feat(parser): extend Parser with parseOrderKeys

2026-09-04 11:23:23 +02:00
1 parent b362403ac5
commit 81653b5d5a
@@ -953,9 +953,59 @@ class Parser(private val tokens: List<Token>) {
        return if (keys.isEmpty()) null else OrderBy(keys)
    }

-    private fun parseLimit(): Int? {
-        if (!match(TokenType.Limit)) return null
-        return expect(TokenType.IntLiteral, "expected an integer after LIMIT").intValue!!.toInt()
+    /** Shared by top-level ORDER BY and a window function's `OVER (... ORDER BY ...)` clause. */
+    private fun parseOrderKeys(): List<OrderKey> {
+        if (!match(TokenType.Order)) return emptyList()
+        expect(TokenType.By, "expected BY after ORDER")
+        val keys = mutableListOf<OrderKey>()
+        do {
+            val expression = parseExpression()
+            val descending = when {
+                match(TokenType.Desc) -> true
+                match(TokenType.Asc) -> false
+                else -> false
+            }
+            val nullsFirst = parseOptionalNullsClause()
+            keys.add(OrderKey(expression, descending, nullsFirst))
+        } while (match(TokenType.Comma))
+        return keys
+    }
+
+    /**
+     * `NULLS FIRST` / `NULLS LAST`, optional after each ORDER BY key's own `ASC`/`DESC`. Like
+     * `ROLLUP`/`CUBE`/`FILTER`/`GENERATED` elsewhere in this parser, `NULLS`/`FIRST`/`LAST` are not
+     * reserved keywords - a column legitimately named any of the three is unaffected outside this
+     * exact position, since [isContextualKeyword] only fires here, right after an order key.
+     */
+    private fun parseOptionalNullsClause(): Boolean? {
+        if (!isContextualKeyword("NULLS")) return null
+        advance()
+        return when {
+            isContextualKeyword("FIRST") -> { advance(); true }
+            isContextualKeyword("LAST") -> { advance(); false }
+            else -> throw error("expected FIRST or LAST after NULLS but found '${describe(peek())}'", peek())
+        }
+    }
+
+    /**
+     * `LIMIT`/`OFFSET` may appear in either order (matching Postgres, unlike the standard's
+     * `OFFSET ... FETCH ...`), but not repeated - a second `LIMIT` or `OFFSET` stops the loop and is
+     * left for whatever follows (typically a parse error from the caller, e.g. a stray token before
+     * `;`) rather than silently overwriting the first.
+     */
+    private fun parseLimitOffset(): Pair<Int?, Int?> {
+        var limit: Int? = null
+        var offset: Int? = null
+        while (true) {
+            if (limit == null && match(TokenType.Limit)) {
+                limit = expect(TokenType.IntLiteral, "expected an integer after LIMIT").intValue!!.toInt()
+            } else if (offset == null && match(TokenType.Offset)) {
+                offset = expect(TokenType.IntLiteral, "expected an integer after OFFSET").intValue!!.toInt()
+            } else {
+                break
+            }
+        }
+        return limit to offset
    }

    private fun parseDelete(): DeleteStatement {
@@ -963,29 +1013,119 @@ class Parser(private val tokens: List<Token>) {
        expect(TokenType.From, "expected FROM after DELETE")
        val tableName = expectIdentifier("expected table name")
        val where = if (match(TokenType.Where)) parseExpression() else null
-        return DeleteStatement(tableName, where)
+        val returning = parseReturning()
+        return DeleteStatement(tableName, where, returning)
    }

