diff options
author | John Hodge <tpg@mutabah.net> | 2016-05-22 22:39:26 +0800 |
---|---|---|
committer | John Hodge <tpg@mutabah.net> | 2016-05-22 22:39:26 +0800 |
commit | 1f9a4180a3b54f85f37919ba4ca709f8e8250bb6 (patch) | |
tree | c00cef798d26bf5f68f1657b01a490533365f015 /src | |
parent | eba9d8f840a5b81f0188d109e63631c9a34c28e9 (diff) | |
download | mrust-1f9a4180a3b54f85f37919ba4ca709f8e8250bb6.tar.gz |
macro_rules - Rework pattern matching to not use try-catch
Diffstat (limited to 'src')
-rw-r--r-- | src/ast/ast.cpp | 3 | ||||
-rw-r--r-- | src/ast/ast.hpp | 21 | ||||
-rw-r--r-- | src/ast/crate.cpp | 2 | ||||
-rw-r--r-- | src/ast/crate.hpp | 4 | ||||
-rw-r--r-- | src/expand/macro_rules.cpp | 8 | ||||
-rw-r--r-- | src/expand/macro_rules.hpp | 2 | ||||
-rw-r--r-- | src/expand/mod.cpp | 2 | ||||
-rw-r--r-- | src/hir/from_ast.cpp | 2 | ||||
-rw-r--r-- | src/hir/hir.hpp | 2 | ||||
-rw-r--r-- | src/macro_rules/eval.cpp (renamed from src/macros.cpp) | 1485 | ||||
-rw-r--r-- | src/macro_rules/macro_rules.hpp (renamed from src/macros.hpp) | 46 | ||||
-rw-r--r-- | src/macro_rules/macro_rules_ptr.hpp | 31 | ||||
-rw-r--r-- | src/macro_rules/mod.cpp | 203 | ||||
-rw-r--r-- | src/macro_rules/parse.cpp (renamed from src/parse/macro_rules.cpp) | 210 | ||||
-rw-r--r-- | src/macro_rules/pattern_checks.hpp | 10 | ||||
-rw-r--r-- | src/main.cpp | 5 | ||||
-rw-r--r-- | src/parse/common.hpp | 2 | ||||
-rw-r--r-- | src/parse/expr.cpp | 1 | ||||
-rw-r--r-- | src/parse/root.cpp | 1 |
19 files changed, 1028 insertions, 1012 deletions
diff --git a/src/ast/ast.cpp b/src/ast/ast.cpp index 23e9bfea..84218591 100644 --- a/src/ast/ast.cpp +++ b/src/ast/ast.cpp @@ -313,6 +313,9 @@ void Module::add_submod(bool is_public, ::std::string name, Module mod, MetaItem DEBUG("mod.m_name = " << name << ", attrs = " << attrs);
this->add_item( is_public, mv$(name), Item::make_Module({mv$(mod)}), mv$(attrs) );
}
+void Module::add_macro(bool is_exported, ::std::string name, MacroRulesPtr macro) {
+ m_macros.push_back( Named<MacroRulesPtr>( mv$(name), mv$(macro), is_exported ) );
+}
void Module::prescan()
{
diff --git a/src/ast/ast.hpp b/src/ast/ast.hpp index b7de73ac..d1e8c6a2 100644 --- a/src/ast/ast.hpp +++ b/src/ast/ast.hpp @@ -19,17 +19,18 @@ #include "../parse/tokentree.hpp"
#include "../types.hpp"
-#include "../macros.hpp"
#include <serialise.hpp>
-#include "pattern.hpp"
-#include "attrs.hpp"
-#include "expr.hpp"
-#include "macro.hpp"
-#include "item.hpp"
+#include <ast/pattern.hpp>
+#include <ast/attrs.hpp>
+#include <ast/expr.hpp>
+#include <ast/item.hpp>
+#include <ast/macro.hpp>
#include "generics.hpp"
+#include <macro_rules/macro_rules_ptr.hpp>
+
namespace AST {
class Crate;
@@ -551,7 +552,7 @@ private: ::std::vector<Module*> m_anon_modules;
::std::vector< NamedNS<const MacroRules*> > m_macro_import_res; // Vec of imported macros (not serialised)
- ::std::vector< Named<MacroRules> > m_macros;
+ ::std::vector< Named<MacroRulesPtr> > m_macros;
public:
char m_index_populated = 0; // 0 = no, 1 = partial, 2 = complete
@@ -599,9 +600,7 @@ public: void add_neg_impl(ImplDef impl) {
m_neg_impls.emplace_back( ::std::move(impl) );
}
- void add_macro(bool is_exported, ::std::string name, MacroRules macro) {
- m_macros.push_back( Named<MacroRules>( move(name), move(macro), is_exported ) );
- }
+ void add_macro(bool is_exported, ::std::string name, MacroRulesPtr macro);
void add_macro_import(::std::string name, const MacroRules& mr) {
m_macro_import_res.push_back( NamedNS<const MacroRules*>( mv$(name), &mr, false ) );
}
@@ -641,7 +640,7 @@ public: ::std::vector<MacroInvocation>& macro_invs() { return m_macro_invocations; }
- const NamedList<MacroRules>& macros() const { return m_macros; }
+ const NamedList<MacroRulesPtr>& macros() const { return m_macros; }
const ::std::vector<NamedNS<const MacroRules*> > macro_imports_res() const { return m_macro_import_res; }
diff --git a/src/ast/crate.cpp b/src/ast/crate.cpp index d40879f3..500598b0 100644 --- a/src/ast/crate.cpp +++ b/src/ast/crate.cpp @@ -325,7 +325,7 @@ const MacroRules* ExternCrate::find_macro_rules(const ::std::string& name) { auto i = m_mr_macros.find(name); if(i != m_mr_macros.end()) - return &i->second; + return &*i->second; return nullptr; } diff --git a/src/ast/crate.hpp b/src/ast/crate.hpp index dbb9ca0d..b88b9307 100644 --- a/src/ast/crate.hpp +++ b/src/ast/crate.hpp @@ -79,7 +79,7 @@ private: class ExternCrate: public Serialisable { - ::std::map< ::std::string, MacroRules > m_mr_macros; + ::std::map< ::std::string, MacroRulesPtr > m_mr_macros; //::MIR::Module m_root_module; @@ -87,6 +87,8 @@ class ExternCrate: public: ExternCrate(); ExternCrate(const char *path); + ExternCrate(const ExternCrate&) = delete; + ExternCrate(ExternCrate&&) = default; const MacroRules* find_macro_rules(const ::std::string& name); diff --git a/src/expand/macro_rules.cpp b/src/expand/macro_rules.cpp index 968bf5f9..1689746c 100644 --- a/src/expand/macro_rules.cpp +++ b/src/expand/macro_rules.cpp @@ -4,6 +4,7 @@ #include "../ast/ast.hpp" #include "../parse/common.hpp" #include "macro_rules.hpp" +#include <macro_rules/macro_rules.hpp> class CMacroRulesExpander: public ExpandProcMacro @@ -15,9 +16,10 @@ class CMacroRulesExpander: if( ident == "" ) ERROR(sp, E0000, "macro_rules! requires an identifier" ); + DEBUG("Parsing macro_rules! " << ident); TTStream lex(tt); auto mac = Parse_MacroRules(lex); - mod.add_macro( false, ident, mac ); + mod.add_macro( false, ident, mv$(mac) ); return ::std::unique_ptr<TokenStream>( new TTStreamO(TokenTree()) ); } @@ -45,7 +47,7 @@ class CMacroUseHandler: { if( mr.name == name ) { DEBUG("Imported " << mr.name); - mod.add_macro_import( mr.name, mr.data ); + mod.add_macro_import( mr.name, *mr.data ); goto _good; } } @@ -67,7 +69,7 @@ class CMacroUseHandler: for( const auto& mr : submod.macros() ) { DEBUG("Imported " << mr.name); - mod.add_macro_import( mr.name, mr.data ); + mod.add_macro_import( mr.name, *mr.data ); } for( const auto& mri : submod.macro_imports_res() ) { diff --git a/src/expand/macro_rules.hpp b/src/expand/macro_rules.hpp index b4e6e1e0..fc4a2851 100644 --- a/src/expand/macro_rules.hpp +++ b/src/expand/macro_rules.hpp @@ -4,7 +4,6 @@ #pragma once #include <synext.hpp> -#include "../macros.hpp" namespace AST { class Expr; @@ -12,5 +11,6 @@ namespace AST { } class TokenTree; class TokenStream; +class MacroRules; extern ::std::unique_ptr<TokenStream> Macro_Invoke(const char* name, const MacroRules& rules, const TokenTree& tt, AST::Module& mod); diff --git a/src/expand/mod.cpp b/src/expand/mod.cpp index 3f0911fe..0cb2c9c7 100644 --- a/src/expand/mod.cpp +++ b/src/expand/mod.cpp @@ -105,7 +105,7 @@ void Expand_Attrs(const ::AST::MetaItems& attrs, AttrStage stage, ::AST::Crate& if( input_ident != "" ) ERROR(mi_span, E0000, "macro_rules! macros can't take an ident"); - auto e = Macro_Invoke(name.c_str(), mr.data, input_tt, mod); + auto e = Macro_Invoke(name.c_str(), *mr.data, input_tt, mod); return e; } } diff --git a/src/hir/from_ast.cpp b/src/hir/from_ast.cpp index 7ab503cf..9595ca02 100644 --- a/src/hir/from_ast.cpp +++ b/src/hir/from_ast.cpp @@ -20,7 +20,7 @@ // - Extract macros from root module for( const auto& mac : crate.m_root_module.macros() ) { //if( mac.data.export ) { - macros.insert( ::std::make_pair( mac.name, mac.data ) ); + macros.insert( ::std::make_pair( mac.name, mv$(*mac.data) ) ); //} } for( const auto& mac : crate.m_root_module.macro_imports_res() ) { diff --git a/src/hir/hir.hpp b/src/hir/hir.hpp index ff0d82a9..cc63626f 100644 --- a/src/hir/hir.hpp +++ b/src/hir/hir.hpp @@ -12,7 +12,7 @@ #include <tagged_union.hpp> -#include <macros.hpp> // DAMNIT - Why can't I have it be incomplete +#include <macro_rules/macro_rules.hpp> // DAMNIT - Why can't I have it be incomplete #include <hir/type.hpp> #include <hir/path.hpp> diff --git a/src/macros.cpp b/src/macro_rules/eval.cpp index a5c90fee..44900807 100644 --- a/src/macros.cpp +++ b/src/macro_rules/eval.cpp @@ -1,846 +1,639 @@ -/*
- */
-#include "common.hpp"
