Projects/bash-ini-parser
1
0
Fork 0
Code Releases Wiki Activity

Almost complete - before cleanup work.

Browse source
This commit is contained in:
Darren 'Tadgy' Austin 2019-07-19 01:55:34 +01:00
commit e59eb107e3
1 changed files with 162 additions and 159 deletions
Download patch file Download diff file Expand all files Collapse all files

321
parse_ini
View file

@ -1,45 +1,5 @@
#!/bin/bash
# http://en.wikipedia.org/wiki/INI_file:
# * Provides a good explanation of the ini format - use this for docs *
# * INI's have 'sections' and 'properties'. Properties have key = value format *
#
# Case insensitivity: Case is not changed, unless option used to covert to lower/upper case.
# Comments: Allow ; and # for comments. Must be on their own line.
# Blank lines: Blank lines are ignored.
# Escape chars: \ at the end of a line will continue it onto next (leading whitespace is removed per normal)
# Ordering: GLOBAL section must be at the top, sections continue until next section or EOF.
# Duplicate names: Duplicate property values overwrite previous values.
# Provide an option to abort/error is duplicate is found?
# Add option to merge duplicates separated by octal byte (\036 ??)
# Duplicate sections are merged. Option to error if dup.
# Global properties: Support. Add to a GLOBAL section?
# Hierarchy: No hierarchy support. Each section is own section.
# Name/value delim: Use = by default. Allow : via option?
# Quoted values: Allow values to be within " and ' to keep literal formatting.
# Whitespace: Whitespace around section labels and []s is removed.
# Whitespace within section labels is kept / translated.
# Whitespace around property names is removed.
# Whitespace within property names is kept as is (spaces squashed - option to override).
# Property values have whitespace between = and data removed.
# Property values are kept as is (no squashing)
# http://www.regular-expressions.info/posixbrackets.html
# http://ajdiaz.wordpress.com/2008/02/09/bash-ini-parser/
# https://github.com/rudimeier/bash_ini_parser/blob/ff9d46a5503bf41b3344af85447e28cbaf95350e/read_ini.sh
# http://tldp.org/LDP/abs/html/
# Specs:
# [section] Can be upper/lower/mixed case (set by options)
# Can only include: '-+_. [:alnum:]'
# # Any single or consecutive occurance of '-+_. ' are converted to a *single* _
# # eg: [foo -+_. bar] becomes [foo_bar] ??
# Any leading/trailing spaces/tabs between the []s and name will be removed.
# Notes:
# * To make env vars available to subsequent programs, use -x|--export.
parser_getopts() {
local DELIM_SET=0
@ -76,9 +36,44 @@ parser_getopts() {
fi
shift
;;
-h|-help|--help)
-e|-export|--export)
shift
DECLARE_SCOPE="-x"
;;
-global-name|--global-name)
shift
if [[ -z "$1" ]]; then
echo "${0##*/}: global name (--global-name) cannot be an empty value" >&2
return 1
elif [[ "${1:0:1}" =~ [[:digit:]] ]]; then
echo "${0##*/}: global name (--global-name) cannot begin with a number" >&2
return 1
elif [[ "$1" =~ [^[:alnum:]_] ]]; then
echo "${0##*/}: only alphanumerics and _ allowed for global name (--global-name)" >&2
else
CURRENT_SECTION="$1"
fi
shift
;;
-h|-\?|-help|--help)
parser_help
return 0
return 2
;;
-l|-local|--local)
shift
DECLARE_SCOPE="-l"
;;
-lowercase|--lowercase)
shift
CONVERT_CASE="-1"
;;
-no-booleans|--no-booleans)
shift
USE_BOOLEANS="0"
;;
-no-squash|--no-squash)
shift
SQUASH_SPACES=0
;;
-p|-prefix|--prefix)
shift
@ -96,16 +91,24 @@ parser_getopts() {
echo "${0##*/}: prefix (-p) cannot begin with a number" >&2
return 1
elif [[ "$1" =~ [^[:alnum:]_] ]]; then
echo "${0##*/}: invalid characters in prefix (-p) - alphanumerics and _ only" >&2
echo "${0##*/}: only alphanumerics and _ allowed for prefix (-p)" >&2
return 1
else
VARIABLE_PREFIX="$1"
fi
shift
;;
-textual-booleans|--textual-booleans)
shift
TEXTUAL_BOOLEANS="1"
;;
-uppercase|--uppercase)
shift
CONVERT_CASE="1"
;;
-v|-version|--version)
parser_version
return 0
return 2
;;
--)
