bc: prepare for char-by-char input handling
authorDenys Vlasenko <vda.linux@googlemail.com>
Wed, 26 Dec 2018 17:32:43 +0000 (18:32 +0100)
committerDenys Vlasenko <vda.linux@googlemail.com>
Wed, 26 Dec 2018 17:32:43 +0000 (18:32 +0100)
function                                             old     new   delta
peek_inbuf                                             -     292    +292
parse_lex_by_checking_eq_sign                          -      26     +26
eat_inbuf                                              -      22     +22
zbc_vm_execute_FILE                                   52      61      +9
bc_lex_lineComment                                    29      30      +1
zbc_lex_number                                       174     172      -2
bc_vm_run                                            104      99      -5
zbc_num_divmod                                       156     150      -6
bc_lex_file                                           24       -     -24
bc_lex_assign                                         26       -     -26
zbc_lex_next                                        1982    1587    -395
------------------------------------------------------------------------------
(add/remove: 3/2 grow/shrink: 2/4 up/down: 350/-458)         Total: -108 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
miscutils/bc.c
testsuite/dc_strings_results.txt

index f4e499f13438e6fb74f2fabd6975e950c1ae8e3d..5789869a70281976c956d47007df5a7847c85bd9 100644 (file)
@@ -713,7 +713,6 @@ dc_LEX_to_INST[] = { // starts at XC_LEX_OP_POWER       // corresponding XC/DC_L
 typedef struct BcParse {
        smallint lex;      // was BcLexType // first member is most used
        smallint lex_last; // was BcLexType
-       bool   lex_newline;
        size_t lex_line;
        const char *lex_inbuf;
        const char *lex_end;
@@ -2741,147 +2740,6 @@ static BC_STATUS zbc_num_parse(BcNum *n, const char *val, unsigned base_t)
 }
 #define zbc_num_parse(...) (zbc_num_parse(__VA_ARGS__) COMMA_SUCCESS)
 
