1 files changed, 354 insertions, 28 deletions

diff --git a/src/string.c b/src/string.c
index 354a409..ae19271 100644
--- a/src/string.c
+++ b/src/string.c
@@ -8,6 +8,8 @@
  *	either version 2.1 or (at your option) any later version.
  */
 #include <errno.h>
+#include <limits.h>
+#include <math.h>
 #include <stdbool.h>
 #include <stddef.h>
 #include <stdint.h>
@@ -25,6 +27,7 @@
  * 			all others pass through
  */
 enum HX_quote_selector {
+	HXQUOTE_ALWAYS,
 	HXQUOTE_ACCEPT,
 	HXQUOTE_REJECT,
 };
@@ -40,11 +43,6 @@ struct HX_quote_rule {
 
 static const char HX_hexenc[16] = "0123456789ABCDEF";
 
-static __inline__ unsigned int min_uint(unsigned int a, unsigned int b)
-{
-	return (a < b) ? a : b;
-}
-
 EXPORT_SYMBOL char *HX_basename(const char *s)
 {
 	const char *p;
@@ -214,6 +212,8 @@ EXPORT_SYMBOL char **HX_split(const char *str, const char *delim,
 		*cp = max;
 
 	ret = malloc(sizeof(char *) * (*cp + 1));
+	if (ret == nullptr)
+		return nullptr;
 	ret[*cp] = NULL;
 
 	{
@@ -381,20 +381,48 @@ EXPORT_SYMBOL char *HX_stpltrim(const char *p)
 	return const_cast1(char *, p);
 }
 
+/**
+ * Helper for substr() function for dealing with negative off/len values
+ * @z:		total length of string
+ * @offset:	n>=0 specifies offset from the start,
+ * 		n<0 specifies offset from the end
+ * @len:	"length"; n>=0 specifies length to copy (from @offset),
+ * 		n<0 specifies the byte relative to the end at which to stop
+ *
+ * @abstart:	(result) absolute start
+ * @retval:	(result) absolute length to copy (from *@abstart)
+ */
+size_t HX_substr_helper(size_t z, long offset, long len, size_t *start)
+{
+	if (offset >= 0)
+		*start = offset;
+	else if (offset == LONG_MIN)
+		*start = z + -static_cast(size_t, LONG_MIN);
+	else
+		*start = z >= static_cast(unsigned long, -offset) ? z + offset : z;
+
+	size_t end;
+	if (len >= 0)
+		end = *start < SIZE_MAX - len ? *start + len : SIZE_MAX;
+	else if (len == LONG_MIN)
+		end = z + -static_cast(unsigned long, LONG_MIN);
+	else
+		end = z >= static_cast(unsigned long, -len) ? z + len : 0;
+	if (end > z)
+		end = z;
+	return end > *start ? end - *start : 0;
+}
+
 /* supports negative offsets like scripting languages */
 EXPORT_SYMBOL char *HX_strmid(const char *expr, long offset, long length)
 {
-	char *buffer;
-
-	if (offset < 0)
-		offset = strlen(expr) + offset;
-	if (length < 0)
-		length = strlen(expr) - offset + length;
-	if ((buffer = malloc(length + 1)) == NULL)
+	size_t start = 0, tocopy = HX_substr_helper(strlen(expr), offset, length, &start);
+	char *buffer = malloc(tocopy + 1);
+	if (buffer == nullptr)
 		return NULL;
-
-	expr += offset;
-	return HX_strlcpy(buffer, expr, length + 1);
+	memcpy(buffer, &expr[start], tocopy);
+	buffer[tocopy] = '\0';
+	return buffer;
 }
 
 EXPORT_SYMBOL char *HX_strndup(const char *src, size_t size)
@@ -605,14 +633,16 @@ HX_quote_sqlbackslash(char *dest, const char *src, const char *trm)
  * Encode @src into BASE-64 according to RFC 4648 and write result to @dest,
  * which must be of appropriate size, plus one for a trailing NUL.
  */
-static char *HX_quote_base64(char *d, const char *s)
+static char *HX_quote_base64(char *d, const char *s, char x1, char x2)
 {
-	static const char a[] =
+	char a[] =
 		"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
-		"abcdefghijklmnopqrstuvwxyz0123456789+/";
+		"abcdefghijklmnopqrstuvwxyz0123456789??";
 	size_t len = strlen(s);
 	char *ret = d;
 
