Конфигурация в скриптах

OpenWrt предоставляет набор стандартных процедур для взаимодействия с UCI для чтения и обработки конфигурационных файлов из скриптов оболочки. Они особенно полезны для написания системных скриптов для /etc/init.d/.

Для работы с UCI configuration files, вам нужно включить в ваш скрипт эту строку, для подключения библиотеки:

. /lib/functions.sh

После этого вы сможете использовать функцию config_load name для загрузки config-файлов.

Эта функция загружает файл с именем name из папки /etc/config/ (в которой обычно лежат конфиги).

Первый способ парсинга UCI config файлов- через т.н. callbacks, которые вызываются для каждой секции (section), опции (option) и списка (list) встреченных при парсинге.

Можно определить три callbacks: config_cb(), option_cb(), и list_cb(). Нужно определить эти функции в скрипте до вызова config_load, но после строки /lib/functions.sh.

Также можно использовать reset_cb() чтобы сбросить все три callbacks в неопределенные функции.

Процедура config_cb вызывается каждый раз, когда при парсинге попадается заголовок UCI секции. Кроме того, config_cb вызывается (без аргументов) каждый раз после выполнения config_load. Это полезно, когда вы формируете значения переменных в процессе парсинга (см. ниже) и хотите, например, записать их в файл по окончании парсинга секции.

При вызове процедуре передается два аргумента:

  1. Type, тип секции, Пр: interface или config interface wan
  2. Name, название секции, Пр: wan для config interface wan или автогенерируемое ID, типаcfg13abef для безымянных секций типа config route
config_cb() { local type=“$1” local name=“$2” # эта команда будет выполнена для каждой секции }

Также дополнительный вызов config_cb (без новой секции) произойдет после окончания работы config_load. Это позволяет обрабатывать секции (sections ) как до, так и после того как все опции (options ) обработаны.

Аналогично config_cb, option_cb вызывается каждый раз когда при парсинге встречается UCI option.

При вызове этой процедуре передается два аргумента:

  1. Name, имя опции, пр: ifname для option ifname eth0
  2. Значение, значение опцее, пр: eth0 для option ifname eth0
option_cb() { local name=“$1” local value=“$2” # команда выполняется для каждой опции }

Внутри option_cb(), ID текущей секции доступно через переменную $CONFIG_SECTION.

Можно определить option_cb внутри config_cb, что позволит определить разные option_cb callbacks для разных типов секций. Это позволяет обрабатывать разные config секции индивидуально.

list_cb works exactly like option_cb above, but gets called each time a list item is encountered.

When called, the procedure receives two arguments:

  1. Name, the list name, e.g. server for list server 192.168.42.1
  2. Value, the list value, e.g. 192.168.42.1 for list server 192.168.42.1
list_cb() { local name=“$1” local value=“$2” # commands to be run for every list item }

Each item of a given list generates a new call to list_cb.

An alternative approach to callback based parsing is iterating the configuration sections with the config_foreach procedure.

The config_foreach procedure takes at least one argument:

  1. Function, name of a previously defined procedure called for each encountered section
  2. Type (optional), only iterate sections of the given type, skip others
  3. Additional arguments (optional), all following arguments are passed to the callback procedure as-is

In the example below, the handle_interface procedure is called for each config interface section in /etc/config/network. The string test is passed as second argument on each invocation.

handle_interface() { local config=“$1” local custom=“$2” # run commands for every interface section } config_load network config_foreach handle_interface interface test

It is possible to abort the iteration from within the callback by returning a non-zero value (return 1).

Within the per-section callback, the config_get or config_set procedures may be used to read or set values belonging to the currently processed section.

The config_get procedure takes at least three arguments:

  1. Name of a variable to store the retrieved value in
  2. ID of the section to read the value from
  3. Name of the option to read the value from
  4. Default (optional), value to return instead if option is unset
... # read the value of “option ifname” into the “iface” variable # $config contains the ID of the current section local iface config_get iface “$config” ifname echo “Interface name is $iface” ...

The config_set procedure takes three arguments:

  1. ID of the section to set the option in
  2. Name of the option to assign the value to
  3. Value to assign
... # set the value of “option auto” to “0” # $config contains the ID of the current section config_set “$config” auto 0 ...

