Annotation of parser3/src/main/compile_tools.h, revision 1.95
1.32 paf 1: /** @file
1.33 paf 2: Parser: compiler support helper functions decls.
3:
1.82 misha 4: Copyright (c) 2001-2009 ArtLebedev Group (http://www.artlebedev.com)
1.58 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.1 paf 6: */
7:
8: #ifndef COMPILE_TOOLS
9: #define COMPILE_TOOLS
1.62 paf 10:
1.95 ! misha 11: static const char * const IDENT_COMPILE_TOOLS_H="$Date: 2009-06-04 09:31:38 $";
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
1.5 paf 90: char error[MAX_STRING];
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),
1.83 misha 117: append(false) {
1.78 misha 118:
119: *cclasses+=aclass;
120: }
1.68 paf 121:
1.79 misha 122: void class_add(){
123: if(cclass_new){
124: cclass=cclass_new;
125: // append to request's classes
126: request.classes().put(cclass->name(), cclass);
127: *cclasses+=cclass;
128: cclass_new=0;
129: append=false;
130: }
131: }
132:
1.81 misha 133: VStateless_class* get_existed_class(VStateless_class* aclass){
134: if(aclass){
135: if(Value* class_value=request.classes().get(aclass->name())){
136: return class_value->get_class();
1.79 misha 137: }
138: }
1.81 misha 139: return 0;
140: }
141:
142: bool reuse_existed_class(VStateless_class* aclass){
143: if(aclass->is_partial()){
144: cclass=aclass;
145: cclass_new=0;
146: append=true;
147: return true;
148: } else {
149: return false;
150: }
1.79 misha 151: }
152:
153: void set_all_vars_local(){
154: if(cclass_new){
1.81 misha 155: cclass_new->set_all_vars_local();
1.79 misha 156: } else {
1.81 misha 157: cclass->set_all_vars_local();
1.79 misha 158: }
159: }
160:
1.68 paf 161: void pos_next_line() {
162: pos.line++;
163: last_line_end_col=pos.col;
164: pos.col=0;
165: }
1.77 paf 166: void pos_next_c(int c) {
1.68 paf 167: if(c=='\t')
168: pos.col=(pos.col+TAB_SIZE)&~(TAB_SIZE-1);
169: else
170: pos.col++;
171: }
172: /// not precise in case of \t in the middle of the text
173: void pos_prev_c() {
174: if(pos.col==0) {
175: --pos.line; pos.col=last_line_end_col;
176: } else
177: --pos.col;
178: }
179: void ungetc() {
180: source--;
181: pos_prev_c();
182: }
1.1 paf 183: };
184:
1.36 paf 185: /// New array // return empty array
1.68 paf 186: inline ArrayOperation* N() {
187: return new ArrayOperation;
1.3 paf 188: }
189:
1.36 paf 190: /// Assembler instruction // append ordinary instruction to ops
1.80 misha 191: inline void O(ArrayOperation& result, OP::OPCODE code) {
1.68 paf 192: result+=Operation(code);
1.3 paf 193: }
194:
1.36 paf 195: /// aPpend 'code_array' to 'result'
1.68 paf 196: inline void P(ArrayOperation& result, ArrayOperation& code_array) {
197: result.append(code_array);
1.11 paf 198: }
1.36 paf 199: /// aPpend part of 'code_array', starting from offset, to 'result'
1.68 paf 200: inline void P(ArrayOperation& result, ArrayOperation& code_array, int offset) {
201: result.append(code_array, offset);
1.3 paf 202: }
1.53 paf 203:
1.84 misha 204: /// aPpend part of 'code_array', starting from offset, to 'result'
205: inline void P(ArrayOperation& result, ArrayOperation& code_array, int offset, int limit) {
206: result.append(code_array, offset, limit);
207: }
208:
1.68 paf 209: /// append cOde Array
1.80 misha 210: inline void OA(ArrayOperation& result, OP::OPCODE code, ArrayOperation* code_array) {
1.68 paf 211: result+=Operation(code); // append OP_CODE
212: result+=Operation(code_array); // append 'code_array'
1.61 paf 213: }
1.1 paf 214:
1.36 paf 215: /**
216: Value Literal // returns array with
1.68 paf 217: - first op: OP_VALUE instruction
