fix(parsing): improve regex patterns for DevInParser and DevInLexer t…

…o support more complex identifiers and whitespace handling #101 The commit addresses issues with the DevInParser and DevInLexer in the ext/devin-lang package. It introduces more flexible regex patterns for identifiers, allowing characters other than the initial '$', '@', or '/'. Additionally, the commit refactors the lexer to handle whitespace more efficiently, using a dedicated WHITE_SPACE token type. The parser definition is updated to leverage the new token types, and the build script is modified to reflect the changes in the generated parser and lexer classes. Finally, a test file is updated to demonstrate the new parsing capabilities.
unit-mesh · Mar 11, 2024 · 7ea39e3 · 7ea39e3
1 parent b2bfd1b
commit 7ea39e3
Show file tree

Hide file tree

Showing 6 changed files with 46 additions and 23 deletions.
diff --git a/build.gradle.kts b/build.gradle.kts
@@ -600,15 +600,12 @@ project(":exts:devin-lang") {
     tasks {
         generateLexer {
             sourceFile.set(file("src/grammar/DevInLexer.flex"))
-//            targetDir.set("src/gen/com/feakin/intellij/lexer")
-            targetOutputDir.set(file("src/gen/cc/unitmesh/language/lexer"))
-//            targetClass.set("_FeakinLexer")
+            targetOutputDir.set(file("src/gen/cc/unitmesh/language"))
             purgeOldFiles.set(true)
         }
 
         generateParser {
             sourceFile.set(file("src/grammar/DevInParser.bnf"))
-//            targetRoot.set("src/gen")
             targetRootOutputDir.set(file("src/gen"))
             pathToParser.set("cc/unitmesh/language/parser/DevInParser.java")
             pathToPsiRoot.set("cc/unitmesh/language/psi")

diff --git a/exts/devin-lang/src/grammar/DevInLexer.flex b/exts/devin-lang/src/grammar/DevInLexer.flex
@@ -1,9 +1,8 @@
-// Copyright 2000-2022 JetBrains s.r.o. and other contributors. Use of this source code is governed by the Apache 2.0 license that can be found in the LICENSE file.
 package cc.unitmesh.language;
 
 import com.intellij.lexer.FlexLexer;
 import com.intellij.psi.tree.IElementType;
-import cc.unitmesh.language.psi.DevInTypes;
+import static cc.unitmesh.language.psi.DevInTypes.*;
 import com.intellij.psi.TokenType;
 
 %%
@@ -23,20 +22,27 @@ import com.intellij.psi.TokenType;
 %eof{  return;
 %eof}
 
-CRLF=\R
-WHITE_SPACE=[\ \n\t\f]
-// $ variable
-STRING=\"([^\\\"\r\n]|\\[^\r\n])*\"?
-IDENTIFIER=[_a-zA-Z][_a-zA-Z0-9]*
+EOL=\R
+WHITE_SPACE=\s+
 
-%state WAITING_VALUE
+IDENTIFIER=[_a-zA-Z][_a-zA-Z0-9]*
+TEXT_SEGMENT=[^[@\\$]_a-zA-Z0-9]+
+WS=[ \t\n\x0B\f\r]
+NEWLINE=\n|\r\n
 
 %%
 <YYINITIAL> {
-    {STRING}           { return DevInTypes.STRING; }
-    {IDENTIFIER}       { return DevInTypes.IDENTIFIER; }
-}
+  {WHITE_SPACE}        { return TokenType.WHITE_SPACE; }
 
-({CRLF}|{WHITE_SPACE})+                                     { yybegin(YYINITIAL); return TokenType.WHITE_SPACE; }
+  "$"                  { return DOLLAR; }
+  "@"                  { return AT; }
+  "/"                  { return SLASH; }
+
+  {IDENTIFIER}         { return IDENTIFIER; }
+  {TEXT_SEGMENT}       { return TEXT_SEGMENT; }
+  {WS}                 { return WS; }
+  {NEWLINE}            { return NEWLINE; }
+
+}
 
 [^]                                                         { return TokenType.BAD_CHARACTER; }
diff --git a/exts/devin-lang/src/grammar/DevInParser.bnf b/exts/devin-lang/src/grammar/DevInParser.bnf
@@ -18,18 +18,25 @@
     DOLLAR          = '$'
     AT              = '@'
     SLASH           = '/'
+    // char should be any character except start with $, @, /
+    IDENTIFIER      = 'regexp:[_a-zA-Z][_a-zA-Z0-9]*'
+    TEXT_SEGMENT    = 'regexp:[^[@\\$]_a-zA-Z0-9]+'
+
+    WS              = 'regexp:\s'
+
+    NEWLINE         = 'regexp:\n|\r\n'
   ]
 }
 
-DevInFile ::= item_*
+DevInFile ::= item*
 
-item_ ::= (useVariable|useAgent|useCommand|STRING)
+private item ::= (useVariable|useAgent|useCommand|TEXT_SEGMENT|NEWLINE)
 
-useVariable ::= '$' IDENTIFIER
 // $use-variable
+useVariable ::= DOLLAR IDENTIFIER WS*
 
 // @use-agent
-useAgent ::= '@' IDENTIFIER
+useAgent ::= AT IDENTIFIER WS*
 
 // /use-command
-useCommand ::= '/' IDENTIFIER
+useCommand ::= SLASH IDENTIFIER WS*
diff --git a/exts/devin-lang/src/main/kotlin/cc/unitmesh/language/DevInParserDefinition.kt b/exts/devin-lang/src/main/kotlin/cc/unitmesh/language/DevInParserDefinition.kt
@@ -2,6 +2,7 @@ package cc.unitmesh.language
 
 import cc.unitmesh.language.parser.DevInParser
 import cc.unitmesh.language.psi.DevInFile
+import cc.unitmesh.language.psi.DevInTypes
 import com.intellij.lang.ASTNode
 import com.intellij.lang.ParserDefinition
 import com.intellij.lang.PsiParser
@@ -44,7 +45,7 @@ internal class DevInParserDefinition : ParserDefinition {
 
     @NotNull
     override fun createElement(node: ASTNode?): PsiElement {
-        TODO()
+        return DevInTypes.Factory.createElement(node)
     }
 
     companion object {

diff --git a/exts/devin-lang/src/test/testData/parser/BasicTest.devin b/exts/devin-lang/src/test/testData/parser/BasicTest.devin
@@ -1 +1 @@
-Gen hello, world @gen
+@gen what's this?
diff --git a/exts/devin-lang/src/test/testData/parser/BasicTest.txt b/exts/devin-lang/src/test/testData/parser/BasicTest.txt
@@ -0,0 +1,12 @@
+FILE
+  DevInUseAgentImpl(USE_AGENT)
+    PsiElement(DevInTokenType.@)('@')
+    PsiElement(DevInTokenType.IDENTIFIER)('gen')
+  PsiWhiteSpace(' ')
+  PsiErrorElement:DevInTokenType.$, DevInTokenType./, DevInTokenType.@, DevInTokenType.NEWLINE, DevInTokenType.TEXT_SEGMENT or DevInTokenType.WS expected, got 'what'
+    PsiElement(DevInTokenType.IDENTIFIER)('what')
+  PsiElement(DevInTokenType.TEXT_SEGMENT)(''')
+  PsiElement(DevInTokenType.IDENTIFIER)('s')
+  PsiWhiteSpace(' ')
+  PsiElement(DevInTokenType.IDENTIFIER)('this')
+  PsiElement(DevInTokenType.TEXT_SEGMENT)('?')