pikapython/src/PikaParser.c

451 lines
14 KiB
C
Raw Normal View History

2021-10-13 18:05:54 +08:00
#include "PikaParser.h"
2021-10-16 19:52:19 +08:00
#include "BaseObj.h"
2021-10-13 18:05:54 +08:00
#include "PikaObj.h"
2021-10-16 17:32:02 +08:00
#include "dataQueue.h"
2021-10-13 18:05:54 +08:00
#include "dataQueueObj.h"
2021-10-16 18:59:46 +08:00
#include "dataStack.h"
2021-10-13 18:05:54 +08:00
#include "dataStrs.h"
char* strsPopTokenWithSkip(Args* buffs,
char* stmts,
char sign,
char skipStart,
char skipEnd) {
2021-10-13 20:07:32 +08:00
int32_t size = strGetSize(stmts);
if (0 == size) {
return NULL;
}
char* strOut = args_getBuff(buffs, size);
int32_t iOut = 0;
int32_t stmtEnd = 0;
uint8_t isGetSign = 0;
int32_t parentheseDeepth = 0;
for (int32_t i = 0; i < size; i++) {
if (skipStart == stmts[i]) {
2021-10-13 20:07:32 +08:00
parentheseDeepth++;
}
if (skipEnd == stmts[i]) {
2021-10-13 20:07:32 +08:00
parentheseDeepth--;
}
if (parentheseDeepth == 0) {
if (sign == stmts[i]) {
2021-10-13 20:07:32 +08:00
stmtEnd = i;
isGetSign = 1;
break;
}
}
}
if (!isGetSign) {
stmtEnd = size;
}
for (int32_t i = 0; i < stmtEnd; i++) {
strOut[i] = stmts[i];
}
memmove(stmts, stmts + stmtEnd + 1, size);
strOut[stmtEnd] = 0;
return strOut;
}
2021-10-20 20:40:45 +08:00
enum StmtType {
REF,
STR,
NUM,
METHOD,
2021-10-20 21:06:13 +08:00
OPERATOR,
2021-10-20 20:40:45 +08:00
NONE,
};
2021-10-20 21:06:13 +08:00
char* strs_deleteBetween(Args* buffs, char* strIn, char begin, char end) {
int32_t size = strGetSize(strIn);
char* strOut = args_getBuff(buffs, size);
2021-10-20 21:21:03 +08:00
uint8_t deepth = 0;
2021-10-20 21:06:13 +08:00
uint32_t iOut = 0;
for (int i = 0; i < size; i++) {
2021-10-20 21:21:03 +08:00
if (end == strIn[i]) {
deepth--;
}
if (0 == deepth) {
2021-10-20 21:06:13 +08:00
strOut[iOut] = strIn[i];
2021-10-20 21:21:03 +08:00
iOut++;
2021-10-20 21:06:13 +08:00
}
if (begin == strIn[i]) {
2021-10-20 21:21:03 +08:00
deepth++;
2021-10-20 21:06:13 +08:00
}
}
strOut[iOut] = 0;
return strOut;
}
2021-10-25 21:53:35 +08:00
uint8_t checkIsEqu(char* str) {
uint32_t size = strGetSize(str) + 1;
for (int i = 0; i + 1 < size; i++) {
if (str[i] == '=' && str[i + 1] == '=') {
return 1;
}
}
return 0;
}
2021-10-20 20:40:45 +08:00
static enum StmtType matchStmtType(char* right) {
2021-10-20 21:06:13 +08:00
Args* buffs = New_strBuff();
enum StmtType stmtType = NONE;
char* rightWithoutSubStmt = strs_deleteBetween(buffs, right, '(', ')');
2021-10-20 21:21:03 +08:00
if (strIsContain(rightWithoutSubStmt, '+') ||
strIsContain(rightWithoutSubStmt, '-') ||
strIsContain(rightWithoutSubStmt, '*') ||
2021-10-20 22:00:53 +08:00
strIsContain(rightWithoutSubStmt, '<') ||
strIsContain(rightWithoutSubStmt, '>') ||