-#include "macros.hpp"
-#include "parse/parseerror.hpp"
-#include "parse/tokentree.hpp"
-#include "parse/common.hpp"
-#include "ast/ast.hpp"
-#include <limits.h>
-
-typedef ::std::map< ::std::string, MacroRules> t_macro_regs;
-
-t_macro_regs g_macro_registrations;
-
-class ParameterMappings
-{
- // MultiMap (layer, name) -> TokenTree
- // - Multiple values are only allowed for layer>0
- typedef ::std::pair<unsigned, unsigned> t_mapping_block;
- struct Mapping
- {
- t_mapping_block block;
- ::std::vector<TokenTree> entries;
- friend ::std::ostream& operator<<(::std::ostream& os, const Mapping& x) {
- os << "(" << x.block.first << ", " << x.block.second << "): '" << x.entries << "')";
- return os;
- }
- };
- struct less_cstr {
- bool operator()(const char *a, const char *b) const { return ::std::strcmp(a,b) < 0; }
- };
-
- typedef ::std::map<const char *, Mapping, less_cstr> t_inner_map;
-
- t_inner_map m_inner;
- unsigned m_layer_count;
-public:
- ParameterMappings():
- m_inner(),
- m_layer_count(0)
- {
- }
- ParameterMappings(ParameterMappings&&) = default;
-
- const t_inner_map& inner_() const {
- return m_inner;
- }
-
- size_t layer_count() const {
- return m_layer_count+1;
- }
-
- void insert(unsigned int layer, const char *name, TokenTree data) {
- if(layer > m_layer_count)
- m_layer_count = layer;
- auto v = m_inner.insert( ::std::make_pair( name, Mapping { {layer, 0}, {} } ) );
- if(v.first->second.block.first != layer) {
- throw ParseError::Generic(FMT("matching '"<<name<<"' at multiple layers"));
- }
- v.first->second.entries.push_back( mv$(data) );
- }
-
- const TokenTree* get(unsigned int layer, unsigned int iteration, const char *name, unsigned int idx) const
- {
- const auto it = m_inner.find( name );
- if( it == m_inner.end() ) {
- DEBUG("m_mappings = " << m_inner);
- return nullptr;
- }
- const auto& e = it->second;
- if( e.block.first < layer ) {
- DEBUG(name<<" higher layer (" << e.block.first << ")");
- return nullptr;
- }
- else if( e.block.first > layer ) {
- throw ParseError::Generic( FMT("'" << name << "' is still repeating at this layer") );
- }
- else if( idx >= e.entries.size() ) {
- DEBUG("Not enough mappings for name " << name << " at layer " << layer << " PI " << iteration);
- return nullptr;
- }
- else {
- DEBUG(name << " #" << idx << ": " << e.entries[idx]);
- return &e.entries[idx];
- }
- }
- unsigned int count(unsigned int layer, unsigned int iteration, const char *name) const
- {
- const auto it = m_inner.find( name );
- if( it == m_inner.end() ) {
- DEBUG("("<<layer<<","<<iteration<<",'"<<name<<"') not found m_mappings = " << m_inner);
- return 0;
- }
- const auto& e = it->second;
-
- if( e.block.first < layer ) {
- DEBUG(name<<" higher layer (" << e.block.first << ")");
- return UINT_MAX;
- }
- else if( e.block.first > layer ) {
- //throw ParseError::Generic( FMT("'" << name << "' is still repeating at this layer") );
- // ignore mismatch
- return UINT_MAX;
- }
- else {
- return it->second.entries.size();
- }
- }
-};
-
-class MacroExpander:
- public TokenStream
-{
-public:
-
-private:
- const RcString m_macro_filename;
-
- const ::std::string m_crate_name;
- const ::std::vector<MacroRuleEnt>& m_root_contents;
- const ParameterMappings m_mappings;
-
-
- struct t_offset {
- unsigned read_pos;
- unsigned loop_index;
- unsigned max_index;
- };
- /// Layer states : Index and Iteration
- ::std::vector< t_offset > m_offsets;
- ::std::vector< size_t > m_layer_iters;
-
- /// Cached pointer to the current layer
- const ::std::vector<MacroRuleEnt>* m_cur_ents; // For faster lookup.
-
- Token m_next_token; // used for inserting a single token into the stream
- ::std::unique_ptr<TTStream> m_ttstream;
-
-public:
- MacroExpander(const MacroExpander& x) = delete;
- //MacroExpander(const MacroExpander& x):
- // m_macro_name( x.m_macro_name ),
- // m_crate_name(x.m_crate_name),
- // m_root_contents(x.m_root_contents),
- // m_mappings(x.m_mappings),
- // m_offsets({ {0,0,0} }),
- // m_cur_ents(&m_root_contents)
- //{
- // prep_counts();
- //}
- MacroExpander(const ::std::string& macro_name, const ::std::vector<MacroRuleEnt>& contents, ParameterMappings mappings, ::std::string crate_name):
- m_macro_filename( FMT("Macro:" << macro_name) ),
- m_crate_name( mv$(crate_name) ),
- m_root_contents(contents),
- m_mappings( mv$(mappings) ),
- m_offsets({ {0,0,0} }),
- m_cur_ents(&m_root_contents)
- {
- prep_counts();
- }
-
- virtual Position getPosition() const override;
- virtual Token realGetToken() override;
-private:
- const MacroRuleEnt& getCurLayerEnt() const;
- const ::std::vector<MacroRuleEnt>* getCurLayer() const;
- void prep_counts();
- unsigned int count_repeats(const ::std::vector<MacroRuleEnt>& ents, unsigned layer, unsigned iter);
-};
-
-class MacroToken:
- public TokenStream
-{
- Token m_tok;
-public:
- MacroToken(Token tok);
- virtual Position getPosition() const override;
- virtual Token realGetToken() override;
-};
-
-class MacroStringify:
- public TokenStream
-{
- Token m_tok;
-public:
- MacroStringify(const TokenTree& input);
- virtual Position getPosition() const override;
- virtual Token realGetToken() override;
-};
-
-::std::unique_ptr<TokenStream> Macro_Invoke_Concat(const TokenTree& input, enum eTokenType exp);
-::std::unique_ptr<TokenStream> Macro_Invoke_Cfg(const TokenTree& input);
-
-void Macro_InitDefaults()
-{
-}
-
-bool Macro_TryPattern(TTStream& lex, const MacroPatEnt& pat)
-{
- DEBUG("pat = " << pat);
- switch(pat.type)
- {
- case MacroPatEnt::PAT_TOKEN: {
- Token tok = lex.getToken();
- if( tok != pat.tok ) {
- PUTBACK(tok, lex);
- return false;
- }
- else {
- PUTBACK(tok, lex);
- return true;
- }
- }
- case MacroPatEnt::PAT_LOOP:
- if( pat.name == "*" )
- return true;
- return Macro_TryPattern(lex, pat.subpats[0]);
- case MacroPatEnt::PAT_BLOCK:
- return LOOK_AHEAD(lex) == TOK_BRACE_OPEN;
- case MacroPatEnt::PAT_IDENT:
- return LOOK_AHEAD(lex) == TOK_IDENT;
- case MacroPatEnt::PAT_TT:
- return LOOK_AHEAD(lex) != TOK_EOF;
- case MacroPatEnt::PAT_PATH:
- return LOOK_AHEAD(lex) == TOK_IDENT
- || LOOK_AHEAD(lex) == TOK_DOUBLE_COLON
- || LOOK_AHEAD(lex) == TOK_RWORD_SUPER;
- case MacroPatEnt::PAT_TYPE:
- try {
- TTStream slex = lex;
- Parse_TT_Type(slex);
- return true;
- }
- catch( const CompileError::Base& e ) {
- return false;
- }
- case MacroPatEnt::PAT_EXPR:
- return Parse_IsTokValue( LOOK_AHEAD(lex) );
- case MacroPatEnt::PAT_STMT:
- try {
- TTStream slex = lex;
- Parse_TT_Stmt(slex);
- return true;
- }
- catch( const CompileError::Base& e ) {
- return false;
- }
- case MacroPatEnt::PAT_PAT:
- try {
- TTStream slex = lex;
- Parse_TT_Pattern(slex);
- return true;
- }
- catch( const CompileError::Base& e ) {
- return false;
- }
- case MacroPatEnt::PAT_META:
- return LOOK_AHEAD(lex) == TOK_IDENT;
- }
- throw ParseError::Todo(lex, FMT("Macro_TryPattern : " << pat));
-}
-
-bool Macro_HandlePattern(TTStream& lex, const MacroPatEnt& pat, unsigned int layer, ParameterMappings& bound_tts)
-{
- TRACE_FUNCTION_F("layer = " << layer);
- Token tok;
- TokenTree val;
-
- if( !Macro_TryPattern(lex, pat) ) {
- DEBUG("FAIL");
- return false;
- }
-
- switch(pat.type)
- {
- case MacroPatEnt::PAT_TOKEN:
- DEBUG("Token " << pat.tok);
- GET_CHECK_TOK(tok, lex, pat.tok.type());
- break;
- case MacroPatEnt::PAT_LOOP:
- //case MacroPatEnt::PAT_OPTLOOP:
- {
- unsigned int match_count = 0;
- DEBUG("Loop");
- for(;;)
- {
- if( ! Macro_TryPattern(lex, pat.subpats[0]) )
- {
- DEBUG("break");
- break;
- }
- for( unsigned int i = 0; i < pat.subpats.size(); i ++ )
- {
- if( !Macro_HandlePattern(lex, pat.subpats[i], layer+1, bound_tts) ) {
- DEBUG("Ent " << i << " failed");
- return false;
- }
- }
- match_count += 1;
- DEBUG("succ");
- if( pat.tok.type() != TOK_NULL )
- {
- if( GET_TOK(tok, lex) != pat.tok.type() )
- {
- lex.putback( mv$(tok) );
- break;
- }
- }
- }
- DEBUG("Done (" << match_count << " matches)");
- break; }
-
- case MacroPatEnt::PAT_TT:
- DEBUG("TT");
- if( GET_TOK(tok, lex) == TOK_EOF )
- throw ParseError::Unexpected(lex, TOK_EOF);