-static void bc_lex_lineComment(void)
-{
-       BcParse *p = &G.prs;
-       // Try: echo -n '#foo' | bc
-       p->lex = XC_LEX_WHITESPACE;
-       while (p->lex_inbuf < p->lex_end && *p->lex_inbuf != '\n')
-               p->lex_inbuf++;
-}
-
-static void bc_lex_whitespace(void)
-{
-       BcParse *p = &G.prs;
-       p->lex = XC_LEX_WHITESPACE;
-       for (;;) {
-               char c = *p->lex_inbuf;
-               if (c == '\n') // this is XC_LEX_NLINE, not XC_LEX_WHITESPACE
-                       break;
-               if (!isspace(c))
-                       break;
-               p->lex_inbuf++;
-       }
-}
-
-static BC_STATUS zbc_lex_number(char start)
-{
-       BcParse *p = &G.prs;
-       const char *buf = p->lex_inbuf;
-       size_t len, i, ccnt;
-       bool pt;
-
-       pt = (start == '.');
-       p->lex = XC_LEX_NUMBER;
-       ccnt = i = 0;
-       for (;;) {
-               char c = buf[i];
-               if (c == '\0')
-                       break;
-               if (c == '\\' && buf[i + 1] == '\n') {
-                       i += 2;
-                       //number_of_backslashes++ - see comment below
-                       continue;
-               }
-               if (!isdigit(c) && (c < 'A' || c > 'F')) {
-                       if (c != '.') break;
-                       // if '.' was already seen, stop on second one:
-                       if (pt) break;
-                       pt = true;
-               }
-               // buf[i] is one of "0-9A-F."
-               i++;
-               if (c != '.')
-                       ccnt = i;
-       }
-       //ccnt is the number of chars in the number string, excluding possible
-       //trailing "[\<newline>].[\<newline>]" (with any number of \<NL> repetitions).
-       //i is buf[i] index of the first not-yet-parsed char after that.
-       p->lex_inbuf += i;
-
-       // This might overestimate the size, if there are "\<NL>"'s
-       // in the number. Subtracting number_of_backslashes*2 correctly
-       // is not that easy: consider that in the case of "NNN.\<NL>"
-       // loop above will count "\<NL>" before it realizes it is not
-       // in fact *inside* the number:
-       len = ccnt + 1; // +1 byte for NUL termination
-
-       // This check makes sense only if size_t is (much) larger than BC_MAX_NUM.
-       if (SIZE_MAX > (BC_MAX_NUM | 0xff)) {
-               if (len > BC_MAX_NUM)
-                       RETURN_STATUS(bc_error("number too long: must be [1,"BC_MAX_NUM_STR"]"));
-       }
-
-       bc_vec_pop_all(&p->lex_strnumbuf);
-       bc_vec_expand(&p->lex_strnumbuf, 1 + len);
-       bc_vec_push(&p->lex_strnumbuf, &start);
-
-       while (ccnt != 0) {
-               // If we have hit a backslash, skip it. We don't have
-               // to check for a newline because it's guaranteed.
-               if (*buf == '\\') {
-                       buf += 2;
-                       ccnt -= 2;
-                       continue;
-               }
-               bc_vec_push(&p->lex_strnumbuf, buf);
-               buf++;
-               ccnt--;
-       }
-
-       bc_vec_pushZeroByte(&p->lex_strnumbuf);
-
-       RETURN_STATUS(BC_STATUS_SUCCESS);
-}
-#define zbc_lex_number(...) (zbc_lex_number(__VA_ARGS__) COMMA_SUCCESS)
-
-static void bc_lex_name(void)
-{
-       BcParse *p = &G.prs;
-       size_t i;
-       const char *buf;
-
-       p->lex = XC_LEX_NAME;
-
-       i = 0;
-       buf = p->lex_inbuf - 1;
-       for (;;) {
-               char c = buf[i];
-               if ((c < 'a' || c > 'z') && !isdigit(c) && c != '_') break;
-               i++;
-       }
-
-#if 0 // We do not protect against people with gigabyte-long names
-       // This check makes sense only if size_t is (much) larger than BC_MAX_STRING.
-       if (SIZE_MAX > (BC_MAX_STRING | 0xff)) {
-               if (i > BC_MAX_STRING)
-                       return bc_error("name too long: must be [1,"BC_MAX_STRING_STR"]");
-       }
-#endif
-       bc_vec_string(&p->lex_strnumbuf, i, buf);
-
-       // Increment the index. We minus 1 because it has already been incremented.
-       p->lex_inbuf += i - 1;
-
-       //return BC_STATUS_SUCCESS;
-}
-
-static void bc_lex_init(void)
-{
-       bc_char_vec_init(&G.prs.lex_strnumbuf);
-}
-
-static void bc_lex_free(void)
-{
-       bc_vec_free(&G.prs.lex_strnumbuf);
-}
-
-static void bc_lex_file(void)
-{
-       G.err_line = G.prs.lex_line = 1;
-       G.prs.lex_newline = false;
-}
-
 static bool bc_lex_more_input(void)
 {
        BcParse *p = &G.prs;
@@ -2996,6 +2854,147 @@ static bool bc_lex_more_input(void)
        return G.input_buffer.len > 1;
 }
 
