// Copyright (c) Herb Sutter
// SPDX-License-Identifier: CC-BY-NC-ND-4.0
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
// THE SOFTWARE.
//===========================================================================
// cppfront
//===========================================================================
#include "sema.h"
#include
#include
namespace cpp2 {
// Defined out of line here just to avoid bringing into the headers,
// so that we can't accidentally start depending on iostreams in the compiler body
auto cmdline_processor::print(std::string_view s, int width) -> void
{
if (width > 0) {
std::cout insert( emit_text_chunks_targets.back()->begin(), text_with_pos(std::string(s), pos) );
}
return;
}
// Otherwise, we'll actually print the string to the output file
// and update our curr_pos position
// Output the string
out void
{
if (curr_pos.colno > 1) {
auto old_pos = curr_pos;
print( "\n" );
assert(curr_pos.lineno == old_pos.lineno+1);
assert(curr_pos.colno == 1);
}
}
// Print a #line directive
//
auto print_line_directive( lineno_t line ) -> void
{
prev_line_info = { curr_pos.lineno, { } };
ensure_at_start_of_new_line();
// Not using print() here because this is transparent to the curr_pos
if (!flag_clean_cpp1) {
out identifier != n.identifier &&
// and this variable was uninitialized
!decl->initializer &&
// and it's either a non-parameter or an out parameter
(!decl->parameter || (decl->parameter && decl->parameter->pass == passing_style::out))
)
{
printer.print_cpp2(".value()", n.position());
}
}
if (add_std_move || add_std_forward) {
printer.print_cpp2(")", n.position());
}
}
//-----------------------------------------------------------------------
//
auto emit(qualified_id_node const& n) -> void
{
// Implicit "cpp2::" qualification of "unique.new" and "shared.new"
if (n.ids.size() == 2 &&
(*n.ids[0].id->identifier == "unique" || *n.ids[0].id->identifier == "shared") &&
*n.ids[1].scope_op == "." &&
*n.ids[1].id->identifier == "new"
)
{
printer.print_cpp2("cpp2::", n.position());
}
auto ident = std::string{};
printer.emit_to_string(&ident);
for (auto const& id : n.ids)
{
if (id.scope_op) {
emit(*id.scope_op);
}
emit(*id.id);
}
printer.emit_to_string();
printer.print_cpp2( ident, n.position() );
}
//-----------------------------------------------------------------------
//
auto emit(id_expression_node const& n, bool is_local_name = true) -> void
{
try_emit(n.id);
try_emit(n.id, false, is_local_name);
}
//-----------------------------------------------------------------------
//
auto emit(
compound_statement_node const& n,
std::vector const& function_prolog = {},
std::vector const& function_epilog = {},
colno_t function_indent = 1
)
-> void
{
auto pos = n.open_brace;
pos.lineno -= std::ssize(function_prolog);
printer.print_cpp2( "{", pos );
if (!function_prolog.empty()) {
printer.ignore_alignment( true, function_indent + 4 );
auto pos = source_position{};
if (!n.statements.empty()) {
pos = n.statements.front()->position();
}
for (auto& loc : function_prolog) {
printer.print_cpp2("\n", pos);
printer.print_cpp2(loc, pos);
}
printer.ignore_alignment( false );
}
for (auto const& x : n.statements) {
assert(x);
emit(*x);
}
if (!function_epilog.empty()) {
printer.ignore_alignment( true, function_indent + 4 );
auto pos = source_position{};
if (!n.statements.empty()) {
pos = n.statements.front()->position();
}
for (auto& loc : function_epilog) {
printer.print_cpp2("\n", pos);
printer.print_cpp2(loc, pos);
}
printer.ignore_alignment( false );
}
printer.print_cpp2( "}", n.close_brace );
}
//-----------------------------------------------------------------------
//
auto emit(inspect_expression_node const& n, bool is_expression) -> void
{
auto constexpr_qualifier = std::string{};
if (n.is_constexpr) {
constexpr_qualifier = "constexpr ";
}
// If this is an expression, it will have an explicit result type,
// and we need to start the lambda that we'll immediately invoke
auto result_type = std::string{};
if (is_expression) {
assert(n.result_type);
printer.emit_to_string(&result_type);
emit(*n.result_type);
printer.emit_to_string();
printer.print_cpp2("[&] () -> " + result_type + " ", n.position());
}
printer.print_cpp2("{ " + constexpr_qualifier + "auto&& __expr = ", n.position());
assert(n.expression);
emit(*n.expression);
printer.print_cpp2(";", n.position());
assert(n.identifier && *n.identifier == "inspect");
assert(!n.alternatives.empty());
auto found_wildcard = false;
for (auto first = true; auto&& alt : n.alternatives)
{
assert(alt && alt->is_as_keyword);
if (!first) {
printer.print_cpp2("else ", alt->position());
}
first = false;
auto id = std::string{};
printer.emit_to_string(&id);
assert(alt->id_expression);
emit(*alt->id_expression);
printer.emit_to_string();
assert (*alt->is_as_keyword == "is" || *alt->is_as_keyword == "as");
// TODO: pick up 'as' next, for now just do 'is'
if (*alt->is_as_keyword == "is")
{
// Stringize the expression-statement now...
