ed, e, red - Edits a file line by line
ed [-p string] [-s] [-x] [-] [file]
e [-p string] [-s] [-x] [-] [file]
red [-p string] [-s] [-x] [-] [file]
The ed command is a line editing program that works on one file at a time
by copying it into a temporary edit buffer and making changes to that copy.
Interfaces documented on this reference page conform to industry standards
Refer to the standards(5) reference page for more information about indus-
try standards and associated tags.
Sets the ed prompt to string. The default for string is null (no
-s Suppresses byte counts that the editor displays with the e, E, r, and w
subcommands, suppresses diagnostic messages for the e, E, and q subcom-
mands, and suppresses the ! (exclamation point) prompt after a
!system_command. This also holds true if you invoke ed with a file
- Suppresses byte counts that the editor displays with the e, E, r, and w
subcommands, suppresses diagnostic messages for the e, E, and q subcom-
mands, and suppresses the ! (exclamation point) prompt after a
!system_command. This also holds true if you invoke ed with a file
name. This form is obsolescent and may be withdrawn in future
If you quit without writing the file, you do not get an error message.
(Generally, before ending the program, q checks to determine whether
the buffer was written to a file since the last time it was changed.
If not, q displays the ? message.)
-x [Tru64 UNIX] Prompts for an encryption key and unencrypts the file.
If the file specified is not encrypted or an incorrect key is entered,
garbled text is displayed.
The name of the file to be edited. If this operand is specified, ed
reads file into the internal buffer before reading standard input.
The ed command does not alter the file itself until you use the write sub-
command. You can specify on the command line the file you want to edit, or
you can use ed subcommands to read a file into the buffer. When ed reads a
new file into the buffer, the contents of that file replace the buffer's
previous contents, if any.
[Tru64 UNIX] There is also a restricted version of ed, the red command,
for use with the restricted shell (rsh). With red, you can edit only files
that reside in the current directory, or in the /tmp directory, and you
cannot use the !system_command subcommand (see Subcommands section).
[Tru64 UNIX] A file may be encrypted using the X subcommand. Care should
be take to remember the encryption key used as there is not a simple way to
unencrypt a file if the key used is lost. Only the first six characters of
the key are significant.
Modes for ed
The ed program operates in one of two modes, command mode and text mode.
In command mode, ed recognizes and executes subcommands. In text mode, ed
adds text to the file buffer, but does not recognize subcommands. You
enter text mode by using the a subcommand, c subcommand, or i subcommand.
To leave text mode, enter . (dot) alone at the beginning of a line.
[Tru64 UNIX] The e command is a supported alias for the ed command.
Syntax for Subcommands
An ed subcommand consists of zero, one, or two addresses, followed by a
single-character subcommand, possibly followed by arguments to that subcom-
mand. These addresses specify one or more lines in the buffer. Because
every subcommand has default addresses, you frequently do not need to
The ed command supports a limited form of special pattern-matching charac-
ters that you can use as regular expressions (REs) to construct pattern
strings. You can use these patterns in addresses to specify lines and in
some subcommands to specify portions of a line.
[Tru64 UNIX] For information about regular expressions (REs), see the grep
There are three types of ed addresses: line number addresses, addresses
relative to the current line, and pattern addresses.
The current line is the point of reference in the buffer and is represented
by a . (dot). When you start the ed program, the current line is the last
line in the buffer. As you edit a file, the current line changes to the
last line affected by a subcommand. The current line is the default
address for several ed commands. (See the Subcommands section to find out
how each subcommand affects the current line.)
Subcommands for determining and changing the current line are described in
the Subcommands section. The following are guidelines for constructing
+ A . (dot) addresses the current line.
+ A $ (dollar sign) addresses the last line of the buffer.
+ The number n addresses the nth line of the buffer.
+ The construct 'x addresses the line marked with a lowercase letter, x,
by the k subcommand.