2021-10-25 21:53:35 +08:00
checkIsEqu(rightWithoutSubStmt) ||
2021-10-20 21:21:03 +08:00
strIsContain(rightWithoutSubStmt, '/')) {
2021-10-20 21:06:13 +08:00
stmtType = OPERATOR;
goto exit;
}
2021-10-20 21:21:03 +08:00
if (strIsContain(rightWithoutSubStmt, '(') ||
strIsContain(rightWithoutSubStmt, ')')) {
2021-10-20 21:06:13 +08:00
stmtType = METHOD;
goto exit;
2021-10-20 20:40:45 +08:00
}
2021-10-20 21:21:03 +08:00
if (strIsContain(rightWithoutSubStmt, '\'') ||
strIsContain(rightWithoutSubStmt, '\"')) {
2021-10-20 21:06:13 +08:00
stmtType = STR;
goto exit;
2021-10-20 20:40:45 +08:00
}
2021-10-20 21:21:03 +08:00
if (rightWithoutSubStmt[0] >= '0' && rightWithoutSubStmt[0] <= '9') {
2021-10-20 21:06:13 +08:00
stmtType = NUM;
goto exit;
2021-10-20 20:40:45 +08:00
}
2021-10-20 21:21:03 +08:00
if (!strEqu(rightWithoutSubStmt, "")) {
2021-10-20 21:06:13 +08:00
stmtType = REF;
goto exit;
2021-10-20 20:40:45 +08:00
}
2021-10-20 21:06:13 +08:00
exit:
args_deinit(buffs);
return stmtType;
2021-10-20 20:40:45 +08:00
}
2021-10-25 21:53:35 +08:00
uint8_t checkIsDirect(char* str) {
/* include '0' */
uint32_t size = strGetSize(str) + 1;
for (int i = 1; i + 1 < size; i++) {
if ((str[i - 1] != '=') && (str[i] == '=') && (str[i + 1] != '=')) {
return 1;
}
}
return 0;
}
2021-10-13 19:41:04 +08:00
AST* AST_parseStmt(AST* ast, char* stmt) {
2021-10-13 18:05:54 +08:00
Args* buffs = New_strBuff();
2021-10-13 18:24:38 +08:00
char* assignment = strsGetFirstToken(buffs, stmt, '(');
2021-10-13 18:05:54 +08:00
char* direct = NULL;
2021-10-13 18:24:38 +08:00
char* method = NULL;
2021-10-13 21:20:45 +08:00
char* ref = NULL;
2021-10-14 09:51:50 +08:00
char* str = NULL;
2021-10-14 15:08:45 +08:00
char* num = NULL;
2021-10-14 15:50:42 +08:00
char* right = NULL;
2021-10-20 20:40:45 +08:00
/* solve direct */
uint8_t directExist = 0;
2021-10-25 21:53:35 +08:00
if (checkIsDirect(assignment)) {
2021-10-13 21:20:45 +08:00
directExist = 1;
}
if (directExist) {
2021-10-13 18:05:54 +08:00
direct = strsGetFirstToken(buffs, assignment, '=');
2021-10-13 20:07:32 +08:00
obj_setStr(ast, (char*)"direct", direct);
2021-10-13 18:05:54 +08:00
}
2021-10-20 20:40:45 +08:00
/* solve right stmt */
if (directExist) {
right = strsGetLastToken(buffs, stmt, '=');
} else {
right = stmt;
}
enum StmtType stmtType = matchStmtType(right);
/* solve method stmt */
2021-10-20 21:33:16 +08:00
if (OPERATOR == stmtType) {
char* rightWithoutSubStmt = strs_deleteBetween(buffs, right, '(', ')');
char operator[2] = {0};
if (strIsContain(rightWithoutSubStmt, '*')) {
operator[0] = '*';
}
if (strIsContain(rightWithoutSubStmt, '/')) {
operator[0] = '/';
}
if (strIsContain(rightWithoutSubStmt, '+')) {
operator[0] = '+';
}
