ed is a line-oriented text editor. It is
used to create, display, modify, and otherwise manipulate text files. If
invoked with a file argument, then a copy of
file is read into the editor's buffer. Changes are
made to this copy and not directly to file itself.
ed, any changes not explicitly saved
w command are lost.
Editing is done in two distinct modes:
When first invoked,
ed is in command mode. In this
mode, commands are read from the standard input and executed to manipulate
the contents of the editor buffer.
A typical command might look like:
which replaces all occurrences of the string "old" with "new".
When an input command, such as
i (insert), or
c (change) is
ed enters input mode. This is the primary
means of adding text to a file. In this mode, no commands are available;
instead, the standard input is written directly to the editor buffer. Lines
consist of text up to and including a newline character. Input mode is
terminated by entering a single period
.’) on a line.
ed commands operate on whole lines or
ranges of lines; e.g., the
d command deletes lines;
m command moves lines, and so on. It is possible
to modify only a portion of a line by means of replacement, as in the
example above. However, even here, the
s command is
applied to whole lines at a time.
ed commands consist of zero or
more line addresses, followed by a single character command and possibly
additional parameters; i.e., commands have the structure:
The address(es) indicate the line or range of lines to be affected by the command. If fewer addresses are given than the command accepts, then default addresses are supplied.
ed commands and line addresses
support basic regular expressions (BREs). See
re_format(7) for more information on regular expressions.
The options are as follows:
- Same as the
- Specifies a command prompt. This may be toggled on and off with the
- Suppress diagnostics. This should be used if
edstandard input is from a script.
- Specifies the name of a file to read. If file is
prefixed with a bang (‘
!’), then it is interpreted as a shell command. In this case, what is read is the standard output of file executed via sh(1). To read a file whose name begins with a bang, prefix the name with a backslash (‘
\’). The default filename is set to file only if it is not prefixed with a bang.
An address represents the number of a line in the buffer.
ed maintains a
address which is typically supplied to commands as the default
address when none is specified. When a file is first read, the current
address is set to the last line of the file. In general, the current address
is set to the last line affected by a command.
A line address is constructed from one of the bases in the list
below, optionally followed by a numeric offset. The offset may include any
combination of digits, operators (e.g.,
^’), and whitespace. Addresses are
read from left to right, and their values are computed relative to the
One exception to the rule that addresses represent line numbers is the address 0 (zero). This means “before the first line”, and is legal wherever it makes sense.
An address range is two addresses separated either by a comma or semi-colon. The value of the first address in a range cannot exceed the value of the second. If only one address is given in a range, then the second address is set to the given address. If an n-tuple of addresses is given where n > 2, then the corresponding range is determined by the last two addresses in the n-tuple. If only one address is expected, then the last address is used.
Each address in a comma-delimited range is interpreted relative to the current address. In a semi-colon-delimited range, the first address is used to set the current address, and the second address is interpreted relative to the first.
The following address symbols are recognized:
- The current line (address) in the buffer.
- The last line in the buffer.
- The nth line in the buffer, where n is a number in the range [0,$].
- - or ^
- The previous line. This is equivalent to -1 and may be repeated with cumulative effect.
- -n or ^n
- The nth previous line, where n is a non-negative number.
- The next line. This is equivalent to +1 and may be repeated with cumulative effect.
- The nth next line, where n is a non-negative number.
- , or %
- The first through last lines in the buffer. This is equivalent to the address range 1,$.
- The current through last lines in the buffer. This is equivalent to the address range .,$.
- The next line containing the regular expression re. The search wraps to the beginning of the buffer and continues down to the current line, if necessary. The second slash can be omitted if it ends a line. "//" repeats the last search.
- The previous line containing the regular expression re. The search wraps to the end of the buffer and continues up to the current line, if necessary. The second question mark can be omitted if it ends a line. "??" repeats the last search.
- The line previously marked by a
k(mark) command, where lc is a lower case letter.
ed commands are single characters,
though some require additional parameters. If a command's parameters extend
over several lines, then each line except for the last must be terminated
with a backslash (‘
In general, at most one command is allowed per line. However, most
commands accept a print suffix, which is any of
l (list), or
(enumerate), to print the last line affected by the command.
ed recognizes the following commands. The
commands are shown together with the default address or address range
supplied if none is specified (in parentheses), and other possible arguments
on the right.
