Readme file for: findgrep
Programme: findgrep
grep's in current as well as in subdirectories.
Perl reg.exp.s can be used. Supports approximate matches.
usage:
findgrep [-lnisefcpdtBhaFSx] <reg.exp.> [<file name reg.exp.>]
Author: Lars Gregersen
E-mail: lg@kt.dtu.dk
Homepage: http://www.gbar.dtu.dk/~matlg/
findgrep webpage: http://www.gbar.dtu.dk/~matlg/findgrep/
1997.10.01, version 1.00: Creation of a great new programme
1997.10.07, version 1.10: added -s option
1997.10.08, version 1.20: added -e, -f and -c options
1997.10.22, version 1.30: added -p option
1997.11.05, version 1.31: changed option -c to -b
1997.11.07, version 1.40: added -P, -d, -t and -B option
1997.12.17, version 1.41: runs nicely with Perl -w option
added -h option
1998.03.12, version 1.50: added -a, --adiff, --ains, --adel
and --asub options
changed the handling of the -t option
now uses Getopt::Long to handle options
env.var. FINDGREP contains default arguments
removed option -P
added option -F
1998.04.09, version 1.60: fixed bug with -i option
fized bug: added 'local $_' to build_matchany
addapted to String::Approx version 2.5
1998.07.28, version 1.61: allowed "cpp,h" when using -e option
will now give an error message and
continue when an unreadable file is met
1999.01.31, version 1.63: added option -S (show paragraph)
-f now handles ..
adapted to String::Approx version 3.05
(works with version 2.5+)
1999.02.07, version 1.64: bug fixes for options -S and -s
-S option may now take an integer as argument
1999.04.19, version 1.70: added option -x
findgrep [-lnisefcpdtBhaFSx] <reg.exp.> [<file name reg.exp.>]
Options:
-l : list file names
-n : line numbers
-i : case insensitive
-s : shorten long lines
-e : treat file name reg.exp. as an extension
-f : use directory names in file name reg.exp.
and convert . to \. and * to .*
-b : only search base directory
-p : write path name with the file names
-d : match file name reg.exp. on path name and
file name (concatenated)
-t : transform file names to lower case
-B : don't search binary files
-h : help screen
-a : approximate search
-F : use pattern file
-S : show paragraph
-x : exclude file/directory pattern
Options can be specified in the environmental variable FINDGREP just
like on the command line. Options given in the FINDGREP variable will
be overridden by the options given on the commandline.
An option can be turned off on the command line bu specifying '--no'
before the option letter. E.g. if you have specified 'FINDGREP=-n' the
numbering can be turned of on the command line using '--non'.
The options have the same effect whether specified on the command line
except for the approximate match modifiers. See approximate match for
a specification.
On DOS systems the following definition in autoexec.bat is often
useful
set FINDGREP=-t
Description of options:
-l : list file names
List only the filenames of the files that have lines that match the
criteria. If used in combination with the -n option the lines numbers
of the first matching line is shown.
-n : line numbers
Show line numbers of matching lines.
-i : case insensitive
Do a case insensitive search. This makes the regular expresions case
insensitive for both the search pattern and the path and file name
pattern.
-s : shorten long lines
Show only a few lines of text around the place where some matching
text is found. This feature is useful when searching binary files
that do not have "standard" end of line chracters.
-e : treat file name reg.exp. as an extension
Use the file name specification as an extension (suffix) instead of a
regular expression. That means that "cpp" will be tranferred into
"\.cpp$". You may specify more than one extension by using "(h|cpp)",
which will search files that have either "h" or "cpp" as extension.
When used with the -f option the last part of the file name
specification will be used as the extension.
-f : use directory names in file name reg.exp.
and convert . to \. and * to .*
Make it possible to specify the base directory for the recursive
search. The remeaning part of the file name specification will have
the characters "." and "*" translated into "\." and ".*" respectively
and in that order. This makes it easier to specify file names in a
'glob' kind of way. But remember that regular expressions are still
used.
Examples:
Search for the pattern "foo" using bar as base directory in all files that matches the pattern "filename"
findgrep -f foo bar/filename
Search for the pattern "foo" in files that start with "bar" and
has the extension "h", but note that a file called "foobar.h.bak"
will also be found by this search criterion. The search criterion
used is "bar.*\.h".
findgrep -f foo "bar*.h"
Search for the pattern "foo" in a file that has "bar" in it followed by a number somewhere using the directory "tmpdir" as the base directory.
findgrep -f foo "tmpdir/bar*\d"
Search for the pattern "foo" in files that matches the pattern bar using the previous directory as base directory.
findgrep -f foo ../bar
-b : only search base directory
Only search the base directory. This makes findgrep work almost as
grep.
