From e08a7cf1aa342ec0b96179ee847cee1a0adabe56 Mon Sep 17 00:00:00 2001
From: Almann Goo <almann.goo@gmail.com>
Date: Thu, 3 Jun 2021 22:42:07 -0700
Subject: [PATCH] Fixes an ambiguity with identifiers and keywords. (#32)

Previously, `SELECTY` would be parsed as keyword `SELECT` identifier
`Y`.  This fixes the negative look-ahead assertion for keywords to
make sure a non-quoted identifier character follows and similarly
adds a look-ahead assertion for the prefix of identifiers to not be
keywords (which includes the previous assertion).

Also fixes a unit test name for identifiers.
---
 partiql-parser/src/partiql.pest | 6 ++++--
 partiql-parser/src/scanner.rs   | 9 ++++++++-
 2 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/partiql-parser/src/partiql.pest b/partiql-parser/src/partiql.pest
index 986f117c..cf2716fe 100644
--- a/partiql-parser/src/partiql.pest
+++ b/partiql-parser/src/partiql.pest
@@ -155,7 +155,7 @@ StringContent = {
 Identifier = { NonQuotedIdentifier | QuotedIdentifier }
 
 NonQuotedIdentifier = @{
-    !Keyword ~ NonQuotedIdentifierStart ~ NonQuotedIdentifierCont*
+    !(Keyword) ~ NonQuotedIdentifierStart ~ NonQuotedIdentifierCont*
 }
 
 NonQuotedIdentifierStart = _{ "_" | "$" | 'a'..'z' | 'A'..'Z' }
@@ -173,7 +173,9 @@ QuotedIdentifierContent = {
 // Keywords
 //
 
-Keyword = { SqlKeyword | PartiqlKeyword }
+Keyword = @{ 
+    (SqlKeyword | PartiqlKeyword) ~ !NonQuotedIdentifierCont
+}
 
 SqlKeyword = _{
       Absolute_
diff --git a/partiql-parser/src/scanner.rs b/partiql-parser/src/scanner.rs
index 0b50cc6c..6492237f 100644
--- a/partiql-parser/src/scanner.rs
+++ b/partiql-parser/src/scanner.rs
@@ -509,7 +509,7 @@ mod test {
             "WHERE" => keyword("WHERE")
         ]
     )]
-    #[case::some_keywords(
+    #[case::some_identifiers(
         scanner_test_case![
             "moo_cow_1999" => identifier("moo_cow_1999"),
             " ",
@@ -518,6 +518,13 @@ mod test {
             "$$$$" => identifier("$$$$")
         ]
     )]
+    #[case::identifiers_with_keywords_in_them(
+        scanner_test_case![
+            "moowhere" => identifier("moowhere"),
+            " ",
+            "selecty" => identifier("selecty"),
+        ]
+    )]
     #[case::quoted_identifiers(
         scanner_test_case![
             "    ",