- else
- PUTBACK(tok, lex);
- val = Parse_TT(lex, false);
- if(0)
- case MacroPatEnt::PAT_PAT:
- val = Parse_TT_Pattern(lex);
- if(0)
- case MacroPatEnt::PAT_TYPE:
- val = Parse_TT_Type(lex);
- if(0)
- case MacroPatEnt::PAT_EXPR:
- val = Parse_TT_Expr(lex);
- if(0)
- case MacroPatEnt::PAT_STMT:
- val = Parse_TT_Stmt(lex);
- if(0)
- case MacroPatEnt::PAT_PATH:
- val = Parse_TT_Path(lex, false); // non-expr mode
- if(0)
- case MacroPatEnt::PAT_BLOCK:
- val = Parse_TT_Block(lex);
- if(0)
- case MacroPatEnt::PAT_META:
- val = Parse_TT_Meta(lex);
- if(0)
- case MacroPatEnt::PAT_IDENT:
- {
- GET_CHECK_TOK(tok, lex, TOK_IDENT);
- val = TokenTree(tok);
- }
- bound_tts.insert( layer, pat.name.c_str(), ::std::move(val) );
- break;
-
- //default:
- // throw ParseError::Todo("full macro pattern matching");
- }
- return true;
-}
-
-::std::unique_ptr<TokenStream> Macro_InvokeRules(const char *name, const MacroRules& rules, const TokenTree& input)
-{
- TRACE_FUNCTION;
-
- // TODO: Use types in `parse/macro_rules.cpp` to evaluate
-
- // 2. Check input token tree against possible variants
- // 3. Bind names
- // 4. Return expander
- int i = 0;
- for(const auto& rule : rules.m_rules)
- {
- Token tok;
- // Create token stream for input tree
- TTStream lex(input);
- ParameterMappings bound_tts;
- // Parse according to rules
- try
- {
- for(const auto& pat : rule.m_pattern)
- {
- if( !Macro_HandlePattern(lex, pat, 0, bound_tts) )
- throw ParseError::Generic(lex, "Macro pattern failed");
- }
-
- //GET_CHECK_TOK(tok, lex, close);
- GET_CHECK_TOK(tok, lex, TOK_EOF);
- DEBUG( rule.m_contents.size() << " rule contents bound to " << bound_tts.inner_().size() << " values - " << name );
- for( const auto& v : bound_tts.inner_() )
- {
- DEBUG("- " << v.first << " = [" << v.second << "]");
- }
-
- DEBUG("TODO: Obtain crate name correctly");
- TokenStream* ret_ptr = new MacroExpander(name, rule.m_contents, mv$(bound_tts), "");
- // HACK! Disable nested macro expansion
- //ret_ptr->parse_state().no_expand_macros = true;
-
- return ::std::unique_ptr<TokenStream>( ret_ptr );
- }
- catch(const CompileError::Base& e)
- {
- DEBUG("Parse of rule " << i << " of " << name <<" failed - " << e.what());
- }
- i ++;
- }
- DEBUG("");
- throw ParseError::Todo(/*source_span, */"Error when macro fails to match");
-}
-
-
-Position MacroExpander::getPosition() const
-{
- return Position(m_macro_filename, 0, m_offsets[0].read_pos);
-}
-Token MacroExpander::realGetToken()
-{
- // Use m_next_token first
- if( m_next_token.type() != TOK_NULL )
- {
- DEBUG("m_next_token = " << m_next_token);
- return ::std::move(m_next_token);
- }
- // Then try m_ttstream
- if( m_ttstream.get() )
- {
- DEBUG("TTStream set");
- Token rv = m_ttstream->getToken();
- if( rv.type() != TOK_EOF )
- return rv;
- m_ttstream.reset();
- }
- //DEBUG("ofs " << m_offsets << " < " << m_root_contents.size());
-
- // Check offset of lowest layer
- while(m_offsets.size() > 0)
- {
- unsigned int layer = m_offsets.size() - 1;
- const auto& ents = *m_cur_ents;
-
- // Obtain current read position in layer, and increment
- size_t idx = m_offsets.back().read_pos++;
-
- // Check if limit has been reached
- if( idx < ents.size() )
- {
- // - If not, just handle the next entry
- const auto& ent = ents[idx];
- // Check type of entry
- // - XXX: Hack for $crate special name
- if( ent.name == "*crate" )
- {
- DEBUG("Crate name hack");
- if( m_crate_name != "" )
- {
- m_next_token = Token(TOK_STRING, m_crate_name);
- return Token(TOK_DOUBLE_COLON);
- }
- }
- // - Expand to a parameter
- else if( ent.name != "" )
- {
- DEBUG("lookup '" << ent.name << "'");
- const TokenTree* tt;
- unsigned int search_layer = layer;
- do {
- unsigned int parent_iter = (search_layer > 0 ? m_layer_iters.at(search_layer-1) : 0);
- const size_t iter_idx = m_offsets.at(search_layer).loop_index;
- tt = m_mappings.get(search_layer, parent_iter, ent.name.c_str(), iter_idx);
- } while( !tt && search_layer-- > 0 );
- if( ! tt )
- {
- throw ParseError::Generic(*this, FMT("Cannot find '" << ent.name << "' for " << layer));
- }
- else
- {
- m_ttstream.reset( new TTStream(*tt) );
- return m_ttstream->getToken();
- }
- }
- // - Descend into a repetition
- else if( ent.subpats.size() != 0 )
- {
- DEBUG("desc: layer = " << layer << ", m_layer_iters = " << m_layer_iters);
- unsigned int layer_iter = m_layer_iters.at(layer);
- unsigned int num_repeats = this->count_repeats(ent.subpats, layer+1, layer_iter);
- if(num_repeats == UINT_MAX)
- num_repeats = 0;
- // New layer
- DEBUG("- NL = " << layer+1 << ", count = " << num_repeats );
- if( num_repeats > 0 )
- {
- // - Push an offset
- m_offsets.push_back( {0, 0, num_repeats} );
- // - Save the current layer
- m_cur_ents = getCurLayer();
- // - Restart loop for new layer
- }
- else
- {
- // Layer empty
- DEBUG("Layer " << layer+1 << " is empty");
- }
- // VVV fall through and continue loop
- }
- // - Emit a raw token
- else
- {
- return ent.tok;
- }
- // Fall through for loop
- }
- else if( layer > 0 )
- {
- // - Otherwise, restart/end loop and fall through
- DEBUG("layer = " << layer << ", m_layer_iters = " << m_layer_iters);
- auto& cur_ofs = m_offsets.back();
- DEBUG("Layer #" << layer << " Cur: " << cur_ofs.loop_index << ", Max: " << cur_ofs.max_index);
- if( cur_ofs.loop_index + 1 < cur_ofs.max_index )
- {
- m_layer_iters.at(layer) ++;
-
- DEBUG("Restart layer");
- cur_ofs.read_pos = 0;
- cur_ofs.loop_index ++;
-
- auto& loop_layer = getCurLayerEnt();
- assert(loop_layer.subpats.size());
- if( loop_layer.tok.type() != TOK_NULL ) {
- DEBUG("- Separator token = " << loop_layer.tok);
- return loop_layer.tok;
- }
- // Fall through and restart layer
- }
- else
- {
- DEBUG("Terminate layer");
- // Terminate loop, fall through to lower layers
- m_offsets.pop_back();
- // - Special case: End of macro, avoid issues
- if( m_offsets.size() == 0 )
- break;
- m_cur_ents = getCurLayer();
- }
- }
- else
- {
- DEBUG("Terminate evaluation");
- m_offsets.pop_back();
- assert( m_offsets.size() == 0 );
- }
- } // while( m_offsets NONEMPTY )
-
- DEBUG("EOF");
- return Token(TOK_EOF);
-}
-
-/// Count the number of names at each layer
-void MacroExpander::prep_counts()
-{
- m_layer_iters.resize(m_mappings.layer_count(), 0);
-}
-const MacroRuleEnt& MacroExpander::getCurLayerEnt() const
-{
- assert( m_offsets.size() > 1 );
-
- const ::std::vector<MacroRuleEnt>* ents = &m_root_contents;
- for( unsigned int i = 0; i < m_offsets.size()-2; i ++ )
- {
- unsigned int ofs = m_offsets[i].read_pos;
- assert( ofs > 0 && ofs <= ents->size() );
- ents = &(*ents)[ofs-1].subpats;
- }
- return (*ents)[m_offsets[m_offsets.size()-2].read_pos-1];
-
-}
-const ::std::vector<MacroRuleEnt>* MacroExpander::getCurLayer() const
-{
- assert( m_offsets.size() > 0 );
- const ::std::vector<MacroRuleEnt>* ents = &m_root_contents;
- for( unsigned int i = 0; i < m_offsets.size()-1; i ++ )
- {
- unsigned int ofs = m_offsets[i].read_pos;
- //DEBUG(i << " ofs=" << ofs << " / " << ents->size());
- assert( ofs > 0 && ofs <= ents->size() );
- ents = &(*ents)[ofs-1].subpats;
- //DEBUG("ents = " << ents);
- }
- return ents;
-}
-unsigned int MacroExpander::count_repeats(const ::std::vector<MacroRuleEnt>& ents, unsigned layer, unsigned iter)
-{
- bool valid = false;
- unsigned int count = 0;
- for(const auto& ent : ents)
- {
- if( ent.name != "" )
- {
- if( ent.name[0] == '*' ) {
- // Ignore meta-vars, they don't repeat
- }
- else {
- auto c = m_mappings.count(layer, iter, ent.name.c_str());
- DEBUG(c << " mappings for " << ent.name << " at (" << layer << ", " << iter << ")");
- if( c == UINT_MAX ) {
- // Ignore
- }
- else if(!valid || count == c) {
- count = c;
- valid = true;
- }
- else {
- // Mismatch!
- throw ParseError::Generic("count_repeats - iteration count mismatch");
- }
- }
- }
- else if( ent.subpats.size() > 0 )
- {
- auto c = this->count_repeats(ent.subpats, layer, iter);
- if( c == UINT_MAX ) {
- }
- else if(!valid || count == c) {
- count = c;
- valid = true;
- }
- else {
- // Mismatch!