# Stop option processing.
@ -137,93 +140,64 @@ parser_help() {
Usage: ${0##*/} [options] <INI file>
Parse an INI-style file into array assignments which can be 'eval'ed into Bash.
Options:
Commonly used options:
-b <char>, --bound <char>
The bound character which delimits the key from the value in a property
line of the INI file. The default is "=". This must be a single
character and cannot be empty value.
-d <char(s)>, --delim <char(s)>
The character(s) to use as a delimiter between the prefix and section name
when defining the arrays. The default is "_", except where prefix is set
to an empty value, in which case the default is also empty. Only
alphanumerics and _ may be used with this option, and it may not begin
with a number if prefix is empty. To use no delimintaor, use '-d ""'.
The character(s) (which may be an empty value) to use as a delimiter
between the prefix and section name when defining the arrays. The
default is "_", except where prefix is set to an empty value, in which
case the default is also empty. Only alphanumerics and _ may be used with
this option, and it may not begin with a number if prefix is empty. The
delimiter may be converted to upper or lower case depending upon the use
of '--uppercase' or '--lowercase'.
-e, --export
When declaring the arrays, export them to the environment.
-h, -?, --help
Show (this) help.
-l, --local
Declare the arrays as being local in scope, instead of the default of
global scope.
-p <prefix>, --prefix <prefix>
The prefix of all the variables set when defining the arrays. The default
is "INI". An empty prefix (denoted by "") implies '-d ""', but this can
be overridden by explicitly specifying a delimiter with '-d'. Only
alphanumerics and _ may be used with this option, and it may not be empty
when delim ('-d) begins with a number.
when delim ('-d') begins with a number.
-v, --version
Show version and copyright information.
-i, --implied-boolean
Options usually require a value (after the =) in order to be set.
With this option, any key without a value contained in the ini file
if assumed to be a boolean 'true' and set accordingly. Likewise, any key
preceeded with 'no_' (eg: no_foo) will set the value of 'foo' to boolean 'false'.
Cannot be used with --no-boolean.
-c, --case-sensitive
Be case sensitive with section names and properties.
Section names and property names will be used as is - no translation.
-d, --delim
The delimiter between the key and value. Must be a single character. Default =
-g, --global-name <name>
INI files can contain an optional implied "global" section - where there
are property names/values before any [section] header. This option
specified what section name the implied "global" section should be given
in the environment variables which are set. The default is 'global'.
-l, --lowercase
Usually, environment variables are converted to all uppercase before being set.
This option forces all environment variables to be converted to lowercase instead.
Note: This only effects the environment variable set with -e, and the section names
read from the ini file. Options are ??????????????????????????????????????
-x, --export
Export environment variables.
--no-boolean
Don't parse 'yes', 'true', 'on', 'no', 'false', 'off' into the corresponding boolean
values, and set the options strictly as is. Incompatible with -i.
--no-squash
Do not squash multiple consecutive occurances of punctuation characters
into a single _ when parsing section names and options. With this option
'foo.-_bar' would become 'foo___bar' rather than 'foo_bar'.
--no-duplicates
If a duplicate section name or option name is found, report error and stop.
Usually sections with the same name will have their options merged, and
duplicate option values will overwrite previous ones.
# -c, --check-only Only validate the ini file, don't parse it into the environment
--check
Check/validate the INI file by running it through the parser. Testing the
ini file will report any problems or syntax errors in the file, but will
not set up the environment variables as would happen in normal parsing.
Any parse errors are reported to stderr. When combined with the --debug
option, every detail of the parsing process is reported to stderr.
