xo-tokenizer xo-reader: + bool literals + if-expr parsing

This commit is contained in:
Roland Conybeare 2025-07-22 07:20:56 -05:00
commit 3573ec800a
23 changed files with 712 additions and 53 deletions

View file

@ -12,6 +12,18 @@ namespace xo {
/** @class expect_expr_xs
* @brief state machine to expect + capture an expression
*
* Examples:
* @text
* def x : i64 = 5 ; // with allow_defs
* lambda (f : x64) { ... } ;
* if (prime(x)) { ... } ;
* 5 + x ;
* @endtext
*
* top exprstate when expect_expr_xs::start() invoked
* will receive parsed expression via
* exprstate::on_expr() or exprstate::on_expr_with_semicolon().
**/
class expect_expr_xs : public exprstate {
public:
@ -41,6 +53,9 @@ namespace xo {
virtual void on_symbol_token(const token_type & tk,
parserstatemachine * p_psm) override;
virtual void on_bool_token(const token_type & tk,
parserstatemachine * p_psm) override;
virtual void on_i64_token(const token_type & tk,
parserstatemachine * p_psm) override;
@ -50,7 +65,7 @@ namespace xo {
/** update exprstate in response to a successfully-parsed subexpression **/
virtual void on_expr(bp<Expression> expr,
parserstatemachine * p_psm) override;
/** update exprstate in response to a successfully-parsed subexpression,
/** update exprstate in response to a successfully-parsed subexpression
* that's terminated by semicolon ';'
**/
virtual void on_expr_with_semicolon(bp<Expression> expr,

View file

@ -17,8 +17,6 @@ namespace xo {
* This used for top-level expressions in a translation unit.
**/
toplevel_batch,
/** nested sequence, for example in function body **/
nested,
};
/** @class exprseq_xs
@ -27,14 +25,14 @@ namespace xo {
* expression sequences come in several types:
* 1. top-level interactive
* 2. top-level batch
* 3. nested
*
* @text
* 1 2 3
* +--------
* def | y y y
* symbol | y n n 1: evaluate as variable
* i64 | y n n 1: evaluate as constant
* 1 2
* +-----
* def | y y
* if | y n 1: eval as expression
* symbol | y n 1: evaluate as variable
* i64 | y n 1: evaluate as constant
*
* @endtext
**/
@ -45,12 +43,18 @@ namespace xo {
static void start(exprseqtype seqtype, parserstatemachine * p_psm);
public:
virtual const char * get_expect_str() const override;
// ----- token input methods -----
virtual void on_def_token(const token_type & tk,
parserstatemachine * p_psm) override;
virtual void on_if_token(const token_type & tk,
parserstatemachine * p_psm) override;
virtual void on_symbol_token(const token_type & tk,
parserstatemachine * p_psm) override;
virtual void on_bool_token(const token_type & tk,
parserstatemachine * p_psm) override;
virtual void on_i64_token(const token_type & tk,
parserstatemachine * p_psm) override;

View file

@ -45,6 +45,11 @@ namespace xo {
**/
let1expr,
/** handle if-else expression
* see @ref if_else_xs
**/
ifexpr,
expect_rhs_expression,
expect_symbol,
expect_type,
@ -177,6 +182,22 @@ namespace xo {
virtual void on_operator_token(const token_type & tk,
parserstatemachine * p_psm);
/** handle incoming if-keyword token **/
virtual void on_if_token(const token_type & tk,
parserstatemachine * p_psm);
/** handle incoming then-keyword token **/
virtual void on_then_token(const token_type & tk,
parserstatemachine * p_psm);
/** handle incoming then-keyword token **/
virtual void on_else_token(const token_type & tk,
parserstatemachine * p_psm);
/** handle incoming bool-literal token **/
virtual void on_bool_token(const token_type & tk,
parserstatemachine * p_psm);
/** handle incoming integer-literal token **/
virtual void on_i64_token(const token_type & tk,
parserstatemachine * p_psm);

