third_party/sqlite/fuzz/sql_query_grammar.proto

// Copyright 2018 The Chromium Authors
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

syntax = "proto2";

import "icu_codes.proto";

package sql_query_grammar;

/* Generates SQL queries in protobuf format, using enums for the tables and
   integers for column names. The column name integers will be wrapped.
   Relies heavily on the sqlite grammar from here:
   https://www.sqlite.org/lang.html

   Yes, it is all in one big blobby proto file, as everything is extremely
   mutually recursive and there is no such thing as circular imports or forward
   declarations in protobuf.
*/

message SQLQuery {
  // No comments generated, nor EXPLAIN queries
  oneof query_oneof {
    Select select = 1;
    CreateTable create_table = 2;
    Insert insert = 3;
    Delete delete = 4;
    CreateFTS3Table fts3_table = 5;
    FTS3SpecificQuery fts_query = 6;
    BeginTransaction begin_txn = 7;
    CommitTransaction commit_txn = 8;
    RollbackStatement rollback_stmt = 9;
    CreateSavePoint create_save_point = 10;
    ReleaseSavePoint release_save_point = 11;
    Analyze analyze = 12;
    Vacuum vacuum = 13;
    Pragma pragma = 14;
    Update update = 15;
    CreateIndex create_index = 16;
    CreateView create_view = 17;
    CreateTrigger create_trigger = 18;
    ReIndex reindex = 19;
    Drop drop = 20;
    AlterTable alter_table = 21;
    AttachDatabase attach_db = 22;
    DetachDatabase detach_db = 23;
    FTS3HiddenTableInsert fts3_insert = 24;
    FTS3HiddenTableUpdate fts3_update = 25;
    FTS3HiddenTableDelete fts3_delete = 26;
  }
}

// ~~~~FTS3~~~

message FTS3SpecificQuery {
  oneof query_oneof {
    FTS3SpecialCommand command = 1;
    // The following exists to increase the probability of hitting MATCH
    // queries.
    FTS3SelectMatch select = 2;
  }
}

message FTS3SelectMatch {
  // SELECT * FROM |table| WHERE |col| MATCH |match_pattern|
  required FTS3Table table = 1;
  required Column col = 2;
  required FTS3MatchFormat match_pattern = 3;
}

message FTS3SpecialCommand {
  // INSERT INTO xyz(xyz) VALUES('optimize');
  // is a special command that optimizes FTS table xyz.
  required FTS3Table table = 1;
  enum Command {
    OPTIMIZE = 0;
    REBUILD = 1;
    INTEGRITY_CHECK = 2;
    MERGE = 3;
    AUTOMERGE = 4;
  }
  required Command command = 2;
  required uint32 val1 = 3;
  required uint32 val2 = 4;
}

message ICULocale {
  required IsoLangCode iso_lang_code = 1;
  // least significant 2 bytes will each be clamped to within 'A' and
  // 'Z' and then concatenated together to form a country code.
  required uint32 country_code = 2;
  // TODO(mpdenton) generate anything else here? Will sqlite care?
}

enum TokenizerType {
  TT_SIMPLE = 0;
  TT_PORTER = 1;
  TT_ICU = 2;
  TT_UNICODE61 = 3;  // FIXME in the future: remove_diacritics, separators,
                     // token_chars.... (not supported in Chrome)
}

// FIXME in the future: maybe also fuzz the tokenizer type/icu locales with
// random strings???
message CreateFTS3Table {
  required bool if_not_exists = 1;
  optional Schema schema = 6;
  required FTS3Table table = 2;
  optional ColumnList col_list = 3;
  optional TokenizerType tokenizer_type = 4;
  required ICULocale locale = 5;  // used if tokenizer_type == TT_ICU.
}

message FTS3Table {
  required uint32 table = 1;
}

// TODO(mpdenton) query by row id?

message FTS3MatchFormat {
  repeated FTS3MatchCompoundFormat ft = 1;
}

message FTS3MatchCompoundFormat {
  required FTS3MatchFormatCore core = 1;
  repeated FTS3CompoundAndCore compound_and_cores = 2;
}

message FTS3CompoundAndCore {
  enum CompoundOp {
    OR = 0;
    // Chrome does not enable the enhanced query syntax. FIXME in the future.
    // AND = 1;
    // NOT = 2;
  }
  required CompoundOp op = 1;
  required FTS3MatchFormatCore core = 2;
}

message FTS3MatchFormatCore {
  oneof ft_oneof {
    FTS3PhraseQuery pq = 1;
    FTS3NearQuery nq = 2;
  }
  required FTS3MatchToken mt_fallback = 3;
}

message FTS3NearQuery {
  // May generate into another NEAR query, this is fine.
  required FTS3MatchFormatCore format_core1 = 1;
  optional uint32 num_tokens_near = 2;
  required FTS3MatchFormatCore format_core2 = 3;
}

message FTS3PhraseQuery {
  required FTS3MatchToken mt = 1;
  repeated FTS3MatchToken extra_mts = 2;
}

message FTS3MatchToken {
  optional Column col = 1;  // search in this column specifically
  required bool negate = 4;
  required string token = 2;  // token (or prefix) to search for
  required bool prefix = 3;   // append star to end
}

// FIXME in the future: https://www.sqlite.org/fts3.html Section 8.2: Query
// tokenizers themselves. Unsupported by Chrome.

message FTS3OffsetsFn {
  required FTS3Table table = 1;
}

message FTS3SnippetsFn {
  required FTS3Table table = 1;
  enum NumArgs {
    A0 = 0;
    A1 = 1;
    A2 = 2;
    A3 = 3;
    A4 = 4;
    A5 = 5;
  }
  required NumArgs num_optional_args = 2;
  required string start_match = 3;
  required string end_match = 4;
  required string ellipses = 5;
  optional uint32 col_number = 6;  // capped by KMaxColumnNumber
  required uint32 num_tokens = 7;  // in [-64, 64].
}

message FTS3MatchInfoFn {
  required FTS3Table table = 1;

  // The following will be wrapped to any of the characters p,c,n,a,,l,s,x,y,b.
  repeated uint32 chars = 2;
}

