mirrors/quickjs
1
0
Fork 0
mirror of https://github.com/bellard/quickjs.git synced 2024-11-22 13:48:11 +08:00
Code Issues Packages Projects Releases Wiki Activity

Improve Date.parse

Browse Source 
- accept many more alternative date/time formats
- add test cases in tests/test_builtin.js
- match month and timezone names case insensitively
- accept AM and PM markers
- recognize US timezone names
- skip parenthesized stuff
- fix almost all v8 test cases
This commit is contained in:
Charlie Gordon 2024-03-03 02:59:08 +01:00
parent 8d64731eb8
commit 35b7b3c379
2 changed files with 123 additions and 40 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

139
quickjs.c
View File

@ -49886,23 +49886,28 @@ static BOOL string_skip_char(const uint8_t *sp, int *pp, int c) {
} }
} }
/* skip spaces, update offset */ /* skip spaces, update offset, return next char */
static void string_skip_spaces(const uint8_t *sp, int *pp) { static int string_skip_spaces(const uint8_t *sp, int *pp) {
while (sp[*pp] == ' ') int c;
while ((c = sp[*pp]) == ' ')
*pp += 1; *pp += 1;
return c;
} }
/* skip dashes dots and commas */ /* skip dashes dots and commas */
static void string_skip_separators(const uint8_t *sp, int *pp) { static int string_skip_separators(const uint8_t *sp, int *pp) {
int c; int c;
while ((c = sp[*pp]) == '-' || c == '.' || c == ',') while ((c = sp[*pp]) == '-' || c == '/' || c == '.' || c == ',')
*pp += 1; *pp += 1;
return c;
} }
/* skip non spaces, update offset */ /* skip a word, stop on spaces, digits and separators, update offset */
static void string_skip_non_spaces(const uint8_t *sp, int *pp) { static int string_skip_until(const uint8_t *sp, int *pp, const char *stoplist) {
while (sp[*pp] != '\0' && sp[*pp] != ' ') int c;
while (!strchr(stoplist, c = sp[*pp]))
*pp += 1; *pp += 1;
return c;
} }
/* parse a numeric field (max_digits = 0 -> no maximum) */ /* parse a numeric field (max_digits = 0 -> no maximum) */
@ -49950,26 +49955,37 @@ static BOOL string_get_milliseconds(const uint8_t *sp, int *pp, int *pval) {
return TRUE; return TRUE;
} }
static BOOL string_get_timezone(const uint8_t *sp, int *pp, int *tzp) { static BOOL string_get_timezone(const uint8_t *sp, int *pp, int *tzp, BOOL strict) {
int tz = 0, sgn, hh, mm, p = *pp; int tz = 0, sgn, hh, mm, p = *pp;
sgn = sp[p]; sgn = sp[p++];
if (sgn == '+' || sgn == '-') { if (sgn == '+' || sgn == '-') {
p++; int n = p;
if (!string_get_digits(sp, &p, &hh, 2, 2)) if (!string_get_digits(sp, &p, &hh, 1, 9))
return FALSE; return FALSE;
string_skip_char(sp, &p, ':'); /* optional separator */ n = p - n;
if (!string_get_digits(sp, &p, &mm, 2, 2)) if (strict && n != 2 && n != 4)
return FALSE; return FALSE;
while (n > 4) {
n -= 2;
hh /= 100;
}
if (n > 2) {
mm = hh % 100;
hh = hh / 100;
} else {
mm = 0;
if (string_skip_char(sp, &p, ':') /* optional separator */
&& !string_get_digits(sp, &p, &mm, 2, 2))
return FALSE;
}
if (hh > 23 || mm > 59) if (hh > 23 || mm > 59)
return FALSE; return FALSE;
tz = hh * 60 + mm; tz = hh * 60 + mm;
if (sgn != '+') if (sgn != '+')
tz = -tz; tz = -tz;
} else } else
if (sgn == 'Z') { if (sgn != 'Z') {
p++;
} else {
return FALSE; return FALSE;
} }
*pp = p; *pp = p;
@ -49977,10 +49993,14 @@ static BOOL string_get_timezone(const uint8_t *sp, int *pp, int *tzp) {
return TRUE; return TRUE;
} }
static uint8_t upper_ascii(uint8_t c) {
return c >= 'a' && c <= 'z' ? c - 'a' + 'Z' : c;
}
static BOOL string_match(const uint8_t *sp, int *pp, const char *s) { static BOOL string_match(const uint8_t *sp, int *pp, const char *s) {
int p = *pp; int p = *pp;
while (*s != '\0') { while (*s != '\0') {
if (sp[p] != (uint8_t)*s++) if (upper_ascii(sp[p]) != upper_ascii(*s++))
return FALSE; return FALSE;
p++; p++;
} }
@ -49993,7 +50013,7 @@ static int find_abbrev(const uint8_t *sp, int p, const char *list, int count) {
for (n = 0; n < count; n++) { for (n = 0; n < count; n++) {
for (i = 0;; i++) { for (i = 0;; i++) {
if (sp[p + i] != (uint8_t)month_names[n * 3 + i]) if (upper_ascii(sp[p + i]) != upper_ascii(list[n * 3 + i]))
break; break;
if (i == 2) if (i == 2)
return n; return n;
@ -50056,8 +50076,10 @@ static BOOL js_date_parse_isostring(const uint8_t *sp, int fields[9], BOOL *is_l
*is_local = TRUE; *is_local = TRUE;
if (!string_get_digits(sp, &p, &fields[3], 2, 2) /* hour */ if (!string_get_digits(sp, &p, &fields[3], 2, 2) /* hour */
|| !string_skip_char(sp, &p, ':') || !string_skip_char(sp, &p, ':')
|| !string_get_digits(sp, &p, &fields[4], 2, 2)) /* minute */ || !string_get_digits(sp, &p, &fields[4], 2, 2)) { /* minute */
return FALSE; fields[3] = 100; // reject unconditionally
return TRUE;
}
if (string_skip_char(sp, &p, ':')) { if (string_skip_char(sp, &p, ':')) {
if (!string_get_digits(sp, &p, &fields[5], 2, 2)) /* second */ if (!string_get_digits(sp, &p, &fields[5], 2, 2)) /* second */
return FALSE; return FALSE;
@ -50067,7 +50089,7 @@ static BOOL js_date_parse_isostring(const uint8_t *sp, int fields[9], BOOL *is_l
/* parse the time zone offset if present: [+-]HH:mm or [+-]HHmm */ /* parse the time zone offset if present: [+-]HH:mm or [+-]HHmm */
if (sp[p]) { if (sp[p]) {
*is_local = FALSE; *is_local = FALSE;
if (!string_get_timezone(sp, &p, &fields[8])) if (!string_get_timezone(sp, &p, &fields[8], TRUE))
return FALSE; return FALSE;
} }
/* error if extraneous characters */ /* error if extraneous characters */
@ -50092,11 +50114,10 @@ static BOOL js_date_parse_otherstring(const uint8_t *sp, int fields[9], BOOL *is
} }
*is_local = TRUE; *is_local = TRUE;
while (sp[p] != '\0') { while (string_skip_spaces(sp, &p)) {
string_skip_spaces(sp, &p);
p_start = p; p_start = p;
if ((c = sp[p]) == '+' || c == '-') { if ((c = sp[p]) == '+' || c == '-') {
if (has_time && string_get_timezone(sp, &p, &fields[8])) { if (has_time && string_get_timezone(sp, &p, &fields[8], FALSE)) {
*is_local = FALSE; *is_local = FALSE;
} else { } else {
p++; p++;
@ -50129,7 +50150,7 @@ static BOOL js_date_parse_otherstring(const uint8_t *sp, int fields[9], BOOL *is
has_year = TRUE; has_year = TRUE;
} else } else
if (val < 1 || val > 31) { if (val < 1 || val > 31) {
fields[0] = val + (val < 100) * 1900; fields[0] = val + (val < 100) * 1900 + (val < 50) * 100;
has_year = TRUE; has_year = TRUE;
} else { } else {
if (num_index == 3) if (num_index == 3)
@ -50140,19 +50161,73 @@ static BOOL js_date_parse_otherstring(const uint8_t *sp, int fields[9], BOOL *is
} else } else
if (string_get_month(sp, &p, &fields[1])) { if (string_get_month(sp, &p, &fields[1])) {
has_mon = TRUE; has_mon = TRUE;
string_skip_until(sp, &p, "0123456789 -/(");
} else } else
if (c == 'Z') { if (c == 'Z') {
*is_local = FALSE; *is_local = FALSE;
p++; p++;
continue; continue;
} else } else
if (string_match(sp, &p, "GMT") || string_match(sp, &p, "UTC")) { if (has_time && string_match(sp, &p, "PM")) {
if (fields[3] < 12)
fields[3] += 12;
continue;
} else
if (has_time && string_match(sp, &p, "AM")) {
if (fields[3] == 12)
fields[3] -= 12;
continue;
} else
if (string_match(sp, &p, "GMT")
|| string_match(sp, &p, "UTC")
|| string_match(sp, &p, "UT")) {
*is_local = FALSE; *is_local = FALSE;
continue; continue;
} else { } else
/* skip a word */ if (string_match(sp, &p, "EDT")) {
string_skip_non_spaces(sp, &p); fields[8] = -4 * 60;
*is_local = FALSE;
continue; continue;
} else
if (string_match(sp, &p, "EST") || string_match(sp, &p, "CDT")) {
fields[8] = -5 * 60;
*is_local = FALSE;
continue;
} else
if (string_match(sp, &p, "CST") || string_match(sp, &p, "MDT")) {
fields[8] = -6 * 60;
*is_local = FALSE;
continue;
} else
if (string_match(sp, &p, "MST") || string_match(sp, &p, "PDT")) {
fields[8] = -7 * 60;
*is_local = FALSE;
continue;
} else
if (string_match(sp, &p, "PST")) {
fields[8] = -8 * 60;
*is_local = FALSE;
continue;
} else
if (c == '(') { /* skip parenthesized phrase */
int level = 0;
while ((c = sp[p]) != '\0') {
p++;
level += (c == '(');
level -= (c == ')');
if (!level)
break;
}
if (level > 0)
return FALSE;
} else
if (c == ')') {
return FALSE;
} else {
if (has_year + has_mon + has_time + num_index)
return FALSE;
/* skip a word */
string_skip_until(sp, &p, " -/(");
} }
string_skip_separators(sp, &p); string_skip_separators(sp, &p);
} }
@ -50176,7 +50251,7 @@ static BOOL js_date_parse_otherstring(const uint8_t *sp, int fields[9], BOOL *is
fields[2] = num[1]; fields[2] = num[1];
} else } else
if (has_mon) { if (has_mon) {
fields[0] = num[1] + (num[1] < 100) * 1900; fields[0] = num[1] + (num[1] < 100) * 1900 + (num[1] < 50) * 100;
fields[2] = num[0]; fields[2] = num[0];
} else { } else {
fields[1] = num[0]; fields[1] = num[0];
@ -50184,7 +50259,7 @@ static BOOL js_date_parse_otherstring(const uint8_t *sp, int fields[9], BOOL *is
} }
break; break;
case 3: case 3:
fields[0] = num[2] + (num[2] < 100) * 1900; fields[0] = num[2] + (num[2] < 100) * 1900 + (num[2] < 50) * 100;
fields[1] = num[0]; fields[1] = num[0];
fields[2] = num[1]; fields[2] = num[1];
break; break;

