Migrate to PetitParser2

Add tests
Remove unnecessary code generator dependency
Add lowercaseCodes and uppercaseCodes to speed up parsing

repository/BaselineOfBioSmalltalk/BaselineOfBioSmalltalk.class.st

@@ -32,7 +32,6 @@ BaselineOfBioSmalltalk >> baseline: spec [
 			projectSpecForCommonUtils: spec;
 			projectSpecForDataDrameInspector: spec;
 			projectSpecForProjectFramework: spec;
-			projectSpecForCodeGenerator: spec;
 			projectSpecForStringExtensions: spec;
 			projectSpecForFileDialog: spec;
 			projectSpecForPolyMath: spec;
@@ -283,16 +282,6 @@ BaselineOfBioSmalltalk >> preLoad [
     super preLoad.
 ]
 
-{ #category : #specs }
-BaselineOfBioSmalltalk >> projectSpecForCodeGenerator: spec [
-
-	spec 
-		baseline: 'CodeGenerator'
-		with: [ spec 
-						repository: 'github://hernanmd/CodeGenerator/repository';
-						loads: #('Smalltalk-Generators') ]
-]
-
 { #category : #specs }
 BaselineOfBioSmalltalk >> projectSpecForCommonUtils: spec [
 
@@ -340,12 +329,9 @@ BaselineOfBioSmalltalk >> projectSpecForOsSubprocess: spec [
 BaselineOfBioSmalltalk >> projectSpecForPetitParser: spec [
 
 	spec
-		project: 'PetitParser' with: [
-			spec
-				className: #ConfigurationOfPetitParser;
-				versionString: '2.0';
-				loads: #('Core' );
-				repository: 'http://smalltalkhub.com/mc/Moose/PetitParser/main/' ].
+		baseline: 'PetitParser2Core'	
+		with: [ spec repository: 'github://kursjan/petitparser2' ].
+
 ]
 
 { #category : #specs }

repository/BioBlast/BioQBlastReader.class.st

@@ -66,7 +66,7 @@ BioQBlastReader >> parserForErrorMsg [
 	 Message ID#24 Error: Failed to read the Blast query: Nucleotide FASTA provided for protein sequence 
 	 Message ID#24 Error: Failed to read the Blast query: Gi/accession mismatch: requested nucleotide, found protein	 "
 
-	^ ( 'Message ID#' asParser negate plus , 'Message ID#' asParser ) ,
+	^ ( 'Message ID#' asPParser negate plus , 'Message ID#' asPParser ) ,
 		( PPPredicateObjectParser anyExceptAnyOf: '</p>' ) plus flatten foldRight: [: a : b | b ]
 
 ]

repository/BioParserTests/BioAccessionParserTest.class.st

@@ -1,7 +1,7 @@
 Class {
 	#name : #BioAccessionParserTest,
 	#superclass : #BioAbstractParserTest,
-	#category : 'BioParserTests-Core'
+	#category : #'BioParserTests-Core'
 }
 
 { #category : #accessing }
@@ -21,64 +21,102 @@ BioAccessionParserTest >> setUp [
 { #category : #testing }
 BioAccessionParserTest >> testParseAccession01 [
 
-	self assert: (( self parser parse: 'gi|555|emb|X65215.1|' ) 		isKindOf: BioAccession).
-	self assert: (( self parser parse: 'gi|226437718|gb|AC150860.6|' )	isKindOf: BioAccession).
-	self assert: (( self parser parse: 'gi|207524544|gb|AC226190.2|' )	isKindOf: BioAccession).
-	self assert: (( self parser parse: 'gb|AC226190.2|' )				isKindOf: BioAccession).
-	self assert: (( self parser parse: 'AC150530.4' )				isKindOf: BioAccession).
-	self assert: (( self parser parse: 'AC150707' )					isKindOf: BioAccession).
-
-	self assert: (( self parser parse: '>gi|198282148|ref|NC_011206.1| Acidithiobacillus ferrooxidans ATCC 53993 chromosome, complete genome' ) 																isKindOf: BioAccession).
-	self assert: (( self parser parse: '>gi|104773257|ref|NC_008054.1| Lactobacillus delbrueckii subsp. bulgaricus ATCC 11842, complete genome' ) 																isKindOf: BioAccession).
-
-
-
+	parseResult := self parser parse: 'gi|555|emb|X65215.1|'.
+	self assert: (parseResult isKindOf: BioAccession).
+	self assert: parseResult name equals: 'X65215'.
+	self assert: parseResult version equals: '1'.
 
