Eclipse Community Forums: TMF (Xtext) » Request Guidance on Best Practice for Parsing/Content Assist

Help

Home

Home » Modeling » TMF (Xtext) » Request Guidance on Best Practice for Parsing/Content Assist

Show: Today's Messages :: Show Polls :: Message Navigator

Request Guidance on Best Practice for Parsing/Content Assist [message #1776811]

Tue, 21 November 2017 20:36

Christoff Erasmus

Messages: 32
Registered: December 2016

Member

Hello,
What is the best practice for dealing with the code snippet below.

Looking for guidance regarding

Parsing
Content Assist

Based on the responses I should be able to figure out ( but pointer would be appreciated)

Validation
QuickFix
Value Converter (if required)

KeywordDATE: {KeywordDATFMT} 
	'DATE' ('('
		//2-Digit Year Formats
		(format='*MDY' separator=('/'|'-'|'.'|','|'&')?)  |
		(format='*DMY' separator=('/'|'-'|'.'|','|'&')?)  |
		(format='*YMD' separator=('/'|'-'|'.'|','|'&')?)  |
		(format='*JUL' separator=('/'|'-'|'.'|','|'&')?)  |
		//4-Digit Year Formats
		(format='*ISO' separator=('-')?) |
		(format='*USA' separator=('/')?) |
		(format='*EUR' separator=('.')?) |
		(format='*JIS' separator=('-')?)			 
	')')?;
;

KeywordTIME: {KeywordTIME}
	'TIME' ('('
		(format='*HMS' separator=(':'|'.'|','|'&')?) | 
		(format='*ISO' separator=('.')?)             |
		(format='*USA' separator=(':')?)             |
		(format='*EUR' separator=('.')?)             |
		(format='*JIS' separator=(':')?)			 
	')')?
;

Report message to a moderator

Re: Request Guidance on Best Practice for Parsing/Content Assist [message #1776812 is a reply to message #1776811]

Tue, 21 November 2017 20:39

Christian Dietrich

Messages: 14665
Registered: July 2009

Senior Member

Hmmm I don't really get your question. Can you please be more specific

Twitter : @chrdietrich
Blog : https://www.dietrich-it.de

Report message to a moderator

Re: Request Guidance on Best Practice for Parsing/Content Assist [message #1776890 is a reply to message #1776812]

Wed, 22 November 2017 14:49

Christoff Erasmus

Messages: 32
Registered: December 2016

Member

@Christian,

The refactored the snippet (see below) has 2 issues
1) The Content Assist does not recommend either the DATSEP or the TIMSEP as options
2) The .mwe2 workflow gives the following error message :
Quote:

error(208): ../bz.erasmus.x400.xrpgle.xtext/src-gen/bz/erasmus/x400/xrpgle/xtext/parser/antlr/lexer/InternalXRPGLELexer.g:74:1: The following token definitions can never be matched because prior tokens match the same input: RULE_TIMSEP

error(208): ../bz.erasmus.x400.xrpgle.xtext.ide/src-gen/bz/erasmus/x400/xrpgle/xtext/ide/contentassist/antlr/lexer/InternalXRPGLELexer.g:74:1: The following token definitions can never be matched because prior tokens match the same input: RULE_TIMSEP

My question:
What is most "correct" way of dealing with the grammar rule?

The Refactored Code Snippet:

KeywordDATE: {KeywordDATE}
	'DATE' ('(' format=DATFMT (separator=DATSEP)? ')')?
;
enum DATFMT:
	MDY = '*MDY' |
	DMY = '*DMY' |
	YMD = '*YMD' |
	JUL = '*JUL' |
	ISO = '*ISO' |
	USA = '*USA' |
	EUR = '*EUR' |
	JIS = '*JIS'
;
terminal DATSEP returns ecore::EChar:
	'/'|'-'|'.'|','|'&'
;
KeywordTIME: {KeywordTIME}
	'TIME' ('(' format=TIMFMT (separator=TIMSEP)? ')')?
;
enum TIMFMT:
	HMS = '*HMS' |
	ISO = '*ISO' |
	USA = '*USA' |
	EUR = '*EUR' |
	JIS = '*JIS'
;
terminal TIMSEP returns ecore::EChar:
	':'|'.'|','|'&'
;

[Updated on: Wed, 22 November 2017 14:51]

Report message to a moderator

Re: Request Guidance on Best Practice for Parsing/Content Assist [message #1776899 is a reply to message #1776890]