auto statement = std::string{};
printer.emit_to_string(&statement);
emit(*alt->statement);
printer.emit_to_string();
// ... and jettison the final ; for an expression-statement
while (!statement.empty() && (statement.back() == ';' || isspace(statement.back()))) {
statement.pop_back();
}
// If this is an inspect-expression, we'll have to wrap each alternative
// in an 'if constexpr' so that its type is ignored for mismatches with
// the inspect-expression's type
auto return_prefix = std::string{};
auto return_suffix = std::string{";"}; // use this to tack the ; back on in the alternative body
if (is_expression) {
return_prefix = "{ if constexpr( requires{" + statement + ";} ) if constexpr( std::is_convertible_v ) return ";
return_suffix += " }";
}
if (id == "_") {
found_wildcard = true;
if (is_expression) {
printer.print_cpp2("return ", alt->position());
}
}
else {
printer.print_cpp2("if " + constexpr_qualifier + "(cpp2::is(__expr)) " + return_prefix, alt->position());
}
printer.print_cpp2(statement, alt->position());
if (is_expression && id != "_") {
assert(alt->statement->statement.index() == statement_node::expression);
printer.print_cpp2("; else return " + result_type + "{}", alt->position());
printer.print_cpp2("; else return " + result_type + "{}", alt->position());
}
printer.print_cpp2(return_suffix, alt->position());
}
else {
errors.emplace_back(
alt->position(),
"(temporary alpha limitation) cppfront is still learning 'inspect' - only simple 'is' alternatives are currently supported"
);
return;
}
}
if (is_expression) {
if (!found_wildcard) {
errors.emplace_back(
n.position(),
"an inspect expression must have an `is _` match-anything wildcard alternative"
);
return;
}
}
else {
printer.print_cpp2("}", n.close_brace);
}
// If this is an expression, finally actually invoke the lambda
if (is_expression) {
printer.print_cpp2("()", n.close_brace);
}
printer.print_cpp2("\n", n.close_brace);
}
//-----------------------------------------------------------------------
//
auto emit(selection_statement_node const& n) -> void
{
assert(n.identifier);
emit(*n.identifier);
printer.print_cpp2(" (", n.position());
printer.add_pad_in_this_line(1);
assert(n.expression);
emit(*n.expression);
printer.print_cpp2(") ", n.position());
printer.add_pad_in_this_line(1);
assert(n.true_branch);
emit(*n.true_branch);
if (n.has_source_false_branch) {
printer.print_cpp2("else ", n.else_pos);
emit(*n.false_branch);
}
}
//-----------------------------------------------------------------------
//
auto emit(iteration_statement_node const& n) -> void
{
assert(n.identifier);
// Handle while
//
if (*n.identifier == "while") {
assert(n.condition && n.statement && !n.range && !n.body);
// We emit Cpp2 while loops as Cpp2 for loops if there's a "next" clause
if (!n.next_expression) {
printer.print_cpp2("while( ", n.position());
emit(*n.condition);
}
else {
printer.print_cpp2("for( ; ", n.position());
emit(*n.condition);
printer.print_cpp2("; ", n.position());
printer.add_pad_in_this_line(-10);
emit(*n.next_expression);
}
printer.print_cpp2(" ) ", n.position());
emit(*n.statement);
return;
}
// Handle do
//
if (*n.identifier == "do") {
assert(n.condition && n.statement && !n.range && !n.body);
printer.print_cpp2("do ", n.position());
emit(*n.statement);
printer.print_cpp2(" while ( ", n.position());
emit(*n.condition);
if (n.next_expression) {
// Gotta say, this feels kind of nifty... short-circuit eval
// and smuggling work into a condition via a lambda, O my...