 ]
 
 { #category : #testing }
 BioAccessionParserTest >> testParseAccession02 [
 
-	self assert: ( self parser parse: 'gi|555|emb|X65215.1|' ) name = 'X65215' .
-	self assert: ( self parser parse: 'gi|226437718|gb|AC150860.6|' ) name = 'AC150860'.
-	self assert: ( self parser parse: 'gi|207524544|gb|AC226190.2|' ) name = 'AC226190'.
-	self assert: ( self parser parse: 'gi|207524544|gb|AC226190.2345|' ) name = 'AC226190'.	
-	self assert: ( self parser parse: 'gi|207524544|gb|AC226190.2|ABCDEFG' ) name = 'AC226190'.	
-	self assert: ( self parser parse: 'gi|207524544|gb|AC226190.2|ABCDEFG|' ) name = 'AC226190'.
-	self assert: ( self parser parse: 'gi|207524544|gb|AC226190.2|ABCDEFG|HIJKL' ) name = 'AC226190'.
-	self assert: ( self parser parse: 'gb|AC226190.2|' ) name = 'AC226190'.
-	self assert: ( self parser parse: 'AC150530.4' ) name = 'AC150530'.
-	self assert: ( self parser parse: 'AC150707' ) name = 'AC150707'
+	parseResult := self parser parse: 'gi|226437718|gb|AC150860.6|'.
+	self assert: (parseResult isKindOf: BioAccession).
+	self assert: parseResult name equals: 'AC150860'.
+	self assert: parseResult version equals: '6'.
 
 ]
 
 { #category : #testing }
 BioAccessionParserTest >> testParseAccession03 [
 
-	self assert: ( self parser parse: 'gi|555|emb|X65215.1|' ) 						version = '1' .
-	self assert: ( self parser parse: 'gi|226437718|gb|AC150860.6|' ) 					version = '6'.
-	self assert: ( self parser parse: 'gi|207524544|gb|AC226190.2|' ) 					version = '2'.
-	self assert: ( self parser parse: 'gi|207524544|gb|AC226190.2345|' ) 				version = '2345'.		
-	self assert: ( self parser parse: 'gi|207524544|gb|AC226190.2|ABCDEFG' ) 			version = '2'.	
-	self assert: ( self parser parse: 'gi|207524544|gb|AC226190.2|ABCDEFG|' ) 			version = '2'.
-	self assert: ( self parser parse: 'gi|207524544|gb|AC226190.2|ABCDEFG|HIJKL' ) 	version = '2'.	
-	self assert: ( self parser parse: 'gb|AC226190.2|' ) 								version = '2'.
-	self assert: ( self parser parse: 'AC150530.4' ) 								version = '4'.
-	self assert: ( self parser parse: 'AC150707' ) 									version isNil
+	parseResult := self parser parse: 'gi|207524544|gb|AC226190.2|'.
+	self assert: (parseResult isKindOf: BioAccession).
+	self assert: parseResult name equals: 'AC226190'.
+	self assert: parseResult version equals: '2'.
+]
+
+{ #category : #testing }
+BioAccessionParserTest >> testParseAccession04 [
+
+	parseResult := self parser parse: 'gb|AC226190.2|'.
+	self assert: (parseResult isKindOf: BioAccession).
+	self assert: parseResult name equals: 'AC226190'.
+	self assert: parseResult version equals: '2'.
+]
+
+{ #category : #testing }
+BioAccessionParserTest >> testParseAccession05 [
+
+	parseResult := self parser parse: 'AC150530.4'.
+	self assert: (parseResult isKindOf: BioAccession).
+	self assert: parseResult name equals: 'AC150530'.
+	self assert: parseResult version equals: '4'.
+]
+
+{ #category : #testing }
+BioAccessionParserTest >> testParseAccession06 [
+
+	parseResult := self parser parse: 'AC150707'.
+	self assert: (parseResult isKindOf: BioAccession).
+	self assert: parseResult name equals: 'AC150707'.
+	self deny: parseResult hasVersion.
+]
 
+{ #category : #testing }
+BioAccessionParserTest >> testParseAccession07 [
+
+	parseResult := self parser parse: '>gi|198282148|ref|NC_011206.1| Acidithiobacillus ferrooxidans ATCC 53993 chromosome, complete genome'.
+	self assert: (parseResult isKindOf: BioAccession).
+	self assert: parseResult name equals: 'NC_011206'.
+	self assert: parseResult version equals: '1'.
+]
+
+{ #category : #testing }
+BioAccessionParserTest >> testParseAccession08 [
 
+	parseResult := self parser parse: '>gi|104773257|ref|NC_008054.1| Lactobacillus delbrueckii subsp. bulgaricus ATCC 11842, complete genome'.
+	self assert: (parseResult isKindOf: BioAccession).
+	self assert: parseResult name equals: 'NC_008054'.
+	self assert: parseResult version equals: '1'.
 ]
 
