X-Git-Url: https://git.m6w6.name/?p=m6w6%2Fext-psi;a=blobdiff_plain;f=src%2Fparser.re;h=69e76c8f124244f9c7920844569dfa5da11c3a80;hp=0bca5b09aca87938a0645828ec222b70e1d1aca6;hb=e16c15d5936c3b57b05e49570fb9526920de8188;hpb=22b4a95d8fd6525e4539f843ba5bebb9a56e89f2 diff --git a/src/parser.re b/src/parser.re index 0bca5b0..69e76c8 100644 --- a/src/parser.re +++ b/src/parser.re @@ -30,8 +30,9 @@ PSI_Parser *PSI_ParserInit(PSI_Parser *P, const char *filename, psi_error_cb err } memset(P, 0, sizeof(*P)); + P->psi.file.fn = strdup(filename); P->fp = fp; - P->fn = strdup(filename); + P->col = 1; P->line = 1; P->error = error; P->flags = flags; @@ -139,29 +140,100 @@ void PSI_ParserFree(PSI_Parser **P) # error BSIZE must be greater than YYMAXFILL #endif +#define PSI_T(n) \ +(n) == PSI_T_NAME ? "NAME" : \ +(n) == PSI_T_PLUS ? "PLUS" : \ +(n) == PSI_T_MINUS ? "MINUS" : \ +(n) == PSI_T_SLASH ? "SLASH" : \ +(n) == PSI_T_ASTERISK ? "ASTERISK" : \ +(n) == PSI_T_TEMP ? "TEMP" : \ +(n) == PSI_T_FREE ? "FREE" : \ +(n) == PSI_T_SET ? "SET" : \ +(n) == PSI_T_LET ? "LET" : \ +(n) == PSI_T_RETURN ? "RETURN" : \ +(n) == PSI_T_LIB ? "LIB" : \ +(n) == PSI_T_INT ? "INT" : \ +(n) == PSI_T_UNSIGNED ? "UNSIGNED" : \ +(n) == PSI_T_EOF ? "EOF" : \ +(n) == PSI_T_QUOTED_STRING ? "QUOTED_STRING" : \ +(n) == PSI_T_EOS ? "EOS" : \ +(n) == PSI_T_STRUCT ? "STRUCT" : \ +(n) == PSI_T_LBRACE ? "LBRACE" : \ +(n) == PSI_T_RBRACE ? "RBRACE" : \ +(n) == PSI_T_COLON ? "COLON" : \ +(n) == PSI_T_LPAREN ? "LPAREN" : \ +(n) == PSI_T_NUMBER ? "NUMBER" : \ +(n) == PSI_T_RPAREN ? "RPAREN" : \ +(n) == PSI_T_BOOL ? "BOOL" : \ +(n) == PSI_T_FLOAT ? "FLOAT" : \ +(n) == PSI_T_STRING ? "STRING" : \ +(n) == PSI_T_CONST ? "CONST" : \ +(n) == PSI_T_NSNAME ? "NSNAME" : \ +(n) == PSI_T_EQUALS ? "EQUALS" : \ +(n) == PSI_T_TYPEDEF ? "TYPEDEF" : \ +(n) == PSI_T_VOID ? "VOID" : \ +(n) == PSI_T_LBRACKET ? "LBRACKET" : \ +(n) == PSI_T_RBRACKET ? "RBRACKET" : \ +(n) == PSI_T_COMMA ? "COMMA" : \ +(n) == PSI_T_ELLIPSIS ? "ELLIPSIS" : \ +(n) == PSI_T_DOUBLE ? "DOUBLE" : \ +(n) == PSI_T_INT8 ? "INT8" : \ +(n) == PSI_T_UINT8 ? "UINT8" : \ +(n) == PSI_T_INT16 ? "INT16" : \ +(n) == PSI_T_UINT16 ? "UINT16" : \ +(n) == PSI_T_INT32 ? "INT32" : \ +(n) == PSI_T_UINT32 ? "UINT32" : \ +(n) == PSI_T_INT64 ? "INT64" : \ +(n) == PSI_T_UINT64 ? "UINT64" : \ +(n) == PSI_T_FUNCTION ? "FUNCTION" : \ +(n) == PSI_T_NULL ? "NULL" : \ +(n) == PSI_T_TRUE ? "TRUE" : \ +(n) == PSI_T_FALSE ? "FALSE" : \ +(n) == PSI_T_DOLLAR ? "DOLLAR" : \ +(n) == PSI_T_CALLOC ? "CALLOC" : \ +(n) == PSI_T_OBJVAL ? "OBJVAL" : \ +(n) == PSI_T_ARRVAL ? "ARRVAL" : \ +(n) == PSI_T_PATHVAL ? "PATHVAL" : \ +(n) == PSI_T_STRLEN ? "STRLEN" : \ +(n) == PSI_T_STRVAL ? "STRVAL" : \ +(n) == PSI_T_FLOATVAL ? "FLOATVAL" : \ +(n) == PSI_T_INTVAL ? "INTVAL" : \ +(n) == PSI_T_BOOLVAL ? "BOOLVAL" : \ +(n) == PSI_T_TO_OBJECT ? "TO_OBJECT" : \ +(n) == PSI_T_TO_ARRAY ? "TO_ARRAY" : \ +(n) == PSI_T_TO_STRING ? "TO_STRING" : \ +(n) == PSI_T_TO_INT ? "TO_INT" : \ +(n) == PSI_T_TO_FLOAT ? "TO_FLOAT" : \ +(n) == PSI_T_TO_BOOL ? "TO_BOOL" : \ +(n) == PSI_T_MIXED ? "MIXED" : \ +(n) == PSI_T_ARRAY ? "ARRAY" : \ +(n) == PSI_T_OBJECT ? "OBJECT" : \ +(n) == PSI_T_AMPERSAND ? "AMPERSAND" : \ + + #define RETURN(t) do { \ P->num = t; \ + if (P->flags & PSI_PARSER_DEBUG) { \ + fprintf(stderr, "PSI> TOKEN: %d %.*s (EOF=%d %s:%zu:%zu)\n", \ + P->num, (int) (P->cur-P->tok), P->tok, P->num == PSI_T_EOF, \ + P->psi.file.fn, P->line, P->col); \ + } \ return t; \ } while(1) -/* DIGIT = [0-9] - DIGITS = DIGIT+ - DECIMALS = (+|-)? DIGIT* "." - digits ::= digits DIGIT. - decimals ::= digits DOT digits. - decimals ::= DOT digits. - decimals ::= digits DOT. - number ::= digits. - number ::= PLUS digits. - number ::= MINUS digits. - number ::= decimals. - number ::= MINUS decimals. - number ::= PLUS decimals. - -*/ +#define ADDCOLS \ + P->col += P->cur - P->tok + +#define NEWLINE \ + P->col = 1; \ + ++P->line; \ + goto nextline + token_t PSI_ParserScan(PSI_Parser *P) { for (;;) { + ADDCOLS; + nextline: P->tok = P->cur; /*!re2c re2c:indent:top = 2; @@ -169,7 +241,7 @@ token_t PSI_ParserScan(PSI_Parser *P) re2c:define:YYCURSOR = P->cur; re2c:define:YYLIMIT = P->lim; re2c:define:YYMARKER = P->mrk; - re2c:define:YYFILL = "{ if (!PSI_ParserFill(P,@@)) RETURN(-1); }"; + re2c:define:YYFILL = "{ if (!PSI_ParserFill(P,@@)) RETURN(PSI_T_EOF); }"; re2c:yyfill:parameter = 0; B = [^a-zA-Z0-9_]; @@ -177,52 +249,9 @@ token_t PSI_ParserScan(PSI_Parser *P) NAME = [a-zA-Z_]W*; NSNAME = (NAME)? ("\\" NAME)+; QUOTED_STRING = "\"" ([^\"])+ "\""; - TRUE = 'TRUE'; - FALSE = 'FALSE'; - NULL = 'NULL'; - MIXED = 'mixed'; - VOID = 'void'; - BOOL = 'bool'; - CHAR = 'char'; - SHORT = 'short'; - INT = 'int'; - LONG = 'long'; - FLOAT = 'float'; - DOUBLE = 'double'; - SIZE_T = 'size_t'; - SINT8 = 'sint8'; - UINT8 = 'uint8'; - SINT16 = 'sint16'; - UINT16 = 'uint16'; - SINT32 = 'sint32'; - UINT32 = 'uint32'; - SINT64 = 'sint64'; - UINT64 = 'uint64'; - STRING = 'string'; - ARRAY = 'array'; - FUNCTION = 'function'; - TYPEDEF = 'typedef'; - STRUCT = 'struct'; - CONST = 'const'; - LIB = 'lib'; - LET = 'let'; - SET = 'set'; - RETURN = 'return'; - FREE = 'free'; - STRLEN = 'strlen'; - STRVAL = 'strval'; - INTVAL = 'intval'; - FLOATVAL = 'floatval'; - BOOLVAL = 'boolval'; - CALLOC = 'calloc'; - TO_ARRAY = 'to_array'; - TO_STRING = 'to_string'; - TO_INT = 'to_int'; - TO_FLOAT = 'to_float'; - TO_BOOL = 'to_bool'; NUMBER = [+-]? [0-9]* "."? [0-9]+ ([eE] [+-]? [0-9]+)?; - "#" .* "\n" { ++P->line; RETURN(PSI_T_COMMENT);} + ("#"|"//") .* "\n" { NEWLINE; } "(" {RETURN(PSI_T_LPAREN);} ")" {RETURN(PSI_T_RPAREN);} ";" {RETURN(PSI_T_EOS);} @@ -234,53 +263,60 @@ token_t PSI_ParserScan(PSI_Parser *P) "]" {RETURN(PSI_T_RBRACKET);} "=" {RETURN(PSI_T_EQUALS);} "$" {RETURN(PSI_T_DOLLAR);} - "*" {RETURN(PSI_T_POINTER);} - "&" {RETURN(PSI_T_REFERENCE);} - [\r\n] { ++P->line; continue; } + "*" {RETURN(PSI_T_ASTERISK);} + "&" {RETURN(PSI_T_AMPERSAND);} + "+" {RETURN(PSI_T_PLUS);} + "-" {RETURN(PSI_T_MINUS);} + "/" {RETURN(PSI_T_SLASH);} + "..." {RETURN(PSI_T_ELLIPSIS);} + [\r\n] { NEWLINE; } [\t ]+ { continue; } - TRUE {RETURN(PSI_T_TRUE);} - FALSE {RETURN(PSI_T_FALSE);} - NULL {RETURN(PSI_T_NULL);} - MIXED {RETURN(PSI_T_MIXED);} - VOID {RETURN(PSI_T_VOID);} - BOOL {RETURN(PSI_T_BOOL);} - CHAR {RETURN(PSI_T_CHAR);} - SHORT {RETURN(PSI_T_SHORT);} - INT {RETURN(PSI_T_INT);} - LONG {RETURN(PSI_T_LONG);} - FLOAT {RETURN(PSI_T_FLOAT);} - DOUBLE {RETURN(PSI_T_DOUBLE);} - SIZE_T {RETURN(PSI_T_SIZE_T);} - SINT8 {RETURN(PSI_T_SINT8);} - UINT8 {RETURN(PSI_T_UINT8);} - SINT16 {RETURN(PSI_T_SINT16);} - UINT16 {RETURN(PSI_T_UINT16);} - SINT32 {RETURN(PSI_T_SINT32);} - UINT32 {RETURN(PSI_T_UINT32);} - SINT64 {RETURN(PSI_T_SINT64);} - UINT64 {RETURN(PSI_T_UINT64);} - STRING {RETURN(PSI_T_STRING);} - ARRAY {RETURN(PSI_T_ARRAY);} - FUNCTION {RETURN(PSI_T_FUNCTION);} - TYPEDEF {RETURN(PSI_T_TYPEDEF);} - STRUCT {RETURN(PSI_T_STRUCT);} - CONST {RETURN(PSI_T_CONST);} - LIB {RETURN(PSI_T_LIB);} - LET {RETURN(PSI_T_LET);} - SET {RETURN(PSI_T_SET);} - RETURN {RETURN(PSI_T_RETURN);} - FREE {RETURN(PSI_T_FREE);} - STRLEN {RETURN(PSI_T_STRLEN);} - STRVAL {RETURN(PSI_T_STRVAL);} - INTVAL {RETURN(PSI_T_INTVAL);} - FLOATVAL {RETURN(PSI_T_FLOATVAL);} - BOOLVAL {RETURN(PSI_T_BOOLVAL);} - CALLOC {RETURN(PSI_T_CALLOC);} - TO_ARRAY {RETURN(PSI_T_TO_ARRAY);} - TO_STRING {RETURN(PSI_T_TO_STRING);} - TO_INT {RETURN(PSI_T_TO_INT);} - TO_FLOAT {RETURN(PSI_T_TO_FLOAT);} - TO_BOOL {RETURN(PSI_T_TO_BOOL);} + 'TRUE' {RETURN(PSI_T_TRUE);} + 'FALSE' {RETURN(PSI_T_FALSE);} + 'NULL' {RETURN(PSI_T_NULL);} + 'MIXED' {RETURN(PSI_T_MIXED);} + 'VOID' {RETURN(PSI_T_VOID);} + 'BOOL' {RETURN(PSI_T_BOOL);} + 'INT' {RETURN(PSI_T_INT);} + 'FLOAT' {RETURN(PSI_T_FLOAT);} + 'DOUBLE' {RETURN(PSI_T_DOUBLE);} + 'INT8_T' {RETURN(PSI_T_INT8);} + 'UINT8_T' {RETURN(PSI_T_UINT8);} + 'INT16_T' {RETURN(PSI_T_INT16);} + 'UINT16_T' {RETURN(PSI_T_UINT16);} + 'INT32_T' {RETURN(PSI_T_INT32);} + 'UINT32_T' {RETURN(PSI_T_UINT32);} + 'INT64_T' {RETURN(PSI_T_INT64);} + 'UINT64_T' {RETURN(PSI_T_UINT64);} + 'UNSIGNED' {RETURN(PSI_T_UNSIGNED);} + 'STRING' {RETURN(PSI_T_STRING);} + 'ARRAY' {RETURN(PSI_T_ARRAY);} + 'OBJECT' {RETURN(PSI_T_OBJECT);} + 'FUNCTION' {RETURN(PSI_T_FUNCTION);} + 'TYPEDEF' {RETURN(PSI_T_TYPEDEF);} + 'STRUCT' {RETURN(PSI_T_STRUCT);} + 'CONST' {RETURN(PSI_T_CONST);} + 'LIB' {RETURN(PSI_T_LIB);} + 'LET' {RETURN(PSI_T_LET);} + 'SET' {RETURN(PSI_T_SET);} + 'RETURN' {RETURN(PSI_T_RETURN);} + 'FREE' {RETURN(PSI_T_FREE);} + 'TEMP' {RETURN(PSI_T_TEMP);} + 'STRLEN' {RETURN(PSI_T_STRLEN);} + 'STRVAL' {RETURN(PSI_T_STRVAL);} + 'PATHVAL' {RETURN(PSI_T_PATHVAL);} + 'INTVAL' {RETURN(PSI_T_INTVAL);} + 'FLOATVAL' {RETURN(PSI_T_FLOATVAL);} + 'BOOLVAL' {RETURN(PSI_T_BOOLVAL);} + 'ARRVAL' {RETURN(PSI_T_ARRVAL);} + 'OBJVAL' {RETURN(PSI_T_OBJVAL);} + 'CALLOC' {RETURN(PSI_T_CALLOC);} + 'TO_OBJECT' {RETURN(PSI_T_TO_OBJECT);} + 'TO_ARRAY' {RETURN(PSI_T_TO_ARRAY);} + 'TO_STRING' {RETURN(PSI_T_TO_STRING);} + 'TO_INT' {RETURN(PSI_T_TO_INT);} + 'TO_FLOAT' {RETURN(PSI_T_TO_FLOAT);} + 'TO_BOOL' {RETURN(PSI_T_TO_BOOL);} NUMBER {RETURN(PSI_T_NUMBER);} NAME {RETURN(PSI_T_NAME);} NSNAME {RETURN(PSI_T_NSNAME);}