- throw ParseError::Generic("count_repeats - iteration count mismatch (subpat)");
- }
- }
- else
- {
- // Don't care
- }
- }
-
- if(valid) {
- return count;
- }
- else {
- return UINT_MAX;
- }
-}
-
-void Macro_Invoke_Concat_Once(::std::string& s, TokenStream& lex, enum eTokenType exp)
-{
- Token tok;
- GET_TOK(tok, lex);
- if( exp == TOK_STRING )
- {
- if( tok.type() == TOK_MACRO )
- {
- // Special case, expand both concat! and stringify! internally
- // TODO: Invoke
- //auto tt = Parse_TT(lex, false);
- //auto slex = Macro_Invoke(lex, tok.str(), tt);
- //Macro_Invoke_Concat_Once(s, (*slex), exp);
- //GET_CHECK_TOK(tok, (*slex), TOK_EOF);
- }
- else if( tok.type() == exp )
- {
- s += tok.str();
- }
- else
- {
- CHECK_TOK(tok, exp);
- }
- }
- else
- {
- CHECK_TOK(tok, exp);
- s += tok.str();
- }
-}
-
-::std::unique_ptr<TokenStream> Macro_Invoke_Concat(const TokenTree& input, enum eTokenType exp)
-{
- Token tok;
- TTStream lex(input);
- ::std::string val;
- do
- {
- Macro_Invoke_Concat_Once(val, lex, exp);
- } while( GET_TOK(tok, lex) == TOK_COMMA );
- CHECK_TOK(tok, TOK_EOF);
-
-
- return ::std::unique_ptr<TokenStream>( new MacroToken( Token(exp, val) ) );
-}
-
-::std::unique_ptr<TokenStream> Macro_Invoke_Cfg(const TokenTree& input)
-{
- Token tok;
- TTStream lex(input);
-
- GET_CHECK_TOK(tok, lex, TOK_IDENT);
- ::std::string var = tok.str();
-
- if( GET_TOK(tok, lex) == TOK_EQUAL )
- {
- GET_CHECK_TOK(tok, lex, TOK_STRING);
- ::std::string val = tok.str();
- GET_CHECK_TOK(tok, lex, TOK_EOF);
- return ::std::unique_ptr<TokenStream>( new MacroToken( Token(TOK_RWORD_FALSE) ) );
- }
- else
- {
- CHECK_TOK(tok, TOK_EOF);
- return ::std::unique_ptr<TokenStream>( new MacroToken( Token(TOK_RWORD_FALSE) ) );
- }
-}
-
-MacroToken::MacroToken(Token tok):
- m_tok(tok)
-{
-}
-Position MacroToken::getPosition() const
-{
- return Position("MacroToken", 0, 0);
-}
-Token MacroToken::realGetToken()
-{
- Token ret = m_tok;
- m_tok = Token(TOK_EOF);
- return ret;
-}
-MacroStringify::MacroStringify(const TokenTree& input)
-{
- Token tok;
- TTStream lex(input);
-
- ::std::string rv;
- while( GET_TOK(tok, lex) != TOK_EOF )
- {
- rv += tok.to_str();
- rv += " ";
- }
-
- m_tok = Token(TOK_STRING, ::std::move(rv));
-}
-Position MacroStringify::getPosition() const
-{
- return Position("Stringify", 0,0);
-}
-Token MacroStringify::realGetToken()
-{
- Token ret = m_tok;
- m_tok = Token(TOK_EOF);
- return ret;
-}
-SERIALISE_TYPE_S(MacroRule, {
- s.item(m_pattern);
- s.item(m_contents);
-});
-
-
-void operator%(Serialiser& s, MacroPatEnt::Type c) {
- switch(c) {
- #define _(v) case MacroPatEnt::v: s << #v; return
- _(PAT_TOKEN);
- _(PAT_TT);
- _(PAT_PAT);
- _(PAT_TYPE);
- _(PAT_EXPR);
- _(PAT_LOOP);
- //_(PAT_OPTLOOP);
- _(PAT_STMT);
- _(PAT_PATH);
- _(PAT_BLOCK);
- _(PAT_META);
- _(PAT_IDENT);
- #undef _
- }
-}
-void operator%(::Deserialiser& s, MacroPatEnt::Type& c) {
- ::std::string n;
- s.item(n);
- #define _(v) else if(n == #v) c = MacroPatEnt::v
- if(0) ;
- _(PAT_TOKEN);
- _(PAT_TT);
- _(PAT_PAT);
- _(PAT_TYPE);
- _(PAT_EXPR);
- _(PAT_LOOP);
- //_(PAT_OPTLOOP);
- _(PAT_STMT);
- _(PAT_PATH);
- _(PAT_BLOCK);
- _(PAT_META);
- _(PAT_IDENT);
- else
- throw ::std::runtime_error( FMT("No conversion for '" << n << "'") );
- #undef _
-}
-SERIALISE_TYPE_S(MacroPatEnt, {
- s % type;
- s.item(name);
- s.item(tok);
- s.item(subpats);
-});
-::std::ostream& operator<<(::std::ostream& os, const MacroPatEnt& x)
-{
- switch(x.type)
- {
- case MacroPatEnt::PAT_TOKEN: os << "=" << x.tok; break;
- case MacroPatEnt::PAT_LOOP: os << "loop w/ " << x.tok << " [" << x.subpats << "]"; break;
- default:
- os << "$" << x.name << ":";
- switch(x.type)
- {
- case MacroPatEnt::PAT_TOKEN: throw "";
- case MacroPatEnt::PAT_LOOP: throw "";
- case MacroPatEnt::PAT_TT: os << "tt"; break;
- case MacroPatEnt::PAT_PAT: os << "pat"; break;
- case MacroPatEnt::PAT_IDENT: os << "ident"; break;
- case MacroPatEnt::PAT_PATH: os << "path"; break;
- case MacroPatEnt::PAT_TYPE: os << "type"; break;
- case MacroPatEnt::PAT_EXPR: os << "expr"; break;
- case MacroPatEnt::PAT_STMT: os << "stmt"; break;
- case MacroPatEnt::PAT_BLOCK: os << "block"; break;
- case MacroPatEnt::PAT_META: os << "meta"; break;
- }
- break;
- }
- return os;
-}
-
-SERIALISE_TYPE_S(MacroRuleEnt, {
- s.item(name);
- s.item(tok);
- s.item(subpats);
-});
-
-::std::ostream& operator<<(::std::ostream& os, const MacroRuleEnt& x)
-{
- if(x.name.size())
- os << "$"<<x.name;
- else if( x.subpats.size() )
- os << "expand w/ " << x.tok << " [" << x.subpats << "]";
- else
- os << "=" << x.tok;
- return os;
-}
-
-SERIALISE_TYPE_S(MacroRules, {
- s.item( m_exported );
- s.item( m_rules );
-});
-
+/* + */ +#include <common.hpp> +#include "macro_rules.hpp" +#include <parse/parseerror.hpp> +#include <parse/tokentree.hpp> +#include <parse/common.hpp> +#include <limits.h> +#include "pattern_checks.hpp" + +class ParameterMappings +{ + // MultiMap (layer, name) -> TokenTree + // - Multiple values are only allowed for layer>0 + typedef ::std::pair<unsigned, unsigned> t_mapping_block; + struct Mapping + { + t_mapping_block block; + ::std::vector<TokenTree> entries; + friend ::std::ostream& operator<<(::std::ostream& os, const Mapping& x) { + os << "(" << x.block.first << ", " << x.block.second << "): '" << x.entries << "')"; + return os; + } + }; + + struct less_cstr { + bool operator()(const char *a, const char *b) const { return ::std::strcmp(a,b) < 0; } + }; + + + ::std::vector<Mapping> m_mappings; + unsigned m_layer_count; + + //::std::map<const char*, Mapping*, less_cstr> m_map; + ::std::map<::std::string, Mapping*> m_map; +public: + ParameterMappings(): + m_layer_count(0) + { + } + ParameterMappings(ParameterMappings&&) = default; + + const ::std::vector<Mapping>& mappings() const { return m_mappings; } + + void dump() const { + DEBUG("m_mappings = {" << m_mappings << "}"); + DEBUG("m_map = {" << m_map << "}"); + } + + size_t layer_count() const { + return m_layer_count+1; + } + + void insert(unsigned int layer, unsigned int name_index, TokenTree data) { + if(layer > m_layer_count) + m_layer_count = layer; + + if( name_index >= m_mappings.size() ) { + m_mappings.resize( name_index + 1 ); + } + auto& mapping = m_mappings[name_index]; + + if( mapping.entries.size() == 0 ) { + mapping.block.first = layer; + } + else if( mapping.block.first != layer) { + throw ParseError::Generic(FMT("matching argument #'"<<name_index<<"' at multiple layers")); + } + else { + } + mapping.entries.push_back( mv$(data) ); + } + + void reindex(const ::std::vector< ::std::string>& new_names) { + m_map.clear(); + + if( new_names.size() < m_mappings.size() ) { + BUG(Span(), "Macro parameter name mismatch - len["<<new_names<<"] != len["<<m_mappings<<"]"); + } + for( unsigned int i = 0; i < m_mappings.size(); i ++ ) + { + m_map.insert( ::std::make_pair(new_names[i].c_str(), &m_mappings[i]) ); + } + } + + const TokenTree* get(unsigned int layer, unsigned int iteration, const char *name, unsigned int idx) const + { + const auto it = m_map.find( name ); + if( it == m_map.end() ) { + DEBUG("m_map = " << m_map); + return nullptr; + } + const auto& e = *it->second; + if( e.block.first < layer ) { + DEBUG(name<<" higher layer (" << e.block.first << ")"); + return nullptr; + } + else if( e.block.first > layer ) { + throw ParseError::Generic( FMT("'" << name << "' is still repeating at this layer") ); + } + else if( idx >= e.entries.size() ) { + DEBUG("Not enough mappings for name " << name << " at layer " << layer << " PI " << iteration); + return nullptr; + } + else { + DEBUG(name << " #" << idx << ": " << e.entries[idx]); + return &e.entries[idx]; + } + } + unsigned int count(unsigned int layer, unsigned int iteration, const char *name) const + { + const auto it = m_map.find( name ); + if( it == m_map.end() ) { + DEBUG("("<<layer<<","<<iteration<<",'"<<name<<"') not found - m_map={" << m_map << "}"); + return 0; + } + const auto& e = *it->second; + + if( e.block.first < layer ) { + DEBUG(name<<" higher layer (" << e.block.first << ")"); + return UINT_MAX; + } + else if( e.block.first > layer ) { + //throw ParseError::Generic( FMT("'" << name << "' is still repeating at this layer") ); + // ignore mismatch + return UINT_MAX; + } + else { + return e.entries.size(); + } + } +}; + +class MacroExpander: + public TokenStream +{ +public: + +private: + const RcString m_macro_filename; + + const ::std::string m_crate_name; + const ::std::vector<MacroRuleEnt>& m_root_contents; + const ParameterMappings m_mappings; + + + struct t_offset { + unsigned read_pos; + unsigned loop_index; + unsigned max_index; + }; + /// Layer states : Index and Iteration + ::std::vector< t_offset > m_offsets; + ::std::vector< size_t > m_layer_iters; + + /// Cached pointer to the current layer + const ::std::vector<MacroRuleEnt>* m_cur_ents; // For faster lookup. + + Token m_next_token; // used for inserting a single token into the stream + ::std::unique_ptr<TTStream> m_ttstream; + +public: + MacroExpander(const MacroExpander& x) = delete; + //MacroExpander(const MacroExpander& x): + // m_macro_name( x.m_macro_name ), + // m_crate_name(x.m_crate_name), + // m_root_contents(x.m_root_contents), + // m_mappings(x.m_mappings), + // m_offsets({ {0,0,0} }), + // m_cur_ents(&m_root_contents) + //{ + // prep_counts(); + //} + MacroExpander(const ::std::string& macro_name, const ::std::vector<MacroRuleEnt>& contents, ParameterMappings mappings, ::std::string crate_name): + m_macro_filename( FMT("Macro:" << macro_name) ), + m_crate_name( mv$(crate_name) ), + m_root_contents(contents), + m_mappings( mv$(mappings) ), + m_offsets({ {0,0,0} }), + m_cur_ents(&m_root_contents) + { + prep_counts(); + } + + virtual Position getPosition() const override; + virtual Token realGetToken() override; +private: + const MacroRuleEnt& getCurLayerEnt() const; + const ::std::vector<MacroRuleEnt>* getCurLayer() const; + void prep_counts(); + unsigned int count_repeats(const ::std::vector<MacroRuleEnt>& ents, unsigned layer, unsigned iter); +}; + +void Macro_InitDefaults() +{ +} + +bool Macro_TryPattern(TTStream& lex, const MacroPatEnt& pat) +{ + DEBUG("pat = " << pat); + Token tok; + switch(pat.type) + { + case MacroPatEnt::PAT_TOKEN: { + GET_TOK(tok, lex); + bool rv = (tok == pat.tok); + PUTBACK(tok, lex); + return rv; + } + case MacroPatEnt::PAT_LOOP: + if( pat.name == "*" ) + return true; + return Macro_TryPattern(lex, pat.subpats[0]); + case MacroPatEnt::PAT_BLOCK: + return LOOK_AHEAD(lex) == TOK_BRACE_OPEN; + case MacroPatEnt::PAT_IDENT: + return LOOK_AHEAD(lex) == TOK_IDENT; + case MacroPatEnt::PAT_TT: + return LOOK_AHEAD(lex) != TOK_EOF; + case MacroPatEnt::PAT_PATH: + return is_token_path( LOOK_AHEAD(lex) ); + case MacroPatEnt::PAT_TYPE: + return is_token_type( LOOK_AHEAD(lex) ); + case MacroPatEnt::PAT_EXPR: + return is_token_expr( LOOK_AHEAD(lex) ); + case MacroPatEnt::PAT_STMT: + return is_token_stmt( LOOK_AHEAD(lex) ); + case MacroPatEnt::PAT_PAT: + return is_token_pat( LOOK_AHEAD(lex) ); + case MacroPatEnt::PAT_META: + return LOOK_AHEAD(lex) == TOK_IDENT; + } + throw ParseError::Todo(lex, FMT("Macro_TryPattern : " << pat)); +} + +bool Macro_HandlePattern(TTStream& lex, const MacroPatEnt& pat, unsigned int layer, ParameterMappings& bound_tts) +{ + TRACE_FUNCTION_F("layer = " << layer); + Token tok; + TokenTree val; + + switch(pat.type) + { + case MacroPatEnt::PAT_TOKEN: + DEBUG("Token " << pat.tok); + GET_CHECK_TOK(tok, lex, pat.tok.type()); + break; + case MacroPatEnt::PAT_LOOP: + //case MacroPatEnt::PAT_OPTLOOP: + { + unsigned int match_count = 0; + DEBUG("Loop"); + for(;;) + { + if( ! Macro_TryPattern(lex, pat.subpats[0]) ) + { + DEBUG("break"); + break; + } + for( unsigned int i = 0; i < pat.subpats.size(); i ++ ) + { + if( !Macro_HandlePattern(lex, pat.subpats[i], layer+1, bound_tts) ) { + DEBUG("Ent " << i << " failed"); + return false; + } + } + match_count += 1; + DEBUG("succ"); + if( pat.tok.type() != TOK_NULL ) + { + if( GET_TOK(tok, lex) != pat.tok.type() ) + { + lex.putback( mv$(tok) ); + break; + } + } + } + DEBUG("Done (" << match_count << " matches)"); + break; } + + case MacroPatEnt::PAT_TT: + DEBUG("TT"); + if( GET_TOK(tok, lex) == TOK_EOF ) + throw ParseError::Unexpected(lex, TOK_EOF); + else + PUTBACK(tok, lex); + val = Parse_TT(lex, false); + if(0) + case MacroPatEnt::PAT_PAT: + val = Parse_TT_Pattern(lex); + if(0) + case MacroPatEnt::PAT_TYPE: + val = Parse_TT_Type(lex); + if(0) + case MacroPatEnt::PAT_EXPR: + val = Parse_TT_Expr(lex); + if(0) + case MacroPatEnt::PAT_STMT: + val = Parse_TT_Stmt(lex); + if(0) + case MacroPatEnt::PAT_PATH: + val = Parse_TT_Path(lex, false); // non-expr mode + if(0) + case MacroPatEnt::PAT_BLOCK: + val = Parse_TT_Block(lex); + if(0) + case MacroPatEnt::PAT_META: + val = Parse_TT_Meta(lex); + if(0) + case MacroPatEnt::PAT_IDENT: + { + GET_CHECK_TOK(tok, lex, TOK_IDENT); + val = TokenTree(tok); + } + bound_tts.insert( layer, pat.name_index, ::std::move(val) ); + break; + + //default: + // throw ParseError::Todo("full macro pattern matching"); + } + return true; +} + +::std::unique_ptr<TokenStream> Macro_InvokeRules(const char *name, const MacroRules& rules, const TokenTree& input) +{ + TRACE_FUNCTION; + + const auto* cur_frag = &rules.m_pattern; + unsigned int cur_frag_ofs = 0; + + ParameterMappings bound_tts; + unsigned int rule_index; + + TTStream lex(input); + while(true) + { + // If not at the end of the fragment, handle that pattern + if( cur_frag_ofs < cur_frag->m_pats_ents.size() ) + { + const auto& pat = cur_frag->m_pats_ents[cur_frag_ofs]; + DEBUG("- try " << pat); + if( !Macro_HandlePattern(lex, pat, 0, bound_tts) ) + throw ParseError::Generic(lex, "Macro pattern failed"); + // Keep going + cur_frag_ofs ++; + } + else + { + // The stream has ended + if( LOOK_AHEAD(lex) == TOK_EOF ) { + // Check if an end is expected here + if( cur_frag->m_pattern_end == ~0u ) { + Token tok = lex.getToken(); + ERROR(tok.get_pos(), E0000, "Unexpected end of macro invocation - " << cur_frag_ofs << " != len [" << cur_frag->m_pats_ents << "]"); + } + // We've found the rule! + rule_index = cur_frag->m_pattern_end; + break; + } + + // Search for which path to take + for(const auto& next : cur_frag->m_next_frags) { + assert(next.m_pats_ents.size() > 0); + if( Macro_TryPattern(lex, next.m_pats_ents.front()) ) { + cur_frag = &next; + cur_frag_ofs = 0; + goto continue_; + } + } + + // No paths matched - error out + { + ::std::stringstream expected; + for(const auto& next : cur_frag->m_next_frags) { + expected << next.m_pats_ents.front() << ", "; + } + + Token tok = lex.getToken(); + ERROR(tok.get_pos(), E0000, "Unexpected token in macro invocation - " << tok << " - expected " << expected.str()); + } + + continue_: + (void)0; + } + } + + const auto& rule = rules.m_rules[rule_index]; + + DEBUG( rule.m_contents.size() << " rule contents bound to " << bound_tts.mappings().size() << " values - " << name ); + bound_tts.reindex( rule.m_param_names ); + for( unsigned int i = 0; i < bound_tts.mappings().size(); i ++ ) + { + DEBUG(" - " << rule.m_param_names[i] << " = [" << bound_tts.mappings()[i] << "]"); + } + bound_tts.dump(); + + DEBUG("TODO: Obtain crate name correctly"); + TokenStream* ret_ptr = new MacroExpander(name, rule.m_contents, mv$(bound_tts), ""); + // HACK! Disable nested macro expansion + //ret_ptr->parse_state().no_expand_macros = true; + + return ::std::unique_ptr<TokenStream>( ret_ptr ); +} + + +Position MacroExpander::getPosition() const +{ + return Position(m_macro_filename, 0, m_offsets[0].read_pos); +} +Token MacroExpander::realGetToken() +{ + // Use m_next_token first + if( m_next_token.type() != TOK_NULL ) + { + DEBUG("m_next_token = " << m_next_token); + return ::std::move(m_next_token); + } + // Then try m_ttstream + if( m_ttstream.get() ) + { + DEBUG("TTStream set"); + Token rv = m_ttstream->getToken(); + if( rv.type() != TOK_EOF ) + return rv; + m_ttstream.reset(); + } + //DEBUG("ofs " << m_offsets << " < " << m_root_contents.size()); + + // Check offset of lowest layer + while(m_offsets.size() > 0) + { + unsigned int layer = m_offsets.size() - 1; + const auto& ents = *m_cur_ents; + + // Obtain current read position in layer, and increment + size_t idx = m_offsets.back().read_pos++; + + // Check if limit has been reached + if( idx < ents.size() ) + { + // - If not, just handle the next entry + const auto& ent = ents[idx]; + // Check type of entry + // - XXX: Hack