| Chapter 19. Here Documents | | Here and now, boys. --Aldous Huxley, Island |
A here document is a special-purposecode block. It uses a form of I/Oredirection to feed a command list toan interactive program or a command, such as ftp, cat,or the ex text editor. COMMAND <<InputComesFromHERE.........InputComesFromHERE |
A limit string delineates (frames)the command list. The special symbol << precedesthe limit string. This has the effect of redirecting the outputof a command block into the stdin of the programor command. It is similar to interactive-program <command-file, where command-filecontains The here document equivalent looks like this: interactive-program <<LimitStringcommand #1command #2...LimitString | Choose a limit string sufficientlyunusual that it will not occur anywhere in the command list andconfuse matters. Note that here documents may sometimesbe used to good effect with non-interactive utilities and commands,such as, for example, wall. Example 19-1. broadcast: Sends message to everyonelogged in #!/bin/bashwall <<zzz23EndOfMessagezzz23E-mail your noontime orders for pizza to the system administrator. (Add an extra dollar for anchovy or mushroom topping.)# Additional message text goes here.# Note: 'wall' prints comment lines.zzz23EndOfMessagezzz23# Could have been done more efficiently by# wall <message-file# However, embedding the message template in a script#+ is a quick-and-dirty one-off solution.exit |
Even such unlikely candidates as the vi text editor lend themselves to here documents. Example 19-2. dummyfile: Creates a 2-line dummyfile #!/bin/bash# Noninteractive use of 'vi' to edit a file.# Emulates 'sed'.E_BADARGS=85if [ -z "$1" ]then echo "Usage: `basename $0` filename" exit $E_BADARGSfiTARGETFILE=$1# Insert 2 lines in file, then save.#--------Begin here document-----------#vi $TARGETFILE <<x23LimitStringx23iThis is line 1 of the example file.This is line 2 of the example file.^[ZZx23LimitStringx23#----------End here document-----------## Note that ^[ above is a literal escape#+ typed by Control-V <Esc>.# Bram Moolenaar points out that this may not work with 'vim'#+ because of possible problems with terminal interaction.exit | The above script could just as effectively have been implemented withex, rather thanvi. Heredocuments containing a list of excommands are common enough to form their own category, known asex scripts. #!/bin/bash# Replace all instances of "Smith" with "Jones" #+ in files with a ".txt" filename suffix. ORIGINAL=SmithREPLACEMENT=Jonesfor word in $(fgrep -l $ORIGINAL *.txt)do # ------------------------------------- ex $word <<EOF :%s/$ORIGINAL/$REPLACEMENT/g :wqEOF # :%s is the "ex" substitution command. # :wq is write-and-quit. # -------------------------------------done |
Analogous to "ex scripts" are cat scripts. Example 19-3. Multi-line message using cat #!/bin/bash# 'echo' is fine for printing single line messages,#+ but somewhat problematic for for message blocks.# A 'cat' here document overcomes this limitation.cat <<End-of-message-------------------------------------This is line 1 of the message.This is line 2 of the message.This is line 3 of the message.This is line 4 of the message.This is the last line of the message.-------------------------------------End-of-message# Replacing line 7, above, with#+ cat > $Newfile <<End-of-message#+ ^^^^^^^^^^#+ writes the output to the file $Newfile, rather than to stdout.exit 0#--------------------------------------------# Code below disabled, due to "exit 0" above.# S.C. points out that the following also works.echo "-------------------------------------This is line 1 of the message.This is line 2 of the message.This is line 3 of the message.This is line 4 of the message.This is the last line of the message.-------------------------------------" # However, text may not include double quotes unless they are escaped. |
The - option to mark a here document limit string(<<-LimitString) suppresses leadingtabs (but not spaces) in the output. This may be useful in makinga script more readable. Example 19-4. Multi-line message, with tabs suppressed #!/bin/bash# Same as previous example, but...# The - option to a here document <<-#+ suppresses leading tabs in the body of the document,#+ but *not* spaces.cat <<-ENDOFMESSAGEThis is line 1 of the message.This is line 2 of the message.This is line 3 of the message.This is line 4 of the message.This is the last line of the message.ENDOFMESSAGE# The output of the script will be flush left.# Leading tab in each line will not show.# Above 5 lines of "message" prefaced by a tab, not spaces.# Spaces not affected by <<- .# Note that this option has no effect on *embedded* tabs.exit 0 |
