeats the previous item between m and n times repeats the previous item m or more times
Examples of regexp:

\w+\s2002 May 2002 / happy 2002 05-02-2002 / May 5th, 2002 \we{2}n With this short explanation you should be able to interpret and create basic regexps. See the References and links for more extensive information. Examples of advanced filters
Below are some examples of filters that take advantage of the possibilities of regular expressions.
Text in message body
Match messages where entire message matches regular expression \n(\n.*)+subject
It is impossible to filter messages that contain the word subject, as all e-mail has a header called Subject.
And unfortunately you can’t make the filter look in the message body only. The filter above is still looking
in the entire message, but it skips the headers thanks to the regexp. The “trick” of this expression is that it
looks for the first blank line, as the headers are always separated from the actual text in this way.
High-risk attachments
Match messages where entire message contains attachment
AND where entire message matches regular expression filename=.+\.(exe|vb\w?|scr|ws\w|pif)
Do you want to see the messages with attachments that could be harmful? Then use this filter and add as
many extensions between the parentheses as you like.
Period of time
Match messages where any header matches regular expression Date:.+(1[5-9]|2[0-5])\sDec\s2001
This will only show messages sent or received between 15 December 2001 and 25 December 2001.
Time zones
Match messages where any header matches regular expression Date:.*(\+0[^1]00|-0\d00)
Separate your messages from different time zones (which could t the same time mean: from abroad) with
a similar regexp. In this case everything from outside the +0100 region will match the filter.
Multiple recipients
Match messages where any header matches regular expression To:.+@(.+@|(\n|.)+Cc:.+@)
Are you interested in e-mail sent to more than one person? This regexp tracks all To headers containing
multiple addresses and all messages with at least one address in both the To and the Cc header.
Jokes
Match messages where subject matches regular expression ([Ff][Ww][Dd]?:\s?){2,}
… where subject matches regular expression !{3,}
… where subject matches regular expression ([Jj]oke|[Ff]unny|[Ll]augh)
… where To header matches regular expression (@.*){5,}
… where entire message matches regular expression ([Ss]ubject(\n|.*)+){3,}
Do you often receive forwarded messages with jokes, funny pictures or links to interesting sites? You can
mark them with the Funny label, but you could also group them on arrival. Note that every single filter
mentioned above is likely to be too inaccurate for this purpose, so you will need to combine at least two of
them with the AND operator.
Filters you don’t need
Some filters can be easily created but aren’t necessary in M2.
Account selector
Match messages where any header contains Alex@operamail.com
If your e-mail address is Alex@operamail.com and you want to find all
messages you have received or sent with that account, you don’t need to
use the filter above. Instead, right-click on the Mail Panel in Hotlist, go to
Show account and select your account there.
Contacts
Match messages where To header contains Peter@aol.com
If you want to find messages that you have sent to or received from
someone, use the active contacts list (in the Mail Panel) or the Contacts
Panel in Hotlist. Double-clicking on one of your contacts will show you all
requested e-mails. You can moreover select whether you only want to
see messages To or messages From that contact in the mail toolbar.
Temporary searches
Match messages where entire message contains this week’s game
Some filters are only useful for a short period of time. Consider using the
search field in the Mail Panel. The results will be shown in a new access
point that you can delete when you don’t need it anymore. Doing it this
way saves you the trouble of making a new view.
Attachments
Match messages where entire message contains filename=
AND where entire message contains attachment
AND where any header does not contain text/plain
You can find messages with attachments using this filter, but why would
you create a view if you can also rely on the Attachments access points?
Are you looking for a specific file? Do a search like described in the
previous paragraph.
Mailing lists
Match messages where any header contains List-Id
Messages that have a List-Id header are placed into the Mailing lists
access point. As you receive additional messages from the same Mailing
list, they are appropriately grouped under the same access point for easy
viewing.
Filters you can’t make
There are a few filters that could be useful but are impossible or too impractical to make.
Messages older than…
Match messages where any header contains 2002
You can make a view like this to try and see old messages only. In this case all messages from 2002 will
be found, but you will also see messages with 2002 in the subject line or in any of the e-mail addresses.
Unfortunately you can’t make a similar filter for messages that are older than a month or a week. Even if
you use advanced filtering, there will be restrictions: the date limits of the search are always static, which
means that a view is only able to show messages from say November 12 to December 12, and unable to
show messages from the last month unless the date limits set by the user coincide with that period. This
implies that the user would need to make daily changes to the filter to keep it up to date.
Note: messages sent or received in the past week, month or year can be shown thanks to the drop-down