--debug
Show full details of the ini file parsing process. Detail is written to
stderr. Unless --test is used with this option, the parser will still
set up the environment as would happen normally,
-h, --help
Show (this) help.
-v, --version
Show version and copyright information.
# -b, --booleans Allow 'yes', 'true', 'on', 'no', 'false', 'off' to be used as values
# and interpited as boolean values. 'yes', 'true', 'on' set option value to "1".
# 'no', 'false', 'off' set option value to "0".
# -?, --???? Interprite the presense of an option name without any value as a boolean
# 'true', and no_<option> as a boolean 'false', setting the option value
# to 1 or 0 accordingly. eg: 'foo' in the ini file would set option foo = 1
# and 'no_foo' would set foo = 0.
# ???? Implies -b ????
# --check Parse the file, report any problems, but don't output the code.
# --debug Show all details of the parsing process to stderr. If --check is used, no code is outputted.
# --?? Declare the arrays as exported.
# --?? Declare the arrays as global (the default).
# --?? Declare the arrays as local.
# --?? Set comment characters. Each char can be used to indicate a comment.
# --?? Treat all problems as errors and stop processing at that point. Need to integrate into code.
# --?? Change bools from 0/1 to false/true???
Lesser used options:
--global-name <name>
The name of the 'global' section used when defining the arrays. Only
alphanumerics and _ may be used with this option, which cannot be empty.
The name may not begin with a number, and may be converted to upper or
lower case depending upon the use of '--uppercase' or '--lowercase'.
--lowercase
When defining the arrays, the case of the prefix ('-p') name, delimiter
and section name is kept as set. With this option all items are converted
to lower case. The case of the propertie's keys/values is not affected.
--no-booleans
Normally, the parser interprites the presence of a key without an
associated value as a boolean. Keys which are proceeded by "no_" are
given a boolean 'false' value, while keys without a "no_" are given a
'true' value. With this option, the presence of a key without a value is
considered a syntax error in the INI file.
--no-squash
Do not squash multiple consecutive blanks (which are later translated to
a _) into a single space while reading section names and properties.
--textual-booleans
When defining the arrays, boolean keys are given a value of "0" or "1"
(representing 'false' and 'true' respectivly). With this option the value
of the key will be the text "false" or "true" instead. Ignored when
'--no-booleans' is in use.
--uppercase
When defining the arrays, the case of the prefix ('-p') name, delimiter
and section name is kept as set. With this option all items are converted
to upper case. The case of the propertie's keys/values is not affected.
Option processing ceases with the first non-option argument, or "--".
EOF
}
@ -243,7 +217,8 @@ parser_version() {
parse_ini() {
# Bash v4.1+ is required.
# Bash v4.4+ is required.
# if [[ -z "${BASH_VERSINFO[0]}" ]] || ((BASH_VERSINFO[0] < 4)); then
if [[ -z "${BASH_VERSINFO[0]}" ]] || ((BASH_VERSINFO[0] < 4)); then
echo "${0##*/}: minimum of bash v4 required" >&2
return 1
@ -255,17 +230,28 @@ parse_ini() {
local CONVERT_CASE="0" # Whether to keep or convert section and key names to upper or loweer case. -1 = covert to lowercase, 0 = keep case, 1 = convert to uppercase.
local CONVERT_CHARS="[:blank:].+-" # Characters from ACCEPTABLE_CHARS in section and key names that should be converted to _. Must be a valid regex bracket expression.
local CURRENT_SECTION="global" # Name used for the 'global' section of the INI file.
local KEYVALUE_DELIM="=" # Delimintator between key and value. Must be a single character.
local DECLARE_SCOPE="-g" # The scope given in the array definitions. "-g" = global scope, "-l" = local scope, "-x" = export values.
local KEYVALUE_DELIM="=" # Delimiter between key and value. Must be a single character.
local SQUASH_SPACES="1" # Whether to squash multiple consecutive blanks into a single space. 0 = don't squash, 1 = do squash.
local TEXTUAL_BOOLEANS="0" # Whether to use "false" and "true" for booleans. 0 = use "0" and "1", 1 = use "false" and "true".
local USE_BOOLEANS="1" # Whether to allow the use of boolean values in the INI file. 0 = don't allow, 1 = do allow.
local VARIABLE_PREFIX="INI" # Prefix for all variables. Note: case is not changed, even with CONVERT_CASE set.
local VARIABLE_DELIM="_" # Delimiter between prefix and section name, unless VARIABLE_PREFIX is empty.
