Professional Information Technology-Programming Book part 105 ppt

To demonstrate the use of string boundaries, look at the following example. Valid XML documents begin with <?xml> and likely have additional attributes (possibly a version number, as in <xml version="1.0" ?>). Following is a simple test to check whether text is an XML document: <?xml version="1.0" encoding="UTF-8" ?> <wsdl:definitions targetNamespace="http://tips.cf" xmlns:impl="http://tips.cf" xmlns:intf="http://tips.cf" xmlns:apachesoap="http://xml.apache.org/xml-soap" <\?xml.*\?> <?xml version="1.0" encoding="UTF-8" ?> <wsdl:definitions targetNamespace="http://tips.cf" xmlns:impl="http://tips.cf" xmlns:intf="http://tips.cf" xmlns:apachesoap="http://xml.apache.org/xml-soap" The pattern appeared to work. <\?xml matches <?xml, .* matches any other text (zero or more instances of .), and \?> matches the end ?>. But this is a very inaccurate test. Look at the example that follows; the same pattern is being used to match text with extraneous text before the XML opening: This is bad, real bad! <?xml version="1.0" encoding="UTF-8" ?> <wsdl:definitions targetNamespace="http://tips.cf" xmlns:impl="http://tips.cf" xmlns:intf="http://tips.cf" xmlns:apachesoap="http://xml.apache.org/xml-soap" <\?xml.*\?> This is bad, real bad! <?xml version="1.0" encoding="UTF-8" ?> <wsdl:definitions targetNamespace="http://tips.cf" xmlns:impl="http://tips.cf" xmlns:intf="http://tips.cf" xmlns:apachesoap="http://xml.apache.org/xml-soap" The pattern <\?xml.*\?> matched the second line of the text. And although the opening XML tag may, in fact, be on the second line of text, this example is definitely invalid (and processing the text as XML could cause all sorts of problems). What is needed is a test that ensures that the opening XML tag is the first actual text in the string, and that's a perfect job for the ^ metacharacter as seen next: <?xml version="1.0" encoding="UTF-8" ?> <wsdl:definitions targetNamespace="http://tips.cf" xmlns:impl="http://tips.cf" xmlns:intf="http://tips.cf" xmlns:apachesoap="http://xml.apache.org/xml-soap" ^\s*<\?xml.*\?> <?xml version="1.0" encoding="UTF-8" ?> <wsdl:definitions targetNamespace="http://tips.cf" xmlns:impl="http://tips.cf" xmlns:intf="http://tips.cf" xmlns:apachesoap="http://xml.apache.org/xml-soap" The opening ^ matches the start of the string; ^\s* therefore matches the start of the string followed by zero or more whitespace characters (thus handling legitimate spaces, tabs, or line breaks before the XML opening). The complete ^\s*<\?xml.*\?> thus matches an opening XML tag with any attributes and correctly handles whitespace, too. Tip The pattern ^\s*<\?xml.*\?> worked, but only because the XML shown in this example is incomplete. Had a complete XML listing been used, you would have seen an example of a greedy quantifier at work. This is, therefore, a great example of when to use .*? instead of just .*. $ is used much the same way. This pattern could be used to check that nothing comes after the closing </html> tag in a Web page: </[Hh][Tt][Mm][Ll]>\s*$ Sets are used for each of the characters H, T, M, and L (so as to be able to handle any combination of upper- or lowercase characters), and \s*$ matches any whitespace followed by the end of a string. Note The pattern ^.*$ is a syntactically correct regular expression; it will almost always find a match, and it is utterly useless. Can you work out what it matches and when it will not find a match? Using Multiline Mode ^ matches the start of a string and $ matches the end of a string—usually. There is an exception, or rather, a way to change this behavior. Many regular expression implementations support the use of special metacharacters that modify the behavior of other metacharacters, and one of these is (?m), which enables multiline mode. Multiline mode forces the regular expression engine to treat line breaks as a string separator, so that ^ matches the start of a string or the start after a line break (a new line), and $ matches the end of a string or the end after a line break. If used, (?m) must be placed at the very front of the pattern, as shown in the following example, which uses a regular expression to locate all JavaScript comments within a block of code: <SCRIPT> function doSpellCheck(form, field) { // Make sure not empty if (field.value == '') { return false; } // Init var windowName='spellWindow'; var spellCheckURL='spell.cfm?formname=comment&fieldname='+field.name; // Done return false; } </SCRIPT> (?m)^\s*//.*$ <SCRIPT> function doSpellCheck(form, field) { // Make sure not empty if (field.value == '') { return false; }