list in the upper right corner of the Mail Toolbar, which is set to Forever by default.
Combine the filters of the subviews
A filter can’t be set to combine the filters used in its subviews. Of course you can do this manually, but
this task may turn out to be too complicated.
Messages from all contacts in a folder
Match messages where any header contains Alex@operamail.com
OR where any header contains Peter@aol.com
OR where any header contains Fred@msn.com
OR where From header contains John@yahoo.com
You can’t set a filter to display all messages To and From contacts who are grouped in a folder in the
Contacts Panel, unless you explicitly insert each name from that folder into the filter.
Filter all unknown “contacts”
It is not possible to make a filter that looks for messages To or From people who are not in your contacts
list.
Attachment size
Match messages where entire message matches regular expression filename=(.|\n){10000,}
This filter finds all messages with attachments larger than approximately 10 000 bytes. Even though it is
perfectly possible to change the file size to e.g. 100 000 bytes, it is not recommended to use such a filter.
M2 will need too much time to apply the filter to existing and incoming e-mail. Instead, select one of the
Attachments access points and sort the messages by size (click on the column header in the message
list).
Street addresses and phone numbers
Match messages where entire message matches regular expression
\d+\s[a-zA-Z\s]+\n a-zA-Z \s]+,\s[A-Z]{2}\s\d{5} Due to the various methods and unpredictable format from recipients, it would be very difficult to filter
messages by address or phone number effectively. The example above will only filter addresses that are
written as a number followed by at least one word (the street name) and then on the next line at least one
word (city) followed by a comma, two capitals (state) and a zip code of five digits. All addresses that aren’t
written in this American format will be missed.
Active threads
Match messages where any header matches regular expression References:.*<oprz4s0f2n9@msn.com>
An Active Thread is automatically generated by M2 when you receive a reply from someone to a
message you have sent (this may happen during the current session or at the next startup). The original
sent message and replies will be grouped into this access point. This usually works better than filtering
messages with the same subject line, because Active Threads recognize the right messages even if the
subject line has been completely changed. E-mail belonging to an Active Thread is namely characterized
by the Message-ID and References header. You can’t make a filter with the same functionality, as
Message-ID’s are unique and rather unpredictable.
Spam filtering
M2 is able to recognize spam. There are two different methods it can use for this: the standard filter and
the learning filter1. The former is activated on installation of Opera and has three predefined Intern filtering
systems: Off, Medium and Strong. You can use any of them in conjunction with rules that you can add
just like with custom views. For effective spam filtering you will need to construct filter rules based on the
types of messages you receive. Below are some examples that you can build from.
Adult sites
Match messages where entire message contains must be 18
… where entire message contains adults only
There are of course many other words that are typical for spam, such as viagra or pr0n.
Message related sentences
Match messages where entire message contains If you no longer wish to
… where entire message contains received this e-mail because
… where entire message contains reply to this message with the word
… where entire message contains to be removed from
The sender often seems to think you requested spam, or that you will be naïve enough to confirm that
your e-mail address is valid by “opting out”.
100% guaranteed spam
Match messages where entire message matches regular expression 100%\s(satisf|guarantee)
… where entire message matches regular expression special\s(savings|offer|introductory)
… where entire message contains guaranteed approval
… where subject contains free!
… where entire message matches regular expression amazing\s(offer|promotion|results|sale|savings)
… where entire message matches regular expression order\s(now|today|your\s(risk\s)?free|risk\sfree)
Ways to filter wonderful offers.
Other
Match messages where entire message does not contain <font
… where entire message contains @www.
… where entire message contains /images/pdf/
… where From header contains .biz
Strange e-mail addresses
Match messages where From header matches regular expression ([a-zA-z]+[0-9]+)+@
Sometimes spam is received from an address similar to mskir9Wl2dx4fiJq@ehjeehcydf.com. It is risky to
use a regexp like the one in the filter above, as it may also match normal addresses. It is very difficult (if
not impossible) for a regular expression to distinguish these strings of random characters from readable
words or names. Consider filtering e-mail addresses that don’t contain any vowels ([^aeiouAEIOU\s]+@)
or that contain more than one number ((\d+[^0-9\s]+){2,}@).
Match messages where To header does not contain Your_address@email.com
and where Cc header does not contain Your_address@email.com
Even though these filters are very suitable for spam, you should keep in mind that messages that have
been sent to you as a blind carbon copy (Bcc header) also fit this description.
1 The learning filter can’t be enabled through Opera’s user interface in version 7.23; it requires editing configuration files (*.ini). It is expected that this type of filter will be available by default in later versions of Opera. Troubleshooting

