{\rtf1\mac\deff2 {\fonttbl{\f0\fswiss Chicago;}{\f2\froman New York;}{\f3\fswiss Geneva;}{\f4\fmodern Monaco;}{\f20\froman Times;}{\f21\fswiss Helvetica;}{\f22\fmodern Courier;}{\f23\ftech Symbol;}{\f200\fnil MitreLogo;}{\f201\fnil MitreSymbol;} {\f2222\fnil MITRE;}{\f14988\fnil BI Garamond BoldItalic;}{\f14989\fnil B Garamond Bold;}{\f14990\fnil I Garamond LightItalic;}{\f14991\fnil Garamond;}{\f15018\fnil BI Palatino BoldItalic;}{\f15019\fnil B Palatino Bold;}{\f15020\fnil I Palatino Italic;} {\f15021\fnil Palatino;}}{\colortbl\red0\green0\blue0;\red0\green0\blue255;\red0\green255\blue255;\red0\green255\blue0;\red255\green0\blue255;\red255\green0\blue0;\red255\green255\blue0;\red255\green255\blue255;}{\stylesheet{\s243\tqc\tx4320\tqr\tx8640 \f156 \sbasedon0\snext243 footer;}{\s244\tqc\tx4320\tqr\tx8640 \f156 \sbasedon0\snext244 header;}{\f156 \sbasedon222\snext0 Normal;}{\s1\fi720 \f156 \sbasedon0\snext1 Description Paragraph;}{\s2\li900\ri720 \f156 \sbasedon0\snext2 Item;}{\s3 \f156\fs36 \sbasedon0\snext3 Chapter Header;}}{\info{\author Rich Siegel}}\margl1440\margr2160\widowctrl\ftnbj\ftnrestart \sectd \linemod0\linex0\cols1\colsx0\endnhere {\header \pard\plain \li180\ri-20\tx7200\tx9619 \f156 {\f20\fs20 \par }}{\footer \pard\plain \li180\ri-20\tx7200\tx9619 \f156 {\f20\fs20 \par }}\pard\plain \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 \f156 {\f21 ____________________________________________________________\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\b\f21\fs28 INTRODUCTION TO GREP\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f21 ____________________________________________________________\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\b\f20\fs20 grep}{\f20\fs20 is a a method of pattern matching that derives from the Unix\'aa system. You are probably familiar with simpl e pattern matching from using word processors; when you ask a word processor to find all instances of the word "black", it is performing a simple pattern match, where each letter has to match literally. Matching strings in this manner is not very hard. \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 But the ability to match strings in a more general manner is both more powerful and more complicated. It allows for sophisticated pattern matching operations, such as matching all words that begin with the letter "P" and end with the letters "er", or dele ting the first word of every line. Grep provides a powerful means of doing this.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par \par \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f21 \page ____________________________________________________________\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\b\f21\fs28 HOW GREP WORKS\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f21 ____________________________________________________________\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 The "grep" mode of searching and replacing is a powerful tool. At the expense of being somewhat slower than normal text searching, using grep allows the user to search for one of a set of many strings instead of a particular string. As a simple example, you can search for any occurrence of an identifier beginning with the letter }{\f22\fs20 P}{\f20\fs20 , or all lines that begin with a left brace.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 A }{\b\f20\fs20 pattern}{\f20\fs20 is a string of characters that, in turn, describes a set of strings of characters. An example of a set of strings is the set of all strings that begin with the letter }{\f22\fs20 P}{\f20\fs20 and end with the letter }{\f22\fs20 r}{\f20\fs20 ; the strings "Ptr" and "ProcPtr" are members of this set. We say that a string is }{\b\f20\fs20 matched}{\f20\fs20 by a pattern if it is a member of the set described by the pattern. Patterns are composed of sub-patterns which are patte rns in themselves; this is how complicated patterns may be formed.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par Some examples of grep patterns:\par \par To replace a Pascal comment with a C comment, you would use \par }{\f22\fs20 \tab \{\\([^\}]*\\)\}\par }{\f20\fs20 to match the comment and\par }{\f22\fs20 \tab /*\\1*/\par }{\f20\fs20 to replace it.