for $crate special name + if( ent.name == "*crate" ) + { + DEBUG("Crate name hack"); + if( m_crate_name != "" ) + { + m_next_token = Token(TOK_STRING, m_crate_name); + return Token(TOK_DOUBLE_COLON); + } + } + // - Expand to a parameter + else if( ent.name != "" ) + { + DEBUG("lookup '" << ent.name << "'"); + const TokenTree* tt; + unsigned int search_layer = layer; + do { + unsigned int parent_iter = (search_layer > 0 ? m_layer_iters.at(search_layer-1) : 0); + const size_t iter_idx = m_offsets.at(search_layer).loop_index; + tt = m_mappings.get(search_layer, parent_iter, ent.name.c_str(), iter_idx); + } while( !tt && search_layer-- > 0 ); + if( ! tt ) + { + throw ParseError::Generic(*this, FMT("Cannot find '" << ent.name << "' for " << layer)); + } + else + { + m_ttstream.reset( new TTStream(*tt) ); + return m_ttstream->getToken(); + } + } + // - Descend into a repetition + else if( ent.subpats.size() != 0 ) + { + DEBUG("desc: layer = " << layer << ", m_layer_iters = " << m_layer_iters); + unsigned int layer_iter = m_layer_iters.at(layer); + unsigned int num_repeats = this->count_repeats(ent.subpats, layer+1, layer_iter); + if(num_repeats == UINT_MAX) + num_repeats = 0; + // New layer + DEBUG("- NL = " << layer+1 << ", count = " << num_repeats ); + if( num_repeats > 0 ) + { + // - Push an offset + m_offsets.push_back( {0, 0, num_repeats} ); + // - Save the current layer + m_cur_ents = getCurLayer(); + // - Restart loop for new layer + } + else + { + // Layer empty + DEBUG("Layer " << layer+1 << " is empty"); + } + // VVV fall through and continue loop + } + // - Emit a raw token + else + { + return ent.tok; + } + // Fall through for loop + } + else if( layer > 0 ) + { + // - Otherwise, restart/end loop and fall through + DEBUG("layer = " << layer << ", m_layer_iters = " << m_layer_iters); + auto& cur_ofs = m_offsets.back(); + DEBUG("Layer #" << layer << " Cur: " << cur_ofs.loop_index << ", Max: " << cur_ofs.max_index); + if( cur_ofs.loop_index + 1 < cur_ofs.max_index ) + { + m_layer_iters.at(layer) ++; + + DEBUG("Restart layer"); + cur_ofs.read_pos = 0; + cur_ofs.loop_index ++; + + auto& loop_layer = getCurLayerEnt(); + assert(loop_layer.subpats.size()); + if( loop_layer.tok.type() != TOK_NULL ) { + DEBUG("- Separator token = " << loop_layer.tok); + return loop_layer.tok; + } + // Fall through and restart layer + } + else + { + DEBUG("Terminate layer"); + // Terminate loop, fall through to lower layers + m_offsets.pop_back(); + // - Special case: End of macro, avoid issues + if( m_offsets.size() == 0 ) + break; + m_cur_ents = getCurLayer(); + } + } + else + { + DEBUG("Terminate evaluation"); + m_offsets.pop_back(); + assert( m_offsets.size() == 0 ); + } + } // while( m_offsets NONEMPTY ) + + DEBUG("EOF"); + return Token(TOK_EOF); +} + +/// Count the number of names at each layer +void MacroExpander::prep_counts() +{ + m_layer_iters.resize(m_mappings.layer_count(), 0); +} +const MacroRuleEnt& MacroExpander::getCurLayerEnt() const +{ + assert( m_offsets.size() > 1 ); + + const ::std::vector<MacroRuleEnt>* ents = &m_root_contents; + for( unsigned int i = 0; i < m_offsets.size()-2; i ++ ) + { + unsigned int ofs = m_offsets[i].read_pos; + assert( ofs > 0 && ofs <= ents->size() ); + ents = &(*ents)[ofs-1].subpats; + } + return (*ents)[m_offsets[m_offsets.size()-2].read_pos-1]; + +} +const ::std::vector<MacroRuleEnt>* MacroExpander::getCurLayer() const +{ + assert( m_offsets.size() > 0 ); + const ::std::vector<MacroRuleEnt>* ents = &m_root_contents; + for( unsigned int i = 0; i < m_offsets.size()-1; i ++ ) + { + unsigned int ofs = m_offsets[i].read_pos; + //DEBUG(i << " ofs=" << ofs << " / " << ents->size()); + assert( ofs > 0 && ofs <= ents->size() ); + ents = &(*ents)[ofs-1].subpats; + //DEBUG("ents = " << ents); + } + return ents; +} +unsigned int MacroExpander::count_repeats(const ::std::vector<MacroRuleEnt>& ents, unsigned layer, unsigned iter) +{ + bool valid = false; + unsigned int count = 0; + for(const auto& ent : ents) + { + if( ent.name != "" ) + { + if( ent.name[0] == '*' ) { + // Ignore meta-vars, they don't repeat + } + else { + auto c = m_mappings.count(layer, iter, ent.name.c_str()); + DEBUG(c << " mappings for " << ent.name << " at (" << layer << ", " << iter << ")"); + if( c == UINT_MAX ) { + // Ignore + } + else if(!valid || count == c) { + count = c; + valid = true; + } + else { + // Mismatch! + throw ParseError::Generic("count_repeats - iteration count mismatch"); + } + } + } + else if( ent.subpats.size() > 0 ) + { + auto c = this->count_repeats(ent.subpats, layer, iter); + if( c == UINT_MAX ) { + } + else if(!valid || count == c) { + count = c; + valid = true; + } + else { + // Mismatch! + throw ParseError::Generic("count_repeats - iteration count mismatch (subpat)"); + } + } + else + { + // Don't care + } + } + + if(valid) { + return count; + } + else { + return UINT_MAX; + } +} + diff --git a/src/macros.hpp b/src/macro_rules/macro_rules.hpp index f10326c5..65822fa5 100644 --- a/src/macros.hpp +++ b/src/macro_rules/macro_rules.hpp @@ -7,13 +7,10 @@ #include <map>
#include <memory>
#include <cstring>
+#include "macro_rules_ptr.hpp"
class MacroExpander;
-namespace AST {
- class Module;
-}
-
class MacroRuleEnt:
public Serialisable
{
@@ -51,6 +48,7 @@ struct MacroPatEnt: public Serialisable
{
::std::string name;
+ unsigned int name_index;
Token tok;
::std::vector<MacroPatEnt> subpats;
@@ -81,8 +79,9 @@ struct MacroPatEnt: {
}
- MacroPatEnt(::std::string name, Type type):
+ MacroPatEnt(::std::string name, unsigned int name_index, Type type):
name( mv$(name) ),
+ name_index( name_index ),
tok(),
type(type)
{
@@ -101,14 +100,33 @@ struct MacroPatEnt: SERIALISABLE_PROTOTYPES();
};
-/// A rule within a macro_rules! blcok
-class MacroRule:
+struct MacroRulesPatFrag:
public Serialisable
{
-public:
- ::std::vector<MacroPatEnt> m_pattern;
+ ::std::vector<MacroPatEnt> m_pats_ents;
+ unsigned int m_pattern_end;
+ ::std::vector< MacroRulesPatFrag > m_next_frags;
+
+ MacroRulesPatFrag():
+ m_pattern_end(~0)
+ {}
+
+ SERIALISABLE_PROTOTYPES();
+};
+
+/// An arm within a macro_rules! blcok
+struct MacroRulesArm:
+ public Serialisable
+{
+ ::std::vector< ::std::string> m_param_names;
::std::vector<MacroRuleEnt> m_contents;
+ MacroRulesArm()
+ {}
+ MacroRulesArm(::std::vector<MacroRuleEnt> contents):
+ m_contents( mv$(contents) )
+ {}
+
SERIALISABLE_PROTOTYPES();
};
@@ -118,20 +136,18 @@ class MacroRules: {
public:
bool m_exported;
- ::std::vector<MacroRule> m_rules;
+ MacroRulesPatFrag m_pattern;
+ ::std::vector<MacroRulesArm> m_rules;
MacroRules()
{
}
- MacroRules( ::std::vector<MacroRule> rules ):
- m_rules( mv$(rules) )
- {
- }
+ virtual ~MacroRules();
SERIALISABLE_PROTOTYPES();
};
extern ::std::unique_ptr<TokenStream> Macro_InvokeRules(const char *name, const MacroRules& rules, const TokenTree& input);
-extern ::std::unique_ptr<TokenStream> Macro_Invoke(const TokenStream& lex, const ::std::string& name, const TokenTree& input);
+extern MacroRulesPtr Parse_MacroRules(TokenStream& lex);
#endif // MACROS_HPP_INCLUDED
diff --git a/src/macro_rules/macro_rules_ptr.hpp b/src/macro_rules/macro_rules_ptr.hpp new file mode 100644 index 00000000..512cbda2 --- /dev/null +++ b/src/macro_rules/macro_rules_ptr.hpp @@ -0,0 +1,31 @@ +/* + */ +#pragma once + +class MacroRules; + +class MacroRulesPtr: + public Serialisable +{ + MacroRules* m_ptr; +public: + MacroRulesPtr() {} + MacroRulesPtr(MacroRules* p): m_ptr(p) {} + MacroRulesPtr(MacroRulesPtr&& x): + m_ptr(x.m_ptr) + { + x.m_ptr = nullptr; + } + MacroRulesPtr& operator=(MacroRulesPtr&& x) + { + m_ptr = x.m_ptr; x.m_ptr = nullptr; + return *this; + } + + ~MacroRulesPtr(); + + MacroRules& operator*() { assert(m_ptr); return *m_ptr; } + const MacroRules& operator*() const { assert(m_ptr); return *m_ptr; } + + SERIALISABLE_PROTOTYPES(); +}; diff --git a/src/macro_rules/mod.cpp b/src/macro_rules/mod.cpp new file mode 100644 index 00000000..e3699a3d --- /dev/null +++ b/src/macro_rules/mod.cpp @@ -0,0 +1,203 @@ +/*
+ */
+#include <common.hpp>
+#include "macro_rules.hpp"
+#include <parse/parseerror.hpp>
+#include <parse/tokentree.hpp>
+#include <parse/common.hpp>
+#include <limits.h>
+
+#include "pattern_checks.hpp"
+
