fix support for pgsql ANALYZE, DO, VACUUM, GRANT/REVOKE (to the extent possible)

Author: richarah

README.md

@@ -10,7 +10,9 @@ Supports stored procedures (PL/pgSQL, T-SQL, MySQL, PL/SQL): where sqlglot falls
 
 Anywhere SQL hits a hot path: proxies, sidecars, migration tools, linters, etc. Also a replacement for every regex that's pretending to be a parser.
 
-#### Inspiration
+#### Why this exists
+
+libsqlglot was born out of a gap in the C++ ecosystem: the lack of native tooling for high-speed, hassle-free parsing & transpiling between dozens of SQL dialects.
 
 Inspired by the original [sqlglot](https://github.com/tobymao/sqlglot), which did the decade-long work of mapping 31+ SQL dialects into an elegant, universal AST. libsqlglot does the comparatively trivial work of compiling it: the algorithm was already O(n), the runtime wasn't.
 
@@ -216,7 +218,7 @@ cmake --build build
 
 ## Architecture
 
-Header-only design: you only pay for what you use. 19 header files, no `.cpp`. See `include/libsqlglot/` for the full layout. Core files: `parser.h` (2958 lines), `generator.h` (1643), `expression.h` (1105, 105 expression types). Entry point is `transpiler.h` (86 lines).
+Header-only design: you only pay for what you use. 19 header files, no `.cpp`. See `include/libsqlglot/` for the full layout. Core files: `parser.h` (3479 lines), `generator.h` (1913), `expression.h` (1248, 115 expression types). Entry point is `transpiler.h` (86 lines).
 
 ### Memory management
 

include/libsqlglot/dialect_reflection.h

@@ -33,7 +33,8 @@ namespace libsqlglot::dialects {
 
 /// Find dialect by name (case-insensitive) - runtime function
 /// Uses compile-time generated mappings for zero-maintenance lookups
-[[nodiscard]] inline Dialect from_name(std::string_view name) {
+/// Returns ANSI dialect if name is unknown (fail-safe, not fail-fast)
+[[nodiscard]] inline Dialect from_name(std::string_view name) noexcept {
     // Convert to lowercase for comparison
     auto to_lower = [](char c) { return (c >= 'A' && c <= 'Z') ? (c - 'A' + 'a') : c; };
 
@@ -66,7 +67,8 @@ namespace libsqlglot::dialects {
     if (lower_name == "tsql" || lower_name == "sqlserver") return Dialect::SQLServer;
     if (lower_name == "cockroach" || lower_name == "cockroachdb") return Dialect::CockroachDB;
 
-    throw std::runtime_error("Unknown dialect: " + std::string(name));
+    // Unknown dialect - return ANSI as safe default (fail-safe)
+    return Dialect::ANSI;
 }
 
 /// Get dialect name as string

include/libsqlglot/expression.h

@@ -91,6 +91,10 @@ enum class ExprType : uint16_t {
     SHOW_STMT,
     DESCRIBE_STMT,
     EXPLAIN_STMT,
+    ANALYZE_STMT,       // ANALYZE table
+    VACUUM_STMT,        // VACUUM table
+    GRANT_STMT,         // GRANT privileges
+    REVOKE_STMT,        // REVOKE privileges
 
     // Advanced SELECT features
     CTE,                // Common Table Expression (WITH clause)
@@ -132,6 +136,7 @@ enum class ExprType : uint16_t {
     CONTINUE_STMT,      // CONTINUE statement
     RETURN_STMT,        // RETURN expression
     BEGIN_END_BLOCK,    // BEGIN ... END block (T-SQL, MySQL, PL/SQL)
+    DO_BLOCK,           // DO $$ ... $$ (PostgreSQL anonymous block)
     EXCEPTION_BLOCK,    // EXCEPTION block (PL/pgSQL, PL/SQL)
     ASSIGNMENT_STMT,    // Variable assignment (SET var = val, var := val)
     DELIMITER_STMT,     // DELIMITER command (MySQL)
@@ -149,6 +154,13 @@ enum class ExprType : uint16_t {
     CREATE_TABLESPACE,  // CREATE TABLESPACE
     CREATE_INDEX_ADV,   // Advanced CREATE INDEX (partial, expression, concurrent)
 
+    // BigQuery ML & Advanced Analytics
+    CREATE_MODEL,       // CREATE MODEL (BigQuery ML)
+    DROP_MODEL,         // DROP MODEL
+    ML_PREDICT,         // ML.PREDICT() function
+    ML_EVALUATE,        // ML.EVALUATE() function
+    ML_TRAINING_INFO,   // ML.TRAINING_INFO() function
+
     // Other
     ALIAS,
     CAST,
@@ -1102,4 +1114,135 @@ struct CreateIndexAdvStmt : Expression {
         : Expression(ExprType::CREATE_INDEX_ADV), where_clause(nullptr) {}
 };
 