+// p->lex_inbuf points to the current string to be parsed.
+// if p->lex_inbuf points to '\0', it's either EOF or it points after
+// last processed line's terminating '\n' (and more reading needs to be done
+// to get next character).
+//
+// If you are in a situation where that is a possibility, call peek_inbuf().
+// If necessary, it performs more reading and changes p->lex_inbuf,
+// then it returns *p->lex_inbuf (which will be '\0' only if it's EOF).
+// After it, just referencing *p->lex_inbuf is valid, and if it wasn't '\0',
+// it's ok to do p->lex_inbuf++ once without end-of-buffer checking.
+//
+// eat_inbuf() is equvalent to "peek_inbuf(); if (c) p->lex_inbuf++":
+// it returns current char and advances the pointer (if not EOF).
+// After eat_inbuf(), referencing p->lex_inbuf[-1] and *p->lex_inbuf is valid.
+//
+// In many cases, you can use fast *p->lex_inbuf instead of peek_inbuf():
+// unless prev char might have been '\n', *p->lex_inbuf is '\0' ONLY
+// on real EOF, not end-of-buffer.
+static char peek_inbuf(void)
+{
+       if (G.prs.lex_inbuf == G.prs.lex_end) {
+               if (G.prs.lex_input_fp)
+                       if (!bc_lex_more_input())
+                               G.prs.lex_input_fp = NULL;
+       }
+       return *G.prs.lex_inbuf;
+}
+static char eat_inbuf(void)
+{
+       char c = peek_inbuf();
+       if (c) G.prs.lex_inbuf++;
+       return c;
+}
+
+static void bc_lex_lineComment(void)
+{
+       BcParse *p = &G.prs;
+       char c;
+
+       // Try: echo -n '#foo' | bc
+       p->lex = XC_LEX_WHITESPACE;
+
+       // We depend here on input being done in whole lines:
+       // '\0' which isn't the EOF can only be seen after '\n'.
+       while ((c = *p->lex_inbuf) != '\n' && c != '\0')
+               p->lex_inbuf++;
+}
+
+static void bc_lex_whitespace(void)
+{
+       BcParse *p = &G.prs;
+
+       p->lex = XC_LEX_WHITESPACE;
+       for (;;) {
+               // We depend here on input being done in whole lines:
+               // '\0' which isn't the EOF can only be seen after '\n'.
+               char c = *p->lex_inbuf;
+               if (c == '\n') // this is XC_LEX_NLINE, not XC_LEX_WHITESPACE
+                       break;
+               if (!isspace(c))
+                       break;
+               p->lex_inbuf++;
+       }
+}
+
+static BC_STATUS zbc_lex_number(char last)
+{
+       BcParse *p = &G.prs;
+       bool pt;
+
+       bc_vec_pop_all(&p->lex_strnumbuf);
+       bc_vec_pushByte(&p->lex_strnumbuf, last);
+
+       pt = (last == '.');
+       p->lex = XC_LEX_NUMBER;
+       for (;;) {
+               // We depend here on input being done in whole lines:
+               // '\0' which isn't the EOF can only be seen after '\n'.
+               char c = *p->lex_inbuf;
+ check_c:
+               if (c == '\0')
+                       break;
+               if (c == '\\' && p->lex_inbuf[1] == '\n') {
+                       p->lex_inbuf += 2;
+                       p->lex_line++;
+                       c = peek_inbuf(); // force next line to be read
+                       goto check_c;
+               }
+               if (!isdigit(c) && (c < 'A' || c > 'F')) {
+                       if (c != '.') break;
+                       // if '.' was already seen, stop on second one:
+                       if (pt) break;
+                       pt = true;
+               }
+               // c is one of "0-9A-F."
+               last = c;
+               bc_vec_push(&p->lex_strnumbuf, p->lex_inbuf);
+               p->lex_inbuf++;
+       }
+       if (last == '.') // remove trailing '.' if any
+               bc_vec_pop(&p->lex_strnumbuf);
+       bc_vec_pushZeroByte(&p->lex_strnumbuf);
+
+       G.err_line = G.prs.lex_line;
+       RETURN_STATUS(BC_STATUS_SUCCESS);
+}
+#define zbc_lex_number(...) (zbc_lex_number(__VA_ARGS__) COMMA_SUCCESS)
+
+static void bc_lex_name(void)
+{
+       BcParse *p = &G.prs;
+       size_t i;
+       const char *buf;
+
+       p->lex = XC_LEX_NAME;
+
+       // Since names can't cross lines with \<newline>,
+       // we depend on the fact that whole line is in the buffer
+       i = 0;
+       buf = p->lex_inbuf - 1;
+       for (;;) {
+               char c = buf[i];
+               if ((c < 'a' || c > 'z') && !isdigit(c) && c != '_') break;
+               i++;
+       }
+
+#if 0 // We do not protect against people with gigabyte-long names
+       // This check makes sense only if size_t is (much) larger than BC_MAX_STRING.
+       if (SIZE_MAX > (BC_MAX_STRING | 0xff)) {
+               if (i > BC_MAX_STRING)
+                       return bc_error("name too long: must be [1,"BC_MAX_STRING_STR"]");
+       }
+#endif
+       bc_vec_string(&p->lex_strnumbuf, i, buf);
+
+       // Increment the index. We minus 1 because it has already been incremented.
+       p->lex_inbuf += i - 1;
+
+       //return BC_STATUS_SUCCESS;
+}
+
 IF_BC(static BC_STATUS zbc_lex_token(void);)
 IF_DC(static BC_STATUS zdc_lex_token(void);)
 #define zbc_lex_token(...) (zbc_lex_token(__VA_ARGS__) COMMA_SUCCESS)
@@ -3007,26 +3006,18 @@ static BC_STATUS zbc_lex_next(void)
        BcStatus s;
 
        p->lex_last = p->lex;
-       if (p->lex_last == XC_LEX_EOF) RETURN_STATUS(bc_error("end of file"));
-
-       p->lex_line += p->lex_newline;
-       G.err_line = p->lex_line;
-       p->lex_newline = false;
+       if (p->lex_last == XC_LEX_EOF)
+               RETURN_STATUS(bc_error("end of file"));
 