printer.print_cpp2(" && [&]{ ", n.position());
emit(*n.next_expression);
printer.print_cpp2(" ; return true; }() ", n.position());
}
printer.print_cpp2(");", n.position());
return;
}
// Handle for
//
if (*n.identifier == "for") {
assert(!n.condition && !n.statement && n.range && n.body);
// TODO: break n.range into subexpressions and lifetime-extend
// each subexpression, because that's the right thing to do
// For now, just use a for-scope auto&&
// Also: using the name 'cpp2_range' for now because '__range' is reserved
// and common enough that it might clash with existing impls
printer.print_cpp2("for ( auto&& cpp2_range = ", n.position());
emit(*n.range);
printer.print_cpp2("; ", n.position());
emit(*n.get_for_parameter());
printer.print_cpp2(" : cpp2_range ) ", n.position());
// If there's a next-expression, smuggle it in via a nested do/while(false) loop
// (nested "continue" will work, but "break" won't until we do extra work to implement
// that using a flag and implementing "break" as "__for_break = true; continue;")
if (n.next_expression) {
printer.print_cpp2(" { do ", n.position());
}
assert(n.body->initializer);
emit(*n.body->initializer);
if (n.next_expression) {
printer.print_cpp2(" while (false); ", n.position());
emit(*n.next_expression);
printer.print_cpp2("; }", n.position());
}
return;
}
assert(!"compiler bug: unexpected case");
}
//-----------------------------------------------------------------------
//
auto emit(return_statement_node const& n) -> void
{
assert(n.identifier);
assert(*n.identifier == "return");
printer.print_cpp2("return ", n.position());
// Return with expression == single anonymous return type
//
if (n.expression) {
emit(*n.expression);
if (function_returns.empty() || function_returns.back() != &single_anon) {
errors.emplace_back(
n.position(),
"return statement with expression must be in a function with a single anonymous return value"
);
return;
}
}
else if (!function_returns.empty() && function_returns.back() == &single_anon) {
errors.emplace_back(
n.position(),
"return statement must have an expression in a function with a single anonymous return value"
);
}
// Return without expression == zero or named return values
//
else if (!function_returns.empty() && function_returns.back()) {
//auto stmt = function_return_name + " { "; // we shouldn't need this with { } init
auto stmt = std::string(" { ");
auto& parameters = function_returns.back()->parameters;
for (bool first = true; auto& param : parameters) {
if (!first) {
stmt += ", ";
}
first = false;
assert(param->declaration->identifier);
printer.emit_to_string(&stmt);
emit(*param->declaration->identifier, true);
printer.emit_to_string();
}
stmt += " }";
printer.print_cpp2(stmt, n.position());
}
printer.print_cpp2("; ", n.position());
}
//-----------------------------------------------------------------------
//
auto build_capture_lambda_intro_for( capture_group& captures, source_position pos ) -> std::string
{
// First calculate the stringized version of each capture expression
// This will let us compare and de-duplicate repeated capture expressions
for (auto& cap : captures)
{
assert(cap.capture_expr->cap_grp == &captures);
printer.emit_to_string(&cap.str);
emit(*cap.capture_expr, true);
printer.emit_to_string();
}
// Then build the capture list, ignoring duplicated expressions
auto lambda_intro = std::string("[");
printer.emit_to_string(&lambda_intro);
auto num = 0;
auto handled = std::vector{};
for (auto& cap : captures)
{
// If we haven't handled a capture that looks like this one
if (std::find(handled.begin(), handled.end(), cap.str) == handled.end())
{
// Remember it
handled.push_back(cap.str);
// And handle it
if (num != 0) { // not first
lambda_intro += ", ";
}
printer.print_cpp2("_"+std::to_string(num)+" = ", pos);
emit(*cap.capture_expr, true);
}
++num;
}
printer.emit_to_string();
lambda_intro += "]";
return lambda_intro;
}
//-----------------------------------------------------------------------
//
auto emit(primary_expression_node const& n) -> void
{
try_emit(n.expr);
try_emit(n.expr);
try_emit(n.expr);
try_emit(n.expr, true);
if (n.expr.index() == primary_expression_node::declaration)
{
auto& decl = std::get(n.expr);
// The usual non-null assertion, plus it should be an anonymous function
assert(decl && !decl->identifier && decl->is(declaration_node::function));
auto lambda_intro = build_capture_lambda_intro_for(decl->captures, n.position());
emit(*decl, lambda_intro);
}
}
//-----------------------------------------------------------------------
//
auto emit(postfix_expression_node& n, bool for_lambda_capture = false) -> void
// note: parameter is deliberately not const because we we will fill
// in the capture .str information, and we may also adjust token