A here document supports parameter andcommand substitution. It is therefore possible to pass differentparameters to the body of the here document, changing its outputaccordingly. Example 19-5. Here document with replaceable parameters #!/bin/bash# Another 'cat' here document, using parameter substitution.# Try it with no command-line parameters, ./scriptname# Try it with one command-line parameter, ./scriptname Mortimer# Try it with one two-word quoted command-line parameter,# ./scriptname "Mortimer Jones" CMDLINEPARAM=1 # Expect at least command-line parameter.if [ $# -ge $CMDLINEPARAM ]then name=$1 # If more than one command-line param, #+ then just take the first.else name="John Doe" # Default, if no command-line parameter.fi RESPONDENT="the author of this fine script" cat <<EndofmessageHello, there, $NAME.Greetings to you, $NAME, from $RESPONDENT.# This comment shows up in the output (why?).Endofmessage# Note that the blank lines show up in the output.# So does the comment.exit |
This is a useful script containing a here document with parameter substitution. Example 19-6. Upload a file pair to Sunsite incoming directory #!/bin/bash# upload.sh# Upload file pair (Filename.lsm, Filename.tar.gz)#+ to incoming directory at Sunsite/UNC (ibiblio.org).# Filename.tar.gz is the tarball itself.# Filename.lsm is the descriptor file.# Sunsite requires "lsm" file, otherwise will bounce contributions.E_ARGERROR=85if [ -z "$1" ]then echo "Usage: `basename $0` Filename-to-upload" exit $E_ARGERRORfi Filename=`basename $1` # Strips pathname out of file name.Server="ibiblio.org" Directory="/incoming/Linux" # These need not be hard-coded into script,#+ but may instead be changed to command-line argument.Password="your.e-mail.address" # Change above to suit.ftp -n $Server <<End-Of-Session# -n option disables auto-logonuser anonymous "$Password" # If this doesn't work, then try: # quote user anonymous "$Password" binarybell # Ring 'bell' after each file transfer.cd $Directoryput "$Filename.lsm" put "$Filename.tar.gz" byeEnd-Of-Sessionexit 0 |
Quoting or escaping the "limit string" at the head of a here document disables parameter substitution within itsbody. The reason for this is that quoting/escaping thelimit string effectively escapes the $,`, and special characters, and causes them tobe interpreted literally. (Thank you, Allen Halsey, for pointingthis out.) Example 19-7. Parameter substitution turned off #!/bin/bash# A 'cat' here-document, but with parameter substitution disabled.NAME="John Doe" RESPONDENT="the author of this fine script" cat <<'Endofmessage'Hello, there, $NAME.Greetings to you, $NAME, from $RESPONDENT.Endofmessage# No parameter substitution when the "limit string" is quoted or escaped.# Either of the following at the head of the here document would have#+ the same effect.# cat <<"Endofmessage" # cat <<Endofmessage# And, likewise:cat <<"SpecialCharTest" Directory listing would followif limit string were not quoted.`ls -l`Arithmetic expansion would take placeif limit string were not quoted.$((5 + 3))A a single backslash would echoif limit string were not quoted.SpecialCharTestexit |
Disabling parameter substitution permits outputting literal text. Generating scripts or even program code is one use for this. Example 19-8. A script that generates another script #!/bin/bash# generate-script.sh# Based on an idea by Albert Reiner.OUTFILE=generated.sh # Name of the file to generate.# -----------------------------------------------------------# 'Here document containing the body of the generated script.(cat <<'EOF'#!/bin/bashecho "This is a generated shell script." # Note that since we are inside a subshell,#+ we can't access variables in the "outside" script.echo "Generated file will be named: $OUTFILE" # Above line will not work as normally expected#+ because parameter expansion has been disabled.# Instead, the result is literal output.a=7b=3let "c = $a * $b" echo "c = $c" exit 0EOF) > $OUTFILE# -----------------------------------------------------------# Quoting the 'limit string' prevents variable expansion#+ within the body of the above 'here document.'# This permits outputting literal strings in the output file.if [ -f "$OUTFILE" ]then chmod 755 $OUTFILE # Make the generated file executable.else echo "Problem in creating file: "$OUTFILE"" fi# This method can also be used for generating#+ C programs, Perl programs, Python programs, Makefiles,#+ and the like.exit 0 |
It is possible to set a variable from the output of a here document.This is actually a devious form of command substitution. variable=$(cat <<SETVARThis variableruns over multiple lines.SETVAR)echo "$variable" |
A here document can supply input to a function in the same script. Example 19-9. Here documents and functions #!/bin/bash# here-function.shGetPersonalData (){ read firstname read lastname read address read city read state read zipcode} # This certainly appears to be an interactive function, but . . .# Supply input to the above function.GetPersonalData <<RECORD001BozoBozeman2726 Nondescript Dr.BozemanMT21226RECORD001echoecho "$firstname $lastname" echo "$address" echo "$city, $state $zipcode" echoexit 0 |
It is possible to use : as a dummy command accepting output from a here document. This, in effect, creates an"anonymous" here document. Example 19-10. "Anonymous" Here Document #!/bin/bash: <<TESTVARIABLES${HOSTNAME?}${USER?}${MAIL?} # Print error message if one of the variables not set.TESTVARIABLESexit $? |
