documents.js 1.63.0 → 1.64.0

package/dist/odb/sql/lexer.js

@@ -0,0 +1,281 @@
+import { HsqldbSqlParseError, HsqldbSqlUnsupportedError } from "./errors.js";
+//#region src/odb/sql/lexer.ts
+const KEYWORDS = /* @__PURE__ */ new Set([
+	"SELECT",
+	"FROM",
+	"WHERE",
+	"GROUP",
+	"BY",
+	"ORDER",
+	"ASC",
+	"DESC",
+	"AND",
+	"OR",
+	"NOT",
+	"IS",
+	"NULL",
+	"LIKE",
+	"IN",
+	"BETWEEN",
+	"TRUE",
+	"FALSE",
+	"COUNT",
+	"SUM",
+	"AVG",
+	"MIN",
+	"MAX",
+	"JOIN",
+	"INNER",
+	"OUTER",
+	"LEFT",
+	"RIGHT",
+	"FULL",
+	"CROSS",
+	"NATURAL",
+	"ON",
+	"USING",
+	"UNION",
+	"INTERSECT",
+	"EXCEPT",
+	"MINUS",
+	"DISTINCT",
+	"ALL",
+	"HAVING",
+	"LIMIT",
+	"OFFSET",
+	"FETCH",
+	"TOP",
+	"AS",
+	"CASE",
+	"WHEN",
+	"THEN",
+	"ELSE",
+	"END",
+	"EXISTS",
+	"ANY",
+	"SOME",
+	"WITH",
+	"INSERT",
+	"UPDATE",
+	"DELETE",
+	"CREATE",
+	"DROP",
+	"ALTER",
+	"INTO",
+	"VALUES",
+	"SET",
+	"ESCAPE",
+	"COLLATE",
+	"NULLS",
+	"FIRST",
+	"LAST"
+]);
+const UNSUPPORTED_SYMBOLS = [
+	["--", "a SQL line comment"],
+	["/*", "a SQL block comment"],
+	["||", "the || string-concatenation operator"],
+	["!=", "the != operator (write <> instead)"],
+	["+", "an arithmetic expression"],
+	["/", "an arithmetic expression"],
+	["%", "an arithmetic expression"],
+	["?", "a parameter placeholder"],
+	[":", "a named parameter"],
+	["@", "a variable reference"],
+	["#", "a variable reference"],
+	["&", "a bitwise operator"],
+	["|", "a bitwise operator"],
+	["^", "a bitwise operator"],
+	["~", "a bitwise operator"],
+	["!", "the ! operator"]
+];
+function isDigit(character) {
+	return character >= "0" && character <= "9";
+}
+function isIdentifierStart(character) {
+	return character >= "A" && character <= "Z" || character >= "a" && character <= "z" || character === "_";
+}
+function isIdentifierPart(character) {
+	return isIdentifierStart(character) || isDigit(character) || character === "$";
+}
+function isWhitespace(character) {
+	return character === " " || character === "	" || character === "\n" || character === "\r" || character === "\f" || character === "\v";
+}
+function readNumber(sql, start) {
+	let cursor = start;
+	while (cursor < sql.length && isDigit(sql.charAt(cursor))) cursor += 1;
+	if (cursor < sql.length && sql.charAt(cursor) === ".") {
+		cursor += 1;
+		while (cursor < sql.length && isDigit(sql.charAt(cursor))) cursor += 1;
+	}
+	if (cursor < sql.length && (sql.charAt(cursor) === "e" || sql.charAt(cursor) === "E")) {
+		const exponentStart = cursor;
+		cursor += 1;
+		if (cursor < sql.length && (sql.charAt(cursor) === "+" || sql.charAt(cursor) === "-")) cursor += 1;
+		if (cursor < sql.length && isDigit(sql.charAt(cursor))) while (cursor < sql.length && isDigit(sql.charAt(cursor))) cursor += 1;
+		else cursor = exponentStart;
+	}
+	const text = sql.slice(start, cursor);
+	const value = Number(text);
+	if (!Number.isFinite(value)) throw new HsqldbSqlParseError(`numeric literal "${text}" is not a finite number`, sql, start);
+	return {
+		value,
+		next: cursor
+	};
+}
+function readStringLiteral(sql, start) {
+	let cursor = start + 1;
+	let value = "";
+	while (cursor < sql.length) {
+		const character = sql.charAt(cursor);
+		if (character === "'") {
+			if (sql.charAt(cursor + 1) === "'") {
+				value += "'";
+				cursor += 2;
+				continue;
+			}
+			return {
+				value,
+				next: cursor + 1
+			};
+		}
+		value += character;
+		cursor += 1;
+	}
+	throw new HsqldbSqlParseError("unterminated string literal", sql, start);
+}
+function readQuotedIdentifier(sql, start) {
+	let cursor = start + 1;
+	let name = "";
+	while (cursor < sql.length) {
+		const character = sql.charAt(cursor);
+		if (character === "\"") {
+			if (sql.charAt(cursor + 1) === "\"") {
+				name += "\"";
+				cursor += 2;
+				continue;
+			}
+			if (name === "") throw new HsqldbSqlParseError("empty quoted identifier (\"\")", sql, start);
+			return {
+				name,
+				next: cursor + 1
+			};
+		}
+		name += character;
+		cursor += 1;
+	}
+	throw new HsqldbSqlParseError("unterminated quoted identifier", sql, start);
+}
+function unsupportedSymbolAt(sql, cursor) {
+	for (const [symbol, construct] of UNSUPPORTED_SYMBOLS) if (sql.startsWith(symbol, cursor)) return construct;
+}
+function tokenizeSql(sql) {
+	const tokens = [];
+	let cursor = 0;
+	while (cursor < sql.length) {
+		const character = sql.charAt(cursor);
+		if (isWhitespace(character)) {
+			cursor += 1;
+			continue;
+		}
+		if (character === "'") {
+			const start = cursor;
+			const literal = readStringLiteral(sql, cursor);
+			cursor = literal.next;
+			tokens.push({
+				kind: "string",
+				value: literal.value,
+				text: sql.slice(start, cursor),
+				start
+			});
+			continue;
+		}
+		if (character === "\"") {
+			const start = cursor;
+			const identifier = readQuotedIdentifier(sql, cursor);
+			cursor = identifier.next;
+			tokens.push({
+				kind: "identifier",
+				name: identifier.name,
+				quoted: true,
+				text: sql.slice(start, cursor),
+				start
+			});
+			continue;
+		}
+		if (isDigit(character) || character === "." && isDigit(sql.charAt(cursor + 1))) {
+			const start = cursor;
+			const numeric = readNumber(sql, cursor);
+			cursor = numeric.next;
+			tokens.push({
+				kind: "number",
+				value: numeric.value,
+				text: sql.slice(start, cursor),
+				start
+			});
+			continue;
+		}
+		if (isIdentifierStart(character)) {
+			const start = cursor;
+			while (cursor < sql.length && isIdentifierPart(sql.charAt(cursor))) cursor += 1;
+			const text = sql.slice(start, cursor);
+			const folded = text.toUpperCase();
+			if (KEYWORDS.has(folded)) tokens.push({
+				kind: "keyword",
+				keyword: folded,
+				text,
+				start
+			});
+			else tokens.push({
+				kind: "identifier",
+				name: folded,
+				quoted: false,
+				text,
+				start
+			});
+			continue;
+		}
+		if (character === "<" || character === ">" || character === "=") {
+			const start = cursor;
+			const twoCharacter = sql.slice(cursor, cursor + 2);
+			if (twoCharacter === "<=" || twoCharacter === ">=" || twoCharacter === "<>") {
+				cursor += 2;
+				tokens.push({
+					kind: "operator",
+					operator: twoCharacter,
+					text: twoCharacter,
+					start
+				});
+				continue;
+			}
+			cursor += 1;
+			tokens.push({
+				kind: "operator",
+				operator: character,
+				text: character,
+				start
+			});
+			continue;
+		}
+		const unsupported = unsupportedSymbolAt(sql, cursor);
+		if (unsupported !== void 0) throw new HsqldbSqlUnsupportedError(unsupported, sql);
+		if (character === "(" || character === ")" || character === "," || character === "." || character === "*" || character === ";" || character === "-") {
+			tokens.push({
+				kind: "punctuation",
+				punctuation: character,
+				text: character,
+				start: cursor
+			});
+			cursor += 1;
+			continue;
+		}
+		throw new HsqldbSqlParseError(`unexpected character "${character}"`, sql, cursor);
+	}
+	tokens.push({
+		kind: "end",
+		text: "",
+		start: sql.length
+	});
+	return tokens;
+}
+//#endregion
+export { tokenizeSql };