// FTS3 Auxiliary function format
message FTS3AuxiliaryFn {
  required FTS3OffsetsFn offsets_fallback = 1;
  oneof fts_aux_fn_oneof {
    FTS3SnippetsFn snippets = 2;
    FTS3MatchInfoFn matchinfo = 3;
  }
}

// Types of hidden tables in FTS3
message FTS3HiddenTable {
  enum HiddenTableVal {
    // CONTENT = 0; // Taken care of in |Table|
    SEGDIR = 1;
    SEGMENTS = 2;
    // stat and docsize are FTS4
  }
  required HiddenTableVal htv = 1;
  required FTS3Table table = 2;
}

// The different columns that go into the FTS3 hidden tables above
enum FTS3HiddenTableColumn {
  FTS3_HT_BLOCKID = 0;
  FTS3_HT_BLOCK = 1;
  FTS3_HT_LEVEL = 2;
  FTS3_HT_IDX = 3;
  FTS3_HT_START_BLOCK = 4;
  FTS3_HT_LEAVES_END_BLOCK = 5;
  FTS3_HT_END_BLOCK = 6;
  FTS3_HT_ROOT = 7;
}

// Some extra stuff to modify the FTS3 Hidden tables, only used for FTS3
// Fuzzing.

// Used to insert single values in.
message FTS3HiddenTableInsert {
  required FTS3HiddenTable fht = 1;
  repeated FTS3HiddenTableColumnValue col_vals = 2;
}

message FTS3HiddenTableUpdate {
  required FTS3HiddenTable fht = 1;
  repeated FTS3HiddenTableColumnValue col_vals = 2;

  // mini WHERE clause with some BinOp
  optional FTS3HiddenTableColumn col_where = 3;
  required BinaryOperator bin_op = 4;
  required Expr comp_expr = 5;
}

message FTS3HiddenTableDelete {
  required FTS3HiddenTable fht = 1;
  optional FTS3HiddenTableColumn col_where = 2;
  required BinaryOperator bin_op = 3;
  required Expr comp_expr = 4;
}

message FTS3HiddenTableColumnValue {
  required FTS3HiddenTableColumn col = 1;
  required Expr expr = 2;
}

// ~~~~TRANSACTIONs~~~~
// TODO(mpdenton) these could be meta-statements that enclose a bunch of other
// statements. Is that worthwhile or should I just let the fuzzer generate them
// haphazardly?
message BeginTransaction {
  enum TransactionType {
    DEFERRED = 0;
    IMMEDIATE = 1;
    EXCLUSIVE = 2;
  }
  optional TransactionType type = 1;
}

message CommitTransaction {
  enum CommitText {
    COMMIT = 0;
    END = 1;
    COMMIT_TRANSACTION = 2;
    END_TRANSACTION = 3;
  }
  required CommitText text = 1;
}

message SavePoint {
  required uint32 savepoint_num = 1;
}

message RollbackStatement {
  optional SavePoint save_point = 1;
}

message CreateSavePoint {
  required SavePoint save_point = 1;
}

message ReleaseSavePoint {
  required SavePoint save_point = 1;
}

// ~~~~ANALYZE~~~~
message Analyze {
  optional Schema schema_name = 1;
  optional Table table_name = 2;
  optional Index index_name = 3;
}

// ~~~~VACUUM~~~~
message Vacuum {
  optional Schema schema = 1;
}

// ~~~~PRAGMA~~~~
message Pragma {
  // FIXME in the future: full list here: https://www.sqlite.org/pragma.html
  // These are the ones I've seen in Chrome
  enum PragmaCommand {
    // QUICK_CHECK = 0;
    // INTEGRITY_CHECK = 1;
    AUTO_VACUUM = 2;
    WRITEABLE_SCHEMA = 3;
    LOCKING_MODE = 4;
    TEMP_STORE = 5;
    PAGE_SIZE_ = 6;
    TABLE_INFO = 7;
    JOURNAL_MODE = 8;
    MMAP_SIZE = 9;
  }
  required PragmaCommand command = 1;
  optional Schema schema = 2;
  required int32 arg1 = 3;
}

// ~~~~CREATE_TABLE~~~~
enum TempModifier {
  TM_TEMP = 3;
  TM_TEMPORARY = 4;
}

message CreateTable {
  optional TempModifier temp_modifier = 1;
  required bool if_not_exists = 2;
  required ExprSchemaTable schema_table = 3;
  oneof create_table_oneof {
    CreateTableOpt1 op1 = 4;
    Select as_select_stmt = 5;  // AS select-stmt
  }
  required CreateTableOpt1 op = 6;  // used only if the above oneof is empty
}

message CreateTableOpt1 {
  required ColumnDef col_def = 1;
  repeated ColumnDef extra_col_defs = 2;
  repeated TableConstraint table_constraints = 3;
  required bool without_rowid = 4;
}

message ColumnDef {
  required Column col = 1;
  optional TypeName type_name = 2;
  repeated ColumnConstraint col_constraints = 3;
}

message TypeName {
  // Things like VARCHAR(100) are simply not enforced by SQLite (this example
  // would end up with a column affinity of TEXT). So I don't make much effort
  // to generate them.
  required CastTypeName ctn = 1;
  optional uint32 sn = 2;
}

message ColumnConstraint {
  optional ColumnConstraintName constraint_name = 1;
  oneof col_constraint_oneof {
    ColConstraintOpt1 opt1 = 2;
    ConflictClause not_null_conf_clause = 3;
    ConflictClause unique_conf_clause = 4;
    Expr check_expr = 5;  // CORPUS specialize??
    ColConstraintOpt2 opt2 = 6;
    CollateType collate = 7;
    ForeignKeyClause fkey_clause = 8;
  }
  required ColConstraintOpt2 opt2_fallback = 9;
}

message ColConstraintOpt1 {
  required AscDesc asc_desc = 1;
  required ConflictClause conflict = 2;
  required bool autoincrement = 3;
}

message ColConstraintOpt2 {
  required LiteralValue lit_val = 1;
  optional Expr expr = 2;  // CORPUS specialize?
}

message ConflictClause {
  enum OnConflict {
    ROLLBACK = 0;
    ABORT = 1;
    FAIL = 2;
    IGNORE = 3;
    REPLACE = 4;
  }
  optional OnConflict on_conflict = 1;
}

