Error Parsing Custom Type Declaration

[Philipp1297]

Description

I encountered an issue when parsing a type declaration in a Structured Text (ST) file. The declaration is as follows:

TYPE T_INTERLOCK : WORD; 
END_TYPE

However, when attempting to parse it using the current configuration, I receive the following error:

Error parsing node T_INTERLOCK in file test_files\baselib\BaseLibProj\BaseLib\POUs\HMI\Datentypen\T_INTERLOCK.TcDUT
(None:1:24: Expected '.' or '\s' or '//' or '(*(.|\n)?*)' or '{' or 'END_TYPE' => 'OCK : WORD; END_TYP')

Steps Taken:

I tried modifying the st_declaration.tx grammar in an attempt to accommodate this type declaration by adding or adjusting the TypeDef rule as follows:

TypeDef:
    comments*=CommentAny?                
    'TYPE'
    name=ID                             
    ('EXTENDS' extends=AnyType)?  
    ':'  
    CommentAny* 
    type=AnyType
    ';'?
    CommentAny* 
    'END_TYPE'
;

AnyType:
    TypeStruct | TypeUnion | TypeEnum | StringType | Fqn | 'WORD'
;

However, making these changes requires modifying the source code, and I’m wondering if there's a better approach.

[Philipp1297]

If changing the PlcDeclaration class to handle string types (like WORD) is an acceptable solution, I can make the necessary changes. But before proceeding, I wanted to confirm if that would be the right approach, or if there is a more elegant or preferable solution that avoids modifying the source code directly.

[Philipp1297]

I tried this inside the st_declaration.tx, but it seems not a proper approach:

TypeDef:
    comments*=CommentAny?               
    'TYPE'
    name=ID                             
    ('EXTENDS' extends=AnyType)?     
    ':'
    CommentAny*                       
    type=BaseTypeOrAnyType
    ';'?                       
    CommentAny*
    'END_TYPE'                          
;

BaseTypeOrAnyType:
    AnyType | BaseType     
;

AnyType:
    TypeStruct | TypeUnion | TypeEnum | StringType | Fqn
;

BaseType:
    WORD
;

WORD:
    'WORD'
;

[RobertoRoos]

Thanks for your report @Philipp1297 !

I'll have to admit I've never seen a TYPE ... declaration in structured text, I didn't know that existed.
EDIT: It's an alias: https://infosys.beckhoff.com/english.php?content=../content/1033/tc3_plc_intro/3928317323.html&id=

Am I understanding correctly this prevents Sphinx from completing, as in your run into an uncaught exception?

There's two things to pick up then:

• TYPE should be supported - the grammar should be updated to be able to parse it and custom TYPES should be handled with new directives in our plugin.
• Failed code parsing with our grammar should cause a bright red error in the log, but the plugin should continue, allowing docs generation to complete at least.

I'll look into either. I can use your suggested grammar modifications as a starting point. But I'll have to admit it might be a while before I can make some progress here.

[Philipp1297]

I have identified additional cases where the grammar is currently not functioning as expected.
I will gather these cases and provide the existing limitations, along with potential grammar or source code changes, here.

[Philipp1297]

Limitation in case of default value for TYPE Decleration:

TYPE E_HMI_BUTTON_FEEDBACK :
(
	NONE := 0,
	ACTIVE := 1,
	PENDING := 2,
	WARNING := 3,
	ERROR := 4
) INT := NONE;
END_TYPE

I fixed this by changing the grammer to this:

TypeDef:
    comments*=CommentAny?               
    'TYPE'
    name=ID                             
    ('EXTENDS' extends=AnyType)?     
    ':'
    CommentAny*                        
    type=BaseTypeOrAnyType
    ';'?                       
    CommentAny*
    'END_TYPE'                          
;

BaseTypeOrAnyType:
    AnyType | BaseType     
;

AnyType:
    TypeStruct | TypeUnion | TypeEnum | StringType | Fqn
;

BaseType:
    WORD | INT
;

WORD:
    'WORD'
;

TypeStruct:
    'STRUCT'
    members*=Variable
    CommentAny*
    // Catch trailing comments here, not at the end of `Variable`
    'END_STRUCT'
;

TypeUnion:
    'UNION'
    members*=Variable
    CommentAny*
    // Catch trailing comments here, not at the end of `Variable`
    'END_UNION'
;


TypeEnum:
    '('
    values*=EnumOption
    CommentAny* 
    ')'
    (base_type=AssignmentValue)? 
    ';'?
;

EnumOption:
    CommentAny*
    name=ID
    (':=' number=INT)?
    (',')?
    (comment=CommentLine)?
;

[Philipp1297]

Limitation in case of the char ';' for string default value like:

S_GET_CURR_MIN_MAX : STRING := 'CURR:LIM:NEG?;:CURR:LIM:POS?$L';

[Philipp1297]

Last (atm) found Limitation

_stHMIValve : ST_HMI_ANALOG_VALVE_DATA;;

Double Semicolon

[RobertoRoos]

Thanks again, I'll look into these. Are these from an open source project by any chance? In order to test the grammar for failures I would ideally just throw it against big existing projects. I'm already doing that for TcUnit for example.

[RobertoRoos]

I think all issues will be fixed with #28 , except for the string escape, which is more tricky. Let's continue that in #29 .