- Appends text to the buffer after the addressed line. Text is entered in input mode. The current address is set to last line entered.
- Changes lines in the buffer. The addressed lines are deleted from the buffer, and text is appended in their place. Text is entered in input mode. The current address is set to last line entered.
- Deletes the addressed lines from the buffer. If there is a line after the deleted range, then the current address is set to this line. Otherwise the current address is set to the line before the deleted range.
- Edits file, and sets the default filename. If file is not specified, then the default filename is used. Any lines in the buffer are deleted before the new file is read. The current address is set to the last line read.
- Edits the standard output of
!command below). The default filename is unchanged. Any lines in the buffer are deleted before the output of command is read. The current address is set to the last line read.
- Edits file unconditionally. This is similar to the
ecommand, except that unwritten changes are discarded without warning. The current address is set to the last line read.
- Sets the default filename to file. If file is not specified, then the default unescaped filename is printed.
- Mark each addressed line matching the regular expression
re for modification. The current address is set to
each marked line in turn, and then the command-list
is executed each time. The command-list can change the current line
number, and it is not changed back after the command-list ended. When a
marked line is changed, it is unmarked and the command-list won't be
executed for it any more. If no lines were matched, the current line
number remains unchanged.
Each command in command-list must be on a separate line, and every line except for the last must be terminated by a backslash (‘\’). Any commands are allowed, except for
V. An empty command-list is equivalent to a
pcommand — unlike for the
Gcommand, where an empty command-list does nothing, and unlike an empty command, which is equivalent to the command
+p. If the command-list is empty, the trailing slash can be omitted.
- Interactively edits the addressed lines matching a regular expression
re. The trailing slash after
re can be omitted. For each matching line, the line
is printed, the current address is set, and the user is prompted to enter
a command-list. At the end of the
Gcommand, the current address is set to the last line affected by (the last) command-list. If no lines were matched, the current line number remains unchanged.
The format of command-list is the same as that of the
gcommand, but an empty command list does nothing. A single ‘&’ repeats the last non-empty command list.
- Toggles the printing of error explanations. By default, explanations are
not printed. It is recommended that
edscripts begin with this command to aid in debugging.
- Prints an explanation of the last error.
- Inserts text in the buffer before the current line. Text is entered in input mode. The current address is set to the last line entered.
- Joins the addressed lines. The addressed lines are deleted from the buffer and replaced by a single line containing their joined text. The current address is set to the resultant line.
- Marks a line with a lower case letter lc. The line
can then be addressed as
'lc (i.e., a single quote followed by lc) in subsequent commands. The mark is not cleared until the line is deleted or otherwise modified.
- Prints the addressed lines unambiguously. The current address is set to the last line printed.
- Moves lines in the buffer. The addressed lines are moved to after the right-hand destination address, which may be the address 0 (zero). The current address is set to the last line moved.
- Prints the addressed lines along with their line numbers. The current address is set to the last line printed.
- Prints the addressed lines. The current address is set to the last line printed.
- Toggles the command prompt on and off. Unless a prompt was specified with
the command-line option
-pstring, the command prompt is by default turned off.
edunconditionally. This is similar to the
qcommand, except that unwritten changes are discarded without warning.
- Reads file to after the addressed line. If file is not specified, then the default filename is used. If there was no default filename prior to the command, then the default filename is set to file. Otherwise, the default filename is unchanged. The current address is set to the last line read.
- Reads to after the addressed line the standard output of
!command below). The default filename is unchanged. The current address is set to the last line read.