# Variables.
local BOOL_VALUE CURRENT_SECTION IGNORE_SECTION=0 INIFD KEY LINE LINENUMBER=0 REPLY VALUE
local BOOL_VALUE DELIM ERR IGNORE_SECTION=0 INIFD KEY LINE LINENUMBER=0 PREFIX REPLY VALUE
declare INIFILE
# Parse options.
parser_getopts "$@" || return $?
parser_getopts "$@"
ERR=$?
if ((ERR == 1)); then
# And error occured.
return 1
elif ((ERR == 2)); then
# Help/version was showed, exit sucessfully.
return 0
fi
# If reading from stdin, don't try to open the FD as it's already open.
if [[ "$INIFILE" == "-" ]]; then
@ -293,11 +279,26 @@ parse_ini() {
# Extglob is required.
shopt -s extglob
# Convert case, if required.
if ((CONVERT_CASE == -1)); then
# Covert to lowercase.
PREFIX="${VARIABLE_PREFIX,,}"
DELIM="${VARIABLE_DELIM,,}"
CURRENT_SECTION="${CURRENT_SECTION,,}"
elif ((CONVERT_CASE == 1)); then
# Convert to uppercase.
PREFIX="${VARIABLE_PREFIX^^}"
DELIM="${VARIABLE_DELIM^^}"
CURRENT_SECTION="${CURRENT_SECTION^^}"
else
# Don't convert.
PREFIX="$VARIABLE_PREFIX"
DELIM="$VARIABLE_DELIM"
fi
# Output the 'global' section definition.
# FIXME: If doing validation only, don't output declaration here.
# FIXME: Need to handle export and local options here.
# FIXME: Handle an empty prefix here, and delim as per key/value printf.
printf "declare -g -A %s%s%s\\n" "$VARIABLE_PREFIX" "$VARIABLE_DELIM" "$CURRENT_SECTION"
printf "declare %s -A %s%s%s\\n" "$DECLARE_SCOPE" "$PREFIX" "$DELIM" "$CURRENT_SECTION"
# Parse the INI file.
while :; do
@ -320,6 +321,7 @@ parse_ini() {
done
# Ignore the line if it's a comment.
# FIXME: Is the printf required here?
[[ "$LINE" =~ ^[[:blank:]]*([$(printf "%q" "$COMMENT_CHARS")].*)*$ ]] && continue
# Process the line.
@ -363,8 +365,7 @@ parse_ini() {
# Output the associative array declaration.
# FIXME: If doing validation only, don't output declaration here.
# FIXME: Need to handle export and local options here.
printf "declare -g -A %s%s%s\\n" "$VARIABLE_PREFIX" "$VARIABLE_DELIM" "$LINE"
printf "declare %s -A %s%s%s\\n" "$DECLARE_SCOPE" "$PREFIX" "$DELIM" "$LINE"
# Keep track of the current section name.
CURRENT_SECTION="$LINE"
@ -387,13 +388,27 @@ parse_ini() {
# Determine the boolean value.
if [[ "${LINE:0:3}" == "no_" ]]; then
LINE="${LINE:3:${#LINE} - 1}"
BOOL_VALUE=0
if ((TEXTUAL_BOOLEANS == 0)); then
BOOL_VALUE=0
else
BOOL_VALUE="false"
fi
else
BOOL_VALUE=1
if ((TEXTUAL_BOOLEANS == 0)); then
BOOL_VALUE=1
else
BOOL_VALUE="true"