| | A variation of the above technique permits "commenting out" blocks of code. | Example 19-11. Commenting out a block of code #!/bin/bash# commentblock.sh: <<COMMENTBLOCKecho "This line will not echo." This is a comment line missing the "#" prefix.This is another comment line missing the "#" prefix.&*@!!++=The above line will cause no error message,because the Bash interpreter will ignore it.COMMENTBLOCKecho "Exit value of above "COMMENTBLOCK" is $?." # 0# No error shown.echo# The above technique also comes in useful for commenting out#+ a block of working code for debugging purposes.# This saves having to put a "#" at the beginning of each line,#+ then having to go back and delete each "#" later.# Note that the use of of colon, above, is optional.echo "Just before commented-out code block." # The lines of code between the double-dashed lines will not execute.# ===================================================================: <<DEBUGXXXfor file in *do cat "$file" doneDEBUGXXX# ===================================================================echo "Just after commented-out code block." exit 0####################################################################### Note, however, that if a bracketed variable is contained within#+ the commented-out code block,#+ then this could cause problems.# for example:#/!/bin/bash : <<COMMENTBLOCK echo "This line will not echo." &*@!!++= ${foo_bar_bazz?} $(rm -rf /tmp/foobar/) $(touch my_build_directory/cups/Makefile)COMMENTBLOCK$ sh commented-bad.shcommented-bad.sh: line 3: foo_bar_bazz: parameter null or not set# The remedy for this is to strong-quote the 'COMMENTBLOCK' in line 49, above. : <<'COMMENTBLOCK'# Thank you, Kurt Pfeifle, for pointing this out. |
| | Yet another twist of this nifty trick makes "self-documenting" scripts possible. | Example 19-12. A self-documenting script #!/bin/bash# self-document.sh: self-documenting script# Modification of "colm.sh".DOC_REQUEST=70if [ "$1" = "-h" -o "$1" = "--help" ] # Request help.then echo; echo "Usage: $0 [directory-name]"; echo sed --silent -e '/DOCUMENTATIONXX$/,/^DOCUMENTATIONXX$/p' "$0" | sed -e '/DOCUMENTATIONXX$/d' exit $DOC_REQUEST; fi: <<DOCUMENTATIONXXList the statistics of a specified directory in tabular format.---------------------------------------------------------------The command-line parameter gives the directory to be listed.If no directory specified or directory specified cannot be read,then list the current working directory.DOCUMENTATIONXXif [ -z "$1" -o ! -r "$1" ]then directory=.else directory="$1" fi echo "Listing of "$directory":"; echo(printf "PERMISSIONS LINKS OWNER GROUP SIZE MONTH DAY HH:MM PROG-NAME" ; ls -l "$directory" | sed 1d) | column -texit 0 | Using a cat script is an alternate way of accomplishing this. DOC_REQUEST=70if [ "$1" = "-h" -o "$1" = "--help" ] # Request help.then # Use a "cat script" . . . cat <<DOCUMENTATIONXXList the statistics of a specified directory in tabular format.---------------------------------------------------------------The command-line parameter gives the directory to be listed.If no directory specified or directory specified cannot be read,then list the current working directory.DOCUMENTATIONXXexit $DOC_REQUESTfi | See also Example A-28, Example A-40, Example A-41, and Example A-42 for more examples of self-documenting scripts.
| | Here documents create temporary files, but these files are deleted after opening and are not accessible to any other process. bash$ bash -c 'lsof -a -p $$ -d0' << EOF> EOFlsof 1213 bozo 0r REG 3,5 0 30386 /tmp/t1213-0-sh (deleted) | | | | Some utilities will not work inside a here document. |
| | The closing limit string, on the final line of a here document, must start in the first character position. There can be no leading whitespace. Trailing whitespace after the limit string likewise causes unexpected behavior. The whitespace prevents the limit string from being recognized. #!/bin/bashecho "----------------------------------------------------------------------" cat <<LimitStringecho "This is line 1 of the message inside the here document." echo "This is line 2 of the message inside the here document." echo "This is the final line of the message inside the here document." LimitString#^^^^Indented limit string. Error! This script will not behave as expected.echo "----------------------------------------------------------------------" # These comments are outside the 'here document',#+ and should not echo.echo "Outside the here document." exit 0echo "This line had better not echo." # Follows an 'exit' command. | | | | Some people very cleverly use a single ! as a limit string. But, that's not necessarily a good idea. # This works.cat <<!Hello!! Three more exclamations !!!!# But . . .cat <<!Hello!Single exclamation point follows!!!# Crashes with an error message.# However, the following will work.cat <<EOFHello!Single exclamation point follows!!EOF# It's safer to use a multi-character limit string. | | For those tasks too complex for a here document, consider using the expect scripting language, which was specifically designed for feeding input into interactive programs.