-p : write path name with the file names instead of
in seperate lines
The normal output of findgrep is to first specify the directory where
the search is made and then to write the file names without directory
names. The -p option changes this so the directory names are written
with the files.
-d : match file name reg.exp. on path name and
file name (concatenated)
Instead of just matching the file name the file name reg.exp. must
match the path name and the file name when they are concatenated
Example:
search for "foo" when the path name and file name will match
"bar.*foo". Will e.g. match "bar/somedir/foo".
findgrep -d foo "bar.*foo"
searh for "foo" using the "dir1" as base directory and looking for
files which have a path name (relative to the current file
directory) and file name that matches "bar*txt". It will e.g. match
"dir1/dir2/bar/readme.txt"
findgrep -fd foo "dir1/bar*txt"
-t : transform file names to lower case
File names and path names are transformed to lower case before using
in tests and when printing. The case is preserved when opening files.
Useful on file systems that do not distinguish between (but allow)
upper and lower case (like Windows 96/NT).
-B : don't search binary files
Don't search files that appear to be binary.
-h : help screen
Print a help screen with the most commonly used options.
-a : approximate search
Search for approximate match. This is good for searching for
misspelled
information and when you don't know the precise spelling of words.
Warning: the approximate search takes much longer time than the
normal
search. Note that String::Approx must be installed for this to work.
--adiff, --ains, --adel --asub
Approximate is defined here as k-differences. One difference
is an insertion, a deletion, or a substitution of one character.
The k in the k-differences is the maximum number of differences. The
default is to allow for one difference. This can be changed using the
modifiers above.
Approximate matching will automatically be used when any of the
modifiers are specified on the command line. This means that the -a
option can be left out. If the modifiers are specified in the FINDGREP
environmental variable the approximate match is not used unless the -a
option is specified. This feature allow one to specify defaults of the
modifers.
Examples
Search for 'Tschaikowskij' with one difference
findgrep -ea Tschaikowskij
Search for 'Tschaikowskij' with one substitution and no deletions or
insertions
findgrep -e --adiff=0 --asub=1 'Tschaikowskij'
FINDGREP contains '--adiff=2' and we want to search for
'Tschaikowskij' with 1 insertion, 2 deletions and 2 substitutions
findgrep --ains=1 Tschaikowskij
-F : use pattern file
Use the pattern in the specified pattern file for searches. A regular
expression cannot be specified on the command line if this option is
used. The pattern in the pattern file are separated by newlines.
Example
The file 'pattern.ptn' contains:
red
blue
green
Use findgrep to find these words in text files:
findgrep -e -Fpattern.ptn txt
-S : show paragraph
The standard output of findgrep is a single line where the matching
text if found. This is out of context with the rest of the text/file.
By using the option -S the entire paragraph where the matching text is
found will be displayed. findgrep uses an empty line to determine where
paragraphs start and end. Note that this option can create a lot of
output if paragraphs are long. This option can be combined with the -s option.
If used in combination with the -n parameter the line number of the first matching line is shown with the file name (note that there can be more than more match in a paragraph); the line number of the lines in the paragraph will be shown as well.
When the -S option is used the number of lines in a paragraph is limited to a default of 100 lines. This number can be changed using the option
-S=<number>
Example
A search for the word "foo" showing only entire paragraphs when the paragraph is smaller than 25 lines.
findgrep -S=25 foo
If paragraphs larger than 25 lines is encountered only the matching line is shown.
-x : exclude file/directory pattern
Use this option to exclude files and/or directories that is otherwise included. File names and directories must first positively match their respectively specified patterns before this pattern is checked.
Example
Search for "foo" in .ini files, but exclude files that contain "system"
findgrep -e -x="system" foo ini
Note that this search expression will search for files in directories containing "system".
Search for "foo" in .ini files, but exclude directories that contain "system"
findgrep -ed -x="/system/" foo ini
Note that this search expression will search for files that contain the phrase "system".
Search for "foo" in .ini files, but exclude paths that contain "system". I.e., do not contain "system" in the directory name or file name.
findgrep -ed -x="system" foo ini