+ The /pattern/ construct addresses the next line that contains a match-
ing string. The search begins with the line after the current line
and stops when it finds a match for the pattern. If necessary, the
search moves to the end of the buffer, wraps around to the beginning
of the buffer, and continues until it either finds a match or returns
to the current line.
+ The ?pattern? construct addresses the previous line that contains a
match for the pattern. The ?pattern? construct, like /pattern/, can
search the entire buffer, but it does so in the opposite direction.
+ An address followed by +number or -number specifies an address plus or
minus the indicated number of lines. (The + (plus sign) is optional.)
+ An address that begins with + or - specifies a line relative to the
current line. For example, -5 is the equivalent of .-5 (five lines
above the current line).
+ An address that ends with - or + specifies the line immediately before
(-) or immediately after (+) the addressed line. Used alone, the -
character addresses the line immediately before the current line. The
+ character addresses the line immediately after the current line;
however, the + character is optional. The + and - characters have a
cumulative effect; for example, the address -- addresses the line two
lines above the current line.
+ For convenience, a , (comma) stands for the address pair 1,$ (first
line through last line) and a ; (semicolon) stands for the pair .,$
(current line through last line).
Commands that do not accept addresses regard the presence of an address as
an error. Commands that do accept addresses can use either given or
default addresses. When given more addresses than it accepts, a command
uses the last (rightmost) ones.
In most cases, a , (comma) separates addresses (for example 2,8). A ;
(semicolon) can also separate addresses. A ; between addresses causes ed
to set the current line to the first address and then calculate the second
address (for example, to set the starting line for a search based on
guidelines 5 and 6). In a pair of addresses, the first must be numerically
smaller than the second.
Limits for ed
The following is a list of ed size limitations:
+ [Tru64 UNIX] 2048 bytes per line.
+ [Tru64 UNIX] 256 bytes per global subcommand list.
+ [Tru64 UNIX] 128 kilobyte buffer size. (The buffer not only contains
the original file, but also editing information.)
[Tru64 UNIX] The maximum number of lines depends on the amount of memory
available to you. The maximum file size depends on the amount of physical
data storage (disk or tape drive) available or on the maximum number of
lines permitted in user memory.
In most cases, only one ed subcommand can be entered on a line. The excep-
tions to this rule are the n, p, and l subcommands, which can be added to
any subcommand except e, E, f, q, Q, r, w, or !.
The e, E, f, r, and w subcommands accept file names as arguments. The ed
command stores the last file name used with a subcommand as a default file
name. The next e, E, f, r, or w given without a file name uses the default
The ed command responds to an error condition with one of two messages: ?
(question mark) or ?file. When the ed command receives an INT signal, it
displays a ? and returns to command mode. When ed reads a file, it dis-
cards NULL characters and all characters after the last newline character.
Unless otherwise noted, all subcommands work by default on the current
line; an address is optional. If you specify an address, you do not type
When used as an address, a . (dot) refers to the current line. When a .
(dot) is shown in the first position on an otherwise blank line, it ter-
minates text mode and returns to command mode. The text parameter denotes
user input in text mode. Note that address need not be a number; it can be
a regular expression of the form /RE/,/RE/ or /RE/;/RE/.
. The a (append) subcommand adds text to the buffer after the addressed
line. Enter a . (dot) to return to command mode. The a subcommand
sets the current line to the last inserted line, or, if no lines were
inserted, to the addressed line. Address 0 (zero) causes the a subcom-
mand to add text to the beginning of the buffer.
. The c (change) subcommand deletes the addressed lines, then replaces
them with new input. Enter a . (dot) to return to command mode. The c
command sets the current line to the last new line of input, or, if
there were none, to the first line that was not deleted.
The d (delete) subcommand removes the addressed lines from the buffer.
The line after the last line deleted becomes the current line. If the
deleted lines were originally at the end of the buffer, the new last
line becomes the current line.
The e (edit) subcommand first deletes any contents from the buffer,
then loads another file into the buffer, sets the current line to the
last line of the buffer, and displays the number of bytes read in to
the buffer. If the buffer was changed since its contents were last
saved (with the w subcommand), e displays ? before it clears the
The e subcommand stores file as the default file name to be used, if
necessary, by subsequent e, E, r, or w subcommands. (See the f subcom-
When the ! (exclamation point) character replaces file, e takes the
rest of the line as a shell (sh) command and reads the command output.