message TableConstraint {
  optional TableConstraintName name = 1;
  oneof table_constraint_oneof {
    TableConstraintOpt1 opt1 = 2;
    Expr check_expr = 3;  // CORPUS specialize?
    TableConstraintOpt2 opt2 = 4;
  }
}

message TableConstraintOpt1 {
  enum ConstraintType {
    PRIMARY_KEY = 0;
    UNIQUE = 1;
  }
  required ConstraintType constraint_type = 1;
  required IndexedColumnList indexed_col_list = 2;
  required ConflictClause conf_clause = 3;
}

message TableConstraintOpt2 {
  required ColumnList cols = 1;
  required ForeignKeyClause fkey_clause = 2;
}

message IndexedColumn {
  optional Expr expr = 1;   // CORPUS specialize?
  required Column col = 2;  // only used if expr non-existent
  optional CollateType collate_type = 3;
  required AscDesc asc_desc = 4;
}

message IndexedColumnList {
  required IndexedColumn indexed_col = 1;
  repeated IndexedColumn extra_indexed_cols = 2;
}

message ForeignKeyClause {
  required Table foreign_table = 1;
  optional ColumnList col_list = 2;
  repeated ForeignKeyClauseCore fkey_cores = 3;
  optional DeferStrategy defer_strat = 4;
}

message ForeignKeyClauseCore {
  optional ForeignKeyClauseNotMatch fkey_not_match = 1;
  // Sqlite doesn't obey MATCH expressions, MATCH SIMPLE is always assumed
}

message ForeignKeyClauseNotMatch {
  enum DeleteOrUpdate {
    DELETE = 0;
    UPDATE = 1;
  }
  required DeleteOrUpdate del_or_update = 1;

  enum Action {
    SET_NULL = 0;
    SET_DEFAULT = 1;
    CASCADE = 2;
    RESTRICT = 3;
    NO_ACTION = 4;
  }
  required Action action = 2;
}

message DeferStrategy {
  required bool not = 1;
  enum DeferStratEnum {
    INITIALLY_DEFERRED = 0;
    INITIALLY_IMMEDIATE = 1;
    NONE = 2;
  }
  required DeferStratEnum strat = 2;
}

// ~~~~DELETE~~~~
// FIXME in the future: SQL_ENABLE_UPDATE_DELETE_LIMIT
message Delete {
  optional WithStatement with = 1;
  required QualifiedTableName qtn = 2;
  optional WhereStatement where = 3;
}

message QualifiedTableName {
  required SchemaTableAsAlias staa = 1;
  required bool indexed = 2;
  required bool not_indexed = 3;
  required Index indexed_by = 4;
}

// ~~~~INSERT~~~~

message UpsertClausePart1 {
  required IndexedColumnList icol_list = 1;
  optional WhereStatement where_stmt = 2;
}

message ColEqualsExpr {
  oneof uclause_p2_oneof {
    Column col = 1;
    ColumnList col_list = 2;
  }
  required Expr expr = 3;
}

message UpsertClausePart2 {
  required ColEqualsExpr cee = 1;
  repeated ColEqualsExpr extra_cees = 2;
  optional WhereStatement where_stmt = 3;
}

message UpsertClause {
  optional UpsertClausePart1 uclause_p1 = 1;
  optional UpsertClausePart2 uclause_p2 = 2;
}

message SchemaTableAsAlias {
  required ExprSchemaTable schema_table = 1;
  optional Table as_table_alias = 2;
}

message Insert {
  optional WithStatement with = 1;
  enum InsertType {
    INSERT = 0;
    REPLACE = 1;
    INSERT_OR_REPLACE = 2;
    INSERT_OR_ROLLBACK = 3;
    INSERT_OR_ABORT = 4;
    INSERT_OR_FAIL = 5;
    INSERT_OR_IGNORE = 6;
  }
  required InsertType insert_type = 2;
  required SchemaTableAsAlias staa = 3;
  optional ColumnList col_list = 5;
  oneof insert_oneof {
    ValuesStatement values = 6;
    Select select = 7;
  }  // if empty, DEFAULT VALUES
  optional UpsertClause upsert_clause = 8;
}

// ~~~~UPDATE~~~~
message Update {
  optional WithStatement with = 1;
  enum UpdateType {
    OR_ROLLBACK = 0;
    OR_ABORT = 1;
    OR_REPLACE = 2;
    OR_FAIL = 3;
    OR_IGNORE = 4;
  }
  optional UpdateType update_type = 2;
  required QualifiedTableName qtn = 3;
  required UpsertClausePart2 ucp2 = 4;
}

// ~~~~CREATE INDEX~~~~
message CreateIndex {
  required bool unique = 1;
  required bool if_not_exists = 2;
  optional Schema schema = 3;
  required Index index = 4;
  required Table table = 5;
  required IndexedColumnList icol_list = 6;
  optional WhereStatement where = 7;
}

// ~~~~CREATE VIEW~~~~
message View {
  required uint32 view = 1;
}
message CreateView {
  optional TempModifier temp_modifier = 1;
  required bool if_not_exists = 2;
  optional Schema schema = 3;
  required View view = 4;
  optional ColumnList col_list = 5;
  required Select select = 6;
}

// ~~~~CREATE TRIGGER~~~~
message Trigger {
  required uint32 trigger = 1;
}

message CreateTrigger {
  optional TempModifier temp_modifier = 1;
  required bool if_not_exists = 2;
  optional Schema schema = 3;
  required Trigger trigger = 4;
  enum TriggerType {
    BEFORE = 0;
    AFTER = 1;
    INSTEAD_OF = 2;
  }
  optional TriggerType trigger_type = 5;
  enum TriggerInstr {
    DELETE = 0;
    UPDATE = 1;
    INSERT = 2;
  }
  optional TriggerInstr trigger_instr = 6;
  required ColumnList col_list = 7;  // CORPUS create corpus item with an Update
                                     // using the same ColumnList as this
  required Table table = 8;
  required bool for_each_row = 9;
  optional ExprComparisonHighProbability when = 10;  // for the WHEN statement
  // There are significant restrictions on update, insert, select, and delete
  // expressions used in triggers. However, we might as well generate normal
  // queries and see if these restrictions are adequately enforced.

