Annotation of parser3/src/main/compile_tools.h, revision 1.119
1.32 paf 1: /** @file
1.33 paf 2: Parser: compiler support helper functions decls.
3:
1.117 moko 4: Copyright (c) 2001-2023 Art. Lebedev Studio (http://www.artlebedev.com)
5: Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <paf@design.ru>
1.1 paf 6: */
7:
8: #ifndef COMPILE_TOOLS
9: #define COMPILE_TOOLS
1.62 paf 10:
1.119 ! moko 11: #define IDENT_COMPILE_TOOLS_H "$Id: compile_tools.h,v 1.118 2024/09/16 23:59:31 moko Exp $"
1.1 paf 12:
1.39 paf 13: #include "pa_opcode.h"
1.1 paf 14: #include "pa_types.h"
1.10 paf 15: #include "pa_vstring.h"
1.14 paf 16: #include "pa_request.h"
1.1 paf 17:
1.68 paf 18: /// used to track source column number
19: #define TAB_SIZE 8
20:
1.1 paf 21: enum lexical_state {
1.48 parser 22: LS_USER, LS_NAME_SQUARE_PART,
1.65 paf 23: LS_USER_COMMENT,
1.7 paf 24: LS_DEF_NAME,
25: LS_DEF_PARAMS,
26: LS_DEF_LOCALS,
27: LS_DEF_COMMENT,
1.18 paf 28: LS_DEF_SPECIAL_BODY,
1.42 parser 29: LS_EXPRESSION_STRING_QUOTED,
30: LS_EXPRESSION_STRING_APOSTROFED,
1.48 parser 31: LS_EXPRESSION_VAR_NAME_WITH_COLON, LS_EXPRESSION_VAR_NAME_WITHOUT_COLON,
1.65 paf 32: LS_EXPRESSION_COMMENT,
1.48 parser 33: LS_VAR_NAME_SIMPLE_WITH_COLON, LS_VAR_NAME_SIMPLE_WITHOUT_COLON,
1.1 paf 34: LS_VAR_NAME_CURLY,
1.22 paf 35: LS_VAR_ROUND,
1.18 paf 36: LS_VAR_SQUARE,
1.1 paf 37: LS_VAR_CURLY,
38: LS_METHOD_NAME,
1.18 paf 39: LS_METHOD_SQUARE,
1.1 paf 40: LS_METHOD_CURLY,
1.22 paf 41: LS_METHOD_ROUND,
1.1 paf 42: LS_METHOD_AFTER
43: };
1.68 paf 44:
45: struct Pos {
1.73 paf 46: int line;
47: int col;
48: Pos(int aline, int acol): line(aline), col(acol) {}
1.68 paf 49: Pos(): line(0), col(0) {}
50:
51: void clear() { line=col=0; }
52: operator bool() { return col!=0; }
53: };
54:
1.38 paf 55: /// compiler status
1.68 paf 56: class Parse_control {
57: const String* main_alias;
58: uint last_line_end_col;
59: public:
60: const String& alias_method(const String& name);
1.38 paf 61: //@{
62: /// @name input
1.68 paf 63: Request& request;
64: VStateless_class* cclass;
1.79 misha 65: VStateless_class* cclass_new;
1.78 misha 66: ArrayClass* cclasses;
1.68 paf 67: const char* source;
68: uint file_no;
69: Pos pos;
1.38 paf 70: //@}
71: //@{
72: /// @name state; initially
1.45 parser 73: bool trim_bof;
1.38 paf 74: int pending_state; ///< i=0
1.69 paf 75: String::Body string; ///< lexical string accumulator
1.68 paf 76: Pos string_start;
1.1 paf 77:
78: #define MAX_LEXICAL_STATES 100
1.38 paf 79: enum lexical_state ls; ///< =LS_USER;
1.54 paf 80: int ls_sp; ///< =0
81: enum lexical_state ls_stack[MAX_LEXICAL_STATES];
1.38 paf 82: int brackets_nestages[MAX_LEXICAL_STATES]; ///< brackets nestage on each state
1.46 parser 83:
1.56 paf 84: bool in_call_value;
1.74 paf 85: bool explicit_result;
1.79 misha 86: bool append;
1.38 paf 87: //@}
1.1 paf 88:
1.36 paf 89: /// output: filled input 'methods' and 'error' if any
1.116 moko 90: const char *error;
1.68 paf 91:
92: Parse_control(Request& arequest,
93: VStateless_class* aclass,
94: const char* asource, const String* amain_alias,
1.73 paf 95: uint afile_no,
96: int line_no_offset):
1.70 paf 97: main_alias(amain_alias),
98: last_line_end_col(0),
99:
1.68 paf 100: request(arequest), // input
101:
102: // we were told the class to compile to?