218: - second op: origin (debug information)
219: - third op: string itself
1.36 paf 220: */
1.68 paf 221: inline ArrayOperation* VL(Value* value, uint file_no, uint line, uint col) {
1.61 paf 222: // empty ops array
1.68 paf 223: ArrayOperation& result=*N();
1.61 paf 224:
225: // append 'value' to 'result'
1.80 misha 226: result+=Operation(OP::OP_VALUE);
1.68 paf 227: result+=Operation(file_no, line, col); // append origin
228: result+=Operation(value); // append 'value'
229:
230: return &result;
1.61 paf 231: }
232:
1.68 paf 233: /// Literal Array to(2) Value @return Value from literal Array OP+origin+Value
1.85 misha 234: Value* LA2V(ArrayOperation& literal_string_array, int offset=0, OP::OPCODE code=OP::OP_VALUE);
1.68 paf 235: /// Literal Array to(2) String @return String value from literal Array OP+origin+String array
1.85 misha 236: inline const String* LA2S(ArrayOperation& literal_string_array, int offset=0, OP::OPCODE code=OP::OP_VALUE) {
237: if(Value* value=LA2V(literal_string_array, offset, code))
1.37 paf 238: return value->get_string();
239: return 0;
1.36 paf 240: }
1.61 paf 241:
1.68 paf 242: inline void change_string_literal_to_write_string_literal(ArrayOperation& literal_string_array) {
1.80 misha 243: literal_string_array.put(0, OP::OP_STRING__WRITE);
1.61 paf 244: }
245:
1.89 misha 246: void maybe_change_string_literal_to_double_literal(ArrayOperation& literal_string_array);
247:
248: void change_string_literal_value(ArrayOperation& literal_string_array, const String& new_value);
249:
250: void changetail_or_append(ArrayOperation& opcodes,
251: OP::OPCODE find, bool with_argument, OP::OPCODE replace, OP::OPCODE notfound);
252:
1.94 misha 253: bool maybe_change_first_opcode(ArrayOperation& opcodes, OP::OPCODE find, OP::OPCODE replace);
1.86 misha 254:
255:
1.92 misha 256: #ifdef OPTIMIZE_BYTECODE_GET_OBJECT_ELEMENT
1.93 misha 257: // OP_VALUE+origin+value+OP_GET_ELEMENT+OP_VALUE+origin+value+OP_GET_ELEMENT => OP_GET_OBJECT_ELEMENT+origin+value+origin+value
1.89 misha 258: inline bool maybe_make_get_object_element(ArrayOperation& opcodes, ArrayOperation& diving_code, size_t divine_count){
1.95 ! misha 259: if(divine_count<8)
1.89 misha 260: return false;
261:
262: assert(diving_code[0].code==OP::OP_VALUE);
263: if(
1.94 misha 264: diving_code[3].code==OP::OP_GET_ELEMENT
1.93 misha 265: && diving_code[4].code==OP::OP_VALUE
1.95 ! misha 266: && diving_code[7].code==OP::OP_GET_ELEMENT
1.89 misha 267: ){
268: O(opcodes, OP::OP_GET_OBJECT_ELEMENT);
1.94 misha 269: P(opcodes, diving_code, 1/*offset*/, 2/*limit*/); // copy first origin+value
1.93 misha 270: P(opcodes, diving_code, 5, 2); // second origin+value
1.95 ! misha 271: if(divine_count>8)
! 272: P(opcodes, diving_code, 8/*offset*/); // tail
1.89 misha 273: return true;
274: }
275: return false;
276: }
1.92 misha 277: #endif
1.88 misha 278:
1.94 misha 279:
1.92 misha 280: #ifdef OPTIMIZE_BYTECODE_GET_OBJECT_VAR_ELEMENT
1.93 misha 281: // 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
1.89 misha 282: inline bool maybe_make_get_object_var_element(ArrayOperation& opcodes, ArrayOperation& diving_code, size_t divine_count){
1.95 ! misha 283: if(divine_count<10)
1.89 misha 284: return false;
285:
286: assert(diving_code[0].code==OP::OP_VALUE);
287: if(
1.95 ! misha 288: diving_code[3].code==OP::OP_GET_ELEMENT
! 289: && diving_code[4].code==OP::OP_WITH_READ
1.89 misha 290: && diving_code[5].code==OP::OP_VALUE
1.95 ! misha 291: && diving_code[8].code==OP::OP_GET_ELEMENT
! 292: && diving_code[9].code==OP::OP_GET_ELEMENT
1.89 misha 293: ){