  // Also, from https://www.sqlite.org/lang_createtrigger.html: "If a BEFORE
  // UPDATE or BEFORE DELETE trigger modifies or deletes a row that was to have
  // been updated or deleted, then the result of the subsequent update or delete
  // operation is undefined. Furthermore, if a BEFORE trigger modifies or
  // deletes a row, then it is undefined whether or not AFTER triggers that
  // would have otherwise run on those rows will in fact run."

  // It is unclear what is meant by "undefined". Are we talking memory
  // corruption? I suppose the fuzzer will find it and we'll see.
  required TypicalQuery tq = 11;
  repeated TypicalQuery extra_tqs = 12;
}

message TypicalQuery {
  oneof tq_oneof {
    Update update = 1;
    Insert insert = 2;
    Select select = 3;
  }
  required Delete delete_fallback = 4;
}

// ~~~~REINDEX~~~~
message ReIndex {
  required bool empty = 5;
  optional CollateType collate_type = 4;  // used if schema does not exist
  optional Schema schema = 1;
  optional Table table = 2;
  required Index index = 3;  // used if table does not exist.
}

// ~~~~DROP *~~~~
message Drop {
  required bool if_exists = 5;
  optional Schema schema = 6;
  oneof drop_oneof {
    Index index = 1;
    Table table = 2;
    Trigger trigger = 3;
  }
  required View view_fallback = 4;
}

// ~~~~ALTER TABLE~~~~
message AlterTable {
  required ExprSchemaTable schema_table = 1;
  required bool column = 2;
  optional Column col = 4;
  required Column col_to = 5;
  optional ColumnDef col_def = 6;
  required Table table_fallback = 3;
}

// ~~~~ATTACH DATABASE~~~~

// SEE: https://www.sqlite.org/inmemorydb.html
// Lots of different options.
// TODO(mpdenton) may want to experiment with on-filesystem main dbs as well...
message AttachDatabase {
  required bool in_memory = 1;
  required bool file_uri = 2;
  optional Schema db_name = 3;
  required bool shared_cache = 4;
  required Schema schema = 5;
}

// ~~~~DETACH DATABASE~~~~
message DetachDatabase {
  required Schema schema = 1;
}

// ~~~~SELECT~~~~
/*
  Select is obviously the most complicated syntax in the language, and the
  fuzzer will likely generate plenty of invalid SELECT statements. As long as
  it does not generate too many, we should still be perfectly fine.

  From sqlite docs:
  Note that there are paths through the syntax diagrams that are not allowed in
  practice. Some examples:

  A VALUES clause can be the first element in a compound SELECT that uses a WITH
  clause, but a simple SELECT that consists of just a VALUES clause cannot be
  preceded by a WITH clause. The WITH clause must occur on the first SELECT of a
  compound SELECT. It cannot follow a compound-operator.


  See https://www.sqlite.org/lang_select.html.
*/
message Select {
  optional WithStatement with = 1;

  // SQL grammar specifies SelectSubStatement here but that just leads
  // to a bunch of unparseable VALUES ... ; statements. So require
  // SelectCore here.
  required SelectStatementCore select_core = 2;
  repeated ExtraSelectSubStatement extra_select_substatements = 3;
  optional OrderByStatement orderby = 4;
  optional LimitStatement limit = 5;
}

message OrderByStatement {
  required ExprOrderingTerm ord_term = 1;
  repeated ExprOrderingTerm extra_ord_terms = 2;
}

message LimitStatement {
  // CORPUS specialize these exprs???
  required Expr limit_expr = 1;
  required bool offset = 2;  // this is only used if second_expr exists
  optional Expr second_expr = 3;
}

message ExtraSelectSubStatement {
  required CompoundOperator compound_op = 1;
  required SelectSubStatement select_substatement = 2;
}

enum CompoundOperator {
  CO_UNION = 0;
  CO_UNION_ALL = 1;
  CO_INTERSECT = 2;
  CO_EXCEPT = 3;
}

message SelectSubStatement {
  oneof select_subexpr_oneof {
    SelectStatementCore select_core = 1;
    ValuesStatement values = 2;
  }
  required ValuesStatement values_fallback = 3;
}

message ValuesStatement {
  required ExprList expr_list = 1;
  repeated ExprList extra_expr_lists = 2;  // CORPUS specialize?
}

message ExprColAlias {
  required Expr expr = 1;
  optional Column col_alias = 2;
  required bool as = 3;
}

message ResultColumn {
  oneof result_col_oneof {
    Column col = 1;
    ExprColAlias eca = 2;
    Table table_star = 3;
    FTS3AuxiliaryFn fts3_fn = 4;  // Only emitted when FUZZ_FTS3 enabled
  }                               // if nothing, use star *
}

message SelectStatementCore {
  enum SelectOrDistinct {
    SELECT_DISTINCT = 0;
    SELECT = 1;
    SELECT_ALL = 2;
  }
  required SelectOrDistinct s_or_d = 1;
  repeated ResultColumn result_columns = 2;
  optional FromStatement from = 3;
  optional WhereStatement where = 4;
  optional GroupByStatement groupby = 5;
  optional WindowStatement window = 6;
}

message WithStatement {
  required bool recursive = 1;
  required CommonTableExpr table_expr = 2;
  repeated CommonTableExpr extra_table_exprs = 3;
}

message FromStatement {
  required TableOrSubqueryOption3 tos3 = 1;
}

message WindowStatement {
  required WindowStatementNaming win = 1;
  repeated WindowStatementNaming extra_wins = 2;
}

message WindowStatementNaming {
  required WindowName name = 1;
  required WindowDefn defn = 2;
}

message GroupByStatement {
  // CORPUS specialize all these exprs????
  required ExprList exprs = 1;
  optional Expr having_expr = 3;
}

message WhereStatement {
  required ExprComparisonHighProbability expr = 1;
}

message CommonTableExpr {
  required Table table = 1;
  repeated Column columns = 2;
  required Select select = 3;
}

