plume 0.1.0 → 0.2.0

data/lib/plume/parser.rb

@@ -0,0 +1,4015 @@
+# frozen_string_literal: true
+
+module Plume
+	class Parser
+		# https://www.sqlite.org/syntaxdiagrams.html
+		# Thrown on every dead end, most of which `maybe` backtracks out of, so the
+		# message is only put together when someone actually reads it
+		class ErrorMessage
+			def initialize(sql, token, beg, fin, expected)
+				@sql, @token, @beg, @fin, @expected = sql, token, beg, fin, expected
+			end
+
+			def msg
+				line = @sql.byteslice(0, @beg).count("\n") + 1
+				where = @sql.strip.include?("\n") ? " line #{line}" : ""
+				what = @token ? "token #{@token}[#{@sql.byteslice(@beg, @fin - @beg).inspect}]" : "end of input"
+				"Unexpected #{what} at#{where}:\n#{Plume.excerpt(@sql, @beg, @fin)}\n  Expected one of: #{@expected.join(", ")}\n"
+			end
+		end
+
+		# see: https://github.com/sqlite/sqlite/blob/master/src/parse.y#L255-L283
+		TOKEN_FALLBACKS = Set[
+			:EXPLAIN,
+			:QUERY,
+			:PLAN,
+			:BEGIN,
+			:DEFERRED,
+			:IMMEDIATE,
+			:EXCLUSIVE,
+			:END,
+			:ROLLBACK,
+			:SAVEPOINT,
+			:RELEASE,
+			:IF,
+			:TEMP,
+			:TEMPORARY,
+			:WITHIN,
+			:WITHOUT,
+			:ABORT,
+			:ACTION,
+			:AFTER,
+			:ANALYZE,
+			:ASC,
+			:ATTACH,
+			:BEFORE,
+			:BY,
+			:CASCADE,
+			:CAST,
+			:CONFLICT,
+			:DATABASE,
+			:DESC,
+			:DETACH,
+			:EACH,
+			:FAIL,
+			:MATCH,
+			:REGEXP,
+			:GLOB,
+			:LIKE,
+			:ROWID,
+			:STRICT,
+			:COLUMN,
+			:DO,
+			:FOR,
+			:IGNORE,
+			:INITIALLY,
+			:INSTEAD,
+			:NO,
+			:KEY,
+			:OF,
+			:OFFSET,
+			:PRAGMA,
+			:RAISE,
+			:RECURSIVE,
+			:REPLACE,
+			:RESTRICT,
+			:ROW,
+			:ROWS,
+			:TRIGGER,
+			:VACUUM,
+			:VIEW,
+			:VIRTUAL,
+			:WITH,
+			:NULLS,
+			:FIRST,
+			:LAST,
+			:CURRENT,
+			:FOLLOWING,
+			:PARTITION,
+			:PRECEDING,
+			:RANGE,
+			:UNBOUNDED,
+			:EXCLUDE,
+			:GROUPS,
+			:OTHERS,
+			:TIES,
+			:GENERATED,
+			:ALWAYS,
+			:MATERIALIZED,
+			:REINDEX,
+			:RENAME,
+			:CTIME_KW,
+		].freeze
+
+		OPERATOR_PRECEDENCE = {
+			BITNOT:  100,
+			UPLUS:   100,
+			UMINUS:  100,
+			COLLATE:  90,
+			CONCAT:   80,
+			PTR1:     80,
+			PTR2:     80,
+			STAR:     70,
+			SLASH:    70,
+			REM:      70,
+			PLUS:     60,
+			MINUS:    60,
+			BITAND:   55,
+			BITOR:    55,
+			LSHIFT:   55,
+			RSHIFT:   55,
+			NOT:      50,
+			LT:       40,
+			GT:       40,
+			LE:       40,
+			GE:       40,
+			EQ:       30,
+			NE:       30,
+			IS:       30,
+			ISNULL:   30,
+			NOTNULL:  30,
+			BETWEEN:  20,
+			IN:       20,
+			MATCH:    20,
+			LIKE:     20,
+			REGEXP:   20,
+			GLOB:     20,
+			AND:      10,
+			OR:        5,
+		}.freeze
+
+		def initialize(sql)
+			@lexer = Lexer.new(sql, skip_spaces: true)
+			@peek_buffer = []
+		end
+
+		def parse
+			result = catch(:ERROR) { sql_stmt_list }
+
+			if ErrorMessage === result
+				raise SyntaxError.new(result.msg)
+			else
+				result
+			end
+		rescue SystemStackError
+			# How deep is too deep depends on the stack the caller has left, so no fixture relies on it
+			raise SyntaxError.new("statement nested too deeply")
+		end
+
+		def sql_stmt_list
+			# Relevant `parse.y` grammar rules:
+			#   input ::= cmdlist
+			#   cmdlist ::= cmdlist ecmd | ecmd
+			#   ecmd ::= SEMI | cmdx SEMI | explain cmdx SEMI
+			#
+			# Syntax diagram:
+			#     ┌──────{ ; }◀─────┐
+			#   ◯─┴┬─▶[ sql-stmt ]─┬┴─▶◯
+			#      └───────▶───────┘
+			#
+			# Simplified grammar:
+			#   SEMI* [sql_stmt SEMI]*
+			#
+			# Relevant SQLite documentation:
+			#   - https://www.sqlite.org/lang.html
+			#
+			# A `sql_stmt_list` is zero or more statements,
+			# where a `;` is the statement separator,
+			# but an empty statement is allowed.
+			#
+			[].tap do |stmts|
+				until current_token.nil?
+					maybe :SEMI
+					stmts << sql_stmt if current_token
+					maybe :SEMI
+				end
+			end
+		end
+
+		def sql_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   ecmd ::= SEMI | cmdx SEMI | explain cmdx SEMI
+				#   explain ::= EXPLAIN | EXPLAIN QUERY PLAN
+				#   cmdx ::= cmd
+				#
+				# Syntax diagram:
+				#   ◯─┬─────────────┬▶─────────────────────┬─┬─▶[ alter-table-stmt ]──────────▶─┬─▶◯
+				#     └─{ EXPLAIN }─┴─▶{ QUERY }─▶{ PLAN }─┘ ├─▶[ analyze-stmt ]──────────────▶─┤
+				#                                            ├─▶[ attach-stmt ]───────────────▶─┤
+				#                                            ├─▶[ begin-stmt ]────────────────▶─┤
+				#                                            ├─▶[ commit-stmt ]───────────────▶─┤
+				#                                            ├─▶[ create-index-stmt ]─────────▶─┤
+				#                                            ├─▶[ create-table-stmt ]─────────▶─┤
+				#                                            ├─▶[ create-trigger-stmt ]───────▶─┤
+				#                                            ├─▶[ create-view-stmt ]──────────▶─┤
+				#                                            ├─▶[ create-virtual-table-stmt ]─▶─┤
+				#                                            ├─▶[ delete-stmt ]───────────────▶─┤
+				#                                            ├─▶[ detach-stmt ]───────────────▶─┤
+				#                                            ├─▶[ drop-index-stmt ]───────────▶─┤
+				#                                            ├─▶[ drop-table-stmt ]───────────▶─┤
+				#                                            ├─▶[ drop-trigger-stmt ]─────────▶─┤
+				#                                            ├─▶[ drop-view-stmt ]────────────▶─┤
+				#                                            ├─▶[ insert-stmt ]───────────────▶─┤
+				#                                            ├─▶[ pragma-stmt ]───────────────▶─┤
+				#                                            ├─▶[ reindex-stmt ]──────────────▶─┤
+				#                                            ├─▶[ release-stmt ]──────────────▶─┤
+				#                                            ├─▶[ rollback-stmt ]─────────────▶─┤
+				#                                            ├─▶[ savepoint-stmt ]────────────▶─┤
+				#                                            ├─▶[ select-stmt ]───────────────▶─┤
+				#                                            ├─▶[ update-stmt ]───────────────▶─┤
+				#                                            └─▶[ vacuum-stmt ]───────────────▶─┘
+				# Simplified grammar:
+				#   [EXPLAIN [QUERY PLAN]] stmt
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_explain.html
+				#   - https://www.sqlite.org/eqp.html
+				#   - https://www.sqlite.org/opcode.html
+				#
+				# A `sql_stmt` is a single SQL statement, optionally preceded by one of two different
+				# explanation prefixes. A simple `EXPLAIN` prefix requests the bytecode instructions,
+				# while `EXPLAIN QUERY PLAN` requests the query plan.
+				#
+				# This method doesn't consume tokens associated with the statements so that each
+				# statement method can independently consume the tokens it needs.
+			#
+			if (explain_kw = maybe_one_of [:EXPLAIN, :QUERY, :PLAN].freeze, :EXPLAIN)
+				return ExplainStatement.concrete(
+					full_source: @lexer.sql,
+					explain_kw:  Token::Keyword(explain_kw),
+					statement:   cmd,
+				)
+			end
+
+			cmd
+		end
+
+		def cmd
+			case current_token
+			when :ALTER							then alter_table_stmt
+			when :ANALYZE						then analyze_stmt
+			when :ATTACH						then attach_stmt
+			when :BEGIN							then begin_stmt
+			when :COMMIT, :END			then commit_stmt
+			when :CREATE
+				case peek
+				when :INDEX		then create_index_stmt
+				when :TABLE		then create_table_stmt
+				when :TRIGGER	then create_trigger_stmt
+				when :VIEW		then create_view_stmt
+				when :VIRTUAL	then create_virtual_table_stmt
+				when :UNIQUE	then create_index_stmt
+				when :TEMP, :TEMPORARY
+					case peek(2)
+					when [:TEMP, :TABLE]				then create_table_stmt
+					when [:TEMP, :TRIGGER]			then create_trigger_stmt
+					when [:TEMP, :VIEW]					then create_view_stmt
+					when [:TEMPORARY, :TABLE]		then create_table_stmt
+					when [:TEMPORARY, :TRIGGER]	then create_trigger_stmt
+					when [:TEMPORARY, :VIEW]		then create_view_stmt
+					else
+						expected! :TABLE, :TRIGGER, :VIEW
+					end
+				else
+					expected! :INDEX, :TABLE, :TRIGGER, :VIEW, :VIRTUAL, :UNIQUE, :TEMP, :TEMPORARY
+				end
+			when :DELETE						then delete_stmt
+			when :DETACH						then detach_stmt
+			when :DROP							then drop_stmt
+			when :INSERT, :REPLACE	then insert_stmt
+			when :PRAGMA						then pragma_stmt
+			when :REINDEX						then reindex_stmt
+			when :RELEASE						then release_stmt
+			when :ROLLBACK					then rollback_stmt
+			when :SAVEPOINT					then savepoint_stmt
+			when :SELECT, :VALUES		then select_stmt
+			when :UPDATE						then update_stmt
+			when :VACUUM						then vacuum_stmt
+			when :WITH							then with_stmt
+			else
+				expected!(:ALTER, :ANALYZE, :ATTACH, :BEGIN, :COMMIT, :END, :CREATE, :DELETE, :DETACH, :DROP, :INSERT, :REPLACE, :PRAGMA, :REINDEX, :RELEASE, :ROLLBACK, :SAVEPOINT, :SELECT, :UPDATE, :VACUUM, :WITH)
+			end
+		end
+
+		# ---------- Statements ----------
+
+		def begin_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= BEGIN transtype trans_opt
+				#   transtype ::= . | DEFERRED | IMMEDIATE | EXCLUSIVE
+				#   trans_opt ::= . | TRANSACTION | TRANSACTION nm
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_transaction.html
+			#
+			begin_kw = require :BEGIN
+			type_kw = maybe_one_of :DEFERRED, :IMMEDIATE, :EXCLUSIVE
+			transaction_kw, name_tk = transaction_name
+
+			BeginStatement.concrete(
+				full_source:    @lexer.sql,
+				begin_kw:       Token::Keyword(begin_kw),
+				type_kw:        Token::Keyword(type_kw),
+				transaction_kw: Token::Keyword(transaction_kw),
+				name_tk:        Token::Identifier(name_tk),
+			)
+		end
+
+		def commit_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= COMMIT|END trans_opt
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_transaction.html
+			#
+			commit_kw = require_one_of :COMMIT, :END
+			transaction_kw, name_tk = transaction_name
+
+			CommitStatement.concrete(
+				full_source:    @lexer.sql,
+				commit_kw:      Token::Keyword(commit_kw),
+				transaction_kw: Token::Keyword(transaction_kw),
+				name_tk:        Token::Identifier(name_tk),
+			)
+		end
+
+		def rollback_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= ROLLBACK trans_opt
+				#   cmd ::= ROLLBACK trans_opt TO savepoint_opt nm
+				#   savepoint_opt ::= SAVEPOINT | .
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_transaction.html
+				#   - https://www.sqlite.org/lang_savepoint.html
+			#
+			rollback_kw = require :ROLLBACK
+			transaction_kw, name_tk = transaction_name
+			if (to_kw = maybe_one_of [:TO, :SAVEPOINT].freeze, :TO)
+				savepoint_tk = name
+			end
+
+			RollbackStatement.concrete(
+				full_source:    @lexer.sql,
+				rollback_kw:    Token::Keyword(rollback_kw),
+				transaction_kw: Token::Keyword(transaction_kw),
+				name_tk:        Token::Identifier(name_tk),
+				to_kw:          Token::Keyword(to_kw),
+				savepoint_tk:   Token::Identifier(savepoint_tk),
+			)
+		end
+
+		# `trans_opt ::= . | TRANSACTION | TRANSACTION nm` - the name is accepted and ignored by SQLite
+		def transaction_name
+			return [] unless (transaction_kw = maybe :TRANSACTION)
+
+			name_tk = maybe { name(except: [:TO]) } unless statement_end?
+			[transaction_kw, name_tk]
+		end
+
+		def savepoint_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= SAVEPOINT nm
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_savepoint.html
+			#
+			savepoint_kw = require :SAVEPOINT
+
+			SavepointStatement.concrete(
+				full_source:  @lexer.sql,
+				savepoint_kw: Token::Keyword(savepoint_kw),
+				name_tk:      Token::Identifier(name),
+			)
+		end
+
+		def release_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= RELEASE savepoint_opt nm
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_savepoint.html
+			#
+			release_kw = require_one_of [:RELEASE, :SAVEPOINT].freeze, :RELEASE
+
+			ReleaseStatement.concrete(
+				full_source: @lexer.sql,
+				release_kw:  Token::Keyword(release_kw),
+				name_tk:     Token::Identifier(name),
+			)
+		end
+
+		def drop_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= DROP TABLE ifexists fullname
+				#   cmd ::= DROP VIEW ifexists fullname
+				#   cmd ::= DROP INDEX ifexists fullname
+				#   cmd ::= DROP TRIGGER ifexists fullname
+				#   ifexists ::= IF EXISTS | .
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_droptable.html
+				#   - https://www.sqlite.org/lang_dropview.html
+				#   - https://www.sqlite.org/lang_dropindex.html
+				#   - https://www.sqlite.org/lang_droptrigger.html
+			#
+			drop_kw = require :DROP
+			type_kw = require_one_of :TABLE, :VIEW, :INDEX, :TRIGGER
+			if_exists_kw = maybe_all_of :IF, :EXISTS
+
+			DropStatement.concrete(
+				full_source:  @lexer.sql,
+				drop_kw:      Token::Keyword(drop_kw),
+				type_kw:      Token::Keyword(type_kw),
+				if_exists_kw: Token::Keyword(if_exists_kw),
+				target:       table_name,
+			)
+		end
+
+		def pragma_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= PRAGMA nm dbnm
+				#   cmd ::= PRAGMA nm dbnm EQ nmnum
+				#   cmd ::= PRAGMA nm dbnm LP nmnum RP
+				#   cmd ::= PRAGMA nm dbnm EQ minus_num
+				#   cmd ::= PRAGMA nm dbnm LP minus_num RP
+				#   nmnum ::= plus_num | nm | ON | DELETE | DEFAULT
+				#
+				# Syntax diagram:
+				#   ◯─▶{ PRAGMA }┬▶{ schema-name }─▶{ . }┬▶{ pragma-name }┬─────────────────────────────┬─▶◯
+				#                └───────────▶───────────┘                ├▶{ = }─▶[ pragma-value ]─────┤
+				#                                                         └▶{ ( }─▶[ pragma-value ]─▶{ ) }┘
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/pragma.html
+			#
+			pragma_kw = require :PRAGMA
+			name_tk = name
+			if (dot_tk = maybe :DOT)
+				schema_tk = name_tk
+				name_tk = name
+			end
+			if (eq_tk = maybe :EQ)
+				value_tk = pragma_value
+			elsif (eq_tk = maybe :LP)
+				value_tk = pragma_value
+				value_rp = require :RP
+			end
+
+			PragmaStatement.concrete(
+				full_source: @lexer.sql,
+				pragma_kw:   Token::Keyword(pragma_kw),
+				schema_tk:   Token::Identifier(schema_tk),
+				dot_tk:      Token::Punctuation(dot_tk),
+				name_tk:     Token::Identifier(name_tk),
+				eq_tk:       Token::Punctuation(eq_tk),
+				value_tk:    value_tk,
+				value_rp:    Token::Punctuation(value_rp),
+			)
+		end
+
+		def pragma_value
+			if current_token in :PLUS | :MINUS | :INTEGER | :FLOAT
+				Token::Numeric(signed_number)
+			elsif (keyword_tk = maybe_one_of :ON, :DELETE, :DEFAULT, :TRUE, :FALSE)
+				Token::Keyword(keyword_tk)
+			else
+				Token::Identifier(name)
+			end
+		end
+
+		def attach_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= ATTACH database_kw_opt expr AS expr key_opt
+				#   key_opt ::= . | KEY expr
+				#   database_kw_opt ::= DATABASE | .
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_attach.html
+			#
+			attach_kw = require_one_of [:ATTACH, :DATABASE].freeze, :ATTACH
+			file = expression
+			as_kw = require :AS
+			schema = expression
+			if (key_kw = maybe :KEY)
+				key = expression
+			end
+
+			AttachStatement.concrete(
+				full_source: @lexer.sql,
+				attach_kw:   Token::Keyword(attach_kw),
+				file:        file,
+				as_kw:       Token::Keyword(as_kw),
+				schema:      schema,
+				key_kw:      Token::Keyword(key_kw),
+				key:         key,
+			)
+		end
+
+		def detach_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= DETACH database_kw_opt expr
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_detach.html
+			#
+			detach_kw = require_one_of [:DETACH, :DATABASE].freeze, :DETACH
+
+			DetachStatement.concrete(
+				full_source: @lexer.sql,
+				detach_kw:   Token::Keyword(detach_kw),
+				schema:      expression,
+			)
+		end
+
+		def analyze_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= ANALYZE | ANALYZE nm dbnm
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_analyze.html
+			#
+			analyze_kw = require :ANALYZE
+			target = table_name unless statement_end?
+
+			AnalyzeStatement.concrete(
+				full_source: @lexer.sql,
+				analyze_kw:  Token::Keyword(analyze_kw),
+				target:      target,
+			)
+		end
+
+		def reindex_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= REINDEX | REINDEX nm dbnm
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_reindex.html
+			#
+			reindex_kw = require :REINDEX
+			target = table_name unless statement_end?
+
+			ReindexStatement.concrete(
+				full_source: @lexer.sql,
+				reindex_kw:  Token::Keyword(reindex_kw),
+				target:      target,
+			)
+		end
+
+		def vacuum_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= VACUUM vinto | VACUUM nm vinto
+				#   vinto ::= INTO expr | .