// column positions when moving operators to prefix notation
{
assert(n.expr);
last_postfix_expr_was_pointer = false;
// Check that this isn't pointer arithmentic
// (initial partial implementation)
if (n.expr->expr.index() == primary_expression_node::id_expression)
{
auto& id = std::get(n.expr->expr);
assert(id);
if (id->id.index() == id_expression_node::unqualified)
{
auto& unqual = std::get(id->id);
assert(unqual);
auto decl = sema.get_declaration_of(*unqual->identifier);
// TODO: Generalize this -- for now we detect only cases of the form "p: *int = ...;"
// We don't recognize pointer types that are deduced, multi-level, or from Cpp1
if (decl && decl->declaration && decl->declaration->pointer_declarator) {
if (n.ops.empty()) {
last_postfix_expr_was_pointer = true;
}
else
{
if (n.ops.front().op->type() == lexeme::PlusPlus ||
n.ops.front().op->type() == lexeme::MinusMinus ||
n.ops.front().op->type() == lexeme::LeftBracket
) {
errors.emplace_back(
n.ops.front().op->position(),
n.ops.front().op->to_string(true) + " - pointer arithmetic is illegal - use std::span or gsl::span instead"
);
violates_bounds_safety = true;
}
else if (n.ops.front().op->type() == lexeme::Tilde) {
errors.emplace_back(
n.ops.front().op->position(),
n.ops.front().op->to_string(true) + " - pointer bitwise manipulation is illegal - use std::bit_cast to convert to raw bytes first"
);
}
}
}
}
}
// Simple case: If there are no .ops, just emit the expression
if (n.ops.empty()) {
emit(*n.expr);
return;
}
// Check to see if it's a capture expression that contains $,
// and if we're not capturing the expression for the lambda
// introducer replace it with the capture name
auto captured_part = std::string{};
if (n.cap_grp && !for_lambda_capture)
{
// First stringize ourselves so that we compare equal against
// the first *cap_grp .str that matches us (which is what the
// lambda introducer generator used to create a lambda capture)
auto my_str = std::string{};
printer.emit_to_string(&my_str);
emit(n, true); // reentrant, but not in this 'if' because for_lambda_capture == true
printer.emit_to_string();
// Look in the capture group to see which capture # we are
auto mynum = 0;
for (auto const& cap : *n.cap_grp) {
if (cap.str == my_str) {
break;
}
++mynum;
}
assert (mynum < n.cap_grp->size() && "could not find this postfix-expression in capture group");
// And then emit that capture number
captured_part += "_" + std::to_string(mynum);
}
// Check to see if it's just a function call with "." syntax,
// and if so use this path to convert it to UFCS
if (// there's a single-token expression followed by . and (
n.expr->get_token() && // if the base expression is a single token
std::ssize(n.ops) >= 2 && // and we're of the form:
n.ops[0].op->type() == lexeme::Dot && // token . id-expr ( expr-list )
n.ops[1].op->type() == lexeme::LeftParen &&
// and either there's nothing after that, or there's just a $ after that
(
std::ssize(n.ops) == 2 ||
(std::ssize(n.ops) == 3 && n.ops[2].op->type() == lexeme::Dollar)
)
)
{
// If we already replaced this with a capture (which contains the UFCS
// work already done when the capture was computed), emit the capture
if (!captured_part.empty()) {
printer.print_cpp2(captured_part, n.position());
return;
}
// Otherwise, do the UFCS work...
// The . has its id_expr
assert (n.ops[0].id_expr);
// The ( has its expr_list and op_close
assert (n.ops[1].expr_list && n.ops[1].op_close);
//--------------------------------------------------------------------
// TODO: When MSVC supports __VA_OPT__ in standard mode without the
// experimental /Zc:preprocessor switch, use this single line
// instead of the dual lines below that special-case _0 args
// AND: Make the similarly noted change in cpp2util.h
//
//printer.print_cpp2("CPP2_UFCS(", n.position());
// If there are no additional arguments, use the CPP2_UFCS_0 version
if (!n.ops[1].expr_list->expressions.empty()) {
printer.print_cpp2("CPP2_UFCS(", n.position());
}
else {
printer.print_cpp2("CPP2_UFCS_0(", n.position());
}
//--------------------------------------------------------------------
// Make the "funcname" the first argument to CPP2_UFCS
emit(*n.ops[0].id_expr);
printer.print_cpp2(", ", n.position());
// Then make the base expression the second argument
emit(*n.expr);
// Then tack on any additional arguments
if (!n.ops[1].expr_list->expressions.empty()) {
printer.print_cpp2(", ", n.position());
push_need_expression_list_parens(false);
emit(*n.ops[1].expr_list);
pop_need_expression_list_parens();
}
printer.print_cpp2(")", n.position());
// And we're done. This path has handled this node, so return...
return;
}
// Otherwise, we're going to have to potentially do some work to change
// some Cpp2 postfix operators to Cpp1 prefix operators, so let's set up...