// ~~~~Join stuff~~~~~
message JoinClause {
  required TableOrSubquery tos = 1;
  repeated JoinClauseCore clauses = 2;
}

message JoinClauseCore {
  required JoinOperator join_op = 1;
  required TableOrSubquery tos = 2;
  required JoinConstraint join_constraint = 3;
}

message JoinOperator {
  required bool comma = 1;
  // the following fields only used if comma is false
  required bool natural = 2;
  enum JoinType {
    LEFT = 0;
    LEFT_OUTER = 1;
    INNER = 2;
    CROSS = 3;
    NONE = 4;
  }
  required JoinType join_type = 3;
}

message JoinConstraint {
  oneof join_constraint_oneof {
    Expr on_expr = 1;
    UsingExpr using_expr = 2;
  }  // fine if empty
}

message UsingExpr {
  required ColumnList col_list = 1;
}

// ~~~~~Table names etc.~~~~~

// First checks if main is set. Then checks if temp is set. Then checks if
// other schema number is set.
message Schema {
  required uint32 schema = 1;
  required bool main = 2;
  required bool temp = 3;
}

message Table {
  required uint32 table = 1;
  optional bool fts3_content_table = 2;  // only used for FTS3 fuzzing
}

message Column {
  required uint32 column = 1;
  optional bool rowid = 2;  // can also have "rowid" column

  // FTS has a hidden column with the same name as the table.
  optional FTS3Table fts3_table = 3;
  // FTS3 tables have "docid" as an alias for "rowid".
  optional bool fts3_docid = 4;
}

message WindowName {
  required uint32 window_name = 1;
}

message ColumnConstraintName {
  required uint32 constraint_name = 1;
}

message TableConstraintName {
  required uint32 constraint_name = 1;
}

message Index {
  required uint32 index = 1;
}

// Example:
// column1, column2, column3
message ColumnList {
  required Column col = 1;
  repeated Column extra_cols = 2;
}

// ~~~~table-or-subquery~~~~
message TableOrSubquery {
  oneof tos_oneof {
    QualifiedTableName qtn = 1;
    TableOrSubqueryOption2 toso2 = 2;
    TableOrSubqueryOption3 toso3 = 3;
    TableOrSubqueryOption4 toso4 = 4;
  }
  required ExprSchemaTable schema_table_expr = 5;  // used if the oneof is empty
}

message TableOrSubqueryOption2 {
  required ExprSchemaTableFn schema_table_fn = 1;
  optional AsTableAlias as_table_alias = 2;
}

message TableOrSubqueryOption3 {
  repeated TableOrSubquery tos_list = 1;  // if empty, use the join clause
  required JoinClause join_clause = 2;
}

message TableOrSubqueryOption4 {
  required Select select = 1;
  optional AsTableAlias as_table_alias = 2;
}

message AsTableAlias {
  required bool as = 2;
  required Table table_alias = 3;
}

// ~~~~Expressions~~~~
message Expr {
  oneof expr_oneof {
    LiteralValue lit_val = 1;
    ComplicatedExpr comp_expr = 2;
  }
}

message NumericLiteral {
  repeated uint32 hex_digits = 1;
  repeated uint32 digits = 2;
  required bool decimal_point = 3;
  repeated uint32 dec_digits = 4;
  repeated uint32 exp_digits = 5;
  required bool negative_exp = 6;
}

message LiteralValue {
  enum SpecialVal {
    VAL_NULL = 0;  // using just "NULL" vauses it not to compile.
    VAL_TRUE = 1;
    VAL_FALSE = 2;
    CURRENT_TIME = 3;
    CURRENT_DATE = 4;
    CURRENT_TIMESTAMP = 5;
  }
  oneof lit_val_oneof {
    int64 num_lit = 1;
    string string_lit = 2;
    bytes blob_lit = 3;
    SpecialVal special_val = 4;
    NumericLiteral numeric_lit = 5;
  }  // If no value, just use int64(1)
}

enum UnaryOperator {
  UNOP_MINUS = 1;
  UNOP_PLUS = 2;
  UNOP_TILDE = 3;
  UNOP_NOT = 4;
}

message UnaryExpr {
  required UnaryOperator unary_op = 1;
  required Expr expr = 2;
}

enum BinaryOperator {
  BINOP_CONCAT = 1;  // double pipe
  BINOP_STAR = 2;
  BINOP_SLASH = 3;
  BINOP_PERCENT = 4;
  BINOP_PLUS = 5;
  BINOP_MINUS = 6;
  BINOP_LELE = 7;  // <<
  BINOP_GRGR = 8;  // >>
  BINOP_AMPERSAND = 9;
  BINOP_PIPE = 10;
  BINOP_LE = 11;
  BINOP_LEQ = 12;
  BINOP_GR = 13;
  BINOP_GREQ = 14;
  BINOP_EQ = 15;
  BINOP_EQEQ = 16;
  BINOP_NOTEQ = 17;
  BINOP_LEGR = 18;  // <> (not equal)
  BINOP_IS = 19;
  BINOP_ISNOT = 20;
  BINOP_IN = 21;
  BINOP_LIKE = 22;
  BINOP_GLOB = 23;
  BINOP_MATCH = 24;
  BINOP_REGEXP = 25;
  BINOP_AND = 26;
  BINOP_OR = 27;
}

message BinaryExpr {
  required Expr lhs = 1;
  required BinaryOperator op = 2;
  required Expr rhs = 3;

  // In FUZZ_FTS3 mode, if |fmt| exists we will use it instead of rhs to
  // help generate better MATCH queries.
  optional FTS3MatchFormat fmt = 4;
}

// Used to inflate the probability that we get a comparison of a column with an
// expr. This is useful, as an example, for WHERE expressions.
message ExprComparisonHighProbability {
  oneof expr_comp_oneof {
    ColumnComparison cc = 1;
    Expr expr = 2;
  }
}

// Special version of expr that only generates predicates with a column on the
// left.
message ColumnComparison {
  required ExprSchemaTableColumn col = 1;
  required BinaryOperator op = 2;
  required Expr expr = 3;
  // In FUZZ_FTS3 mode, if |fmt| exists we will use it instead of rhs to
  // help generate better MATCH queries.
  optional FTS3MatchFormat fmt = 4;
}

message ExprSchemaTableColumn {
  optional Schema schema = 1;
  optional Table table = 2;
  required Column col = 3;
}

