diff options
| author | Jemma Issroff <[email protected]> | 2023-08-29 16:13:15 -0400 |
|---|---|---|
| committer | GitHub <[email protected]> | 2023-08-29 13:13:15 -0700 |
| commit | 80dc570a451ed493797fa68b59a8dd60c1fa51a6 (patch) | |
| tree | dc9d22df2a72a3ba99299b2b65b8f58382bf8d92 | |
| parent | b435161404ed960e02fadfd1c3d983d1fbfb91a0 (diff) | |
Compile more YARP node types (#8322)
* Add several more node simple types to YARP's compiler:
Nodes include: DefinedNode, EmbeddedStatementsNode,
LocalVariableReadNode, LocalVariableWriteNode, MultiWriteNode,
OptionalParameterNode, SplatNode, YieldNode
* Add AssocSplatNode, RangeNode
* Add RangeNode, other helpers for future nodes
* Add ArrayNode, HashNode, static literal helpers
* Add branch conditionals
* Add IfNode, UnlessNode
* Add ScopeNode
* NEW_ISEQ and NEW_CHILD_ISEQ implemented for YARP
* Add nodes that depend on ScopeNode
* Addressed PR comments
| -rw-r--r-- | iseq.c | 46 | ||||
| -rw-r--r-- | yarp/yarp_compiler.c | 900 |
2 files changed, 945 insertions, 1 deletions
@@ -945,6 +945,52 @@ rb_iseq_new_with_opt(const rb_ast_body_t *ast, VALUE name, VALUE path, VALUE rea VALUE rb_iseq_compile_yarp_node(rb_iseq_t * iseq, const yp_node_t * yarp_pointer, yp_parser_t *parser); rb_iseq_t * +yp_iseq_new_with_opt(yp_node_t *node, yp_parser_t *parser, VALUE name, VALUE path, VALUE realpath, + int first_lineno, const rb_iseq_t *parent, int isolated_depth, + enum rb_iseq_type type, const rb_compile_option_t *option) +{ + rb_iseq_t *iseq = iseq_alloc(); + rb_compile_option_t new_opt; + + if (option) { + new_opt = *option; + } + else { + new_opt = COMPILE_OPTION_DEFAULT; + } + + VALUE script_lines = Qnil; + + rb_code_location_t code_loc; + + if (node) { + yp_line_column_t start_line_col = yp_newline_list_line_column(&(parser->newline_list), node->location.start); + yp_line_column_t end_line_col= yp_newline_list_line_column(&(parser->newline_list), node->location.end); + code_loc = (rb_code_location_t) { + .beg_pos = { + .lineno = (int)start_line_col.line, + .column = (int)start_line_col.column + }, + .end_pos = { + .lineno = (int)end_line_col.line, + .column = (int)end_line_col.column + }, + }; + } + + // TODO: node_id + int node_id = -1; + prepare_iseq_build(iseq, name, path, realpath, first_lineno, &code_loc, node_id, + parent, isolated_depth, type, script_lines, &new_opt); + + rb_iseq_compile_yarp_node(iseq, node, parser); + + finish_iseq_build(iseq); + + return iseq_translate(iseq); +} + +rb_iseq_t * rb_iseq_new_with_callback( const struct rb_iseq_new_with_callback_callback_func * ifunc, VALUE name, VALUE path, VALUE realpath, diff --git a/yarp/yarp_compiler.c b/yarp/yarp_compiler.c index 81c5a36d58..7157c2b9fa 100644 --- a/yarp/yarp_compiler.c +++ b/yarp/yarp_compiler.c @@ -1,5 +1,22 @@ #include "yarp.h" +#define OLD_ISEQ NEW_ISEQ +#undef NEW_ISEQ + +#define NEW_ISEQ(node, name, type, line_no) \ + yp_new_child_iseq(iseq, (node), parser, rb_fstring(name), 0, (type), (line_no)) + +#define OLD_CHILD_ISEQ NEW_CHILD_ISEQ +#undef NEW_CHILD_ISEQ + +#define NEW_CHILD_ISEQ(node, name, type, line_no) \ + yp_new_child_iseq(iseq, (node), parser, rb_fstring(name), iseq, (type), (line_no)) + +rb_iseq_t * +yp_iseq_new_with_opt(yp_scope_node_t *node, yp_parser_t *parser, VALUE name, VALUE path, VALUE realpath, + int first_lineno, const rb_iseq_t *parent, int isolated_depth, + enum rb_iseq_type type, const rb_compile_option_t *option); + static VALUE parse_number(const yp_node_t *node) { const char *start = node->location.start; @@ -42,6 +59,265 @@ parse_location_symbol(yp_location_t *location) { return parse_symbol(location->start, location->end); } +static int +yp_optimizable_range_item_p(yp_node_t *node) +{ + return (!node || node->type == YP_NODE_INTEGER_NODE || node->type == YP_NODE_NIL_NODE); +} + +static bool +yp_static_node_literal_p(yp_node_t *node) +{ + switch(node->type) { + case YP_NODE_FALSE_NODE: + case YP_NODE_FLOAT_NODE: + case YP_NODE_IMAGINARY_NODE: + case YP_NODE_INTEGER_NODE: + case YP_NODE_NIL_NODE: + case YP_NODE_RATIONAL_NODE: + case YP_NODE_SELF_NODE: + case YP_NODE_STRING_NODE: + case YP_NODE_SOURCE_ENCODING_NODE: + case YP_NODE_SOURCE_FILE_NODE: + case YP_NODE_SOURCE_LINE_NODE: + case YP_NODE_SYMBOL_NODE: + case YP_NODE_TRUE_NODE: + return true; + default: + return false; + } +} + +static inline VALUE +yp_static_literal_value(yp_node_t *node) +{ + switch(node->type) { + case YP_NODE_NIL_NODE: + return Qnil; + case