Plugin: yanlr

• infos = Information about the yanlr plugin is in keys below
• infos/author = René Schwaiger sanss.nosp@m.ecou.nosp@m.rs@me.nosp@m..com
• infos/licence = BSD
• infos/needs = directoryvalue yamlsmith
• infos/provides = storage/yaml
• infos/recommends =
• infos/placements = getstorage
• infos/status = maintained unittest preview experimental unfinished nodoc concept discouraged
• infos/metadata =
• infos/description = This storage plugin use a parser generated by ANTLR to read YAML files

Yan LR

Introduction

This plugin uses ANTLR to generate a parser for the YAML serialization format. Apart from the hand written lexer, all other parts of the plugin use standard ANTLR tools, such as

• an ANTLR parser grammar,
• a listener, that reacts to matches in the grammar building a key set, and
• an error listener that reacts to syntax errors

Dependencies

The plugin requires

• ANTLR 4.7.1 or later (antlr4), and
• ANTLR 4’s C++ runtime (antlr4-cpp-runtime or libantlr4-runtime-dev)

. If packages for those libraries are not available for your system, you can install them manually. For more information about that please take a look at ANTLR’s homepage and at the ReadMe of the ANTLR C++ runtime.

Examples

Mappings

`` <h1>Mount plugin to namespaceuser/tests/yanlr` sudo kdb mount config.yaml user/tests/yanlr yanlr

Manually add some mappings to the configuration file

printf 'all : circles presuppose
' > kdb file user/tests/yanlr printf 'hello : world
' >> kdb file user/tests/yanlr

kdb ls /tests/yanlr #> user/tests/yanlr/all #> user/tests/yanlr/hello

kdb get user/tests/yanlr/all #> circles presuppose

Store value at root of mountpoint

kdb set user/tests/yanlr 'Mount Eerie' kdb get user/tests/yanlr #> Mount Eerie

Add new key-value pairs

Yan LR actually uses the YAML Smith plugin to write data

kdb set user/tests/yanlr/brand new kdb set user/tests/yanlr/brand/new eyes kdb set user/tests/yanlr/dance/gavin 'Dance!'

kdb ls /tests/yanlr #> user/tests/yanlr #> user/tests/yanlr/all #> user/tests/yanlr/brand #> user/tests/yanlr/brand/new #> user/tests/yanlr/dance/gavin #> user/tests/yanlr/hello

kdb get /tests/yanlr/hello #> world kdb get user/tests/yanlr/brand #> new kdb get /tests/yanlr/dance/gavin #> Dance!

Undo modifications to the key database

kdb rm -r user/tests/yanlr sudo kdb umount user/tests/yanlr

### Arrays

Mount plugin to cascading namespace /tests/yanlr

sudo kdb mount config.yaml user/tests/yanlr yanlr

Manually add a sequences to the configuration file

printf 'primes:
' > kdb file user/tests/yanlr printf ' - two
' >> kdb file user/tests/yanlr printf ' - three
' >> kdb file user/tests/yanlr printf ' - five
' >> kdb file user/tests/yanlr

kdb ls user/tests/yanlr #> user/tests/yanlr/primes #> user/tests/yanlr/primes/#0 #> user/tests/yanlr/primes/#1 #> user/tests/yanlr/primes/#2

kdb set user/tests/yanlr/primes/#3 seven

Retrieve index of last array element

kdb getmeta user/tests/yanlr/primes array #> #3

Undo modifications to the key database

kdb rm -r user/tests/yanlr sudo kdb umount user/tests/yanlr

### Error Messages

Mount plugin

sudo kdb mount config.yaml user/tests/yanlr yanlr

Manually add syntactically incorrect data

printf – 'key: - element 1
' > kdb file user/tests/yanlr printf – '- element 2 # Incorrect Indentation!' >> kdb file user/tests/yanlr

The plugin reports the location of the error

kdb ls user/tests/yanlr

RET: 5

STDERR: .*/config.yaml:2:1: mismatched input '- ' expecting end of map.*

Let us look at the error message more closely.

Since the location of config.yaml depends on the current user and OS,

we store the text before config.yaml as user/tests/error/prefix.

kdb set user/tests/error "$(2>&1 kdb ls user/tests/yanlr)" kdb set user/tests/error/prefix "$(kdb get user/tests/error | grep 'config.yaml' | head -1 | sed -E 's/(.*)config.yaml.*/\1/')" kdb get user/tests/error/prefix

We also store the length of the prefix, so we can remove it from every

line of the error message.

kdb set user/tests/error/prefix/length "$(kdb get user/tests/error/prefix | wc -c | sed 's/[ ]*//g')"

Since we only want to look at the “reason” of the error, we

remove the other part of the error message with head and tail.

kdb get user/tests/error | tail -n11 | head -n6 | cut -c"$(kdb get user/tests/error/prefix/length | tr -d '\n')"- #> config.yaml:2:1: mismatched input '- ' expecting end of map #> - element 2 # Incorrect Indentation! #> ^^ #> config.yaml:2:37: extraneous input 'end of map' expecting end of document #> - element 2 # Incorrect Indentation! #> ^

Fix syntax error

printf – 'key: - element 1
' > kdb file user/tests/yanlr printf – ' - element 2 # Fixed!' >> kdb file user/tests/yanlr

kdb ls user/tests/yanlr #> user/tests/yanlr/key #> user/tests/yanlr/key/#0 #> user/tests/yanlr/key/#1

Undo modifications

kdb rm -r user/tests/error kdb rm -r user/tests/yanlr sudo kdb umount user/tests/yanlr

## Limitations

- The plugin does **not support**

  - plain scalars that span **multiple lines**,
  - **special characters** inside double quoted scalars,
  - other **line endings** than `\n` (Unix line endings),
  - **block scalars**,
  - **flow collections**,
  - **tags**,
  - **anchors & aliases**,
  - **multiple documents**, and
  - document **start and end markers**

  .

- Yan LR does not provide write support for data. Please use the [YAML Smith](@ref src_plugins_yamlsmith_README_md) plugin for that purpose.

### Comments

The [lexer](/home/markus/Projekte/Elektra/current/src/plugins/yanlr/yaml_lexer.cpp) does currently tokenize comments. Consequently the [plugin grammar](@ref src_plugins_yanlr_YAML_g4) of the plugin does also match comments. However, the [listener](/home/markus/Projekte/Elektra/current/src/plugins/yanlr/listener.cpp) does currently **ignore comments**.

### Indentation

The lexer does not check for incorrect indentation. Consequently the following YAML data:



```yaml
        value

will produce a plain scalar containing a tab character followed by the text value. The correct behavior would be to report an error, since YAML does not allow tab characters in indentation.

Error Messages

Visualized error messages (containing ^ markers) might not point to the correct error location, if the input contains tabs or unicode characters with a length other than 1.