fi
fi
# Output the associative array element definition.
printf "%s%s%s+=([\"%s\"]=\"%s\")\\n" "$VARIABLE_PREFIX" "${VARIABLE_PREFIX:+$VARIABLE_DELIM}" "$CURRENT_SECTION" "$LINE" "$BOOL_VALUE"
if ((USE_BOOLEANS == 1)); then
# printf "%s%s%s+=([\"%s\"]=\"%s\")\\n" "$PREFIX" "${PREFIX:+$DELIM}" "$CURRENT_SECTION" "$LINE" "$BOOL_VALUE"
printf "%s%s%s[\"%s\"]=\"%s\"\\n" "$PREFIX" "${PREFIX:+$DELIM}" "$CURRENT_SECTION" "$LINE" "$BOOL_VALUE"
else
echo "${0##*/}: line $LINENUMBER: key without a value - ignoring property"
continue
fi
elif ((IGNORE_SECTION == 0)); then # Process the property definition as a key/value pair.
# Remove trailing whitespace from key part.
LINE="${LINE/+([[:blank:]])$KEYVALUE_DELIM/$KEYVALUE_DELIM}"
@ -405,6 +420,9 @@ parse_ini() {
KEY="${LINE%%$KEYVALUE_DELIM*}"
VALUE="${LINE#*$KEYVALUE_DELIM}"
# Squash multiple consecutive blanks into a single space.
((SQUASH_SPACES == 1)) && KEY="${KEY//+([[:blank:]])/ }"
# If the value starts with a " or ' it must end with same.
if [[ "${VALUE:0:1}" =~ [\"\'] ]]; then
if [[ "${VALUE:0:1}" == "${VALUE: -1:1}" ]]; then
@ -418,42 +436,27 @@ parse_ini() {
# Output the associative array element definition.
# FIXME: If doing validation only, don't output declaration here.
# FIXME: Need to handle export and local options here.
# FIXME: Need to make sure multiple keys with the same name add to the element, not replace it.
# FIXME: Have an option to have repeat sections/properties over-write previous ones rather than append.
# FIXME: Need to handle bash <4.4 (by CLI option): declare -n foo="$prefix$delim$section"; $prefix$delim$section["$key"]="${foo["$key"]}$value"
# FIXME: Need to handle bash <4.4 (by CLI option?): declare -n foo="$prefix$delim$section"; $prefix$delim$section["$key"]="${foo["$key"]}$value"
# For bash 4.4+.
printf "%s%s%s+=([\"%s\"]+=\"%s\")\\n" "$VARIABLE_PREFIX" "${VARIABLE_PREFIX:+$VARIABLE_DELIM}" "$CURRENT_SECTION" "$KEY" "$VALUE"
# printf "%s%s%s+=([\"%s\"]+=\"%s\")\\n" "$PREFIX" "${PREFIX:+$DELIM}" "$CURRENT_SECTION" "$KEY" "$VALUE"
# For bash 4.0+
printf "%s%s%s[\"%s\"]+=\"%s\"\\n" "$PREFIX" "${PREFIX:+$DELIM}" "$CURRENT_SECTION" "$KEY" "$VALUE"
else
# FIXME: Make this debug output only.
echo "Skipping line $LINENUMBER"
true
fi
# printf "<%s = %s>\\n" "$KEY" "$VALUE"
# if first non-whitespace char after the first = is " or ', check the last non-whitespace char on the line.
# if that character is a matching " or ', skip to normal processing.
# if that character doesn't match the opening " or ', go to continued line processing
# else (no opening " or ') check the last non-whitespace char on the line; if its a \ (line continuation marker)
# go to continued line processing
# fi
# Continued line processing
# Notes: If within a " or ' block, keep whitespace as entered - don't strip from begining of line.
# If here from a continueation marker, remove leading whitespace.
# Will need a flag to show if we're looking for an ending " or '
# Normal processing:
# Escape chars?
# Close file descriptor for ini file
# clean up the environment
# IFS=$INI_SAVED_IFS
# Remove any variables begining INI_
done
# Close file descriptor for INI file.
if ((INIFD != 1)); then
exec {INIFD}<&-
fi
# Clean up the environment.
unset INIFILE
}
parse_ini "$@"