- Replaces text in the addressed lines matching a regular expression
re with replacement. By
default, only the first match in each line is replaced. If the
g(global) suffix is given, then every match is replaced. The n suffix, where n is a positive number, causes only the nth match to be replaced. It is an error if no substitutions are performed on any of the addressed lines. The current address is set to the last line affected.
re and replacement may be delimited by any character other than space and newline (see the
scommand below). If one or two of the last delimiters is omitted, then the last line affected is printed as though the print suffix
An unescaped ‘
&’ in replacement is replaced by the currently matched text. The character sequence \m, where m is a number in the range [1,9], is replaced by the mth backreference expression of the matched text. If replacement consists of a single ‘
%’, then replacement from the last substitution is used. Newlines may be embedded in replacement if they are escaped with a backslash (‘
- Repeats the last substitution. This form of the
scommand accepts a count suffix n, or any combination of the characters
p. If a count suffix n is given, then only the nth match is replaced. The
rsuffix causes the regular expression of the last search to be used instead of that of the last substitution. The
gsuffix toggles the global suffix of the last substitution. The
psuffix toggles the print suffix of the last substitution. The current address is set to the last line affected.
- Copies (i.e., transfers) the addressed lines to after the right-hand destination address, which may be the address 0 (zero). The current address is set to the last line copied.
- Undoes the last command and restores the current address to what it was
before the command. The global commands
Vare treated as a single command by undo.
uis its own inverse.
- The same as the
gcommand, except that it applies command-list to each of the addressed lines not matching the regular expression re.
- The same as the
Gcommand, except that it interactively edits the addressed lines not matching the regular expression re.
- Writes the addressed lines to file. Any previous contents of file are lost without warning. If there is no default filename, then the default filename is set to file, otherwise it is unchanged. If no filename is specified, then the default filename is used. The current address is unchanged.
- Writes the addressed lines to file, and then
- Writes the addressed lines to the standard input of
!command below). The default filename and current address are unchanged.
- Appends the addressed lines to the end of file. This
is similar to the
wcommand, except that the previous contents of file are not clobbered. The current address is unchanged.
- Scrolls n lines at a time starting at addressed line. If n is not specified, then the current window size is used. The current address is set to the last line printed.
- Prints the line number of the addressed line.
- An address without a command prints the addressed line and sets the current address to that line. If the address is also omitted, it defaults to the next line (+).
- Executes command via
sh(1). If the
first character of command is ‘!’,
then it is replaced by text of the previous
eddoes not process command for ‘\’ (backslash) escapes. However, an unescaped ‘%’ is replaced by the default filename. When the shell returns from execution, a ‘!’ is printed to the standard output. The current line is unchanged.
- If the current buffer has changed since it was last written,
edattempts to write the buffer to the file ed.hup. Nothing is written to the currently remembered file, and
- When an interrupt occurs,
edprints ‘?\n’ and returns to command mode. If interrupted during text input, the text already input is written to the current buffer, as if text input had been normally terminated.
- This signal is ignored.
- The screen is resized.
- buffer file
edattempts to write the buffer if the terminal hangs up
ed utility exits 0 on success,
and >0 if an error occurs.
When an error occurs,
ed prints a
‘?’ and either returns to command mode or exits if its input
is from a script. An explanation of the last error can be printed with the
h (help) command.
g (global) command masks any
errors from failed searches and substitutions, it can be used to perform
conditional operations in scripts; e.g.,
replaces any occurrences of "old" with "new".
u (undo) command occurs in a global
command list, then the command list is executed only once.
If diagnostics are not disabled, attempting to quit
ed or edit another file before writing a modified
buffer results in an error. If the command is entered a second time, it
succeeds, but any changes to the buffer are lost.
sed(1), sh(1), vi(1), re_format(7)
B. W. Kernighan, A Tutorial Introduction to the UNIX Text Editor.
B. W. Kernighan, Advanced Editing on UNIX.
B. W. Kernighan and P. J. Plauger, Software Tools in Pascal, 1981, Addison-Wesley.
ed utility is compliant with the
IEEE Std 1003.1-2008 (“POSIX.1”)
s (to repeat the last
z as well as the address specifier ‘%’
are extensions to that specification.
The IEEE Std 1003.1-2008
(“POSIX.1”) specification says the ‘^’
address specifier is neither required nor prohibited; additionally, it says
behaviour for the
- option is
ed command appeared in
Version 1 AT&T UNIX.
ed processes file
arguments for backslash escapes, i.e., in a filename, any characters
preceded by a backslash (‘
If a text (non-binary) file is not terminated by a newline
ed appends one on reading/writing
it. In the case of a binary file,
ed does not append
a newline on reading/writing.