Note that values changed with config_set are only kept in memory. Subsequent calls to config_get will return the updated values but the underlying configuration files are not altered. If you want to alter values, use the uci_* functions from /lib/config/uci.sh which are automatically included by /etc/functions.sh.

If the name of a configuration section is known in advance (it is named), options can be read directly without using a section iterator callback.

The example below reads “option proto” from the “config interface wan” section.

... local proto config_get proto wan proto echo “Current WAN protocol is $proto” ...

Some UCI configurations may contain list options in the form:

... list network lan list network wifi ...

Calling config_get on the network list will return the list values separated by space, lan wifi in this example.

However, this behaviour might break values if the list items itself contain spaces like illustrated below:

... list animal 'White Elephant' list animal 'Mighty Unicorn' ...

The config_get approach would return the values in the form White Elephant Mighty Unicorn and the original list items are not clearly separated anymore.

To circumvent this problem, the config_list_foreach iterator can be used. It works similar to config_foreach but operates on list values instead of config sections.

The config_list_foreach procedure takes at least three arguments:

  1. ID of the section to read the list from
  2. Name of the list option to read items from
  3. Procedure to call for each list item
  4. Additional arguments (optional), all following arguments are passed to the callback procedure as-is
# handle list items in a callback # $config contains the ID of the section handle_animal() { local value=“$1” # do something with $value } config_list_foreach “$config” animal handle_animal

Note: for editing value, use the uci command line tool.

Boolean options may contain various values to signify a true value like on, true, enabled or 1.

The config_get_bool procedure simplifies the process of reading a boolean option and casting it to a plain integer value (1 for true and 0 for false).

At least three arguments are expected by the procedure:

  1. Name of a variable to store the retrieved value in
  2. ID of the section to read the value from
  3. Name of the option to read the value from
  4. Default (optional), boolean value to return if option is unset

It depends on what you want to parse.

If, for a given section, all options should be treated in the same way (say, write them to a config file), whatever their order or name, then callbacks are a good option. It allows you to parse all options without having to know their name in advance.

For instance, consider this script:

config_cb() {
    local type="$1"
    local name="$2"
    if [ "$type" = "mysection" ]
    then
        option_cb() {                                            
            local option="$1"                                
            local value="$2"                                 
            echo "${option//_/-} $value" >> /var/etc/myfile.conf
        }
    else {
        option_cb() { return; }
    }
}

This would parse a configuration like this one:

config mysection foo option link_quality auto option rxcost 256 option hello_inverval 4

And generate a file containing:

link-quality auto rxcost 256 hello-interval 4

On the other hand, if different options play a very different role, then it may be more convenient to pick each option explicitly.

For instance, babeld has the following syntax for filters:

{in|out|redistribute} selector* [allow|deny|metric n]

where a selector can be one of:

local
ip prefix
proto p
if interface
...

At most one instance of each selector type is allowed in a filter. The UCI configuration looks like this:

config filter
    option ip '172.22.0.0/15'
    option local 'true'
    option type 'redistribute'
    option action 'metric 128'

The result of this section should be:

redistribute ip 172.22.0.0/15 local metric 128

That is, the type should come first, then selectors (with a special case for local, which doesn't take an argument), and finally the action. The basic parsing method is the following:

parse_filter() {
    local section="$1"
    local _buffer
    local _value

    config_get _value "$section" "type"
    append _buffer "$_value"

    config_get _value "$section" "ip"
    append _buffer "ip $_value"
    config_get _value "$section" "proto"
    append _buffer "proto $_value"
    ...

    config_get_bool _value "$section" "local" 0
    [ "$_value" -eq 1 ] && append _buffer "local"

    config_get _value "$section" "action"
    append _buffer "$_value"

    echo "$_buffer" >> /var/etc/babeld.conf
}
config_load babeld
config_foreach parse_filter filter

Of course, it is possible to mix the two styles of parsing, for instance parsing one section with callbacks and another section explicitly. Don't forget to reset or unset the callbacks (reset_cb or option_cb() { return; }) for sections you don't want to parse with callbacks.

This website uses cookies. By using the website, you agree with storing cookies on your computer. Also you acknowledge that you have read and understand our Privacy Policy. If you do not agree leave the website.More information about cookies
  • Last modified: 2021/07/24 05:37
  • by someothertime