+	a[62] = x1;
+	a[63] = x2;
 	while (len > 0) {
 		if (len >= 3) {
 			len -= 3;
@@ -761,6 +791,8 @@ static size_t HX_quoted_size(const char *s, unsigned int type)
 	case HXQUOTE_LDAPRDN:
 		return HX_qsize_bsr(s, HX_quote_rules[type].chars, 2);
 	case HXQUOTE_BASE64:
+	case HXQUOTE_BASE64URL:
+	case HXQUOTE_BASE64IMAP:
 		return (strlen(s) + 2) / 3 * 4;
 	case HXQUOTE_URIENC:
 		return HX_qsize_bsa(s, HX_quote_rules[type].chars, 2);
@@ -772,7 +804,7 @@ static size_t HX_quoted_size(const char *s, unsigned int type)
 EXPORT_SYMBOL char *HX_strquote(const char *src, unsigned int type,
     char **free_me)
 {
-	const struct HX_quote_rule *rule;
+	const struct HX_quote_rule *rule = nullptr;
 	bool do_quote;
 	char *tmp;
 
@@ -781,15 +813,19 @@ EXPORT_SYMBOL char *HX_strquote(const char *src, unsigned int type,
 		return NULL;
 	}
 	/* If quote_chars is NULL, it is clear all chars are to be encoded. */
-	rule = &HX_quote_rules[type];
-	if (type >= ARRAY_SIZE(HX_quote_rules) || rule->chars == NULL)
+	if (type >= ARRAY_SIZE(HX_quote_rules)) {
 		do_quote = true;
-	else if (rule->selector == HXQUOTE_REJECT)
-		do_quote = strpbrk(src, rule->chars) != NULL;
-	else if (rule->selector == HXQUOTE_ACCEPT)
-		do_quote = HX_strchr2(src, rule->chars) != NULL;
-	else
-		do_quote = false;
+	} else {
+		rule = &HX_quote_rules[type];
+		if (rule->selector == HXQUOTE_ALWAYS)
+			do_quote = true;
+		else if (rule->selector == HXQUOTE_REJECT)
+			do_quote = strpbrk(src, rule->chars) != NULL;
+		else if (rule->selector == HXQUOTE_ACCEPT)
+			do_quote = HX_strchr2(src, rule->chars) != NULL;
+		else
+			do_quote = false;
+	}
 	/*
 	 * free_me == NULL implies that we always allocate, even if
 	 * there is nothing to quote.
@@ -819,7 +855,11 @@ EXPORT_SYMBOL char *HX_strquote(const char *src, unsigned int type,
 	case HXQUOTE_LDAPRDN:
 		return HX_quote_ldap(*free_me, src, rule->chars);
 	case HXQUOTE_BASE64:
-		return HX_quote_base64(*free_me, src);
+		return HX_quote_base64(*free_me, src, '+', '/');
+	case HXQUOTE_BASE64URL:
+		return HX_quote_base64(*free_me, src, '-', '_');
+	case HXQUOTE_BASE64IMAP:
+		return HX_quote_base64(*free_me, src, '+', ',');
 	case HXQUOTE_URIENC:
 		return HX_quote_urlenc(*free_me, src);
 	case HXQUOTE_SQLSQUOTE:
@@ -836,3 +876,289 @@ EXPORT_SYMBOL char *HX_strupper(char *orig)
 		*expr = HX_toupper(*expr);
 	return orig;
 }
+
+EXPORT_SYMBOL char *HX_unit_size(char *buf, size_t bufsize,
+    unsigned long long size, unsigned int divisor, unsigned int cutoff)
+{
+	static const char unit_names[] = "\0kMGTPEZYRQ";
+	unsigned int unit_idx = 0;
+	if (divisor == 0)
+		divisor = 1000;
+	if (cutoff == 0) {
+		cutoff = 10000;
+		if (cutoff < divisor)
+			cutoff = divisor;
+	}
+	while (unit_idx < ARRAY_SIZE(unit_names) - 1 && size >= cutoff) {
+		++unit_idx;
+		size /= divisor;
+	}
+	snprintf(buf, bufsize, "%llu%.1s", size, &unit_names[unit_idx]);
+	return buf;
+}
+
+static inline unsigned long long p_90(unsigned long long x)
+{
+	/* Perform x*9/10, but without the risk of overflow. */
+	return x - x / 10 - !!(x % 10);
+}
+
+EXPORT_SYMBOL char *HX_unit_size_cu(char *buf, size_t bufsize,
