Skip to content

Latest commit

 

History

History
298 lines (222 loc) · 11.8 KB

README.md

File metadata and controls

298 lines (222 loc) · 11.8 KB
Raw
  • infos = Information about type plugin is in keys below
  • infos/author = Klemens Böswirth k.boeswirth+git@gmail.com
  • infos/licence = BSD
  • infos/provides = check
  • infos/needs =
  • infos/placements = postgetstorage presetstorage
  • infos/status = recommended maintained unittest tested nodep libc
  • infos/metadata = check/type type check/enum check/enum/# check/enum/delimiter check/boolean/true check/boolean/false
  • infos/description = type checker using COBRA data types

Introduction

This plugin is a type checker plugin using the CORBA data types.

A common and successful type system happens to be CORBA. The system is well suited because of the many well-defined mappings it provides to other programming languages.

The type checker plugin supports these types: short, unsigned_short, long, unsigned_long, long_long, unsigned_long_long, float, double, char, wchar, boolean, any, enum, string, wstring and octet.

  • Checking any will always be successful, regardless of the content.
  • string matches any non-empty key value.
  • octet and char are equivalent to each other.
  • enum will do enum checking as described below.
  • boolean only allows the values 1 and 0. See also Normalization.
  • To use wchar and wstring the function mbstowcs(3) must be able convert the key value into a wide character string. wstrings can be of any non-zero length, wchar must have exactly length 1.

Enums

If a key is set to the type enum the plugin will look for the metadata array check/enum/#.

For example:

check/enum = #3
check/enum/#0 = small
check/enum/#1 = middle
check/enum/#2 = large
check/enum/#3 = huge

Only the values listed in this array will be accepted. The array indices don't have to be continuous, using e.g. only #1, #2 and #4 is also allowed. Just make sure check/enum is set to the largest index in the array.

Furthermore check/enum/delimiter may contain a separator character, that separates multiple allowed occurrences. If check/enum/delimiter contain more than a single character validation will fail.

For example:

check/enum/delimiter = _

Then the value middle_small would validate. middle_small_small would be allowed as well, because multi-values are treated like bitfields.

Normalization

Some types support normalization in addition to the standard type checking. This means that an extended set of allowed values is normalized into a different (in most cases standardized) set before type checking. The normalized values will be passed on from kdbGet to the rest of Elektra and your application. During kdbSet changed values are also normalized before type checking and at the end of kdbSet, if everything was successful, the values are restored to the ones originally provided by the user (no matter if they were changed before kdbSet or already present in kdbGet).

The full normalization/restore procedure can be described by the following cases:

  • Case 1: The Key existed in kdbGet and is unchanged between kdbGet and kdbSet

    This is the easiest and most obvious case. The value is normalized in kdbGet and the original value is restored in kdbSet, so that the underlying storage file remains unchanged (w.r.t. the key in question).

  • Case 2: The Key didn't exist in kdbGet, i.e. it was added

    Here the value is normalized to verify the type and then restored immediately (all inside of kdbSet).

  • Case 3: The Key existed in kdbGet, but its value was changed between kdbGet and kdbSet

    This is essential the same as Case 2. keySetString removes the origvalue metadata used to store the original value, so as far as this plugin is concerned, the key didn't exist in kdbGet.

Note: If normalization is used, often times you will get a normalization error instead of a type checking error.

Booleans

The values that are accepted as booleans are configured during mounting:

sudo kdb mount typetest.dump user/tests/type dump type booleans=#1 booleans/#0/true=a booleans/#0/false=b booleans/#1/true=t booleans/#1/false=f

The above line defines that the array of allowed boolean pairs. booleans=#1 defines the last element of the array as #1. For each element # the keys booleans/#/true and booleans/#/false define the true and false value respectively. True values are normalized to 1, false values to 0.

Even though we didn't specify them the values 1 and 0 are still accepted. The normalized values are always okay to use. If no configuration is given, the allowed values default to 1, yes, on, true, enabled and enable as true values and 0, no, off, false, disabled and disable as false values.

The accepted values can also be overridden on a per-key-basis. Simply add the metakey check/boolean/true and check/boolean/false to set the accepted true and false values. Only a single true/false value can be chosen. This is intended for use cases, where normally you prefer to use only e.g. 1, 0 and true, false, but what to override that for a key where e.g. enabled and disabled make more sense contextually (e.g. for something like /log/debug). Because of this intention restoring also works differently, when check/boolean/true and check/boolean/false are used. In this case we will always restore to the chosen override values.

Note: The values 1 and 0 are accepted, even if overrides are used. This means you can set a key with overrides to 0 (or 1) and during kdbSet it will be restored to the false (or true) override value. (This is useful for the high-level API.)

It is an error to specify only one of booleans/#/true and booleans/#/false or check/boolean/true and check/boolean/false. Boolean always come in pairs!

You can also change how values shall be restored in kdbSet:

sudo kdb mount typetest.dump user/tests/type dump type booleans=#0 booleans/#0/true=true booleans/#0/false=false boolean/restoreas=#0

The config key boolean/restoreas must be a valid index of the booleans array. If boolean/restoreas was set the chosen boolean pair will be used when values are restored in kdbSet. So in the above example the plugin accepts 1, true, 0 and false as boolean values, on kdbGet it turns true into 1 and false into 0 and on kdbSet it turns 1 into true and 0 into false.

