#include #include "cs_vm.hh" #include "cs_std.hh" #include "cs_parser.hh" #include #include #include namespace cscript { std::string_view codegen_state::get_str() { size_t nl; char const *beg = source; source = parse_string( cs, std::string_view{source, std::size_t(send - source)}, nl ); current_line += nl - 1; auto ret = std::string_view{beg, std::size_t(source - beg)}; return ret.substr(1, ret.size() - 2); } charbuf codegen_state::get_str_dup() { charbuf buf{cs}; unescape_string(std::back_inserter(buf), get_str()); return buf; } std::string_view codegen_state::read_macro_name() { char const *op = source; char c = current(); if (!isalpha(c) && (c != '_')) { return std::string_view{}; } for (; isalnum(c) || (c == '_'); c = current()) { next_char(); } return std::string_view{op, std::size_t(source - op)}; } char codegen_state::skip_until(std::string_view chars) { char c = current(); while (c && (chars.find(c) == std::string_view::npos)) { next_char(); c = current(); } return c; } char codegen_state::skip_until(char cf) { char c = current(); while (c && (c != cf)) { next_char(); c = current(); } return c; } static bool is_hspace(char c) { return (c == ' ') || (c == '\t') || (c == '\r'); } void codegen_state::skip_comments() { for (;;) { for (char c = current(); is_hspace(c); c = current()) { next_char(); } if (current() == '\\') { char c = current(1); if ((c != '\r') && (c != '\n')) { throw error(cs, "invalid line break"); } /* skip backslash */ next_char(); /* skip CR or LF */ next_char(); /* when CR, try also skipping LF; covers \r, \n, \r\n */ if ((c == '\r') && (current(1) == '\n')) { next_char(); } /* skip whitespace on new line */ continue; } if ((current() != '/') || (current(1) != '/')) { return; } while (current() != '\n') { next_char(); } } } std::string_view codegen_state::get_word() { char const *beg = source; source = parse_word( cs, std::string_view{source, std::size_t(send - source)} ); return std::string_view{beg, std::size_t(source - beg)}; } static inline int ret_code(int type, int def = 0) { if (type >= VAL_ANY) { return (type == VAL_STRING) ? BC_RET_STRING : def; } return type << BC_INST_RET; } static void compilestatements( codegen_state &gs, int rettype, int brak = '\0', int prevargs = 0 ); static inline std::pair compileblock( codegen_state &gs, std::string_view p, size_t line, int rettype = BC_RET_NULL, int brak = '\0' ); void codegen_state::gen_int(std::string_view word) { gen_int(parse_int(word)); } void codegen_state::gen_float(std::string_view word) { gen_float(parse_float(word)); } void codegen_state::gen_value(int wordtype, std::string_view word, int line) { switch (wordtype) { case VAL_ANY: if (!word.empty()) { gen_str(word); } else { gen_null(); } break; case VAL_STRING: gen_str(word); break; case VAL_FLOAT: gen_float(word); break; case VAL_INT: gen_int(word); break; case VAL_COND: if (!word.empty()) { compileblock(*this, word, line); } else { gen_null(); } break; case VAL_CODE: compileblock(*this, word, line); break; case VAL_IDENT: gen_ident(word); break; default: break; } } static inline void compileblock(codegen_state &gs) { gs.code.push_back(BC_INST_EMPTY); } static inline std::pair compileblock( codegen_state &gs, std::string_view p, size_t line, int rettype, int brak ) { size_t start = gs.code.size(); gs.code.push_back(BC_INST_BLOCK); gs.code.push_back(BC_INST_OFFSET | ((start + 2) << 8)); size_t retline = line; if (!p.empty()) { char const *op = gs.source, *oe = gs.send; size_t oldline = gs.current_line; gs.source = p.data(); gs.send = p.data() + p.size(); gs.current_line = line; compilestatements(gs, VAL_ANY, brak); p = std::string_view{gs.source, std::size_t(gs.send - gs.source)}; retline = gs.current_line; gs.source = op; gs.send = oe; gs.current_line = oldline; } if (gs.code.size() > start + 2) { gs.code.push_back(BC_INST_EXIT | rettype); gs.code[start] |= uint32_t(gs.code.size() - (start + 1)) << 8; } else { gs.code.resize(start); gs.code.push_back(BC_INST_EMPTY | rettype); } return std::make_pair(p, retline); } static inline void compileunescapestr(codegen_state &gs) { auto str = gs.get_str(); gs.code.push_back(BC_INST_VAL | BC_RET_STRING); gs.code.reserve( gs.code.size() + str.size() / sizeof(uint32_t) + 1 ); size_t bufs = (gs.code.capacity() - gs.code.size()) * sizeof(uint32_t); auto alloc = std_allocator{gs.cs}; auto *buf = alloc.allocate(bufs + 1); char *wbuf = unescape_string(&buf[0], str); memset( &buf[wbuf - buf], 0, sizeof(uint32_t) - (wbuf - buf) % sizeof(uint32_t) ); gs.code.back() |= (wbuf - buf) << 8; uint32_t *ubuf = reinterpret_cast(buf); gs.code.append(ubuf, ubuf + ((wbuf - buf) / sizeof(uint32_t) + 1)); alloc.deallocate(buf, bufs + 1); } static bool compilearg( codegen_state &gs, int wordtype, int prevargs = MaxResults, charbuf *word = nullptr ); static void compilelookup(codegen_state &gs, int ltype, int prevargs = MaxResults) { charbuf lookup{gs.cs}; gs.next_char(); switch (gs.current()) { case '(': case '[': if (!compilearg(gs, VAL_STRING, prevargs)) { goto invalid; } break; case '$': compilelookup(gs, VAL_STRING, prevargs); break; case '\"': lookup = gs.get_str_dup(); lookup.push_back('\0'); goto lookupid; default: { lookup.append(gs.get_word()); if (lookup.empty()) goto invalid; lookup.push_back('\0'); lookupid: ident *id = gs.cs.new_ident(lookup.str_term()); if (id) { switch (id->get_type()) { case ident_type::IVAR: gs.code.push_back( BC_INST_IVAR | ret_code(ltype, BC_RET_INT) | (id->get_index() << 8) ); switch (ltype) { case VAL_POP: gs.code.pop_back(); break; case VAL_CODE: gs.code.push_back(BC_INST_COMPILE); break; case VAL_IDENT: gs.code.push_back(BC_INST_IDENT_U); break; } return; case ident_type::FVAR: gs.code.push_back( BC_INST_FVAR | ret_code(ltype, BC_RET_FLOAT) | (id->get_index() << 8) ); switch (ltype) { case VAL_POP: gs.code.pop_back(); break; case VAL_CODE: gs.code.push_back(BC_INST_COMPILE); break; case VAL_IDENT: gs.code.push_back(BC_INST_IDENT_U); break; } return; case ident_type::SVAR: switch (ltype) { case VAL_POP: return; default: gs.code.push_back( BC_INST_SVAR | ret_code(ltype, BC_RET_STRING) | (id->get_index() << 8) ); break; } goto done; case ident_type::ALIAS: switch (ltype) { case VAL_POP: return; case VAL_COND: gs.code.push_back( (id->get_index() < MaxArguments ? BC_INST_LOOKUP_MARG : BC_INST_LOOKUP_M ) | (id->get_index() << 8) ); break; case VAL_CODE: case VAL_IDENT: gs.code.push_back( (id->get_index() < MaxArguments ? BC_INST_LOOKUP_MARG : BC_INST_LOOKUP_M ) | BC_RET_STRING | (id->get_index() << 8) ); break; default: gs.code.push_back( (id->get_index() < MaxArguments ? BC_INST_LOOKUP_ARG : BC_INST_LOOKUP ) | ret_code(ltype, BC_RET_STRING) | (id->get_index() << 8) ); break; } goto done; case ident_type::COMMAND: { int comtype = BC_INST_COM, numargs = 0; if (prevargs >= MaxResults) { gs.code.push_back(BC_INST_ENTER); } auto fmt = static_cast(id)->get_args(); for (char c: fmt) { switch (c) { case 'S': gs.gen_str(); numargs++; break; case 's': gs.gen_str(std::string_view{}); numargs++; break; case 'i': gs.gen_int(); numargs++; break; case 'b': gs.gen_int(std::numeric_limits::min()); numargs++; break; case 'f': gs.gen_float(); numargs++; break; case 