24
tests/test_builtin.js
View File

@ -600,6 +600,14 @@ function test_date()
assert((new Date("2020-01-01T01:01:01.9999Z")).toISOString(), assert((new Date("2020-01-01T01:01:01.9999Z")).toISOString(),
"2020-01-01T01:01:01.999Z"); "2020-01-01T01:01:01.999Z");
assert(Date.UTC(2017), 1483228800000);
assert(Date.UTC(2017, 9), 1506816000000);
assert(Date.UTC(2017, 9, 22), 1508630400000);
assert(Date.UTC(2017, 9, 22, 18), 1508695200000);
assert(Date.UTC(2017, 9, 22, 18, 10), 1508695800000);
assert(Date.UTC(2017, 9, 22, 18, 10, 11), 1508695811000);
assert(Date.UTC(2017, 9, 22, 18, 10, 11, 91), 1508695811091);
assert(Date.UTC(NaN), NaN); assert(Date.UTC(NaN), NaN);
assert(Date.UTC(2017, NaN), NaN); assert(Date.UTC(2017, NaN), NaN);
assert(Date.UTC(2017, 9, NaN), NaN); assert(Date.UTC(2017, 9, NaN), NaN);
@ -609,14 +617,14 @@ function test_date()
assert(Date.UTC(2017, 9, 22, 18, 10, 11, NaN), NaN); assert(Date.UTC(2017, 9, 22, 18, 10, 11, NaN), NaN);
assert(Date.UTC(2017, 9, 22, 18, 10, 11, 91, NaN), 1508695811091); assert(Date.UTC(2017, 9, 22, 18, 10, 11, 91, NaN), 1508695811091);
assert(Date.UTC(2017), 1483228800000); // TODO: Fix rounding errors on Windows/Cygwin.
assert(Date.UTC(2017, 9), 1506816000000); if (!(typeof os !== 'undefined' && ['win32', 'cygwin'].includes(os.platform))) {
assert(Date.UTC(2017, 9, 22), 1508630400000); // from test262/test/built-ins/Date/UTC/fp-evaluation-order.js
assert(Date.UTC(2017, 9, 22, 18), 1508695200000); assert(Date.UTC(1970, 0, 1, 80063993375, 29, 1, -288230376151711740), 29312,
assert(Date.UTC(2017, 9, 22, 18, 10), 1508695800000); 'order of operations / precision in MakeTime');
assert(Date.UTC(2017, 9, 22, 18, 10, 11), 1508695811000); assert(Date.UTC(1970, 0, 213503982336, 0, 0, 0, -18446744073709552000), 34447360,
assert(Date.UTC(2017, 9, 22, 18, 10, 11, 91), 1508695811091); 'precision in MakeDate');
}
//assert(Date.UTC(2017 - 1e9, 9 + 12e9), 1506816000000); // node fails this //assert(Date.UTC(2017 - 1e9, 9 + 12e9), 1506816000000); // node fails this
assert(Date.UTC(2017, 9, 22 - 1e10, 18 + 24e10), 1508695200000); assert(Date.UTC(2017, 9, 22 - 1e10, 18 + 24e10), 1508695200000);
assert(Date.UTC(2017, 9, 22, 18 - 1e10, 10 + 60e10), 1508695800000); assert(Date.UTC(2017, 9, 22, 18 - 1e10, 10 + 60e10), 1508695800000);