If no booleans array was given the allowed values for boolean/restore are:

  • #0 for yes/no
  • #1 for true/false
  • #2 for on/off
  • #3 for enabled/disabled
  • #4 for enable/disable

Enums

Enums also support normalization. Contrary to boolean normalization, enum normalization is always configured on a per-key-basis.

Simply set the metakey check/enum/normalize to 1 in order to normalize the string values to there indexes. Any other value is ignored.

Take for example a key with the following enum configuration:

check/enum = #3
check/enum/#0 = small
check/enum/#1 = medium
check/enum/#3 = huge

The value small will be normalized to 0, medium to 1 and huge to 3. During restore the values 0, 1 and 3 will be restored to small, medium and huge.

If you use normalization, you can pass string values or indices to kdbGet and kdbSet, but you will always get back indices from kdbGet and string values from kdbSet. (Therefore you can seamlessly use elektraGetUnsignedLongLong from high-level API for normalized enums.)

The plugin also supports normalizing enums that use check/enum/delimiter, however be careful which indexes you use in this case. The indexes of all values are simple bitwise or-ed (using |). In the above example small_medium would be normalized to 1 (0 | 1 == 1), the same value as medium. This means during restore the value emitted will be medium.

A version that would work with delimiter and normalization is:

check/enum = #4
check/enum/#0 = none
check/enum/#1 = small
check/enum/#2 = medium
check/enum/#4 = huge

Here small_medium is normalized to 3, which is a unique value. During restore with delimiters the values might not be restored to there original form, but may be restored to an equivalent representation. e.g. small_none may be restored to just small or small_medium may be restored to medium_small This has technical reasons and we do not guarantee any restriction on what representation is produced during restore, other than the normalized value being the same as for the user provided representation.

IMPORTANT: Do not use normalization together with enums, whose string values start with digits (e.g. check/enum/#0 = 1abc). This breaks normalization! Indices are differentiated from string value by whether they start with a digit.

Example

#Mount the plugin
sudo kdb mount typetest.dump user/tests/type dump type

#Store a character value
kdb set user/tests/type/key a

#Only allow character values
kdb setmeta user/tests/type/key type char
kdb get user/tests/type/key
#> a

#If we store another character everything works fine
kdb set user/tests/type/key b
kdb get user/tests/type/key
#> b

#If we try to store a string Elektra will not change the value
kdb set user/tests/type/key 'Not a char'
# RET:5
# ERROR:52
kdb get user/tests/type/key
#> b

#Undo modifications to the database
kdb rm user/tests/type/key
sudo kdb umount user/tests/type

For enums:

# Backup-and-Restore:/tests/enum

sudo kdb mount typeenum.ecf user/tests/type dump type

# valid initial value + setup valid enum list
kdb set user/tests/type/value middle
kdb setmeta user/tests/type/value check/enum '#2'
kdb setmeta user/tests/type/value 'check/enum/#0' 'low'
kdb setmeta user/tests/type/value 'check/enum/#1' 'middle'
kdb setmeta user/tests/type/value 'check/enum/#2' 'high'
kdb setmeta user/tests/type/value type enum

# should succeed
kdb set user/tests/type/value low

# should fail with error 52
kdb set user/tests/type/value no
# RET:5
# ERROR:52

Or with multi-enums:

# valid initial value + setup array with valid enums
kdb set user/tests/type/multivalue middle_small
kdb setmeta user/tests/type/multivalue check/enum/#0 small
kdb setmeta user/tests/type/multivalue check/enum/#1 middle
kdb setmeta user/tests/type/multivalue check/enum/#2 large
kdb setmeta user/tests/type/multivalue check/enum/#3 huge
kdb setmeta user/tests/type/multivalue check/enum/delimiter _
kdb setmeta user/tests/type/multivalue check/enum "#3"
kdb setmeta user/tests/type/multivalue type enum

# should succeed
kdb set user/tests/type/multivalue small_middle

# should fail with error 52
kdb set user/tests/type/multivalue all_small
# RET:5
# ERROR:52

# cleanup
kdb rm -r user/tests/type
sudo kdb umount user/tests/type

For booleans:

# Mount plugin
sudo kdb mount config.ecf user/tests/type dump type

# By default the plugin uses `1` (true) and `0` (false) to represent boolean values
kdb set user/tests/type/truthiness false
kdb setmeta user/tests/type/truthiness type boolean
kdb get user/tests/type/truthiness
#> 0

# The plugin does not change ordinary values
kdb set user/tests/type/key value
kdb get user/tests/type/key
#> value

# Undo changes
kdb rm -r user/tests/type
sudo kdb umount user/tests/type
sudo kdb mount config.ecf user/tests/type dump type
kdb set user/tests/type/truthiness 0
kdb setmeta user/tests/type/truthiness type boolean
 kdb set user/tests/type/truthiness yes
# RET: 0
 # Undo changes
kdb rm -r user/tests/type
sudo kdb umount user/tests/type

Limitations

Records are part of other plugins.

The CORBA type system also has its limits. The types string and enum can be unsatisfactory. While string is too general and makes no limit on how the sequence of characters is structured, the enumeration is too finite. For example, it is not possible to say that a string is not allowed to have a specific symbol in it. Combine this plugin with other type checker plugins to circumvent such limitations.