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_vacuum.html
+			#
+			vacuum_kw = require :VACUUM
+			schema_tk = maybe { name } unless current_token.nil? || current_token in :SEMI | :INTO
+			if (into_kw = maybe :INTO)
+				into = expression
+			end
+
+			VacuumStatement.concrete(
+				full_source: @lexer.sql,
+				vacuum_kw:   Token::Keyword(vacuum_kw),
+				schema_tk:   Token::Identifier(schema_tk),
+				into_kw:     Token::Keyword(into_kw),
+				into:        into,
+			)
+		end
+
+		def create_index_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= createkw uniqueflag INDEX ifnotexists nm dbnm ON nm LP sortlist RP where_opt
+				#   uniqueflag ::= UNIQUE | .
+				#
+				# Syntax diagram:
+				#   ◯─▶{ CREATE }┬▶{ UNIQUE }┬▶{ INDEX }┬▶{ IF }─▶{ NOT }─▶{ EXISTS }┬┐
+				#                └─────▶─────┘          └───────────────▶────────────┘│
+				#   ┌──────────────────────────────────◀──────────────────────────────┘
+				#   └┬▶{ schema-name }─▶{ . }┬▶{ index-name }─▶{ ON }─▶{ table-name }─▶{ ( }┐
+				#    └──────────▶────────────┘                                              │
+				#   ┌───────────────────────────────────◀───────────────────────────────────┘
+				#   └┬▶[ indexed-column ]┬▶{ ) }┬────────────────────────┬─▶◯
+				#    └──────{ , }◀───────┘      └▶{ WHERE }─▶[ expr ]────┘
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_createindex.html
+				#   - https://www.sqlite.org/partialindex.html
+				#   - https://www.sqlite.org/expridx.html
+			#
+			create_kw = require :CREATE
+			unique_kw = maybe :UNIQUE
+			index_kw = require :INDEX
+			if_not_exists_kw = maybe_all_of :IF, :NOT, :EXISTS
+			index = table_name
+			on_kw = require :ON
+			table_tk = name
+			columns_lp = require :LP
+			columns = require_some(trailing_sep: :COMMA) { ordering_term }
+			columns_rp = require :RP
+			if (where_kw = maybe :WHERE)
+				condition = expression
+			end
+
+			CreateIndexStatement.concrete(
+				full_source:      @lexer.sql,
+				create_kw:        Token::Keyword(create_kw),
+				unique_kw:        Token::Keyword(unique_kw),
+				index_kw:         Token::Keyword(index_kw),
+				if_not_exists_kw: Token::Keyword(if_not_exists_kw),
+				index:            index,
+				on_kw:            Token::Keyword(on_kw),
+				table_tk:         Token::Identifier(table_tk),
+				columns_lp:       Token::Punctuation(columns_lp),
+				columns:          columns,
+				columns_rp:       Token::Punctuation(columns_rp),
+				where_kw:         Token::Keyword(where_kw),
+				condition:        condition,
+			)
+		end
+
+		def alter_table_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= ALTER TABLE fullname RENAME TO nm
+				#   cmd ::= ALTER TABLE add_column_fullname ADD kwcolumn_opt columnname carglist
+				#   cmd ::= ALTER TABLE fullname DROP kwcolumn_opt nm
+				#   cmd ::= ALTER TABLE fullname RENAME kwcolumn_opt nm TO nm
+				#   kwcolumn_opt ::= . | COLUMNKW
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_altertable.html
+			#
+			alter_table_kw = require_all_of :ALTER, :TABLE
+			table = table_name
+
+			if current_token == :RENAME && peek == :TO
+				action = :RENAME_TABLE
+				action_kw = require :RENAME
+				to_kw = require :TO
+				new_name_tk = name
+			elsif (action_kw = maybe_one_of [:RENAME, :COLUMN].freeze, :RENAME)
+				action = :RENAME_COLUMN
+				column_tk = name
+				to_kw = require :TO
+				new_name_tk = name
+			elsif (action_kw = maybe_one_of [:ADD, :COLUMN].freeze, :ADD)
+				action = :ADD_COLUMN
+				column_definition = column_def
+			elsif (action_kw = maybe_one_of [:DROP, :COLUMN].freeze, :DROP)
+				action = :DROP_COLUMN
+				column_tk = name
+			else
+				expected!(:RENAME, :ADD, :DROP)
+			end
+
+			AlterTableStatement.concrete(
+				full_source:       @lexer.sql,
+				alter_table_kw:    Token::Keyword(alter_table_kw),
+				table:             table,
+				action:            action,
+				action_kw:         Token::Keyword(action_kw),
+				column_tk:         Token::Identifier(column_tk),
+				to_kw:             Token::Keyword(to_kw),
+				new_name_tk:       Token::Identifier(new_name_tk),
+				column_definition: column_definition,
+			)
+		end
+
+		def create_virtual_table_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   create_vtab ::= createkw VIRTUAL TABLE ifnotexists nm dbnm USING nm
+				#   cmd ::= create_vtab | create_vtab LP vtabarglist RP
+				#   vtabarglist ::= vtabarg | vtabarglist COMMA vtabarg
+				#   vtabarg ::= . | vtabarg vtabargtoken
+				#   vtabargtoken ::= ANY | lp anylist RP
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_createvtab.html
+			#
+			create_virtual_table_kw = require_all_of :CREATE, :VIRTUAL, :TABLE
+			if_not_exists_kw = maybe_all_of :IF, :NOT, :EXISTS
+			table = table_name
+			using_kw = require :USING
+			module_tk = name
+			if (arguments_lp = maybe :LP)
+				arguments, arguments_rp = module_arguments
+			end
+
+			CreateVirtualTableStatement.concrete(
+				full_source:             @lexer.sql,
+				create_virtual_table_kw: Token::Keyword(create_virtual_table_kw),
+				if_not_exists_kw:        Token::Keyword(if_not_exists_kw),
+				table:                   table,
+				using_kw:                Token::Keyword(using_kw),
+				module_tk:               Token::Identifier(module_tk),
+				arguments_lp:            Token::Punctuation(arguments_lp),
+				arguments:               arguments,
+				arguments_rp:            Token::Punctuation(arguments_rp),
+			)
+		end
+
+		# Collects the text of each comma-separated argument up to the closing parenthesis,
+		# whatever tokens it is made of
+		def module_arguments
+			arguments = []
+			depth = 0
+			first = last = nil
+
+			loop do
+				expected!(:RP) if current_token.nil?
+
+				if depth.zero? && current_token in :COMMA | :RP
+					arguments << (first ? @lexer.sql.byteslice(first, last - first) : "")
+					first = last = nil
+					return [arguments, advance] if current_token == :RP
+
+					advance
+					next
+				end
+
+				depth += 1 if current_token == :LP
+				depth -= 1 if current_token == :RP
+				_tok, beg, fin = advance
+				first ||= beg
+				last = fin
+			end
+		end
+
+		def create_trigger_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= createkw trigger_decl BEGIN trigger_cmd_list END
+				#   trigger_decl ::= temp TRIGGER ifnotexists nm dbnm trigger_time trigger_event ON fullname foreach_clause when_clause
+				#   trigger_time ::= BEFORE | AFTER | INSTEAD OF | .
+				#   trigger_event ::= DELETE | INSERT | UPDATE | UPDATE OF idlist
+				#   foreach_clause ::= . | FOR EACH ROW
+				#   when_clause ::= . | WHEN expr
+				#   trigger_cmd_list ::= trigger_cmd_list trigger_cmd SEMI | trigger_cmd SEMI
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_createtrigger.html
+			#
+			create_kw = require :CREATE
+			temp_kw = maybe_one_of :TEMP, :TEMPORARY
+			trigger_kw = require :TRIGGER
+			if_not_exists_kw = maybe_all_of :IF, :NOT, :EXISTS
+			trigger = table_name
+			time_kw = maybe_one_of :BEFORE, :AFTER, [:INSTEAD, :OF].freeze
+			if (event_kw = maybe_all_of :UPDATE, :OF)
+				columns = require_some(trailing_sep: :COMMA) { column_reference }
+			else
+				event_kw = require_one_of :DELETE, :INSERT, :UPDATE
+			end
+			on_kw = require :ON
+			table = table_name
+			for_each_row_kw = maybe_all_of :FOR, :EACH, :ROW
+			if (when_kw = maybe :WHEN)
+				condition = expression
+			end
+			begin_kw = require :BEGIN
+			statements = require_some do
+				statement =
+					case current_token
+					when :UPDATE then update_stmt
+					when :INSERT, :REPLACE then insert_stmt
+					when :DELETE then delete_stmt
+					when :WITH then with_stmt
+					else select_stmt
+					end
+				require :SEMI
+				statement
+			end
+			end_kw = require :END
+
+			CreateTriggerStatement.concrete(
+				full_source:      @lexer.sql,
+				create_kw:        Token::Keyword(create_kw),
+				temp_kw:          Token::Keyword(temp_kw),
+				trigger_kw:       Token::Keyword(trigger_kw),
+				if_not_exists_kw: Token::Keyword(if_not_exists_kw),
+				trigger:          trigger,
+				time_kw:          Token::Keyword(time_kw),
+				event_kw:         Token::Keyword(event_kw),
+				columns:          columns,
+				on_kw:            Token::Keyword(on_kw),
+				table:            table,
+				for_each_row_kw:  Token::Keyword(for_each_row_kw),
+				when_kw:          Token::Keyword(when_kw),
+				condition:        condition,
+				begin_kw:         Token::Keyword(begin_kw),
+				statements:       statements,
+				end_kw:           Token::Keyword(end_kw),
+			)
+		end
+
+		def create_table_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= create_table create_table_args
+				#   create_table ::= createkw temp TABLE ifnotexists nm dbnm
+				#   create_table_args ::= LP columnlist conslist_opt RP table_option_set | AS select
+				#   createkw ::= CREATE
+				#   temp ::= . | TEMP
+				#   ifnotexists ::= . | IF NOT EXISTS
+				#   columnlist ::= columnlist COMMA columnname carglist
+				#   nm ::= idj | STRING
+				#   dbnm ::= . | DOT nm
+				#
+				# Syntax diagram:
+				#   ◯─▶{ CREATE }┬───────▶──────┬▶{ TABLE }┬▶{ IF }─▶{ NOT }─▶{ EXISTS }─┐
+				#                ├▶{ TEMP }─────┤          │                             │
+				#                └▶{ TEMPORARY }┘          │                             │
+				#   ┌─────────────────────◀────────────────┴──────────────────◀──────────┘
+				#   ├─▶{ schema-name }─▶{ . }─┬▶{ table-name }┬───────▶{ AS }─▶[ select-stmt ]───▶─────┐
+				#   └────────▶────────────────┘               │                                        │
+				#   ┌────────────────────◀────────────────────┘                     ┌────────▶─────────┼─▶◯
+				#   └▶{ ( }─┬▶[ column-def ]─┬▶┬─────────────────────────────┬▶{ ) }┴▶[ table-options ]┘
+				#           └─────{ , }◀─────┘ └[ table-constraint ]◀─{ , }◀─┘
+				#
+				# Simplified grammar:
+				#   CREATE [TEMP] TABLE [IF NOT EXISTS] name [DOT name] (
+				#     LP column_def [, column_def]* [table_constraint [, table_constraint]*] RP [table_options]
+				#     | AS select_stmt
+				#   )
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_createtable.html
+				#
+				# A `create_table_stmt` creates a table.
+			#
+			create_kw         = require      :CREATE
+			temp_kw           = maybe_one_of :TEMP, :TEMPORARY
+			table_kw          = require      :TABLE
+			if_not_exists_kw  = maybe_all_of :IF, :NOT, :EXISTS
+			table_name        =              table_name()
+
+			if maybe :AS
+				as_select = select_stmt
+
+				CreateTableStatement.concrete(
+					full_source:      @lexer.sql,
+					create_kw:        Token::Keyword(create_kw),
+					temp_kw:          Token::Keyword(temp_kw),
+					table_kw:         Token::Keyword(table_kw),
+					if_not_exists_kw: Token::Keyword(if_not_exists_kw),
+					table:            table_name,
+					select_statement: as_select
+				)
+			elsif (columns_lp = maybe :LP)
+				columns         = require_some(trailing_sep: :COMMA) { column_def }
+				# inlined `table_constraint`
+				constraints     = maybe_some(trailing_sep: [:COMMA, nil].freeze) do
+					#
+						# Relevant `parse.y` grammar rules:
+						#   conslist_opt ::= . | COMMA conslist.
+						#   conslist ::= conslist tconscomma tcons | tcons
+						#   tconscomma ::= . | COMMA
+						#   tcons ::= CONSTRAINT nm |
+						#             PRIMARY KEY LP sortlist autoinc RP onconf |
+						#             UNIQUE LP sortlist RP onconf |
+						#             CHECK LP expr RP onconf |
+						#             FOREIGN KEY LP eidlist RP REFERENCES nm eidlist_opt refargs defer_subclause_opt
+						#
+						# Syntax diagram:
+						#   ◯─▶┬▶{ CONSTRAINT }─▶{ name }─┐
+						#      ├─────────────◀────────────┘
+						#      ├─▶{ PRIMARY }─▶{ KEY }───▶{ ( }┬▶[ indexed-column ]┬┬▶{ AUTOINCREMENT }┬▶{ ) }─▶[ conflict-clause ]───┬─▶◯
+						#      │                               └───────{ , }◀──────┘└────────▶─────────┘                              │
+						#      ├─▶{ UNIQUE }──▶{ ( }┬▶[ indexed-column ]┬▶{ ) }─▶[ conflict-clause ]────────────────────────────────▶─┤
+						#      │                    └───────{ , }◀──────┘                                                             │
+						#      ├─▶{ CHECK }─▶{ ( }─▶[ expr ]─▶{ ) }─────────────────────────────────────────────────────────────────▶─┤
+						#      └─▶{ FOREIGN }─▶{ KEY }─▶{ ( }┬▶[ indexed-column ]┬▶{ ) }─▶[ foreign-key-clause ]────────────────────▶─┘
+						#                                    └──────{ , }◀───────┘
+						#
+						# Simplified grammar:
+						#
+						# Relevant SQLite documentation:
+						#   - https://www.sqlite.org/syntax/table-constraint.html
+						#
+						# A set of SQL constraints for each table.
+						# SQLite supports UNIQUE, NOT NULL, CHECK and FOREIGN KEY constraints.
+					#
+					if (constraint_kw     = maybe :CONSTRAINT)
+						constraint_name     = name
+					end
+
+					if    (unique_kw      = maybe          :UNIQUE)
+						unique_lp           = require :LP
+						unique_columns      = require_some(trailing_sep: :COMMA) { indexed_column }
+						unique_rp           = require        :RP
+						on_conflict         = maybe        { conflict_clause }
+
+						UniqueTableConstraint.concrete(
+							full_source:     @lexer.sql,
+							constraint_kw:   Token::Keyword(constraint_kw),
+							name_tk:         Token::Identifier(constraint_name),
+							unique_kw:       Token::Keyword(unique_kw),
+							unique_lp:       Token::Punctuation(unique_lp),
+							columns:         unique_columns,
+							unique_rp:       Token::Punctuation(unique_rp),
+							conflict_clause: on_conflict,
+						)
+					elsif (check_kw       = maybe          :CHECK)
+						check_lp            = require        :LP
+						check               =                expression
+						check_rp            = require        :RP
+
+						CheckTableConstraint.concrete(
+							full_source:     @lexer.sql,
+							constraint_kw:   Token::Keyword(constraint_kw),
+							name_tk:         Token::Identifier(constraint_name),
+							check_kw:        Token::Keyword(check_kw),
+							check_lp:        Token::Punctuation(check_lp),
+							expression:      check,
+							check_rp:        Token::Punctuation(check_rp),
+						)
+					elsif (primary_key_kw = maybe_all_of   :PRIMARY, :KEY)
+						primary_key_lp      = require :LP
+						primary_key_columns = require_some(trailing_sep: :COMMA) { indexed_column }
+						autoincrement_kw    = maybe          :AUTOINCREMENT
+						primary_key_rp      = require        :RP
+						on_conflict         = maybe        { conflict_clause }
+
+						PrimaryKeyTableConstraint.concrete(
+							full_source:      @lexer.sql,
+							constraint_kw:    Token::Keyword(constraint_kw),
+							name_tk:          Token::Identifier(constraint_name),
+							primary_key_kw:   Token::Keyword(primary_key_kw),
+							primary_key_lp:   Token::Punctuation(primary_key_lp),
+							columns:          primary_key_columns,
+							autoincrement_kw: Token::Keyword(autoincrement_kw),
+							primary_key_rp:   Token::Punctuation(primary_key_rp),
+							conflict_clause:  on_conflict,
+						)
+					elsif (foreign_key_kw = maybe_all_of   :FOREIGN, :KEY)
+						foreign_key_lp      = require :LP
+						foreign_key_columns = require_some(trailing_sep: :COMMA) { indexed_column }
+						foreign_key_rp      = require        :RP
+						clause              =                foreign_key_clause
+
+						ForeignKeyTableConstraint.concrete(
+							full_source:        @lexer.sql,
+							constraint_kw:      Token::Keyword(constraint_kw),
+							name_tk:            Token::Identifier(constraint_name),
+							foreign_key_kw:     Token::Keyword(foreign_key_kw),
+							foreign_key_lp:     Token::Punctuation(foreign_key_lp),
+							columns:            foreign_key_columns,
+							foreign_key_rp:     Token::Punctuation(foreign_key_rp),
+							foreign_key_clause: clause,
+						)
+					else
+						if constraint_name
+							NoOpTableConstraint.concrete(
+								full_source:   @lexer.sql,
+								constraint_kw: Token::Keyword(constraint_kw),
+								name_tk:       Token::Identifier(constraint_name),
+							)
+						else
+							expected!(:PRIMARY, :UNIQUE, :CHECK, :FOREIGN)
+						end
+					end
+				end
+				columns_rp      = require :RP
+				# inlined `table_options`
+				options         = maybe_some(trailing_sep: :COMMA) do
+					#
+						# Relevant `parse.y` grammar rules:
+						#   table_option_set ::= . | table_option | table_option_set COMMA table_option
+						#   table_option ::= WITHOUT nm  // nm must be "ROWID"
+						#   table_option ::= nm          // nm must be "STRICT"
+						#
+						# Syntax diagram:
+						#       ┌─▶{ WITHOUT }─▶{ ROWID }─┐
+						#   ◯─┬─┤                         ├─┬─▶◯
+						#     │ └─▶{ STRICT }─────────────┘ │
+						#     └──────◀───────────────{ , }◀─┘
+						#
+						# Simplified grammar:
+						#   [WITHOUT ROWID | STRICT] [, [WITHOUT ROWID | STRICT]]*
+						#
+						# Relevant SQLite documentation:
+						#   - https://www.sqlite.org/syntax/table-options.html
+						#   - https://www.sqlite.org/withoutrowid.html
+						#   - https://www.sqlite.org/stricttables.html
+						#
+						# Whether the table is a WITHOUT ROWID table.
+						# Whether the table is subject to STRICT type checking.
+					#
+					if maybe(:STRICT)
+						StrictTableOption.new
+					elsif maybe_all_of(:WITHOUT, :ROWID)
+						WithoutRowidTableOption.new
+					else
+						expected!(:STRICT, :WITHOUT)
+					end
+				end
+
+				CreateTableStatement.concrete(
+					full_source:      @lexer.sql,
+					create_kw:        Token::Keyword(create_kw),
+					temp_kw:          Token::Keyword(temp_kw),
+					table_kw:         Token::Keyword(table_kw),
+					if_not_exists_kw: Token::Keyword(if_not_exists_kw),
+					table:            table_name,
+					columns_lp:       Token::Punctuation(columns_lp),
+					columns:          columns,
+					columns_rp:       Token::Punctuation(columns_rp),
+					options:          options,
+					constraints:      constraints,
+				)
+			else
+				expected!(:AS, :LP)
+			end
+		end
+
+		def create_view_stmt
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= createkw temp VIEW ifnotexists nm dbnm eidlist_opt AS select
+				#
+				# Syntax diagram:
+				#   ◯─▶{ CREATE }┬───────▶──────┬▶{ VIEW }┬▶{ IF }─▶{ NOT }─▶{ EXISTS }─┐
+				#                ├▶{ TEMP }─────┤         │                             │
+				#                └▶{ TEMPORARY }┘         │                             │
+				#   ┌─────────────────────◀───────────────┴──────────────────◀──────────┘
+				#   ├─▶{ schema-name }─▶{ . }─┬▶{ view-name }┬─────────────────────────────────┬▶{ AS }─▶[ select-stmt ]─▶◯
+				#   └────────▶────────────────┘              └▶{ ( }┬▶{ column-name }┬▶{ ) }─┘
+				#                                                   └──────{ , }◀────┘
+				#
+				# Simplified grammar:
+				#   CREATE [TEMP] VIEW [IF NOT EXISTS] [schema .] name [(column [, column]*)] AS select_stmt
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_createview.html
+				#
+				# A view is a named select statement that can be queried like a table.
+			#
+			create_kw        = require      :CREATE
+			temp_kw          = maybe_one_of :TEMP, :TEMPORARY
+			view_kw          = require      :VIEW
+			if_not_exists_kw = maybe_all_of :IF, :NOT, :EXISTS
+			view             = table_name
+			if (columns_lp = maybe :LP)
+				columns    = require_some(trailing_sep: :COMMA) { indexed_column }
+				columns_rp = require :RP
+			end
+			as_kw            = require :AS
+			select           = select_stmt
+
+			CreateViewStatement.concrete(
+				full_source:      @lexer.sql,
+				create_kw:        Token::Keyword(create_kw),
+				temp_kw:          Token::Keyword(temp_kw),
+				view_kw:          Token::Keyword(view_kw),
+				if_not_exists_kw: Token::Keyword(if_not_exists_kw),
+				view:             view,
+				columns_lp:       Token::Punctuation(columns_lp),
+				columns:          columns,
+				columns_rp:       Token::Punctuation(columns_rp),
+				as_kw:            Token::Keyword(as_kw),
+				select:           select,
+			)
+		end
+
+		def insert_stmt(with = nil)
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= with insert_cmd INTO xfullname idlist_opt select upsert
+				#   cmd ::= with insert_cmd INTO xfullname idlist_opt DEFAULT VALUES returning
+				#   insert_cmd ::= INSERT orconf | REPLACE
+				#   orconf ::= . | OR resolvetype
+				#   resolvetype ::= raisetype | IGNORE | REPLACE
+				#   idlist_opt ::= . | LP idlist RP
+				#   upsert ::= . | RETURNING selcollist |
+				#              ON CONFLICT LP sortlist RP where_opt DO UPDATE SET setlist where_opt upsert |
+				#              ON CONFLICT LP sortlist RP where_opt DO NOTHING upsert |
+				#              ON CONFLICT DO NOTHING returning |
+				#              ON CONFLICT DO UPDATE SET setlist where_opt returning
+				#
+				# Syntax diagram:
+				#   ◯─▶┬▶[ with-clause ]─┐
+				#      ├───────◀─────────┘
+				#      ├─▶{ REPLACE }─────────────────────────┐
+				#      └─▶{ INSERT }┬────────────────────────┬┴▶{ INTO }─▶[ qualified-table-name ]┐
+				#                   └▶{ OR }─▶{ resolution }─┘                                    │
+				#   ┌────────────────────────────────◀────────────────────────────────────────────┘
+				#   ├─▶{ ( }┬▶{ column-name }┬▶{ ) }┐
+				#   │       └──────{ , }◀────┘      │
+				#   ├───────────────◀───────────────┘
+				#   ├─▶[ select-stmt ]─┬───────────────────┬───────────┬─▶[ returning-clause ]─▶◯
+				#   │                  └▶[ upsert-clause ]─┘           │
+				#   └─▶{ DEFAULT }─▶{ VALUES }─────────────────────────┘
+				#
+				# Simplified grammar:
+				#   [with_clause] (INSERT [OR resolution] | REPLACE) INTO table [AS alias] [(column [, column]*)]
+				#   (select_stmt [upsert_clause]* | DEFAULT VALUES) [returning_clause]
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_insert.html
+				#   - https://www.sqlite.org/lang_upsert.html
+				#
+				# Adds rows to a table, from a VALUES list, a select, or the column defaults.
+			#
+			insert_kw = require_one_of(
+				[:INSERT, :OR, :ROLLBACK].freeze, [:INSERT, :OR, :ABORT].freeze, [:INSERT, :OR, :FAIL].freeze,
+				[:INSERT, :OR, :IGNORE].freeze, [:INSERT, :OR, :REPLACE].freeze, :INSERT, :REPLACE,
+			)
+			into_kw = require :INTO
+			table = qualified_table_name(indexed: false)
+			if (columns_lp = maybe :LP)
+				columns = require_some(trailing_sep: :COMMA) { column_reference }
+				columns_rp = require :RP
+			end
+
+			unless (default_values_kw = maybe_all_of :DEFAULT, :VALUES)
+				select = select_stmt
+				upserts = maybe_some { upsert_clause }
+			end
+			returning = returning_clause
+
+			InsertStatement.concrete(
+				full_source:       @lexer.sql,
+				with_clause:       with,
+				insert_kw:         Token::Keyword(insert_kw),
+				into_kw:           Token::Keyword(into_kw),
+				table:             table,
+				columns_lp:        Token::Punctuation(columns_lp),
+				columns:           columns,
+				columns_rp:        Token::Punctuation(columns_rp),
+				default_values_kw: Token::Keyword(default_values_kw),
+				select:            select,
+				upserts:           upserts,
+				returning:         returning,
+			)
+		end
+
+		def update_stmt(with = nil)
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= with UPDATE orconf xfullname indexed_opt SET setlist from where_opt_ret orderby_opt limit_opt
+				#   where_opt_ret ::= . | WHERE expr | RETURNING selcollist | WHERE expr RETURNING selcollist
+				#
+				# Syntax diagram:
+				#   ◯─▶┬▶[ with-clause ]─┐
+				#      ├───────◀─────────┘
+				#      └─▶{ UPDATE }┬────────────────────────┬▶[ qualified-table-name ]─▶{ SET }─┐
+				#                   └▶{ OR }─▶{ resolution }─┘                                   │
+				#   ┌─────────────────────────────────◀──────────────────────────────────────────┘
+				#   └┬▶{ column-name }──────────────────┬▶{ = }─▶[ expr ]┬┬───────────────────────────┬┐
+				#    ├▶{ ( }┬▶{ column-name }┬▶{ ) }────┘                │└▶{ FROM }─▶[ join-clause ]─┘│
+				#    │      └──────{ , }◀────┘                           │                             │
+				#    └───────────────────────{ , }◀──────────────────────┘                             │
+				#   ┌──────────────────────────────────────◀───────────────────────────────────────────┘
+				#   └┬▶{ WHERE }─▶[ expr ]┬┬─▶[ returning-clause ]┬─▶[ ORDER BY … ]─▶[ LIMIT … ]─▶◯
+				#    └──────────▶─────────┘└──────────▶───────────┘
+				#
+				# Simplified grammar:
+				#   [with_clause] UPDATE [OR resolution] table [AS alias] [INDEXED BY index | NOT INDEXED]
+				#   SET assignment [, assignment]* [FROM join_clause] [WHERE expr] [returning_clause]
+				#   [ORDER BY ordering_term [, ordering_term]*] [LIMIT expr [(OFFSET | COMMA) expr]]
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_update.html
+				#
+				# Changes columns of existing rows.
+			#
+			update_kw = require_one_of(
+				[:UPDATE, :OR, :ROLLBACK].freeze, [:UPDATE, :OR, :ABORT].freeze, [:UPDATE, :OR, :FAIL].freeze,
+				[:UPDATE, :OR, :IGNORE].freeze, [:UPDATE, :OR, :REPLACE].freeze, :UPDATE,
+			)
+			table = qualified_table_name(indexed: true)
+			set_kw = require :SET
+			assignments = set_list
+			if (from_kw = maybe :FROM)
+				source = join_clause
+			end
+			if (where_kw = maybe :WHERE)
+				condition = expression
+			end
+			returning = returning_clause
+			tail = order_and_limit
+
+			UpdateStatement.concrete(
+				full_source: @lexer.sql,
+				with_clause: with,
+				update_kw:   Token::Keyword(update_kw),
+				table:       table,
+				set_kw:      Token::Keyword(set_kw),
+				assignments: assignments,
+				from_kw:     Token::Keyword(from_kw),
+				source:      source,
+				where_kw:    Token::Keyword(where_kw),
+				condition:   condition,
+				returning:   returning,
+				**tail,
+			)
+		end
+
+		def delete_stmt(with = nil)
+			#
+				# Relevant `parse.y` grammar rules:
+				#   cmd ::= with DELETE FROM xfullname indexed_opt where_opt_ret orderby_opt limit_opt
+				#
+				# Syntax diagram:
+				#   ◯─▶┬▶[ with-clause ]─┐
+				#      ├───────◀─────────┘
+				#      └─▶{ DELETE }─▶{ FROM }─▶[ qualified-table-name ]┬▶{ WHERE }─▶[ expr ]┬┐
+				#                                                       └──────────▶─────────┘│
+				#   ┌──────────────────────────────────────◀──────────────────────────────────┘
+				#   └┬─▶[ returning-clause ]┬─▶[ ORDER BY … ]─▶[ LIMIT … ]─▶◯
+				#    └──────────▶───────────┘
+				#
+				# Simplified grammar:
+				#   [with_clause] DELETE FROM table [AS alias] [INDEXED BY index | NOT INDEXED]
+				#   [WHERE expr] [returning_clause] [ORDER BY …] [LIMIT …]
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_delete.html
+				#
+				# Removes rows from a table.
+			#
+			delete_kw = require :DELETE
+			from_kw = require :FROM
+			table = qualified_table_name(indexed: true)
+			if (where_kw = maybe :WHERE)
+				condition = expression
+			end
+			returning = returning_clause
+			tail = order_and_limit
+
+			DeleteStatement.concrete(
+				full_source: @lexer.sql,
+				with_clause: with,
+				delete_kw:   Token::Keyword(delete_kw),
+				from_kw:     Token::Keyword(from_kw),
+				table:       table,
+				where_kw:    Token::Keyword(where_kw),
+				condition:   condition,
+				returning:   returning,
+				**tail,
+			)
+		end
+
+		# A WITH clause can lead a SELECT, INSERT, UPDATE or DELETE - which one is only
+		# known once the clause is behind us
+		def with_stmt
+			with = with_clause
+
+			case current_token
+			when :SELECT, :VALUES   then select_stmt(with)
+			when :INSERT, :REPLACE  then insert_stmt(with)
+			when :UPDATE            then update_stmt(with)
+			when :DELETE            then delete_stmt(with)
+			else
+				expected!(:SELECT, :VALUES, :INSERT, :REPLACE, :UPDATE, :DELETE)
+			end
+		end
+
+		def select_stmt(with = nil)
+			#
+				# Relevant `parse.y` grammar rules:
+				#   select ::= WITH wqlist selectnowith |
+				#              WITH RECURSIVE wqlist selectnowith |
+				#              selectnowith
+				#   selectnowith ::= oneselect |
+				#                    selectnowith multiselect_op oneselect
+				#   multiselect_op ::= UNION | UNION ALL | EXCEPT | INTERSECT
+				#   orderby_opt ::= . | ORDER BY sortlist
+				#   limit_opt ::= . | LIMIT expr | LIMIT expr OFFSET expr | LIMIT expr COMMA expr
+				#
+				# Syntax diagram:
+				#   ◯─▶┬▶[ with-stmt ]─┐
+				#      ├───────◀───────┘
+				#   ┌──┴──────────────▶[ select-core ]───────────────────┐
+				#   └──────────◀────[ compound-operator ]◀───────────────┤
+				#    ┌──────────────────◀────────────────────────────────┘
+				#    ├─▶{ ORDER }─▶{ BY }┬▶[ ordering-term ]─┬┐
+				#    ▼                   └───────{ , }◀──────┘│
+				#    ├────────────────◀───────────────────────┘
+				#    ├─▶{ LIMIT }─▶[ expr ]─▶┬───────────────────────▶─┐
+				#    │                       ├─▶{ OFFSET }─▶[ expr ]─▶─┤
+				#    │                       └─▶{ , }─▶[ expr ]──────▶─┤
+				#    └───────────────────────────────────────────────▶─┴───────▶◯
+				#
+				# Simplified grammar:
+				#   [with_clause] select_core [compound_operator select_core]*
+				#   [ORDER BY ordering_term [, ordering_term]*]
+				#   [LIMIT expr [(OFFSET | COMMA) expr]]
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_select.html
+				#
+				# The SELECT statement is used to query the database.
+			#
+			with ||= with_clause if current_token == :WITH
+			body = select_core
+
+			while (operator_tk = maybe_one_of [:UNION, :ALL].freeze, :UNION, :INTERSECT, :EXCEPT)
+				body = CompoundSelect.concrete(
+					full_source: @lexer.sql,
+					operator_tk: Token::Keyword(operator_tk),
+					left:        body,
+					right:       select_core,
+				)
+			end
+
+			SelectStatement.concrete(
+				full_source: @lexer.sql,
+				with_clause: with,
+				body:        body,
+				**order_and_limit,
+			)
+		end
+
+		def with_clause
+			#
+				# Relevant `parse.y` grammar rules:
+				#   with ::= WITH wqlist | WITH RECURSIVE wqlist
+				#   wqlist ::= wqitem | wqlist COMMA wqitem
+				#   wqitem ::= withnm eidlist_opt wqas LP select RP
+				#   wqas ::= AS | AS MATERIALIZED | AS NOT MATERIALIZED
+				#   eidlist_opt ::= . | LP eidlist RP
+				#
+				# Syntax diagram:
+				#   ◯─▶{ WITH }─┬──────────────────┬▶[ common-table-expression ]┬─▶◯
+				#               └─▶{ RECURSIVE }─▶─┘└───────────{ , }◀──────────┘
+				#
+				#   ◯─▶{ table-name }┬────────────────────────────────────┬▶{ AS }┐
+				#                    └▶{ ( }┬▶{ column-name }┬▶{ ) }──────┘       │
+				#                           └──────{ , }◀────┘                    │
+				#   ┌──────────────────────────────◀──────────────────────────────┘
+				#   ├────────────────▶────────────────┬▶{ ( }─▶[ select-stmt ]─▶{ ) }─▶◯
+				#   ├─▶{ NOT }─┬─▶{ MATERIALIZED }──▶─┘
+				#   └────▶─────┘
+				#
+				# Simplified grammar:
+				#   WITH [RECURSIVE] name [(column [, column]*)] AS [[NOT] MATERIALIZED] (select_stmt)
+				#                    [, name [(column [, column]*)] AS [[NOT] MATERIALIZED] (select_stmt)]*
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_with.html
+				#
+				# Common table expressions act like temporary views that exist only for a single statement.
+			#
+			with_kw = require :WITH
+			recursive_kw = maybe :RECURSIVE
+
+			ctes = require_some(trailing_sep: :COMMA) do
+				name_tk = name
+				if (columns_lp = maybe :LP)
+					columns = require_some(trailing_sep: :COMMA) { indexed_column }
+					columns_rp = require :RP
+				end
+				as_kw = require :AS
+				materialized_tk = maybe_one_of :MATERIALIZED, [:NOT, :MATERIALIZED].freeze
+				select_lp = require :LP
+				select = select_stmt
+				select_rp = require :RP
+
+				CommonTableExpression.concrete(
+					full_source:     @lexer.sql,
+					name_tk:         Token::Identifier(name_tk),
+					columns_lp:      Token::Punctuation(columns_lp),
+					columns:         columns,
+					columns_rp:      Token::Punctuation(columns_rp),
+					as_kw:           Token::Keyword(as_kw),
+					materialized_tk: Token::Keyword(materialized_tk),
+					select_lp:       Token::Punctuation(select_lp),
+					select:          select,
+					select_rp:       Token::Punctuation(select_rp),
+				)
+			end
+
+			WithClause.concrete(
+				full_source:  @lexer.sql,
+				with_kw:      Token::Keyword(with_kw),
+				recursive_kw: Token::Keyword(recursive_kw),
+				ctes:         ctes,
+			)
+		end
+
+		# ---------- Clauses ----------
+
+		def expression(min_precedence = 0)
+			#
+				# Relevant `parse.y` grammar rules:
+				#
+				# Syntax diagram:
+				#   ◯┬─▶[ literal-value ]───────────────────────────────────────────────────────────────────────────────────┬─▶◯
+				#    ├─▶{ bind-parameter }────────────────────────────────────────────────────────────────────────────────▶─┤
+				#    ├────────────────────────┬──────────────────────┐                                                      │
+				#    ├─▶{ schema-name }─▶{ . }┴▶{ table-name }─▶{ . }┴▶{ column-name }────────────────────────────────────▶─┤
+				#    ├─▶{ unary-operator }─▶[ expr ]──────────────────────────────────────────────────────────────────────▶─┤
+				#    ├─▶[ expr ]─▶{ binary-operator }─▶[ expr ]───────────────────────────────────────────────────────────▶─┤
+				#    ├─▶{ function-name }─▶{ ( }─▶[ function-arguments ]─▶{ ) }┬───────────────────┬┬─────────────────────▶─┤
+				#    │       ┌──{ , }◀─┐                                       └▶[ filter-clause ]─┘└▶[ over-clause ]─────▶─┤
+				#    ├─▶{ ( }┴▶[ expr ]┴▶{ ) }────────────────────────────────────────────────────────────────────────────▶─┤
+				#    ├─▶{ CAST }─▶{ ( }─▶[ expr ]─▶{ AS }─▶[ type-name ]─▶{ ) }───────────────────────────────────────────▶─┤
+				#    ├─▶[ expr ]─▶{ COLLATE }─▶{ collation-name }─────────────────────────────────────────────────────────▶─┤
+				#    ├─▶[ expr ]┬▶{ NOT }┬┬▶{ LIKE }─────▶[ expr ]─┬────▶────────────────┬▶┬──────────────────────────────▶─┤
+				#    │          └────▶───┘├▶{ GLOB }───┐           └▶{ ESCAPE }─▶[ expr ]┘ │                                │
+				#    │                    ├▶{ REGEXP }─┼─▶[ expr ]────────────▶────────────┘                                │
+				#    │                    └▶{ MATCH }──┘                                                                    │
+				#    ├─▶[ expr ]┬▶{ ISNULL }───────┬──────────────────────────────────────────────────────────────────────▶─┤
+				#    │          ├▶{ NOTNULL }──────┤                                                                        │
+				#    │          └▶{ NOT }─▶{ NULL }┘                                                                        │
+				#    ├─▶[ expr ]─▶{ IS }┬▶{ NOT }┬┬▶{ DISTINCT }─▶{ FROM }┬▶[ expr ]──────────────────────────────────────▶─┤
+				#    │                  └────▶───┘└────────────▶──────────┘                                                 │
+				#    ├─▶[ expr ]┬▶{ NOT }┬▶{ BETWEEN }─▶[ expr ]─▶{ AND }─▶[ expr ]───────────────────────────────────────▶─┤
+				#    │          └────▶───┘               ┌─────────────────┐                                                │
+				#    ├─▶[ expr ]┬▶{ NOT }┬▶{ IN }┬▶{ ( }─┼▶[ select-stmt ]─┼▶{ ) }────────────────────────────────────────▶─┤
+				#    │          └────▶───┘       │       └─┬──▶[ expr ]──┬─┘                                                │
+				#    │                           │         └────{ , }◀───┘ ┌▶{ table-name }───────────────────────────────▶─┤
+				#    │                           ├────────────────────────┐│                           ┌──{ , }──┐          │
+				#    │                           └─▶{ schema-name }─▶{ . }┴┴▶{ table-function }─▶{ ( }┬┴▶[ expr ]┴┬▶{ ) }─▶─┤
+				#    ├────▶────┬───────────┐                                                          └─────▶─────┘         │
+				#    ├─▶{ NOT }┴▶{ EXISTS }┴▶{ ( }─▶[ select-stmt ]─▶{ ) }────────────────────────────────────────────────▶─┤
+				#    ├─▶{ CASE }┬▶[ expr ]┬┬▶{ WHEN }─▶[ expr ]─▶{ THEN }─▶[ expr ]┬┬▶{ ELSE }─▶[ expr ]┬▶{ END }─────────▶─┤
+				#    │          └─────▶───┘└────────────────◀──────────────────────┘└─────────▶─────────┘                   │
+				#    └─▶[ raise-function ]────────────────────────────────────────────────────────────────────────────────▶─┘
+				#
+				# Simplified grammar:
+				#
+				# Relevant SQLite documentation:
+				#   -
+				#
+				# [description]
+			#
+
+			left = basic_expression
+
+			while (operator = current_token)
+				operator_precedence = OPERATOR_PRECEDENCE[operator] || -1
+				break if operator_precedence < min_precedence
+				next_precedence = operator_precedence + 1
+
+				negated = maybe :NOT
+
+				if (in_tk = maybe :IN)
+					if (value_lp = maybe :LP)
+						if (value_rp = maybe :RP)
+							right = ParentheticalExpression.concrete(
+								full_source: @lexer.sql,
+								value_lp: Token::Punctuation(value_lp),
+								value: nil,
+								value_rp: Token::Punctuation(value_rp),
+							)
+						elsif subquery_start?
+							select = select_stmt
+							value_rp = require :RP
+
+							right = SubqueryExpression.concrete(
+								full_source: @lexer.sql,
+								select_lp: Token::Punctuation(value_lp),
+								select: select,
+								select_rp: Token::Punctuation(value_rp),
+							)
+						elsif (e = maybe { expression })
+							exprs = require_some(given: e, trailing_sep: :COMMA) do
+								expression
+							end
+							value_rp = require :RP
+
+							right = ParentheticalExpression.concrete(
+								full_source: @lexer.sql,
+								value_lp: Token::Punctuation(value_lp),
+								value: exprs,
+								value_rp: Token::Punctuation(value_rp),
+							)
+						else
+							expected!(:RP, "select-stmt", "expr")
+						end
+					elsif (ref = maybe { identifier })
+						if maybe :DOT
+							name = identifier
+
+							if (arguments_lp = maybe :LP)
+								arguments = maybe_some(trailing_sep: :COMMA) { expression }
+								arguments_rp = require :RP
+
+								right = FunctionReference.concrete(
+									full_source: @lexer.sql,
+									schema_tk: Token::Identifier(ref),
+									function_tk: Token::Identifier(name),
+									arguments_lp: Token::Punctuation(arguments_lp),
+									arguments: FunctionArguments.concrete(
+										full_source: @lexer.sql,
+										expressions: arguments
+									),
+									arguments_rp: Token::Punctuation(arguments_rp),
+								)
+							else
+								right = TableName.concrete(
+									full_source: @lexer.sql,
+									schema_tk: Token::Identifier(ref),
+									table_tk: Token::Identifier(name),
+								)
+							end
+						elsif (arguments_lp = maybe :LP)
+							arguments = maybe_some(trailing_sep: :COMMA) { expression }
+							arguments_rp = require :RP
+
+							right = FunctionReference.concrete(
+								full_source: @lexer.sql,
+								function_tk: Token::Identifier(ref),
+								arguments_lp: Token::Punctuation(arguments_lp),
+								arguments: FunctionArguments.concrete(
+									full_source: @lexer.sql,
+									expressions: arguments
+								),
+								arguments_rp: Token::Punctuation(arguments_rp),
+							)
+						else
+							table = ref
+
+							right = TableName.concrete(
+								full_source: @lexer.sql,
+								table_tk: Token::Identifier(table),
+							)
+						end
+					else
+						expected!(:LP, :ID)
+					end
+
+					left = BinaryExpression.concrete(
+						full_source: @lexer.sql,
+						operator_tk: Token::Punctuation(token_span_from(negated, in_tk)),
+						left: left,
+						right: right,
+					)
+				elsif (between_kw = maybe :BETWEEN)
+					middle = expression(next_precedence)
+					and_kw = require :AND
+					right = expression(next_precedence)
+					left = TernaryExpression.concrete(
+						full_source: @lexer.sql,
+						operator_tk: Token::Keyword(token_span_from(negated, between_kw)),
+						left: left,
+						middle: middle,
+						and_kw: Token::Keyword(and_kw),
+						right: right,
+					)
+				elsif (like_kw = maybe :LIKE)
+					expression_node = negated ? NotLikeExpression : LikeExpression
+					right = expression(next_precedence)
+					if (escape_kw = maybe :ESCAPE)
+						escape = expression(next_precedence)
+					end
+					left = expression_node.concrete(
+						full_source: @lexer.sql,
+						operator_tk: Token::Keyword(token_span_from(negated, like_kw)),
+						left: left,
+						right: right,
+						escape_kw: Token::Keyword(escape_kw),
+						escape: escape,
+					)
+				elsif (match_tk = maybe_one_of :GLOB, :REGEXP, :MATCH)
+					operator_tk = token_span_from(negated, match_tk)
+					right = expression(next_precedence)
+
+					left_attr = (Node === left) ? { left: left } : { left_tk: left }
+					right_attr = (Node === right) ? { right: right } : { right_tk: right }
+
+					left = BinaryExpression.concrete(
+						full_source: @lexer.sql,
+						operator_tk: Token::Punctuation(operator_tk),
+						**left_attr,
+						**right_attr,
+					)
+				end
+				next if in_tk || between_kw || like_kw || match_tk
+
+				if negated && (null_kw = maybe :NULL)
+					left = UnaryExpression.concrete(
+						full_source: @lexer.sql,
+						operator_tk: Token::Punctuation(token_span_from(negated, null_kw)),
+						operand: left,
+					)
+					next
+				end
+
+				expected!(:IN, :BETWEEN, :LIKE, :GLOB, :REGEXP, :MATCH, :NULL) if negated
+
+				if (operator_tk = maybe :ISNULL)
+					operand_attr = (Node === left) ? { operand: left } : { operand_tk: left }
+
+					left = UnaryExpression.concrete(
+						full_source: @lexer.sql,
+						operator_tk: Token::Punctuation(operator_tk),
+						**operand_attr,
+					)
+				elsif (operator_tk = maybe :NOTNULL)
+					operand_attr = (Node === left) ? { operand: left } : { operand_tk: left }
+
+					left = UnaryExpression.concrete(
+						full_source: @lexer.sql,
+						operator_tk: Token::Punctuation(operator_tk),
+						**operand_attr,
+					)
+				elsif (is_tk = maybe :IS)
+					negated_tk = maybe :NOT
+					distinct_from_tk = maybe_all_of :DISTINCT, :FROM
+
+					operator_tk = token_span_from(is_tk, negated_tk, distinct_from_tk)
+					right = expression(next_precedence)
+
+					left_attr = (Node === left) ? { left: left } : { left_tk: left }
+					right_attr = (Node === right) ? { right: right } : { right_tk: right }
+
+					left = BinaryExpression.concrete(
+						full_source: @lexer.sql,
+						operator_tk: Token::Punctuation(operator_tk),
+						**left_attr,
+						**right_attr,
+					)
+				elsif (collate_kw = maybe :COLLATE)
+					expression_attr = (Node === left) ? { expression: left } : { expression_tk: left }
+					sequence = identifier_or_string
+
+					left = CollationExpression.concrete(
+						full_source: @lexer.sql,
+						collate_kw: Token::Keyword(collate_kw),
+						**expression_attr,
+						sequence_tk: Token::Identifier(sequence),
+					)
+				else
+					operator_tk = require operator
+					right = expression(next_precedence)
+
+					left_attr = (Node === left) ? { left: left } : { left_tk: left }
+					right_attr = (Node === right) ? { right: right } : { right_tk: right }
+
+					left = BinaryExpression.concrete(
+						full_source: @lexer.sql,
+						operator_tk: Token::Punctuation(operator_tk),
+						**left_attr,
+						**right_attr,
+					)
+				end
+			end
+
+			left
+		end
+
+		def column_def
+			#
+				# Used by:
+				#   - alter_table_stmt
+				#   - create_table_stmt
+				#
+				# Relevant `parse.y` grammar rules:
+				#   columnlist ::= columnname carglist
+				#   columnname ::= nm typetoken
+				#   carglist ::= . | carglist ccons
+				#
+				# Syntax diagram:
+				#   ◯─▶{ column-name }─┬─▶[ type-name ]─┬▶┬─▶───────────────────▶──┬─▶◯
+				#                      └────────▶───────┘ └─[ column-constraint ]◀─┘
+				#
+				# Simplified grammar:
+				#   name [type] [constraint [constraint]*]
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_createtable.html#column_definitions
+				#   - https://www.sqlite.org/syntax/column-def.html
+				#
+				# Each column definition consists of the name of the column,
+				# optionally followed by the declared type of the column,
+				# then one or more optional column constraints.
+			#
+			column_name = name
+			column_type = maybe { type_name }
+			constraints = maybe_some(trailing_sep: nil) do
+				#
+					# Relevant `parse.y` grammar rules:
+					#   ccons ::= CONSTRAINT nm |
+					#             DEFAULT term |
+					#             DEFAULT LP expr RP |
+					#             DEFAULT PLUS term |
+					#             DEFAULT MINUS term |
+					#             DEFAULT id |
+					#             NULL onconf |
+					#             NOT NULL onconf |
+					#             PRIMARY KEY sortorder onconf autoinc |
+					#             UNIQUE onconf |
+					#             CHECK LP expr RP |
+					#             REFERENCES nm eidlist_opt refargs |
+					#             defer_subclause |
+					#             COLLATE ids |
+					#             GENERATED ALWAYS AS generated |
+					#             AS generated
+					#   term ::= NULL | FLOAT | BLOB | STRING | INTEGER | CTIME_KW | QNUMBER
+					#   id ::= ID | INDEXED
+					#   ids ::= ID | STRING
+					#   nm ::= idj | STRING
+					#   eidlist_opt ::= . | LP eidlist RP
+					#   eidlist ::= eidlist COMMA nm collate sortorder |
+					# 	            nm collate sortorder
+					#   onconf ::= . | ON CONFLICT resolvetype
+					#   sortorder ::= . | ASC | DESC
+					#   refargs ::= . | refargs refarg
+					#   refarg ::= MATCH nm |
+					# 	           ON INSERT refact |
+					# 	           ON DELETE refact |
+					# 		         ON UPDATE refact
+					#   generated ::= LP expr RP |
+					#                 LP expr RP ID
+					#   autoinc ::= . | AUTOINCR.
+					#
+					# Syntax diagram:
+					#   ◯─▶┬▶{ CONSTRAINT }─▶{ name }─┬───────────────────────────────▶───────────────────────┐
+					#      ├─────────────◀────────────┘                                                       │
+					#      ├─▶{ PRIMARY }─▶{ KEY }──┬─────▶──────┬─▶[ conflict-clause ]┬──────────▶───────────┼─▶◯
+					#      │                        ├─▶{ ASC }──▶┤                     └─▶{ AUTOINCREMENT }─▶─┤
+					#      ├─────▶───┐              └─▶{ DESC }─▶┘                                            │
+					#      ├─▶{ NOT }┴▶{ NULL }─▶[ conflict-clause ]────────────────────────────────────────▶─┤
+					#      ├─▶{ UNIQUE }─▶[ conflict-clause ]───────────────────────────────────────────────▶─┤
+					#      ├─▶{ CHECK }─▶{ ( }─▶[ expr ]─▶{ ) }─────────────────────────────────────────────▶─┤
+					#      ├─▶{ DEFAULT }─▶┬▶{ ( }─▶[ expr ]─▶{ ) }──┬──────────────────────────────────────▶─┤
+					#      │               ├─▶[ literal-value ]────▶─┤                                        │
+					#      │               └─▶[ signed-number ]────▶─┘                                        │
+					#      ├─▶{ COLLATE }─▶{ collation-name }───────────────────────────────────────────────▶─┤
+					#      ├─▶[ foreign-key-clause ]────────────────────────────────────────────────────────▶─┤
+					#      ├─▶{ GENERATED }─▶{ ALWAYS }┬▶{ AS }─▶{ ( }─▶[ expr ]─▶{ ) }┬────────────────────▶─┤
+					#      └──────────────▶────────────┘                               ├─▶{ STORED }────────▶─┤
+					#                                                                  └─▶{ VIRTUAL }───────▶─┘
+					#
+					# Simplified grammar:
+					#   [CONSTRAINT name] (
+					#     [PRIMARY KEY [ASC|DESC] [conflict_clause] [AUTOINCREMENT]]
+					#     [NOT NULL [conflict_clause]]
+					#     [NULL [conflict_clause]]
+					#     [UNIQUE [conflict_clause]]
+					#     [CHECK LP expression RP [conflict_clause]]
+					#     [COLLATE collation-name]
+					#     [foreign-key-clause]
+					#     [GENERATED ALWAYS AS (expr) STORED | VIRTUAL]
+					#   )
+					#
+					# Relevant SQLite documentation:
+					#   - https://www.sqlite.org/lang_createtable.html#column_definitions
+					#   - https://www.sqlite.org/syntax/column-constraint.html
+					#
+					# Included in the definition of "column constraints" for the purposes of the previous statement
+					# are the COLLATE and DEFAULT clauses, even though these are not really constraints in the sense
+					# that they do not restrict the data that the table may contain. The other constraints -
+					# NOT NULL, CHECK, UNIQUE, PRIMARY KEY and FOREIGN KEY constraints - impose restrictions on the table data.
+				#
+				if (constraint_kw = maybe :CONSTRAINT)
+					constraint_name = name
+				end
+
+				if    (primary_key_kw = maybe_all_of :PRIMARY, :KEY)
+					direction_tk        = maybe_one_of :ASC, :DESC
+					on_conflict         = maybe      { conflict_clause }
+					autoincrement_kw    = maybe        :AUTOINCREMENT
+
+					PrimaryKeyColumnConstraint.concrete(
+						full_source:      @lexer.sql,
+						constraint_kw:    Token::Keyword(constraint_kw),
+						name_tk:          Token::Identifier(constraint_name),
+						primary_key_kw:   Token::Keyword(primary_key_kw),
+						direction_tk:     Token::Keyword(direction_tk),
+						conflict_clause:  on_conflict,
+						autoincrement_kw: Token::Keyword(autoincrement_kw),
+					)
+				elsif (not_null_kw    = maybe_all_of :NOT, :NULL)
+					on_conflict         = maybe      { conflict_clause }
+
+					NotNullColumnConstraint.concrete(
+						full_source:     @lexer.sql,
+						constraint_kw:   Token::Keyword(constraint_kw),
+						name_tk:         Token::Identifier(constraint_name),
+						not_null_kw:     Token::Keyword(not_null_kw),
+						conflict_clause: on_conflict,
+					)
+				elsif (null_kw        = maybe        :NULL)
+					on_conflict         = maybe      { conflict_clause }
+
+					NullColumnConstraint.concrete(
+						full_source:     @lexer.sql,
+						constraint_kw:   Token::Keyword(constraint_kw),
+						name_tk:         Token::Identifier(constraint_name),
+						null_kw:         Token::Keyword(null_kw),
+						conflict_clause: on_conflict,
+					)
+				elsif (unique_kw      = maybe        :UNIQUE)
+					on_conflict         = maybe      { conflict_clause }
+
+					UniqueColumnConstraint.concrete(
+						full_source:     @lexer.sql,
+						constraint_kw:   Token::Keyword(constraint_kw),
+						name_tk:         Token::Identifier(constraint_name),
+						unique_kw:       Token::Keyword(unique_kw),
+						conflict_clause: on_conflict,
+					)
+				elsif (check_kw       = maybe        :CHECK)
+					check_lp            = require      :LP
+					check               =              expression
+					check_rp            = require      :RP
+
+					CheckColumnConstraint.concrete(
+						full_source:   @lexer.sql,
+						constraint_kw: Token::Keyword(constraint_kw),
+						name_tk:       Token::Identifier(constraint_name),
+						check_kw:      Token::Keyword(check_kw),
+						check_lp:      Token::Punctuation(check_lp),
+						expression:    check,
+						check_rp:      Token::Punctuation(check_rp),
+					)
+				elsif (default_kw     = maybe        :DEFAULT)
+					if (default_lp      = maybe        :LP)
+						value         = expression
+						default_rp    = require :RP
+
+						value_attr = (Node === value) ? { value: value } : { value_tk: value }
+
+						DefaultColumnConstraint.concrete(
+							full_source:   @lexer.sql,
+							constraint_kw: Token::Keyword(constraint_kw),
+							name_tk:       Token::Identifier(constraint_name),
+							default_kw:    Token::Keyword(default_kw),
+							default_lp:    Token::Punctuation(default_lp),
+							default_rp:    Token::Punctuation(default_rp),
+							**value_attr,
+						)
+					elsif (number       = maybe      { signed_number })
+						DefaultColumnConstraint.concrete(
+							full_source:   @lexer.sql,
+							constraint_kw: Token::Keyword(constraint_kw),
+							name_tk:       Token::Identifier(constraint_name),
+							default_kw:    Token::Keyword(default_kw),
+							value_tk:      Token::Numeric(number),
+						)
+					elsif (value        = maybe      { literal_value })
+						DefaultColumnConstraint.concrete(
+							full_source:   @lexer.sql,
+							constraint_kw: Token::Keyword(constraint_kw),
+							name_tk:       Token::Identifier(constraint_name),
+							default_kw:    Token::Keyword(default_kw),
+							value_tk:      value,
+						)
+					# A sign on something that is not a number, like `DEFAULT -'hello'`
+					elsif (current_token in :PLUS | :MINUS) && (value = maybe { basic_expression })
+						DefaultColumnConstraint.concrete(
+							full_source:   @lexer.sql,
+							constraint_kw: Token::Keyword(constraint_kw),
+							name_tk:       Token::Identifier(constraint_name),
+							default_kw:    Token::Keyword(default_kw),
+							value:         value,
+						)
+					elsif (id           = maybe      { identifier })
+						DefaultColumnConstraint.concrete(
+							full_source:   @lexer.sql,
+							constraint_kw: Token::Keyword(constraint_kw),
+							name_tk:       Token::Identifier(constraint_name),
+							default_kw:    Token::Keyword(default_kw),
+							value_tk:      Token::Identifier(id),
+						)
+					else
+						expected! :LP, "literal-value", "signed-number"
+					end
+				elsif (collate_kw     = maybe        :COLLATE)
+					collation = name
+
+					CollateColumnConstraint.concrete(
+						full_source:   @lexer.sql,
+						constraint_kw: Token::Keyword(constraint_kw),
+						name_tk:       Token::Identifier(constraint_name),
+						collate_kw:    Token::Keyword(collate_kw),
+						collation_tk:  Token::Identifier(collation),
+					)
+				elsif :REFERENCES == current_token
+					clause = foreign_key_clause
+
+					ForeignKeyColumnConstraint.concrete(
+						full_source:        @lexer.sql,
+						constraint_kw:      Token::Keyword(constraint_kw),
+						name_tk:            Token::Identifier(constraint_name),
+						foreign_key_clause: clause,
+					)
+				elsif (as_kw = maybe_all_of(:GENERATED, :ALWAYS, :AS) || maybe(:AS))
+					as_lp      = require :LP
+					default    = expression
+					as_rp      = require :RP
+					type_tk    = maybe_one_of :STORED, :VIRTUAL
+
+					GeneratedAsColumnConstraint.concrete(
+						full_source:        @lexer.sql,
+						constraint_kw:      Token::Keyword(constraint_kw),
+						name_tk:            Token::Identifier(constraint_name),
+						as_kw:              Token::Keyword(as_kw),
+						as_lp:              Token::Punctuation(as_lp),
+						expression:         default,
+						as_rp:              Token::Punctuation(as_rp),
+						type_tk:            Token::Keyword(type_tk),
+					)
+				else
+					if constraint_name
+						NoOpColumnConstraint.concrete(
+							full_source:   @lexer.sql,
+							constraint_kw: Token::Keyword(constraint_kw),
+							name_tk:       Token::Identifier(constraint_name),
+						)
+					else
+						expected! :PRIMARY, :NOT, :NULL, :UNIQUE, :CHECK, :DEFAULT, :COLLATE, :REFERENCES, :GENERATED
+					end
+				end
+			end
+
+			ColumnDefinition.concrete(
+				full_source: @lexer.sql,
+				name_tk:     Token::Identifier(column_name),
+				type:        column_type,
+				constraints: constraints,
+			)
+		end
+
+		def select_core
+			#
+				# Relevant `parse.y` grammar rules:
+				#   oneselect ::= SELECT distinct selcollist from where_opt groupby_opt having_opt orderby_opt limit_opt |
+				#                 SELECT distinct selcollist from where_opt groupby_opt having_opt window_clause orderby_opt limit_opt |
+				#                 values
+				#   distinct ::= DISTINCT | ALL | .
+				#   from ::= . | FROM seltablist
+				#   where_opt ::= . | WHERE expr
+				#   groupby_opt ::= . | GROUP BY nexprlist
+				#   having_opt ::= . | HAVING expr
+				#   window_clause ::= WINDOW windowdefn_list
+				#   windowdefn_list ::= windowdefn | windowdefn_list COMMA windowdefn
+				#   windowdefn ::= nm AS LP window RP
+				#   values ::= VALUES LP nexprlist RP | values COMMA LP nexprlist RP
+				#
+				# Syntax diagram:
+				#   ◯─┬───▶{ SELECT }─┬───────▶──────┬─┬▶[ result-column ]─┐
+				#     │               ├▶{ DISTINCT }─┤ └─────{ , }◀────────┤
+				#     │               └▶{ ALL }─────▶┘                     │
+				#     │  ┌──────────────────────────◀──────────────────────┘
+				#     │  ├─▶{ FROM }┬─┬▶[ table-or-subquery ]─┬─┬─▶───┐
+				#     │  │          │ └────────{ , }◀─────────┘ ▲     ▼
+				#     │  │          └──▶[ join-clause ]─────────┘     │
+				#     │  ├──────────────────────◀─────────────────────┘
+				#     │  ├─▶{ WHERE }─▶[ expr ]─┐
+				#     │  ├──────────◀───────────┘
+				#     │  ├────────────▶─────────────────┐
+				#     │  ├─▶{ GROUP }─▶{ BY }┬▶[ expr ]┬┴▶┬─▶{ HAVING }─▶[ expr ]─┬─┐
+				#     │  ▼                   └──{ , }◀─┘  └────────▶──────────────┘ ▼
+				#     │  ├─────────────────────────────◀────────────────────────────┘
+				#     │  ├─▶{ WINDOW }┬▶{ window-name }─▶{ AS }─▶[ window-defn ]─┐
+				#     │  │            └───────────────────{ , }◀─────────────────┤
+				#     │  └──────────────────────▶─────────────────────────────▶──┴─┬───▶◯
+				#     │                    ┌──{ , }◀─┐                             │
+				#     └─▶{ VALUES }┬▶{ ( }─┴▶[ expr ]┴─▶{ ) }─┬─────────────────▶──┘
+				#                  └──────────{ , }◀──────────┘
+				#
+				# Simplified grammar:
+				#   SELECT [DISTINCT | ALL] result_column [, result_column]*
+				#   [FROM join_clause] [WHERE expr] [GROUP BY expr [, expr]* [HAVING expr]]
+				#   [WINDOW name AS window_defn [, name AS window_defn]*]
+				#   | VALUES (expr [, expr]*) [, (expr [, expr]*)]*
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_select.html#select_core
+				#
+				# A single SELECT or VALUES, which compound operators then glue together.
+			#
+			if (select_kw = maybe :SELECT)
+				type_tk = maybe_one_of :DISTINCT, :ALL
+				columns = require_some(trailing_sep: :COMMA) { result_column }
+				if (from_kw = maybe :FROM)
+					source = join_clause
+				end
+				if (where_kw = maybe :WHERE)
+					condition = expression
+				end
+				if (group_by_kw = maybe_all_of :GROUP, :BY)
+					groupings = require_some(trailing_sep: :COMMA) { expression }
+				end
+				# SQLite accepts HAVING without GROUP BY, treating the whole result as one group
+				if (having_kw = maybe :HAVING)
+					having = expression
+				end
+				if (window_kw = maybe :WINDOW)
+					windows = require_some(trailing_sep: :COMMA) { window_definition }
+				end
+
+				SelectCore.concrete(
+					full_source: @lexer.sql,
+					select_kw:   Token::Keyword(select_kw),
+					type_tk:     Token::Keyword(type_tk),
+					columns:     columns,
+					from_kw:     Token::Keyword(from_kw),
+					source:      source,
+					where_kw:    Token::Keyword(where_kw),
+					condition:   condition,
+					group_by_kw: Token::Keyword(group_by_kw),
+					groupings:   groupings,
+					having_kw:   Token::Keyword(having_kw),
+					having:      having,
+					window_kw:   Token::Keyword(window_kw),
+					windows:     windows,
+				)
+			elsif (values_kw = maybe :VALUES)
+				rows = require_some(trailing_sep: :COMMA) do
+					row_lp = require :LP
+					values = require_some(trailing_sep: :COMMA) { expression }
+					row_rp = require :RP
+
+					ValuesRow.concrete(
+						full_source: @lexer.sql,
+						row_lp:      Token::Punctuation(row_lp),
+						values:      values,
+						row_rp:      Token::Punctuation(row_rp),
+					)
+				end
+
+				ValuesClause.concrete(
+					full_source: @lexer.sql,
+					values_kw:   Token::Keyword(values_kw),
+					rows:        rows,
+				)
+			else
+				expected!(:SELECT, :VALUES)
+			end
+		end
+
+		def window_definition
+			#
+				# Relevant `parse.y` grammar rules:
+				#   windowdefn ::= nm AS LP window RP
+				#
+				# Syntax diagram:
+				#   ◯─▶{ window-name }─▶{ AS }─▶[ window-defn ]─▶◯
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/windowfunctions.html#window_chaining
+				#
+				# A named window from the WINDOW clause, which OVER clauses may refer to or extend.
+			#
+			name_tk = name
+			as_kw = require :AS
+			window_lp = require :LP
+			window = window_body
+			window_rp = require :RP
+
+			WindowDefinition.concrete(
+				full_source: @lexer.sql,
+				name_tk:     Token::Identifier(name_tk),
+				as_kw:       Token::Keyword(as_kw),
+				window_lp:   Token::Punctuation(window_lp),
+				**window,
+				window_rp:   Token::Punctuation(window_rp),
+			)
+		end
+
+		# The inside of a window-defn parenthesis, shared by OVER (…) and WINDOW name AS (…)
+		def window_body
+			base_window_name_tk = maybe { identifier(except: [:PARTITION, :ORDER, :RANGE, :ROWS, :GROUPS]) }
+			if (partition_by_kw = maybe_all_of(:PARTITION, :BY))
+				partitions = require_some(trailing_sep: :COMMA) { expression }
+			end
+			if (order_by_kw = maybe_all_of(:ORDER, :BY))
+				orderings = require_some(trailing_sep: :COMMA) { ordering_term }
+			end
+			if current_token in :RANGE | :ROWS | :GROUPS
+				frame = frame_spec
+			end
+
+			{
+				base_window_name_tk: Token::Identifier(base_window_name_tk),
+				partition_by_kw:     Token::Keyword(partition_by_kw),
+				partitions:          partitions,
+				order_by_kw:         Token::Keyword(order_by_kw),
+				orderings:           orderings,
+				frame:               frame,
+			}
+		end
+
+		def filter_clause
+			# Relevant `parse.y` grammar rules:
+			#
+			# Syntax diagram:
+			#   ◯─▶{ FILTER }─▶{ ( }─▶{ WHERE }─▶[ expr ]─▶{ ) }─▶◯
+			#
+			# Simplified grammar:
+			#
+			# Relevant SQLite documentation:
+			#   -
+			#
+			# [description]
+			#
+			require_all_of :FILTER, :LP, :WHERE
+			condition = expression
+			require :RP
+
+			condition
+		end
+
+		def over_clause
+			#
+				# Relevant `parse.y` grammar rules:
+				#   over_clause ::= OVER LP window RP |
+				#                   OVER nm
+				#   window ::= nm frame_opt |
+				#              frame_opt |
+				#              nm ORDER BY sortlist frame_opt |
+				#              ORDER BY sortlist frame_opt |
+				#              nm PARTITION BY nexprlist orderby_opt frame_opt |
+				#              PARTITION BY nexprlist orderby_opt frame_opt
+				#   frame_opt ::= range_or_rows BETWEEN frame_bound_s AND frame_bound_e frame_exclude_opt |
+				#                 range_or_rows frame_bound_s frame_exclude_opt  |
+				#                 .
+				#   range_or_rows ::= RANGE | ROWS | GROUPS
+				#   frame_bound_s ::= frame_bound |
+				#                     UNBOUNDED PRECEDING
+				#   frame_bound_e ::= frame_bound |
+				#                     UNBOUNDED FOLLOWING
+				#   frame_exclude_opt ::= . |
+				#                         EXCLUDE frame_exclude
+				#   frame_bound ::= CURRENT ROW |
+				#                   expr PRECEDING | FOLLOWING
+				#   frame_exclude ::= NO OTHERS | CURRENT ROW | GROUP | TIES
+				#   filter_clause ::= FILTER LP WHERE expr RP
+				#
+				# Syntax diagram:
+				#   ◯─▶{ OVER }┬─▶{ window-name }───────────────────────────────┬─▶◯
+				#              └─▶{ ( }┬───────────▶────────────┐               │
+				#                      └─▶{ base-window-name }─▶┤               │
+				#                   ┌───────────────◀───────────┘               │
+				#                   ├─▶{ PARTITION }─▶{ BY }┬[ expr ]┐          │
+				#                   │                       └─{ , }◀─┤          │
+				#                   ├────────────◀───────────────────┘          │
+				#                   ├─▶{ ORDER }─▶{ BY }┬[ ordering-term ]┐     │
+				#                   │                   └──────{ , }◀─────┤     │
+				#                   ├─────────────◀───────────────────────┘     │
+				#                   ├─▶[ frame-spec ]┬─▶──────────────────{ ) }─┘
+				#                   └─────────▶──────┘
+				#
+				# Simplified grammar:
+				#
+				# Relevant SQLite documentation:
+				#   -
+				#
+				# [description]
+			#
+			over_kw = require :OVER
+			if (window_lp = maybe :LP)
+				window = window_body
+				window_rp = require :RP
+
+				OverClause.concrete(
+					full_source: @lexer.sql,
+					over_kw: Token::Keyword(over_kw),
+					window_lp: Token::Punctuation(window_lp),
+					**window,
+					window_rp: Token::Punctuation(window_rp),
+				)
+			else
+				window_name_tk = identifier
+
+				OverClause.concrete(
+					full_source: @lexer.sql,
+					over_kw: Token::Keyword(over_kw),
+					window_name_tk: Token::Identifier(window_name_tk),
+				)
+			end
+		end
+
+		def foreign_key_clause
+			#
+				# Relevant `parse.y` grammar rules:
+				#   ccons ::= REFERENCES nm eidlist_opt refargs.
+				#   eidlist_opt ::= . | LP eidlist RP
+				#   eidlist ::= eidlist COMMA nm collate sortorder | nm collate sortorder
+				#   refargs ::= . | refargs refarg
+				#   refarg ::= MATCH nm |
+				#              ON INSERT refact |
+				#              ON DELETE refact |
+				#              ON UPDATE refact
+				#   refact ::= SET NULL |
+				#              SET DEFAULT |
+				#              CASCADE |
+				#              RESTRICT |
+				#              NO ACTION
+				#   defer_subclause_opt ::= . | defer_subclause
+				#   defer_subclause ::= NOT DEFERRABLE init_deferred_pred_opt |
+				#                       DEFERRABLE init_deferred_pred_opt
+				#   init_deferred_pred_opt ::= . | INITIALLY DEFERRED | INITIALLY IMMEDIATE
+				#
+				# Syntax diagram:
+				#                                                ┌─────{ , }◀─────┐
+				#   ◯─▶{ REFERENCES }─▶{ foreign-table }─┬─▶{ ( }┴▶{ column-name }┴▶{ ) }──┐
+				#    ┌───────────────────────────────────┴──────────◀──┬───────────────────┘
+				#    ├─▶{ ON }┬─▶{ DELETE }─┬┬─▶{ SET }─▶{ NULL }────▶─┤
+				#    │        └─▶{ UPDATE }─┘├─▶{ SET }─▶{ DEFAULT }─▶─┤
+				#    │                       ├─▶{ CASCADE }──────────▶─┤
+				#    │                       ├─▶{ RESTRICT }─────────▶─┤
+				#    │                       └─▶{ NO }─▶{ ACTION }───▶─┤
+				#    ├─▶{ MATCH }─▶{ name }────────────────────────────┘
+				#    ├────────────┬─▶{ DEFERRABLE }┬───────────────────────────────▶─┐
+				#    ├─{ NOT }──▶─┘                ├─▶{ INITIALLY }─▶{ DEFERRED }──▶─┤
+				#    │                             └─▶{ INITIALLY }─▶{ IMMEDIATE }─▶─┤
+				#    └───────────────────────────────────────────────────────────────┴──────▶◯
+				#
+				# Simplified grammar:
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/syntax/foreign-key-clause.html
+				#   - https://www.sqlite.org/foreignkeys.html
+				#
+				# [description]
+			#
+			references_kw = require :REFERENCES
+			foreign_table = name(except: [:ON, :MATCH, :DEFERRABLE, :NOT])
+			columns = nil
+			if (columns_lp = maybe :LP)
+				columns = require_some(trailing_sep: :COMMA) { indexed_column }
+				columns_rp = require :RP
+			end
+			foreign_key_actions = maybe_some(trailing_sep: nil) do
+				on_kw      = require        :ON
+				trigger_kw = require_one_of :DELETE, :UPDATE
+				action_kw  = require_one_of :CASCADE, :RESTRICT, [:SET, :NULL].freeze, [:SET, :DEFAULT].freeze, [:NO, :ACTION].freeze
+				ForeignKeyAction.concrete(
+					full_source: @lexer.sql,
+					on_kw:       Token::Keyword(on_kw),
+					trigger_kw:  Token::Keyword(trigger_kw),
+					action_kw:   Token::Keyword(action_kw),
+				)
+			end
+			match_clauses = maybe_some(trailing_sep: nil) do
+				match_kw = require :MATCH
+				type_tk  = name(except: [:ON, :MATCH, :DEFERRABLE, :NOT])
+				MatchClause.concrete(
+					full_source: @lexer.sql,
+					match_kw:    Token::Keyword(match_kw),
+					type_tk:     Token::Identifier(type_tk),
+				)
+			end
+
+			deferrable_span = maybe_one_of(
+				[:NOT, :DEFERRABLE, :INITIALLY, :IMMEDIATE].freeze,
+				[:NOT, :DEFERRABLE, :INITIALLY, :DEFERRED].freeze,
+				[:DEFERRABLE, :INITIALLY, :IMMEDIATE].freeze,
+				[:DEFERRABLE, :INITIALLY, :DEFERRED].freeze,
+				[:NOT, :DEFERRABLE].freeze,
+				:DEFERRABLE,
+			)
+
+			ForeignKeyClause.concrete(
+				full_source: @lexer.sql,
+				references_kw: Token::Keyword(references_kw),
+				table_tk: Token::Identifier(foreign_table),
+				columns_lp: Token::Punctuation(columns_lp),
+				columns: columns,
+				columns_rp: Token::Punctuation(columns_rp),
+				actions: foreign_key_actions,
+				match_clauses: match_clauses,
+				deferrable_span: Token::Keyword(deferrable_span),
+			)
+		end
+
+		def conflict_clause
+			#
+				# Used by:
+				#   - column_def
+				#
+				# Relevant `parse.y` grammar rules:
+				#   onconf ::= . | ON CONFLICT resolvetype.
+				#   orconf ::= . | OR resolvetype.
+				#   resolvetype ::= raisetype |
+				#                   IGNORE |
+				#                   REPLACE
+				#   raisetype ::= ROLLBACK |
+				#                 ABORT |
+				#                 FAIL
+				#
+				# Syntax diagram:
+				#   ◯─▶┬────────────────────────────────────────┬─────▶◯
+				#      └─▶{ ON }─▶{ CONFLICT }┬─▶{ ROLLBACK }─▶─┤
+				#                             ├─▶{ ABORT }────▶─┤
+				#                             ├─▶{ FAIL }─────▶─┤
+				#                             ├─▶{ IGNORE }───▶─┤
+				#                             └─▶{ REPLACE }──▶─┘
+				#
+				# Simplified grammar:
+				#   [ON CONFLICT [ROLLBACK | ABORT | FAIL | IGNORE | REPLACE]]
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_conflict.html
+				#
+				# The ON CONFLICT clause is a non-standard extension specific to SQLite
+				# that can appear in many other SQL commands. There are five conflict resolution
+				# algorithm choices: ROLLBACK, ABORT, FAIL, IGNORE, and REPLACE.
+			#
+			return unless (on_conflict_kw  = maybe_all_of :ON, :CONFLICT)
+
+			if (resolution_kw = maybe_one_of :ROLLBACK, :ABORT, :FAIL, :IGNORE, :REPLACE)
+				ConflictClause.concrete(
+					full_source:    @lexer.sql,
+					on_conflict_kw: Token::Keyword(on_conflict_kw),
+					resolution_kw:  Token::Keyword(resolution_kw),
+				)
+			else
+				expected! :ROLLBACK, :ABORT, :FAIL, :IGNORE, :REPLACE
+			end
+		end
+
+		def basic_expression
+			if (operator_tk = maybe :BITNOT)
+				operand = expression(OPERATOR_PRECEDENCE[:BITNOT])
+				operand_attr = (Node === operand) ? { operand: operand } : { operand_tk: operand }
+
+				UnaryExpression.concrete(
+					full_source: @lexer.sql,
+					operator_tk: Token::Punctuation(operator_tk),
+					**operand_attr,
+				)
+			elsif (operator_tk = maybe :PLUS)
+				operand = expression(OPERATOR_PRECEDENCE[:UPLUS])
+				operand_attr = (Node === operand) ? { operand: operand } : { operand_tk: operand }
+
+				UnaryExpression.concrete(
+					full_source: @lexer.sql,
+					operator_tk: Token::Punctuation(operator_tk),
+					**operand_attr,
+				)
+			elsif (operator_tk = maybe :MINUS)
+				operand = expression(OPERATOR_PRECEDENCE[:UMINUS])
+				operand_attr = (Node === operand) ? { operand: operand } : { operand_tk: operand }
+
+				UnaryExpression.concrete(
+					full_source: @lexer.sql,
+					operator_tk: Token::Punctuation(operator_tk),
+					**operand_attr,
+				)
+			elsif (operator_tk = maybe :NOT)
+				operand = expression(OPERATOR_PRECEDENCE[:NOT])
+				operand_attr = (Node === operand) ? { operand: operand } : { operand_tk: operand }
+
+				UnaryExpression.concrete(
+					full_source: @lexer.sql,
+					operator_tk: Token::Punctuation(operator_tk),
+					**operand_attr,
+				)
+			elsif (value_lp = maybe :LP)
+				if subquery_start?
+					select = select_stmt
+					select_rp = require :RP
+
+					return SubqueryExpression.concrete(
+						full_source: @lexer.sql,
+						select_lp: Token::Punctuation(value_lp),
+						select: select,
+						select_rp: Token::Punctuation(select_rp),
+					)
+				end
+
+				value = expression(0)
+				# A row value like `(a, b) = (1, 2)`
+				if current_token == :COMMA
+					value = require_some(given: value, trailing_sep: :COMMA) { expression(0) }
+				end
+				value_rp = require :RP
+
+				value_attr = (Node === value || Array === value) ? { value: value } : { value_tk: value }
+
+				ParentheticalExpression.concrete(
+					full_source: @lexer.sql,
+					value_lp: Token::Punctuation(value_lp),
+					**value_attr,
+					value_rp: Token::Punctuation(value_rp),
+				)
+			elsif (cast_kw = maybe :CAST)
+				expression_lp = require :LP
+				e = expression
+				as_kw = require :AS
+				t = type_name
+				expression_rp = require :RP
+
+				CastExpression.concrete(
+					full_source: @lexer.sql,
+					cast_kw: Token::Keyword(cast_kw),
+					expression_lp: Token::Punctuation(expression_lp),
+					expression: e,
+					as_kw: Token::Keyword(as_kw),
+					as: t,
+					expression_rp: Token::Punctuation(expression_rp),
+				)
+			elsif (case_kw = maybe :CASE)
+				if current_token == :WHEN
+					conditions = require_some(leading_sep: :WHEN) do |when_kw|
+						case_predicate = expression
+						then_kw = require :THEN
+						case_consequence = expression
+
+						predicate_attr = (Node === case_predicate) ? { predicate: case_predicate } : { predicate_tk: case_predicate }
+						consequence_attr = (Node === case_consequence) ? { consequence: case_consequence } : { consequence_tk: case_consequence }
+
+						CaseCondition.concrete(
+							full_source: @lexer.sql,
+							when_kw: Token::Keyword(when_kw),
+							**predicate_attr,
+							then_kw: Token::Keyword(then_kw),
+							**consequence_attr,
+						)
+					end
+					if (else_kw = maybe :ELSE)
+						else_clause = expression
+					end
+					end_kw = require :END
+
+					else_clause_attr = (Node === else_clause) ? { else_clause: else_clause } : { else_clause_tk: else_clause }
+
+					CaseExpression.concrete(
+						full_source: @lexer.sql,
+						case_kw: Token::Keyword(case_kw),
+						conditions: conditions,
+						else_kw: Token::Keyword(else_kw),
+						**else_clause_attr,
+						end_kw: Token::Keyword(end_kw),
+					)
+				else
+					predicate = expression
+					conditions = require_some(leading_sep: :WHEN) do |when_kw|
+						case_predicate = expression
+						then_kw = require :THEN
+						case_consequence = expression
+
+						CaseCondition.concrete(
+							full_source: @lexer.sql,
+							when_kw: Token::Keyword(when_kw),
+							predicate: case_predicate,
+							then_kw: Token::Keyword(then_kw),
+							consequence: case_consequence,
+						)
+					end
+					if (else_kw = maybe :ELSE)
+						else_clause = expression
+					end
+					end_kw = require :END
+
+					CaseExpression.concrete(
+						full_source: @lexer.sql,
+						case_kw: Token::Keyword(case_kw),
+						predicate: predicate,
+						conditions: conditions,
+						else_kw: Token::Keyword(else_kw),
+						else_clause: else_clause,
+						end_kw: Token::Keyword(end_kw),
+					)
+				end
+			elsif (exists_kw = maybe :EXISTS)
+				select_lp = require :LP
+				select = select_stmt
+				select_rp = require :RP
+
+				ExistsExpression.concrete(
+					full_source: @lexer.sql,
+					exists_kw: Token::Keyword(exists_kw),
+					select_lp: Token::Punctuation(select_lp),
+					select: select,
+					select_rp: Token::Punctuation(select_rp),
+				)
+			# A string followed by a dot is a quoted table name, as in `'t'.col`
+			elsif !(current_token == :STRING && peek == :DOT) && (literal_tk = maybe { literal_value })
+				LiteralExpression.concrete(
+					full_source: @lexer.sql,
+					value_tk: literal_tk,
+				)
+			elsif (id = maybe { name(except: [:RAISE]) })
+				case current_token
+				in :IN
+					ColumnName.concrete(
+						full_source: @lexer.sql,
+						column_tk: Token::Identifier(id),
+					)
+				in :LP
+					arguments_lp = require :LP
+					arguments = function_arguments
+					arguments_rp = require :RP
+					if (within_group_kw = maybe_all_of :WITHIN, :GROUP)
+						within_lp = require :LP
+						require_all_of :ORDER, :BY
+						within_group = expression
+						within_rp = require :RP
+					end
+					filter = maybe { filter_clause }
+					over = maybe { over_clause }
+
+					FunctionReference.concrete(
+						full_source: @lexer.sql,
+						function_tk: Token::Identifier(id),
+						arguments_lp: Token::Punctuation(arguments_lp),
+						arguments: arguments,
+						arguments_rp: Token::Punctuation(arguments_rp),
+						within_group_kw: Token::Keyword(within_group_kw),
+						within_lp: Token::Punctuation(within_lp),
+						within_group: within_group,
+						within_rp: Token::Punctuation(within_rp),
+						filter_clause: filter,
+						over_clause: over,
+					)
+				in :DOT
+					schema_or_table_name = id
+					dot = require :DOT
+					table_or_column_name = name
+					if (dot2 = maybe :DOT)
+						column_name = name
+						ColumnName.concrete(
+							full_source: @lexer.sql,
+							schema_tk: Token::Identifier(schema_or_table_name),
+							dot1: Token::Punctuation(dot),
+							table_tk: Token::Identifier(table_or_column_name),
+							dot2: Token::Punctuation(dot2),
+							column_tk: Token::Identifier(column_name),
+						)
+					else
+						ColumnName.concrete(
+							full_source: @lexer.sql,
+							table_tk: Token::Identifier(schema_or_table_name),
+							dot2: Token::Punctuation(dot),
+							column_tk: Token::Identifier(table_or_column_name),
+						)
+					end
+				else
+					ColumnName.concrete(
+						full_source: @lexer.sql,
+						column_tk: Token::Identifier(id),
+					)
+				end
+			elsif :RAISE == current_token
+				raise_function
+			# SQLite lexes "#9" as a variable too, but "#" and a digit only ever names a register
+			# inside its own nested parses, so as a parameter it is a syntax error
+			elsif :VARIABLE == current_token && !current_value.match?(/\A#\d/)
+				Variable.concrete(
+					full_source: @lexer.sql,
+					name_tk: Token::Variable(require(:VARIABLE)),
+				)
+			else
+				expected!("<basic expression>")
+			end
+		end
+
+		def raise_function
+			#
+				# Relevant `parse.y` grammar rules:
+				#    expr ::= RAISE LP IGNORE RP | RAISE LP raisetype COMMA expr RP
+				#    raisetype ::= ROLLBACK | ABORT | FAIL
+				#
+				# Syntax diagram:
+				#   ◯─▶{ RAISE }─▶{ ( }┬─▶{ IGNORE }────────────────────▶─────────┬▶{ ) }──▶◯
+				#                      ├─▶{ ROLLBACK }─┬▶{ , }─▶{ error-message }─┘
+				#                      ├─▶{ ABORT }──▶─┤
+				#                      └─▶{ FAIL }───▶─┘
+				#
+				# Simplified grammar:
+				#   RAISE LP (IGNORE | (ROLLBACK | ABORT | FAIL) COMMA expr) RP
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/syntax/raise-function.html
+				#   - https://www.sqlite.org/lang_createtrigger.html#the_raise_function
+				#
+				# The special RAISE expression that may _only_ occur in trigger programs.
+			#
+			raise_kw = require :RAISE
+			raise_lp = require :LP
+			if (ignore_kw = maybe :IGNORE)
+				raise_rp = require :RP
+
+				RaiseExpression.concrete(
+					full_source: @lexer.sql,
+					raise_kw: Token::Keyword(raise_kw),
+					raise_lp: Token::Punctuation(raise_lp),
+					type_kw: Token::Keyword(ignore_kw),
+					raise_rp: Token::Punctuation(raise_rp),
+				)
+			elsif (rollback_kw = maybe :ROLLBACK)
+				comma_tk = require :COMMA
+				error_message = expression
+				error_message.leading = Token::Punctuation(comma_tk)
+				raise_rp = require :RP
+
+				RaiseExpression.concrete(
+					full_source: @lexer.sql,
+					raise_kw: Token::Keyword(raise_kw),
+					raise_lp: Token::Punctuation(raise_lp),
+					type_kw: Token::Keyword(rollback_kw),
+					raise_rp: Token::Punctuation(raise_rp),
+					error_message: error_message,
+				)
+			elsif (abort_kw = maybe :ABORT)
+				comma_tk = require :COMMA
+				error_message = expression
+				error_message.leading = Token::Punctuation(comma_tk)
+				raise_rp = require :RP
+
+				RaiseExpression.concrete(
+					full_source: @lexer.sql,
+					raise_kw: Token::Keyword(raise_kw),
+					raise_lp: Token::Punctuation(raise_lp),
+					type_kw: Token::Keyword(abort_kw),
+					raise_rp: Token::Punctuation(raise_rp),
+					error_message: error_message,
+				)
+			elsif (fail_kw = maybe :FAIL)
+				comma_tk = require :COMMA
+				error_message = expression
+				error_message.leading = Token::Punctuation(comma_tk)
+				raise_rp = require :RP
+
+				RaiseExpression.concrete(
+					full_source: @lexer.sql,
+					raise_kw: Token::Keyword(raise_kw),
+					raise_lp: Token::Punctuation(raise_lp),
+					type_kw: Token::Keyword(fail_kw),
+					raise_rp: Token::Punctuation(raise_rp),
+					error_message: error_message,
+				)
+			else
+				expected!(:IGNORE, :ROLLBACK, :ABORT, :FAIL)
+			end
+		end
+
+		def function_arguments
+			#
+				# Relevant `parse.y` grammar rules:
+				#   expr ::= idj LP distinct exprlist RP
+				#            idj LP distinct exprlist ORDER BY sortlist RP
+				#            idj LP STAR RP
+				#            idj LP distinct exprlist RP filter_over
+				#            idj LP distinct exprlist ORDER BY sortlist RP filter_over
+				#            idj LP STAR RP filter_over
+				#            idj LP distinct exprlist RP WITHIN GROUP LP ORDER BY expr RP filter_over
+				#   distinct ::= DISTINCT | ALL | .
+				#   exprlist ::= nexprlist | .
+				#   nexprlist ::= nexprlist COMMA expr | expr
+				#
+				# Syntax diagram:
+				#       ┌─▶{ DISTINCT }─▶┐ ┌──{ , }◀─┐
+				#   ◯─▶─┼────────────────┴▶┴▶[ expr ]┼───────────┬─┬─▶◯
+				#       │ ┌──────────────────────────┘           │ │
+				#       │ └▶{ ORDER }─▶{ BY }┬▶[ ordering-term ]─┤ │
+				#       │                    └───────{ , }◀──────┘ │
+				#       └─────────────────┬▶{ * }┬▶────────────────┘
+				#                         └───▶──┘
+				#
+				# Simplified grammar:
+				#
+				# Relevant SQLite documentation:
+				#   -
+				#
+				# [description]
+			#
+			if (distinct_kw = maybe_one_of :DISTINCT, :ALL)
+				expressions = require_some(trailing_sep: :COMMA) { expression }
+				if (order_by_kw = maybe_all_of :ORDER, :BY)
+					order_by = require_some(trailing_sep: :COMMA) { ordering_term }
+				end
+
+				FunctionArguments.concrete(
+					full_source: @lexer.sql,
+					distinct_kw: Token::Keyword(distinct_kw),
+					expressions: expressions,
+					order_by_kw: Token::Keyword(order_by_kw),
+					order_by: order_by,
+				)
+			elsif maybe :STAR
+				StarFunctionArgument.new
+			elsif (e = maybe { expression })
+				expressions = require_some(given: e, trailing_sep: :COMMA) { expression }
+				if (order_by_kw = maybe_all_of :ORDER, :BY)
+					order_by = require_some(trailing_sep: :COMMA) { ordering_term }
+				end
+
+				FunctionArguments.concrete(
+					full_source: @lexer.sql,
+					expressions: expressions,
+					order_by_kw: Token::Keyword(order_by_kw),
+					order_by: order_by,
+				)
+			elsif :RP == current_token
+				EmptyFunctionArgument.new
+			else
+				expected!(:DISTINCT, :STAR, :RP, "expr")
+			end
+		end
+
+		def frame_spec
+			# Relevant `parse.y` grammar rules:
+			#
+			# Syntax diagram:
+			#   ◯┬─▶{ RANGE }──┬┬▶{ BETWEEN }┬─▶{ UNBOUNDED }─▶{ PRECEDING }┬▶{ AND }┬─▶[ expr ]─▶{ PRECEDING }────▶─┐
+			#    ├─▶{ ROWS }───┤│            ├─▶[ expr ]─▶{ PRECEDING }─────┤        ├─▶{ CURRENT }─▶{ ROW }───────▶─┤
+			#    └─▶{ GROUPS }─┘│            ├─▶{ CURRENT }─▶{ ROW }────────┤        ├─▶[ expr ]─▶{ FOLLOWING }────▶─┤
+			#                   │            └─▶[ expr ]─▶{ FOLLOWING }─────┘        └─▶{ UNBOUNDED }─▶{ FOLLOWING }─┤
+			#                   ├─▶{ UNBOUNDED }─▶{ PRECEDING }────────────────────────────────────────────────────▶─┤
+			#                   ├─▶[ expr ]─▶{ PRECEDING }─────────────────────────────────────────────────────────▶─┤
+			#                   └─▶{ CURRENT }─▶{ ROW }────────────────────────────────────────────────────────────▶─┤
+			#                                           ┌────────────────────────────────────────────────────────────┘
+			#                                           ├─▶{ EXCLUDE }─▶{ NO }─▶{ OTHERS }───▶─┐
+			#                                           ├─▶{ EXCLUDE }─▶{ CURRENT }─▶{ ROW }─▶─┤
+			#                                           ├─▶{ EXCLUDE }─▶{ GROUP }────────────▶─┤
+			#                                           ├─▶{ EXCLUDE }─▶{ TIES }─────────────▶─┤
+			#                                           └──────────────────────────────────────┴─────────▶◯
+			#
+			# Simplified grammar:
+			#
+			# Relevant SQLite documentation:
+			#   -
+			#
+			# [description]
+			#
+			if current_token in :RANGE | :ROWS | :GROUPS
+				type_tk = require current_token
+				if (between_kw = maybe :BETWEEN)
+					if (unbounded_kw = maybe :UNBOUNDED)
+						precedence = 1
+						preceding_kw = maybe :PRECEDING
+						starting_boundary = FrameBoundary.concrete(
+							full_source: @lexer.sql,
+							type_tk: Token::Keyword(preceding_kw),
+							value_tk: Token::Keyword(unbounded_kw),
+						)
+					elsif (current_row_kw = maybe_all_of :CURRENT, :ROW)
+						precedence = 3
+						starting_boundary = FrameBoundary.concrete(
+							full_source: @lexer.sql,
+							type_tk: Token::Keyword(current_row_kw),
+						)
+					elsif (e = maybe { expression })
+						if (preceding_kw = maybe :PRECEDING)
+							precedence = 2
+							starting_boundary = FrameBoundary.concrete(
+								full_source: @lexer.sql,
+								type_tk: Token::Keyword(preceding_kw),
+								value: e
+							)
+						elsif (following_kw = maybe :FOLLOWING)
+							precedence = 4
+							starting_boundary = FrameBoundary.concrete(
+								full_source: @lexer.sql,
+								type_tk: Token::Keyword(following_kw),
+								value: e
+							)
+						else
+							expected!(:PRECEDING, :FOLLOWING)
+						end
+					else
+						expected!("UNBOUNDED PRECEDING", "CURRENT ROW", "expr")
+					end
+					and_kw = require :AND
+					if (current_row_kw = maybe_all_of :CURRENT, :ROW)
+						expected!("UNBOUNDED FOLLOWING", "expr") if 3 < precedence
+						ending_boundary = FrameBoundary.concrete(
+							full_source: @lexer.sql,
+							type_tk: Token::Keyword(current_row_kw),
+						)
+					elsif (unbounded_kw = maybe :UNBOUNDED)
+						following_kw = maybe :FOLLOWING
+						ending_boundary = FrameBoundary.concrete(
+							full_source: @lexer.sql,
+							type_tk: Token::Keyword(following_kw),
+							value_tk: Token::Keyword(unbounded_kw),
+						)
+					elsif (e = maybe { expression })
+						if (preceding_kw = maybe :PRECEDING)
+							expected!("CURRENT ROW", "UNBOUNDED FOLLOWING", "expr") if 2 < precedence
+							ending_boundary = FrameBoundary.concrete(
+								full_source: @lexer.sql,
+								type_tk: Token::Keyword(preceding_kw),
+								value: e
+							)
+						elsif (following_kw = maybe :FOLLOWING)
+							expected!("UNBOUNDED FOLLOWING") if 4 < precedence
+							ending_boundary = FrameBoundary.concrete(
+								full_source: @lexer.sql,
+								type_tk: Token::Keyword(following_kw),
+								value: e
+							)
+						else
+							expected!(:PRECEDING, :FOLLOWING)
+						end
+					else
+						expected!("CURRENT ROW", "UNBOUNDED FOLLOWING", "expr")
+					end
+				elsif (unbounded_kw = maybe :UNBOUNDED)
+					preceding_kw = maybe :PRECEDING
+					starting_boundary = FrameBoundary.concrete(
+						full_source: @lexer.sql,
+						type_tk: Token::Keyword(preceding_kw),
+						value_tk: Token::Keyword(unbounded_kw),
+					)
+					ending_boundary = nil
+				elsif (current_row_kw = maybe_all_of :CURRENT, :ROW)
+					starting_boundary = FrameBoundary.concrete(
+						full_source: @lexer.sql,
+						type_tk: Token::Keyword(current_row_kw),
+					)
+					ending_boundary = nil
+				elsif (e = maybe { expression })
+					preceding_kw = require :PRECEDING
+					starting_boundary = FrameBoundary.concrete(
+						full_source: @lexer.sql,
+						type_tk: Token::Keyword(preceding_kw),
+						value: e
+					)
+					ending_boundary = nil
+				else
+					expected!(:BETWEEN, :UNBOUNDED, :CURRENT, "expr")
+				end
+
+				if (exclude_kw = maybe :EXCLUDE)
+					exclude_tk = require_one_of(:GROUP, :TIES, [:NO, :OTHERS].freeze, [:CURRENT, :ROW].freeze)
+				end
+
+				FrameSpec.concrete(
+					full_source: @lexer.sql,
+					type_tk: Token::Keyword(type_tk),
+					between_kw: Token::Keyword(between_kw),
+					and_kw: Token::Keyword(and_kw),
+					starting_boundary: starting_boundary,
+					ending_boundary: ending_boundary,
+					exclude_kw: Token::Keyword(exclude_kw),
+					exclude_tk: Token::Keyword(exclude_tk),
+				)
+			else
+				expected!(:RANGE, :ROWS, :GROUPS)
+			end
+		end
+
+		# ---------- Objects ----------
+
+		def type_name
+			#
+				# Used by:
+				#   - column_def
+				#   - expression
+				#
+				# Relevant `parse.y` grammar rules:
+				#   typetoken ::= . | typename. | typename LP signed RP | typename LP signed COMMA signed RP
+				#   typename ::= ids | typename ids
+				#   signed ::= plus_num | minus_num
+				#   plus_num ::= PLUS number | number
+				#   minus_num ::= MINUS number
+				#   number ::= INTEGER | FLOAT
+				#
+				# Syntax diagram:
+				#   ◯─┬▶{ name }─┬┬──────────────────────────────▶─────────────────────────────┬─▶◯
+				#     └────◀─────┘├─▶{ ( }─▶[ signed-number ]─▶{ ) }─────────────────────────▶─┤
+				#                 └─▶{ ( }─▶[ signed-number ]─▶{ , }─▶[ signed-number ]─▶{ ) }─┘
+				#
+				# Simplified grammar:
+				#   name [name]* [LP signed_number [, signed_number] RP]
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/syntax/type-name.html
+				#   - https://www.sqlite.org/datatype3.html
+				#   - https://www.sqlite.org/flextypegood.html
+				#
+				# A typetoken is really zero or more tokens that form a type name such
+				# as can be found after the column name in a CREATE TABLE statement.
+				# Multiple tokens are concatenated to form the value of the typetoken.
+			#
+			text = scan do
+				require_some(trailing_sep: nil) do
+					identifier_or_string(
+						# Don't consume tokens that mark the start of a `column_constraint`
+						except: Set[:CONSTRAINT, :PRIMARY, :NOT, :NULL, :UNIQUE, :CHECK, :DEFAULT, :COLLATE, :REFERENCES, :GENERATED, :AS]
+					)
+				end
+				if maybe :LP
+					signed_number
+					signed_number if maybe :COMMA
+					require :RP
+				end
+			end
+
+			ColumnType.concrete(
+				full_source: @lexer.sql,
+				text_span: Token::Keyword(text),
+			)
+		end
+
+		def table_name
+			#
+				# Used by:
+				#   - drop table
+				#   - delete
+				#   - update
+				#   - insert
+				#   - alter table
+				#
+				# Relevant `parse.y` grammar rules:
+				#   fullname ::= nm | nm DOT nm
+				#
+				# Syntax diagram:
+				#   ◯─▶┬▶{ schema-name }─▶{ . }┬▶{ object-name }─▶◯
+				#      └───────────▶───────────┘
+				#
+				# Simplified grammar:
+				#   name [DOT name]
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_createtable.html
+				#   - https://www.sqlite.org/lang_createvtab.html
+				#
+				# A table or virtual table name, where the schema name is optional.
+			#
+
+			if peek == :DOT
+				schema_tk = name
+				dot_tk    = require :DOT
+				table_tk  = name
+
+				TableName.concrete(
+					full_source: @lexer.sql,
+					schema_tk:   Token::Identifier(schema_tk),
+					dot_tk:      Token::Punctuation(dot_tk),
+					table_tk:    Token::Identifier(table_tk),
+				)
+			else
+				table_tk = name
+
+				TableName.concrete(
+					full_source: @lexer.sql,
+					table_tk:    Token::Identifier(table_tk),
+				)
+			end
+		end
+
+		# `xfullname` plus the optional `indexed_opt` - the target table of INSERT, UPDATE and DELETE.
+		# Unlike in FROM, an alias here needs the AS
+		def qualified_table_name(indexed:)
+			#
+				# Relevant `parse.y` grammar rules:
+				#   xfullname ::= nm | nm DOT nm | nm DOT nm AS nm | nm AS nm
+				#   indexed_opt ::= . | indexed_by
+				#   indexed_by ::= INDEXED BY nm | NOT INDEXED
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/syntax/qualified-table-name.html
+			#
+			table_tk = name
+			if (dot_tk = maybe :DOT)
+				schema_tk = table_tk
+				table_tk = name
+			end
+			if (as_kw = maybe :AS)
+				alias_tk = name
+			end
+			indexed_kw, index_tk = indexed_by if indexed
+
+			TableName.concrete(
+				full_source: @lexer.sql,
+				schema_tk:   Token::Identifier(schema_tk),
+				dot_tk:      Token::Punctuation(dot_tk),
+				table_tk:    Token::Identifier(table_tk),
+				as_kw:       Token::Keyword(as_kw),
+				alias_tk:    Token::Identifier(alias_tk),
+				indexed_kw:  Token::Keyword(indexed_kw),
+				index_tk:    Token::Identifier(index_tk),
+			)
+		end
+
+		def column_reference
+			ColumnName.concrete(
+				full_source: @lexer.sql,
+				column_tk:   Token::Identifier(name),
+			)
+		end
+
+		def set_list
+			#
+				# Relevant `parse.y` grammar rules:
+				#   setlist ::= setlist COMMA nm EQ expr | setlist COMMA LP idlist RP EQ expr |
+				#               nm EQ expr | LP idlist RP EQ expr
+			#
+			require_some(trailing_sep: :COMMA) do
+				if (columns_lp = maybe :LP)
+					columns = require_some(trailing_sep: :COMMA) { column_reference }
+					columns_rp = require :RP
+				else
+					columns = [column_reference]
+				end
+				eq_tk = require :EQ
+				value = expression
+
+				Assignment.concrete(
+					full_source: @lexer.sql,
+					columns_lp:  Token::Punctuation(columns_lp),
+					columns:     columns,
+					columns_rp:  Token::Punctuation(columns_rp),
+					eq_tk:       Token::Punctuation(eq_tk),
+					value:       value,
+				)
+			end
+		end
+
+		def upsert_clause
+			#
+				# Syntax diagram:
+				#   ◯─▶{ ON }─▶{ CONFLICT }┬───────────────────────────────────────────────────────────┬┐
+				#                          └▶{ ( }┬▶[ indexed-column ]┬▶{ ) }┬────────────────────────┤│
+				#                                 └───────{ , }◀──────┘      └▶{ WHERE }─▶[ expr ]─┘  ││
+				#   ┌───────────────────────────────────────────◀────────────────────────────────────┘│
+				#   └▶{ DO }┬▶{ NOTHING }─────────────────────────────────────────────────────────────┬─▶◯
+				#           └▶{ UPDATE }─▶{ SET }─▶[ assignment … ]┬───────────────────────┬──────────┘
+				#                                                  └▶{ WHERE }─▶[ expr ]──┘
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_upsert.html
+			#
+			on_conflict_kw = require_all_of :ON, :CONFLICT
+			if (target_lp = maybe :LP)
+				targets = require_some(trailing_sep: :COMMA) { ordering_term }
+				target_rp = require :RP
+				if (target_where_kw = maybe :WHERE)
+					target_condition = expression
+				end
+			end
+			do_kw = require :DO
+			action_kw = require_one_of :NOTHING, :UPDATE
+			if action_kw[0] == :UPDATE
+				set_kw = require :SET
+				assignments = set_list
+				if (where_kw = maybe :WHERE)
+					condition = expression
+				end
+			end
+
+			UpsertClause.concrete(
+				full_source:      @lexer.sql,
+				on_conflict_kw:   Token::Keyword(on_conflict_kw),
+				target_lp:        Token::Punctuation(target_lp),
+				targets:          targets,
+				target_rp:        Token::Punctuation(target_rp),
+				target_where_kw:  Token::Keyword(target_where_kw),
+				target_condition: target_condition,
+				do_kw:            Token::Keyword(do_kw),
+				action_kw:        Token::Keyword(action_kw),
+				set_kw:           Token::Keyword(set_kw),
+				assignments:      assignments,
+				where_kw:         Token::Keyword(where_kw),
+				condition:        condition,
+			)
+		end
+
+		def returning_clause
+			#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/lang_returning.html
+			#
+			return unless (returning_kw = maybe :RETURNING)
+
+			ReturningClause.concrete(
+				full_source:  @lexer.sql,
+				returning_kw: Token::Keyword(returning_kw),
+				columns:      require_some(trailing_sep: :COMMA) { result_column },
+			)
+		end
+
+		# The statement-level tail of SELECT. UPDATE and DELETE take it too when SQLite is built
+		# with SQLITE_ENABLE_UPDATE_DELETE_LIMIT, so we accept it there as well
+		def order_and_limit
+			if (order_by_kw = maybe_all_of :ORDER, :BY)
+				orderings = require_some(trailing_sep: :COMMA) { ordering_term }
+			end
+			if (limit_kw = maybe :LIMIT)
+				limit = expression
+				if (offset_tk = maybe :OFFSET)
+					offset = expression
+				elsif (offset_tk = maybe :COMMA)
+					# `LIMIT 5, 10` means skip 5 and take 10, the reverse of what it looks like
+					offset = limit
+					limit = expression
+				end
+			end
+
+			{
+				order_by_kw: Token::Keyword(order_by_kw),
+				orderings:   orderings,
+				limit_kw:    Token::Keyword(limit_kw),
+				limit:       limit,
+				offset_tk:   Token::Keyword(offset_tk),
+				offset:      offset,
+			}
+		end
+
+		def result_column
+			#
+				# Relevant `parse.y` grammar rules:
+				#   selcollist ::= sclp expr as | sclp STAR | sclp nm DOT STAR
+				#   sclp ::= selcollist COMMA | .
+				#   as ::= AS nm | ids | .
+				#   nm ::= idj | STRING
+				#
+				# Syntax diagram:
+				#   ◯┬─▶[ expr ]─┬─────────────▶───────────────┬─▶◯
+				#    │           ├─▶{ AS }┬▶{ column-alias }─▶─┤
+				#    │           └────▶───┘                    │
+				#    ├─▶{ * }────────────────────────────────▶─┤
+				#    └─▶{ table-name }─▶{ . }─▶{ * }─────────▶─┘
+				#
+				# Simplified grammar:
+				#   expr [[AS] alias] | * | table.*
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/syntax/result-column.html
+				#
+				# One of the columns a select core returns.
+			#
+			if (star_tk = maybe :STAR)
+				StarResultColumn.concrete(
+					full_source: @lexer.sql,
+					star_tk:     Token::Punctuation(star_tk),
+				)
+			elsif peek(2) == [:DOT, :STAR]
+				table_tk = name
+				dot_tk = require :DOT
+				star_tk = require :STAR
+
+				StarResultColumn.concrete(
+					full_source: @lexer.sql,
+					table_tk:    Token::Identifier(table_tk),
+					dot_tk:      Token::Punctuation(dot_tk),
+					star_tk:     Token::Punctuation(star_tk),
+				)
+			else
+				result = expression
+				as_kw, alias_tk = as_alias
+				result_attr = (Node === result) ? { result: result } : { result_tk: result }
+
+				ResultColumn.concrete(
+					full_source: @lexer.sql,
+					**result_attr,
+					as_kw:       Token::Keyword(as_kw),
+					alias_tk:    Token::Identifier(alias_tk),
+				)
+			end
+		end
+
+		# `as ::= AS nm | ids | .` - without the AS only an identifier or a string will do,
+		# so keywords like INDEXED or the join keywords are never mistaken for an alias
+		def as_alias
+			if (as_kw = maybe :AS)
+				[as_kw, name]
+			else
+				[nil, maybe { identifier_or_string }]
+			end
+		end
+
+		# indexed_by ::= INDEXED BY nm | NOT INDEXED
+		def indexed_by
+			if (indexed_kw = maybe_all_of :INDEXED, :BY)
+				[indexed_kw, name]
+			else
+				[maybe_all_of(:NOT, :INDEXED), nil]
+			end
+		end
+
+		def join_clause
+			#
+				# Relevant `parse.y` grammar rules:
+				#   seltablist ::= stl_prefix nm dbnm as on_using |
+				#                  stl_prefix nm dbnm as indexed_by on_using |
+				#                  stl_prefix nm dbnm LP exprlist RP as on_using |
+				#                  stl_prefix LP select RP as on_using |
+				#                  stl_prefix LP seltablist RP as on_using
+				#   stl_prefix ::= seltablist joinop | .
+				#   joinop ::= COMMA | JOIN | JOIN_KW JOIN | JOIN_KW nm JOIN | JOIN_KW nm nm JOIN
+				#   on_using ::= ON expr | USING LP idlist RP | .
+				#
+				# Syntax diagram:
+				#   ◯─▶[ table-or-subquery ]┬───────────────────────────────▶────────────────────────────────┬─▶◯
+				#                           └┬▶[ join-operator ]─▶[ table-or-subquery ]─▶[ join-constraint ]┬┘
+				#                            └───────────────────────────────◀──────────────────────────────┘
+				#
+				# Simplified grammar:
+				#   table_or_subquery [join_operator table_or_subquery [join_constraint]]*
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/syntax/join-clause.html
+				#
+				# Everything after FROM. A comma between sources is a join operator as well.
+			#
+			left = table_or_subquery
+			joins = []
+
+			while (operator = join_operator)
+				source = table_or_subquery
+				constraint = join_constraint
+
+				joins << Join.concrete(
+					full_source: @lexer.sql,
+					operator:    operator,
+					source:      source,
+					constraint:  constraint,
+				)
+			end
+
+			return left if joins.empty?
+
+			JoinClause.concrete(
+				full_source: @lexer.sql,
+				left:        left,
+				joins:       joins,
+			)
+		end
+
+		JOIN_KEYWORDS = Set[:NATURAL, :LEFT, :RIGHT, :FULL, :INNER, :OUTER, :CROSS].freeze
+
+		def join_operator
+			#
+				# Syntax diagram:
+				#   ◯┬──────────────▶─────────────────────────────────────────┬─▶◯
+				#    ├─▶{ , }──────────────────────────────────────────────▶─┤
+				#    ├─▶{ NATURAL }─┬┬─▶{ LEFT }──┬┬────────────┬┬▶{ JOIN }─┘
+				#    └──────▶───────┘├─▶{ RIGHT }─┤└▶{ OUTER }─▶┘│
+				#                    ├─▶{ FULL }──┘              │
+				#                    ├─▶{ INNER }───────────────▶┤
+				#                    └─▶{ CROSS }───────────────▶┘
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/syntax/join-operator.html
+			#
+			if (comma_tk = maybe :COMMA)
+				return JoinOperator.concrete(
+					full_source: @lexer.sql,
+					operator_tk: Token::Punctuation(comma_tk),
+				)
+			end
+
+			keywords = []
+			keywords << advance while keywords.size < 3 && JOIN_KEYWORDS.include?(current_token)
+			join_tk = keywords.any? ? require(:JOIN) : maybe(:JOIN)
+			return unless join_tk
+
+			JoinOperator.concrete(
+				full_source: @lexer.sql,
+				operator_tk: Token::Keyword(token_span_from(*keywords, join_tk)),
+			)
+		end
+
+		def join_constraint
+			#
+				# Syntax diagram:
+				#   ◯─┬─▶{ ON }─▶[ expr ]─────────────────────────────▶─┬─▶◯
+				#     ├─▶{ USING }─▶{ ( }┬▶{ column-name }┬▶{ ) }────▶─┤
+				#     │                  └──────{ , }◀────┘             │
+				#     └────────────────────────▶────────────────────────┘
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/syntax/join-constraint.html
+			#
+			if (on_kw = maybe :ON)
+				JoinOnConstraint.concrete(
+					full_source: @lexer.sql,
+					on_kw:       Token::Keyword(on_kw),
+					expression:  expression,
+				)
+			elsif (using_kw = maybe :USING)
+				columns_lp = require :LP
+				columns = require_some(trailing_sep: :COMMA) { column_reference }
+				columns_rp = require :RP
+
+				JoinUsingConstraint.concrete(
+					full_source: @lexer.sql,
+					using_kw:    Token::Keyword(using_kw),
+					columns_lp:  Token::Punctuation(columns_lp),
+					columns:     columns,
+					columns_rp:  Token::Punctuation(columns_rp),
+				)
+			end
+		end
+
+		def table_or_subquery
+			#
+				# Relevant `parse.y` grammar rules:
+				#   seltablist ::= stl_prefix nm dbnm as on_using |
+				#                  stl_prefix nm dbnm as indexed_by on_using |
+				#                  stl_prefix nm dbnm LP exprlist RP as on_using |
+				#                  stl_prefix LP select RP as on_using |
+				#                  stl_prefix LP seltablist RP as on_using
+				#   indexed_by ::= INDEXED BY nm | NOT INDEXED
+				#   as ::= AS nm | ids | .
+				#   dbnm ::= . | DOT nm
+				#
+				# Syntax diagram:
+				#                                              ┌───────┬────────────────┐
+				#   ◯┬─────────────────────────┬─{ table-name }┴▶{ AS }┴▶{ table-alias }┴┬──────────┬─▶◯
+				#    ├─▶{ schema-name }─▶{ . }─┤   ┌───────────────────◀─────────────────┘          │
+				#    │                         │   ├─▶{ INDEXED }─▶{ BY }─▶{ index-name }─────────▶─┤
+				#    │                         │   └─▶{ NOT }─▶{ INDEXED }────────────────────────▶─┤
+				#    │                         │                                 ┌──{ , }◀─┐        │
+				#    ├─▶───────────────────────┴───{ table-function-name }─▶{ ( }┴▶[ expr ]┴▶{ ) }┬─┤
+				#    │                                               ┌────────┬───────◀───────────┘ │
+				#    │                                ┌──────────────┴▶{ AS }┬┴▶{ table-alias }───▶─┤
+				#    ├─▶{ ( }─▶[ select-stmt ]─▶{ ) }─┴──────────────────────┴────────────────────▶─┤
+				#    └─▶{ ( }┬┬▶[ table-or-subquery ]┬┬▶{ ) }─────────────────────────────────────▶─┘
+				#            │└─────────{ , }◀───────┘│
+				#            └─▶────[ join-clause ]───┘
+				#
+				# Simplified grammar:
+				#   [schema .] table [[AS] alias] [INDEXED BY index | NOT INDEXED]
+				#   | [schema .] function (expr [, expr]*) [[AS] alias]
+				#   | (select_stmt) [[AS] alias]
+				#   | (join_clause) [[AS] alias]
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/syntax/table-or-subquery.html
+				#
+				# One source of rows in a FROM clause.
+			#
+			if (source_lp = maybe :LP)
+				if subquery_start?
