Finding and replacing characters using wildcards

Overview

The theory

The practice

Tips for advanced users

Gremlins to be aware of (for advanced users)

(to return to top, press Ctrl+Home)

1.

? and *

The two most basic wildcard characters are ? and *. These are essentially similar in use.

? is used to represent a single character and * represents any number of characters. On their own, these have limited use.

s?t will find sat, set, sit, sat and any other combination of three characters beginning with “s” and ending with “t”. It will also find that combination of letters within a word, thus it would locate the relevant (highlighted) part of inset etc.

s*t will find all the above, but will also find “secret”, “serpent”, “sailing boat” and“sign over document”, etc.

Note: Unlike ?, which always looks for the missing character, * will also find occurrences of the two adjacent characters (here “s” & “t”) with no text between them e.g. “streaker”. It is a blunt instrument and must be used with care, or with other characters to tie it down to only the text you require. There is no limit to the amount of text that the * character might match. It can end up matching all of your text in a multi-megabyte document!

2.

@

@ is used to find one or more occurrences of the previous character. For example, lo@t will find lot or loot, ful@ will find ful or full etc.

3.

< >

Used with any of the above (or any other combination of wildcards and characters), you can use the tags < and > to mark the start and end of a word, respectively. Thus, building on the example we used for the “*” character: <s*t> would find“secret” and “serpent” and “sailing boat”, but not “sailing boats” or “sign over documents”. It will also find “'set” in “tea-set” , but not “set” in “toolset”.

Again, beware of using “*”, as <s*t> will find any block of text from a word starting with “s” to the end of the next word in the document that ends with “t”, which may not always be what you had in mind.

The <> tags can be used in pairs, as above; or individually, as appropriate. For instance, ful@> will find “full” and the appropriate part of “wilful”, but will not find“wilfully”.

4.

[ ]

Square brackets are always used in pairs and are used to identify specific characters or ranges of characters. For example:

[abc] will find any of the letters a, b, or c. [F] will find upper case “F”.

[A-Z] will find any upper case letter; [0-9] will find any single number; [13579] will find any odd numbers; [0-9A-z] will find any numbers or letters.

You can use any character or series of characters in a range [ ], including the space character. Characters are processed in alphanumeric order – lowest first. If you are uncertain which character is lower than another, look in the Insert + Symbol dialog.

5.

\

If you wish to search for a character that has a special meaning in wildcard searches – the obvious example being “?” – then you can do so by putting a backslash in front of it:

[\?] will not find “\” followed by any character; but will find “?” instead.

If you wish to find the backslash itself then you need to precede that with a backslash [\\]. It's best if you always put the double-backslash into its own range, as in [\\] – sometimes it doesn't work, otherwise.

As previously mentioned, the following is a list of the characters that have a special meaning in wildcard searches ( [ ] { } < > ( ) - @ ? ! * \ ).

6.

[!]

[!] is very similar to [ ] except in this case it finds any character not listed in the box so [!o] would find every character except “o”. You can use ranges of characters in exactly the same was as with [ ], thus [!A-Z] will find everything except upper case letters.

7.

{ }

Curly brackets are used for counting occurrences of the previous character or expression.

{n} This finds exactly the number “n” of occurrences of the previous character (so for example, a{2} will find “aa”).

{n,} finds at least the number “n” occurrences; so a{2,} will find “aa” and “aaaa”).

{n,m} finds text containing between “n” and “m” occurrences of the previous character or expression; so a{2,3} will find “aa” and “aaa”, but only the first 3 characters in “aaaa” ).

Note: Counting can be used with individual characters or more usefully with sets of characters e.g. [deno]{4} will match done, node, eden); or with bracketed groups: (ts, ){3} will match ts, ts, ts, .

(Unfortunately, there is no wildcard to search for “zero or more occurrences” in Word wildcard searches; [!^13]{0,} does not work).

8.

( )

Round brackets have no effect on the search pattern, but are used to divide the pattern into logical sequences where you wish to re-assemble those sequences in a different order during the replace – or to replace only part of that sequence. They must be used in pairs and are addressed by number in the replacement e.g.