// Separate this out to inflate the probability of having a literal value
message ComplicatedExpr {
  // Don't want bind-parameter, unless fuzzing sql_bind
  oneof complicated_expr_oneof {
    ExprSchemaTableColumn expr_stc = 2;
    UnaryExpr unary_expr = 3;
    BinaryExpr binary_expr = 4;
    Fn fn_expr = 5;
    ParenthesesExpr par_expr = 6;
    CastExpr cast_expr = 7;
    CollateExpr collate_expr = 8;
    Expr1 expr1 = 9;
    ExprNullTests expr_null_tests = 10;
    ExprIs expr_is = 11;
    ExprBetween expr_between = 12;
    ExprIn expr_in = 17;
    ExprExists expr_exists = 13;
    ExprCase expr_case = 14;
    ExprRaiseFn expr_raise = 15;
  }
  required LiteralValue lit_val = 16;  // used if oneof is empty.
}

message ExprRaiseFn {
  required bool ignore = 3;
  enum RaiseFnEnum {
    ROLLBACK = 0;
    ABORT = 1;
    FAIL = 2;
  }
  required RaiseFnEnum raise_fn = 1;
  required string error_msg = 2;
}

message ExprCase {
  optional Expr expr = 1;
  required ExprWhenThen when_then = 2;
  repeated ExprWhenThen extra_when_thens = 3;
  optional Expr else_expr = 4;
}

message ExprWhenThen {
  required Expr when_expr = 1;
  required Expr then_expr = 2;
}

message ExprExists {
  required bool not = 1;
  required bool exists = 2;
  required Select select = 3;
}

message ExprIn {
  required Expr expr = 5;
  required bool not = 1;
  oneof exprin_oneof {
    ExprInParen expr_in_paren = 2;
    ExprSchemaTable schema_table = 3;
    ExprSchemaTableFn schema_table_fn = 4;
  }
}

message ExprSchemaTable {
  optional Schema schema_name = 1;
  required Table table_name = 2;
}

message ExprSchemaTableFn {
  required TableFn table_fn = 2;
  // FIXME in the future add more. For now, no exprs.
}

message ExprInParen {
  oneof exprin_paren_oneof {
    Select select = 1;
    ExprList exprs = 2;
  }  // if zero, can just emit closed parentheses
}

message ExprList {
  required Expr expr = 1;
  repeated Expr extra_exprs = 2;
}

message Expr1 {
  required Expr expr1 = 5;
  required bool not = 1;
  enum PossibleKeywords {
    LIKE = 0;
    GLOB = 1;
    REGEXP = 2;
    MATCH = 3;
  }
  required PossibleKeywords keyword = 2;
  required Expr expr2 = 3;
  optional Expr escape_expr = 4;  // CORPUS specialize?
}

message ExprNullTests {
  required Expr expr = 1;
  enum PossibleKeywords {
    ISNULL = 0;
    NOTNULL = 1;
    NOT_NULL = 2;
  }
  required PossibleKeywords keyword = 2;
}

message ExprIs {
  required bool not = 1;
  required Expr expr1 = 2;
  required Expr expr2 = 3;
}

message ExprBetween {
  required bool not = 1;
  required Expr expr1 = 2;
  required Expr expr2 = 3;
  required Expr expr3 = 4;
}

enum CollateType {
  COLLATE_BINARY = 1;
  COLLATE_NOCASE = 2;
  COLLATE_RTRIM = 3;
}

message CollateExpr {
  required Expr expr = 1;
  required CollateType collate_type = 2;
}

message CastTypeName {
  enum CastTypeNameEnum {
    BLOB = 0;
    TEXT = 1;
    REAL = 2;
    INTEGER = 3;
    NUMERIC = 4;
  }
  required CastTypeNameEnum type_enum = 1;
}

message CastExpr {
  required Expr expr = 1;
  required CastTypeName type_name = 2;
}

message ParenthesesExpr {
  required Expr expr = 1;
  repeated Expr other_exprs = 2;
}

message Fn {
  oneof fn_oneof {
    SimpleFn simple_fn = 1;
    FTS3AuxiliaryFn fts_aux_fn = 2;
    DateAndTimeFn dat_fn = 3;
    AggregateFn aggregate_fn = 4;
    Printf printf = 5;
  }
  // FIXME in the future: JSON functions. Not used in Chrome.
}

// Aggregate FNs
message AggregateFn {
  optional bool count_star = 7;
  enum FnName {
    AVG = 0;
    COUNT = 1;
    GROUP_CONCAT = 2;
    MAX = 3;
    MIN = 4;
    SUM = 5;
    TOTAL = 6;
  }
  required FnName fn_name = 6;
  required bool distinct = 5;
  optional Column col1 = 1;
  optional Column col2 = 2;
  required Expr expr1 = 3;
  optional Expr expr2 = 4;
}

// Date and Time Functions
message DateAndTimeFn {
  optional SimpleDateAndTimeFn simple = 1;
  required StrftimeFn strftime = 2;
}

message StrftimeFn {
  repeated StrftimeFormat fmts = 1;
  required TimeString time_string = 2;
  repeated TimeModifier modifiers = 3;
}

message StrftimeFormat {
  enum Substitution {
    D = 0;
    F = 1;
    H = 2;
    J = 3;
    M = 4;
    S = 5;
    W = 6;
    Y = 7;
  }
  required bool lowercase = 1;
  optional Substitution subs = 2;
  required string bytes = 3;
}

message SimpleDateAndTimeFn {
  enum FnName {
    DATE = 0;
    TIME = 1;
    DATETIME = 2;
    JULIANDAY = 3;
  }
  required FnName fn_name = 1;
  required TimeString time_string = 2;
  repeated TimeModifier modifiers = 3;
}

message HoursStuff {
  optional uint32 hh = 1;
  optional uint32 mm = 2;
  optional uint32 ss = 3;
  optional uint32 sss = 4;
}

message TimeString {
  optional uint32 yyyy = 1;
  optional uint32 mm = 2;
  optional uint32 dd = 3;
  optional HoursStuff hs = 4;
  required bool extra_t = 6;