Wed, 22 November 2017 15:23

Christian Dietrich

Messages: 14665
Registered: July 2009

Senior Member

this error says your grammar is ambious. there is no general solutiion to this since it depends on the individual case. so a best practice is the wrong thing to ask for. your original snippet did no even contain TIMSEP
but having the same stuff as keywords and as terminals wont work. thus your solution seems legit (grammar snippets quite incomplete)

Twitter : @chrdietrich
Blog : https://www.dietrich-it.de

Report message to a moderator

Re: Request Guidance on Best Practice for Parsing/Content Assist [message #1776909 is a reply to message #1776899]

Wed, 22 November 2017 17:02

Christoff Erasmus

Messages: 32
Registered: December 2016

Member

@Christian,

The ambiguity is caused when DATFMT and TIMFMT is defined together (refer option2 in the grammar below).

Your previous post, confused me a bit. Do you recommend option1 or option2 or maybe a different approach?

My end-goal is to parse the "Date Keyword" from the RPGLE language, but I am struggling with the translation of the info on IBM's knowledge center into xtext grammer.
Links to the Knowledge center:
* Date Keyword - Parsing
* Date Data Type - Validation

More complete grammer

Model:
variabledeclarations+=StandAloneField
;

StandAloneField:
	'DCL-S' name=ID type=DataType keywords+=StandAloneFieldKeywords* (';')? 
;
StandAloneFieldKeywords:
	placeholder= 'StandAloneFieldKeywords'
;
DataType:
//	CharacterDataType	|
//	NumericDataType		|
	DateTimeDataType	
;
DateTimeDataType:
//	KeywordTIMESTAMP |
//	KeywordTIME      |
	KeywordDATE			

;
//---------------------------------------------------------------------------
// Option 1:
KeywordDATE: {KeywordDATE} 
	'DATE' ('('
		//2-Digit Year Formats
		(format='*MDY' separator=('/'|'-'|'.'|','|'&')?) |
		(format='*DMY' separator=('/'|'-'|'.'|','|'&')?) |
		(format='*YMD' separator=('/'|'-'|'.'|','|'&')?) |
		(format='*JUL' separator=('/'|'-'|'.'|','|'&')?) |
		//4-Digit Year Formats
		(format='*ISO' separator=('-')?)	|
		(format='*USA' separator=('/')?)	|
		(format='*EUR' separator=('.')?)	|
		(format='*JIS' separator=('-')?)			 
	')')?
;
KeywordTIME: {KeywordTIME}
	'TIME' ('('
		(format='*HMS' separator=(':'|'.'|','|'&')?)	| 
		(format='*ISO' separator=('.')?)				|
		(format='*USA' separator=(':')?)			 	|
		(format='*EUR' separator=('.')?)			 	|
		(format='*JIS' separator=(':')?)			 
	')')?
;
//------------------------------------------------------------------------
// Option2:
KeywordDATE: {KeywordDATE}
	'DATE' ('(' format=DATFMT (separator=DATSEP)? ')')?
;
enum DATFMT:
	ISO = '*ISO' |
	MDY = '*MDY' |
	DMY = '*DMY' |
	YMD = '*YMD' |
	JUL = '*JUL' |
	USA = '*USA' |
	EUR = '*EUR' |
	JIS = '*JIS'
;
terminal DATSEP returns ecore::EChar:
	'/'|'-'|'.'|','|'&'
;
KeywordTIME: {KeywordTIME}
	'TIME' ('(' format=TIMFMT (separator=TIMSEP)? ')')?
;
enum TIMFMT:
	ISO = '*ISO' |
	HMS = '*HMS' |
	USA = '*USA' |
	EUR = '*EUR' |
	JIS = '*JIS'
;
terminal TIMSEP returns ecore::EChar:
	':'|'.'|','|'&'
;

Report message to a moderator

Re: Request Guidance on Best Practice for Parsing/Content Assist [message #1776910 is a reply to message #1776909]

Wed, 22 November 2017 17:40

Christian Dietrich

Messages: 14665
Registered: July 2009

Senior Member

id prefer option 2, maybe replaced by a datatype rule (+ custom content assist) e.g.

// no terminal keyword
TIMSEP returns ecore::EChar:
':'|'.'|','|'&'
;

Twitter : @chrdietrich
Blog : https://www.dietrich-it.de

Report message to a moderator