\par To change all words that begin with the letter}{\f22\fs20 P }{\f20\fs20 to begin with the letter}{\f22\fs20 Q}{\f20\fs20 , you would use\par }{\f22\fs20 \tab \\\par }{\f20\fs20 to match the word and\par }{\f22\fs20 \tab Q\\1\par }{\f20\fs20 to replace it.\par \par To change a list of names; ie:\par }{\f22\fs20 \tab FrameRect\par \tab PaintRect\par \tab EmptyRect\par \par }{\f20\fs20 to a list of names, followed by strings containing those names; i.e.}{\f22\fs20 \par \tab FrameRect,\tab "FrameRect",\par \tab PaintRect,\tab "PaintRect",\par \tab EmptyRect,\tab "EmptyRect",\par \par }{\f20\fs20 you would use\par }{\f22\fs20 \tab \\([A-Za-z][A-Za-z]*\\)\par }{\f20\fs20 to match the name and\par }{\f22\fs20 \tab \\1,\tab "\\1",\par }{\f20\fs20 to replace it.\par \par \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 You don't have to understand how these work now; in fact, it would be surprising if you do. The following section goes through the grep pattern matching and replacement rules step by step, so that by the end of it you should be able to understand how eac h of these grep patterns works and be able to make your own.\page \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f21 ____________________________________________________________\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\b\f21\fs28 PATTERN MATCHING AND REPLACEMENT RULES\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f21 ____________________________________________________________\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\b\f20\fs20 A note on notation:}{\f20\fs20 Writing about patterns and strings can be very confusing, since patterns and strings are made up of characters, as is this text. Therefore, we use certain typographical conventions to distinguish various usages.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 All literal characters will be in the courier font; therefore,}{\f22\fs20 a }{\f20\fs20 and}{\f22\fs20 xyz }{\f20\fs20 refer to those literal strings of characters.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 All patterns, when talked about in the abstract, will be italicized; therefore, }{\i\f20\fs20 p}{\f20\fs20 and }{\i\f20\fs20 q}{\f20\fs20 refer to abstract patterns.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 All strings, when talked about in the abstract, will be Greek letters; therefore, \'a7 and \'b5 refer to abstract strings.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 Sometimes we will be referring to parts of strings or patterns within longer ones. In these cases, the parts that are being referred to will be underlined. Therefore, in the string}{\f22\fs20 xx}{\f22\fs20\ul aab}{\f22\fs20 x}{\f20\fs20 , only the sub-string}{\f22\fs20 aab }{\f20\fs20 is actually being referred to; the other letters are used for context.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 In the examples, a string that can occur anywhere in a line will be preceded and followed by an ellipsis (}{\f22\fs20 \'c9)}{\f20\fs20 ; i.e. }{\f22\fs20 \'c9xyz\'c9}{\f20\fs20 . If it can occur only at the beginning of the line, it will only be followed by an ellipsis; i.e., }{\f22\fs20 xyz\'c9}{\f20\fs20 . Similarly, if it can occur only at the end of the line, it will be preceded but not followed by an ellipsis.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 In some cases, the state of case sensitivity affects the results of a pattern match. In the examples we have noted when this is this case.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\b\f21\fs28\ul Pattern matching}{\f20\fs28 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\b\f21 Simple matching}{\f21 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par 1. Any character, with certain exceptions described below, is a pattern that matches itself.\par \par }{\b\f20\fs20 Examples:\par }{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx9619 {\f20\fs20 \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab }{\f20\fs20\ul Text}{\f20\fs20 \tab }{\f20\fs20\ul With case sensitivity}{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f22\fs20 X}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9X\'c9\tab \par }{\f20\fs20 \tab doesn't match\tab }{\f22\fs20 \'c9x\'c9\tab }{\f20\fs20 on}{\f22\fs20 \par }{\f20\fs20 \tab but matches\tab }{\f22\fs20 \'c9x\'c9\tab }{\f20\fs20 off\par \par \par \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 2.