How should I start making a filter?
First, examine two or more of the messages you want to see in your new view. Then try to discover what
makes them different from all other messages and specify the search string. After that you should make
the search string as short as possible. If you need to see messages that contain military expenditure
database, then one of these words will probably suffice for the filter. Or, if you are looking for e-mail
addresses in the message body, you will notice that the regular expression \n(\n.*)+@ will be as effective
as \n(\n.*)+\w+(\.\w+)?@\w+(\.\w+)?\.\w{2,3}.
Finally, make a choice between the filtering options contains, does not contain and matches regexp, of
which the latter is preferable if there are any uncertainties in your search string. Consider the differences
between the following: “all messages with subject World Championship” and “all messages with a subject
of two words”. In the first case you will most likely use a regular filter, in the second case you need a
regexp.

I created a filter, but it’s filtering every message into the view. What went wrong?
The most common reason is that your filter includes something that is common in all messages. Try to
refine your filter and/or add a does not contain rule to avoid false positives. You should also be aware of
the amount of information stored in the headers that you don’t see; there are many character combina-
tions that could match your search (copy a message with <c> and paste it into a text editor to see this).
I created a filter but it’s not finding any messages that I know exist. What could be missing?
Check your filter for typos and spelling errors. Make sure you are filtering the proper field. For example
you can’t filter a specific contact using the Subject field.
If your filter contains several
rules, you might try removing
some of them to isolate the
problem to a specific rule.
Start building the filters by
adding one rule at a time if
you’re unfamiliar with M2.
Also, keep in mind that M2
processes AND before OR
conjunctions.
If you are using regular
expressions, check the syntax
and use the program referred
to in one of the links found
later in this guide to verify you
are using the proper operators
for your criteria. A regexp
should be as short as possible
while matching as much infor-
mation as possible.
The logical operators I use
do not do what I want.
As stated before, AND goes
before OR. For example filter 1
AND filter 2 OR filter 3 is
logically grouped as (filter 1
AND filter 2) OR filter 3. If you
need filter 1 AND (filter 2 OR filter 3) you should implement filter 1 AND filter 2 OR filter 1 AND filter 3.
I have a complicated set of filters in mind, but I don’t know how to combine them efficiently.
Suppose you want to combine a set of filters in the following way: filter 1 AND (filter 2 OR filter 3 OR …). If
the number of rules is not too large you can solve this as described in the previous paragraph.
When you have more than four filters you should first try and reformulate them as to obtain a simpler set.
This may not be possible, though, and then you should consider creating a view that has only filter 1 and
then adding a subview with (filter 2 OR filter 3 OR …) – don’t forget to mark Inherit filter from parent. This
solution has some disadvantages but it may prove to be the best way to avoid mistakes in the filters.
Another workaround is to implement (filter 2 OR filter 3 OR …) in a regular expression so that the filter is
reduced to only two lines: filter 1 AND (filter 2 | filter 3 | …).
My filter is working fine except messages from a specific contact keep appearing in my view.
You can add a rule similar to AND where From header does not contain Lisa@example.com. This will
keep those specific messages from appearing.
I updated my filters but messages that matched the previous filters are still in my view.
You will need to remove the existing messages as the filter will not update and remove them for you.
Don’t delete the messages – they will be removed from all views if you do. (In case you already did: see
your trash folder and try to undelete the messages you wanted to keep.) Use <CTRL>+<x> instead. Then
apply the filter to existing messages.
I’m having trouble making my regexp work.
Keep regular expressions simple. You can make more complex filters as you gain experience. A very
useful tool istest your regexp on any text (see References and links). You
should preferably use one or more of the messages you would like to match with your filter.
A regexp does not recognize words: every line is seen as a string of characters. The expression and\w*
will also match wand, standing and chandelier. If you want to find any word starting with and, then insert a
space at the beginning of the expression: \sand\w*.
Don’t leave (white)spaces in your regexp. A space is represented by \s, for example \d\w{2}\s\w{3,}.
Regular expressions are case sensitive: and does not filter And. If you’re not sure about the use of
capitals in your e-mail, use something like this: [aA]nd.
Be aware of the special function of the backslash. It triggers the special use of letters, and it disables the
function of other characters. Are you looking for c:\windows in an e-mail? Use c:\\windows in your regexp.
Avoid using \W, \D and \S between brackets [ ] – it could be confusing.
Use the . (dot) as little as possible. Replace it by \w, \d, \s, \W, \D, \S, a bracketed definition [ ] or a
specific character if you can.
There is a difference between * and + but both are only valid until the end of the current line, so you
should try to predict line breaks (\n) if you want to extend your search to multiple lines.
The \w variable applies to both letters and digits. Letters only are described by [a-zA-z].
Every regexp in a filter can be interpreted as (\n|.*)*regexp(\n|.*)*. It is thus unnecessary to end or start a
regexp with .*, as in .*\d\d\s[jJ]an(uary)?.
References and links

Mail & News
- A comprehensive tutorial from Opera Software.
- A complete guide and tutorial by Mark Schenk.
- Read and post questions about M2.
Regular Expressions
n download to test and develop regular expressions.
- Examples and tutorials.
- A compact reference page.
Glossary
access point
any entry in the Mail Panel that shows a class of e-mail messages any person of whom the e-mail address has been stored in Opera a (set of) rule(s) / expression(s) that is only true for a part of all e-mail messages the information about the sender, recipients etc stored in an e-mail message Opera’s sidebar that consists of Panels such as the Mail Panel (toggle with F4) used to combine filtering rules: AND demands that both rules apply; OR allows that only one of them is true (or regexp) a string describing a class of texts a custom made access point that shows messages matching the view’s filter

Source: http://mata.wz.cz/temp/FilterGuide.pdf