Regular Expressions for Translators: Anchors

Have you ever needed to find some text that appears at the beginning or at the end of a segment? How about some text that appears in the middle of a word? Regular expression anchors allow you to do just that.

Download Nora’s Regular Expressions for Translators Cheat Sheet

Here are a few examples where anchors are used to filter segments with SDL Trados Studio's display filter.

First, have a look at the unfiltered text.

In the first example below, I have used a simple regular expression to filter on target segments that have the string "tornillo" at the beginning. Notice that the word "tornillos" is also included, as there is no indication of a word boundary in the regex.

Now, I've filtered to display target segments that end in the word "perno".

The start of segment and end of segment anchors can be used together to enclose the entire contents of the segment, as shown below.

Next, I will use the word boundary anchor, which indicates where a word should start or end. In this example, I'm filtering on the word "tornillo" followed by a word boundary, which means that this won't match the word "tornillos".

When the word boundary anchor is used right before the string "tornill", the filter finds all instances of "tornillo" and "tornillos", but not "atornillada" (segment 4), as the string doesn't appear right after the word boundary in that instance.

The last anchor in the list is the non word boundary anchor. When I use it along with the word "tornillo", I find only instances where the word "tornillos" is found, as the regex means that "tornillo" should not be followed by a word boundary.

Using the non word boundary anchor right before the string "tornill" displays the segment that has the word "atornillada" in it, as the regex indicates that there should not be a word boundary right before "tornill".

As a last example, have a look at this regex that finds source segments that end in the string "bolt", which includes a segment that ends in the word "bolt" and another one that ends in the word "thunderbolt".

If I combine word boundary and end of segment anchors, the filter will display only the segment that ends in the word "bolt", as the word boundary anchors have excluded the word "thunderbolt".

I hope that these simple examples will inspire you to use anchors to find specific text in a variety of use cases. 

Regular Expressions for Translators: Four Applications in SDL Trados Studio

In regular expressions (regex), a new line break (or soft return) and a tab are represented with the following special characters:

Download Nora’s Regular Expressions for Translators Cheat Sheet

In a program like SDL Trados Studio, regular expressions can be used to:

                                        1) Filter on segments that match a certain regex

                                        2) Find text that matches a regex

                                        3) Create verification settings

                                        4) Add new segmentation rules to a TM

Let's use the new line and tab special characters to look at a few examples of these applications.

1) Filtering on segments that contain a new line break

This is achieved by using the regular Display Filter (found in the Review tab), which has regex enabled by default, or the Advanced Display Filter, where regular expressions must be enabled by checking a box.

Tip: For even more powerful filtering, download the Community Advanced Display Filter from the SDL app store.

So, if we have a document that looks like this:

Entering the "new line" regex character in the Display Filter search box produces the following filtered results:

2) Finding a tab

To do this, the regular expressions checkbox in the Find dialog box must be checked. This example shows the results of the search.

Once we learn that we can use regex in the Find dialog box, a natural question is whether the same can be done in a replace operation. The answer is a bit disappointing: while the Find field accepts all kinds of regular expressions, the regex syntax accepted by the Replace field is very limited, so, in short, no, you can't do the same in the replace field, that is, you can't replace a tab character with a new line character using regex, for example. In fact, if you enter "\n" in the replace field, that will be interpreted literally as "a backslash followed by an n", and that´s exactly what will be used in the replacement.

3) Creating verification settings

SDL Trados Studio's out-of-the-box verification options include the ability to add regex patterns to flag potential errors. In the example below, a rule has been created to tell Studio that when a new line character is found in the source, it should also be present in the target.

With the rule in place, once the verification is run, the program will identify any instances where there is a new line character in the source but not in the target.

4) Adding new segmentation rules to a TM

There are some cases where creating new pattern-based segmentation rules is desirable. A new segmentation rule for line breaks (soft returns), for example, would look like this (there's a dot in the "After break" section, even though it's hard to see):

After the rule has been added to the TM, files that are added to the project will be segmented at every line break, in addition to the usual segmentation. So, for our example above, if we remove the file from the project and add it back after the rule has been added, the new segmentation would look like this:

Final words
While the examples in this article use only the tab and new line characters, all kinds of complex regex patterns can be used in the four features that make use of regular expressions in Studio (display filter, search, verification and segmentation), and while linguists don't need to be computer programmers, investing some time to learn the basics of regex will help them save time and work more efficiently.

Regular Expressions for Translators: Escaping Metacharacters

If you’ve ever attempted to use a question mark or an asterisk in SDL Trados Studio’s display filter, you may have been surprised to get an error message that looks like this:

In fact, several of the characters below will trigger this message, while others will simply return unwanted results. 

You can test this by creating a simple Word file that contains these characters, opening it in Studio and attempting to filter using each character. I’ve indicated the results you’ll get below.

Why is this? Because the display filter has regular expressions enabled by default, and all of these characters have special meanings when creating regular expressions. To learn more about the meaning of each character, have a look at my Regular Expressions for Translators Cheat Sheet.

Download Nora’s Regular Expressions for Translators Cheat Sheet

So, does this mean you can’t use any of these characters in the display filter? Not exactly. All you need to do is “escape” each of these characters whenever you want them to be matched literally. You actually do this by using one of those metacharacters: the backslash.

The screenshot below shows that after escaping the question mark character in the display filter, there is no error message and the filter is properly applied, displaying only the segments that contain a question mark.

Metacharacters that don’t need to be escaped

As you may have inferred from the test above, there are three metacharacters that don’t really need to be escaped: {, < and >. This is because their special meaning only applies when they are used in very specific ways. However, while it may be important for a computer programmer writing code to avoid escaping metacharacters when it's not required, for a linguist looking to quickly filter content while translating, editing or proofreading this is not so critical, so if it’s hard to remember which metacharacters to escape and which not to, simply escape them all.

Final notes 

• SDL Trados Studio uses the .NET regex flavor
• While regular expressions are the default for the regular display filter in SDL Trados Studio, they are optional in the Advanced Display Filter and in the Community Advanced Display Filter
• Escape sequences can be used in Find operations but not in Replace patterns 
• To match a literal backslash, use another backslash to escape it: \\