19.1. Here Strings
A here string can be considered as a stripped-down form of a here document.
It consists of nothing more than COMMAND <<< $WORD,
where $WORD is expanded and fed to the stdin of COMMAND.
As a simple example, consider this alternative to the echo-grep construction. # Instead of:if echo "$VAR" | grep -q txt # if [[ $VAR = *txt* ]]# etc.# Try:if grep -q "txt" <<< "$VAR" then # ^^^ echo "$VAR contains the substring sequence "txt"" fi# Thank you, Sebastian Kaminski, for the suggestion. |
Or, in combination with read: String="This is a string of words." read -r -a Words <<< "$String" # The -a option to "read" #+ assigns the resulting values to successive members of an array.echo "First word in String is: ${Words[0]}" # Thisecho "Second word in String is: ${Words[1]}" # isecho "Third word in String is: ${Words[2]}" # aecho "Fourth word in String is: ${Words[3]}" # stringecho "Fifth word in String is: ${Words[4]}" # ofecho "Sixth word in String is: ${Words[5]}" # words.echo "Seventh word in String is: ${Words[6]}" # (null) # Past end of $String.# Thank you, Francisco Lobo, for the suggestion. | It is, of course, possible to feed the output of a here string into the stdin of a loop. # As Seamus points out . . .ArrayVar=( element0 element1 element2 {A..D} )while read element ; do echo "$element" 1>&2done <<< $(echo ${ArrayVar[*]})# element0 element1 element2 A B C D |
Example 19-13. Prepending a line to a file #!/bin/bash# prepend.sh: Add text at beginning of file.## Example contributed by Kenny Stauffer,#+ and slightly modified by document author.E_NOSUCHFILE=85read -p "File: " file # -p arg to 'read' displays prompt.if [ ! -e "$file" ]then # Bail out if no such file. echo "File $file not found." exit $E_NOSUCHFILEfiread -p "Title: " titlecat - $file <<<$title > $file.newecho "Modified file is $file.new" exit # Ends script execution. from 'man bash': Here Strings A variant of here documents, the format is: <<<word The word is expanded and supplied to the command on its standard input. Of course, the following also works: sed -e '1i Title: ' $file | Example 19-14. Parsing a mailbox #!/bin/bash# Script by Francisco Lobo,#+ and slightly modified and commented by ABS Guide author.# Used in ABS Guide with permission. (Thank you!)# This script will not run under Bash versions -lt 3.0.E_MISSING_ARG=87if [ -z "$1" ]then echo "Usage: $0 mailbox-file" exit $E_MISSING_ARGfimbox_grep() # Parse mailbox file.{ declare -i body=0 match=0 declare -a date sender declare mail header value while IFS= read -r mail# ^^^^ Reset $IFS.# Otherwise "read" will strip leading & trailing space from its input. do if [[ $mail =~ "^From " ]] # Match "From" field in message. then (( body = 0 )) # "Zero out" variables. (( match = 0 )) unset date elif (( body )) then (( match )) # echo "$mail" # Uncomment above line if you want entire body #+ of message to display. elif [[ $mail ]]; then IFS=: read -r header value <<< "$mail" # ^^^ "here string" case "$header" in [Ff][Rr][Oo][Mm] ) [[ $value =~ "$2" ]] && (( match++ )) ; # Match "From" line. [Dd][Aa][Tt][Ee] ) read -r -a date <<< "$value" ; # ^^^ # Match "Date" line. [Rr][Ee][Cc][Ee][Ii][Vv][Ee][Dd] ) read -r -a sender <<< "$value" ; # ^^^ # Match IP Address (may be spoofed). esac else (( body++ )) (( match )) && echo "MESSAGE ${date:+of: ${date[*]} }" # Entire $date array ^ echo "IP address of sender: ${sender[1]}" # Second field of "Received" line ^ fi done < "$1" # Redirect stdout of file into loop.}mbox_grep "$1" # Send mailbox file to function.exit $?# Exercises:# ---------# 1) Break the single function, above, into multiple functions,#+ for the sake of readability.# 2) Add additional parsing to the script, checking for various keywords.$ mailbox_grep.sh scam_mail MESSAGE of Thu, 5 Jan 2006 08:00:56 -0500 (EST) IP address of sender: 196.3.62.4 | Exercise: Find other uses for here strings, such as, for example, feeding input to dc. |
| |