103: cclass(aclass), // until changed with @CLASS would consider operators loading
1.79 misha 104: cclass_new(0),
1.78 misha 105: cclasses(new ArrayClass(1)),
1.70 paf 106: source(asource),
1.68 paf 107: file_no(afile_no),
1.73 paf 108: pos(line_no_offset, 0),
1.68 paf 109:
110: // initialize state
111: trim_bof(true),
112: pending_state(0),
113: ls(LS_USER),
114: ls_sp(0),
1.74 paf 115: in_call_value(false),
1.79 misha 116: explicit_result(false),
1.116 moko 117: append(false),
118: error("") {
1.78 misha 119:
120: *cclasses+=aclass;
121: }
1.68 paf 122:
1.118 moko 123: /// false if exception should be rised
1.104 moko 124: bool class_add(){
1.79 misha 125: if(cclass_new){
126: cclass=cclass_new;
127: *cclasses+=cclass;
128: cclass_new=0;
129: append=false;
1.104 moko 130: // append to request's classes
1.119 ! moko 131: return request.add_class(cclass->type(), cclass);
1.79 misha 132: }
1.118 moko 133: return true;
1.79 misha 134: }
135:
1.81 misha 136: VStateless_class* get_existed_class(VStateless_class* aclass){
1.101 misha 137: // checking existence of the class during processing @OPTIONS\npartial
1.112 moko 138: // can't use get_class because it will call @autouse[] if the class wasn't loaded
1.99 misha 139: if(aclass)
1.112 moko 140: return request.classes().get(aclass->type());
1.81 misha 141: return 0;
142: }
143:
144: bool reuse_existed_class(VStateless_class* aclass){
145: if(aclass->is_partial()){
146: cclass=aclass;
147: cclass_new=0;
148: append=true;
149: return true;
150: } else {
151: return false;
152: }
1.79 misha 153: }
154:
155: void set_all_vars_local(){
1.102 misha 156: (cclass_new ? cclass_new : cclass)->set_all_vars_local();
157: }
158:
159: void set_methods_call_type(Method::Call_type call_type){
160: (cclass_new ? cclass_new : cclass)->set_methods_call_type(call_type);
161: }
162:
163: Method::Call_type get_methods_call_type(){
164: return (cclass_new ? cclass_new : cclass)->get_methods_call_type();
1.79 misha 165: }
166:
1.68 paf 167: void pos_next_line() {
168: pos.line++;
169: last_line_end_col=pos.col;
170: pos.col=0;
171: }
1.77 paf 172: void pos_next_c(int c) {
1.68 paf 173: if(c=='\t')
174: pos.col=(pos.col+TAB_SIZE)&~(TAB_SIZE-1);
175: else
176: pos.col++;
177: }
178: /// not precise in case of \t in the middle of the text
179: void pos_prev_c() {
180: if(pos.col==0) {
181: --pos.line; pos.col=last_line_end_col;
182: } else
183: --pos.col;
184: }
185: void ungetc() {
186: source--;
187: pos_prev_c();
188: }
1.1 paf 189: };
190:
1.36 paf 191: /// New array // return empty array
1.68 paf 192: inline ArrayOperation* N() {
193: return new ArrayOperation;
1.3 paf 194: }
195:
1.36 paf 196: /// Assembler instruction // append ordinary instruction to ops
1.80 misha 197: inline void O(ArrayOperation& result, OP::OPCODE code) {
1.68 paf 198: result+=Operation(code);
1.3 paf 199: }
200:
1.36 paf 201: /// aPpend 'code_array' to 'result'
1.68 paf 202: inline void P(ArrayOperation& result, ArrayOperation& code_array) {
203: result.append(code_array);
1.11 paf 204: }
1.36 paf 205: /// aPpend part of 'code_array', starting from offset, to 'result'
1.68 paf 206: inline void P(ArrayOperation& result, ArrayOperation& code_array, int offset) {
207: result.append(code_array, offset);
1.3 paf 208: }
1.53 paf 209:
1.84 misha 210: /// aPpend part of 'code_array', starting from offset, to 'result'
211: inline void P(ArrayOperation& result, ArrayOperation& code_array, int offset, int limit) {
212: result.append(code_array, offset, limit);
213: }
214:
1.68 paf 215: /// append cOde Array
1.97 misha 216: inline void OA(ArrayOperation& result, ArrayOperation* code_array) {
217: result+=Operation(code_array); // append 'code_array'
218: }
219:
1.80 misha 220: inline void OA(ArrayOperation& result, OP::OPCODE code, ArrayOperation* code_array) {
1.68 paf 221: result+=Operation(code); // append OP_CODE
222: result+=Operation(code_array); // append 'code_array'
1.61 paf 223: }
1.1 paf 224:
1.36 paf 225: /**
226: Value Literal // returns array with
1.68 paf 227: - first op: OP_VALUE instruction
228: - second op: origin (debug information)
229: - third op: string itself
1.36 paf 230: */
1.68 paf 231: inline ArrayOperation* VL(Value* value, uint file_no, uint line, uint col) {
1.61 paf 232: // empty ops array
1.68 paf 233: ArrayOperation& result=*N();
1.61 paf 234:
235: // append 'value' to 'result'
1.80 misha 236: result+=Operation(OP::OP_VALUE);
1.68 paf 237: result+=Operation(file_no, line, col); // append origin
238: result+=Operation(value); // append 'value'
239:
240: return &result;
1.61 paf 241: }
242:
1.68 paf 243: /// Literal Array to(2) Value @return Value from literal Array OP+origin+Value
1.85 misha 244: Value* LA2V(ArrayOperation& literal_string_array, int offset=0, OP::OPCODE code=OP::OP_VALUE);
1.102 misha 245:
1.68 paf 246: /// Literal Array to(2) String @return String value from literal Array OP+origin+String array
1.85 misha 247: inline const String* LA2S(ArrayOperation& literal_string_array, int offset=0, OP::OPCODE code=OP::OP_VALUE) {
248: if(Value* value=LA2V(literal_string_array, offset, code))
1.37 paf 249: return value->get_string();
250: return 0;
1.36 paf 251: }
1.61 paf 252:
1.68 paf 253: inline void change_string_literal_to_write_string_literal(ArrayOperation& literal_string_array) {
1.80 misha 254: literal_string_array.put(0, OP::OP_STRING__WRITE);
1.61 paf 255: }
256:
1.89 misha 257: void maybe_change_string_literal_to_double_literal(ArrayOperation& literal_string_array);
258:
259: void change_string_literal_value(ArrayOperation& literal_string_array, const String& new_value);
260:
1.106 misha 261: inline bool change(ArrayOperation& opcodes, int pos, OP::OPCODE find, OP::OPCODE replace) {
262: if(pos>=0) {
263: Operation& op=opcodes.get_ref(pos);
264: if(op.code==find) {
265: op.code=replace;
266: return true;
267: }
268: }
269: return false;
270: }
1.89 misha 271:
1.106 misha 272: inline void change_or_append(ArrayOperation& opcodes, int pos, OP::OPCODE find, OP::OPCODE replace, OP::OPCODE notfound) {
273: if(change(opcodes, pos, find, replace))
274: return;
1.86 misha 275:
1.106 misha 276: opcodes+=Operation(notfound);
1.108 moko 277: }
1.106 misha 278:
279: bool change_first(ArrayOperation& opcodes, OP::OPCODE find, OP::OPCODE replace);
1.86 misha 280:
1.92 misha 281: #ifdef OPTIMIZE_BYTECODE_GET_OBJECT_ELEMENT
1.93 misha 282: // OP_VALUE+origin+value+OP_GET_ELEMENT+OP_VALUE+origin+value+OP_GET_ELEMENT => OP_GET_OBJECT_ELEMENT+origin+value+origin+value
1.114 moko 283: inline bool maybe_make_get_object_element(ArrayOperation& opcodes, ArrayOperation& diving_code, size_t diving_count){
1.89 misha 284: if(
1.114 moko 285: diving_count>=8
286: && diving_code[0].code==OP::OP_VALUE
287: && diving_code[3].code==OP::OP_GET_ELEMENT
1.93 misha 288: && diving_code[4].code==OP::OP_VALUE
1.95 misha 289: && diving_code[7].code==OP::OP_GET_ELEMENT
1.89 misha 290: ){
291: O(opcodes, OP::OP_GET_OBJECT_ELEMENT);
1.114 moko 292: P(opcodes, diving_code, 1 /*offset*/, 2 /*limit*/); // copy first origin+value
1.93 misha 293: P(opcodes, diving_code, 5, 2); // second origin+value
1.114 moko 294: if(diving_count>8)
295: P(opcodes, diving_code, 8 /*offset*/); // tail
1.89 misha 296: return true;
297: }
298: return false;
299: }
1.92 misha 300: #endif
1.88 misha 301:
1.94 misha 302:
1.92 misha 303: #ifdef OPTIMIZE_BYTECODE_GET_OBJECT_VAR_ELEMENT