  optional uint32 dddddddddd = 7;

  required bool now = 8;

  optional string random_bytes = 9;

  required bool z = 13;
  required bool plus = 14;
  optional bool tz_plus = 10;
  optional uint32 tz_hh = 11;
  optional uint32 tz_mm = 12;
}

message TimeModifier {
  enum NumberedModifiers {
    DAYS = 0;
    HOURS = 1;
    MINUTES = 2;
    SECONDS = 3;
    MONTHS = 4;
    YEARS = 5;
  }
  required uint32 num = 4;
  optional uint32 dot_num = 5;
  optional NumberedModifiers nm = 1;
  enum OtherModifiers {
    START_OF_MONTH = 0;
    START_OF_YEAR = 1;
    START_OF_DAY = 2;
    WEEKDAY = 3;
    UNIXEPOCH = 4;
    LOCALTIME = 5;
    UTC = 6;
  }
  required OtherModifiers om = 2;
  required uint32 weekday = 3;
}

// Simple Functions finish
message SimpleFn {
  oneof simple_fn_oneof {
    ZeroArgFn zero_arg_fn = 1;
    OneArgFn one_arg_fn = 2;
    TwoArgFn two_arg_fn = 3;
    ThreeArgFn three_arg_fn = 4;
    VarNumFn varnum_fn = 5;
    CharFn char_fn = 6;
  }
}

// Zero arguments fn
enum ZeroArgFn {
  ZFN_CHANGES = 0;
  ZFN_LAST_INSERT_ROWID = 1;
  ZFN_RANDOM = 2;
  ZFN_SQLITE_SOURCE_ID = 3;
  ZFN_SQLITE_VERSION = 4;
  ZFN_TOTAL_CHANGES = 5;
}

message OneArgFn {
  enum OneArgFnEnum {
    ABS = 1;
    HEX = 2;
    LENGTH = 3;
    LIKELY = 4;
    LOWER = 5;
    LTRIM = 6;
    LOAD_EXTENSION = 7;
    QUOTE = 8;
    ROUND = 9;
    RTRIM = 10;
    RANDOMBLOB = 11;
    SOUNDEX = 12;
    SQLITE_COMPILE_OPTION_GET = 13;
    SQLITE_COMPILE_OPTION_USED = 14;
    SQLITE_OFFSET = 15;
    TRIM = 16;
    TYPEOF = 17;
    UNICODE_ = 18;
    UNLIKELY = 19;
    UPPER = 20;
    ZEROBLOB = 21;
  }
  required OneArgFnEnum fn_enum = 1;
  required Expr arg1 = 2;
}

message TwoArgFn {
  enum TwoArgFnEnum {
    GLOB = 1;
    IFNULL = 2;
    INSTR = 3;
    LIKE = 4;
    LIKELIHOOD = 5;
    LOAD_EXTENSION = 6;
    LTRIM = 7;
    NULLIF = 8;
    ROUND = 9;
    RTRIM = 10;
    SUBSTR = 11;
    TRIM = 12;
  }
  required TwoArgFnEnum fn_enum = 1;
  required Expr arg1 = 2;
  required Expr arg2 = 3;
}

message ThreeArgFn {
  enum ThreeArgFnEnum {
    LIKE = 0;
    REPLACE = 1;
    SUBSTR = 2;
  }
  required ThreeArgFnEnum fn_enum = 1;
  required Expr arg1 = 2;
  required Expr arg2 = 3;
  required Expr arg3 = 4;
}

// Fns with two args required + an arbitrary number of other args.
message VarNumFn {
  enum VarNumFnEnum {
    COALESCE = 1;
    MAX = 2;
    MIN = 3;
  }
  required VarNumFnEnum fn_enum = 1;
  required Expr arg1 = 2;
  required Expr arg2 = 3;
  repeated Expr other_args = 4;
}

message CharFn {
  required uint64 char = 1;
  repeated uint64 extra_chars = 2;
}

message Printf {
  repeated string strings = 1;
  repeated PrintfFormatSpecifier specifiers = 2;
  repeated Expr exprs = 3;
}

message PrintfFormatSpecifier {
  enum Flags {
    MINUS = 0;
    PLUS = 1;
    SPACE = 2;
    ZERO = 3;
    HASH = 4;
    COMMA = 5;
    BANG = 6;
  }
  repeated Flags flags = 1;
  optional uint32 precision = 2;
  optional uint32 width = 3;
  optional bool width_star = 4;
  // The length modifiers make no difference for the sqlite function.
  optional uint32 length = 5;
  enum SubType {
    I = 0;
    D = 1;
    U = 2;
    F = 3;
    E = 4;
    G = 5;
    X = 6;
    O = 7;
    S = 8;
    Z = 9;
    C = 10;
    P = 11;
    N = 12;
    Q = 13;
    W = 14;
  }
  optional bool percent = 8;
  required SubType sub_type = 6;
  required bool lowercase = 7;
}

// Window fns!
// FIXME in the future: this may only be in the result set and the ORDER_BY
// clause of a SELECT statement.
message WindowFnInvocation {
  oneof window_fn_oneof {
    ZeroArgWinFn zero_arg_fn = 1;
    OneArgWinFn one_arg_fn = 2;
    NthValueFn nth_value_fn = 4;
    MiscWinFn misc_fn = 3;
  }

  // Used if the above oneof produces zero values.
  required ZeroArgWinFn fallback_zero_arg_fn = 5;

