CMPUT 201, Winter 1996 - (C) David Southwell

Regular Expressions
c
^
$
.
[abc]
[d-k]
[^xyz]
Any character; use \c if a specialdenotes the start of the linedenotes the end of the lineany single charactera, b or c are matchedd,e,f,g,h,i,j or k are matchedany character not in the set...
Some single character matching codes...Operators...*xy\(...\)\n
After an expression, this willmatch n duplications (n can be 0)expression x followed byexpression ydelimits a subexpressionn : 0-9  \n matches the nth sub-expression matched this line
7
Regular Expressions...are used for pattern matching strings.W
e can use strcmp(char *, char *) for
precise string comparison, but very oftenwe would like to be able to find stringswhich conform to certain constraints.Regular expressions prescribe a patternin an array of characters.  W
e can tell
the system what our regular expressionis, and then ask it if particular stringsconform to it.They are found throughout the UNIXenvironment, either at the commandline (through GREP - Get REgularexPression), or through system callsfrom within user programs...as we shallsee...
CMPUT 201, Winter 1996 - (C) David Southwell
10
Regular ExpressionsSome examples...SupercomputerSuper
.*
[13579]^Once upon a time[A-Z][a-z]*[qQ][^u].*\([0-9][0-9]*\)\1[^aeiou][aeiuo]*So, how do we access this from withinour C programs ?First, we must set up your environmentso that the dynamic linker can find thesystem libraries for the manipulation ofregular expressions...LD_LIBRAR
Y_P
A
TH /usr/ucblib
CMPUT 201, Winter 1996 - (C) David Southwell
9
Regular ExpressionsUNIX grepgrep <regex> <files>...scans all the characters in the filesspecified, and prints out each line wherethere is at least 1 match with theregular expression given.eg.
cd /usr/includegrep 
'.printf
' *.h
Note that you will need touse [
'] if your regular expression
includes codes interpreted bythe shell before grep sees them.
UNIX is a line oriented OS - nearly allthe utilities work on text files, and on aline by line basis.
CMPUT 201, Winter 1996 - (C) David Southwell
12
Regular ExpressionsExample program :#include <stdio.h>char *re_comp(char *);int re_exec(char *);main(){char *return_value;char input_string[256];return_value = re_comp("[A-Z][a-z]*");if (return_value != (char *)NULL)
{puts(return_value);exit(1);}
CMPUT 201, Winter 1996 - (C) David Southwell
1
1
Regular ExpressionsAlso, modify the makefile so that theappropriate libraries are visible from gcc.Use the flags -L/usr/ucblib -lucbIt appears that the function headers arenot available from falun.  Y
ou should use
the following to keep the type checkerhappy :char *re_comp(char *);int re_exex(char *);The pattern matching must be veryfast, and the raw text form of theregular expressions require someparsing, to take them down to a moreef
ficient representation.  W
e use the
re_comp() call to do this.  The re_exec()call is used to check the string againstour regular expression.
CMPUT 201, Winter 1996 - (C) David Southwell
14
Where to get these slidesCopies of these notes can be foundfrom my web page :
http://web.cs.ualberta.ca/~daves
They are in compressed postscript form,so to print them out...
- download them from the web, by
clicking on them.
- uncompress, with gunzip...
gunzip 201_dave.ps.gz
- and print directly to a postscript
printer
...
lpr -P<your printer> 201_dave.ps
CMPUT 201, Winter 1996 - (C) David Southwell
13
Regular Expressionsdo
{scanf("%s", input_string);} while (re_exec(input_string));
printf("%s wasn
't a capitalised name!\n",
input_string);
exit(0);}The loop takes a string from the console,and checks it with the previouslysubmitted regular expression; this willreturn a 1 for a match, or a 0 for nomatch.  The loop continues until nomatch is found - the program terminates.re_exec continues to use the initialregular expression until re_comp() iscalled with a new one.