YP_NODE_TRUE_NODE: + return Qtrue; + case YP_NODE_FALSE_NODE: + return Qfalse; + // TODO: Implement this method for the other literal nodes described above + default: + rb_bug("This node type doesn't have a literal value"); + } +} + +static void +yp_compile_branch_condition(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const yp_node_t *cond, + LABEL *then_label, LABEL *else_label, const char *src, bool popped, yp_compile_context_t *compile_context); + +static void +yp_compile_logical(rb_iseq_t *iseq, LINK_ANCHOR *const ret, yp_node_t *cond, + LABEL *then_label, LABEL *else_label, const char *src, bool popped, yp_compile_context_t *compile_context) +{ + yp_parser_t *parser = compile_context->parser; + yp_newline_list_t newline_list = parser->newline_list; + int lineno = (int)yp_newline_list_line_column(&newline_list, cond->location.start).line; + NODE dummy_line_node = generate_dummy_line_node(lineno, lineno); + + DECL_ANCHOR(seq); + INIT_ANCHOR(seq); + LABEL *label = NEW_LABEL(lineno); + if (!then_label) then_label = label; + else if (!else_label) else_label = label; + + yp_compile_branch_condition(iseq, seq, cond, then_label, else_label, src, popped, compile_context); + + if (LIST_INSN_SIZE_ONE(seq)) { + INSN *insn = (INSN *)ELEM_FIRST_INSN(FIRST_ELEMENT(seq)); + if (insn->insn_id == BIN(jump) && (LABEL *)(insn->operands[0]) == label) + return; + } + if (!label->refcnt) { + ADD_INSN(seq, &dummy_line_node, putnil); + } + else { + ADD_LABEL(seq, label); + } + ADD_SEQ(ret, seq); + return; +} + +static void yp_compile_node(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, const char * src, bool popped, yp_compile_context_t *context); + +static void +yp_compile_branch_condition(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const yp_node_t *cond, + LABEL *then_label, LABEL *else_label, const char *src, bool popped, yp_compile_context_t *compile_context) +{ + yp_parser_t *parser = compile_context->parser; + yp_newline_list_t newline_list = parser->newline_list; + int lineno = (int) yp_newline_list_line_column(&newline_list, cond->location.start).line; + NODE dummy_line_node = generate_dummy_line_node(lineno, lineno); + +again: + switch (YP_NODE_TYPE(cond)) { + case YP_NODE_AND_NODE: { + yp_and_node_t *and_node = (yp_and_node_t *)cond; + yp_compile_logical(iseq, ret, and_node->left, NULL, else_label, src, popped, compile_context); + cond = and_node->right; + goto again; + } + case YP_NODE_OR_NODE: { + yp_or_node_t *or_node = (yp_or_node_t *)cond; + yp_compile_logical(iseq, ret, or_node->left, then_label, NULL, src, popped, compile_context); + cond = or_node->right; + goto again; + } + case YP_NODE_FALSE_NODE: + case YP_NODE_NIL_NODE: + ADD_INSNL(ret, &dummy_line_node, jump, else_label); + return; + case YP_NODE_FLOAT_NODE: + case YP_NODE_IMAGINARY_NODE: + case YP_NODE_INTEGER_NODE: + case YP_NODE_LAMBDA_NODE: + case YP_NODE_RATIONAL_NODE: + case YP_NODE_REGULAR_EXPRESSION_NODE: + case YP_NODE_STRING_NODE: + case YP_NODE_SYMBOL_NODE: + case YP_NODE_TRUE_NODE: + ADD_INSNL(ret, &dummy_line_node, jump, then_label); + return; + // TODO: Several more nodes in this case statement + default: + { + DECL_ANCHOR(cond_seq); + INIT_ANCHOR(cond_seq); + + yp_compile_node(iseq, cond, cond_seq, src, Qfalse, compile_context); + ADD_SEQ(ret, cond_seq); + } + break; + } + ADD_INSNL(ret, &dummy_line_node, branchunless, else_label); + ADD_INSNL(ret, &dummy_line_node, jump, then_label); + return; +} + +static void +yp_compile_if(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, const char * src, bool popped, yp_compile_context_t *compile_context) { + yp_parser_t *parser = compile_context->parser; + yp_statements_node_t *node_body; + yp_node_t *node_else; + yp_node_t *predicate; + + if (node->type == YP_NODE_IF_NODE) { + yp_if_node_t *if_node = (yp_if_node_t *)node; + node_body = if_node->statements; + node_else = if_node->consequent; + predicate = if_node->predicate; + } + else { + yp_unless_node_t *unless_node = (yp_unless_node_t *)node; + node_body = unless_node->statements; + node_else = (yp_node_t *)(unless_node->consequent); + predicate = unless_node->predicate; + } + + const int line = (int)yp_newline_list_line_column(&(parser->newline_list), node->location.start).line; + NODE line_node = generate_dummy_line_node(line, line); + + DECL_ANCHOR(cond_seq); + + LABEL *then_label, *else_label, *end_label; + + INIT_ANCHOR(cond_seq); + then_label = NEW_LABEL(line); + else_label = NEW_LABEL(line); + end_label = 0; + + yp_compile_branch_condition(iseq, cond_seq, predicate, then_label, else_label, src, popped, compile_context); + ADD_SEQ(ret, cond_seq); + + if (then_label->refcnt) { + ADD_LABEL(ret, then_label); + + DECL_ANCHOR(then_seq); + INIT_ANCHOR(then_seq); + if (node_body) { + yp_compile_node(iseq, (yp_node_t *)node_body, then_seq, src, popped, compile_context); + } + else { + if (!popped) { + ADD_INSN(ret, &line_node, putnil); + } + } + + if (else_label->refcnt) { + end_label = NEW_LABEL(line); + ADD_INSNL(then_seq, &line_node, jump, end_label); + if (!popped) { + ADD_INSN(then_seq, &line_node, pop); + } + } + ADD_SEQ(ret, then_seq); + } + + if (else_label->refcnt) { + ADD_LABEL(ret, else_label); + + DECL_ANCHOR(else_seq); + INIT_ANCHOR(else_seq); + if (node_else) { + yp_compile_node(iseq, (yp_node_t *)(((yp_else_node_t *)node_else)->statements), else_seq, src, popped, compile_context); + } + else { + if (!popped) { + ADD_INSN(ret, &line_node, putnil); + } + } + + ADD_SEQ(ret, else_seq); + } + + if (end_label) { + ADD_LABEL(ret, end_label); + } + + return; +} + +static rb_iseq_t * +yp_new_child_iseq(rb_iseq_t *iseq, yp_scope_node_t * node, yp_parser_t *parser, + VALUE name, const rb_iseq_t *parent, enum rb_iseq_type type, int line_no) +{ + debugs("[new_child_iseq]> ---------------------------------------\n"); + int isolated_depth = ISEQ_COMPILE_DATA(iseq)->isolated_depth; + rb_iseq_t * ret_iseq = yp_iseq_new_with_opt(node, parser, name, + rb_iseq_path(iseq), rb_iseq_realpath(iseq), + line_no, parent, + isolated_depth ? isolated_depth + 1 : 0, + type, ISEQ_COMPILE_DATA(iseq)->option); + debugs("[new_child_iseq]< ---------------------------------------\n"); + return ret_iseq; +} + + +static int +yp_compile_class_path(LINK_ANCHOR *const ret, rb_iseq_t *iseq, const yp_node_t *constant_path_node, const NODE *line_node) +{ + if (constant_path_node->type == YP_NODE_CONSTANT_PATH_NODE) { + if (((yp_constant_path_node_t *)constant_path_node)->parent) { + /* Bar::Foo */ + // TODO: yp_compile_node(ret, "nd_else->nd_head", cpath->nd_head)); + return VM_DEFINECLASS_FLAG_SCOPED; + } + else { + /* toplevel class ::Foo */ + ADD_INSN1(ret, line_node, putobject, rb_cObject); + return VM_DEFINECLASS_FLAG_SCOPED; + } + } + else { + /* class at cbase Foo */ + ADD_INSN1(ret, line_node, putspecialobject, + INT2FIX(VM_SPECIAL_OBJECT_CONST_BASE)); + return 0; + } +} + /* * Compiles a YARP node into instruction sequences * @@ -97,12 +373,46 @@ yp_compile_node(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, } return; } + case YP_NODE_ARRAY_NODE: { + yp_array_node_t *array_node = (yp_array_node_t *) node; + yp_node_list_t elements = array_node->elements; + if (elements.size == 1 && yp_static_node_literal_p(elements.nodes[0])) { + VALUE ary = rb_ary_hidden_new(1); + rb_ary_push(ary, yp_static_literal_value(elements.nodes[0])); + OBJ_FREEZE(ary); + + ADD_INSN1(ret, &dummy_line_node, duparray, ary); + } + else { + for (size_t index = 0; index < elements.size; index++) { + yp_compile_node(iseq, elements.nodes[index], ret, src, popped, compile_context); + } + + if (!popped) { + ADD_INSN1(ret, &dummy_line_node, newarray, INT2FIX(elements.size)); + } + } + + return; + } case YP_NODE_ASSOC_NODE: { yp_assoc_node_t *assoc_node = (yp_assoc_node_t *) node; yp_compile_node(iseq, assoc_node->key, ret, src, popped, compile_context); yp_compile_node(iseq, assoc_node->value, ret, src, popped, compile_context); return; } + case YP_NODE_ASSOC_SPLAT_NODE: { + yp_assoc_splat_node_t *assoc_splat_node = (yp_assoc_splat_node_t *)node; + yp_compile_node(iseq, assoc_splat_node->value, ret, src, popped, compile_context); + + // TODO: Not sure this is accurate, look at FLUSH_CHUNK in the compiler + ADD_INSN1(ret, &dummy_line_node, newarraykwsplat, INT2FIX(0)); + + if (popped) { + ADD_INSN(ret, &dummy_line_node, pop); + } + return; + } case YP_NODE_BEGIN_NODE: { yp_begin_node_t *begin_node = (yp_begin_node_t *) node; if (begin_node->statements) { @@ -110,6 +420,87 @@ yp_compile_node(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, } return; } + case YP_NODE_CALL_NODE: { + yp_call_node_t *call_node = (yp_call_node_t *) node; + + ID mid = parse_string_symbol(&call_node->name); + int flags = 0; + int orig_argc = 0; + + if (call_node->receiver == NULL) { + ADD_INSN(ret, &dummy_line_node, putself); + } else { + yp_compile_node(iseq, call_node->receiver, ret, src, popped, compile_context); + } + + if (call_node->arguments == NULL) { + if (flags & VM_CALL_FCALL) { + flags |= VM_CALL_VCALL; + } + } else { + yp_arguments_node_t *arguments = call_node->arguments; + yp_compile_node(iseq, (yp_node_t *) arguments, ret, src, popped, compile_context); + orig_argc = (int)arguments->arguments.size; + } + + VALUE block_iseq = Qnil; + if (call_node->block != NULL) { + // Scope associated with the block + yp_scope_node_t scope_node; + yp_scope_node_init((yp_node_t *)call_node->block, &scope_node); + + const rb_iseq_t *block_iseq = NEW_CHILD_ISEQ(&scope_node, make_name_for_block(iseq), ISEQ_TYPE_BLOCK, lineno); + ISEQ_COMPILE_DATA(iseq)->current_block = block_iseq; + ADD_SEND_WITH_BLOCK(ret, &dummy_line_node, mid, INT2FIX(orig_argc), block_iseq); + } + else { + if (block_iseq == Qnil && flags == 0) { + flags |= VM_CALL_ARGS_SIMPLE; + } + + if (call_node->receiver == NULL) { + flags |= VM_CALL_FCALL; + + if (block_iseq == Qnil && call_node->arguments == NULL) { + flags |= VM_CALL_VCALL; + } + } + + ADD_SEND_WITH_FLAG(ret, &dummy_line_node, mid, INT2NUM(orig_argc), INT2FIX(flags)); + } + return; + } + case YP_NODE_CLASS_NODE: { + yp_class_node_t *class_node = (yp_class_node_t *)node; + yp_scope_node_t scope_node; + yp_scope_node_init((yp_node_t *)class_node, &scope_node); + + ID class_id = parse_string_symbol(&class_node->name); + + VALUE class_name = rb_str_freeze(rb_sprintf("<class:%"PRIsVALUE">", rb_id2str(class_id))); + + const rb_iseq_t *class_iseq = NEW_CHILD_ISEQ(&scope_node, class_name, ISEQ_TYPE_CLASS, lineno); + + // TODO: Once we merge constant path nodes correctly, fix this flag + const int flags = VM_DEFINECLASS_TYPE_CLASS | + (class_node->superclass ? VM_DEFINECLASS_FLAG_HAS_SUPERCLASS : 0) | + yp_compile_class_path(ret, iseq, class_node->constant_path, &dummy_line_node); + + if (class_node->superclass) { + yp_compile_node(iseq, class_node->superclass, ret, src, popped, compile_context); + } + else { + ADD_INSN(ret, &dummy_line_node, putnil); + } + + ADD_INSN3(ret, &dummy_line_node, defineclass, ID2SYM(class_id), class_iseq, INT2FIX(flags)); + RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)class_iseq); + + if (popped) { + ADD_INSN(ret, &dummy_line_node, pop); + } + return; + } case YP_NODE_CLASS_VARIABLE_READ_NODE: if (!popped) { ID cvar_name = parse_node_symbol((yp_node_t *)node); @@ -176,6 +567,41 @@ yp_compile_node(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, ADD_INSN1(ret, &dummy_line_node, setconstant, ID2SYM(constant_name)); return; } + case YP_NODE_DEF_NODE: { + yp_def_node_t *def_node = (yp_def_node_t *) node; + ID method_name = parse_location_symbol(&def_node->name_loc); + yp_scope_node_t scope_node; + yp_scope_node_init((yp_node_t *)def_node, &scope_node); + rb_iseq_t *method_iseq = NEW_ISEQ(&scope_node, rb_id2str(method_name), ISEQ_TYPE_METHOD, lineno); + + ADD_INSN2(ret, &dummy_line_node, definemethod, ID2SYM(method_name), method_iseq); + RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)method_iseq); + + if (!popped) { + ADD_INSN1(ret, &dummy_line_node, putobject, ID2SYM(method_name)); + } + return; + } + case YP_NODE_DEFINED_NODE: { + ADD_INSN(ret, &dummy_line_node, putself); + yp_defined_node_t *defined_node = (yp_defined_node_t *)node; + // TODO: Correct defined_type + enum defined_type dtype = DEFINED_CONST; + VALUE sym; + + sym = parse_number(defined_node->value); + + ADD_INSN3(ret, &dummy_line_node, defined, INT2FIX(dtype), sym, rb_iseq_defined_string(dtype)); + return; + } + case YP_NODE_EMBEDDED_STATEMENTS_NODE: { + yp_embedded_statements_node_t *embedded_statements_node = (yp_embedded_statements_node_t *)node; + + if (embedded_statements_node->statements) + yp_compile_node(iseq, (yp_node_t *) (embedded_statements_node->statements), ret, src, popped, compile_context); + // TODO: Concatenate the strings that exist here + return; + } case YP_NODE_EMBEDDED_VARIABLE_NODE: { yp_embedded_variable_node_t *embedded_node = (yp_embedded_variable_node_t *)node; yp_compile_node(iseq, embedded_node->variable, ret, src, popped, compile_context); @@ -186,6 +612,47 @@ yp_compile_node(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, ADD_INSN1(ret, &dummy_line_node, putobject, Qfalse); } return; + case YP_NODE_FLIP_FLOP_NODE: { + // TODO: The labels here are wrong, figure out why..... + yp_flip_flop_node_t *flip_flop_node = (yp_flip_flop_node_t *)node; + + LABEL *lend = NEW_LABEL(lineno); + LABEL *then_label = NEW_LABEL(lineno); + LABEL *else_label = NEW_LABEL(lineno); + //TODO: int again = type == NODE_FLIP2; + int again = 0; + + rb_num_t cnt = ISEQ_FLIP_CNT_INCREMENT(ISEQ_BODY(iseq)->local_iseq) + + VM_SVAR_FLIPFLOP_START; + VALUE key = INT2FIX(cnt); + + ADD_INSN2(ret, &dummy_line_node, getspecial, key, INT2FIX(0)); + ADD_INSNL(ret, &dummy_line_node, branchif, lend); + + yp_compile_node(iseq, flip_flop_node->left, ret, src, popped, compile_context); + /* *flip == 0 */ + ADD_INSNL(ret, &dummy_line_node, branchunless, else_label); + ADD_INSN1(ret, &dummy_line_node, putobject, Qtrue); + ADD_INSN1(ret, &dummy_line_node, setspecial, key); + if (!again) { + ADD_INSNL(ret, &dummy_line_node, jump, then_label); + } + + /* *flip == 1 */ + ADD_LABEL(ret, lend); + yp_compile_node(iseq, flip_flop_node->right, ret, src, popped, compile_context); + ADD_INSNL(ret, &dummy_line_node, branchunless, then_label); + ADD_INSN1(ret, &dummy_line_node, putobject, Qfalse); + ADD_INSN1(ret, &dummy_line_node, setspecial, key); + ADD_INSNL(ret, &dummy_line_node, jump, then_label); + ADD_LABEL(ret, then_label); + ADD_INSN1(ret, &dummy_line_node, putobject, Qtrue); + ADD_INSNL(ret, &dummy_line_node, jump, lend); + ADD_LABEL(ret, else_label); + ADD_INSN1(ret, &dummy_line_node, putobject, Qfalse); + ADD_LABEL(ret, lend); + return; + } case YP_NODE_FLOAT_NODE: { if (!popped) { ADD_INSN1(ret, &dummy_line_node, putobject, parse_number(node)); @@ -208,6 +675,37 @@ yp_compile_node(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, ADD_INSN1(ret, &dummy_line_node, setglobal, ID2SYM(ivar_name)); return; } + case YP_NODE_HASH_NODE: { + yp_hash_node_t *hash_node = (yp_hash_node_t *) node; + yp_node_list_t elements = hash_node->elements; + + if (elements.size == 1) { + assert(elements.nodes[0]->type == YP_NODE_ASSOC_NODE); + yp_assoc_node_t *assoc_node = (yp_assoc_node_t *) elements.nodes[0]; + + if (yp_static_node_literal_p(assoc_node->key) && + yp_static_node_literal_p(assoc_node->value)) { + VALUE hash = rb_hash_new_with_size(1); + hash = rb_obj_hide(hash); + OBJ_FREEZE(hash); + ADD_INSN1(ret, &dummy_line_node, duphash, hash); + return; + } + } + + for (size_t index = 0; index < elements.size; index++) { + yp_compile_node(iseq, elements.nodes[index], ret, src, popped, compile_context); + } + + if (!popped) { + ADD_INSN1(ret, &dummy_line_node, newhash, INT2FIX(elements.size * 2)); + } + return; + } + case YP_NODE_IF_NODE: { + yp_compile_if(iseq, node, ret, src, popped, compile_context); + return; + } case YP_NODE_IMAGINARY_NODE: { if (!popped) { ADD_INSN1(ret, &dummy_line_node, putobject, parse_number(node)); @@ -312,10 +810,110 @@ yp_compile_node(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, ADD_INSN1(ret, &dummy_line_node, newhash, INT2FIX(elements.size * 2)); return; } + case YP_NODE_LAMBDA_NODE: { + yp_scope_node_t scope_node; + yp_scope_node_init((yp_node_t *)node, &scope_node); + + const rb_iseq_t *block = NEW_CHILD_ISEQ(&scope_node, make_name_for_block(iseq), ISEQ_TYPE_BLOCK, lineno); + VALUE argc = INT2FIX(0); + + ADD_INSN1(ret, &dummy_line_node, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE)); + ADD_CALL_WITH_BLOCK(ret, &dummy_line_node, idLambda, argc, block); + RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)block); + + if (popped) { + ADD_INSN(ret, &dummy_line_node, pop); + } + return; + } + case YP_NODE_LOCAL_VARIABLE_READ_NODE: { + yp_local_variable_read_node_t *local_read_node = (yp_local_variable_read_node_t *) node; + + yp_constant_id_t constant_id = local_read_node->name; + st_data_t local_index; + + for(uint32_t i = 0; i < local_read_node->depth; i++) { + compile_context = compile_context->previous; + iseq = (rb_iseq_t *)ISEQ_BODY(iseq)->parent_iseq; + } + + int num_params = ISEQ_BODY(iseq)->param.size; + + if (!st_lookup(compile_context->index_lookup_table, constant_id, &local_index)) { + rb_bug("This local does not exist"); + } + + int index = num_params - (int)local_index; + + if (!popped) { + ADD_GETLOCAL(ret, &dummy_line_node, index, local_read_node->depth); + } + return; + } + case YP_NODE_LOCAL_VARIABLE_WRITE_NODE: { + yp_local_variable_write_node_t *local_write_node = (yp_local_variable_write_node_t *) node; + yp_compile_node(iseq, local_write_node->value, ret, src, false, compile_context); + + if (!popped) { + ADD_INSN(ret, &dummy_line_node, dup); + } + + yp_constant_id_t constant_id = local_write_node->name; + size_t stack_index; + + if (!st_lookup(compile_context->index_lookup_table, constant_id, &stack_index)) { + rb_bug("This local doesn't exist"); + } + + unsigned int num_params = ISEQ_BODY(iseq)->param.size; + size_t index = num_params - stack_index; + + ADD_SETLOCAL(ret, &dummy_line_node, (int)index, local_write_node->depth); + return; + } case YP_NODE_MISSING_NODE: { rb_bug("A yp_missing_node_t should not exist in YARP's AST."); return; } + case YP_NODE_MODULE_NODE: { + yp_module_node_t *module_node = (yp_module_node_t *)node; + yp_scope_node_t scope_node; + yp_scope_node_init((yp_node_t *)module_node, &scope_node); + + ID module_id = parse_string_symbol(&module_node->name); + VALUE module_name = rb_str_freeze(rb_sprintf("<module:%"PRIsVALUE">", rb_id2str(module_id))); + + const rb_iseq_t *module_iseq = NEW_CHILD_ISEQ(&scope_node, module_name, ISEQ_TYPE_CLASS, lineno); + + const int flags = VM_DEFINECLASS_TYPE_MODULE | + yp_compile_class_path(ret, iseq, module_node->constant_path, &dummy_line_node); + + ADD_INSN (ret, &dummy_line_node, putnil); + ADD_INSN3(ret, &dummy_line_node, defineclass, ID2SYM(module_id), module_iseq, INT2FIX(flags)); + RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)module_iseq); + + if (popped) { + ADD_INSN(ret, &dummy_line_node, pop); + } + return; + } + case YP_NODE_MULTI_WRITE_NODE: { + yp_multi_write_node_t *multi_write_node = (yp_multi_write_node_t *)node; + yp_compile_node(iseq, multi_write_node->value, ret, src, popped, compile_context); + + // TODO: int flag = 0x02 | (NODE_NAMED_REST_P(restn) ? 0x01 : 0x00); + int flag = 0x00; + + ADD_INSN(ret, &dummy_line_node, dup); + ADD_INSN2(ret, &dummy_line_node, expandarray, INT2FIX(multi_write_node->targets.size), INT2FIX(flag)); + yp_node_list_t node_list = multi_write_node->targets; + + for (size_t index = 0; index < node_list.size; index++) { + yp_compile_node(iseq, node_list.nodes[index], ret, src, popped, compile_context); + } + + return; + } case YP_NODE_NIL_NODE: if (!popped) { ADD_INSN(ret, &dummy_line_node, putnil); @@ -340,6 +938,24 @@ yp_compile_node(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, return; } + case YP_NODE_OPTIONAL_PARAMETER_NODE: { + yp_optional_parameter_node_t *optional_parameter_node = (yp_optional_parameter_node_t *)node; + yp_compile_node(iseq, optional_parameter_node->value, ret, src, false, compile_context); + + yp_constant_id_t constant_id = optional_parameter_node->name; + + size_t param_number; + if (!st_lookup(compile_context->index_lookup_table, constant_id, ¶m_number)) { + rb_bug("This local doesn't exist"); + } + + unsigned int num_params = ISEQ_BODY(iseq)->param.size; + int index = (int) (num_params - param_number); + + ADD_SETLOCAL(ret, &dummy_line_node, index, 0); + + return; + } case YP_NODE_PARENTHESES_NODE: { yp_parentheses_node_t *parentheses_node = (yp_parentheses_node_t *) node; @@ -363,6 +979,42 @@ yp_compile_node(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, ADD_INSN(ret, &dummy_line_node, leave); return; } + case YP_NODE_RANGE_NODE: { + yp_range_node_t *range_node = (yp_range_node_t *) node; + bool exclusive = (range_node->operator_loc.end - range_node->operator_loc.start) == 3; + + if (yp_optimizable_range_item_p(range_node->left) && yp_optimizable_range_item_p(range_node->right)) { + if (!popped) { + yp_node_t *left = range_node->left; + yp_node_t *right = range_node->right; + VALUE val = rb_range_new( + left && left->type == YP_NODE_INTEGER_NODE ? parse_number(left) : Qnil, + right && right->type == YP_NODE_INTEGER_NODE ? parse_number(right) : Qnil, + exclusive + ); + ADD_INSN1(ret, &dummy_line_node, putobject, val); + RB_OBJ_WRITTEN(iseq, Qundef, val); + } + } + else { + if (range_node->left == NULL) { + ADD_INSN(ret, &dummy_line_node, putnil); + } else { + yp_compile_node(iseq, range_node->left, ret, src, popped, compile_context); + } + + if (range_node->right == NULL) { + ADD_INSN(ret, &dummy_line_node, putnil); + } else { + yp_compile_node(iseq, range_node->right, ret, src, popped, compile_context); + } + + if (!popped) { + ADD_INSN1(ret, &dummy_line_node, newrange, INT2FIX(exclusive)); + } + } + return; + } case YP_NODE_RETURN_NODE: { yp_arguments_node_t *arguments = ((yp_return_node_t *)node)->arguments; @@ -381,9 +1033,154 @@ yp_compile_node(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, } return; } + case YP_NODE_SCOPE_NODE: { + yp_scope_node_t *scope_node = (yp_scope_node_t *)node; + yp_constant_id_list_t locals = scope_node->locals; + + yp_parameters_node_t *parameters_node = (yp_parameters_node_t *)scope_node->parameters; + yp_node_list_t requireds_list = YP_EMPTY_NODE_LIST; + yp_node_list_t optionals_list = YP_EMPTY_NODE_LIST; + + + if (parameters_node) { + requireds_list = parameters_node->requireds; + optionals_list = parameters_node->optionals; + } + + size_t size = locals.size; + + // Index lookup table buffer size is only the number of the locals + st_table *index_lookup_table = st_init_numtable(); + + VALUE idtmp = 0; + rb_ast_id_table_t *tbl = ALLOCV(idtmp, sizeof(rb_ast_id_table_t) + size * sizeof(ID)); + tbl->size = (int)size; + + // First param gets 0, second param 1, param n... + // Calculate the local index for all locals + for (size_t i = 0; i < size; i++) { + yp_constant_id_t constant_id = locals.ids[i]; + ID local = compile_context->constants[constant_id - 1]; + tbl->ids[i] = local; + st_insert(index_lookup_table, constant_id, i); + } + + yp_compile_context_t scope_compile_context = { + .parser = parser, + .previous = compile_context, + .constants = compile_context->constants, + .index_lookup_table = index_lookup_table + }; + + ISEQ_BODY(iseq)->param.lead_num = (int)requireds_list.size; + ISEQ_BODY(iseq)->param.opt_num = (int)optionals_list.size; + // TODO: Set all the other nums (good comment by lead_num illustrating what they are) + ISEQ_BODY(iseq)->param.size = (unsigned int)size; + + if (optionals_list.size) { + LABEL **opt_table = (LABEL **)ALLOC_N(VALUE, optionals_list.size + 1); + LABEL *label; + + // TODO: Should we make an api for NEW_LABEL where you can pass + // a pointer to the label it should fill out? We already + // have a list of labels allocated above so it seems wasteful + // to do the copies. + for (size_t i = 0; i < optionals_list.size; i++) { + label = NEW_LABEL(lineno); + opt_table[i] = label; + ADD_LABEL(ret, label); + yp_node_t *optional_node = optionals_list.nodes[i]; + yp_compile_node(iseq, optional_node, ret, src, false, &scope_compile_context); + } + + // Set the last label + label = NEW_LABEL(lineno); + opt_table[optionals_list.size] = label; + ADD_LABEL(ret, label); + + ISEQ_BODY(iseq)->param.flags.has_opt = TRUE; + ISEQ_BODY(iseq)->param.opt_table = (const VALUE *)opt_table; + } + + iseq_set_local_table(iseq, tbl); + + switch (ISEQ_BODY(iseq)->type) { + case ISEQ_TYPE_BLOCK: + { + LABEL *start = ISEQ_COMPILE_DATA(iseq)->start_label = NEW_LABEL(0); + LABEL *end = ISEQ_COMPILE_DATA(iseq)->end_label = NEW_LABEL(0); + + start->rescued = LABEL_RESCUE_BEG; + end->rescued = LABEL_RESCUE_END; + + ADD_TRACE(ret, RUBY_EVENT_B_CALL); + NODE dummy_line_node = generate_dummy_line_node(ISEQ_BODY(iseq)->location.first_lineno, -1); + ADD_INSN (ret, &dummy_line_node, nop); + ADD_LABEL(ret, start); + + yp_compile_node(iseq, (yp_node_t *)(scope_node->body), ret, src, popped, &scope_compile_context); + + ADD_LABEL(ret, end); + ADD_TRACE(ret, RUBY_EVENT_B_RETURN); + ISEQ_COMPILE_DATA(iseq)->last_line = ISEQ_BODY(iseq)->location.code_location.end_pos.lineno; + + /* wide range catch handler must put at last */ + ADD_CATCH_ENTRY(CATCH_TYPE_REDO, start, end, NULL, start); + ADD_CATCH_ENTRY(CATCH_TYPE_NEXT, start, end, NULL, end); + break; + } + default: + if (scope_node->body) { + yp_compile_node(iseq, (yp_node_t *)(scope_node->body), ret, src, popped, &scope_compile_context); + } + else { + ADD_INSN(ret, &dummy_line_node, putnil); + } + } + + free(index_lookup_table); + + ADD_INSN(ret, &dummy_line_node, leave); + return; + } case YP_NODE_SELF_NODE: ADD_INSN(ret, &dummy_line_node, putself); return; + case YP_NODE_SINGLETON_CLASS_NODE: { + yp_singleton_class_node_t *singleton_class_node = (yp_singleton_class_node_t *)node; + yp_scope_node_t scope_node; + yp_scope_node_init((yp_node_t *)singleton_class_node, &scope_node); + + const rb_iseq_t *singleton_class = NEW_ISEQ(&scope_node, rb_fstring_lit("singleton class"), + ISEQ_TYPE_CLASS, lineno); + + yp_compile_node(iseq, singleton_class_node->expression, ret, src, popped, compile_context); + ADD_INSN(ret, &dummy_line_node, putnil); + ID singletonclass; + CONST_ID(singletonclass, "singletonclass"); + + ADD_INSN3(ret, &dummy_line_node, defineclass, + ID2SYM(singletonclass), singleton_class, + INT2FIX(VM_DEFINECLASS_TYPE_SINGLETON_CLASS)); + RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)singleton_class); + + if (popped) { + ADD_INSN(ret, &dummy_line_node, pop); + } + + return; + } + case YP_NODE_SPLAT_NODE: { + yp_splat_node_t *splat_node = (yp_splat_node_t *)node; + yp_compile_node(iseq, splat_node->expression, ret, src, popped, compile_context); + + ADD_INSN1(ret, &dummy_line_node, splatarray, Qtrue); + + if (popped) { + ADD_INSN(ret, &dummy_line_node, pop); + } + return; + } case YP_NODE_STATEMENTS_NODE: { yp_statements_node_t *statements_node = (yp_statements_node_t *) node; yp_node_list_t node_list = statements_node->body; @@ -436,6 +1233,80 @@ yp_compile_node(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, return; } + case YP_NODE_UNLESS_NODE: { + yp_compile_if(iseq, node, ret, src, popped, compile_context); + return; + } + case YP_NODE_WHILE_NODE: { + yp_while_node_t *while_node = (yp_while_node_t *)node; + + LABEL *prev_start_label = ISEQ_COMPILE_DATA(iseq)->start_label; + LABEL *prev_end_label = ISEQ_COMPILE_DATA(iseq)->end_label; + LABEL *prev_redo_label = ISEQ_COMPILE_DATA(iseq)->redo_label; + int prev_loopval_popped = ISEQ_COMPILE_DATA(iseq)->loopval_popped; + + // TODO: Deal with ensures in here + LABEL *next_label = ISEQ_COMPILE_DATA(iseq)->start_label = NEW_LABEL(lineno); /* next */ + LABEL *redo_label = ISEQ_COMPILE_DATA(iseq)->redo_label = NEW_LABEL(lineno); /* redo */ + LABEL *break_label = ISEQ_COMPILE_DATA(iseq)->end_label = NEW_LABEL(lineno); /* break */ + LABEL *end_label = NEW_LABEL(lineno); + LABEL *adjust_label = NEW_LABEL(lineno); + + LABEL *next_catch_label = NEW_LABEL(lineno); + LABEL *tmp_label = NULL; + + ISEQ_COMPILE_DATA(iseq)->loopval_popped = 0; + + // begin; end while true + if (while_node->base.flags & YP_LOOP_FLAGS_BEGIN_MODIFIER) { + ADD_INSNL(ret, &dummy_line_node, jump, next_label); + } + else { + // while true; end + tmp_label = NEW_LABEL(lineno); + ADD_INSNL(ret, &dummy_line_node, jump, tmp_label); + } + + ADD_LABEL(ret, adjust_label); + ADD_INSN(ret, &dummy_line_node, putnil); + ADD_LABEL(ret, next_catch_label); + ADD_INSN(ret, &dummy_line_node, pop); + ADD_INSNL(ret, &dummy_line_node, jump, next_label); + if (tmp_label) ADD_LABEL(ret, tmp_label); + + ADD_LABEL(ret, redo_label); + if (while_node->statements) { + yp_compile_node(iseq, (yp_node_t *)while_node->statements, ret, src, Qfalse, compile_context); + } + + ADD_LABEL(ret, next_label); + + yp_compile_branch_condition(iseq, ret, while_node->predicate, redo_label, end_label, src, popped, compile_context); + + ADD_LABEL(ret, end_label); + ADD_ADJUST_RESTORE(ret, adjust_label); + + ADD_INSN(ret, &dummy_line_node, putnil); + + ADD_LABEL(ret, break_label); + + if (popped) { + ADD_INSN(ret, &dummy_line_node, pop); + } + + ADD_CATCH_ENTRY(CATCH_TYPE_BREAK, redo_label, break_label, NULL, + break_label); + ADD_CATCH_ENTRY(CATCH_TYPE_NEXT, redo_label, break_label, NULL, + next_catch_label); + ADD_CATCH_ENTRY(CATCH_TYPE_REDO, redo_label, break_label, NULL, + ISEQ_COMPILE_DATA(iseq)->redo_label); + + ISEQ_COMPILE_DATA(iseq)->start_label = prev_start_label; + ISEQ_COMPILE_DATA(iseq)->end_label = prev_end_label; + ISEQ_COMPILE_DATA(iseq)->redo_label = prev_redo_label; + ISEQ_COMPILE_DATA(iseq)->loopval_popped = prev_loopval_popped; + return; + } case YP_NODE_X_STRING_NODE: { yp_x_string_node_t *xstring_node = (yp_x_string_node_t *) node; ADD_INSN(ret, &dummy_line_node, putself); @@ -443,6 +1314,28 @@ yp_compile_node(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret, ADD_SEND_WITH_FLAG(ret, &dummy_line_node, rb_intern("`"), INT2NUM(1), INT2FIX(VM_CALL_FCALL | VM_CALL_ARGS_SIMPLE)); return; } + case YP_NODE_YIELD_NODE: { + unsigned int flag = 0; + struct rb_callinfo_kwarg *keywords = NULL; + + VALUE argc = INT2FIX(0); + + ADD_INSN1(ret, &dummy_line_node, invokeblock, new_callinfo(iseq, 0, FIX2INT(argc), flag, keywords, FALSE)); + + if (popped) { + ADD_INSN(ret, &dummy_line_node, pop); + } + + int level = 0; + const rb_iseq_t *tmp_iseq = iseq; + for (; tmp_iseq != ISEQ_BODY(iseq)->local_iseq; level++ ) { + tmp_iseq = ISEQ_BODY(tmp_iseq)->parent_iseq; + } + + if (level > 0) access_outer_variables(iseq, level, rb_intern("yield"), true); + + return; + } default: rb_raise(rb_eNotImpError, "node type %s not implemented", yp_node_type_to_str(node->type)); return; @@ -457,6 +1350,11 @@ rb_translate_yarp(rb_iseq_t *iseq, const yp_node_t *node, LINK_ANCHOR *const ret yp_compile_node(iseq, node, ret, node->location.start, false, compile_context); iseq_set_sequence(iseq, ret); - return Qnil; } + +#undef NEW_ISEQ +#define NEW_ISEQ OLD_ISEQ + +#undef NEW_CHILD_ISEQ +#define NEW_CHILD_ISEQ OLD_CHILD_ISEQ |