Merge pull request #355 from alephium/number_follow_compiler

Follow rules in `Lexer.number` parser
alephium · Jan 16, 2025 · 22401c2 · 22401c2
2 parents 57a9478 + 1d1be2c
commit 22401c2
Show file tree

Hide file tree

Showing 3 changed files with 68 additions and 4 deletions.
diff --git a/...cess/src/main/scala/org/alephium/ralph/lsp/access/compiler/parser/soft/NumberParser.scala b/...cess/src/main/scala/org/alephium/ralph/lsp/access/compiler/parser/soft/NumberParser.scala
@@ -29,8 +29,7 @@ object NumberParser {
   private def number[Unknown: P]: P[SoftAST.CodeString] =
     P {
       Index ~
-        isNumber ~
-        (!Token.AlphLowercase.lexeme ~ CharIn("[0-9a-zA-Z]._+\\-")).rep(1).! ~
+        numberOrHex.! ~
         Index
     } map {
       case (from, number, to) =>
@@ -40,7 +39,34 @@ object NumberParser {
         )
     }
 
-  private def isNumber[Unknown: P]: P[Unit] =
-    P(&(CharIn("+\\-").? ~ CharIn("[0-9]")))
+  /**
+   * Parses characters that <u><b>could</b></u> (not strict) form a number.
+   *
+   * Follows rules defined by the parser [[org.alephium.ralph.Lexer.integer]].
+   * This is to handle all cases that are compiled successfully in ralphc.
+   *
+   * For example, the following are valid numbers in ralphc:
+   *
+   * {{{
+   *   let num0 = 1_
+   *   let num1 = 1_._
+   *   let num2 = 1_.____0
+   *   let num3 = 1_._0___0
+   * }}}
+   *
+   * Current parser code in [[org.alephium.ralph.Lexer.integer]], update accordingly:
+   * {{{
+   *   Index ~ (CharsWhileIn("0-9_") ~ ("." ~ CharsWhileIn("0-9_")).? ~
+   *     ("e" ~ "-".? ~ CharsWhileIn("0-9")).?).! ~
+   *     CharsWhileIn(" ", 0).! ~ token(Keyword.alph).?.!
+   * }}}
+   */
+  private def numberOrHex[Unknown: P]: P[Unit] =
+    P {
+      CharIn("\\+", "\\-").? ~
+        CharsWhileIn("0-9_") ~
+        ("." ~ CharsWhileIn("0-9_")).? ~
+        (!Token.AlphLowercase.lexeme ~ (StringIn("e-", "E-") | CharIn("0-9a-zA-Z_"))).rep
+    }
 
 }
diff --git a/.../src/test/scala/org/alephium/ralph/lsp/access/compiler/parser/soft/NumberParserSpec.scala b/.../src/test/scala/org/alephium/ralph/lsp/access/compiler/parser/soft/NumberParserSpec.scala
@@ -208,9 +208,43 @@ class NumberParserSpec extends AnyWordSpec with Matchers {
               )
           }
         }
+      }
+    }
+  }
+
+  "infix numbers with tail negative signs" should {
+    "not be parsed as a number, they are infix operations" in {
+      // These are not numbers, they are infix operations.
+      // The minus sign within the number is for scientific numbers `e-` only.
+      Seq(
+        "1-1",
+        "1- 1"
+      ) foreach {
+        infixOperation =>
+          // it only parses the first number 1, ignoring the sign and the tail number.
+          parseNumberNoCodeCheck(infixOperation) shouldBe
+            Number(
+              index = indexOf(">>1<<"),
+              text = "1"
+            )
 
+          // parser it from the root SoftParser parses it as an infix operation and not a number.
+          val infix = parseSoft(infixOperation)
+          infix.parts should have size 1
+          infix.parts.head.part shouldBe a[SoftAST.InfixExpression]
       }
     }
   }
 
+  "special cases" in {
+
+    /**
+     * See documentation of [[NumberParser.numberOrHex]]
+     */
+    assertSimpleNumber("1_")
+    assertSimpleNumber("1_._")
+    assertSimpleNumber("1_.____0")
+    assertSimpleNumber("1_._0___0")
+  }
+
 }
diff --git a/...access/src/test/scala/org/alephium/ralph/lsp/access/compiler/parser/soft/TestParser.scala b/...access/src/test/scala/org/alephium/ralph/lsp/access/compiler/parser/soft/TestParser.scala
@@ -78,6 +78,10 @@ object TestParser {
   def parseNumber(code: String): SoftAST.Number =
     runSoftParser(NumberParser.parseOrFail(_))(code)
 
+  /** Run the parser but perform no input code check */
+  def parseNumberNoCodeCheck(code: String): SoftAST.Number =
+    runAnyParser(NumberParser.parseOrFail(_))(code)
+
   def parseBString(code: String): SoftAST.BString =
     runSoftParser(BStringParser.parseOrFail(_))(code)