'F': gs.code.push_back(BC_INST_DUP | BC_RET_FLOAT); numargs++; break; case 'E': case 'T': case 't': gs.gen_null(); numargs++; break; case 'e': compileblock(gs); numargs++; break; case 'r': gs.gen_ident(); numargs++; break; case '$': gs.gen_ident(id); numargs++; break; case 'N': gs.gen_int(-1); numargs++; break; case 'C': comtype = BC_INST_COM_C; goto compilecomv; case 'V': comtype = BC_INST_COM_V; goto compilecomv; case '1': case '2': case '3': case '4': break; } } gs.code.push_back( comtype | ret_code(ltype) | (id->get_index() << 8) ); gs.code.push_back( (prevargs >= MaxResults ? BC_INST_EXIT : BC_INST_RESULT_ARG ) | ret_code(ltype) ); goto done; compilecomv: gs.code.push_back( comtype | ret_code(ltype) | (numargs << 8) | (id->get_index() << 13) ); gs.code.push_back( (prevargs >= MaxResults ? BC_INST_EXIT : BC_INST_RESULT_ARG ) | ret_code(ltype) ); goto done; } default: goto invalid; } } gs.gen_str(lookup.str_term()); break; } } switch (ltype) { case VAL_COND: gs.code.push_back(BC_INST_LOOKUP_MU); break; case VAL_CODE: case VAL_IDENT: gs.code.push_back(BC_INST_LOOKUP_MU | BC_RET_STRING); break; default: gs.code.push_back(BC_INST_LOOKUP_U | ret_code(ltype)); break; } done: switch (ltype) { case VAL_POP: gs.code.push_back(BC_INST_POP); break; case VAL_CODE: gs.code.push_back(BC_INST_COMPILE); break; case VAL_COND: gs.code.push_back(BC_INST_COND); break; case VAL_IDENT: gs.code.push_back(BC_INST_IDENT_U); break; } return; invalid: switch (ltype) { case VAL_POP: break; case VAL_NULL: case VAL_ANY: case VAL_WORD: case VAL_COND: gs.gen_null(); break; default: gs.gen_value(ltype); break; } } static bool compileblockstr(codegen_state &gs, char const *str, char const *send) { int startc = gs.code.size(); gs.code.push_back(BC_INST_VAL | BC_RET_STRING); gs.code.reserve(gs.code.size() + (send - str) / sizeof(uint32_t) + 1); auto alloc = std_allocator{gs.cs}; auto asz = ((send - str) / sizeof(uint32_t) + 1) * sizeof(uint32_t); char *buf = alloc.allocate(asz); int len = 0; while (str < send) { std::string_view chrs{"\r/\"@]"}; char const *orig = str; str = std::find_first_of(str, send, chrs.begin(), chrs.end()); memcpy(&buf[len], orig, str - orig); len += (str - orig); if (str == send) { goto done; } switch (*str) { case '\r': ++str; break; case '\"': { char const *start = str; str = parse_string( gs.cs, std::string_view{str, send} ); memcpy(&buf[len], start, std::size_t(str - start)); len += (str - start); break; } case '/': if (((str + 1) != send) && str[1] == '/') { str = std::find(str, send, '\n'); } else { buf[len++] = *str++; } break; case '@': case ']': if (str < send) { buf[len++] = *str++; } else { goto done; } break; } } done: memset(&buf[len], '\0', sizeof(uint32_t) - len % sizeof(uint32_t)); uint32_t *ubuf = reinterpret_cast(buf); gs.code.append(ubuf, ubuf + (len / sizeof(uint32_t) + 1)); gs.code[startc] |= len << 8; alloc.deallocate(buf, asz); return true; } static bool compileblocksub(codegen_state &gs, int prevargs) { charbuf lookup{gs.cs}; switch (gs.current()) { case '(': if (!compilearg(gs, VAL_ANY, prevargs)) { return false; } break; case '[': if (!compilearg(gs, VAL_STRING, prevargs)) { return false; } gs.code.push_back(BC_INST_LOOKUP_MU); break; case '\"': lookup = gs.get_str_dup(); lookup.push_back('\0'); goto lookupid; default: { lookup.append(gs.read_macro_name()); if (lookup.empty()) { return false; } lookup.push_back('\0'); lookupid: ident *id = gs.cs.new_ident(lookup.str_term()); if (id) { switch (id->get_type()) { case ident_type::IVAR: gs.code.push_back(BC_INST_IVAR | (id->get_index() << 8)); goto done; case