if (strIsContain(rightWithoutSubStmt, '-')) {
operator[0] = '-';
}
2021-10-20 22:00:53 +08:00
if (strIsContain(rightWithoutSubStmt, '<')) {
operator[0] = '<';
}
if (strIsContain(rightWithoutSubStmt, '>')) {
operator[0] = '>';
}
2021-10-25 21:53:35 +08:00
if (checkIsEqu(rightWithoutSubStmt)) {
operator[0] = '=';
operator[1] = '=';
2021-10-20 22:00:53 +08:00
}
2021-10-20 21:33:16 +08:00
obj_setStr(ast, (char*)"operator", operator);
2021-10-20 21:53:10 +08:00
char* rightBuff = strsCopy(buffs, right);
char* subStmt1 =
strsPopTokenWithSkip(buffs, rightBuff, operator[0], '(', ')');
2021-10-20 21:53:10 +08:00
char* subStmt2 = rightBuff;
2021-10-25 21:53:35 +08:00
if (operator[1] == '=') {
subStmt2 = rightBuff + 1;
}
2021-10-20 21:33:16 +08:00
queueObj_pushObj(ast, (char*)"stmt");
AST_parseStmt(queueObj_getCurrentObj(ast), subStmt1);
queueObj_pushObj(ast, (char*)"stmt");
AST_parseStmt(queueObj_getCurrentObj(ast), subStmt2);
goto exit;
}
2021-10-20 20:40:45 +08:00
if (METHOD == stmtType) {
method = strsGetFirstToken(buffs, right, '(');
2021-10-13 21:20:45 +08:00
obj_setStr(ast, (char*)"method", method);
2021-10-20 20:40:45 +08:00
char* subStmts = strsCut(buffs, stmt, '(', ')');
2021-10-13 20:18:12 +08:00
while (1) {
char* subStmt =
strsPopTokenWithSkip(buffs, subStmts, ',', '(', ')');
2021-10-13 20:18:12 +08:00
if (NULL == subStmt) {
break;
}
queueObj_pushObj(ast, (char*)"stmt");
AST_parseStmt(queueObj_getCurrentObj(ast), subStmt);
2021-10-13 20:07:32 +08:00
}
2021-10-14 09:51:50 +08:00
goto exit;
}
2021-10-20 20:40:45 +08:00
/* solve reference stmt */
if (REF == stmtType) {
2021-10-14 15:50:42 +08:00
ref = right;
2021-10-13 21:20:45 +08:00
obj_setStr(ast, (char*)"ref", ref);
2021-10-14 09:51:50 +08:00
goto exit;
}
2021-10-20 20:40:45 +08:00
/* solve str stmt */
if (STR == stmtType) {
2021-10-14 15:50:42 +08:00
str = right;
2021-10-14 09:51:50 +08:00
str = strsDeleteChar(buffs, str, '\'');
str = strsDeleteChar(buffs, str, '\"');
obj_setStr(ast, (char*)"str", str);
goto exit;
2021-10-13 20:07:32 +08:00
}
2021-10-20 20:40:45 +08:00
/* solve number stmt */
if (NUM == stmtType) {
2021-10-14 15:50:42 +08:00
num = right;
2021-10-14 15:08:45 +08:00
obj_setStr(ast, (char*)"num", num);
goto exit;
}
2021-10-13 18:05:54 +08:00
exit:
args_deinit(buffs);
return ast;
}
2021-10-16 17:32:02 +08:00
static int32_t getPyLineBlockDeepth(char* line) {
uint32_t size = strGetSize(line);
2021-10-16 19:52:19 +08:00
for (int i = 0; i < size; i++) {
2021-10-16 17:32:02 +08:00
if (line[i] != ' ') {
uint32_t spaceNum = i;
if (0 == spaceNum % 4) {
return spaceNum / 4;
}
/* space Num is not 4N, error*/
return -1;
}
}
2021-10-19 09:13:10 +08:00
return 0;
2021-10-16 17:32:02 +08:00
}
2021-10-16 18:59:46 +08:00
AST* pikaParseLine(char* line, Stack* blockStack) {
2021-10-13 19:41:04 +08:00
AST* ast = New_queueObj();
Args* buffs = New_strBuff();
2021-10-19 22:26:34 +08:00
line = strsDeleteChar(buffs, line, '\r');
2021-10-16 17:48:01 +08:00
uint8_t blockDeepth = getPyLineBlockDeepth(line);
2021-10-16 19:52:19 +08:00
uint8_t blockDeepthLast = blockDeepth;
if (NULL != blockStack) {
blockDeepthLast = args_getInt(blockStack, "top");
for (int i = 0; i < blockDeepthLast - blockDeepth; i++) {
QueueObj* exitBlock = obj_getObj(ast, "exitBlock", 0);
if (NULL == exitBlock) {
obj_newObj(ast, "exitBlock", "", New_TinyObj);
2021-10-16 19:52:19 +08:00
exitBlock = obj_getObj(ast, "exitBlock", 0);
queueObj_init(exitBlock);
}
char buff[10] = {0};
char* blockType = stack_popStr(blockStack, buff);
queueObj_pushStr(exitBlock, blockType);
}
}
2021-10-16 17:48:01 +08:00
obj_setInt(ast, "blockDeepth", blockDeepth);
char* lineStart = line + blockDeepth * 4;
char* stmt = lineStart;
if (0 == strncmp(lineStart, (char*)"while ", 6)) {
stmt = strsCut(buffs, lineStart, ' ', ':');
2021-10-16 18:13:52 +08:00
obj_setStr(ast, "contralFlow", "while");
2021-10-16 19:09:19 +08:00
if (NULL != blockStack) {
stack_pushStr(blockStack, "while");
}
2021-10-16 17:48:01 +08:00
}
if (0 == strncmp(lineStart, (char*)"if ", 3)) {
stmt = strsCut(buffs, lineStart, ' ', ':');
2021-10-16 18:13:52 +08:00
obj_setStr(ast, "contralFlow", "if");
2021-10-16 19:09:19 +08:00
if (NULL != blockStack) {
stack_pushStr(blockStack, "if");
}
2021-10-16 17:48:01 +08:00
}
stmt = strsGetCleanCmd(buffs, stmt);
2021-10-13 19:41:04 +08:00
ast = AST_parseStmt(ast, stmt);
goto exit;
exit:
args_deinit(buffs);
return ast;
}
2021-10-13 22:02:53 +08:00
2021-10-16 19:09:19 +08:00
char* pikaParseLineToAsm(Args* buffs, char* line, Stack* blockStack) {
AST* ast = pikaParseLine(line, blockStack);
2021-10-14 10:44:59 +08:00
char* pikaAsm = AST_toPikaAsm(ast, buffs);
AST_deinit(ast);
return pikaAsm;
}
2021-10-27 15:57:49 +08:00
static Arg* saveSingleAsm(Args* buffs, Arg* pikaAsmBuff, char* singleAsm) {
2021-10-27 15:56:25 +08:00
char* pikaAsm = arg_getStr(pikaAsmBuff);
pikaAsm = strsAppend(buffs, pikaAsm, singleAsm);
arg_deinit(pikaAsmBuff);
pikaAsmBuff = arg_setStr(NULL, "", pikaAsm);
return pikaAsmBuff;
}
static char* getOutAsm(Args* outBuffs, Arg* pikaAsmBuff) {
return strsCopy(outBuffs, arg_getStr(pikaAsmBuff));
}
2021-10-17 19:07:33 +08:00
char* pikaParseMultiLineToAsm(Args* outBuffs, char* multiLine) {
Stack* blockStack = New_Stack();
2021-10-17 19:35:50 +08:00
Arg* pikaAsmBuff = arg_setStr(NULL, "", "");
2021-10-27 15:24:42 +08:00
uint32_t lineOffset = 0;
uint32_t multiLineSize = strGetSize(multiLine);
uint8_t isSave = 0;
2021-10-27 15:56:25 +08:00
if (strCountSign(multiLine, '\n') > 1) {
2021-10-27 15:24:42 +08:00
isSave = 1;
}
2021-10-17 19:07:33 +08:00
while (1) {
Args* singleRunBuffs = New_strBuff();
2021-10-27 15:56:25 +08:00
char* line =
strsGetFirstToken(singleRunBuffs, multiLine + lineOffset, '\n');
2021-10-27 15:24:42 +08:00
uint32_t lineSize = strGetSize(line);
lineOffset = lineOffset + lineSize + 1;
2021-10-17 19:07:33 +08:00
char* singleAsm = pikaParseLineToAsm(singleRunBuffs, line, blockStack);
2021-10-27 15:57:49 +08:00
pikaAsmBuff = saveSingleAsm(singleRunBuffs, pikaAsmBuff, singleAsm);
2021-10-17 19:07:33 +08:00
args_deinit(singleRunBuffs);
2021-10-27 15:44:35 +08:00
if (lineOffset >= multiLineSize) {
2021-10-17 19:07:33 +08:00
break;
}
}
2021-10-27 15:56:25 +08:00
char* outAsm = getOutAsm(outBuffs, pikaAsmBuff);
2021-10-17 19:35:50 +08:00
arg_deinit(pikaAsmBuff);
2021-10-17 19:07:33 +08:00
stack_deinit(blockStack);
2021-10-27 15:56:25 +08:00
return outAsm;
2021-10-17 19:07:33 +08:00
}
2021-10-13 22:17:40 +08:00
char* AST_appandPikaAsm(AST* ast, AST* subAst, Args* buffs, char* pikaAsm) {
2021-10-13 22:02:53 +08:00
uint32_t deepth = obj_getInt(ast, "deepth");
2021-10-13 21:20:45 +08:00
while (1) {
2021-10-13 22:02:53 +08:00
QueueObj* subStmt = queueObj_popObj(subAst);
2021-10-13 21:20:45 +08:00
if (NULL == subStmt) {
break;
}
2021-10-13 22:02:53 +08:00
obj_setInt(ast, "deepth", deepth + 1);
2021-10-13 22:17:40 +08:00
pikaAsm = AST_appandPikaAsm(ast, subStmt, buffs, pikaAsm);
2021-10-13 21:20:45 +08:00
}
2021-10-13 22:02:53 +08:00
char* method = obj_getStr(subAst, "method");
2021-10-20 21:42:32 +08:00
char* operator= obj_getStr(subAst, "operator");
2021-10-13 22:02:53 +08:00
char* ref = obj_getStr(subAst, "ref");
char* direct = obj_getStr(subAst, "direct");
2021-10-14 09:51:50 +08:00
char* str = obj_getStr(subAst, "str");
2021-10-14 15:08:45 +08:00
char* num = obj_getStr(subAst, "num");
2021-10-13 21:20:45 +08:00
if (NULL != ref) {
char buff[32] = {0};
2021-10-13 21:53:01 +08:00
sprintf(buff, "%d REF %s\n", deepth, ref);
2021-10-13 22:17:40 +08:00
pikaAsm = strsAppend(buffs, pikaAsm, buff);
2021-10-13 21:20:45 +08:00
}
2021-10-20 21:42:32 +08:00
if (NULL != operator) {
char buff[32] = {0};
sprintf(buff, "%d OPT %s\n", deepth, operator);
pikaAsm = strsAppend(buffs, pikaAsm, buff);
}
2021-10-13 21:20:45 +08:00
if (NULL != method) {
char buff[32] = {0};
2021-10-13 21:53:01 +08:00
sprintf(buff, "%d RUN %s\n", deepth, method);
2021-10-13 22:17:40 +08:00
pikaAsm = strsAppend(buffs, pikaAsm, buff);
2021-10-13 21:20:45 +08:00
}
2021-10-14 09:51:50 +08:00
if (NULL != str) {
char buff[32] = {0};
sprintf(buff, "%d STR %s\n", deepth, str);
pikaAsm = strsAppend(buffs, pikaAsm, buff);
}
2021-10-14 15:50:42 +08:00
if (NULL != num) {
2021-10-13 21:20:45 +08:00
char buff[32] = {0};
2021-10-14 15:50:42 +08:00
sprintf(buff, "%d NUM %s\n", deepth, num);
2021-10-13 22:17:40 +08:00
pikaAsm = strsAppend(buffs, pikaAsm, buff);
2021-10-13 21:20:45 +08:00
}
2021-10-14 15:50:42 +08:00
if (NULL != direct) {
2021-10-14 15:08:45 +08:00
char buff[32] = {0};
2021-10-14 15:50:42 +08:00
sprintf(buff, "%d OUT %s\n", deepth, direct);
2021-10-14 15:08:45 +08:00
pikaAsm = strsAppend(buffs, pikaAsm, buff);
}
2021-10-13 22:02:53 +08:00
obj_setInt(ast, "deepth", deepth - 1);
2021-10-13 22:17:40 +08:00
return pikaAsm;
2021-10-13 21:20:45 +08:00
}
2021-10-17 16:19:43 +08:00
static char* addBlockDeepth(AST* ast,
Args* buffs,
char* pikaAsm,
uint8_t deepthOffset) {
2021-10-16 17:32:02 +08:00
pikaAsm = strsAppend(buffs, pikaAsm, (char*)"B");
char buff[11];
2021-10-17 16:19:43 +08:00
pikaAsm = strsAppend(
buffs, pikaAsm,
fast_itoa(buff, obj_getInt(ast, "blockDeepth") + deepthOffset));
2021-10-16 17:32:02 +08:00
pikaAsm = strsAppend(buffs, pikaAsm, (char*)"\n");
2021-10-19 09:13:10 +08:00
return pikaAsm;
2021-10-16 19:52:19 +08:00
}
char* AST_toPikaAsm(AST* ast, Args* buffs) {
Args* runBuffs = New_strBuff();
char* pikaAsm = strsCopy(runBuffs, "");
QueueObj* exitBlock = obj_getObj(ast, "exitBlock", 0);
if (NULL != exitBlock) {
while (1) {
2021-10-17 16:19:43 +08:00
uint8_t blockTypeNum = obj_getInt(exitBlock, "top") -
obj_getInt(exitBlock, "bottom") - 1;
2021-10-16 19:52:19 +08:00
char* blockType = queueObj_popStr(exitBlock);
if (NULL == blockType) {
break;
}
2021-10-17 18:08:14 +08:00
/* goto the while start when exit while block */
2021-10-16 19:52:19 +08:00
if (strEqu(blockType, "while")) {
2021-10-17 18:08:14 +08:00
pikaAsm = addBlockDeepth(ast, buffs, pikaAsm, blockTypeNum);
2021-10-16 19:52:19 +08:00
pikaAsm = strsAppend(buffs, pikaAsm, (char*)"0 JMP -1\n");
}
}
}
2021-10-17 16:19:43 +08:00
pikaAsm = addBlockDeepth(ast, buffs, pikaAsm, 0);
2021-10-13 22:02:53 +08:00
obj_setInt(ast, "deepth", 0);
2021-10-20 21:42:32 +08:00
/* parse ast to asm main process */
2021-10-14 10:05:55 +08:00
pikaAsm = AST_appandPikaAsm(ast, ast, runBuffs, pikaAsm);
2021-10-20 21:42:32 +08:00
2021-10-16 18:13:52 +08:00
if (strEqu(obj_getStr(ast, "contralFlow"), "while")) {
2021-10-16 18:39:09 +08:00
pikaAsm = strsAppend(runBuffs, pikaAsm, "0 JEZ 2\n");
2021-10-16 18:13:52 +08:00
}
2021-10-17 18:08:14 +08:00
if (strEqu(obj_getStr(ast, "contralFlow"), "if")) {
pikaAsm = strsAppend(runBuffs, pikaAsm, "0 JEZ 1\n");
}
2021-10-14 10:05:55 +08:00
pikaAsm = strsCopy(buffs, pikaAsm);
args_deinit(runBuffs);
return pikaAsm;
2021-10-13 18:05:54 +08:00
}
int32_t AST_deinit(AST* ast) {
return obj_deinit(ast);
}