  // FIXME in the future: may just replace with a WHERE expr.
  optional ExprFilter expr_filter = 6;
  optional WindowDefn win_defn = 7;
  required WindowName win_name = 8;  // used only if win_defn above is unused.
}

message WindowDefn {
  // FIXME in the future: change to specialized expr
  optional ExprList partition_exprs = 1;
  optional MultipleOrderingTerm ordering_terms = 2;
  optional ExprFrameSpec frame_spec = 3;
}

message ExprFrameSpec {
  enum RangeRows {
    RANGE = 0;
    ROWS = 1;
  }
  required RangeRows range_rows = 1;
  required FrameSpecSubExpr left_expr = 2;
  // if this exists, then this is a BETWEEN statement.
  optional FrameSpecSubExprRight right_expr = 3;
}

message FrameSpecSubExpr {
  enum Which {
    UNBOUNDED_PRECEDING = 0;
    EXPR_PRECEDING = 1;
    CURRENT_ROW = 2;
    EXPR_FOLLOWING = 3;
  }
  required Which which = 1;
  // only used if which is EXPR_PRECEDING or EXPR_FOLLOWING
  optional Expr expr = 2;  // CORPUS specialize? integer?
}

message FrameSpecSubExprRight {
  enum Which {
    UNBOUNDED_FOLLOWING = 0;
    EXPR_PRECEDING = 1;
    CURRENT_ROW = 2;
    EXPR_FOLLOWING = 3;
  }
  required Which which = 1;
  // only used if which is EXPR_PRECEDING or EXPR_FOLLOWING
  optional Expr expr = 2;  // CORPUS specialize? integer?
}

message MultipleOrderingTerm {
  repeated ExprOrderingTerm terms = 1;
}

message ExprOrderingTerm {
  required Expr expr = 1;  // CORPUS specialize???
  optional CollateType collate_type = 2;
  required AscDesc asc_desc = 3;
}

enum AscDesc {
  ASCDESC_NONE = 0;
  ASCDESC_ASC = 1;
  ASCDESC_DESC = 2;
}

message ExprFilter {
  required Expr expr = 1;
}

message ZeroArgWinFn {
  enum ZeroArgWinFnEnum {
    ROW_NUMBER = 0;
    RANK = 1;
    DENSE_RANK = 2;
    PERCENT_RANK = 3;
    CUME_DIST = 4;
  }
  required ZeroArgWinFnEnum win_fn = 1;
}

message OneArgWinFn {
  enum OneArgWinFnEnum {
    NTILE = 0;
    FIRST_VALUE = 1;
    LAST_VALUE = 2;
  }
  required OneArgWinFnEnum win_fn = 1;
  required Expr expr = 2;
}

message NthValueFn {
  required Expr expr = 1;
  required uint32 row_num = 2;
}

message MiscWinFn {
  enum MiscWinFnEnum {
    LAG = 0;
    LEAD = 1;
  }
  required MiscWinFnEnum fn = 1;
  required Expr expr = 2;
  optional uint32 offset = 3;
  optional Expr default = 4;  // unused if offset non-existent
}
// FIXME in the future: all aggregate functions may be used as window aggregate
// functions!

// Table-valued FNs, including pragma table-valued fns.
enum PragmaFnZeroArgOneResult {
  PFN_ZO_APPLICATION_ID = 0;
  PFN_ZO_AUTO_VACUUM = 1;
  PFN_ZO_AUTOMATIC_INDEX = 2;
  PFN_ZO_BUSY_TIMEOUT = 3;
  PFN_ZO_CACHE_SIZE = 4;
  PFN_ZO_CACHE_SPILL = 5;
  PFN_ZO_CELL_SIZE_CHECK = 6;
  PFN_ZO_CHECKPOINT_FULL_FSYNC = 7;
  PFN_ZO_COLLATION_LIST = 8;
  PFN_ZO_COMPILE_OPTIONS = 9;
  PFN_ZO_COUNT_CHANGES = 10;
  PFN_ZO_DATA_VERSION = 11;
  PFN_ZO_DATABASE_LIST = 12;
  PFN_ZO_DEFAULT_CACHE_SIZE = 13;
  PFN_ZO_DEFER_FOREIGN_KEYS = 14;
  PFN_ZO_EMPTY_RESULT_CALLBACKS = 15;
  PFN_ZO_ENCODING = 16;
  PFN_ZO_FOREIGN_KEY_CHECK = 17;
  PFN_ZO_FOREIGN_KEYS = 18;
  PFN_ZO_FREELIST_COUNT = 19;
  PFN_ZO_FULL_COLUMN_NAMES = 20;
  PFN_ZO_FULLFSYNC = 21;
  PFN_ZO_FUNCTION_LIST = 22;
  PFN_ZO_IGNORE_CHECK_CONSTRAINTS = 23;
  PFN_ZO_INTEGRITY_CHECK = 24;
  PFN_ZO_JOURNAL_SIZE_LIMIT = 25;
  PFN_ZO_LEGACY_ALTER_TABLE = 26;
  PFN_ZO_LEGACY_FILE_FORMAT = 27;
  PFN_ZO_LOCK_STATUS = 28;
  PFN_ZO_LOCKING_MODE = 29;
  PFN_ZO_MAX_PAGE_COUNT = 30;
  PFN_ZO_MODULE_LIST = 31;
  PFN_ZO_PAGE_COUNT = 32;
  PFN_ZO_PAGE_SIZE = 33;
  PFN_ZO_PRAGMA_LIST = 34;
  PFN_ZO_QUERY_ONLY = 35;
  PFN_ZO_QUICK_CHECK = 36;
  PFN_ZO_READ_UNCOMMITTED = 37;
  PFN_ZO_RECURSIVE_TRIGGERS = 38;
  PFN_ZO_REVERSE_UNORDERED_SELECTS = 39;
  PFN_ZO_SCHEMA_VERSION = 40;
  PFN_ZO_SECURE_DELETE = 41;
  PFN_ZO_SHORT_COLUMN_NAMES = 42;
  PFN_ZO_SOFT_HEAP_LIMIT = 43;
  PFN_ZO_SQL_TRACE = 44;
  PFN_ZO_STATS = 45;
  PFN_ZO_SYNCHRONOUS = 46;
  PFN_ZO_TEMP_STORE = 47;
  PFN_ZO_THREADS = 48;
  PFN_ZO_USER_VERSION = 49;
  // omit vdbe_* debug pragmas
  PFN_ZO_WRITEABLE_SCHEMA = 50;
}

message TableFn {
  required PragmaFnZeroArgOneResult no_arg_one_result = 10;
  oneof pragma_fn_oneof {
    Table foreign_key_list = 1;
    Index index_info = 2;
    Table index_list = 3;
    Index index_xinfo = 4;
    uint32 integrity_check = 5;
    uint32 optimize = 6;
    uint32 quick_check = 7;
    Table table_info = 8;
    Table table_xinfo = 9;
  }
}