--- a/src/highlighter.c Sun Mar 02 12:47:31 2025 +0100
+++ b/src/highlighter.c Sun Mar 02 16:06:24 2025 +0100
@@ -75,6 +75,107 @@
return 1;
}
+static size_t check_number(const char *str) {
+ /* this function is not precise, but a good over-approximation */
+ size_t i = 0;
+ if (str[0] == '+' || str[0] == '-') {
+ i++;
+ }
+ bool hex = str[i] == '0' && (str[i + 1] == 'x' || str[i + 1] == 'X');
+ bool bin = str[i] == '0' && (str[i + 1] == 'b' || str[i + 1] == 'B');
+ if (hex || bin) {
+ i += 2;
+ }
+ bool flt = false;
+ bool exp = false;
+ bool dot = false;
+ bool digit_seen = false;
+ if (str[i] == '.') {
+ dot = true;
+ flt = true;
+ i++;
+ }
+ char exp_char_low = hex ? 'p' : 'e';
+ char exp_char_up = hex ? 'P' : 'E';
+ while (str[i] != '\0' && str[i] != '\n') {
+ /* ignore grouping char */
+ if (str[i] == '\'') {
+ i++;
+ continue;
+ }
+ /* binary is always integer, nothing else allowed */
+ if (bin) {
+ if (str[i] != '0' && str[i] != '1') {
+ break;
+ } else {
+ i++;
+ digit_seen = true;
+ }
+ } else {
+ /* detect decimal and exponent separators */
+ if ((!dot && str[i] == '.') ||
+ (!exp && digit_seen &&
+ (str[i] == exp_char_low || str[i] == exp_char_up)
+ )
+ ) {
+ if (str[i] == '.') {
+ dot = true;
+ } else {
+ exp = true;
+ /* a sign may directly follow */
+ if (str[i+1] == '+' || str[i+1] == '-') {
+ i++;
+ }
+ }
+ flt = true;
+ i++;
+ continue;
+ }
+ /* check for allowed digits */
+ if ((str[i] >= '0' && str[i] <= '9') || (hex && (
+ (str[i] >= 'a' && str[i] <= 'f')
+ || (str[i] >= 'A' && str[i] <= 'F')
+ ))) {
+ digit_seen = true;
+ i++;
+ } else {
+ break;
+ }
+ }
+ }
+ /* have we seen at least one digit? */
+ if (!digit_seen) return 0;
+
+ /* check if we are already done (over-approximation) */
+ if (!isalpha(str[i])) return i;
+
+ /* check suffixes (must check with decreasing length) */
+ const char *const flt_suffixes[] = {
+ "f128", "bf16", "F128", "BF16",
+ "f16", "f32", "f64", "F16", "F32", "F64",
+ "df", "DF", "dd", "DD", "dl", "DL",
+ "d", "D", "f", "l", "F", "L",
+ };
+ const unsigned flt_suffixes_len = 22;
+ const char *const int_suffixes[] = {
+ "ull", "ULL",
+ "ul", "UL", "ll", "LL", "wb", "WB",
+ "u", "U", "l", "L",
+ };
+ const unsigned int_suffixes_len = 12;
+ const char * const *allowed_suffixes = flt ? flt_suffixes : int_suffixes;
+ const unsigned allowed_suffixes_len = flt ? flt_suffixes_len : int_suffixes_len;
+ for (unsigned j = 0 ; j < allowed_suffixes_len ; j++) {
+ cxstring suffix = cx_str(allowed_suffixes[j]);
+ const char *testee = str+i;
+ if (memcmp(testee, suffix.ptr, suffix.length) == 0) {
+ return i+suffix.length;
+ }
+ }
+ /* no suffix matched */
+ return 0;
+}
+
/* Plaintext Highlighter */
void c2html_plain_highlighter(char const *src, CxBuffer *dest,
@@ -246,6 +347,22 @@
} else {
if (isstring) {
put_htmlescaped(dest, c);
+ } else if (wbuf->size == 0 &&
+ (isdigit(c) || c == '+' || c == '-' || c == '.')
+ ) {
+ /* might be a number */
+ size_t numlen = check_number(src+sp);
+ if (numlen > 0) {
+ start_span("number");
+ put_htmlescapedstr(dest, cx_strn(src+sp, numlen));
+ stop_span;
+ sp += numlen - 1;
+ c = src[sp];
+ continue;
+ } else {
+ /* start a new buffered word */
+ cxBufferPut(wbuf, c);
+ }
} else if (isalnum(c) || c == '_' || c == '#') {
/* buffer the current word */
cxBufferPut(wbuf, c);
@@ -271,8 +388,14 @@
if (closespan) {
stop_span;
}
+
+ /* reset word buffer */
+ wbuf->pos = wbuf->size = 0;
+
+ /* re-test current char */
+ c = src[--sp];
+ continue;
}
- wbuf->pos = wbuf->size = 0; /* reset word buffer */
/* write current character */
put_htmlescaped(dest, c);
@@ -367,6 +490,23 @@
} else {
if (isstring) {
put_htmlescaped(dest, c);
+ } else if (wbuf->size == 0 &&
+ (isdigit(c) || c == '+' || c == '-' || c == '.')
+ ) {
+ /* might be a number */
+ size_t numlen = check_number(src+sp);
+ if (numlen > 0) {
+ cxBufferPutString(dest,
+ "<span class=\"c2html-number\">");
+ put_htmlescapedstr(dest, cx_strn(src+sp, numlen));
+ cxBufferPutString(dest, "</span>");
+ sp += numlen - 1;
+ c = src[sp];
+ continue;
+ } else {
+ /* start a new buffered word */
+ cxBufferPut(wbuf, c);
+ }
} else if (isalnum(c) || c == '_' || c == '@') {
/* buffer the current word */
cxBufferPut(wbuf, c);
@@ -395,8 +535,14 @@
if (closespan) {
cxBufferPutString(dest, "</span>");
}
+
+ /* reset word buffer */
+ wbuf->pos = wbuf->size = 0;
+
+ /* re-test current char */
+ c = src[--sp];
+ continue;
}
- wbuf->pos = wbuf->size = 0; /* reset buffer */
/* write current character */
put_htmlescaped(dest, c);