The e subcommand does not store the name of the shell command as a
default file name.
The E (Edit) subcommand works like e, with one exception: E does not
check for changes made to the buffer since the last w subcommand.
The f (file name) subcommand changes the default file name (the stored
name of the last file used) to file, if file is given. If file is not
given, the f subcommand prints the default file name.
The g (global) subcommand first marks every line that matches the pat-
tern. Then, for each marked line, this subcommand sets the current
line to that line and executes subcommand_list. Place a single subcom-
mand, or the first subcommand of a list, on the same line with the g
subcommand; place subsequent subcommands on separate lines. Except for
the last line, end each of these lines with a \ (backslash).
The subcommand_list can include the a, i, and c subcommands and their
input. If the last command in subcommand_list is normally the . (dot)
that ends input mode, the . (dot) is optional. If there is no
subcommand_list, ed displays the current line. The subcommand_list
cannot include the g, G, v, V, or ! subcommands.
The g subcommand is similar to the v subcommand, which executes
subcommand_list for every line that does not contain a match for the
pattern. This subcommand defaults to the entire file, not to the
The interactive G (Global) subcommand first marks every line that
matches the pattern, then displays the first marked line, sets the
current line to that line, and waits for a subcommand. The G subcommand
accepts any but the following ed subcommands: a, c, i, g, G, v, V, and
!. After the subcommand finishes, G displays the next marked line, and
so on. The G subcommand takes a newline character as a null subcom-
mand. The && subcommand causes G to execute the previous subcommand
again, if there is one. Subcommands executed within the G subcommand
can address and change any lines in the buffer. The G subcommand can
be terminated by pressing the Interrupt key sequence. This subcommand
defaults to the entire file, not to the current line.
h The h (help) subcommand displays a short message that explains the rea-
son for the most recent ? notification. The current line number is
H The H (Help) subcommand causes ed to enter a mode in which help mes-
sages (see the h command) are displayed for all subsequent ? notifica-
tions. The H subcommand toggles this mode, and is initially set to
"off". The current line number is unchanged.
. The i (insert) subcommand inserts text before the addressed line and
sets the current line to the last inserted line. Enter . (dot) to
return to command mode. If no lines are inserted, i sets the current
line to the addressed line. This subcommand differs from the a subcom-
mand only in the placement of the input text. Address 0 (zero) is not
legal for this subcommand.
The j (join) subcommand joins contiguous lines by removing the inter-
vening newline characters. If given only one address, j does nothing.
(For splitting lines, see the s subcommand.) Lines that exceed the
line length limit cannot be joined. If lines are joined, the current
line number is set to the address of the joined line; otherwise, the
current line number is unchanged.
The k (mark) subcommand marks the addressed line with name x, which
must be a lowercase letter. The address 'x (single quotation mark
before the marking character) then addresses this line. The k subcom-
mand does not change the current line. Marks attached to lines are
deleted with the line.
The l (list) subcommand displays the addressed lines. The l subcommand
wraps long lines and, unlike the p subcommand, represents nonprinting
characters as 3-digit octal numbers with a \ (backslash) preceding each
byte in the character. The following characters, however, are written
as escape sequences:
\r Carriage return
\v Vertical tab
A $ (dollar sign) character is placed at the end of each line, so that
a real (literal) $ at the end of a line cannot be misinterpreted.
An l subcommand can be appended to any ed subcommand except: e, E, f,
q, Q, r, w, or !.
The m (move) subcommand repositions the addressed lines. The first
moved line follows the line addressed by address3. Address 0 (zero)
for address3 causes m to move the addressed lines to the beginning of
the file. The line specified by address3 cannot be one of the lines to
be moved. The m subcommand sets the current line to the last moved
The n (number) subcommand displays the addressed lines, each preceded
by its line number and a tab character (displayed as spaces); the n
subcommand leaves the current line at the last line displayed. An n
subcommand can be appended to any ed subcommand except e, E, f, q, Q,
r, w, or !.