294: O(opcodes, OP::OP_GET_OBJECT_VAR_ELEMENT);
1.94 misha 295: P(opcodes, diving_code, 1/*offset*/, 2/*limit*/); // copy first origin+value
1.93 misha 296: P(opcodes, diving_code, 6, 2); // second origin+value
1.95 ! misha 297: if(divine_count>10)
! 298: P(opcodes, diving_code, 10/*offset*/); // tail
1.89 misha 299: return true;
300: }
301: return false;
302: }
1.92 misha 303: #endif
1.68 paf 304:
1.94 misha 305:
306: // OP_VALUE+origin+self+OP_GET_ELEMENT+OP_VALUE+origin+value+OP_GET_ELEMENT => OP_WITH_SELF__VALUE__GET_ELEMENT+origin+value
307: #ifdef OPTIMIZE_BYTECODE_GET_SELF_ELEMENT
308: inline bool maybe_make_with_self_get_element(ArrayOperation& opcodes, ArrayOperation& diving_code, size_t divine_count){
1.95 ! misha 309: if(divine_count<8)
1.94 misha 310: return false;
311:
312: assert(diving_code[0].code==OP::OP_VALUE);
313: if(
314: diving_code[3].code==OP::OP_GET_ELEMENT
315: && diving_code[4].code==OP::OP_VALUE
1.95 ! misha 316: && diving_code[7].code==OP::OP_GET_ELEMENT
1.94 misha 317: ){
318: O(opcodes, OP::OP_WITH_SELF__VALUE__GET_ELEMENT);
319: P(opcodes, diving_code, 5/*offset*/, 2/*limit*/); // copy second origin+value. we know that the first one is "self"
1.95 ! misha 320: if(divine_count>8)
! 321: P(opcodes, diving_code, 8/*offset*/); // tail
1.94 misha 322: return true;
323: }
324: return false;
325: }
326: #endif
327:
328:
1.91 misha 329: #ifdef OPTIMIZE_BYTECODE_CONSTRUCT
1.93 misha 330: inline bool maybe_optimize_construct(ArrayOperation& opcodes, ArrayOperation& var_ops, ArrayOperation& expr_ops){
331: size_t expr_count=expr_ops.count();
332: OP::OPCODE construct_op=expr_ops[expr_count-1].code;
333: size_t construct=(construct_op==OP::OP_CONSTRUCT_VALUE)?0x01:(construct_op==OP::OP_CONSTRUCT_EXPR)?0x02:0x00;
334: if(construct){
335: P(opcodes, expr_ops, 0/*offset*/, expr_count-1/*limit*/); // copy constructor body without CONSTRUCT_(VALUE|EXPR)
336:
1.94 misha 337: size_t with=0x00;
338: switch(var_ops[0].code){
339: case OP::OP_WITH_ROOT:
340: {
341: with=0x10;
342: break;
343: }
344: case OP::OP_WITH_WRITE:
345: {
346: with=0x20;
347: break;
348: }
349: case OP::OP_WITH_SELF:
350: {
351: with=0x30;
352: break;
353: }
354: }
1.93 misha 355:
356: if(with && var_ops[1].code==OP::OP_VALUE && var_ops.count()==4){
357: OP::OPCODE code=OP::OP_VALUE; // calm down compiler. will be reassigned for sure.
358: switch( with | construct ) {
359: case 0x11:
1.92 misha 360: {
1.93 misha 361: code=OP::OP_WITH_ROOT__VALUE__CONSTRUCT_VALUE;
1.92 misha 362: break;
363: }
1.93 misha 364: case 0x12:
1.92 misha 365: {
1.93 misha 366: code=OP::OP_WITH_ROOT__VALUE__CONSTRUCT_EXPR;
1.92 misha 367: break;
368: }
1.93 misha 369: case 0x21:
1.92 misha 370: {
1.93 misha 371: code=OP::OP_WITH_WRITE__VALUE__CONSTRUCT_VALUE;
1.92 misha 372: break;
373: }
1.93 misha 374: case 0x22:
1.92 misha 375: {
1.93 misha 376: code=OP::OP_WITH_WRITE__VALUE__CONSTRUCT_EXPR;
1.92 misha 377: break;
378: }
1.94 misha 379: case 0x31:
380: {
381: code=OP::OP_WITH_SELF__VALUE__CONSTRUCT_VALUE;
382: break;
383: }
384: case 0x32:
385: {
386: code=OP::OP_WITH_SELF__VALUE__CONSTRUCT_EXPR;
387: break;
388: }
1.93 misha 389: }
390: O(opcodes, code);
391: P(opcodes, var_ops, 2/*offset*/, 2/*limit*/); // copy origin+value
392: } else {
393: P(opcodes, var_ops);
394: O(opcodes, construct_op);
1.89 misha 395: }
1.93 misha 396: return true;
1.89 misha 397: }
398: return false;
399: }
1.92 misha 400: #endif
1.61 paf 401:
1.1 paf 402:
1.68 paf 403: void push_LS(Parse_control& pc, lexical_state new_state);
404: void pop_LS(Parse_control& pc);
1.1 paf 405:
406: #endif
E-mail: