Matching Text Patterns
Like locators, patterns are a type of parameter frequently required by Selenese commands. Examples of commands which require patterns are verifyTextPresent, verifyTitle, verifyAlert, assertConfirmation, verifyText, and verifyPrompt. And as has been mentioned above, link locators can utilize a pattern. Patterns allow one to describe, via the use of special characters, what text is expected rather than having to specify that text exactly.There are three types of patterns: globbing, regular expressions, and exact.
=> Globbing Patterns
Most people are familiar with globbing as it is utilized in filename expansion at a DOS or Unix/Linux command line such as ls *.c. In this case, globbing is used to display all the files ending with a .c extension that exist in the current directory. Globbing is fairly limited. Only two special characters are supported in the Selenium implementation:
* which translates to “match anything,” i.e., nothing, a single character, or many characters.
[ ] (character class) which translates to “match any single character found inside the square brackets.” A dash (hyphen) can be used as a shorthand to specify a range of characters (which are contiguous in the ASCII character set). A few examples will make the functionality of a character class clear:
[aeiou] matches any lowercase vowel
[0-9] matches any digit
[a-zA-Z0-9] matches any alphanumeric character
In most other contexts, globbing includes a third special character, the ?. However, Selenium globbing patterns only support the asterisk and character class.
To specify a globbing pattern parameter for a Selenese command, one can prefix the pattern with a glob:
label. However, because globbing patterns are the default, one can also omit the label and specify just
the pattern itself.
Below is an example of two commands that use globbing patterns. The actual link text on the page being tested was “Film/Television Department”; by using a pattern rather than the exact text, the click command will work even if the link text is changed to “Film & Television Department” or “Film and Television Department”. The glob pattern’s asterisk will match “anything or nothing” between the word “Film” and the word “Television”.
click link=glob:Film*Television Department
verifyTitle glob:*Film*Television*
The actual title of the page reached by clicking on the link was “De Anza Film And Television Department - Menu”. By using a pattern rather than the exact text, the verifyTitle will pass as long as the two words “Film” and “Television” appear (in that order) anywhere in the page’s title. For example, if the page’s owner should shorten the title to just “Film & Television Department,” the test would still pass. Using a pattern for both a link and a simple test that the link worked (such as the verifyTitle above does) can greatly reduce the maintenance for such test cases.
* which translates to “match anything,” i.e., nothing, a single character, or many characters.
[ ] (character class) which translates to “match any single character found inside the square brackets.” A dash (hyphen) can be used as a shorthand to specify a range of characters (which are contiguous in the ASCII character set). A few examples will make the functionality of a character class clear:
[aeiou] matches any lowercase vowel
[0-9] matches any digit
[a-zA-Z0-9] matches any alphanumeric character
In most other contexts, globbing includes a third special character, the ?. However, Selenium globbing patterns only support the asterisk and character class.
To specify a globbing pattern parameter for a Selenese command, one can prefix the pattern with a glob:
label. However, because globbing patterns are the default, one can also omit the label and specify just
the pattern itself.
Below is an example of two commands that use globbing patterns. The actual link text on the page being tested was “Film/Television Department”; by using a pattern rather than the exact text, the click command will work even if the link text is changed to “Film & Television Department” or “Film and Television Department”. The glob pattern’s asterisk will match “anything or nothing” between the word “Film” and the word “Television”.
click link=glob:Film*Television Department
verifyTitle glob:*Film*Television*
The actual title of the page reached by clicking on the link was “De Anza Film And Television Department - Menu”. By using a pattern rather than the exact text, the verifyTitle will pass as long as the two words “Film” and “Television” appear (in that order) anywhere in the page’s title. For example, if the page’s owner should shorten the title to just “Film & Television Department,” the test would still pass. Using a pattern for both a link and a simple test that the link worked (such as the verifyTitle above does) can greatly reduce the maintenance for such test cases.
=> Regular Expression Patterns
Regular expression patterns are the most powerful of the three types of patterns that Selenese supports.
Regular expressions are also supported by most high-level programming languages, many text editors, and a host of tools, including the Linux/Unix command-line utilities grep, sed, and awk. In Selenese, regular expression patterns allow a user to perform many tasks that would be very difficult otherwise.
For example, suppose your test needed to ensure that a particular table cell contained nothing but a number. regexp: [0-9]+ is a simple pattern that will match a decimal number of any length. Whereas Selenese globbing patterns support only the * and [ ] (character class) features, Selenese regular expression patterns offer the same wide array of special characters that exist in JavaScript. Below are a subset of those special characters:
Regular expressions are also supported by most high-level programming languages, many text editors, and a host of tools, including the Linux/Unix command-line utilities grep, sed, and awk. In Selenese, regular expression patterns allow a user to perform many tasks that would be very difficult otherwise.
For example, suppose your test needed to ensure that a particular table cell contained nothing but a number. regexp: [0-9]+ is a simple pattern that will match a decimal number of any length. Whereas Selenese globbing patterns support only the * and [ ] (character class) features, Selenese regular expression patterns offer the same wide array of special characters that exist in JavaScript. Below are a subset of those special characters:
Regular expression patterns in Selenese need to be prefixed with either regexp: or regexpi:. The former is case-sensitive; the latter is case-insensitive.
A few examples will help clarify how regular expression patterns can be used with Selenese commands.
The first one uses what is probably the most commonly used regular expression pattern–.* (“dot star”).
This two-character sequence can be translated as “0 or more occurrences of any character” or more simply, “anything or nothing.” It is the equivalent of the one-character globbing pattern * (a single asterisk).
click link=regexp:Film.*Television Department
verifyTitle regexp:.*Film.*Television.*
The example above is functionally equivalent to the earlier example that used globbing patterns for this same test. The only differences are the prefix (regexp: instead of glob:) and the “anything or nothing” pattern (.* instead of just *).
The more complex example below tests that the Yahoo! Weather page for Anchorage, Alaska contains info on the sunrise time:
open http://weather.yahoo.com/forecast/USAK0012.html
verifyText Present regexp:Sunrise: *[0-9]{1,2}:[0-9]{2} [ap]m
Let’s examine the regular expression above one part at a time:
A few examples will help clarify how regular expression patterns can be used with Selenese commands.
The first one uses what is probably the most commonly used regular expression pattern–.* (“dot star”).
This two-character sequence can be translated as “0 or more occurrences of any character” or more simply, “anything or nothing.” It is the equivalent of the one-character globbing pattern * (a single asterisk).
click link=regexp:Film.*Television Department
verifyTitle regexp:.*Film.*Television.*
The example above is functionally equivalent to the earlier example that used globbing patterns for this same test. The only differences are the prefix (regexp: instead of glob:) and the “anything or nothing” pattern (.* instead of just *).
The more complex example below tests that the Yahoo! Weather page for Anchorage, Alaska contains info on the sunrise time:
open http://weather.yahoo.com/forecast/USAK0012.html
verifyText Present regexp:Sunrise: *[0-9]{1,2}:[0-9]{2} [ap]m
Let’s examine the regular expression above one part at a time:
=> Exact Patterns
The exact type of Selenium pattern is of marginal usefulness. It uses no special characters at all. So, if one needed to look for an actual asterisk character (which is special for both globbing and regular expression patterns), the exact pattern would be one way to do that. For example, if one wanted to select an item labeled “Real *” from a dropdown, the following code might work or it might not. The asterisk in the glob:Real * pattern will match anything or nothing. So, if there was an earlier select option labeled “Real Numbers,” it would be the option selected rather than the “Real *” option.
select | //select | glob:Real *
In order to ensure that the “Real *” item would be selected, the exact: prefix could be used to create an exact pattern as shown below:
select | //select | exact:Real *
But the same effect could be achieved via escaping the asterisk in a regular expression pattern:
select | //select | regexp:Real \*
It’s rather unlikely that most testers will ever need to look for an asterisk or a set of square brackets with characters inside them (the character class for globbing patterns). Thus, globbing patterns and regular expression patterns are sufficient for the vast majority of us.
select | //select | glob:Real *
In order to ensure that the “Real *” item would be selected, the exact: prefix could be used to create an exact pattern as shown below:
select | //select | exact:Real *
But the same effect could be achieved via escaping the asterisk in a regular expression pattern:
select | //select | regexp:Real \*
It’s rather unlikely that most testers will ever need to look for an asterisk or a set of square brackets with characters inside them (the character class for globbing patterns). Thus, globbing patterns and regular expression patterns are sufficient for the vast majority of us.
No comments:
Post a Comment