NAME
sed - the stream editor
SYNOPSIS
sed [-n] [-g] [-e script] [-f sfile] [file ...]
DESCRIPTION
Sed copies the named files (standard input default) to the
standard output, edited according to a script of commands.
An -e option supplies a single edit command from the next
argument; if there are several of these they are executed in
the order in which they appear. If there is just one -e
option and no -f's, the -e flag may be omitted.
An -f option causes commands to be taken from the file
"sfile"; if there are several of these they are executed in
the order in which they appear; -e and -f commands may be
mixed.
The -g option causes sed to act as though every substitute
command in the script has a g suffix.
The -n option suppresses the default output.
A script consists of commands, one per line, of the follow-
ing form:
[address [, address] ] function [arguments]
Normally sed cyclically copies a line of input into a
current text buffer, then applies all commands whose
addresses select the buffer in sequence, then copies the
buffer to standard output and clears it.
The -n option suppresses normal output (so that only p and w
output is done). Also, some commands (n, N) do their own
line reads, and some others (d, D) cause all commands fol-
lowing in the script to be skipped (the D command also
suppresses the clearing of the current text buffer that
would normally occur before the next cycle).
It is also helpful to know that there's a second buffer
(called the `hold space' that can be copied or appended to
or from or swapped with the current text buffer.
An address is: a decimal numeral (which matches the line it
numbers where line numbers start at 1 and run cumulatively
across files), or a `$' that addresses the last line of
input, or a context address, which is a `/regular expres-
sion/', in the style of ed(1) modified thus:
(1) The escape sequence `\n' matches a newline embedded in
the buffer, and `\t' matches a tab.
(2) A command line with no addresses selects every buffer.
(3) A command line with one address selects every buffer
that matches that address.
(4) A command line with two addresses selects the inclusive
range from the first input buffer that matches the
first address through the next input buffer that
matches the second. (If the second address is a number
less than or equal to the line number first selected,
only one line is selected.) Once the second address is
matched sed starts looking for the first one again;
thus, any number of these ranges will be matched.
The negation operator '!' can prefix a command to apply it
to every line not selected by the address(es).
In the following list of functions, the maximum number of
addresses permitted for each function is indicated in
parentheses.
An argument denoted "text" consists of one or more lines,
with all but the last ending with `\' to hide the newline.
Backslashes in text are treated like backslashes in the
replacement string of an s command and may be used to pro-
tect initial whitespace (blanks and tabs) against the strip-
ping that is done on every line of the script.
An argument denoted "rfile" or "wfile" must be last on the
command line. Each wfile is created before processing
begins. There can be at most 10 distinct wfile arguments.
a "text" (1)
Append. Place text on output before reading the next
input line.
b "label" (2)
Branch to the `:' command bearing the label. If no
label is given, branch to the end of the script.
c "text" (2)
Change. Delete the current text buffer. With 0 or 1
address, or at the end of a 2-address range, place text
on the output. Start the next cycle.
d (2)
Delete the current text buffer. Start the next cycle.
D (2)
Delete the first line of the current text buffer (all
chars up to the first newline). Start the next cycle.
g (2)
Replace the contents of the current text buffer with
the contents of the hold space.
G (2)
Append the contents of the hold space to the current
text buffer.
h (2)
Copy the current text buffer into the hold space.
H (2)
Append a copy of the current text buffer to the hold
space.
i "text" (1)
Insert. Place text on the standard output.
l (2)
List. Sends the pattern space to standard output. A
"w" option may follow as in the s command below. Non-
printable characters expand to:
\b -- backspace (ASCII 08)
\t -- tab (ASCII 09)
\n -- newline (ASCII 10)
\r -- return (ASCII 13)
\e -- escape (ASCII 27)
\xx -- the ASCII character corresponding to 2 hex digits xx.
n (2)
Copy the current text buffer to standard output. Read
the next line of input into it.
N (2)
Append the next line of input to the current text
buffer, inserting an embedded newline between the two.
The current line number changes.
p (2)
Print. Copy the current text buffer to the standard
output.
P (2)
Copy the first line of the current text buffer (all
chars up to the first newline) to standard output.
q (1)
Quit. Branch to the end of the script. Do not start a
new cycle.
r "rfile" (1)
Read the contents of rfile. Place them on the output
before reading the next input line.
s /regular-expression/replacement/flags (2)
Substitute the replacement for instances of the regular
expression in the current text buffer. Any character
may be used instead of `/'. For a fuller description
see ed (1). Flags is zero or more of the following:
g -- Global. Substitute for all nonoverlapping
instances of the string rather than just the first
one.
p -- Print the pattern space if a replacement was made.
w -- Write. Append the current text buffer to a file
argument as in a w command if a replacement is
made. Standard output is used if no file argument
is given
t "label" (2)
Branch-if-test. Branch to the : command with the given
label if any substitutes have been made since the most
recent read of an input line or execution of a t or T.
If no label is given, branch to the end of the script.
T "label" (2)
Branch-on-error. Branch to the : command with the
given label if no substitutes have succeeded since the
last input line or t or T command. Branch to the end
of the script if no label is given.
w "wfile" (2)
Write. Append the current text buffer to wfile.
W "wfile" (2)
Write first line. Append first line of the current
text buffer to wfile.
x (2)
Exchange the contents of the current text buffer and
hold space.
y /string1/string2/ (2)
Translate. Replace each occurrence of a character in
string1 with the corresponding character in string2.
The lengths of these strings must be equal.
! "command" (2)
All-but. Apply the function (or group, if function is
{) only to lines not selected by the address(es).
: "label" (0)
This command does nothing but hold a label for b and t
commands to branch to.
= (1)
Place the current line number on the standard output as
a line.
{ (2)
Execute the following commands through a matching `}'
only when the current line matches the address or
address range given.
An empty command is ignored.
PORTABILITY
This tool was reverse-engineered from BSD 4.1 UNIX sed, and
(as far as the author's knowledge and tests can determine)
is compatible with it. All documented features of BSD 4.1
sed are supported.
One undocumented feature (a leading 'n' in the first comment
having the same effect as an -n command-line option) has
been omitted.
The following bugs and limitations have been fixed:
o There is no hidden length limit (40 in BSD sed) on w
file names.
o There is no limit (8 in BSD sed) on the length of
labels.
o The exchange command now works for long pattern and
hold spaces.
The following enhancements to existing commands have been
made:
o a, i commands don't insist on a leading backslash-\n in
the text.
o r, w commands don't insist on whitespace before the
filename.
o The g, p and P options on s commands may be given in
any order.
Some enhancements to regular-expression syntax have been
made:
o \t is recognized in REs (and elsewhere) as an escape
for tab.
o In an RE, + calls for 1..n repeats of the previous pat-
tern.
The following are completely new features:
o The l command (list, undocumented and weaker in BSD)
o The W command (write first line of pattern space to
file).
o The T command (branch on last substitute failed).
o Trailing comments are now allowed on command lines.
In addition, sed's error messages have been made more
specific and informative.
The implementation is also significantly smaller and faster
than BSD 4.1 sed. It uses only the standard I/O library and
exit(3).
NOTE
This is a freeware component of the GNU and MINIX operating
systems. The user is hereby granted permission to use,
modify, reproduce and distribute it subject to the following
conditions:
1. The authorship notice appearing in each source file may
not be altered or deleted.
2. The object form may not be distributed without source.
SEE ALSO
cgrep(1), fgrep(1), grep(1), lex(1), regexp(5), awk(9).
AUTHOR
Eric S. Raymond <esr@snark.thyrsus.com>