auto prefix = std::vector{};
auto suffix = std::vector{};
auto emitted_n = false;
auto last_was_prefixed = false;
auto saw_dollar = false;
for (auto i = n.ops.rbegin(); i != n.ops.rend(); ++i)
{
assert(i->op);
// If we already captured a part as a _## lambda capture,
// skip the part of this expression before the $ symbol
//
if (!captured_part.empty()) {
if (i->op->type() == lexeme::Dollar) {
break;
}
}
// Else skip the part of this expression after the $ symbol
else if (for_lambda_capture) {
if (i->op->type() == lexeme::Dollar) {
saw_dollar = true;
continue;
}
if (!saw_dollar) {
continue;
}
}
// Handle the Cpp2 postfix operators that are prefix in Cpp1
//
if (i->op->type() == lexeme::MinusMinus ||
i->op->type() == lexeme::PlusPlus ||
i->op->type() == lexeme::Multiply ||
i->op->type() == lexeme::Ampersand ||
i->op->type() == lexeme::Tilde
)
{
adjust_remaining_token_columns_on_this_line_visitor v(i->op->position(), 0 - i->op->length());
n.visit(v, 0);
if (!last_was_prefixed && i != n.ops.rbegin()) { // omit some needless parens
prefix.emplace_back( "(", i->op->position() );
}
prefix.emplace_back( i->op->to_string(true), i->op->position());
// Enable null dereference checks
if (flag_safe_null_pointers && i->op->type() == lexeme::Multiply) {
prefix.emplace_back( "cpp2::assert_not_null(", i->op->position() );
}
if (flag_safe_null_pointers && i->op->type() == lexeme::Multiply) {
suffix.emplace_back( ")", i->op->position() );
}
if (!last_was_prefixed && i != n.ops.rbegin()) { // omit some needless parens
suffix.emplace_back( ")", i->op->position() );
}
last_was_prefixed = true;
}
// Handle the suffix operators that remain suffix
//
else {
assert(i->op);
last_was_prefixed = false;
// Enable subscript bounds checks
if (flag_safe_subscripts && i->op->type() == lexeme::LeftBracket) {
suffix.emplace_back( ")", i->op->position() );
}
else if (i->op_close) {
suffix.emplace_back( i->op_close->to_string(true), i->op_close->position() );
}
if (i->id_expr) {
auto print = std::string{};
printer.emit_to_string(&print);
emit(*i->id_expr, false /*not a local name*/);
printer.emit_to_string();
suffix.emplace_back( print, i->id_expr->position() );
}
if (i->expr_list) {
auto text = std::vector{};
printer.emit_to_text_chunks(&text);
push_need_expression_list_parens(false);
emit(*i->expr_list);
pop_need_expression_list_parens();
printer.emit_to_text_chunks();
for (auto&& e: text) {
suffix.push_back(e);
}
}
// Enable subscript bounds checks
if (flag_safe_subscripts && i->op->type() == lexeme::LeftBracket) {
prefix.emplace_back( "cpp2::assert_in_bounds(", i->op->position() );
suffix.emplace_back( ", ", i->op->position() );
}
else {
suffix.emplace_back( i->op->to_string(true), i->op->position() );
}
}
}
// Print the prefixes (in forward order)
for (auto& e : prefix) {
printer.print_cpp2(e.text, n.position());
}
// If this is an --, ++, or &, don't add std::move on the lhs
// even if this is a definite last use (only do that when an rvalue is okay)
if (n.ops.front().op->type() == lexeme::MinusMinus ||
n.ops.front().op->type() == lexeme::PlusPlus ||
n.ops.front().op->type() == lexeme::Ampersand
)
{
suppress_move_from_last_use = true;
}
// Now print the core expression -- or the captured_part in its place
if (captured_part.empty()) {
emit(*n.expr);
}
else {
printer.print_cpp2(captured_part, n.position());
}
suppress_move_from_last_use = false;
// Print the suffixes (in reverse order)
while (!suffix.empty()) {
printer.print_cpp2(suffix.back().text, suffix.back().pos);
suffix.pop_back();
}
}
//-----------------------------------------------------------------------
//
auto emit(prefix_expression_node const& n) -> void
{
auto suffix = std::string{};
for (auto const& x : n.ops) {
assert(x);
if (x->type() == lexeme::Not) {
printer.print_cpp2("!(", n.position());
printer.add_pad_in_this_line(-3);
suffix += ")";
}
else {
printer.print_cpp2(*x, x->position());
}
}
assert(n.expr);
push_need_expression_list_parens(true);
emit(*n.expr);
pop_need_expression_list_parens();
printer.print_cpp2(suffix, n.position());
}
//-----------------------------------------------------------------------
//
template<
String Name,
typename Term
>
auto emit(binary_expression_node const& n) -> void
{
assert(n.expr);
// Handle is/as expressions
// TODO: Generalize
if (!n.terms.empty() && *n.terms.front().op == "is")
{
if (n.terms.size() > 1) {
errors.emplace_back(
n.position(),
"(temporary alpha limitation) this compiler is just starting to learn 'is' and only supports a single is-expression (no chaining with other is/as)"
);
return;
}
printer.print_cpp2("cpp2::is(", n.position());
emit(*n.expr);
printer.print_cpp2(")", n.position());
return;
}
if (!n.terms.empty() && *n.terms.front().op == "as")
{
if (n.terms.size() > 1) {
errors.emplace_back(
n.position(),
"(temporary alpha limitation) this compiler is just starting to learn 'as' and only supports a single as-expression (no chaining with other is/as)"
);
return;
}