+    unsigned long long orig_size, unsigned int divisor)
+{
+	/* No floating point. Take that, coreutils! */
+	static const char unit_names[] = "\0kMGTPEZYRQ";
+	unsigned int unit_idx = 0, last_rem = 0;
+	unsigned long long size = orig_size, gpow = 1, grand_rem;
+	if (divisor == 0)
+		divisor = 1000;
+
+	while (unit_idx < ARRAY_SIZE(unit_names) - 1 && size >= divisor) {
+		++unit_idx;
+		last_rem = size % divisor;
+		size /= divisor;
+		gpow *= divisor;
+	}
+	if (unit_idx == 0) {
+		snprintf(buf, bufsize, "%llu%.1s", size, &unit_names[unit_idx]);
+		return buf;
+	}
+	grand_rem = orig_size - size * gpow;
+	if (grand_rem != 0) {
+		if (grand_rem > p_90(gpow)) {
+			++size;
+			last_rem = 0;
+		} else {
+			last_rem *= 10;
+			last_rem /= divisor;
+			++last_rem;
+			if (last_rem == 10 || (size >= 10 && last_rem > 0)) {
+				++size;
+				last_rem = 0;
+			}
+		}
+		if (unit_idx < ARRAY_SIZE(unit_names) - 1 && size == divisor) {
+			/* ++size from above may brought size to @divisor again */
+			++unit_idx;
+			size /= divisor;
+		}
+	}
+	if (size >= 10 && last_rem == 0)
+		snprintf(buf, bufsize, "%llu%.1s", size, &unit_names[unit_idx]);
+	else
+		snprintf(buf, bufsize, "%llu.%01u%.1s", size, last_rem, &unit_names[unit_idx]);
+	return buf;
+}
+
+static unsigned int suffix_power(char u)
+{
+	switch (HX_toupper(u)) {
+	case 'K': return 1;
+	case 'M': return 2;
+	case 'G': return 3;
+	case 'T': return 4;
+	case 'P': return 5;
+	case 'E': return 6;
+	case 'Z': return 7;
+	case 'Y': return 8;
+	case 'R': return 9;
+	case 'Q': return 10;
+	default: return 0;
+	}
+}
+
+EXPORT_SYMBOL double HX_strtod_unit(const char *s, char **out_end, unsigned int exponent)
+{
+	char *end;
+	double q;
+
+	while (HX_isspace(*s))
+		++s;
+	q = strtod(s, &end);
+	if (exponent == 0)
+		exponent = 1000;
+	if (end == s) {
+		if (out_end != nullptr)
+			*out_end = end;
+		return q;
+	}
+	while (HX_isspace(*end))
+		++end;
+	unsigned int pwr = suffix_power(*end);
+	if (pwr == 0) {
+		if (out_end != nullptr)
+			*out_end = const_cast(char *, end);
+		return q;
+	}
+	if (out_end != nullptr)
+		*out_end = const_cast(char *, end + 1);
+	return q * pow(exponent, pwr);
+}
+
+EXPORT_SYMBOL unsigned long long HX_strtoull_unit(const char *s,
+    char **out_end, unsigned int exponent)
+{
+	char *end;
+	unsigned long long ipart;
+	unsigned int pwr = 0;
+
+	while (HX_isspace(*s))
+		++s;
+	ipart = strtoull(s, &end, 10);
+	if (*end == '.') {
+		double q = HX_strtod_unit(s, out_end, exponent);
+		bool lo_ok = q >= nextafter(-static_cast(double, ULLONG_MAX), 0);
+		bool hi_ok = q <= nextafter(static_cast(double, ULLONG_MAX), 0);
+		if (!hi_ok || !lo_ok)
+			return ULLONG_MAX;
+		return q;
+	}
+	if (exponent == 0)
+		exponent = 1000;
+	while (HX_isspace(*end))
+		++end;
+	pwr = suffix_power(*end);
+	if (pwr == 0) {
+		if (out_end != nullptr)
+			*out_end = end;
+		return ipart;
+	}
+	if (out_end != nullptr)
+		*out_end = const_cast(char *, end + 1);
+	while (pwr-- > 0) {
+		if (ipart >= ULLONG_MAX / exponent) {
+			errno = ERANGE;
+			return ULLONG_MAX;
+		}
+		ipart *= exponent;
+	}
+	return ipart;
+}
+
+#define SECONDS_PER_YEAR 31557600
+#define SECONDS_PER_MONTH 2629800
+
+static const struct {
+	const char name[8];
+	unsigned int len;
+	uint32_t mult;
+} time_multiplier[] = {
+	{"seconds", 7, 1},
+	{"second",  6, 1},
+	{"sec",     3, 1},
+	{"s",       1, 1},
+	{"minutes", 7, 60},
+	{"minute",  6, 60},
+	{"min",     3, 60},
+	{"hours",   5, 3600},
+	{"hour",    4, 3600},
+	{"h",       1, 3600},
+	{"days",    4, 86400},
+	{"day",     3, 86400},
+	{"d",       1, 86400},
+	{"weeks",   5, 604800},
+	{"week",    4, 604800},
+	{"months",  6, SECONDS_PER_MONTH},
+	{"month",   5, SECONDS_PER_MONTH},
+	{"years",   5, SECONDS_PER_YEAR},
+	{"year",    4, SECONDS_PER_YEAR},
+	{"y",       1, SECONDS_PER_YEAR},
+};
+
+EXPORT_SYMBOL unsigned long long HX_strtoull_sec(const char *s, char **out_end)
+{
+	unsigned long long seconds = 0;
+
+	while (*s != '\0') {
+		while (HX_isspace(*s))
+			++s;
+		if (*s == '-') {
+			break;
+		}
+		char *end = nullptr;
+		unsigned long long num = strtoull(s, &end, 10);
+		if (end == s)
+			break;
+		s = end;
+		while (HX_isspace(*s))
+			++s;
+		if (!HX_isalpha(*s)) {
+			seconds += num;
+			continue;
+		}
+		unsigned int i;
+		for (i = 0; i < ARRAY_SIZE(time_multiplier); ++i)
+			if (strncmp(s, time_multiplier[i].name,
+			    time_multiplier[i].len) == 0 &&
+			    !HX_isalpha(s[time_multiplier[i].len]))
+				break;
+		if (i == ARRAY_SIZE(time_multiplier))
+			break;
+		seconds += num * time_multiplier[i].mult;
+		s += time_multiplier[i].len;
+	}
+	if (out_end != nullptr)
+		*out_end = const_cast(char *, s);
+	return seconds;
+}
+
+EXPORT_SYMBOL char *HX_unit_seconds(char *out, size_t outsize,
+    unsigned long long secs, unsigned int flags)
+{
+	unsigned long years = 0, months = 0, weeks = 0, days, hours, mins;
+	char buf[HXSIZEOF_Z64+4];
+	if (flags & HXUNIT_YEARS) {
+		years = secs / SECONDS_PER_YEAR;
+		secs %= SECONDS_PER_YEAR;
+	}
+	if (flags & HXUNIT_MONTHS) {
+		months = secs / SECONDS_PER_MONTH;
+		secs %= SECONDS_PER_MONTH;
+	}
+	if (flags & HXUNIT_WEEKS) {
+		weeks = secs / 604800;
+		secs %= 604800;
+	}
+	days = secs / 86400;
+	secs %= 86400;
+	hours = secs / 3600;
+	secs %= 3600;
+	mins = secs / 60;
+	secs %= 60;
+	*out = '\0';
+	if (years > 0) {
+		snprintf(buf, ARRAY_SIZE(buf), "%luy", years);
+		HX_strlcat(out, buf, outsize);
+	}
+	if (months == 1) {
+		HX_strlcat(out, "1month", outsize);
+	} else if (months > 0) {
+		snprintf(buf, ARRAY_SIZE(buf), "%lumonths", months);
+		HX_strlcat(out, buf, outsize);
+	}
+	if (weeks == 1) {
+		HX_strlcat(out, "1week", outsize);
+	} else if (weeks > 0) {
+		snprintf(buf, ARRAY_SIZE(buf), "%luweeks", weeks);
+		HX_strlcat(out, buf, outsize);
+	}
+	if (days > 0) {
+		snprintf(buf, ARRAY_SIZE(buf), "%lud", days);
+		HX_strlcat(out, buf, outsize);
+	}
+	if (hours > 0) {
+		snprintf(buf, ARRAY_SIZE(buf), "%luh", hours);
+		HX_strlcat(out, buf, outsize);
+	}
+	if (mins > 0) {
+		snprintf(buf, ARRAY_SIZE(buf), "%lumin", mins);
+		HX_strlcat(out, buf, outsize);
+	}
+	if (secs > 0 ||
+	    (years == 0 && months == 0 && weeks == 0 &&
+	    days == 0 && hours == 0 && mins == 0)) {
+		snprintf(buf, ARRAY_SIZE(buf), "%llus", secs);
+		HX_strlcat(out, buf, outsize);
+	}
+	return out;
+}