+					select = select_stmt
+					source_rp = require :RP
+					as_kw, alias_tk = as_alias
+
+					SubquerySource.concrete(
+						full_source: @lexer.sql,
+						select_lp:   Token::Punctuation(source_lp),
+						select:      select,
+						select_rp:   Token::Punctuation(source_rp),
+						as_kw:       Token::Keyword(as_kw),
+						alias_tk:    Token::Identifier(alias_tk),
+					)
+				else
+					source = join_clause
+					source_rp = require :RP
+					as_kw, alias_tk = as_alias
+
+					ParentheticalSource.concrete(
+						full_source: @lexer.sql,
+						source_lp:   Token::Punctuation(source_lp),
+						source:      source,
+						source_rp:   Token::Punctuation(source_rp),
+						as_kw:       Token::Keyword(as_kw),
+						alias_tk:    Token::Identifier(alias_tk),
+					)
+				end
+			else
+				schema_tk = nil
+				object_tk = name
+				if (dot_tk = maybe :DOT)
+					schema_tk = object_tk
+					object_tk = name
+				end
+
+				if (arguments_lp = maybe :LP)
+					arguments = maybe_some(trailing_sep: :COMMA) { expression }
+					arguments_rp = require :RP
+					as_kw, alias_tk = as_alias
+
+					TableFunction.concrete(
+						full_source:  @lexer.sql,
+						schema_tk:    Token::Identifier(schema_tk),
+						dot_tk:       Token::Punctuation(dot_tk),
+						function_tk:  Token::Identifier(object_tk),
+						arguments_lp: Token::Punctuation(arguments_lp),
+						arguments:    arguments,
+						arguments_rp: Token::Punctuation(arguments_rp),
+						as_kw:        Token::Keyword(as_kw),
+						alias_tk:     Token::Identifier(alias_tk),
+					)
+				else
+					as_kw, alias_tk = as_alias
+					indexed_kw, index_tk = indexed_by
+
+					TableName.concrete(
+						full_source: @lexer.sql,
+						schema_tk:   Token::Identifier(schema_tk),
+						dot_tk:      Token::Punctuation(dot_tk),
+						table_tk:    Token::Identifier(object_tk),
+						as_kw:       Token::Keyword(as_kw),
+						alias_tk:    Token::Identifier(alias_tk),
+						indexed_kw:  Token::Keyword(indexed_kw),
+						index_tk:    Token::Identifier(index_tk),
+					)
+				end
+			end
+		end
+
+		def literal_value
+			#
+				# Relevant `parse.y` grammar rules:
+				#   term ::= NULL | FLOAT | BLOB | STRING | INTEGER | CTIME_KW | QNUMBER
+				#
+				# Syntax diagram:
+				#   ◯─┬─▶{ numeric-literal }────▶─┬─▶◯
+				#     ├─▶{ string-literal }─────▶─┤
+				#     ├─▶{ blob-literal }───────▶─┤
+				#     ├─▶{ NULL }───────────────▶─┤
+				#     ├─▶{ TRUE }───────────────▶─┤
+				#     ├─▶{ FALSE }──────────────▶─┤
+				#     ├─▶{ CURRENT_TIME }───────▶─┤
+				#     ├─▶{ CURRENT_DATE }───────▶─┤
+				#     └─▶{ CURRENT_TIMESTAMP }──▶─┘
+				#
+				# Simplified grammar:
+				#   NULL | CURRENT_TIME | CURRENT_DATE | CURRENT_TIMESTAMP | INTEGER | FLOAT | STRING | BLOB | TRUE | FALSE | QNUMBER
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/syntax/literal-value.html
+				#
+				# A value
+			#
+			case current_token
+			when :NULL
+				tk = require :NULL
+				Token::Null(tk)
+			when :CURRENT_TIME
+				tk = require :CURRENT_TIME
+				Token::Current(tk)
+			when :CURRENT_DATE
+				tk = require :CURRENT_DATE
+				Token::Current(tk)
+			when :CURRENT_TIMESTAMP
+				tk = require :CURRENT_TIMESTAMP
+				Token::Current(tk)
+			when :TRUE
+				tk = require :TRUE
+				Token::True(tk)
+			when :FALSE
+				tk = require :FALSE
+				Token::False(tk)
+			when :QNUMBER
+				tk = require_qnumber
+				Token::Numeric(tk)
+			when :INTEGER
+				tk = require :INTEGER
+				Token::Numeric(tk)
+			when :FLOAT
+				tk = require :FLOAT
+				Token::Numeric(tk)
+			when :STRING
+				tk = require :STRING
+				Token::String(tk)
+			when :BLOB
+				tk = require :BLOB
+				Token::Blob(tk)
+			else
+				expected!(:INTEGER, :FLOAT, :QNUMBER, :STRING, :BLOB, :NULL, :TRUE, :FALSE, :CURRENT_TIME, :CURRENT_DATE, :CURRENT_TIMESTAMP)
+			end
+		end
+
+		def indexed_column
+			# Used by:
+			#   - ccons, tcons, creat-view, cte
+			#
+			# Relevant `parse.y` grammar rules:
+			#   eidlist ::= eidlist COMMA nm collate sortorder | nm collate sortorder
+			#   collate ::= . | COLLATE ids.
+			#   sortorder ::= . | ASC | DESC
+			#   nm ::= idj | STRING
+			#
+			# Syntax diagram:
+			#   ◯──▶{ column-name }─┬─▶───────────────────────────────┬┬────────▶───┬──▶◯
+			#                       └─▶{ COLLATE }─▶{ collation-name }┘├─▶{ ASC }─▶─┤
+			#                                                          └─▶{ DESC }──┘
+			#
+			# Simplified grammar:
+			#   name [COLLATE identifier] [ASC | DESC]
+			#
+			# Relevant SQLite documentation:
+			#   - https://www.sqlite.org/syntax/indexed-column.html
+			#
+			# The eidlist non-terminal (Expression Id List) generates an ExprList
+			# from a list of identifiers.  The identifier names are in ExprList.a[].zName.
+			# This list is stored in an ExprList rather than an IdList so that it
+			# can be easily sent to sqlite3ColumnsExprList().
+			#
+			# IMPORTANT COMPATIBILITY NOTE:  Some prior versions of SQLite accepted
+			# COLLATE clauses and ASC or DESC keywords on ID lists in inappropriate
+			# places - places that might have been stored in the sqlite_schema table.
+			# Those extra features were ignored.  But because they might be in some
+			# (busted) old databases, we need to continue parsing them when loading
+			# historical schemas.
+			#
+			column_name  = name(except: [:COLLATE, :ASC, :DESC].freeze)
+			if (collate_kw = maybe :COLLATE)
+				collation_tk = identifier_or_string(except: [:ASC, :DESC].freeze)
+			end
+			direction_tk = maybe_one_of :ASC, :DESC
+
+			IndexedColumn.concrete(
+				full_source: @lexer.sql,
+				name_tk:      Token::Identifier(column_name),
+				collate_kw:   Token::Keyword(collate_kw),
+				collation_tk: Token::Keyword(collation_tk),
+				direction_tk: Token::Keyword(direction_tk)
+			)
+		end
+
+		def indexed_expression
+			# Used by:
+			#   - tcons, upsert, expr (ORDER BY), create-index, window
+			#
+			# Relevant `parse.y` grammar rules:
+			#   sortlist  ::= sortlist COMMA expr sortorder nulls | expr sortorder nulls
+			#   sortorder ::= . | ASC | DESC
+			#   nulls     ::= . | NULLS FIRST | NULLS LAST
+			#
+			# Syntax diagram:
+			#   ◯─▶[ expr ]─┬────────▶───┬─┬────────────▶────────────┬───▶◯
+			#               ├─▶{ ASC }─▶─┤ ├─▶{ NULLS }─▶{ FIRST }─▶─┤
+			#               └─▶{ DESC }──┘ └─▶{ NULLS }─▶{ LAST }────┘
+			#
+			# Simplified grammar:
+			#   expression [ASC | DESC] [NULLS FIRST | NULLS LAST]
+			#
+			# Relevant SQLite documentation:
+			#   - https://www.sqlite.org/syntax/indexed-column.html
+			#   - https://www.sqlite.org/expridx.html
+			#
+			# the sortlist non-terminal stores a list of expression where each
+			# expression is optionally followed by ASC or DESC to indicate the
+			# sort order.
+			#
+			expr         = expression
+			direction_tk = maybe_one_of :ASC, :DESC
+			nulls_tk     = maybe_one_of [:NULLS, :FIRST].freeze, [:NULLS, :LAST].freeze
+
+			IndexedExpression.new(
+				expression: expr,
+				direction_tk: Token::Keyword(direction_tk),
+				nulls_tk: Token::Keyword(nulls_tk)
+			)
+		end
+
+		def ordering_term
+			#
+				# Relevant `parse.y` grammar rules:
+				#   sortlist ::= sortlist COMMA expr sortorder nulls |
+				#                expr sortorder nulls
+				#   sortorder ::= ASC | DESC | .
+				#   nulls ::= NULLS FIRST | NULLS LAST | .
+				#   eidlist ::= eidlist COMMA nm collate sortorder |
+				#               nm collate sortorder
+				#   collate ::= . |
+				#               COLLATE ids.
+				#
+				# Syntax diagram:
+				#   ◯─▶[ expr ]┬───────────────▶────────────────┬▶┬──────▶─────┬▶┬──────────▶─────────────┬─▶◯
+				#              └▶{ COLLATE }─▶{ collation-name }┘ ├▶{ ASC }──▶─┤ ├▶{ NULLS }─▶{ FIRST }─▶─┤
+				#                                                 └▶{ DESC }─▶─┘ └▶{ NULLS }─▶{ LAST }──▶─┘
+				#
+				# Simplified grammar:
+				#
+				# Relevant SQLite documentation:
+				#   -
+				#
+				# [description]
+			#
+			e = expression
+			if (collate_kw = maybe(:COLLATE))
+				collation = identifier
+			end
+			if current_token in :ASC | :DESC
+				direction = require current_token
+			end
+			if (nulls_kw = maybe(:NULLS))
+				nulls = require current_token
+			end
+
+			OrderingTerm.concrete(
+				full_source: @lexer.sql,
+				expression: e,
+				collate_kw: Token::Keyword(collate_kw),
+				collation_tk: Token::Identifier(collation),
+				direction_tk: Token::Keyword(direction),
+				nulls_kw: Token::Keyword(nulls_kw),
+				nulls_tk: Token::Keyword(nulls),
+			)
+		end
+
+		def signed_number
+			#
+				# Used by:
+				#   - pragma_value
+				#   - type_name
+				#
+				# Relevant `parse.y` grammar rules:
+				#   signed ::= plus_num | minus_num
+				#   plus_num ::= PLUS number | number
+				#   minus_num ::= MINUS number
+				#   number ::= INTEGER | FLOAT
+				#
+				# Syntax diagram:
+				#   ◯─┬───▶────┬─▶{ numeric-literal }─▶◯
+				#     ├─▶{ + }─┤
+				#     └─▶{ - }─┘
+				#
+				# Simplified grammar:
+				#   [PLUS | MINUS] INTEGER | FLOAT
+				#
+				# Relevant SQLite documentation:
+				#   - https://www.sqlite.org/syntax/signed-number.html
+				#
+				# A signed number is a numeric literal with an optional sign.
+			#
+			sign_tk  = maybe_one_of   :PLUS, :MINUS
+			value_tk = require_one_of :INTEGER, :FLOAT
+
+			return value_tk unless sign_tk
+
+			sign_tk, beg, _fin = sign_tk
+			value_tk, _beg, fin = value_tk
+
+			[[sign_tk, value_tk], beg, fin]
+		end
+
+		# ---------- Token Classes ----------
+
+		def name(except: [])
+			# Relevant `parse.y` grammar rules:
+			#   // The name of a column or table can be any of the following:
+			#   nm ::= idj | STRING
+			#
+			if current_token in :STRING
+				require current_token
+			elsif (idj = maybe { identifier_or_join_keyword(except: except) })
+				idj
+			else
+				expected!(:STRING, "idj")
+			end
+		end
+
+		def identifier(except: [])
+			# Relevant `parse.y` grammar rules:
+			#   // An IDENTIFIER can be a generic identifier, or one of several
+			#   // keywords.  Any non-standard keyword can also be an identifier.
+			#   %token_class id  ID | INDEXED
+			#
+			if current_token in :ID | :INDEXED
+				require current_token
+			elsif !except.include?(current_token) && TOKEN_FALLBACKS.include?(current_token)
+				require current_token
+			else
+				expected!(:ID, :INDEXED, "fallback")
+			end
+		end
+
+		def identifier_or_string(except: [])
+			# Relevant `parse.y` grammar rules:
+			#   // And "ids" is an identifer-or-string.
+			#   %token_class ids  ID | STRING
+			#
+			if current_token in :ID | :STRING
+				require current_token
+			elsif !except.include?(current_token) && TOKEN_FALLBACKS.include?(current_token)
+				require current_token
+			else
+				expected!(:ID, :STRING, "fallback")
+			end
+		end
+
+		def identifier_or_join_keyword(except: [])
+			# Relevant `parse.y` grammar rules:
+			#   // An identifier or a join-keyword
+			#   %token_class idj  ID | INDEXED | JOIN_KW
+			#
+			if current_token in :ID | :INDEXED | :CROSS | :FULL | :INNER | :LEFT | :NATURAL | :OUTER | :RIGHT
+				require current_token
+			elsif !except.include?(current_token) && TOKEN_FALLBACKS.include?(current_token)
+				require current_token
+			else
+				expected!(:ID, :INDEXED, :JOIN_KW, "fallback")
+			end
+		end
+
+		private
+
+		# ---------- Parsing Helpers ----------
+
+		def require(token)
+			if token == current_token
+				advance
+			else
+				expected!(token)
+			end
+		end
+
+		def maybe(token = nil) # &block
+			if block_given?
+				start_buffer = @peek_buffer.dup
+				start_lexer_cursor = @lexer.cursor
+				start_lexer_pos = @lexer.anchor
+
+				result = catch(:ERROR) { yield }
+
+				if ErrorMessage === result
+					@peek_buffer = start_buffer
+					@lexer.cursor = start_lexer_cursor
+					@lexer.anchor = start_lexer_pos
+					nil
+				else
+					result
+				end
+			elsif token == current_token
+				advance
+			end
+		end
+
+		# The span of whatever the block consumes. Positions come from the buffered tokens - the
+		# lexer itself may already be several tokens ahead, looking ahead for us
+		def scan(&block)
+			beg = ensure_buffer[0][1]
+			block.call
+			[[], beg, @last_consumed_fin]
+		end
+
+		def require_all_of(*tokens)
+			i, len = 0, tokens.length
+			start_pos, end_pos = nil
+			while i < len
+				if tokens[i] == current_token
+					_tok, beg, fin = advance
+					start_pos = beg if i == 0 # first iteration
+					end_pos = fin if i == (len - 1) # last iteration
+
+					i += 1
+				else
+					# TODO: only show the tokens from `i` onwards
+					expected!(tokens)
+				end
+			end
+
+			[tokens, start_pos, end_pos].freeze
+		end
+
+		def maybe_all_of(*tokens)
+			advancing = true
+			i, len = 0, tokens.length
+			buf = ensure_buffer(len)
+			while i < len
+				tok, * = buf[i]
+				if tokens[i] != tok
+					advancing = false
+					break
+				end
+				i += 1
+			end
+
+			require_all_of(*tokens) if advancing
+		end
+
+		def maybe_one_of(*tokens)
+			i, len = 0, tokens.length
+			has_nil = false
+			while i < len
+				check = tokens[i]
+				if Symbol === check && (result = maybe check)
+					return result
+				elsif Array === check && (result = maybe_all_of(*check))
+					return result
+				elsif check.nil?
+					has_nil = true
+					i += 1
+				else
+					i += 1
+				end
+			end
+
+			has_nil ? false : nil
+		end
+
+		def require_one_of(*tokens)
+			maybe_one_of(*tokens) || expected!(*tokens)
+		end
+
+		def require_some(trailing_sep: nil, leading_sep: nil, given: nil, &)
+			a = []
+
+			if leading_sep
+				sep_tk = (Array === leading_sep) ? require_one_of(*leading_sep) : require(leading_sep)
+				val = yield(sep_tk)
+				insert = if Node === val
+					val.leading = Token::Punctuation(sep_tk)
+					val
+				else
+					[sep_tk, val]
+				end
+			elsif trailing_sep
+				val = (given || yield)
+				sep_tk = (Array === trailing_sep) ? maybe_one_of(*trailing_sep) : maybe(trailing_sep)
+				insert = if Node === val
+					val.trailing = Token::Punctuation(sep_tk) if sep_tk
+					val
+				else
+					[val, sep_tk]
+				end
+			else
+				val = (given || yield)
+				insert = val
+			end
+			a << insert
+
+			collect(into: a, trailing_sep: trailing_sep, leading_sep: leading_sep, &)
+
+			return token_span_from(*a) if a.all?(Array)
+
+			a
+		end
+
+		def maybe_some(trailing_sep: nil, leading_sep: nil, &)
+			a = []
+
+			if leading_sep
+				sep_tk = (Array === leading_sep) ? maybe_one_of(*leading_sep) : maybe(leading_sep)
+				return if sep_tk.nil?
+
+				val = yield(sep_tk)
+				insert = if Node === val
+					val.leading = Token::Punctuation(sep_tk)
+					val
+				else
+					[sep_tk, val]
+				end
+			elsif trailing_sep
+				val = maybe { yield }
+				return a if val.nil?
+
+				sep_tk = (Array === trailing_sep) ? maybe_one_of(*trailing_sep) : maybe(trailing_sep)
+				insert = if Node === val
+					val.trailing = Token::Punctuation(sep_tk) if sep_tk
+					val
+				else
+					[val, sep_tk]
+				end
+			else
+				insert = maybe { yield }
+			end
+
+			if insert
+				a << insert
+
+				collect(into: a, trailing_sep: trailing_sep, leading_sep: leading_sep, &)
+
+				return token_span_from(*a) if a.all?(Array)
+
+				a
+			else
+				a
+			end
+		end
+
+		def collect(into: [], trailing_sep: nil, leading_sep: nil, &block)
+			break_next = false
+			while true
+				break if break_next
+
+				if leading_sep
+					sep_tk = (Array === leading_sep) ? maybe_one_of(*leading_sep) : maybe(leading_sep)
+					break if sep_tk.nil?
+
+					val = yield(sep_tk)
+
+					insert = if Node === val
+						val.leading = Token::Punctuation(sep_tk)
+						val
+					else
+						[sep_tk, val]
+					end
+
+				elsif trailing_sep
+					val = maybe { yield }
+					break if val.nil?
+
+					sep_tk = (Array === trailing_sep) ? maybe_one_of(*trailing_sep) : maybe(trailing_sep)
+					break_next = sep_tk.nil?
+
+					insert = if Node === val
+						val.trailing = Token::Punctuation(sep_tk) if sep_tk
+						val
+					else
+						[val, sep_tk]
+					end
+
+				else
+					val = maybe { yield }
+					break if val.nil?
+
+					insert = val
+				end
+
+				into << insert
+			end
+
+			into
+		end
+
+		def token_span_from(*spans)
+			spans.compact!
+			i, len = 0, spans.length
+			start_pos, end_pos = nil
+			tokens = []
+			while i < len
+				tok, beg, fin = spans[i]
+				(Array === tok) ? (tokens.concat(tok)) : (tokens << tok)
+				start_pos = beg if i == 0 # first iteration
+				end_pos = fin if i == (len - 1) # last iteration
+
+				i += 1
+			end
+
+			[tokens, start_pos, end_pos]
+		end
+
+		# Access the current token type in the lexer stream.
+		# Reads but doesn't consume the token at the head of the buffer.
+		def current_token
+			buf = ensure_buffer
+			buf[0][0]
+		end
+
+		# SQLite lexes any number holding a "_" as a QNUMBER, and accepts it as a term only if every
+		# "_" sits between two digits - hex digits after 0x
+		def require_qnumber
+			value = current_value
+			misplaced = value.match?(/\A0x/i) ? /(?<!\h)_|_(?!\h)/ : /(?<!\d)_|_(?!\d)/
+			expected!(:INTEGER, :FLOAT) if value.match?(misplaced)
+			require :QNUMBER
+		end
+
+		def statement_end?
+			current_token.nil? || current_token == :SEMI
+		end
+
+		def subquery_start?
+			current_token in :SELECT | :VALUES | :WITH
+		end
+
+		# Access the string value for the current token.
+		# Reads but doesn't consume the start_pos and end_pos values at the head of the buffer.
+		def current_value
+			buf = ensure_buffer
+			sql = @lexer.sql
+			_tk, beg, fin = buf[0]
+
+			sql.byteslice(beg, (fin - beg))
+		end
+
+		# Peek at the next `n` tokens in the lexer stream.
+		# Does not advance the lexer cursor and does not consider the current token.
+		def peek(n = 1)
+			buf = ensure_buffer(n + 1)
+			return buf[1][0] if n == 1 # symbol element of the 2nd token
+
+			[].tap do |a|
+				i = 1
+				while i <= n
+					tok, * = buf[i]
+					a << tok
+					i += 1
+				end
+			end
+		end
+
+		# Advance the implicit cursor pointing to the first token in the buffer by `n` tokens.
+		# Returns the new token at the head of the buffer.
+		def advance
+			ensure_buffer
+			token = @peek_buffer.shift
+			@last_consumed_fin = token[2]
+			token
+		end
+
+		# Ensure that the `@peek_buffer` has at least `size` tokens.
+		# Note: the buffer is a nested array of `[token, start_pos, end_pos]` triples
+		def ensure_buffer(size = 1)
+			lexer = @lexer
+			@peek_buffer.tap do |buf|
+				while buf.size < size
+					buf << [lexer.next_token, lexer.anchor, lexer.cursor].freeze
+				end
+			end
+		end
+
+		def expected!(*expected)
+			token, beg, fin = ensure_buffer[0]
+			throw :ERROR, ErrorMessage.new(@lexer.sql, token, beg, fin, expected)
+		end
+	end
+end