printer.print_cpp2("cpp2::as(", n.position());
emit(*n.expr);
printer.print_cpp2(")", n.position());
return;
}
// If this is an assignment expression, don't add std::move on the lhs
// even if this is a definite last use (only do that when an rvalue is okay)
if (!n.terms.empty() && is_assignment_operator(n.terms.front().op->type())) {
suppress_move_from_last_use = true;
}
emit(*n.expr);
suppress_move_from_last_use = false;
// Check that this isn't an illegal pointer operation
// (initial partial implementation)
if (!n.terms.empty() && last_postfix_expr_was_pointer)
{
auto rhs_post = n.get_second_postfix_expression_node();
assert(rhs_post && rhs_post->expr);
auto rhs_tok = rhs_post->expr->get_token();
if (is_assignment_operator(n.terms.front().op->type()) && rhs_tok &&
(*rhs_tok == "nullptr" || is_digit(((std::string_view)*rhs_tok)[0]))
)
{
errors.emplace_back(
n.terms.front().op->position(),
n.terms.front().op->to_string(true) + " - pointer assignment from null or integer is illegal"
);
violates_lifetime_safety = true;
}
else if (
*n.terms.front().op == "+" || *n.terms.front().op == "+=" ||
*n.terms.front().op == "-" || *n.terms.front().op == "-="
)
{
errors.emplace_back(
n.terms.front().op->position(),
n.terms.front().op->to_string(true) + " - pointer arithmetic is illegal - use std::span or gsl::span instead"
);
violates_bounds_safety = true;
}
}
for (auto const& x : n.terms) {
assert(x.op);
assert(x.expr);
// Normally we'll just emit the operator, but if this is an
// assignment that's a definite initialization, change it to
// a .construct() call
if (x.op->type() == lexeme::Assignment && in_definite_init) {
printer.print_cpp2( ".construct(", n.position() );
emit(*x.expr);
printer.print_cpp2( ")", n.position() );
}
else {
printer.print_cpp2(" ", n.position());
emit(*x.op);
printer.print_cpp2(" ", n.position());
emit(*x.expr);
}
}
}
//-----------------------------------------------------------------------
//
auto emit(expression_node const& n) -> void
{
assert(n.expr);
push_need_expression_list_parens(true);
emit(*n.expr);
pop_need_expression_list_parens();
}
//-----------------------------------------------------------------------
//
auto emit(expression_list_node const& n) -> void
{
if (should_add_expression_list_parens() && !n.expressions.empty() && !n.inside_initializer) {
printer.print_cpp2("(", n.position());
}
auto first = true;
for (auto const& x : n.expressions) {
if (!first) {
printer.print_cpp2(", ", n.position());
}
first = false;
auto offset = 0;
if (x.pass != passing_style::in) {
assert(to_string_view(x.pass) == "out" || to_string_view(x.pass) == "move");
if (to_string_view(x.pass) == "out") {
printer.print_cpp2("&", n.position());
offset = -3; // because we're replacing "out " (followed by at least one space) with "&"
}
else if (to_string_view(x.pass) == "move") {
printer.print_cpp2("std::move(", n.position());
offset = 6; // because we're replacing "move " (followed by at least one space) with "std::move("
}
}
assert(x.expr);
adjust_remaining_token_columns_on_this_line_visitor v(x.expr->position(), offset);
x.expr->visit(v, 0);
emit(*x.expr);
if (to_string_view(x.pass) == "move") {
printer.print_cpp2(")", n.position());
}
}
if (should_add_expression_list_parens() && !n.expressions.empty() && !n.inside_initializer) {
printer.print_cpp2(")", n.position());
}
// We want to consume only one of these
consumed_expression_list_parens();
}
//-----------------------------------------------------------------------
//
auto emit(expression_statement_node const& n, bool can_have_semicolon, source_position function_body_start = {}, bool function_void_ret = false ) -> void
{
assert(n.expr);
if (function_body_start != source_position{}) {
printer.print_cpp2(" { ", function_body_start);
if (!function_void_ret) {
printer.print_cpp2("return ", n.position());
}
}
emit(*n.expr);
if (n.has_semicolon && can_have_semicolon) {
printer.print_cpp2(";", n.position());
}
if (function_body_start != source_position{}) {
printer.print_cpp2(" }", n.position());
}
}
//-----------------------------------------------------------------------
//
auto emit(
statement_node const& n,
bool can_have_semicolon = true,
source_position function_body_start = {},
bool function_void_ret = false,
std::vector const& function_prolog = {},
std::vector const& function_epilog = {},
colno_t function_indent = 1
)
-> void
{
// TODO: If there's a let on this statement, generate a block scope
if (n.let) {
// TODO
}
printer.disable_indent_heuristic_for_next_text();
try_emit(n.statement, function_prolog, function_epilog, function_indent);
// NOTE: Reset preemption here because
// - for compound statements written as "= { ... }", we want to keep the
// preempted position which moves the { to where the = was
// - but for other statement types, we want to get rid of any leftover
// preemption (ideally there wouldn't be any, but sometimes there is
// and it should not apply to what we're about to emit)
printer.preempt_position({});