        // Loop until failure or we don't have whitespace. This
        // is so the parser doesn't get inundated with whitespace.
        // Comments are also XC_LEX_WHITESPACE tokens and eaten here.
        s = BC_STATUS_SUCCESS;
        do {
-               if (p->lex_inbuf == p->lex_end) {
+               if (*p->lex_inbuf == '\0') {
                        p->lex = XC_LEX_EOF;
-                       if (!G.prs.lex_input_fp)
+                       if (peek_inbuf() == '\0')
                                RETURN_STATUS(BC_STATUS_SUCCESS);
-                       if (!bc_lex_more_input()) {
-                               G.prs.lex_input_fp = NULL;
-                               RETURN_STATUS(BC_STATUS_SUCCESS);
-                       }
-                       // here it's guaranteed that p->lex_ibuf is below p->lex_end
                }
                p->lex_next_at = p->lex_inbuf;
                dbg_lex("next string to parse:'%.*s'",
@@ -3126,80 +3117,70 @@ static BC_STATUS zbc_lex_identifier(void)
 static BC_STATUS zbc_lex_string(void)
 {
        BcParse *p = &G.prs;
-       size_t len, nls, i;
 
        p->lex = XC_LEX_STR;
-
-       nls = 0;
-       i = 0;
+       bc_vec_pop_all(&p->lex_strnumbuf);
        for (;;) {
-               char c = p->lex_inbuf[i];
+               char c = peek_inbuf(); // strings can cross lines
                if (c == '\0') {
-                       p->lex_inbuf += i;
-                       RETURN_STATUS(bc_error("unterminated string"));
+                       RETURN_STATUS(bc_error("unterminated string1"));
                }
                if (c == '"')
                        break;
-               nls += (c == '\n');
-               i++;
-       }
-
-       len = i;
-       // This check makes sense only if size_t is (much) larger than BC_MAX_STRING.
-       if (SIZE_MAX > (BC_MAX_STRING | 0xff)) {
-               if (len > BC_MAX_STRING)
-                       RETURN_STATUS(bc_error("string too long: must be [1,"BC_MAX_STRING_STR"]"));
+               if (c == '\n')
+                       p->lex_line++;
+               bc_vec_push(&p->lex_strnumbuf, p->lex_inbuf);
+               p->lex_inbuf++;
        }
-       bc_vec_string(&p->lex_strnumbuf, len, p->lex_inbuf);
+       bc_vec_pushZeroByte(&p->lex_strnumbuf);
+       p->lex_inbuf++;
 
-       p->lex_inbuf += i + 1;
-       p->lex_line += nls;
        G.err_line = p->lex_line;
-
        RETURN_STATUS(BC_STATUS_SUCCESS);
 }
 #define zbc_lex_string(...) (zbc_lex_string(__VA_ARGS__) COMMA_SUCCESS)
 
-static void bc_lex_assign(unsigned with_and_without)
+static void parse_lex_by_checking_eq_sign(unsigned with_and_without)
 {
        BcParse *p = &G.prs;
        if (*p->lex_inbuf == '=') {
+               // ^^^ not using peek_inbuf() since '==' etc can't be split across lines
                p->lex_inbuf++;
                with_and_without >>= 8; // store "with" value
        } // else store "without" value
        p->lex = (with_and_without & 0xff);
 }
-#define bc_lex_assign(with, without) \
-       bc_lex_assign(((with)<<8)|(without))
+#define parse_lex_by_checking_eq_sign(with, without) \
+       parse_lex_by_checking_eq_sign(((with)<<8)|(without))
 
 static BC_STATUS zbc_lex_comment(void)
 {
        BcParse *p = &G.prs;
-       size_t i, nls = 0;
-       const char *buf = p->lex_inbuf;
 
        p->lex = XC_LEX_WHITESPACE;
-       i = 0; /* here lex_inbuf[0] is the '*' of opening comment delimiter */
+       // here lex_inbuf is at '*' of opening comment delimiter
        for (;;) {
-               char c = buf[++i];
+               char c;
+
+               p->lex_inbuf++;
+               c = peek_inbuf();
  check_star:
                if (c == '*') {
-                       c = buf[++i];
+                       p->lex_inbuf++;
+                       c = peek_inbuf();
                        if (c == '/')
                                break;
                        goto check_star;
                }
                if (c == '\0') {
-                       p->lex_inbuf += i;
                        RETURN_STATUS(bc_error("unterminated comment"));
                }
-               nls += (c == '\n');
+               if (c == '\n')
+                       p->lex_line++;
        }
+       p->lex_inbuf++; // skip trailing '/'
 