The p (print) subcommand displays the addressed lines and sets the
current line set to the last line displayed. A p subcommand can be
appended to any ed subcommand except: e, E, f, q, Q, r, w, or !. For
example, the subcommand dp deletes the current line and displays the
new current line.
P The P (Prompt) subcommand turns the ed prompt string * or the string
specified by the -p option on or off. Initially, P is off.
q The q (quit) subcommand exits the ed program. Before ending the pro-
gram, q checks to determine whether the buffer was written to a file
since the last time it was changed. If not, q displays the ? message.
You do not get more than one prompt in a row; a second consecutive q
quits the ed program without displaying a prompt.
Q The Q (Quit) subcommand exits the ed program without checking for
changes to the buffer since the last w subcommand (compare with the q
The r (read) subcommand reads a file into the buffer after the
addressed line; r does not delete the previous contents of the buffer.
When entered without file, r reads the default file, if any, into the
buffer (see e and f subcommands). The r subcommand does not change the
default file name. Address 0 (zero) causes r to read a file in at the
beginning of the buffer. After it reads a file successfully, r,
displays the number of bytes read into the buffer and sets the current
line to the last line read.
If ! (exclamation point) replaces file in a r subcommand, r takes the
rest of the line as a shell (sh) command whose output is to be read.
The r subcommand does not store the names of shell commands as default
The s (substitute) subcommand searches each addressed line for a string
that matches the pattern and then replaces the string with the speci-
fied replacement string. Without a number n or the global indicator g,
s replaces only the first matching string on each addressed line. With
n, s replaces the nth occurrence of pattern on the addressed line.
With the g indicator, s replaces every occurrence of the matching
string on each addressed line.
If s does not find a match for the pattern, it returns the error mes-
sage ?. Any character except a space or a newline character can
separate (delimit) the pattern and replacement. The s subcommand sets
the current line to the last line changed.
An && (ampersand) in the replacement string is a special symbol that has
the same value as the pattern string. So, for example, the subcommand
s/out/&&ing/ has the same effect as the subcommand s/out/outing/ and
replaces out with outing on the current line. A backslash before the
ampersand (\&&) removes this special meaning of && in replacement.
The % (percent sign), when used by itself as replacement, causes s to
use the previous replacement again. The % character does not have this
special meaning if it is part of a longer replacement or if it is pre-
ceded by a \ (backslash).
Lines can be split by substituting newline characters into them. In
replacement, the sequence \<<Return>> quotes the newline character (not
displayed) and moves the cursor to the nextline for the remainder of
The value of flags can be the following:
Substitutes for the countth occurrence only of the regular expres-
sion that is found on each addressed line.
g Substitutes globally for all nonoverlapping instances of the regu-
lar expression, instead of just substituting for the first
l Displays the final line in which a substitution was made in the
format specified for the l subcommand.
n Displays the final line in which a substitution was made in the
format specified for the n subcommand.
p Displays the final line in which a substitution was made in the
format specified for the p subcommand.
The t (transfer) subcommand inserts a copy of the addressed lines after
address3. The t subcommand accepts address 0 (for inserting lines at
the beginning of the buffer). The t subcommand sets the current line to
the last line copied.
u The u (undo) subcommand restores the buffer to the state it was in
before it was last modified by an ed subcommand. The commands that u
can undo are a, c, d, g, G, i, j, m, r, s, t, u, v, and V. All changes
made to the buffer by a g, G, v, or V global command are undone as a
single change. The current line number is set to the value it had
before the command being undone started.
The v subcommand executes the subcommands in subcommand_list for each
line that does not contain a match for the pattern. The v subcommand
is a complement for the global subcommand g, which executes
subcommand_list for every line that does contain a match for the pat-
The V subcommand first marks every line that does not match the pat-
tern, then displays the first marked line, sets the current line to
that line, and waits for a subcommand. The V subcommand complements
the G subcommand, which marks the lines that do match the pattern.