}{\i\f20\fs20 \tab }{\f20\fs20 A pattern }{\i\f20\fs20 x}{\f20\fs20 followed by a pattern }{\i\f20\fs20 y}{\f20\fs20 forms a pattern }{\i\f20\fs20 xy}{ \f20\fs20 that matches any string \'a7\'b5 where \'a7 can be matched by }{\i\f20\fs20 x}{\f20\fs20 and \'b5 can be matched by }{\i\f20\fs20 y}{\f20\fs20 . We can, of course, take the compound pattern }{\i\f20\fs20 xy}{\f20\fs20 and concatenate yet another pattern }{\i\f20\fs20 z}{\f20\fs20 onto it, forming the pattern }{\i\f20\fs20 xyz}{\f20\fs20 .\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }{\b\f20\fs20 Examples:\par }{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab \tab }{\f20\fs20\ul Text}{\f20\fs20 \tab }{\f20\fs20\ul With case sensitivity}{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f22\fs20 XY}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9XY\'c9\par }{\f20\fs20 \par \tab }{\f22\fs20 Ptr}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9Ptr\'c9}{\f20\fs20 \par \tab \tab doesn't match\tab }{\f22\fs20 \'c9ptr\'c9\tab }{\f20\fs20 on}{\f22\fs20 \par }{\f20\fs20 \tab \tab but does match\tab }{\f22\fs20 \'c9ptr\'c9\tab }{\f20\fs20 off}{\f22\fs20 \par }{\f20\fs20 \par \par \par 3.\tab The character}{\f22\fs20 . }{\f20\fs20 is a pattern that will match any character.\par \par \tab }{\b\f20\fs20 Examples:\par }{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab \tab }{\f20\fs20\ul Text}{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f22\fs20 P.r}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9Ptr\'c9\tab \par \tab \tab }{\f20\fs20 and matches\tab }{\f22\fs20 \'c9P.r\'c9\tab }{\f20\fs20 \par }{\f22\fs20 \par }{\f20\fs20 \tab }{\f22\fs20 ..}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9ab\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \tab \tab }{\f20\fs20 and matches\tab }{\f22\fs20 \'c9a.\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \par }{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 4.\tab The character}{\f22\fs20 \\ }{\f20\fs20 followed by any character except}{\f22\fs20 (}{\f20\fs20 ,}{\f22\fs20 )}{\f20\fs20 ,}{\f22\fs20 <}{ \f20\fs20 ,}{\f22\fs20 >}{\f20\fs20 ,}{\f22\fs20 }{\f20\fs20 or one of the digits }{\f22\fs20 1}{\f20\fs20 -}{\f22\fs20 9}{\f20\fs20 is a pattern that matches that character.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par \tab }{\b\f20\fs20 Examples:\par }{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab \tab }{\f20\fs20\ul Text}{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f22\fs20 P\\.r}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9P.r\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \tab \tab }{\f20\fs20 but doesn't match}{\f22\fs20 \'c9Ptr\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \par }{\f20\fs20 \tab }{\f22\fs20 P\\\\r}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9P\\r\'c9}{\f20\fs20 \tab }{\f22\fs20 \par }{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 5.\tab A string of characters }{\i\f20\fs20 s}{\f20\fs20 surrounded by square brackets (}{\f22\fs20 [ }{\f20\fs20 and a}{\f22\fs20 ]) }{\f20\fs20 forms a pattern}{\f22\fs20 [}{\i\f22\fs20 s}{\f22\fs20 ] }{\f20\fs20 that matches a single instance of one of the characters in the string }{\i\f20\fs20 s}{\f20\fs20 . Note that the case sensitivity flag does }{\f20\fs20\ul not}{\f20\fs20 apply to characters between square brackets: letters must match exactly.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par \tab }{\b\f20\fs20 Examples:\par }{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab \tab \tab }{\f20\fs20\ul Text}{\f20\fs20 \par \tab }{\f22\fs20 [abc]}{\f20\fs20 \tab matches\tab \tab }{\f22\fs20 \'c9}{\f22\fs20\ul a}{\f22\fs20 b\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \tab \tab }{\f20\fs20 and matches\tab \tab }{\f22\fs20 \'c9x}{\f22\fs20\ul b}{\f22\fs20 \'c9}{\f20\fs20 \tab }{\f22\fs20 \par \tab \tab }{\f20\fs20 but doesn't match\tab }{\f22\fs20 \'c9}{\f22\fs20\ul ab}{\f22\fs20 \'c9}{\f20\fs20 \tab }{\f22\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f22\fs20 \par }{\f20\fs20 \tab }{\f22\fs20 [abc][xyz]}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9ax\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \tab \tab }{\f20\fs20 but doesn't match\tab }{\f22\fs20 \'c9ab\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \par }{\f20\fs20 \tab }{\f22\fs20 [abc]x}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9bx\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \tab \tab }{\f20\fs20 but doesn't match}{\f22\fs20 \tab \'c9Bx}{\b\f22\fs20 \'c9}{\f20\fs20 \par \par \par \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 5a.