+// ============================================================================
+// NEW UTILITY & PERMISSIONS STATEMENTS
+// ============================================================================
+
+/// DO block (PostgreSQL anonymous code block)
+struct DoBlockStmt : Expression {
+    std::string language;                         // plpgsql, sql, etc.
+    std::vector<Expression*> statements;          // Block body
+
+    DoBlockStmt()
+        : Expression(ExprType::DO_BLOCK) {}
+};
+
+/// ANALYZE statement
+struct AnalyzeStmt : Expression {
+    std::string table;                            // Optional table name
+    std::vector<std::string> columns;             // Optional column list
+
+    AnalyzeStmt()
+        : Expression(ExprType::ANALYZE_STMT) {}
+};
+
+/// VACUUM statement (PostgreSQL-specific)
+struct VacuumStmt : Expression {
+    std::string table;                            // Optional table name
+    std::vector<std::string> columns;             // Optional column list (requires ANALYZE)
+
+    // Boolean options
+    bool full = false;
+    bool freeze = false;
+    bool verbose = false;
+    bool analyze = false;
+    bool disable_page_skipping = false;
+    bool skip_locked = false;
+    bool truncate = true;                         // Default is true
+
+    // Integer options
+    int parallel_workers = -1;                    // -1 means not specified
+    int buffer_usage_limit = -1;                  // -1 means not specified (in KB)
+
+    // Index cleanup: true/false/auto
+    enum class IndexCleanup { AUTO, ON, OFF };
+    IndexCleanup index_cleanup = IndexCleanup::AUTO;
+
+    // Use parenthesized syntax (PostgreSQL 9.0+) vs legacy syntax
+    bool use_parenthesized_syntax = true;
+
+    VacuumStmt()
+        : Expression(ExprType::VACUUM_STMT) {}
+};
+
+/// GRANT statement
+struct GrantStmt : Expression {
+    enum class PrivilegeType { SELECT, INSERT, UPDATE, DELETE, ALL, EXECUTE, USAGE };
+
+    std::vector<PrivilegeType> privileges;
+    std::string object_type;                      // TABLE, DATABASE, SCHEMA, FUNCTION, etc.
+    std::string object_name;
+    std::vector<std::string> grantees;            // Users/roles to grant to
+    bool with_grant_option = false;
+
+    GrantStmt()
+        : Expression(ExprType::GRANT_STMT) {}
+};
+
+/// REVOKE statement
+struct RevokeStmt : Expression {
+    using PrivilegeType = GrantStmt::PrivilegeType;
+
+    std::vector<PrivilegeType> privileges;
+    std::string object_type;
+    std::string object_name;
+    std::vector<std::string> grantees;
+    bool cascade = false;
+
+    RevokeStmt()
+        : Expression(ExprType::REVOKE_STMT) {}
+};
+
+// ============================================================================
+// BIGQUERY ML
+// ============================================================================
+
+/// CREATE MODEL statement (BigQuery ML)
+struct CreateModelStmt : Expression {
+    std::string model_name;
+    std::string model_type;                       // LINEAR_REG, LOGISTIC_REG, etc.
+    std::vector<std::pair<std::string, std::string>> options;  // Model options
+    SelectStmt* training_query;                   // Training data query
+    bool or_replace = false;
+    bool if_not_exists = false;
+
+    CreateModelStmt()
+        : Expression(ExprType::CREATE_MODEL), training_query(nullptr) {}
+};
+
+/// DROP MODEL statement
+struct DropModelStmt : Expression {
+    std::string model_name;
+    bool if_exists = false;
+
+    DropModelStmt()
+        : Expression(ExprType::DROP_MODEL) {}
+};
+
+/// ML.PREDICT function
+struct MLPredictExpr : Expression {
+    std::string model_name;
+    SelectStmt* input_query;                      // Input data for prediction
+
+    MLPredictExpr()
+        : Expression(ExprType::ML_PREDICT), input_query(nullptr) {}
+};
+
+/// ML.EVALUATE function
+struct MLEvaluateExpr : Expression {
+    std::string model_name;
+    SelectStmt* evaluation_query;                 // Evaluation data
+
+    MLEvaluateExpr()
+        : Expression(ExprType::ML_EVALUATE), evaluation_query(nullptr) {}
+};
+
+/// ML.TRAINING_INFO function
+struct MLTrainingInfoExpr : Expression {
+    std::string model_name;
+
+    MLTrainingInfoExpr()
+        : Expression(ExprType::ML_TRAINING_INFO) {}
+};
+
 } // namespace libsqlglot