The w (write) subcommand copies the addressed lines from the buffer to
the file named in file. If the file does not exist, the w subcommand
creates it with permission mode 666 (read and write permission for
everyone), unless the umask setting specifies another file creation
mode. (For information about file permissions, see umask in sh and
The w subcommand does not change the default file name (unless file is
the first file name used since you started ed). If you do not provide
a file name, ed uses the default file name, if any (see the e, E, and f
subcommands). The w subcommand does not change the current line.
If ed successfully writes the file, it displays the number of charac-
ters written. When ! (exclamation point) replaces file, ed takes the
rest of the line as a shell (sh) command whose output is to be read; w
does not save shell command names as default file names (same effect as
!). The use of the write subcommand with ! (exclamation point) is not
considered the last w command that wrote the entire buffer. Thus, this
alone does not prevent the warning to the user if an attempt is made to
destroy the editor buffer through the e or q subcommands.
The address 0 is not a legal address for the w subcommand. Therefore,
it is not possible to create an empty file with ed.
X The X subcommand prompts you to enter an encryption key. Only the
first six characters of this key are significant. When the ed commands
write a file after the encryption key is specified, the output is writ-
ten in encrypted form. Subsequent edits of the file require the use the
-x option to display the file in its unencrypted form.
Without an address, the = (equal sign) subcommand displays the current
line number. With the address $, = (equal sign) displays the number of
the last line in the buffer. The = subcommand does not change the
current line and cannot be included in a g or v subcommand list.
The ! (exclamation point) subcommand allows system commands to be run
from within ed. Anything following ! on an ed subcommand line is
interpreted as a system command. Within the text of that command
string, ed replaces the unescaped character % with the current file
name, if there is one.
By default, the subcommands are executed by /usr/bin/sh. To support
single user mode operation, subcommands can be executed by /sbin/sh. To
change the default behavior, modify your $SHELL environmental variable
When used as the first character of a shell command (after the ! that
runs a subshell), ed replaces the ! character with the previous system
command; for example, the command !! repeats the previous system com-
mand. If the command interpreter (the sh command) expands the command
string, ed echoes the expanded line. The ! subcommand does not change
the current line. If any replacements of % and/or ! are performed, the
modified line is written to standard output before the command is exe-
The ed editor interprets a number alone on a line as an address and
displays the addressed line. Addresses can be absolute (line numbers
or $) or relative to the current line (+number or -number). Entering a
newline character (a blank line) is equivalent to +1p and is useful for
stepping forward through the buffer one line at a time.
The following exit values are returned:
0 Successful completion.
>>0 An error occurred.
The following actions are taken upon receipt of signals:
The ed command interrupts its current activity, writes the string ?\n
to standard output, and returns to command mode.
If the buffer is not empty and has changed since the last write, the ed
command attempts to write a copy of the buffer in a file. First, the
file named ed.hup in the current directory is used; if that fails, the
file named ed.hup in the directory named by the HOME environment vari-
able is used. In any case, the ed command exits without returning to
The following environment variables affect the execution of ed:
Provides a default value for the internationalization variables that
are unset or null. If LANG is unset or null, the corresponding value
from the default locale is used. If any of the internationalization
variables contain an invalid setting, the utility behaves as if none of
the variables had been defined.
If set to a non-empty string value, overrides the values of all the
other internationalization variables.
Determines the locale for the interpretation of sequences of bytes of
text data as characters (for example, single-byte as opposed to multi-
byte characters in arguments) and the behavior of character classes
within regular expressions.
Determines the locale for the format and contents of diagnostic mes-
sages written to standard error.
Determines the location of message catalogues for the processing of
Recovery file written if SIGHUP is received during operation.
Commands: chmod(1), edit(1), ex(1), grep(1), rsh(1), sed(1), Bourne shell
sh(1b), POSIX shell sh(1p), stty(1), vi(1)
Command and Shell User's Guide