-       p->lex_inbuf += i + 1;
-       p->lex_line += nls;
        G.err_line = p->lex_line;
-
        RETURN_STATUS(BC_STATUS_SUCCESS);
 }
 #define zbc_lex_comment(...) (zbc_lex_comment(__VA_ARGS__) COMMA_SUCCESS)
@@ -3209,7 +3190,7 @@ static BC_STATUS zbc_lex_token(void)
 {
        BcParse *p = &G.prs;
        BcStatus s = BC_STATUS_SUCCESS;
-       char c = *p->lex_inbuf++;
+       char c = eat_inbuf();
        char c2;
 
        // This is the workhorse of the lexer.
@@ -3217,11 +3198,10 @@ static BC_STATUS zbc_lex_token(void)
 //             case '\0': // probably never reached
 //                     p->lex_inbuf--;
 //                     p->lex = XC_LEX_EOF;
-//                     p->lex_newline = true;
 //                     break;
                case '\n':
+                       p->lex_line++;
                        p->lex = XC_LEX_NLINE;
-                       p->lex_newline = true;
                        break;
                case '\t':
                case '\v':
@@ -3231,7 +3211,7 @@ static BC_STATUS zbc_lex_token(void)
                        bc_lex_whitespace();
                        break;
                case '!':
-                       bc_lex_assign(XC_LEX_OP_REL_NE, BC_LEX_OP_BOOL_NOT);
+                       parse_lex_by_checking_eq_sign(XC_LEX_OP_REL_NE, BC_LEX_OP_BOOL_NOT);
                        if (p->lex == BC_LEX_OP_BOOL_NOT) {
                                s = zbc_POSIX_does_not_allow_bool_ops_this_is_bad("!");
                                if (s) RETURN_STATUS(s);
@@ -3246,7 +3226,7 @@ static BC_STATUS zbc_lex_token(void)
                        bc_lex_lineComment();
                        break;
                case '%':
-                       bc_lex_assign(BC_LEX_OP_ASSIGN_MODULUS, XC_LEX_OP_MODULUS);
+                       parse_lex_by_checking_eq_sign(BC_LEX_OP_ASSIGN_MODULUS, XC_LEX_OP_MODULUS);
                        break;
                case '&':
                        c2 = *p->lex_inbuf;
@@ -3265,7 +3245,7 @@ static BC_STATUS zbc_lex_token(void)
                        p->lex = (BcLexType)(c - '(' + BC_LEX_LPAREN);
                        break;
                case '*':
-                       bc_lex_assign(BC_LEX_OP_ASSIGN_MULTIPLY, XC_LEX_OP_MULTIPLY);
+                       parse_lex_by_checking_eq_sign(BC_LEX_OP_ASSIGN_MULTIPLY, XC_LEX_OP_MULTIPLY);
                        break;
                case '+':
                        c2 = *p->lex_inbuf;
@@ -3273,7 +3253,7 @@ static BC_STATUS zbc_lex_token(void)
                                p->lex_inbuf++;
                                p->lex = BC_LEX_OP_INC;
                        } else
-                               bc_lex_assign(BC_LEX_OP_ASSIGN_PLUS, XC_LEX_OP_PLUS);
+                               parse_lex_by_checking_eq_sign(BC_LEX_OP_ASSIGN_PLUS, XC_LEX_OP_PLUS);
                        break;
                case ',':
                        p->lex = BC_LEX_COMMA;
@@ -3284,7 +3264,7 @@ static BC_STATUS zbc_lex_token(void)
                                p->lex_inbuf++;
                                p->lex = BC_LEX_OP_DEC;
                        } else
-                               bc_lex_assign(BC_LEX_OP_ASSIGN_MINUS, XC_LEX_OP_MINUS);
+                               parse_lex_by_checking_eq_sign(BC_LEX_OP_ASSIGN_MINUS, XC_LEX_OP_MINUS);
                        break;
                case '.':
                        if (isdigit(*p->lex_inbuf))
@@ -3299,7 +3279,7 @@ static BC_STATUS zbc_lex_token(void)
                        if (c2 == '*')
                                s = zbc_lex_comment();
                        else
-                               bc_lex_assign(BC_LEX_OP_ASSIGN_DIVIDE, XC_LEX_OP_DIVIDE);
+                               parse_lex_by_checking_eq_sign(BC_LEX_OP_ASSIGN_DIVIDE, XC_LEX_OP_DIVIDE);
                        break;
                case '0':
                case '1':
@@ -3323,13 +3303,13 @@ static BC_STATUS zbc_lex_token(void)
                        p->lex = BC_LEX_SCOLON;
                        break;
                case '<':
-                       bc_lex_assign(XC_LEX_OP_REL_LE, XC_LEX_OP_REL_LT);
+                       parse_lex_by_checking_eq_sign(XC_LEX_OP_REL_LE, XC_LEX_OP_REL_LT);
                        break;
                case '=':
-                       bc_lex_assign(XC_LEX_OP_REL_EQ, BC_LEX_OP_ASSIGN);
+                       parse_lex_by_checking_eq_sign(XC_LEX_OP_REL_EQ, BC_LEX_OP_ASSIGN);
                        break;
                case '>':
-                       bc_lex_assign(XC_LEX_OP_REL_GE, XC_LEX_OP_REL_GT);
+                       parse_lex_by_checking_eq_sign(XC_LEX_OP_REL_GE, XC_LEX_OP_REL_GT);
                        break;
                case '[':
                case ']':
@@ -3343,7 +3323,7 @@ static BC_STATUS zbc_lex_token(void)
                                s = bc_error_bad_character(c);
                        break;
                case '^':
-                       bc_lex_assign(BC_LEX_OP_ASSIGN_POWER, XC_LEX_OP_POWER);
+                       parse_lex_by_checking_eq_sign(BC_LEX_OP_ASSIGN_POWER, XC_LEX_OP_POWER);
                        break;
                case 'a':
                case 'b':
@@ -3422,43 +3402,30 @@ static BC_STATUS zdc_lex_register(void)
 static BC_STATUS zdc_lex_string(void)
 {
        BcParse *p = &G.prs;
-       size_t depth, nls, i;
+       size_t depth;
 
