Verbs that act after backtracking and failure

by

Before reading this answer, you should be familiar with the mechanism of backtracking, atomic groups, and possessive quantifiers. You can find information about these notions and features in the Friedl book and following these links: www.regular-expressions.info, www.rexegg.com

All the test has been made with a global search (with the preg_match_all() function).

(*FAIL) (or the shorthand (*F))

baabo caaco daado

caac(*FAIL)|aa.|caaco|co

[0] => aab
[1] => caaco
[2] => aad

(*FAIL) causes exactly the same behavior as a “bad character” in a pattern. If you replace it with an “R”, for example, you obtain exactly the same result: caacR|aa.|caaco|co. To be more general, you can indeed replace (*FAIL) with an “always failing subpattern” like: (?!), (?=a(?<!a)),…

a (first from "baabo") : Without surprise, the first result is found by the second alternative. (aab)

c (first) : The regex engine encounters the first “c” and tries the first alternative and finds: caac, but the subpattern is forced to fail. Then the regex engine (always from the first “c”) tries the second alternative that fails, the third alternative succeeds. (caaco)

a (first from "daado") : the third result is found by the second alternative. (aad)

(*SKIP)

baabo caaco daado

caa(*SKIP)c(*FAIL)|aa.|caaco|co

[0] => aab
[1] => co
[2] => aad

This verb defines a point beyond which the regex engine is not allowed to backtrack when the subpattern fails later. Consequently, all the characters found before with the subpattern are consumed, once and for all, and can not be used for another part of the pattern (alternative).

a (first from "baabo") : the first result is found by the second alternative. (aab)

c (first) : the regex engine finds caac as in the first case, then fails (cause of the (*FAIL) verb), backtracks to the second “c” but is not allowed to backtrack to characters that are previously matched (“caa”) before the (*SKIP) verb.
c (second) : Now, the regex engine tries always the first alternative but in this new position and fails since there is an “o” and not an “a” after, then it backtracks to this second “c”. Note that in this case, these characters are not consumed as previously because the subpattern has failed before to have reached the (*SKIP) verb.
The second alternative is tested and fails (doesn’t begin with a “c”). The third alternative fails too because the next character is an “o” and not an “a”.
The fourth alternative succeeds and gives the second result. (co)

a (first from "daado") : the third result is found by the second alternative. (aad)

(*PRUNE)

baabo caaco daado

caa(*PRUNE)c(*FAIL)|aa.|caaco|co

[0] => aab
[1] => aac
[2] => aad

This verb is different from (*SKIP) because it doesn’t forbid using all the previous matched characters, but skips the first matched character by the subpattern (or forbids a subpattern to start with) if the subpattern will fail later.

a (first from "baabo") : the first result is found by the second alternative. (aab)

c (first) : the regex engine finds caac as in the first case, then fails, but now backtracks to the first “a” from “caaco” since the first “c” is skipped.
a (first from "caaco") : the first alternative is tried and fails, the second succeeds and gives the second result. (aac)

a (first from "daado") : the third result is found by the second alternative. (aad)

More Related Contents:

  1. string pattern matching using perl regex [closed]
  2. Non greedy (reluctant) regex matching in sed?
  3. Match whitespace but not newlines
  4. The recognizing power of “modern” regexes
  5. How do you debug a regex? [closed]
  6. Escaping a forward slash in a regular expression
  7. How can I use a variable in the replacement side of the Perl substitution operator?
  8. When is \G useful application in a regex?
  9. Why is a character class faster than alternation?
  10. “vertical” regex matching in an ASCII “image”
  11. What regex can match sequences of the same character?
  12. Is it possible to define a pattern and reuse it to capture multiple groups?
  13. How to extract string following a pattern with grep, regex or perl [duplicate]
  14. What’s the technical reason for “lookbehind assertion MUST be fixed length” in regex?
  15. Regex to match any character including new lines
  16. How do I handle special characters in a Perl regex?
  17. replace a part of a string with REGEXP in sqlite3
  18. Why do ^ and $ not work as expected?
  19. Why can’t you use repetition quantifiers in zero-width look behind assertions?
  20. How can I find the location of a regex match in Perl?
  21. How can I exclude some characters from a class?
  22. Can someone explain Possessive Quantifiers to me? (Regular Expressions)
  23. What is the difference between \1 and $1 in a Perl regex?
  24. How can I use a variable for a regex pattern without interpreting meta characters?
  25. Is There a Way to Match Any Unicode Alphabetic Character?
  26. Capturing text before and after a C-style code block with a Perl regular expression
  27. Can I substitute multiple items in a single regular expression in VIM or Perl?
  28. How to use different separators (/ , |) in a regular expression
  29. Counting number of occurrences of a string inside another (Perl)
  30. How to replace a string in an existing file in Perl

Leave a Comment