\tab The pattern}{\f22\fs20 [^}{\f20\fs20 \'a7}{\f22\fs20 ] }{\f20\fs20 matches any character that is not in the string \'a7 . Special characters will be taken literally in this context. Again, case sensitivity doesn't apply to characters between square brackets.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par \tab }{\b\f20\fs20 Examples:\par }{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab \tab \tab }{\f20\fs20\ul Text\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f22\fs20 [^abc]}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9x\'c9\par \tab \tab \tab }{\f20\fs20 and matches\tab }{\f22\fs20 \'c9A\'c9\par \tab \tab }{\f20\fs20 but doesn't match\tab }{\f22\fs20 \'c9a\'c9\par \tab \par }{\f20\fs20 \tab }{\f22\fs20 [^abc]a}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9xa\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \tab \tab }{\f20\fs20 but doesn't match\tab }{\f22\fs20 \'c9aa\'c9}{\f20\fs20 \tab }{\f22\fs20 \par }{\f20\fs20 \par \tab }{\f22\fs20 [^.]a}{\f20\fs20 \tab \tab matches\tab }{\f22\fs20 \'c9xa\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \tab \tab }{\f20\fs20 but doesn't match\tab }{\f22\fs20 \'c9.a\'c9}{\f20\fs20 \tab }{\f22\fs20 \par }{\f20\fs20 \par \par \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 5b.\tab If a string of three characters in the form}{\f22\fs20 [}{\i\f22\fs20 a}{\f22\fs20 -}{\i\f22\fs20 b}{\f22\fs20 ] }{\f20\fs20 occurs in in the pattern }{\i\f20\fs20 p}{\f20\fs20 , this represents all of the characters from }{\i\f22\fs20 a}{\f20\fs20 to }{\i\f22\fs20 b}{\f20\fs20 inclusive. All special characters are taken literally; i.e.,}{\f22\fs20 [!-.] }{\f20\fs20 denotes the characters from }{\f22\fs20 ! to .}{\f20\fs20 . Notice that the only way to include the character}{\f22\fs20 ] }{\f20\fs20 in }{\i\f20\fs20 p}{\f20\fs20 is to make it the very first character. Likewise, the only way to include the character}{\f22\fs20 - }{\f20\fs20 in }{\i\f20\fs20 p}{\f20\fs20 is to have it either at the very beginning or the very end of }{\i\f20\fs20 p}{\f20\fs20 . Single characters and ranges may both be used between brackets.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par \tab }{\b\f20\fs20 Examples:\par }{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab \tab \tab }{\f20\fs20\ul Text}{\f20\fs20 \par \tab }{\f22\fs20 [a-c]}{\f20\fs20 \tab \tab matches\tab }{\f22\fs20 \'c9}{\f22\fs20\ul a}{\f22\fs20 c\'c9}{\f20\fs20 \tab }{\f22\fs20 \par }{\f20\fs20 \tab \tab \tab and matches\tab }{\f22\fs20 \'c9x}{\f22\fs20\ul c}{\f22\fs20 \'c9}{\f20\fs20 \tab }{\f22\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f22\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f22\fs20 [1x-z]a}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9}{\f22\fs20\ul 1a}{\f22\fs20 \'c9}{\f20\fs20 \tab }{\f22\fs20 \par \tab \tab \tab }{\f20\fs20 and matches\tab }{\f22\fs20 \'c9}{\f22\fs20\ul x}{\f22\fs20 a\'c9}{\f20\fs20 \tab }{\f22\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f22\fs20 [-x-z]a}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9}{\f22\fs20\ul -a}{\f22\fs20 \'c9\par \tab \tab \tab }{\f20\fs20 and matches\tab \'c9}{\f20\fs20\ul xa}{\f20\fs20 \'c9}{\f22\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 6.\tab Any pattern }{\i\f20\fs20 p}{\f20\fs20 formed by any combination of rules 1 or 3-5b followed by a}{\f22\fs20 * }{\f20\fs20 forms the pattern}{ \f22\fs20 }{\i\f20\fs20 p}{\f22\fs20 * }{\f20\fs20 that matches zero or more consecutive occurrences of characters matched by}{\f22\fs20 }{\i\f20\fs20 p}{\f20\fs20 .