1.93 misha 304: // OP_VALUE+origin+value+OP_GET_ELEMENT+OP_WITH_READ+OP_VALUE+origin+value+OP_GET_ELEMENT+OP_GET_ELEMENT => OP_GET_OBJECT_VAR_ELEMENT+origin+value+origin+value
1.114 moko 305: inline bool maybe_make_get_object_var_element(ArrayOperation& opcodes, ArrayOperation& diving_code, size_t diving_count){
1.89 misha 306: if(
1.114 moko 307: diving_count==10
308: && diving_code[0].code==OP::OP_VALUE
309: && diving_code[3].code==OP::OP_GET_ELEMENT
1.95 misha 310: && diving_code[4].code==OP::OP_WITH_READ
1.89 misha 311: && diving_code[5].code==OP::OP_VALUE
1.95 misha 312: && diving_code[8].code==OP::OP_GET_ELEMENT
313: && diving_code[9].code==OP::OP_GET_ELEMENT
1.89 misha 314: ){
315: O(opcodes, OP::OP_GET_OBJECT_VAR_ELEMENT);
1.114 moko 316: P(opcodes, diving_code, 1 /*offset*/, 2 /*limit*/); // copy first origin+value
1.93 misha 317: P(opcodes, diving_code, 6, 2); // second origin+value
1.89 misha 318: return true;
319: }
320: return false;
321: }
1.92 misha 322: #endif
1.68 paf 323:
1.94 misha 324:
1.114 moko 325: bool maybe_make_self(ArrayOperation& opcodes, ArrayOperation& diving_code, size_t diving_count);
1.94 misha 326:
1.106 misha 327: #ifdef OPTIMIZE_BYTECODE_GET_ELEMENT__SPECIAL
328: bool maybe_append_simple_diving_code(ArrayOperation& code, ArrayOperation& diving_code);
329:
330: bool is_special_element(ArrayOperation& opcodes);
331: #endif
1.94 misha 332:
1.91 misha 333: #ifdef OPTIMIZE_BYTECODE_CONSTRUCT
1.93 misha 334: inline bool maybe_optimize_construct(ArrayOperation& opcodes, ArrayOperation& var_ops, ArrayOperation& expr_ops){
335: size_t expr_count=expr_ops.count();
336: OP::OPCODE construct_op=expr_ops[expr_count-1].code;
337: size_t construct=(construct_op==OP::OP_CONSTRUCT_VALUE)?0x01:(construct_op==OP::OP_CONSTRUCT_EXPR)?0x02:0x00;
338: if(construct){
339: P(opcodes, expr_ops, 0/*offset*/, expr_count-1/*limit*/); // copy constructor body without CONSTRUCT_(VALUE|EXPR)
340:
1.94 misha 341: size_t with=0x00;
342: switch(var_ops[0].code){
343: case OP::OP_WITH_ROOT:
344: {
345: with=0x10;
346: break;
347: }
348: case OP::OP_WITH_WRITE:
349: {
350: with=0x20;
351: break;
352: }
353: case OP::OP_WITH_SELF:
354: {
355: with=0x30;
356: break;
357: }
1.107 moko 358: default: break;
1.94 misha 359: }
1.93 misha 360:
361: if(with && var_ops[1].code==OP::OP_VALUE && var_ops.count()==4){
362: OP::OPCODE code=OP::OP_VALUE; // calm down compiler. will be reassigned for sure.
363: switch( with | construct ) {
364: case 0x11:
1.92 misha 365: {
1.93 misha 366: code=OP::OP_WITH_ROOT__VALUE__CONSTRUCT_VALUE;
1.92 misha 367: break;
368: }
1.93 misha 369: case 0x12:
1.92 misha 370: {
1.93 misha 371: code=OP::OP_WITH_ROOT__VALUE__CONSTRUCT_EXPR;
1.92 misha 372: break;
373: }
1.93 misha 374: case 0x21:
1.92 misha 375: {
1.93 misha 376: code=OP::OP_WITH_WRITE__VALUE__CONSTRUCT_VALUE;
1.92 misha 377: break;
378: }
1.93 misha 379: case 0x22:
1.92 misha 380: {
1.93 misha 381: code=OP::OP_WITH_WRITE__VALUE__CONSTRUCT_EXPR;
1.92 misha 382: break;
383: }
1.94 misha 384: case 0x31:
385: {
386: code=OP::OP_WITH_SELF__VALUE__CONSTRUCT_VALUE;
387: break;
388: }
389: case 0x32:
390: {
391: code=OP::OP_WITH_SELF__VALUE__CONSTRUCT_EXPR;
392: break;
393: }
1.93 misha 394: }
395: O(opcodes, code);
396: P(opcodes, var_ops, 2/*offset*/, 2/*limit*/); // copy origin+value
397: } else {
398: P(opcodes, var_ops);
399: O(opcodes, construct_op);
1.89 misha 400: }
1.93 misha 401: return true;
1.89 misha 402: }
403: return false;
404: }
1.92 misha 405: #endif
1.61 paf 406:
1.102 misha 407: Method::Call_type GetMethodCallType(Parse_control& pc, ArrayOperation& literal_string_array);
1.1 paf 408:
1.68 paf 409: void push_LS(Parse_control& pc, lexical_state new_state);
410: void pop_LS(Parse_control& pc);
1.1 paf 411:
412: #endif
E-mail: