basics-reviewed.bash

BASH Shell 编程 经典教程 《高级SHELL脚本编程》中文版

#!/bin/bash
# basics-reviewed.bash

# File extension == *.bash == specific to Bash

#   Copyright (c) Michael S. Zick, 2003; All rights reserved.
#   License: Use in any form, for any purpose.
#   Revision: $ID$
#
#              Edited for layout by M.C.
#   (author of the "Advanced Bash Scripting Guide")


#  This script tested under Bash versions 2.04, 2.05a and 2.05b.
#  It may not work with earlier versions.
#  This demonstration script generates one --intentional--
#+ "command not found" error message. See line 394.

#  The current Bash maintainer, Chet Ramey, has fixed the items noted
#+ for an upcoming version of Bash.



        ###-------------------------------------------###
        ###  Pipe the output of this script to 'more' ###
        ###+ else it will scroll off the page.        ###
        ###                                           ###
        ###  You may also redirect its output         ###
        ###+ to a file for examination.               ###  
        ###-------------------------------------------###



#  Most of the following points are described at length in
#+ the text of the foregoing "Advanced Bash Scripting Guide."
#  This demonstration script is mostly just a reorganized presentation.
#      -- msz

# Variables are not typed unless otherwise specified.

#  Variables are named. Names must contain a non-digit.
#  File descriptor names (as in, for example: 2>&1)
#+ contain ONLY digits.

# Parameters and Bash array elements are numbered.
# (Parameters are very similar to Bash arrays.)

# A variable name may be undefined (null reference).
unset VarNull

# A variable name may be defined but empty (null contents).
VarEmpty=''                         # Two, adjacent, single quotes.

# A variable name my be defined and non-empty
VarSomething='Literal'

# A variable may contain:
#   * A whole number as a signed 32-bit (or larger) integer
#   * A string
# A variable may also be an array.

#  A string may contain embedded blanks and may be treated
#+ as if it where a function name with optional arguments.

#  The names of variables and the names of functions
#+ are in different namespaces.


#  A variable may be defined as a Bash array either explicitly or
#+ implicitly by the syntax of the assignment statement.
#  Explicit:
declare -a ArrayVar



# The echo command is a built-in.
echo $VarSomething

# The printf command is a built-in.
# Translate %s as: String-Format
printf %s $VarSomething         # No linebreak specified, none output.
echo                            # Default, only linebreak output.




# The Bash parser word breaks on whitespace.
# Whitespace, or the lack of it is significant.
# (This holds true in general; there are, of course, exceptions.)




# Translate the DOLLAR_SIGN character as: Content-Of.

# Extended-Syntax way of writing Content-Of:
echo ${VarSomething}

#  The ${ ... } Extended-Syntax allows more than just the variable
#+ name to be specified.
#  In general, $VarSomething can always be written as: ${VarSomething}.

# Call this script with arguments to see the following in action.



#  Outside of double-quotes, the special characters @ and *
#+ specify identical behavior.
#  May be pronounced as: All-Elements-Of.

#  Without specification of a name, they refer to the
#+ pre-defined parameter Bash-Array.



# Glob-Pattern references
echo $*                         # All parameters to script or function
echo ${*}                       # Same

# Bash disables filename expansion for Glob-Patterns.
# Only character matching is active.


# All-Elements-Of references
echo $@                         # Same as above
echo ${@}                       # Same as above




#  Within double-quotes, the behavior of Glob-Pattern references
#+ depends on the setting of IFS (Input Field Separator).
#  Within double-quotes, All-Elements-Of references behave the same.


#  Specifying only the name of a variable holding a string refers
#+ to all elements (characters) of a string.


#  To specify an element (character) of a string,
#+ the Extended-Syntax reference notation (see below) MAY be used.




#  Specifying only the name of a Bash array references
#+ the subscript zero element,
#+ NOT the FIRST DEFINED nor the FIRST WITH CONTENTS element.

#  Additional qualification is needed to reference other elements,
#+ which means that the reference MUST be written in Extended-Syntax.
#  The general form is: ${name[subscript]}.

#  The string forms may also be used: ${name:subscript}
#+ for Bash-Arrays when referencing the subscript zero element.


# Bash-Arrays are implemented internally as linked lists,
#+ not as a fixed area of storage as in some programming languages.


#   Characteristics of Bash arrays (Bash-Arrays):
#   --------------------------------------------

#   If not otherwise specified, Bash-Array subscripts begin with
#+  subscript number zero. Literally: [0]
#   This is called zero-based indexing.
###
#   If not otherwise specified, Bash-Arrays are subscript packed
#+  (sequential subscripts without subscript gaps).
###
#   Negative subscripts are not allowed.
###
#   Elements of a Bash-Array need not all be of the same type.
###
#   Elements of a Bash-Array may be undefined (null reference).
#       That is, a Bash-Array my be "subscript sparse."
###
#   Elements of a Bash-Array may be defined and empty (null contents).
###
#   Elements of a Bash-Array may contain:
#     * A whole number as a signed 32-bit (or larger) integer
#     * A string
#     * A string formated so that it appears to be a function name
#     + with optional arguments
###
#   Defined elements of a Bash-Array may be undefined (unset).
#       That is, a subscript packed Bash-Array may be changed
#   +   into a subscript sparse Bash-Array.
###
#   Elements may be added to a Bash-Array by defining an element
#+  not previously defined.
###
# For these reasons, I have been calling them "Bash-Arrays".
# I'll return to the generic term "array" from now on.
#     -- msz




#  Demo time -- initialize the previously declared ArrayVar as a
#+ sparse array.
#  (The 'unset ... ' is just documentation here.)

unset ArrayVar[0]                   # Just for the record
ArrayVar[1]=one                     # Unquoted literal
ArrayVar[2]=''                      # Defined, and empty
unset ArrayVar[3]                   # Just for the record
ArrayVar[4]='four'                  # Quoted literal



# Translate the %q format as: Quoted-Respecting-IFS-Rules.
echo
echo '- - Outside of double-quotes - -'
###
printf %q ${ArrayVar[*]}            # Glob-Pattern All-Elements-Of
echo
echo 'echo command:'${ArrayVar[*]}
###
printf %q ${ArrayVar[@]}            # All-Elements-Of
echo
echo 'echo command:'${ArrayVar[@]}

# The use of double-quotes may be translated as: Enable-Substitution.

# There are five cases recognized for the IFS setting.

echo
echo '- - Within double-quotes - Default IFS of space-tab-newline - -'
IFS=$'\x20'$'\x09'$'\x0A'           #  These three bytes,
                                    #+ in exactly this order.


printf %q "${ArrayVar[*]}"          # Glob-Pattern All-Elements-Of
echo
echo 'echo command:'"${ArrayVar[*]}"
###
printf %q "${ArrayVar[@]}"          # All-Elements-Of
echo
echo 'echo command:'"${ArrayVar[@]}"


echo
echo '- - Within double-quotes - First character of IFS is ^ - -'
# Any printing, non-whitespace character should do the same.
IFS='^'$IFS                         # ^ + space tab newline
###
printf %q "${ArrayVar[*]}"          # Glob-Pattern All-Elements-Of
echo
echo 'echo command:'"${ArrayVar[*]}"
###
printf %q "${ArrayVar[@]}"          # All-Elements-Of
echo
echo 'echo command:'"${ArrayVar[@]}"


echo
echo '- - Within double-quotes - Without whitespace in IFS - -'
IFS='^:%!'
###
printf %q "${ArrayVar[*]}"          # Glob-Pattern All-Elements-Of
echo
echo 'echo command:'"${ArrayVar[*]}"
###
printf %q "${ArrayVar[@]}"          # All-Elements-Of
echo
echo 'echo command:'"${ArrayVar[@]}"


echo
echo '- - Within double-quotes - IFS set and empty - -'
IFS=''
###
printf %q "${ArrayVar[*]}"          # Glob-Pattern All-Elements-Of
echo
echo 'echo command:'"${ArrayVar[*]}"
###
printf %q "${ArrayVar[@]}"          # All-Elements-Of
echo
echo 'echo command:'"${ArrayVar[@]}"


echo
echo '- - Within double-quotes - IFS undefined - -'
unset IFS
###
printf %q "${ArrayVar[*]}"          # Glob-Pattern All-Elements-Of
echo
echo 'echo command:'"${ArrayVar[*]}"
###
printf %q "${ArrayVar[@]}"          # All-Elements-Of
echo
echo 'echo command:'"${ArrayVar[@]}"


# Put IFS back to the default.
# Default is exactly these three bytes.
IFS=$'\x20'$'\x09'$'\x0A'           # In exactly this order.

# Interpretation of the above outputs:
#   A Glob-Pattern is I/O; the setting of IFS matters.
###
#   An All-Elements-Of does not consider IFS settings.
###
#   Note the different output using the echo command and the
#+  quoted format operator of the printf command.


#  Recall:
#   Parameters are similar to arrays and have the similar behaviors.
###
#  The above examples demonstrate the possible variations.
#  To retain the shape of a sparse array, additional script
#+ programming is required.
###
#  The source code of Bash has a routine to output the
#+ [subscript]=value   array assignment format.
#  As of version 2.05b, that routine is not used,
#+ but that might change in future releases.



# The length of a string, measured in non-null elements (characters):
echo
echo '- - Non-quoted references - -'
echo 'Non-Null character count: '${#VarSomething}' characters.'

# test='Lit'$'\x00''eral'           # $'\x00' is a null character.
# echo ${#test}                     # See that?



#  The length of an array, measured in defined elements,
#+ including null content elements.
echo
echo 'Defined content count: '${#ArrayVar[@]}' elements.'
# That is NOT the maximum subscript (4).
# That is NOT the range of the subscripts (1 . . 4 inclusive).
# It IS the length of the linked list.
###
#  Both the maximum subscript and the range of the subscripts may
#+ be found with additional script programming.

# The length of a string, measured in non-null elements (characters):
echo
echo '- - Quoted, Glob-Pattern references - -'
echo 'Non-Null character count: '"${#VarSomething}"' characters.'

#  The length of an array, measured in defined elements,
#+ including null-content elements.
echo
echo 'Defined element count: '"${#ArrayVar[*]}"' elements.'

#  Interpretation: Substitution does not effect the ${# ... } operation.
#  Suggestion:
#  Always use the All-Elements-Of character
#+ if that is what is intended (independence from IFS).



#  Define a simple function.
#  I include an underscore in the name
#+ to make it distinctive in the examples below.
###
#  Bash separates variable names and function names
#+ in different namespaces.
#  The Mark-One eyeball isn't that advanced.
###
_simple() {
    echo -n 'SimpleFunc'$@          #  Newlines are swallowed in
}                                   #+ result returned in any case.


# The ( ... ) notation invokes a command or function.
# The $( ... ) notation is pronounced: Result-Of.


# Invoke the function _simple
echo
echo '- - Output of function _simple - -'
_simple                             # Try passing arguments.
echo
# or
(_simple)                           # Try passing arguments.
echo

echo '- Is there a variable of that name? -'
echo $_simple not defined           # No variable by that name.

# Invoke the result of function _simple (Error msg intended)

###
$(_simple)                          # Gives an error message:
#                          line 394: SimpleFunc: command not found
#                          ---------------------------------------

echo
###

#  The first word of the result of function _simple
#+ is neither a valid Bash command nor the name of a defined function.
###
# This demonstrates that the output of _simple is subject to evaluation.
###
# Interpretation:
#   A function can be used to generate in-line Bash commands.


# A simple function where the first word of result IS a bash command:
###
_print() {
    echo -n 'printf %q '$@
}

echo '- - Outputs of function _print - -'
_print parm1 parm2                  # An Output NOT A Command.
echo

$(_print parm1 parm2)               #  Executes: printf %q parm1 parm2
                                    #  See above IFS examples for the
                                    #+ various possibilities.
echo

$(_print $VarSomething)             # The predictable result.
echo



# Function variables
# ------------------

echo
echo '- - Function variables - -'
# A variable may represent a signed integer, a string or an array.
# A string may be used like a function name with optional arguments.

# set -vx                           #  Enable if desired
declare -f funcVar                  #+ in namespace of functions

funcVar=_print                      # Contains name of function.
$funcVar parm1                      # Same as _print at this point.
echo

funcVar=$(_print )                  # Contains result of function.
$funcVar                            # No input, No output.
$funcVar $VarSomething              # The predictable result.
echo

funcVar=$(_print $VarSomething)     #  $VarSomething replaced HERE.
$funcVar                            #  The expansion is part of the
echo                                #+ variable contents.

funcVar="$(_print $VarSomething)"   #  $VarSomething replaced HERE.
$funcVar                            #  The expansion is part of the
echo                                #+ variable contents.

#  The difference between the unquoted and the double-quoted versions
#+ above can be seen in the "protect_literal.sh" example.
#  The first case above is processed as two, unquoted, Bash-Words.
#  The second case above is processed as one, quoted, Bash-Word.




# Delayed replacement
# -------------------

echo
echo '- - Delayed replacement - -'
funcVar="$(_print '$VarSomething')" # No replacement, single Bash-Word.
eval $funcVar                       # $VarSomething replaced HERE.
echo

VarSomething='NewThing'
eval $funcVar                       # $VarSomething replaced HERE.
echo

# Restore the original setting trashed above.
VarSomething=Literal

#  There are a pair of functions demonstrated in the
#+ "protect_literal.sh" and "unprotect_literal.sh" examples.
#  These are general purpose functions for delayed replacement literals
#+ containing variables.





# REVIEW:
# ------

#  A string can be considered a Classic-Array of elements (characters).
#  A string operation applies to all elements (characters) of the string
#+ (in concept, anyway).
###
#  The notation: ${array_name[@]} represents all elements of the
#+ Bash-Array: array_name.
###