\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }{\b\f20\fs20 \tab Examples:\par }{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab \tab \tab }{\f20\fs20\ul Text}{\f20\fs20 \tab }{\f20\fs20\ul With case sensitivity}{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f22\fs20 [a-c]*}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9a\par \tab \tab \tab }{\f20\fs20 and matches\tab }{\f22\fs20 \'c9acbca\par \tab \tab \tab }{\f20\fs20 and matches\tab nothing\par }{\f22\fs20 \par }{\f20\fs20 \tab }{\f22\fs20 A[a-z]*}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9A\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \tab \tab \tab }{\f20\fs20 and matches\tab }{\f22\fs20 \'c9Abcb\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \tab \tab \tab }{\f20\fs20 and matches\tab }{\f22\fs20 \'c9abc\'c9}{\f20\fs20 \tab on}{\f22\fs20 \par \tab \tab }{\f20\fs20 but doesn't match\tab }{\f22\fs20 \'c9abc\'c9}{\f20\fs20 \tab off}{\f22\fs20 \par \par }{\f20\fs20 \tab }{\f22\fs20 .*}{\f20\fs20 \tab \tab matches\tab anything from\par \tab \tab \tab \tab \tab beginning of a \par \tab \tab \tab \tab \tab line to the end of\par \tab \tab \tab \tab \tab the line\par }{\f22\fs20 \par }{\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \tab }{\f22\fs20 [abc]*}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9}{\f22\fs20\ul b}{\f22\fs20 \par \tab \tab \tab }{\f20\fs20 and matches\tab }{\f22\fs20 \'c9}{\f22\fs20\ul ab}{\f22\fs20 \par \tab \tab }{\f20\fs20 but doesn't match just\tab }{\f22\fs20 \'c9}{\f22\fs20\ul a}{\f22\fs20 b\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f22\fs20 \tab \tab }{\f20\fs20 (because it matches\par \tab \tab the longest string\par \tab \tab possible)}{\f22\fs20 \par \par }{\f20\fs20 \tab (}{\f22\fs20 .*)}{\f20\fs20 \tab \tab matches\tab }{\f22\fs20 \'c9(aaa)\'c9}{\f20\fs20 \tab \par }{\f22\fs20 \tab \tab \tab }{\f20\fs20 and matches\tab }{\f22\fs20 \'c9()\'c9}{\f20\fs20 \tab \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 \page {\b\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\b\f20\fs20 A closer example:\par \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 Let us examine more closely how the pattern}{\f22\fs20 (.*) }{\f20\fs20 matches text. This pattern will match any string that is enclosed in parentheses. This includes the string}{\f22\fs20 ()}{\f20\fs20 , since the sub-pattern}{\f22\fs20 .* }{\f20\fs20 will match the empty string between the}{\f22\fs20 ( }{\f20\fs20 and the}{\f22\fs20 )}{\f20\fs20 . But what about the string}{\f22\fs20 (())}{\f20\fs20 ? Since the pattern}{\f22\fs20 .* }{\f20\fs20 will match any number of occurrences of all characters, won't it match the}{\f22\fs20 (() }{\f20\fs20 and cause the last}{\f22\fs20 ) }{\f20\fs20 in the string to fail to match? Or conversely, won't the sub-pattern}{\f22\fs20 (.* }{\f20\fs20 match the whole string, leaving the}{\f22\fs20 ) }{\f20\fs20 at the end of the pattern unmatched?\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 The answer to this is that any pattern of the form}{\f22\fs20 }{\i\f20\fs20 p}{\f22\fs20 * }{\f20\fs20 in a pattern}{\f22\fs20 }{\i\f20\fs20 p}{ \f22\fs20 *}{\i\f20\fs20 y}{\f22\fs20 }{\f20\fs20 will match the }{\f20\fs20\ul largest}{\f20\fs20 number of occurrences of whatever }{\i\f20\fs20 p}{\f20\fs20 matches }{\f20\fs20\ul that still allows a match to }{\i\f20\fs20\ul y}{\f20\fs20 . Therefore, in matching}{\f22\fs20 (()) }{\f20\fs20 against the pattern}{\f22\fs20 (.*), }{\f20\fs20 only the inner parentheses in the string}{\f22\fs20 (()) }{\f20\fs20 will be matched by the sub-pattern}{\f22\fs20 .*}{\f20\fs20 .\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par \par \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\b\f21 Remembering sub-strings\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 We now have the ability to form patterns that are composed of sub-patterns, and will find it useful to "remember" sub-strings matched by sub-patterns and to be able to match against those substrings.\par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 \par }\pard \li180\ri-20\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx7200\tx9619 {\f20\fs20 7.\tab A pattern surrounded by}{\f22\fs20 \\( }{\f20\fs20 and}{\f22\fs20 \\) }{\f20\fs20 is a pattern that matches whatever the sub-pattern matches. This is useful for matching two or more instances of the same string and when doing replacements.\par \par \tab }{\b\f20\fs20 Example:\par }{\f20\fs20 \par \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab \tab \tab }{\f20\fs20\ul Text}{\f20\fs20 \par \tab }{\f22\fs20 \\(abc\\)}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9abc}{\f20\fs20 \par \tab }{\f22\fs20 \\(ab(\\)}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9ab(\par \par \par }{\f20\fs20 \par 8.\tab A}{\f22\fs20 \\ }{\f20\fs20 followed by }{\i\f20\fs20 n}{\f20\fs20 , where }{\i\f20\fs20 n}{\f20\fs20 is one of the digits }{\f22\fs20 1-9}{\f20\fs20 , is a pattern that matches whatever was matched by the sub-pattern beginning with the "nth" occurrence of }{\f22\fs20 \\(}{\f20\fs20 . A pattern}{\f22\fs20 \\}{\i\f20\fs20 n}{\f20\fs20 may be followed by an}{\f22\fs20 *}{\f20\fs20 , and forms a pattern}{\f22\fs20 }{\f20\fs20 \\}{\i\f20\fs20 n}{\f22\fs20 * }{\f20\fs20 that matches zero or more occurrences of whatever }{\f22\fs20 \\}{\i\f20\fs20 n}{\f20\fs20 matches.\par \par \tab }{\b\f20\fs20 Examples:\par }{\f20\fs20 \par \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab \tab \tab }{\f20\fs20\ul Text}{\f20\fs20 \par \tab }{\f22\fs20 \\(abc\\)\\1}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9abcabc\'c9}{\f20\fs20 \tab }{\f22\fs20 \par \par }{\f20\fs20 \tab }{\f22\fs20 \\(a.c\\)\\1}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9axcaxc\'c9}{\f20\fs20 \tab }{\f22\fs20 \par }{\f20\fs20 \tab \tab \tab but not\tab \tab }{\f22\fs20 \'c9axcazc\'c9}{\f20\fs20 \tab }{\f22\fs20 \par }{\f20\fs20 \tab \tab \tab nor\tab \tab }{\f22\fs20 \'c9axcaXc\'c9}{\f20\fs20 \tab }{\f22\fs20 \par }{\b\f20\fs20 \par }{\f20\fs20 Note that in this last pattern, the sub-pattern}{\f22\fs20 \\1 }{\f20\fs20 does not imply a re-application of the sub-pattern}{\f22\fs20 a.c}{\f20\fs20 , but what}{\f22\fs20 a.c }{\f20\fs20 matches. If}{\f22\fs20 \\(a.c\\) }{\f20\fs20 was matched with the string}{\f22\fs20 axc}{\f20\fs20 , then the sub-pattern}{\f22\fs20 \\1 }{\f20\fs20 would try to match the literal string}{\f22\fs20 axc }{\f20\fs20 against the remainder of the search string. Therefore, the pattern}{\f22\fs20 \\ (a.c\\)\\1 }{\f20\fs20 will match}{\f22\fs20 axcaxc}{\f20\fs20 , but will not match}{\f22\fs20 axcazc}{\f20\fs20 .\par \par \par \page \par }{\b\f21 Constraining matches}{\b\f20 \par }{\b\f20\fs20 \par }{\f20\fs20 Sometimes it is useful to be able to "constrain" patterns to match only if certain conditions in the context outside the string matched are met.\par \par 9.\tab A pattern surrounded by}{\f22\fs20 \\< }{\f20\fs20 and}{\f22\fs20 \\> }{\f20\fs20 is a pattern that matches whatever is matched by the sub-pattern, provided that the first and last characters of the matched string can be matched by}{\f22\fs20 [A-Za-z0-9_] }{\f20\fs20 and that the characters immediately surrounding the matched string cannot be matched by}{\f22\fs20 [A-Za-z0-9_] }{\f20\fs20 (i.e., can be matched by}{\f22\fs20 [^A-Za-z0-9_]}{\f20\fs20 ).\par \par \tab This is used to match any string that matches the sub-pattern only if the matched string begins and ends on a "word" boundary (a "word" being a C identifier).\par \par \tab }{\b\f20\fs20 Examples:\par }{\f20\fs20 \par \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab \tab \tab }{\f20\fs20\ul Text}{\f20\fs20 \par \tab }{\f22\fs20 \\}{\f20\fs20 \tab matches\tab }{\f22\fs20 \'c9+}{\f22\fs20\ul ab}{\f22\fs20 +\'c9}{\f20\fs20 \tab }{\f22\fs20 \par }{\f20\fs20 \tab \tab but doesn't match\tab }{\f22\fs20 \'c9}{\f22\fs20\ul +ab+}{\f22\fs20 \'c9\par \tab }{\f20\fs20 \tab and doesn't match}{\f22\fs20 \tab \'c9+}{\f22\fs20\ul ab}{\f22\fs20 c+}{\f20\fs20 \par }{\f22\fs20 \par }{\f20\fs20 \par \par 10.\tab A pattern }{\i\f20\fs20 p}{\f20\fs20 that is preceded by a}{\f22\fs20 ^ }{\f20\fs20 forms a pattern}{\f22\fs20 ^}{\i\f22\fs20 p}{\f20\fs20 . If the pattern}{\f22\fs20 ^}{\i\f22\fs20 p}{\f20\fs20 is not preceded by any other pattern, it matches whatever }{\i\f20\fs20 p}{\f20\fs20 matches as long as the first character matched by }{\i\f20\fs20 p}{\f20\fs20 occurs at the beginning of a line. If the pattern}{\f22\fs20 ^}{\i\f22\fs20 p}{ \f20\fs20 is preceded by another pattern, then the}{\f22\fs20 ^ }{\f20\fs20 is taken literally.\par \par \tab }{\b\f20\fs20 Examples:\par }{\f20\fs20 \par \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab \tab \tab }{\f20\fs20\ul Text}{\f20\fs20 \par \tab }{\f22\fs20 ^ab*}{\f20\fs20 \tab matches\tab \tab }{\f22\fs20\ul ab}{\f22\fs20 \'c9}{\f20\fs20 \tab \par \tab \tab but doesn't match\tab }{\f22\fs20 x}{\f22\fs20\ul ab}{\f22\fs20 \'c9}{\f20\fs20 \tab \par \par }{\f22\fs20 \tab ab^ab*}{\f20\fs20 \tab matches\tab }{\f22\fs20 ab^ab\'c9}{\f20\fs20 \tab \par \par \par \par 11.\tab A pattern }{\i\f20\fs20 p}{\f20\fs20 that is followed by a}{\f22\fs20 $ }{\f20\fs20 forms a pattern }{\i\f20\fs20 p}{\f22\fs20 $}{\f20\fs20 . If the pattern }{\i\f20\fs20 p}{\f22\fs20 $}{\f20\fs20 is not followed by any other pattern, it matches whatever }{\i\f20\fs20 p}{\f20\fs20 matches as long as the last character matched by }{\i\f20\fs20 p}{\f20\fs20 occurs at the end of a line. If the pattern }{\i\f20\fs20 p}{\f22\fs20 $ }{\f20\fs20 is followed by another pattern, then the }{\f22\fs20 $}{\f20\fs20 is taken literally.\par \par \tab }{\b\f20\fs20 Examples:\par }{\f20\fs20 \par \tab }{\f20\fs20\ul Pattern}{\f20\fs20 \tab \tab \tab \tab }{\f20\fs20\ul Text}{\f20\fs20 \par \tab }{\f22\fs20 ab$}{\f20\fs20 \tab matches\tab \tab }{\f22\fs20 \'c9ab}{\f20\fs20 \tab }{\f22\fs20 \par }{\f20\fs20 \tab \tab but doesn't match\tab }{\f22\fs20 \'c9}{\f22\fs20\ul ab}{\f22\fs20 x}{\f20\fs20 \tab }{\f22\fs20 \par \par }{\f20\fs20 \tab }{\f22\fs20 ab$ab}{\f20\fs20 \tab matches\tab \tab }{\f22\fs20 \'c9ab$ab\'c9\par }{\f20\fs20 \par \tab }{\f22\fs20 ^ab$}{\f20\fs20 \tab matches\tab \tab }{\f22\fs20 ab}{\f20\fs20 \par \tab \tab but doesn't match\tab }{\f22\fs20 ab\'c9\par \par \par }{\f20\fs20 \tab Note that the characters }{\f22\fs20 ^}{\f20\fs20 and }{\f22\fs20 $}{\f20\fs20 constrain pattern matches to begin or end at line boundaries, and so can be combined to constr ain a pattern to match an entire line only (as in the above example).\par \par We mentioned at the beginning the ability to search for any identifier beginning with the letter }{\f22\fs20 P}{\f20\fs20 . This would be accomplished with the pattern }{\f22\fs20 \\<[Pp][A-Za-z0-9_]*\\>}{\f20\fs20 . Note that, if you have case sensitivity is off, then the patterns}{\f22\fs20 \\ }{\f20\fs20 and }{\f22\fs20 \\}{\f20\fs20 would match the same strings. Also, if word-match is on, then any of these patterns with the}{\f22\fs20 \\<}{\f20\fs20 and}{\f22\fs20 \\>}{\f20\fs20 removed will match the same strings.\par \par \par \par }{\b\f21 Replacement\par }{\f20\fs20 \par Grep provides not only a more sophisticated method of searching, but a sophisticated method of replacing as well. In a replacement string, the following substitutions are made before any text replacement occurs:\par \par 1.\tab Each occurrence of the character }{\f22\fs20 &}{\f20\fs20 is replaced with whatever was last matched by the pattern.\par \par }{\b\f20\fs20 \tab Examples:\par }{\f20\fs20 \par \tab }{\f20\fs20\ul "Find" string}{\f20\fs20 \tab }{\f20\fs20\ul "Replace" string}{\f20\fs20 \tab }{\f20\fs20\ul Original text}{\f20\fs20 \tab }{\f20\fs20\ul Result}{\f20\fs20 \par }{\f22\fs20 \tab abc}{\f20\fs20 \tab \tab }{\f22\fs20 +&}{\f20\fs20 \tab \tab }{\f22\fs20 \'c9abc\'c9}{\f20\fs20 \tab \tab }{\f22\fs20 \'c9+abc\'c9}{\f20\fs20 \par }{\f22\fs20 \tab abc}{\f20\fs20 \tab \tab }{\f22\fs20 &&}{\f20\fs20 \tab \tab }{\f22\fs20 \'c9abc\'c9}{\f20\fs20 \tab \tab }{\f22\fs20 \'c9abcabc\'c9}{\f20\fs20 \par \tab \par \par 2.\tab Each occurrence of a string of the form }{\f22\fs20 \\}{\i\f20\fs20 n}{\f20\fs20 , where }{\i\f20\fs20 n}{\f20\fs20 is one of the digits }{\f22\fs20 1}{\f20\fs20 -}{\f22\fs20 9}{\f20\fs20 , is repla ced by whatever was last matched by the sub-pattern beginning with the }{\i\f22\fs20 n}{\f20\fs20 th occurrence of }{\f22\fs20 \\(}{\f20\fs20 .\par \par }{\b\f20\fs20 \tab Examples:\par }{\f20\fs20 \par \tab }{\f20\fs20\ul "Find" string}{\f20\fs20 \tab \tab }{\f20\fs20\ul "Replace" string}{\f20\fs20 \tab }{\f20\fs20\ul Original text}{\f20\fs20 \tab }{\f20\fs20\ul Result}{\f20\fs20 \par }{\f22\fs20 \tab \\(a*\\)\\(b*\\)}{\f20\fs20 \tab }{\f22\fs20 \\1\\2}{\f20\fs20 \tab \tab }{\f22\fs20 aabb\'c9}{\f20\fs20 \tab \tab }{\f22\fs20 aabb\'c9\par \tab \par \tab \\(a*\\)\\(b*\\)}{\f20\fs20 \tab }{\f22\fs20 \\2\\1}{\f20\fs20 \tab \tab }{\f22\fs20 aabb\'c9}{\f20\fs20 \tab \tab }{\f22\fs20 bbaa\'c9}{\f20\fs20 \par \tab \par \par \par \par 3.\tab Each occurrence of a string of the form }{\f22\fs20 \\}{\i\f22\fs20 p}{\f20\fs20 , where }{\i\f22\fs20 p}{\f20\fs20 is other than one of the digits 1-9, is replaced by }{\i\f22\fs20 p}{\f20\fs20 .\par \par }{\b\f20\fs20 \tab Examples:\par }{\f20\fs20 \par \tab }{\f20\fs20\ul "Find" string}{\f20\fs20 \tab \tab }{\f20\fs20\ul "Replace" string}{\f20\fs20 \tab }{\f20\fs20\ul Original text}{\f20\fs20 \tab }{\f20\fs20\ul Result}{\f20\fs20 \par }{\f22\fs20 \tab \\(a*\\)\\(b*\\)}{\f20\fs20 \tab }{\f22\fs20 \\1&\\2\\}{\f20\fs20 \tab }{\f22\fs20 aabb\'c9}{\f20\fs20 \tab \tab }{\f22\fs20 aa&bb\'c9\par \tab \par \tab \\(a*\\)\\(b*\\)}{\f20\fs20 \tab }{\f22\fs20 \\\\\\2\\1\\\\}{\f20\fs20 \tab }{\f22\fs20 aabb\'c9}{\f20\fs20 \tab \tab }{\f22\fs20 \\bbaa\\\'c9}{\f20\fs20 \par \par \par \tab This allows you to not only be able to search for a string satisfying a complex set of conditions, but also to be able to do a subsequent replacement that varies depending on the string that is matched.\par \par \par \page \par }{\b\f21\fs28 Some Examples\par }{\b\f20\fs20 \par }{\f20\fs20 \bullet \tab Suppose that you have written a program that is to become a Macintosh application (i.e., it uses the Macintosh ToolBox instead of stdio for the user interface). Suppose also that you have discovered that you have forgotten to put a }{\f22\fs20 \\p}{ \f20\fs20 at the beginning of your string constants, so that your program is trying to pass C strings instead of Pascal strings to the ToolBox (which only knows how to deal with Pascal strings). You can easily change all your C strings to Pascal strings by specifying }{\f22\fs20 "\\(.*\\)" }{\f20\fs20 as the search pattern and }{ \f22\fs20 "\\\\p\\1" }{\f20\fs20 as the replacement string.\par \par \bullet \tab Suppose you decided to reverse the two arguments of the function "foo". You might try the pattern }{\f22\fs20 foo(\\([^,]*\\),\\([^)]*\\))}{\f20\fs20 as the search pattern and }{\f22\fs20 foo(\\2, \\1)}{\f20\fs20 as the replacement pattern. How does the search pattern work?\par \par \tab Let's assume we're trying to match some text that looks like }{\f22\fs20 foo(1,*bar)\par }{\f20\fs20 \par \bullet \tab }{\f22\fs20\ul foo(}{\f22\fs20 \\([^,]*\\),\\([^)]*\\))\tab }{\f20\fs20 matches}{\f22\fs20 \tab }{\f22\fs20\ul foo(}{\f22\fs20 1,*bar)\par }{\f20\fs20 \bullet \tab }{\f22\fs20 foo(}{\f22\fs20\ul \\([^,]*\\)}{\f22\fs20 ,\\([^)]*\\))\tab }{\f20\fs20 matches}{\f22\fs20 \tab foo(}{\f22\fs20\ul 1}{\f22\fs20 ,*bar)\par }{\f20\fs20 \bullet \tab }{\f22\fs20 foo(\\([^,]*\\)}{\f22\fs20\ul ,}{\f22\fs20 \\([^)]*\\))\tab }{\f20\fs20 matches}{\f22\fs20 \tab foo(1}{\f22\fs20\ul ,}{\f22\fs20 *bar)\par }{\f20\fs20 \bullet \tab }{\f22\fs20 foo(\\([^,]*\\),}{\f22\fs20\ul \\([^)]*\\)}{\f22\fs20 )\tab }{\f20\fs20 matches}{\f22\fs20 \tab foo(1,}{\f22\fs20\ul *bar}{\f22\fs20 )\par }{\f20\fs20 \bullet \tab }{\f22\fs20 foo(\\([^,]*\\),\\([^)]*\\)}{\f22\fs20\ul )}{\f22\fs20 \tab }{\f20\fs20 matches}{\f22\fs20 \tab foo(1,*bar}{\f22\fs20\ul )}{\f22\fs20 \par }{\f20\fs20 \par \tab Since }{\f22\fs20 \\([^,]*\\) }{\f20\fs20 matched }{\f22\fs20 1}{\f20\fs20 and }{\f22\fs20 \\([^)]*\\) }{\f20\fs20 matched}{\f22\fs20 *bar}{\f20\fs20 , the two arguments to foo, the replacement pattern }{\f22\fs20 foo(\\2, \\1)}{\f20\fs20 will result in }{\f22\fs20 foo(*bar,\~1)}{\f20\fs20 \par \par \tab This, unfortunately, won't work in the case of }{\f22\fs20 foo(1,(*bar)+2)}{\f20\fs20 , since }{\f22\fs20 \\([^)]*\\)}{\f20\fs20 will match only up to the first right parenthesis, leaving }{\f22\fs20 +2)}{\f20\fs20 unmatched. If we're sure that all calls to foo end with a semi-colon, however, we can change our pattern to }{\f22\fs20 foo(\\([^,]*\\),\\([^;]*\\)); . }{\f20\fs20 In this pattern, instead of trying to match the second argument by matching everything up to the first right parenthesis, we match everything up to the}{\f22\fs20 );}{\f20\fs20 which terminates the invocation of foo.\par \par In this example we showed how to analyze a grep pattern by examining sub-patterns. This is a good way of figuring out how to build a pattern as well. grep can be thought of as a small and rather cryptic programming language, with each pattern a program and sub-pattern a statement in this language. If you try to create a grep pattern by testing a small sub-pattern, then adding and testing additional sub-patterns until the complete pattern is built, you may find building complex grep patterns not nearly as daunting as you first thought.\par \par \par }}