(John) (Smith) replaced by \2 \1 (note the spaces in the search and replace strings) – will produce Smith John

or replaced by \2 alone will give Smith.

Note: The placeholders \1, \2 etc., can also be used in the search string to identify recurring text. e.g.

Fred Fred could be written (Fred) \1.

Round brackets are perhaps the most useful aspect of complex wildcard search and replace operations.

9.

^

The ^ (“caret”) character is not specific to wildcard searches but it sometimes has to be used slightly differently from normal, when searching for wildcards.

In the Find and Replace dialog, if you click in the “Find what” and“Replace with” boxes, and click the“Special” button, you will see a list of supported special characters that you can use; select one and a code will be inserted in the box for you. The ones near the bottom of the list insert special ^ codes, such as ^tfor a tab. Once you know what the codes are, you can type them straight in without using the“Special” button.  Unfortunately, the range of options available from the “Special”button when you do a wildcard search is more limited than in an ordinary search. Some notable examples, and their workarounds:    You may wish to identify a character string by means of a paragraph mark . The normal search string for this would be^p. ^p does not work in wildcard search strings! It must, however, be used in replace strings, but when searching, you must use the substitute code^13.

For most other special characters and objects, you can use the same codes as in simple searches (^l = manual line break, ^g = graphic, ^^ = caret ...).

In many cases, characters can be addressed using their character numbers. The obvious example is ^13 for “paragraph”, as mentioned above. Unfortunately, this has not been implemented in a completely predictable or reliable manner, and if you have a need to search using character numbers, ensure that the search works with a test sample before committing to it.

John Smith

Brian Jones

Alan Williams

At its simplest, we have here two words – John and Smith. They can be represented by<*>[space]<*> – where [space] represents a single press of the spacebar.

Add the round brackets (<*>)[space](<*>) and replace with \2[space]\1

Run the search on the column of names and all are swapped. Run it again and they are swapped back.

If some of the names contained middle names and/or initials, you would first have to convert your table to text (separated by paragraph marks). Select Table + Convert Table to Text.

Or if there is more than one column in your table, paste your “Name” column into a new document, and then convert that column to text.

You could then replace:
(<*) ([! ]@)^13

with:
\2, \1^p

This would convert:

Finally, convert the text back to a table. (Select Table + Convert Text to Table).

If there was more than one column in your original table, then when converting the text back to a table, be sure to select “Paragraphs”, where it says “Separate text at”. Then, in your original table, delete the old column and paste in the new one.

[0-9] looks for any single digit number, but dates can have two numbers; so to restrict that to two, we use the “count” function {}. We want to find dates with 1 or 2 numbers so:
[0-9]{1,2}

Next bit is the ordinal “th” – Ordinals will be “nd”, “st”, “rd”, or “th” so identify those letters specifically:
[dhnrst]

There will always be two letters, so restrict the count to 2:
[dhnrst]{2}

Next comes the space. You can insert a space with the spacebar [space].

The month always begins with one of the following capital letters – AFJMNSOD. We don't know how many letters this month has so we can use the blanket “*” to represent the rest. And we are only interested in that word so we will tie it down with <> brackets:
<[AFJMNSOD]*>

There's another space [space] followed by the year. Years will always have four numbers so:
[0-9]{4}

Finally, add the round brackets to provide a logical break-up of the sequence:
([0-9]{1,2})([dhnrst]{2})[space](<[AFJMNSOD]*>)[space]([0-9]{4})

and replace with:
\3[space]\1,[space]\4

(where [space] represents pressing the spacebar once) to re-order the sequence.

US style manuals specify that ordinals should not be used in dates, but if you did want to keep the ordinals (so converting 7th August 2001 to August 7th, 2001), you could replace
([0-9]{1,2}[dhnrst]{2})[space](<[AFJMNSOD]*>)[space]([0-9]{4})

with:
\2[space]\1,[space]\3

You can use the same logic in reverse to convert August 7th, 2001 to 7th August 2001; or  to  convert August 7, 2001 to 7 August 2001. Unfortunately you can't addthe ordinals using a  wildcard Find & Replace, if they're not there to begin with – you would need to use a macro if you wanted to do that. 