// This only has a whitespace effect in the generated Cpp1 code, but it's
// aesthetic and aesthetics are important in this case -- we want to keep
// the original source's personal whitespace formatting style as much as we can
try_emit(n.statement, can_have_semicolon, function_body_start, function_void_ret);
try_emit(n.statement);
try_emit(n.statement);
try_emit(n.statement);
try_emit(n.statement);
try_emit(n.statement);
try_emit(n.statement, false);
}
//-----------------------------------------------------------------------
//
auto emit(parameter_declaration_node const& n, bool returns = false) -> void
{
// Can't declare functions as parameters -- only pointers to functions which are objects
assert( n.declaration );
assert( n.declaration->is(declaration_node::object) );
auto const& id_expr = *std::get(n.declaration->type);
auto unqid = std::get_if(&id_expr.id);
auto is_wildcard = unqid && *(*unqid)->identifier == "_";
// First any prefix
if (!returns && !is_wildcard)
{
switch (n.pass) {
break;case passing_style::in : printer.print_cpp2( "cpp2::in", n.position() );
break;case passing_style::copy : printer.print_cpp2( "", n.position() );
break;case passing_style::inout : printer.print_cpp2( "&", n.position() );
break;case passing_style::out : printer.print_cpp2( ">", n.position() );
break;case passing_style::move : printer.print_cpp2( "&&", n.position() );
break;case passing_style::forward: printer.print_cpp2( "&&", n.position() );
break;default: ;
}
printer.preempt_position( n.position() );
}
printer.print_cpp2( " ", n.declaration->identifier->position() );
emit( *n.declaration->identifier );
if (!returns && n.declaration->initializer) {
printer.print_cpp2( " = ", n.declaration->initializer->position() );
emit(*n.declaration->initializer);
}
//TODO - when we get to classes and inheritance
//o void
{
assert(n.parameters);
emit(*n.parameters);
// Add implicit noexcept when we implement proper EH
// to handle calling Cpp1 code that throws
//if (!n.throws) {
// printer.add_pad_in_this_line(-25);
// printer.print_cpp2( " noexcept", n.position() );
//}
if (n.returns.index() == function_type_node::empty) {
if (ident) {
printer.print_cpp2( " -> void", n.position() );
}
}
else if (n.returns.index() == function_type_node::id) {
printer.print_cpp2( " -> ", n.position() );
auto& r = std::get(n.returns);
assert(r);
emit(*r);
}
else {
printer.print_cpp2( " -> ", n.position() );
function_return_name = {};
printer.emit_to_string(&function_return_name);
assert(ident);
printer.print_cpp2( *ident, ident->position() );
printer.print_cpp2( "__ret", ident->position() );
printer.emit_to_string();
printer.print_cpp2( function_return_name, ident->position() );
}
}
//-----------------------------------------------------------------------
//
auto emit(declaration_node const& n, std::string const& in_expression_intro = "") -> void
{
// If this is a function that has multiple return values,
// first we need to emit the struct that contains the returns
if (printer.doing_declarations_only() && n.is(declaration_node::function))
{
auto& func = std::get(n.type);
assert(func);
if (func->returns.index() == function_type_node::list) {
auto& r = std::get(func->returns);
assert(r);
assert(std::ssize(r->parameters) > 0);
printer.ignore_alignment( true, n.position().colno );
printer.print_cpp2( "struct ", n.position() );
printer.ignore_alignment( true, n.position().colno + 4 );
printer.print_cpp2( *n.identifier->identifier, n.position() );
printer.print_cpp2( "__ret ", n.position() );
emit(*r, true);
printer.print_cpp2( "\n", n.position() );
printer.ignore_alignment( false );
}
}
// Function
if (n.is(declaration_node::function))
{
// Start fresh (there may be one spurious leftover
// requires-condition created during the declarations pass)
function_requires_conditions = {};
auto& func = std::get(n.type);
assert(func);
// If this is at expression scope, we can't emit "[[nodiscard]] auto name"
// so print the provided intro instead, which will be a lambda-capture-list
if (in_expression_intro != "") {
assert (!n.identifier);
printer.print_cpp2(in_expression_intro, n.position());
emit( *func, nullptr );
}
else {
assert (n.identifier);
if (func->returns.index() != function_type_node::empty) {
printer.print_cpp2( "[[nodiscard]] ", n.position() );
}
printer.print_cpp2( "auto ", n.position() );
printer.print_cpp2( *n.identifier->identifier, n.identifier->position() );
emit( *func, n.identifier->identifier );
}
// Function declaration
if (printer.doing_declarations_only()) {
printer.print_cpp2( ";\n", n.position() );
return;
}
if (func->returns.index() == function_type_node::list) {
auto& r = std::get(func->returns);
function_returns.push_back(r.get());
}
else if (func->returns.index() == function_type_node::id) {
function_returns.push_back(&single_anon); // use special value as a note
}
else {
function_returns.push_back(nullptr); // no return type at all
}
// Function body
assert( n.initializer );