        p->lex = XC_LEX_STR;
        bc_vec_pop_all(&p->lex_strnumbuf);
 
-       nls = 0;
        depth = 1;
-       i = 0;
        for (;;) {
-               char c = p->lex_inbuf[i];
+               char c = peek_inbuf();
                if (c == '\0') {
-                       p->lex_inbuf += i;
-                       RETURN_STATUS(bc_error("string end could not be found"));
-               }
-               nls += (c == '\n');
-               if (i == 0 || p->lex_inbuf[i - 1] != '\\') {
-                       if (c == '[') depth++;
-                       if (c == ']')
-                               if (--depth == 0)
-                                       break;
+                       RETURN_STATUS(bc_error("unterminated string"));
                }
-               bc_vec_push(&p->lex_strnumbuf, &p->lex_inbuf[i]);
-               i++;
+               if (c == '[') depth++;
+               if (c == ']')
+                       if (--depth == 0)
+                               break;
+               if (c == '\n')
+                       p->lex_line++;
+               bc_vec_push(&p->lex_strnumbuf, p->lex_inbuf);
+               p->lex_inbuf++;
        }
-       i++;
-
        bc_vec_pushZeroByte(&p->lex_strnumbuf);
-       // This check makes sense only if size_t is (much) larger than BC_MAX_STRING.
-       if (SIZE_MAX > (BC_MAX_STRING | 0xff)) {
-               if (i > BC_MAX_STRING)
-                       RETURN_STATUS(bc_error("string too long: must be [1,"BC_MAX_STRING_STR"]"));
-       }
+       p->lex_inbuf++; // skip trailing ']'
 
-       p->lex_inbuf += i;
-       p->lex_line += nls;
        G.err_line = p->lex_line;
-
        RETURN_STATUS(BC_STATUS_SUCCESS);
 }
 #define zdc_lex_string(...) (zdc_lex_string(__VA_ARGS__) COMMA_SUCCESS)
@@ -3486,7 +3453,7 @@ static BC_STATUS zdc_lex_token(void)
        }
 
        s = BC_STATUS_SUCCESS;
-       c = *p->lex_inbuf++;
+       c = eat_inbuf();
        if (c >= '%' && c <= '~'
         && (p->lex = dc_char_to_LEX[c - '%']) != XC_LEX_INVALID
        ) {
@@ -3507,15 +3474,14 @@ static BC_STATUS zdc_lex_token(void)
                        // commands are not executed on pressing <enter>).
                        // IOW: typing "1p<enter>" should print "1" _at once_,
                        // not after some more input.
+                       p->lex_line++;
                        p->lex = XC_LEX_NLINE;
-                       p->lex_newline = true;
                        break;
                case '\t':
                case '\v':
                case '\f':
                case '\r':
                case ' ':
-                       p->lex_newline = 0; // was (c == '\n')
                        bc_lex_whitespace();
                        break;
                case '!':
@@ -3702,7 +3668,7 @@ static void bc_parse_free(void)
        IF_BC(bc_vec_free(&p->exits);)
        IF_BC(bc_vec_free(&p->conds);)
        IF_BC(bc_vec_free(&p->ops);)
-       bc_lex_free();
+       bc_vec_free(&G.prs.lex_strnumbuf);
 }
 
 static void bc_parse_create(size_t fidx)
@@ -3710,7 +3676,7 @@ static void bc_parse_create(size_t fidx)
        BcParse *p = &G.prs;
        memset(p, 0, sizeof(BcParse));
 
-       bc_lex_init();
+       bc_char_vec_init(&G.prs.lex_strnumbuf);
        IF_BC(bc_vec_init(&p->exits, sizeof(size_t), NULL);)
        IF_BC(bc_vec_init(&p->conds, sizeof(size_t), NULL);)
        IF_BC(bc_vec_init(&p->ops, sizeof(BcLexType), NULL);)
@@ -4753,11 +4719,13 @@ static BcStatus bc_parse_expr_empty_ok(uint8_t flags)
                switch (t) {
                        case BC_LEX_OP_INC:
                        case BC_LEX_OP_DEC:
+                               dbg_lex("%s:%d LEX_OP_INC/DEC", __func__, __LINE__);
                                s = zbc_parse_incdec(&prev, &paren_expr, &nexprs, flags);
                                rprn = bin_last = false;
                                //get_token = false; - already is
                                break;
                        case XC_LEX_OP_MINUS:
+                               dbg_lex("%s:%d LEX_OP_MINUS", __func__, __LINE__);
                                s = zbc_parse_minus(&prev, ops_bgn, rprn, &nexprs);
                                rprn = false;
                                //get_token = false; - already is
@@ -4770,6 +4738,7 @@ static BcStatus bc_parse_expr_empty_ok(uint8_t flags)
                        case BC_LEX_OP_ASSIGN_PLUS:
                        case BC_LEX_OP_ASSIGN_MINUS:
                        case BC_LEX_OP_ASSIGN:
+                               dbg_lex("%s:%d LEX_ASSIGNxyz", __func__, __LINE__);
                                if (prev != XC_INST_VAR && prev != XC_INST_ARRAY_ELEM
                                 && prev != XC_INST_SCALE && prev != XC_INST_IBASE
                                 && prev != XC_INST_OBASE && prev != BC_INST_LAST
@@ -4794,6 +4763,7 @@ static BcStatus bc_parse_expr_empty_ok(uint8_t flags)
                        case BC_LEX_OP_BOOL_NOT:
                        case BC_LEX_OP_BOOL_OR:
                        case BC_LEX_OP_BOOL_AND:
+                               dbg_lex("%s:%d LEX_OP_xyz", __func__, __LINE__);
                                if (((t == BC_LEX_OP_BOOL_NOT) != bin_last)
                                 || (t != BC_LEX_OP_BOOL_NOT && prev == XC_INST_BOOL_NOT)
                                ) {
@@ -4808,6 +4778,7 @@ static BcStatus bc_parse_expr_empty_ok(uint8_t flags)
                                bin_last = (t != BC_LEX_OP_BOOL_NOT);
                                break;
                        case BC_LEX_LPAREN:
+                               dbg_lex("%s:%d LEX_LPAREN", __func__, __LINE__);
                                if (BC_PARSE_LEAF(prev, rprn))
                                        return bc_error_bad_expression();
                                bc_vec_push(&p->ops, &t);
@@ -4817,6 +4788,7 @@ static BcStatus bc_parse_expr_empty_ok(uint8_t flags)
                                rprn = bin_last = false;
                                break;
                        case BC_LEX_RPAREN:
+                               dbg_lex("%s:%d LEX_RPAREN", __func__, __LINE__);
                                if (bin_last || prev == XC_INST_BOOL_NOT)
                                        return bc_error_bad_expression();
                                if (nparens == 0) {
@@ -4833,6 +4805,7 @@ static BcStatus bc_parse_expr_empty_ok(uint8_t flags)
                                bin_last = false;
                                break;
                        case XC_LEX_NAME:
+                               dbg_lex("%s:%d LEX_NAME", __func__, __LINE__);
                                if (BC_PARSE_LEAF(prev, rprn))
                                        return bc_error_bad_expression();
                                s = zbc_parse_name(&prev, flags & ~BC_PARSE_NOCALL);
@@ -4842,6 +4815,7 @@ static BcStatus bc_parse_expr_empty_ok(uint8_t flags)
                                nexprs++;
                                break;
                        case XC_LEX_NUMBER:
+                               dbg_lex("%s:%d LEX_NUMBER", __func__, __LINE__);
                                if (BC_PARSE_LEAF(prev, rprn))
                                        return bc_error_bad_expression();
                                bc_parse_pushNUM();
@@ -4854,6 +4828,7 @@ static BcStatus bc_parse_expr_empty_ok(uint8_t flags)
                        case BC_LEX_KEY_IBASE:
                        case BC_LEX_KEY_LAST:
                        case BC_LEX_KEY_OBASE:
+                               dbg_lex("%s:%d LEX_IBASE/LAST/OBASE", __func__, __LINE__);
                                if (BC_PARSE_LEAF(prev, rprn))
                                        return bc_error_bad_expression();
                                prev = (char) (t - BC_LEX_KEY_IBASE + XC_INST_IBASE);
@@ -4865,6 +4840,7 @@ static BcStatus bc_parse_expr_empty_ok(uint8_t flags)
                                break;
                        case BC_LEX_KEY_LENGTH:
                        case BC_LEX_KEY_SQRT:
+                               dbg_lex("%s:%d LEX_LEN/SQRT", __func__, __LINE__);
                                if (BC_PARSE_LEAF(prev, rprn))
                                        return bc_error_bad_expression();
                                s = zbc_parse_builtin(t, flags, &prev);
@@ -4874,6 +4850,7 @@ static BcStatus bc_parse_expr_empty_ok(uint8_t flags)
                                nexprs++;
                                break;
                        case BC_LEX_KEY_READ:
+                               dbg_lex("%s:%d LEX_READ", __func__, __LINE__);
                                if (BC_PARSE_LEAF(prev, rprn))
                                        return bc_error_bad_expression();
                                s = zbc_parse_read();
@@ -4884,6 +4861,7 @@ static BcStatus bc_parse_expr_empty_ok(uint8_t flags)
                                nexprs++;
                                break;
                        case BC_LEX_KEY_SCALE:
+                               dbg_lex("%s:%d LEX_SCALE", __func__, __LINE__);
                                if (BC_PARSE_LEAF(prev, rprn))
                                        return bc_error_bad_expression();
                                s = zbc_parse_scale(&prev, flags);
@@ -5348,7 +5326,7 @@ static BC_STATUS zbc_program_read(void)
 
        sv_parse = G.prs; // struct copy
        bc_parse_create(BC_PROG_READ);
-       //bc_lex_file(&G.prs.l); - not needed, error line info is not printed for read()
+       //G.err_line = G.prs.lex_line = 1; - not needed, error line info is not printed for read()
 
        s = zbc_parse_text_init(buf.v);
        if (s) goto exec_err;
@@ -6950,7 +6928,7 @@ static BC_STATUS zbc_vm_execute_FILE(FILE *fp, const char *filename)
 
        G.prs.lex_filename = filename;
        G.prs.lex_input_fp = fp;
-       bc_lex_file();
+       G.err_line = G.prs.lex_line = 1;
 
        do {
                s = zbc_vm_process("");
@@ -7232,7 +7210,6 @@ static BC_STATUS zbc_vm_exec(void)
                // We know that internal library is not buggy,
                // thus error checking is normally disabled.
 # define DEBUG_LIB 0
-               bc_lex_file();
                s = zbc_vm_process(bc_lib);
                if (DEBUG_LIB && s) RETURN_STATUS(s);
        }
index d606637ccf06fc74539d45bd7ca3cf7bed1b7a76..e49b9b288d25ae6ec45dffac4622280dfd3069d3 100644 (file)
@@ -1,9 +1,9 @@
 13
 Hello, World!
-16
-Hello, \[ World!
-16
-Hello, \] World!
+Hello, \[ World!]ZpR
+[Hello, \[ World!]pR
+[Hello, \] World!]ZpR
+[Hello, \] World!
 1
 2
 3