View file

@ -0,0 +1,97 @@
/** @file if_else_xs.hpp
*
* author: Roland Conybeare, Jul 2025
**/
#pragma once
#include "xo/expression/IfExpr.hpp"
#include "exprstate.hpp"
#include "xo/indentlog/print/ppdetail_atomic.hpp"
namespace xo {
namespace scm {
/**
* if test-expr then then-expr else else-expr ;
* ^ ^ ^ ^ ^ ^ ^
* | | | | | | |
* | if_1 if_2 if_3 if_4 if_5 if_6
* if_0
*
* if_0 --on_if_token()--> if_1
* if_1 --on_expr()--> if_2
* if_2 --on_then_token()--> if_3
* if_3 --on_expr()--> if_4
* if_4 --on_else_token()--> if_5
* --on_semicolon_token()--> (done)
* if_5 --on_expr()-->if_6
* if_6 --on_semicolon_token()--> (done)
**/
enum class ifexprstatetype {
invalid = -1,
if_0,
if_1,
if_2,
if_3,
if_4,
if_5,
if_6,
n_ifexprstatetype,
};
extern const char * ifexprstatetype_descr(ifexprstatetype x);
std::ostream &
operator<<(std::ostream & os, ifexprstatetype x);
/** @class if_else_xs
* @brief state to provide parsing of a conditional expression
**/
class if_else_xs : public exprstate {
public:
using IfExprAccess = xo::ast::IfExprAccess;
public:
if_else_xs(rp<IfExprAccess> if_expr);
virtual ~if_else_xs() = default;
static const if_else_xs * from(const exprstate * x) { return dynamic_cast<const if_else_xs *>(x); }
static void start(parserstatemachine * p_psm);
// ----- inherited from exprstate -----
virtual const char * get_expect_str() const override;
virtual void on_if_token(const token_type & tk,
parserstatemachine * p_psm) override;
virtual void on_then_token(const token_type & tk,
parserstatemachine * p_psm) override;
virtual void on_else_token(const token_type & tk,
parserstatemachine * p_psm) override;
virtual void on_semicolon_token(const token_type & tk,
parserstatemachine * p_psm) override;
virtual void on_expr(bp<Expression> expr,
parserstatemachine * p_psm) override;
virtual void on_expr_with_semicolon(bp<Expression> expr,
parserstatemachine * p_psm) override;
virtual void print(std::ostream & os) const override;
private:
static std::unique_ptr<if_else_xs> make();
private:
ifexprstatetype ifxs_type_;
/** scaffold output expression here **/
rp<IfExprAccess> if_expr_;
};
} /*namespace scm*/
} /*namespace xo*/
/** end if_else_xs.hpp **/

View file

@ -112,6 +112,9 @@ namespace xo {
* | symbol-literal
* | struct-literal
*
*
* boolean-literal = true | false
*
* variable-expr = $varname
* apply-expr = fn-expr(arg-expr(1), .., arg-expr(n))
* fn-expr = expression
@ -155,8 +158,10 @@ namespace xo {
public:
/** create parser in initial state;
* parser is ready to receive tokens via @ref include_token
*
* @p debug_flag true to enable debug logging
**/
parser();
explicit parser(bool debug_flag);
/** true if parser is at top-level, i.e. ready for next top-level expression **/
bool is_at_toplevel() const { return stack_size() == 0; }
@ -243,6 +248,9 @@ namespace xo {
/** parser result state **/
parser_result result_;
/** enable/disable debug logging **/
bool debug_flag_;
}; /*parser*/
inline std::ostream &

View file

@ -28,10 +28,12 @@ namespace xo {
public:
parserstatemachine(exprstatestack * p_stack,
envframestack * p_env_stack,
parser_result * p_result)
parser_result * p_result,
bool debug_flag)
: p_stack_{p_stack},
p_env_stack_{p_env_stack},
p_result_{p_result}
p_result_{p_result},
debug_flag_{debug_flag}
{}
//const parser_result & result() const { return result_; }
@ -44,6 +46,8 @@ namespace xo {
exprstate & top_exprstate();
void push_exprstate(std::unique_ptr<exprstate> x);
bool debug_flag() const { return debug_flag_; }
/** lookup variable name in lexical context represented by
* this psm. nullptr if not found
**/
@ -54,9 +58,13 @@ namespace xo {
**/
void upsert_var(bp<Variable> x);
/** @return available variable bindings in current parsing state **/
bp<LocalEnv> top_envframe() const;
/** push frame @p x (with new variable bindings) onto environment stack **/
void push_envframe(const rp<LocalEnv> & x);
/** @return pop innermost environment frame and return it **/
rp<LocalEnv> pop_envframe();
/** @return number of stacked environment frames **/
size_t env_stack_size() const { return p_env_stack_->size(); }
// ----- parsing outputs -----
@ -71,6 +79,8 @@ namespace xo {
void on_operator_token(const token_type & tk);
void on_leftbrace_token(const token_type & tk);
void on_rightbrace_token(const token_type & tk);
void on_then_token(const token_type & tk);
void on_else_token(const token_type & tk);
// ----- parsing error -----
@ -91,6 +101,8 @@ namespace xo {
envframestack * p_env_stack_ = nullptr;
/** parser result object **/
parser_result * p_result_ = nullptr;
/** enable debug logging **/
bool debug_flag_ = false;
};
inline std::ostream &

View file

@ -95,6 +95,10 @@ namespace xo {
parserstatemachine * p_psm) override;
virtual void on_rightparen_token(const token_type & tk,
parserstatemachine * p_psm) override;
virtual void on_then_token(const token_type & tk,
parserstatemachine * p_psm) override;
virtual void on_else_token(const token_type & tk,
parserstatemachine * p_psm) override;
/* entry point for an infix operator token */
virtual void on_operator_token(const token_type & tk,
@ -121,7 +125,7 @@ namespace xo {
*
* where f determined by @ref op_type_
**/
rp<Expression> assemble_expr();
rp<Expression> assemble_expr(parserstatemachine * p_psm);
private:
/** populate an expression here, may be followed by an operator **/

View file

@ -78,7 +78,7 @@ namespace xo {
using span_type = tokenizer_type::span_type;
public:
reader() = default;
explicit reader(bool debug_flag);
/** call once before calling .read_expr()
* for a new interactive session