auto function_return_locals = std::vector{};
auto function_epilog = std::vector{};
for (auto&& c : func->contracts) {
auto print = std::string();
printer.emit_to_string(&print);
emit(*c);
printer.emit_to_string();
function_return_locals.push_back(print);
}
if (func->returns.index() == function_type_node::list)
{
auto& r = std::get(func->returns);
assert(r);
for (auto& param : r->parameters)
{
assert(param && param->declaration);
auto& decl = *param->declaration;
assert(decl.type.index() == declaration_node::object);
auto& id_expr = std::get(decl.type);
assert(id_expr);
auto loc = std::string{};
if (!decl.initializer) {
loc += (" cpp2::deferred_init");
}
loc += " ";
loc += ((std::string_view)*decl.identifier->identifier);
if (decl.initializer)
{
std::string init;
printer.emit_to_string(&init);
printer.print_cpp2 ( " = ", decl.initializer->position() );
if (decl.initializer->statement.index() != statement_node::expression) {
errors.emplace_back(
decl.initializer->position(),
"return value initializer must be an expression"
);
return;
}
auto& expr = std::get(decl.initializer->statement);
assert(expr);
emit(*decl.initializer);
printer.emit_to_string();
loc += init;
}
loc += ";";
function_return_locals.push_back(loc);
}
}
//function_epilog.push_back("/*EPILOG-TEST*/");
printer.preempt_position( n.equal_sign );
// TODO: something like this to get rid of extra blank lines
// inside the start of bodies of functions that have
// multiple contracts
//printer.skip_lines( std::ssize(function_return_locals) );
// If processing the parameters generated any requires conditions,
// emit them here
if (!function_requires_conditions.empty()) {
printer.ignore_alignment( true, n.position().colno + 4 );
printer.print_extra("\n");
for (auto const& req : function_requires_conditions) {
printer.print_extra("requires " + req);
}
function_requires_conditions = {};
printer.ignore_alignment( false );
}
emit(
*n.initializer,
true, func->position(), n.identifier && func->returns.index() == function_type_node::empty,
function_return_locals, function_epilog, n.position().colno
);
function_returns.pop_back();
}
// Object with optional initializer
else if (!printer.doing_declarations_only() && n.is(declaration_node::object))
{
auto& type = std::get(n.type);
// Emit "auto" for deduced types (of course)
if (type->id.index() == id_expression_node::empty) {
assert(n.initializer);
printer.print_cpp2("auto", n.position());
}
// Otherwise, emit the type
else {
// If there isn't an initializer, use cpp2::deferred_init
if (!n.initializer) {
printer.print_cpp2( "cpp2::deferred_init", n.position() );
}
}
printer.print_cpp2( " ", n.position());
assert(n.identifier);
emit(*n.identifier);
// If there's an initializer, emit it
if (n.initializer)
{
printer.add_pad_in_this_line(-100);
printer.print_cpp2( " { ", n.position() );
push_need_expression_list_parens(false);
assert( n.initializer );
emit( *n.initializer, false );
pop_need_expression_list_parens();
printer.print_cpp2( " }", n.position() );
}
printer.print_cpp2( "; ", n.position() );
}
}
//-----------------------------------------------------------------------
// print_errors
//
auto print_errors() -> void
{
if (!errors.empty()) {
// Delete the output file
printer.abandon();
}
for (auto&& error : errors) {
error.print(std::cerr, strip_path(sourcefile));
}
if (violates_lifetime_safety) {
std::cerr bool
{
return errors.empty();
}
//-----------------------------------------------------------------------
// debug_print
//
auto debug_print() -> void
{
// Only create debug output files if we managed to load the source file.
//
if (source_loaded)
{
auto out_source = std::ofstream{ sourcefile+"-source" };
source.debug_print( out_source );
auto out_tokens = std::ofstream{ sourcefile+"-tokens" };
tokens.debug_print( out_tokens );
auto out_parse = std::ofstream{ sourcefile+"-parse" };
auto tree_printer = parse_tree_printer{out_parse };
parser.visit ( tree_printer );
auto out_symbols = std::ofstream{ sourcefile+"-symbols" };
sema.debug_print ( out_symbols );
}
}
//-----------------------------------------------------------------------
// has_cpp1: pass through
//
auto has_cpp1() const -> bool {
return source.has_cpp1();
}
//-----------------------------------------------------------------------
// has_cpp2: pass through
//
auto has_cpp2() const -> bool {
return source.has_cpp2();
}
};
}
//===========================================================================
// main - driver
//===========================================================================
using namespace std;
using namespace cpp2;
static auto enable_debug_output_files = false;
static cmdline_processor::register_flag cmd_noline(
9,
"debug",
"Emit compiler debug output files",
[]{ enable_debug_output_files = true; }
);
auto main(int argc, char* argv[]) -> int
{
cmdline.set_args(argc, argv);
cmdline.process_flags();
if (cmdline.help_was_requested()) {
return 0;
}
if (cmdline.arguments().empty()) {
std::cout