    private fun parseUpdate(): UpdateStatement {
        expect(TokenType.Update, "expected UPDATE")
        val tableName = expectIdentifier("expected table name")
        expect(TokenType.Set, "expected SET after table name")
+        val assignments = parseAssignmentList()
+        val where = if (match(TokenType.Where)) parseExpression() else null
+        val returning = parseReturning()
+        return UpdateStatement(tableName, assignments, where, returning)
+    }
+
+    private fun parseAssignmentList(): List<Assignment> {
        val assignments = mutableListOf<Assignment>()
        do {
            val columnName = expectIdentifier("expected column name")
            expect(TokenType.Equal, "expected '=' in assignment")
-            assignments.add(Assignment(columnName, parseExpression()))
+            val value = if (match(TokenType.Default)) null else parseExpression()
+            assignments.add(Assignment(columnName, value))
        } while (match(TokenType.Comma))
-        val where = if (match(TokenType.Where)) parseExpression() else null
-        return UpdateStatement(tableName, assignments, where)
+        return assignments
+    }
+
+    /**
+     * `MERGE INTO target [[AS] alias] USING source ON condition (WHEN ...)+`. [parseFromItem] is
+     * reused verbatim for the `USING` source: it is exactly a `FROM`-shaped item (bare table name
+     * or a parenthesized, aliased derived-table subquery), not a MERGE-specific grammar production.
+     */
+    private fun parseMerge(): MergeStatement {
+        expect(TokenType.Merge, "expected MERGE")
+        expect(TokenType.Into, "expected INTO after MERGE")
+        val targetTable = expectIdentifier("expected target table name")
+        val targetAlias = parseOptionalAlias()
+        expect(TokenType.Using, "expected USING after MERGE target")
+        val source = parseFromItem()
+        expect(TokenType.On, "expected ON after MERGE USING source")
+        val onCondition = parseExpression()
+        val whenClauses = mutableListOf<MergeWhenClause>()
+        while (match(TokenType.When)) {
+            whenClauses.add(parseMergeWhenClause())
+        }
+        if (whenClauses.isEmpty()) {
+            throw error("MERGE requires at least one WHEN clause", peek())
+        }
+        val returning = parseReturning()
+        return MergeStatement(targetTable, targetAlias, source, onCondition, whenClauses, returning)
+    }
+
+    private fun parseOptionalAlias(): String? = when {
+        match(TokenType.As) -> expectIdentifier("expected alias after AS")
+        check(TokenType.Identifier) -> advance().lexeme
+        else -> null
+    }
+
+    /**
+     * `WHEN [NOT] MATCHED [AND condition] THEN action` - the `AND condition` sits between the
+     * MATCHED/NOT MATCHED keyword and THEN in both categories, matching Postgres. A NOT MATCHED
+     * action is always INSERT (there is no target row for NOT MATCHED to UPDATE or DELETE);
+     * [MergeNotMatchedInsert.values] uses the same "bare DEFAULT keyword or expression" per-slot
+     * grammar as plain INSERT's own VALUES list ([parseInsert]).
+     */
+    private fun parseMergeWhenClause(): MergeWhenClause {
+        val matched = when {
+            match(TokenType.Matched) -> true
+            match(TokenType.Not) -> {
+                expect(TokenType.Matched, "expected MATCHED after NOT")
+                false
+            }
+            else -> throw error("expected MATCHED or NOT MATCHED after WHEN but found '${describe(peek())}'", peek())
+        }
+        val condition = if (match(TokenType.And)) parseExpression() else null
+        expect(TokenType.Then, "expected THEN after WHEN ${if (matched) "MATCHED" else "NOT MATCHED"}")
+        if (matched) {
+            return when {
+                match(TokenType.Update) -> {
+                    expect(TokenType.Set, "expected SET after WHEN MATCHED THEN UPDATE")
+                    MergeMatchedUpdate(condition, parseAssignmentList())
+                }
+                match(TokenType.Delete) -> MergeMatchedDelete(condition)
+                else -> throw error("expected UPDATE or DELETE after WHEN MATCHED THEN but found '${describe(peek())}'", peek())
+            }
+        }
+        expect(TokenType.Insert, "expected INSERT after WHEN NOT MATCHED THEN")
+        var columns: List<String>? = null
+        if (match(TokenType.LeftParen)) {
+            val names = mutableListOf<String>()
+            do {
+                names.add(expectIdentifier("expected column name"))
+            } while (match(TokenType.Comma))
+            expect(TokenType.RightParen, "expected ')' after column list")
+            columns = names
+        }
+        expect(TokenType.Values, "expected VALUES after WHEN NOT MATCHED THEN INSERT")
+        expect(TokenType.LeftParen, "expected '(' before value list")
+        val values = mutableListOf<Expression?>()
+        do {
+            values.add(if (match(TokenType.Default)) null else parseExpression())
+        } while (match(TokenType.Comma))
+        expect(TokenType.RightParen, "expected ')' after value list")
+        return MergeNotMatchedInsert(condition, columns, values)
    }

    private fun parseDrop(): Statement {
        expect(TokenType.Drop, "expected DROP")
        return when {
-            match(TokenType.Table) -> DropTableStatement(expectIdentifier("expected table name"))
+            match(TokenType.Table) -> DropTableStatement(expectIdentifier("expected table name"), parseDropMode())
            match(TokenType.Index) -> DropIndexStatement(expectIdentifier("expected index name"))
-            else -> throw error("expected TABLE or INDEX after DROP but found '${describe(peek())}'", peek())
+            match(TokenType.View) -> DropViewStatement(expectIdentifier("expected view name"), parseDropMode())
+            match(TokenType.Trigger) -> DropTriggerStatement(expectIdentifier("expected trigger name"))
+            match(TokenType.Sequence) -> DropSequenceStatement(expectIdentifier("expected sequence name"))
+            else -> throw error("expected TABLE, INDEX, VIEW, TRIGGER, or SEQUENCE after DROP but found '${describe(peek())}'", peek())
        }
    }