Regular Expressions: features, filler, and repetitions
                  Stephanie Lockwood-Childs 9/24/09

* Common uses for regular expressions

  Match/no-match result
    Syntax validation

  Show matching entries
    Search patterns
    
  Modify matching entries
    Find and replace
    Strip out unwanted data
    General reformatting
  
* Demonstration problems

  Find and replace
  Capitalize all instances of "linux"
  
  Strip out unwanted data
  Equivalent of "basename" utility on list of paths
  Equivalent of "dirname" utility on list of paths
  
  Search patterns
  Find C functions returning type int
  
  Syntax validation
  First name, optional middle initial, last name
  
  General reformatting
  Convert dates from MM/DD/YY to YYYY-MM-DD format
  
* General approach
  Make up example inputs
  Pick out match features
  Create candidate expression
  Test on examples
  Refine until examples work
  Think of more devious examples
  Fix until devious examples work
  
- Make up example inputs
  
- Pick out match features
  Define match boundaries
  Mandatory and optional components
  Alternatives
  Repetition of components
  
- Create candidate expression
  Regular expression syntax
  Regex != Pathname expansion
  Basic syntax vs extended syntax: choices in back-slashification
  
         Basic requires slash    Meaning
    .              N             ANY character
    [:class:]      N             Character inside pre-defined set named "class" **
    [abc]          N             Character inside given set
    [^abc]         N             Character NOT inside given set
    (abc)          Y             Cluster into pattern group
    a|b            Y             Alternative patterns/groups
    a?             Y             Repeat count: occurs zero or once
    a*             N             Repeat count: occurs ANY number of times (including 0)
    a+             Y             Repeat count: occurs one or more times
    a{2,5}         Y             Repeat count: occurs some number within range
    ^abc           N             Anchor: beginning of line
    abc$           N             Anchor: end of line
  
  ** there is no character class named "class", merely my placeholder for the real classes
     of which the most important are: alpha, digit, alnum, lower, upper, blank, and space
  
  Reasons for grouping
  Apply count to whole group              (abc)?
  Use group as alternative                (abc)|(xyz)
  Remember group for replacement phase
  
  Match rules
  Matching starts from the left
  Greedy vs. non-greedy matching -- greedy by default
  
* Tips

  Develop complicated patterns in extended syntax then back-slashify if needed

  Choose carefully between "any number" and "one or more"
  echo "HUUMA" | egrep --colour "(U*)|Z"
  echo "HUUMA" | egrep --colour "(U+)|Z"

  Relevant man pages: "man 7 regex", "man perlre"