+bool is_token_path(eTokenType tt) {
+ switch(tt)
+ {
+ case TOK_IDENT:
+ case TOK_DOUBLE_COLON:
+ case TOK_LT:
+ case TOK_DOUBLE_LT:
+ case TOK_RWORD_SELF:
+ case TOK_RWORD_SUPER:
+ return true;
+ default:
+ return false;
+ }
+}
+bool is_token_pat(eTokenType tt) {
+ if( is_token_path(tt) )
+ return true;
+ switch( tt )
+ {
+ case TOK_PAREN_OPEN:
+ case TOK_SQUARE_OPEN:
+
+ case TOK_AMP:
+ case TOK_RWORD_BOX:
+ case TOK_RWORD_REF:
+ case TOK_RWORD_MUT:
+ case TOK_STRING:
+ case TOK_INTEGER:
+ case TOK_CHAR:
+ return true;
+ default:
+ return false;
+ }
+}
+bool is_token_type(eTokenType tt) {
+ if( is_token_path(tt) )
+ return true;
+ switch( tt )
+ {
+ case TOK_PAREN_OPEN:
+ case TOK_SQUARE_OPEN:
+ case TOK_STAR:
+ case TOK_AMP:
+ return true;
+ default:
+ return false;
+ }
+}
+bool is_token_expr(eTokenType tt) {
+ if( is_token_path(tt) )
+ return true;
+ switch( tt )
+ {
+ case TOK_AMP:
+ case TOK_STAR:
+ case TOK_PAREN_OPEN:
+ case TOK_MACRO:
+ case TOK_DASH:
+
+ case TOK_INTEGER:
+ case TOK_STRING:
+ return true;
+ default:
+ return false;
+ }
+}
+bool is_token_stmt(eTokenType tt) {
+ if( is_token_expr(tt) )
+ return true;
+ switch( tt )
+ {
+ case TOK_BRACE_OPEN:
+ case TOK_RWORD_LET:
+ return true;
+ default:
+ return false;
+ }
+}
+
+MacroRulesPtr::~MacroRulesPtr()
+{
+ if(m_ptr)
+ {
+ delete m_ptr;
+ m_ptr = nullptr;
+ }
+}
+SERIALISE_TYPE(MacroRulesPtr::, "MacroRulesPtr", {
+},{
+})
+
+SERIALISE_TYPE_S(MacroRulesArm, {
+})
+SERIALISE_TYPE_S(MacroRulesPatFrag, {
+})
+
+void operator%(Serialiser& s, MacroPatEnt::Type c) {
+ switch(c) {
+ #define _(v) case MacroPatEnt::v: s << #v; return
+ _(PAT_TOKEN);
+ _(PAT_TT);
+ _(PAT_PAT);
+ _(PAT_TYPE);
+ _(PAT_EXPR);
+ _(PAT_LOOP);
+ _(PAT_STMT);
+ _(PAT_PATH);
+ _(PAT_BLOCK);
+ _(PAT_META);
+ _(PAT_IDENT);
+ #undef _
+ }
+}
+void operator%(::Deserialiser& s, MacroPatEnt::Type& c) {
+ ::std::string n;
+ s.item(n);
+ #define _(v) else if(n == #v) c = MacroPatEnt::v
+ if(0) ;
+ _(PAT_TOKEN);
+ _(PAT_TT);
+ _(PAT_PAT);
+ _(PAT_TYPE);
+ _(PAT_EXPR);
+ _(PAT_LOOP);
+ //_(PAT_OPTLOOP);
+ _(PAT_STMT);
+ _(PAT_PATH);
+ _(PAT_BLOCK);
+ _(PAT_META);
+ _(PAT_IDENT);
+ else
+ throw ::std::runtime_error( FMT("No conversion for '" << n << "'") );
+ #undef _
+}
+SERIALISE_TYPE_S(MacroPatEnt, {
+ s % type;
+ s.item(name);
+ s.item(tok);
+ s.item(subpats);
+});
+::std::ostream& operator<<(::std::ostream& os, const MacroPatEnt& x)
+{
+ switch(x.type)
+ {
+ case MacroPatEnt::PAT_TOKEN: os << "=" << x.tok; break;
+ case MacroPatEnt::PAT_LOOP: os << "loop w/ " << x.tok << " [" << x.subpats << "]"; break;
+ default:
+ os << "$" << x.name << ":";
+ switch(x.type)
+ {
+ case MacroPatEnt::PAT_TOKEN: throw "";
+ case MacroPatEnt::PAT_LOOP: throw "";
+ case MacroPatEnt::PAT_TT: os << "tt"; break;
+ case MacroPatEnt::PAT_PAT: os << "pat"; break;
+ case MacroPatEnt::PAT_IDENT: os << "ident"; break;
+ case MacroPatEnt::PAT_PATH: os << "path"; break;
+ case MacroPatEnt::PAT_TYPE: os << "type"; break;
+ case MacroPatEnt::PAT_EXPR: os << "expr"; break;
+ case MacroPatEnt::PAT_STMT: os << "stmt"; break;
+ case MacroPatEnt::PAT_BLOCK: os << "block"; break;
+ case MacroPatEnt::PAT_META: os << "meta"; break;
+ }
+ break;
+ }
+ return os;
+}
+
+SERIALISE_TYPE_S(MacroRuleEnt, {
+ s.item(name);
+ s.item(tok);
+ s.item(subpats);
+});
+
+::std::ostream& operator<<(::std::ostream& os, const MacroRuleEnt& x)
+{
+ if(x.name.size())
+ os << "$"<<x.name;
+ else if( x.subpats.size() )
+ os << "expand w/ " << x.tok << " [" << x.subpats << "]";
+ else
+ os << "=" << x.tok;
+ return os;
+}
+
+MacroRules::~MacroRules()
+{
+}
+SERIALISE_TYPE_S(MacroRules, {
+ s.item( m_exported );
+ s.item( m_rules );
+});
+
diff --git a/src/parse/macro_rules.cpp b/src/macro_rules/parse.cpp index 94f55e2d..2d8ad374 100644 --- a/src/parse/macro_rules.cpp +++ b/src/macro_rules/parse.cpp @@ -1,11 +1,22 @@ +/* + */ #include <common.hpp> #include "../parse/common.hpp" #include "../parse/parseerror.hpp" -#include "../macros.hpp" +#include "macro_rules.hpp" +#include "pattern_checks.hpp" -MacroRules Parse_MacroRules(TokenStream& lex); +MacroRulesPtr Parse_MacroRules(TokenStream& lex); -::std::vector<MacroPatEnt> Parse_MacroRules_Pat(TokenStream& lex, bool allow_sub, enum eTokenType open, enum eTokenType close) +/// A rule within a macro_rules! blcok +class MacroRule +{ +public: + ::std::vector<MacroPatEnt> m_pattern; + ::std::vector<MacroRuleEnt> m_contents; +}; + +::std::vector<MacroPatEnt> Parse_MacroRules_Pat(TokenStream& lex, bool allow_sub, enum eTokenType open, enum eTokenType close, ::std::vector< ::std::string>& names) { TRACE_FUNCTION; Token tok; @@ -36,31 +47,36 @@ MacroRules Parse_MacroRules(TokenStream& lex); GET_CHECK_TOK(tok, lex, TOK_COLON); GET_CHECK_TOK(tok, lex, TOK_IDENT); ::std::string type = tok.str(); + + unsigned int idx = ::std::find( names.begin(), names.end(), name ) - names.begin(); + if( idx == names.size() ) + names.push_back( name ); + if(0) ; else if( type == "tt" ) - ret.push_back( MacroPatEnt(name, MacroPatEnt::PAT_TT) ); + ret.push_back( MacroPatEnt(name, idx, MacroPatEnt::PAT_TT) ); else if( type == "pat" ) - ret.push_back( MacroPatEnt(name, MacroPatEnt::PAT_PAT) ); + ret.push_back( MacroPatEnt(name, idx, MacroPatEnt::PAT_PAT) ); else if( type == "ident" ) - ret.push_back( MacroPatEnt(name, MacroPatEnt::PAT_IDENT) ); + ret.push_back( MacroPatEnt(name, idx, MacroPatEnt::PAT_IDENT) ); else if( type == "path" ) - ret.push_back( MacroPatEnt(name, MacroPatEnt::PAT_PATH) ); + ret.push_back( MacroPatEnt(name, idx, MacroPatEnt::PAT_PATH) ); else if( type == "expr" ) - ret.push_back( MacroPatEnt(name, MacroPatEnt::PAT_EXPR) ); + ret.push_back( MacroPatEnt(name, idx, MacroPatEnt::PAT_EXPR) ); else if( type == "ty" ) - ret.push_back( MacroPatEnt(name, MacroPatEnt::PAT_TYPE) ); + ret.push_back( MacroPatEnt(name, idx, MacroPatEnt::PAT_TYPE) ); else if( type == "meta" ) - ret.push_back( MacroPatEnt(name, MacroPatEnt::PAT_META) ); + ret.push_back( MacroPatEnt(name, idx, MacroPatEnt::PAT_META) ); else if( type == "block" ) - ret.push_back( MacroPatEnt(name, MacroPatEnt::PAT_BLOCK) ); + ret.push_back( MacroPatEnt(name, idx, MacroPatEnt::PAT_BLOCK) ); else throw ParseError::Generic(lex, FMT("Unknown fragment type '" << type << "'")); break; } case TOK_PAREN_OPEN: if( allow_sub ) { - auto subpat = Parse_MacroRules_Pat(lex, true, TOK_PAREN_OPEN, TOK_PAREN_CLOSE); + auto subpat = Parse_MacroRules_Pat(lex, true, TOK_PAREN_OPEN, TOK_PAREN_CLOSE, names); enum eTokenType joiner = TOK_NULL; GET_TOK(tok, lex); if( tok.type() != TOK_PLUS && tok.type() != TOK_STAR ) @@ -204,7 +220,8 @@ MacroRule Parse_MacroRules_Var(TokenStream& lex) throw ParseError::Unexpected(lex, tok); } // - Pattern entries - rule.m_pattern = Parse_MacroRules_Pat(lex, true, tok.type(), close); + ::std::vector< ::std::string> names; + rule.m_pattern = Parse_MacroRules_Pat(lex, true, tok.type(), close, names); GET_CHECK_TOK(tok, lex, TOK_FATARROW); @@ -223,104 +240,15 @@ MacroRule Parse_MacroRules_Var(TokenStream& lex) return rule; } -struct UnifiedPatFrag -{ - ::std::vector<MacroPatEnt> m_pats_ents; - unsigned int m_pattern_end; - ::std::vector< UnifiedPatFrag > m_next_frags; - - UnifiedPatFrag(): - m_pattern_end(~0) - {} - - UnifiedPatFrag split_at(unsigned int remaining_count) { - UnifiedPatFrag rv; - for(unsigned int i = remaining_count; i < m_pats_ents.size(); i ++) - rv.m_pats_ents.push_back( mv$(m_pats_ents[i]) ); - m_pats_ents.resize(remaining_count); - rv.m_pattern_end = m_pattern_end; m_pattern_end = ~0; - rv.m_next_frags = mv$(m_next_frags); - return rv; - } -}; - -struct UnifiedMacroRules -{ - UnifiedPatFrag m_pattern; -}; - -bool is_token_path(eTokenType tt) { - switch(tt) - { - case TOK_IDENT: - case TOK_DOUBLE_COLON: - case TOK_LT: - case TOK_DOUBLE_LT: - case TOK_RWORD_SELF: - case TOK_RWORD_SUPER: - return true; - default: - return false; - } -} -bool is_token_pat(eTokenType tt) { - if( is_token_path(tt) ) - return true; - switch( tt ) - { - case TOK_PAREN_OPEN: - case TOK_SQUARE_OPEN: - - case TOK_AMP: - case TOK_RWORD_BOX: - case TOK_RWORD_REF: - case TOK_RWORD_MUT: - case TOK_STRING: - case TOK_INTEGER: - case TOK_CHAR: - return true; - default: - return false; - } -} -bool is_token_type(eTokenType