(*^13)\1\1 will match any sequence of three identical paragraphs.

If you replace:

(*^13)\1

with

\1

it will delete all consecutive duplicate paragraphs in the document. Repeat until nothing is found, to delete all duplicate paragraphs in the document (as long as you have sorted the text first).

To delete duplicate rows in a table (provided you have no merged cells), you can convert the table to text (Table + Convert to Text, using a tab delimiter); delete the duplicate paragraphs using the above method, then convert the text back to a table.

You can paste any (Unicode) character (unfortunately not characters from decorative (Symbol) fonts) into your search expressions. So copying the first and last characters from the Greek or Cyrillic subsets into a search:

would match any Greek character:

matches any Cyrillic word:

(“Can you help me please?”)

Note: in Word 97, the characters sometimes display in the dialog box as squares, but they do work.

In Word 2000+, you can type in Unicode characters with the Alt-key (make sure NumLock is on, then hold down the Alt-key and type the numbers on the numeric keypad). Since all characters from decorative fonts (Symbol-, Wingdings-fonts ...) are kept in a special code page from &HF000 to &HF0FF, you can search for them with [Alt61472-Alt61695]. (See also Finding and replacing symbols). 

[a-c] will not only match “a”, “b”, “c”, but also:

[a-à] will not match all characters from U+0061 (a) to U+00E0 (à). 

For some discussion on Unicode sorting and wildcards see the Unicode Regular Expression Guidelines. In general, even if the sorting used in Word is not very transparent, you usually get the results you would expect.

If you are using VBA and doing string comparisons, the “Like” operator (covered in VBA Help) gives you much of the functionality of a wildcard search. 

If you are using VBA, you might want to look at the RegExp object (from VBScript, which you can include in your macro projects, and which offers some features not included in Word wildcards). Choose Tools + References in the VB editor, and check“Microsoft VBScript Regular Expressions”. You can get help if you call the VBScript editor and search its online help for RegExp. A list of supported wildcards is given in the help on the Pattern-property.

Sometimes Word will get confused if it encounters “escaped” brackets \( or \). For example:
(\\)
will match any character, not only a backslash.

Workaround: use
([\\])
instead.

([a-z]\() throws an error – it should find an “a(”.

Workaround: Use ([a-z][\(]) instead.

Word starts matching again after the previous match/replacement; so for example^13*^13 will match only every second paragraph in a Replace operation.

If that isn't what you want, you could (in this example) use *^13 instead. 

Other times, if this isn't what you want, you have to do multiple Find & Replaces. For example, supposing you wanted to subscript all the numbers in any chemical formula such as C2H4. You could tag the numbers that have to be subscripted; you want to subscript any group of numbers ([0-9]@) that follow a letter > ([A-Za-z]), and that are not followed by another number ([!0-9]): 

Find what: ([A-Za-z])([0-9]@)([!0-9])

Replace with: \1_\2\3 

That search will for example match “C2H” in “C2H4” (letter + number + non-number), and then continue searching after the “H”. So you need to run this replacement twice, before doing the Find and& Replace that applies the formatting.

But sometimes, Word finding every second instance of your search string may bewhat you want, and you can make use of the feature. (For instance, if you want to format the text between two identical tags).

Not a bug but still annoying: You have to escape any special character even if you type its code; so ^92 will have the same problems as typing the backslash.

The construction {0,} (find zero or more of the preceding item) is refused as incorrect syntax. This concept is available in Unix regular expression matching, so it's a curious omission.

You don’t always have to “escape” the special characters, if the context makes it clear that the special meaning isn’t wanted. [abc-] matches “-”, and [)(] matches “)” or“(”. This may sometimes make your searches behave differently from what you expected.

See also:
Flush bad karma from Word's find facility after an unsuccessful wildcard search
and
How to prevent the built-in BrowseNext and RepeatFind commands from creating bad karma for wildcard searches