ident_type::FVAR: gs.code.push_back(BC_INST_FVAR | (id->get_index() << 8)); goto done; case ident_type::SVAR: gs.code.push_back(BC_INST_SVAR | (id->get_index() << 8)); goto done; case ident_type::ALIAS: gs.code.push_back( (id->get_index() < MaxArguments ? BC_INST_LOOKUP_MARG : BC_INST_LOOKUP_M ) | (id->get_index() << 8) ); goto done; default: break; } } gs.gen_str(lookup.str_term()); gs.code.push_back(BC_INST_LOOKUP_MU); done: break; } } return true; } static void compileblockmain(codegen_state &gs, int wordtype, int prevargs) { char const *start = gs.source; size_t curline = gs.current_line; int concs = 0; for (int brak = 1; brak;) { switch (gs.skip_until("@\"/[]")) { case '\0': throw error(gs.cs, "missing \"]\""); return; case '\"': gs.get_str(); break; case '/': gs.next_char(); if (gs.current() == '/') { gs.skip_until('\n'); } break; case '[': gs.next_char(); brak++; break; case ']': gs.next_char(); brak--; break; case '@': { char const *esc = gs.source; int level = 0; while (gs.current() == '@') { ++level; gs.next_char(); } if (brak > level) { continue; } else if (brak < level) { throw error(gs.cs, "too many @s"); return; } if (!concs && prevargs >= MaxResults) { gs.code.push_back(BC_INST_ENTER); } if (concs + 2 > MaxArguments) { gs.code.push_back(BC_INST_CONC_W | BC_RET_STRING | (concs << 8)); concs = 1; } if (compileblockstr(gs, start, esc)) { concs++; } if (compileblocksub(gs, prevargs + concs)) { concs++; } if (concs) { start = gs.source; curline = gs.current_line; } else if (prevargs >= MaxResults) { gs.code.pop_back(); } break; } default: gs.next_char(); break; } } if (gs.source - 1 > start) { if (!concs) { switch (wordtype) { case VAL_POP: return; case VAL_CODE: case VAL_COND: { auto ret = compileblock(gs, std::string_view{ start, std::size_t(gs.send - start) }, curline, BC_RET_NULL, ']'); gs.source = ret.first.data(); gs.send = ret.first.data() + ret.first.size(); gs.current_line = ret.second; return; } case VAL_IDENT: gs.gen_ident(std::string_view{ start, std::size_t((gs.source - 1) - start) }); return; } } compileblockstr(gs, start, gs.source - 1); if (concs > 1) { concs++; } } if (concs) { if (prevargs >= MaxResults) { gs.code.push_back(BC_INST_CONC_M | ret_code(wordtype) | (concs << 8)); gs.code.push_back(BC_INST_EXIT | ret_code(wordtype)); } else { gs.code.push_back(BC_INST_CONC_W | ret_code(wordtype) | (concs << 8)); } } switch (wordtype) { case VAL_POP: if (concs || gs.source - 1 > start) { gs.code.push_back(BC_INST_POP); } break; case VAL_COND: if (!concs && gs.source - 1 <= start) { gs.gen_null(); } else { gs.code.push_back(BC_INST_COND); } break; case VAL_CODE: if (!concs && gs.source - 1 <= start) { compileblock(gs); } else { gs.code.push_back(BC_INST_COMPILE); } break; case VAL_IDENT: if (!concs && gs.source - 1 <= start) { gs.gen_ident(); } else { gs.code.push_back(BC_INST_IDENT_U); } break; case VAL_STRING: case VAL_NULL: case VAL_ANY: case VAL_WORD: if (!concs && gs.source - 1 <= start) { gs.gen_str(); } break; default: if (!concs) { if (gs.source - 1 <= start) { gs.gen_value(wordtype); } else { gs.code.push_back(BC_INST_FORCE | (wordtype << BC_INST_RET)); } } break; } } static bool compilearg( codegen_state &gs, int wordtype, int prevargs, charbuf *word ) { gs.skip_comments(); switch (gs.current()) { case '\"': switch (wordtype) { case VAL_POP: gs.get_str(); break; case VAL_COND: { size_t line = gs.current_line; auto s = gs.get_str_dup(); if (!s.empty()) { s.push_back('\0'); compileblock(gs, s.str_term(), line); } else { gs.gen_null(); } break; } case VAL_CODE: { auto s = gs.get_str_dup(); s.push_back('\0'); compileblock(gs, s.str_term(), gs.current_line); break; } case VAL_WORD: if (word) { *word = std::move(gs.get_str_dup()); } break; case VAL_ANY: case VAL_STRING: compileunescapestr(gs); break; default: { size_t line = gs.current_line; auto s = gs.get_str_dup(); s.push_back('\0'); gs.gen_value(wordtype, s.str_term(), line); break; } } return true; case '$': compilelookup(gs, wordtype, prevargs); return true; case '(': gs.next_char(); if (prevargs >= MaxResults) { gs.code.push_back(BC_INST_ENTER); compilestatements(gs, VAL_ANY, ')'); gs.code.push_back(BC_INST_EXIT | ret_code(wordtype)); } else { size_t start = gs.code.size(); compilestatements(gs, VAL_ANY, ')', prevargs); if (gs.code.size() > start) { gs.code.push_back(BC_INST_RESULT_ARG | ret_code(wordtype)); } else { gs.gen_value(wordtype); return true; } } switch (wordtype) { case VAL_POP: gs.code.push_back(BC_INST_POP); break; case VAL_COND: gs.code.push_back(BC_INST_COND); break; case VAL_CODE: gs.code.push_back(BC_INST_COMPILE); break; case VAL_IDENT: gs.code.push_back(BC_INST_IDENT_U); break; } return true; case '[': gs.next_char(); compileblockmain(gs, wordtype, prevargs); return true; default: switch (wordtype) { case VAL_POP: { return !gs.get_word().empty(); } case VAL_COND: { size_t line = gs.current_line; auto s = gs.get_word(); if (s.empty()) { return false; } compileblock(gs, s, line); return true; } case VAL_CODE: { size_t line = gs.current_line; auto s = gs.get_word(); if (s.empty()) { return false; } compileblock(gs, s, line); return true; } case VAL_WORD: { auto w = gs.get_word(); if (word) { word->clear(); word->append(w); } return !w.empty(); } default: { size_t line = gs.current_line; auto s = gs.get_word(); if (s.empty()) { return false; } gs.gen_value(wordtype, s, line); return true; } } } } static void compile_cmd( codegen_state &gs, command_impl *id, bool &more, int rettype, int prevargs ) { int comtype = BC_INST_COM, numargs = 0, fakeargs = 0; bool rep = false; auto fmt = id->get_args(); for (auto it = fmt.begin(); it != fmt.end(); ++it) { switch (*it) { case 's': /* string */ if (more) { more = compilearg(gs, VAL_STRING, prevargs + numargs); } if (!more) { if (rep) { break; } gs.gen_str(std::string_view{}); fakeargs++; } else if ((it + 1) == fmt.end()) { int numconc = 1; while ((numargs + numconc) < MaxArguments) { more = compilearg( gs, VAL_STRING, prevargs + numargs + numconc ); if (!more) { break; } numconc++; } if (numconc > 1) { gs.code.push_back(BC_INST_CONC | BC_RET_STRING | (numconc << 8)); } } numargs++; break; case 'i': /* integer */ if (more) { more = compilearg(gs, VAL_INT, prevargs + numargs); } if (!more) { if (rep) { break; } gs.gen_int(); fakeargs++; } numargs++; break; case 'b': /* integer, INT_MIN default */ if (more) { more = compilearg(gs, VAL_INT, prevargs + numargs); } if (!more) { if (rep) { break; } gs.gen_int(std::numeric_limits::min()); fakeargs++; } numargs++; break; case 'f': /* float */ if (more) { more = compilearg(gs, VAL_FLOAT, prevargs + numargs); } if (!more) { if (rep) { break; } gs.gen_float(); fakeargs++; } numargs++; break; case 'F': /* float, prev-argument default */ if (more) { more = compilearg(gs, VAL_FLOAT, prevargs + numargs); } if (!more) { if (rep) { break; } gs.code.push_back(BC_INST_DUP | BC_RET_FLOAT); fakeargs++; } numargs++; break; case 't': /* any arg */ if (more) { more = compilearg( gs, VAL_ANY, prevargs + numargs ); } if (!more) { if (rep) { break; } gs.gen_null(); fakeargs++; } numargs++; break; case 'E': /* condition */ if (more) { more = compilearg(gs, VAL_COND, prevargs + numargs); } if (!more) { if (rep) { break; } gs.gen_null(); fakeargs++; } numargs++; break; case 'e': /* code */ if (more) { more = compilearg(gs, VAL_CODE, prevargs + numargs); } if (!more) { if (rep) { break; } compileblock(gs); fakeargs++; } numargs++; break; case 'r': /* ident */ if (more) { more = compilearg(gs, VAL_IDENT, prevargs + numargs); } if (!more) { if (rep) { break; } gs.gen_ident(); fakeargs++; } numargs++; break; case '$': /* self */ gs.gen_ident(id); numargs++; break; case 'N': /* number of arguments */ gs.gen_int(numargs - fakeargs); numargs++; break; case 'C': /* concatenated string */ comtype = BC_INST_COM_C; if (more) { while (numargs < MaxArguments) { more = compilearg(gs, VAL_ANY, prevargs + numargs); if (!more) { break; } numargs++; } } goto compilecomv; case 'V': /* varargs */ comtype = BC_INST_COM_V; if (more) { while (numargs < MaxArguments) { more = compilearg(gs, VAL_ANY, prevargs + numargs); if (!more) { break; } numargs++; } } goto compilecomv; case '1': /* vararg repetition */ case '2': case '3': case '4': if (more && (numargs < MaxArguments)) { int numrep = *it - '0' + 1; it -= numrep; rep = true; } else { while (numargs > MaxArguments) { gs.code.push_back(BC_INST_POP); --numargs; } } break; } } gs.code.push_back(comtype | ret_code(rettype) | (id->get_index() << 8)); return; compilecomv: gs.code.push_back( comtype | ret_code(rettype) | (numargs << 8) | (id->get_index() << 13) ); } static void compile_alias(codegen_state &gs, alias *id, bool &more, int prevargs) { int numargs = 0; while (numargs < MaxArguments) { more = compilearg(gs, VAL_ANY, prevargs + numargs); if (!more) { break; } ++numargs; } gs.code.push_back( (id->get_index() < MaxArguments ? BC_INST_CALL_ARG : BC_INST_CALL) | (numargs << 8) | (id->get_index() << 13) ); } static void compile_local(codegen_state &gs, bool &more, int prevargs) { int numargs = 0; if (more) { while (numargs < MaxArguments) { more = compilearg(gs, VAL_IDENT, prevargs + numargs); if (!more) { break; } numargs++; } } if (more) { while ((more = compilearg(gs, VAL_POP))); } gs.code.push_back(BC_INST_LOCAL | (numargs << 8)); } static void compile_do( codegen_state &gs, bool &more, int prevargs, int rettype, int opcode ) { if (more) { more = compilearg(gs, VAL_CODE, prevargs); } gs.code.push_back((more ? opcode : BC_INST_NULL) | ret_code(rettype)); } static void compile_if( codegen_state &gs, ident *id, bool &more, int prevargs, int rettype ) { if (more) { more = compilearg(gs, VAL_ANY, prevargs); } if (!more) { gs.code.push_back(BC_INST_NULL | ret_code(rettype)); } else { int start1 = gs.code.size(); more = compilearg(gs, VAL_CODE, prevargs + 1); if (!more) { gs.code.push_back(BC_INST_POP); gs.code.push_back(BC_INST_NULL | ret_code(rettype)); } else { int start2 = gs.code.size(); more = compilearg(gs, VAL_CODE, prevargs + 2); uint32_t inst1 = gs.code[start1]; uint32_t op1 = inst1 & ~BC_INST_RET_MASK; uint32_t len1 = start2 - (start1 + 1); if (!more) { if (op1 == (BC_INST_BLOCK | (len1 << 8))) { gs.code[start1] = (len1 << 8) | BC_INST_JUMP_B | BC_INST_FLAG_FALSE; gs.code[start1 + 1] = BC_INST_ENTER_RESULT; gs.code[start1 + len1] = ( gs.code[start1 + len1] & ~BC_INST_RET_MASK ) | ret_code(rettype); return; } compileblock(gs); } else { uint32_t inst2 = gs.code[start2]; uint32_t op2 = inst2 & ~BC_INST_RET_MASK; uint32_t len2 = gs.code.size() - (start2 + 1); if (op2 == (BC_INST_BLOCK | (len2 << 8))) { if (op1 == (BC_INST_BLOCK | (len1 << 8))) { gs.code[start1] = ((start2 - start1) << 8) | BC_INST_JUMP_B | BC_INST_FLAG_FALSE; gs.code[start1 + 1] = BC_INST_ENTER_RESULT; gs.code[start1 + len1] = ( gs.code[start1 + len1] & ~BC_INST_RET_MASK ) | ret_code(rettype); gs.code[start2] = (len2 << 8) | BC_INST_JUMP; gs.code[start2 + 1] = BC_INST_ENTER_RESULT; gs.code[start2 + len2] = ( gs.code[start2 + len2] & ~BC_INST_RET_MASK ) | ret_code(rettype); return; } else if (op1 == (BC_INST_EMPTY | (len1 << 8))) { gs.code[start1] = BC_INST_NULL | (inst2 & BC_INST_RET_MASK); gs.code[start2] = (len2 << 8) | BC_INST_JUMP_B | BC_INST_FLAG_TRUE; gs.code[start2 + 1] = BC_INST_ENTER_RESULT; gs.code[start2 + len2] = ( gs.code[start2 + len2] & ~BC_INST_RET_MASK ) | ret_code(rettype); return; } } } gs.code.push_back(BC_INST_COM | ret_code(rettype) | (id->get_index() << 8)); } } } static void compile_and_or( codegen_state &gs, ident *id, bool &more, int prevargs, int rettype ) { int numargs = 0; if (more) { more = compilearg(gs, VAL_COND, prevargs); } if (!more) { gs.code.push_back( ((id->get_raw_type() == ID_AND) ? BC_INST_TRUE : BC_INST_FALSE) | ret_code(rettype) ); } else { numargs++; int start = gs.code.size(), end = start; while (numargs < MaxArguments) { more = compilearg(gs, VAL_COND, prevargs + numargs); if (!more) { break; } numargs++; if ((gs.code[end] & ~BC_INST_RET_MASK) != ( BC_INST_BLOCK | (uint32_t(gs.code.size() - (end + 1)) << 8) )) { break; } end = gs.code.size(); } if (more) { while (numargs < MaxArguments) { more = compilearg(gs, VAL_COND, prevargs + numargs); if (!more) { break; } numargs++; } gs.code.push_back( BC_INST_COM_V | ret_code(rettype) | (numargs << 8) | (id->get_index() << 13) ); } else { uint32_t op = (id->get_raw_type() == ID_AND) ? (BC_INST_JUMP_RESULT | BC_INST_FLAG_FALSE) : (BC_INST_JUMP_RESULT | BC_INST_FLAG_TRUE); gs.code.push_back(op); end = gs.code.size(); while ((start + 1) < end) { uint32_t len = gs.code[start] >> 8; gs.code[start] = ((end - (start + 1)) << 8) | op; gs.code[start + 1] = BC_INST_ENTER; gs.code[start + len] = ( gs.code[start + len] & ~BC_INST_RET_MASK ) | ret_code(rettype); start += len + 1; } } } } static void compilestatements(codegen_state &gs, int rettype, int brak, int prevargs) { charbuf idname{gs.cs}; for (;;) { gs.skip_comments(); idname.clear(); size_t curline = gs.current_line; bool more = compilearg(gs, VAL_WORD, prevargs, &idname); if (!more) { goto endstatement; } gs.skip_comments(); if (gs.current() == '=') { switch (gs.current(1)) { case '/': if (gs.current(2) != '/') { break; } [[fallthrough]]; case ';': case ' ': case '\t': case '\r': case '\n': case '\0': gs.next_char(); if (!idname.empty()) { idname.push_back('\0'); ident *id = gs.cs.new_ident(idname.str_term()); if (id) { switch (id->get_type()) { case ident_type::ALIAS: more = compilearg(gs, VAL_ANY, prevargs); if (!more) { gs.gen_str(); } gs.code.push_back( (id->get_index() < MaxArguments ? BC_INST_ALIAS_ARG : BC_INST_ALIAS ) | (id->get_index() << 8) ); goto endstatement; case ident_type::IVAR: more = compilearg(gs, VAL_INT, prevargs); if (!more) { gs.gen_int(); } gs.code.push_back( BC_INST_IVAR1 | (id->get_index() << 8) ); goto endstatement; case ident_type::FVAR: more = compilearg(gs, VAL_FLOAT, prevargs); if (!more) { gs.gen_float(); } gs.code.push_back( BC_INST_FVAR1 | (id->get_index() << 8) ); goto endstatement; case ident_type::SVAR: more = compilearg(gs, VAL_STRING, prevargs); if (!more) { gs.gen_str(); } gs.code.push_back( BC_INST_SVAR1 | (id->get_index() << 8) ); goto endstatement; default: break; } } gs.gen_str(idname.str_term()); } more = compilearg(gs, VAL_ANY); if (!more) { gs.gen_str(); } gs.code.push_back(BC_INST_ALIAS_U); goto endstatement; } } if (idname.empty()) { noid: int numargs = 0; while (numargs < MaxArguments) { more = compilearg(gs, VAL_ANY, prevargs + numargs); if (!more) { break; } ++numargs; } gs.code.push_back(BC_INST_CALL_U | (numargs << 8)); } else { idname.push_back('\0'); ident *id = gs.cs.get_ident(idname.str_term()); if (!id) { if (is_valid_name(idname.str_term())) { gs.gen_str(idname.str_term()); goto noid; } switch (rettype) { case VAL_ANY: { std::string_view end = idname.str_term(); integer_type val = parse_int(end, &end); if (!end.empty()) { gs.gen_str(idname.str_term()); } else { gs.gen_int(val); } break; } default: gs.gen_value(rettype, idname.str_term(), curline); break; } gs.code.push_back(BC_INST_RESULT); } else { switch (id->get_raw_type()) { case ID_ALIAS: compile_alias( gs, static_cast(id), more, prevargs ); break; case ID_COMMAND: compile_cmd( gs, static_cast(id), more, rettype, prevargs ); break; case ID_LOCAL: compile_local(gs, more, prevargs); break; case ID_DO: compile_do(gs, more, prevargs, rettype, BC_INST_DO); break; case ID_DOARGS: compile_do(gs, more, prevargs, rettype, BC_INST_DO_ARGS); break; case ID_IF: compile_if(gs, id, more, prevargs, rettype); break; case ID_BREAK: gs.code.push_back(BC_INST_BREAK | BC_INST_FLAG_FALSE); break; case ID_CONTINUE: gs.code.push_back(BC_INST_BREAK | BC_INST_FLAG_TRUE); break; case ID_RESULT: if (more) { more = compilearg(gs, VAL_ANY, prevargs); } gs.code.push_back( (more ? BC_INST_RESULT : BC_INST_NULL) | ret_code(rettype) ); break; case ID_NOT: if (more) { more = compilearg(gs, VAL_ANY, prevargs); } gs.code.push_back( (more ? BC_INST_NOT : BC_INST_TRUE) | ret_code(rettype) ); break; case ID_AND: case ID_OR: compile_and_or(gs, id, more, prevargs, rettype); break; case ID_IVAR: if (!(more = compilearg(gs, VAL_INT, prevargs))) { gs.code.push_back(BC_INST_PRINT | (id->get_index() << 8)); } else if (!(id->get_flags() & IDENT_FLAG_HEX) || !( more = compilearg(gs, VAL_INT, prevargs + 1) )) { gs.code.push_back(BC_INST_IVAR1 | (id->get_index() << 8)); } else if (!( more = compilearg(gs, VAL_INT, prevargs + 2) )) { gs.code.push_back(BC_INST_IVAR2 | (id->get_index() << 8)); } else { gs.code.push_back(BC_INST_IVAR3 | (id->get_index() << 8)); } break; case ID_FVAR: if (!(more = compilearg(gs, VAL_FLOAT, prevargs))) { gs.code.push_back(BC_INST_PRINT | (id->get_index() << 8)); } else { gs.code.push_back(BC_INST_FVAR1 | (id->get_index() << 8)); } break; case ID_SVAR: if (!(more = compilearg(gs, VAL_STRING, prevargs))) { gs.code.push_back(BC_INST_PRINT | (id->get_index() << 8)); } else { int numargs = 0; do { ++numargs; } while (numargs < MaxArguments && ( more = compilearg( gs, VAL_ANY, prevargs + numargs ) )); if (numargs > 1) { gs.code.push_back( BC_INST_CONC | BC_RET_STRING | (numargs << 8) ); } gs.code.push_back(BC_INST_SVAR1 | (id->get_index() << 8)); } break; } } } endstatement: if (more) { while (compilearg(gs, VAL_POP)); } switch (gs.skip_until(")];/\n")) { case '\0': if (gs.current() != brak) { throw error(gs.cs, "missing \"%c\"", char(brak)); return; } return; case ')': case ']': if (gs.current() == brak) { gs.next_char(); return; } throw error(gs.cs, "unexpected \"%c\"", gs.current()); return; case '/': gs.next_char(); if (gs.current() == '/') { gs.skip_until('\n'); } goto endstatement; default: gs.next_char(); break; } } } void codegen_state::gen_main(std::string_view s, int ret_type) { source = s.data(); send = s.data() + s.size(); code.push_back(BC_INST_START); compilestatements(*this, VAL_ANY); code.push_back(BC_INST_EXIT | ((ret_type < VAL_ANY) ? (ret_type << BC_INST_RET) : 0)); } } /* namespace cscript */