 { #category : #testing }
 BioAccessionParserTest >> testTokenizeAccession01 [
 
-	self assert: ( BioParser tokenizeAccession: 'gi|555|emb|X65215.1|' ) = #('X65215' '1') .
-	self assert: ( BioParser tokenizeAccession: 'gi|226437718|gb|AC150860.6|' ) = #('AC150860' '6').
-	self assert: ( BioParser tokenizeAccession: 'gi|207524544|gb|AC226190.2|' ) = #('AC226190' '2').
-	self assert: ( BioParser tokenizeAccession: 'gi|207524544|gb|AC226190.2345|' ) = #('AC226190' '2345').
-	self assert: ( BioParser tokenizeAccession: 'gb|AC226190.2|' ) = #('AC226190' '2').
-	self assert: ( BioParser tokenizeAccession: 'AC150530.4' ) = #('AC150530' '4').
-	self assert: ( BioParser tokenizeAccession: 'AC150707' ) = #('AC150707')
+	self
+		assert: (BioParser tokenizeAccession: 'gi|555|emb|X65215.1|')
+		equals: #( 'X65215' '1' ).
+	self
+		assert: (BioParser tokenizeAccession: 'gi|226437718|gb|AC150860.6|')
+		equals: #( 'AC150860' '6' ).
+	self
+		assert: (BioParser tokenizeAccession: 'gi|207524544|gb|AC226190.2|')
+		equals: #( 'AC226190' '2' ).
+	self
+		assert:
+		(BioParser tokenizeAccession: 'gi|207524544|gb|AC226190.2345|')
+		equals: #( 'AC226190' '2345' ).
+	self
+		assert: (BioParser tokenizeAccession: 'gb|AC226190.2|')
+		equals: #( 'AC226190' '2' ).
+	self
+		assert: (BioParser tokenizeAccession: 'AC150530.4')
+		equals: #( 'AC150530' '4' ).
+	self
+		assert: (BioParser tokenizeAccession: 'AC150707')
+		equals: #( 'AC150707' )
 ]
 
 { #category : #testing }

repository/BioParserTests/BioDNANucleotideParserTest.class.st

@@ -0,0 +1,65 @@
+Class {
+	#name : #BioDNANucleotideParserTest,
+	#superclass : #BioAbstractParserTest,
+	#category : #'BioParserTests-Basic'
+}
+
+{ #category : #accessing }
+BioDNANucleotideParserTest >> setUp [
+
+	super setUp.
+	parser := #dnaLetter asPParser.
+
+]
+
+{ #category : #testing }
+BioDNANucleotideParserTest >> testDNALetterMatchA [
+
+	self assert: (parser matches: 'a').
+	self assert: (parser matches: 'A').
+
+]
+
+{ #category : #testing }
+BioDNANucleotideParserTest >> testDNALetterMatchC [
+
+	self assert: (parser matches: 'c').
+	self assert: (parser matches: 'C').
+
+]
+
+{ #category : #testing }
+BioDNANucleotideParserTest >> testDNALetterMatchG [
+
+	self assert: (parser matches: 'g').
+	self assert: (parser matches: 'G').
+
+]
+
+{ #category : #testing }
+BioDNANucleotideParserTest >> testDNALetterMatchN [
+
+	self assert: (parser matches: 'N').
+	self assert: (parser matches: 'n').
+]
+
+{ #category : #testing }
+BioDNANucleotideParserTest >> testDNALetterMatchT [
+
+	self assert: (parser matches: 't').
+	self assert: (parser matches: 'T').
+
+]
+
+{ #category : #testing }
+BioDNANucleotideParserTest >> testDNANucelotideEmpty [
+
+	self deny: (parser matches: String empty)
+]
+
+{ #category : #testing }
+BioDNANucleotideParserTest >> testDNANucleotideCharacter [
+
+	self should: [ parser matches: $a ] 		raise: MessageNotUnderstood.
+	self should: [ parser matches: nil ] 		raise: MessageNotUnderstood.
+]

repository/BioParserTests/BioDNASequenceParserTest.class.st

@@ -0,0 +1,44 @@
+Class {
+	#name : #BioDNASequenceParserTest,
+	#superclass : #BioAbstractParserTest,
+	#category : #'BioParserTests-Basic'
+}
+
+{ #category : #accessing }
+BioDNASequenceParserTest >> setUp [
+
+	super setUp.
+	parser := #dnaSequence asPParser.
+
+]
+
+{ #category : #testing }
+BioDNASequenceParserTest >> testDNASeqMatchString [
+
+	self assert: (parser matches: 'actg').
+	self assert: (parser matches: 'ACTG').	
+
+]
+
+{ #category : #testing }
+BioDNASequenceParserTest >> testDNASeqMatches [
+
+	self assert: (parser matches: 'a').
+	self assert: (parser matches: 'A').
+	self assert: (parser matches: 'N').
+	self assert: (parser matches: 'n').
+
+]
+
+{ #category : #testing }
+BioDNASequenceParserTest >> testDNASeqNumber [
+
+	self deny: (parser matches: '8743').
+]
+
+{ #category : #testing }
+BioDNASequenceParserTest >> testDNASeqParseEmpty [
+
+	self deny: (parser matches: String empty).
+
+]