tt) { - if( is_token_path(tt) ) - return true; - switch( tt ) - { - case TOK_PAREN_OPEN: - case TOK_SQUARE_OPEN: - case TOK_STAR: - case TOK_AMP: - return true; - default: - return false; - } -} -bool is_token_expr(eTokenType tt) { - if( is_token_path(tt) ) - return true; - switch( tt ) - { - case TOK_AMP: - case TOK_STAR: - case TOK_PAREN_OPEN: - case TOK_MACRO: - case TOK_DASH: - - case TOK_INTEGER: - case TOK_STRING: - return true; - default: - return false; - } -} - bool patterns_are_same(const Span& sp, const MacroPatEnt& left, const MacroPatEnt& right) { if( left.type > right.type ) return patterns_are_same(sp, right, left); + //if( left.name != right.name ) { + // TODO(sp, "Handle different binding names " << left << " != " << right); + //} + // NOTE: left.type <= right.type switch(right.type) { @@ -414,16 +342,9 @@ bool patterns_are_same(const Span& sp, const MacroPatEnt& left, const MacroPatEn switch(left.type) { case MacroPatEnt::PAT_TOKEN: - switch(left.tok.type()) - { - case TOK_BRACE_OPEN: - case TOK_RWORD_LET: + if( is_token_stmt(left.tok.type()) ) ERROR(sp, E0000, "Incompatible macro fragments " << right << " used with " << left); - default: - if( is_token_expr(left.tok.type()) ) - ERROR(sp, E0000, "Incompatible macro fragments " << right << " used with " << left); - return false; - } + return false; case MacroPatEnt::PAT_STMT: return true; default: @@ -443,29 +364,56 @@ bool patterns_are_same(const Span& sp, const MacroPatEnt& left, const MacroPatEn return false; } // Matches meta/attribute fragments. - // - Compatible with everythin but a literal #[ token case MacroPatEnt::PAT_META: switch(left.type) { case MacroPatEnt::PAT_TOKEN: - if( left.tok.type() == TOK_ATTR_OPEN ) + if( left.tok.type() == TOK_IDENT ) ERROR(sp, E0000, "Incompatible macro fragments"); return false; case MacroPatEnt::PAT_META: return true; default: - return false; + ERROR(sp, E0000, "Incompatible macro fragments " << right << " used with " << left); } } throw ""; } -MacroRules Parse_MacroRules(TokenStream& lex) +MacroRulesPatFrag split_fragment_at(MacroRulesPatFrag& frag, unsigned int remaining_count) +{ + MacroRulesPatFrag rv; + for(unsigned int i = remaining_count; i < frag.m_pats_ents.size(); i ++) + rv.m_pats_ents.push_back( mv$(frag.m_pats_ents[i]) ); + frag.m_pats_ents.resize(remaining_count); + rv.m_pattern_end = frag.m_pattern_end; frag.m_pattern_end = ~0; + rv.m_next_frags = mv$(frag.m_next_frags); + return rv; +} + +void enumerate_names(const ::std::vector<MacroPatEnt>& pats, ::std::vector< ::std::string>& names) { + for( const auto& pat : pats ) + { + if( pat.type == MacroPatEnt::PAT_LOOP ) { + enumerate_names(pat.subpats, names); + } + else if( pat.name != "" ) { + auto b = names.begin(); + auto e = names.end(); + if( ::std::find(b, e, pat.name) == e ) { + names.push_back( pat.name ); + } + } + } +} + +MacroRulesPtr Parse_MacroRules(TokenStream& lex) { TRACE_FUNCTION_F(""); Token tok; + // Parse the patterns and replacements ::std::vector<MacroRule> rules; while( GET_TOK(tok, lex) != TOK_EOF ) { @@ -478,14 +426,18 @@ MacroRules Parse_MacroRules(TokenStream& lex) } } - UnifiedPatFrag root_frag; + MacroRulesPatFrag root_frag; + ::std::vector<MacroRulesArm> rule_arms; // Re-parse the patterns into a unified form for(unsigned int rule_idx = 0; rule_idx < rules.size(); rule_idx ++) { const auto& rule = rules[rule_idx]; + MacroRulesArm arm( mv$(rule.m_contents) ); - UnifiedPatFrag* cur_frag = &root_frag; + enumerate_names(rule.m_pattern, arm.m_param_names); + + auto* cur_frag = &root_frag; unsigned int frag_ofs = 0; for( const auto& pat : rule.m_pattern ) { @@ -509,7 +461,7 @@ MacroRules Parse_MacroRules(TokenStream& lex) } // If not, create a new frag if( ! found ) { - cur_frag->m_next_frags.push_back( UnifiedPatFrag() ); + cur_frag->m_next_frags.push_back( MacroRulesPatFrag() ); cur_frag = &cur_frag->m_next_frags.back(); cur_frag->m_pats_ents.push_back( pat ); } @@ -518,12 +470,12 @@ MacroRules Parse_MacroRules(TokenStream& lex) } else if( ! patterns_are_same(sp, cur_frag->m_pats_ents[frag_ofs], pat) ) { // Difference, split the block. - auto new_frag = cur_frag->split_at(frag_ofs); + auto new_frag = split_fragment_at(*cur_frag, frag_ofs); assert( cur_frag->m_next_frags.size() == 0 ); cur_frag->m_next_frags.push_back( mv$(new_frag) ); // - Update cur_frag to a newly pushed fragment, and push this pattern to it - cur_frag->m_next_frags.push_back( UnifiedPatFrag() ); + cur_frag->m_next_frags.push_back( MacroRulesPatFrag() ); cur_frag = &cur_frag->m_next_frags.back(); cur_frag->m_pats_ents.push_back( pat ); frag_ofs = 1; @@ -537,15 +489,21 @@ MacroRules Parse_MacroRules(TokenStream& lex) // If this pattern ended before the current fragment ended if( frag_ofs < cur_frag->m_pats_ents.size() ) { // Split the current fragment - auto new_frag = cur_frag->split_at(frag_ofs); + auto new_frag = split_fragment_at(*cur_frag, frag_ofs); assert( cur_frag->m_next_frags.size() == 0 ); cur_frag->m_next_frags.push_back( mv$(new_frag) ); // Keep cur_frag the same } cur_frag->m_pattern_end = rule_idx; + + rule_arms.push_back( mv$(arm) ); } // TODO: use `root_frag` above for the actual evaluation - return MacroRules( mv$(rules) ); + auto rv = new MacroRules(); + rv->m_pattern = mv$(root_frag); + rv->m_rules = mv$(rule_arms); + + return MacroRulesPtr(rv); } diff --git a/src/macro_rules/pattern_checks.hpp b/src/macro_rules/pattern_checks.hpp new file mode 100644 index 00000000..820a9ef4 --- /dev/null +++ b/src/macro_rules/pattern_checks.hpp @@ -0,0 +1,10 @@ +/* + */ +#pragma once + +extern bool is_token_path(eTokenType tt); +extern bool is_token_pat(eTokenType tt); +extern bool is_token_type(eTokenType tt); +extern bool is_token_expr(eTokenType tt); +extern bool is_token_stmt(eTokenType tt); + diff --git a/src/main.cpp b/src/main.cpp index 963491d5..8bd944ef 100644 --- a/src/main.cpp +++ b/src/main.cpp @@ -22,6 +22,7 @@ int g_debug_indent_level = 0; bool debug_enabled()
{
+ //return g_cur_phase != "Parse";
return g_cur_phase != "Parse" && g_cur_phase != "Expand";
}
::std::ostream& debug_output(int indent, const char* function)
@@ -145,14 +146,16 @@ int main(int argc, char *argv[]) });
// Typecheck / type propagate module (type annotations of all values)
// - Check all generic conditions (ensure referenced trait is valid)
- // > Also mark parameter with applicable traits
+ // > Binds the trait path to the actual trait definition
CompilePhaseV("TypecheckBounds", [&]() {
//Typecheck_GenericBounds(crate);
});
// - Check all generic parameters match required conditions (without doing full typeck)
+ // >
CompilePhaseV("TypecheckParams", [&]() {
//Typecheck_GenericParams(crate);
});
+ // TODO: Evaluate all constants (or MIR them then evaluate)
// - Full function typeck
CompilePhaseV("TypecheckExpr", [&]() {
//Typecheck_Expr(crate);
diff --git a/src/parse/common.hpp b/src/parse/common.hpp index 9ba8b8cd..d7f84315 100644 --- a/src/parse/common.hpp +++ b/src/parse/common.hpp @@ -44,8 +44,6 @@ extern ::AST::MacroInvocation Parse_MacroInvocation(ProtoSpan ps, ::AST::MetaIte extern TypeRef Parse_Type(TokenStream& lex);
extern AST::Pattern Parse_Pattern(TokenStream& lex, bool is_refutable);
-extern MacroRules Parse_MacroRules(TokenStream& lex);
-
extern void Parse_Impl_Item(TokenStream& lex, AST::Impl& impl);
extern void Parse_Mod_Item(TokenStream& lex, bool file_controls_dir, const ::std::string& file_path, AST::Module& mod, bool is_public, AST::MetaItems meta_items);
extern void Parse_ModRoot_Items(TokenStream& lex, AST::Module& mod, bool file_controls_dir, const ::std::string& path);
diff --git a/src/parse/expr.cpp b/src/parse/expr.cpp index e047c408..7094cf96 100644 --- a/src/parse/expr.cpp +++ b/src/parse/expr.cpp @@ -13,7 +13,6 @@ #include "parseerror.hpp"
#include "../ast/ast.hpp"
#include "common.hpp"
-#include "../macros.hpp"
#include <iostream>
#include "tokentree.hpp"
diff --git a/src/parse/root.cpp b/src/parse/root.cpp index dd3d8904..4f856fbc 100644 --- a/src/parse/root.cpp +++ b/src/parse/root.cpp @@ -13,7 +13,6 @@ #include "../ast/crate.hpp"
#include "parseerror.hpp"
#include "common.hpp"
-#include "../macros.hpp"
#include <cassert>
::std::string dirname(::std::string input) {
|