greenhouse

greenhouse / herbarium / herbarium.c2037 lines · 44218B · raw

/* imports */
#include <ctype.h>
#include <dirent.h>
#include <errno.h>
#include <fcntl.h>
#include <iconv.h>
#include <limits.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <strings.h>
#include <time.h>
#include <unistd.h>
#include <sys/stat.h>

/* macros */
#define LENGTH(X) (sizeof(X) / sizeof((X)[0]))
#define HOUR      3600L
#define DAY       (24 * HOUR)
#define MAXTIME   253402300799LL
#define FIELDLEN  998
#define MAXREFS   16
#define NONE      ((size_t)-1)
#define REPL      "\xef\xbf\xbd"

/* enums */
typedef struct {
	char *s;
	size_t n, cap;
} Buf;

typedef struct {
	char *file;
	char *id;
	char *irt;
	char *refs[MAXREFS];
	size_t nrefs;
	char *subject;
	char *name;
	char *addr;
	long long mtime;
	long long date;
	int tz;
	size_t parent;
	size_t root;
	size_t seen;
} Msg;

typedef struct {
	const char *body;
	size_t len;
	char type[64];
	char cte[32];
	char charset[64];
	char boundary[128];
	char name[512];
} Part;

typedef struct {
	const char *name;
	const char *url;
} Service;

typedef struct {
	size_t root;
	size_t n;
	long long last;
} Thread;

/* function declarations */
static void attachment(const Part *p);
static void b64(const char *s, size_t n, Buf *out);
static const char *boundary(const char *s, const char *end, const char *b,
                            size_t blen);
static int bprintf(char *buf, size_t len, const char *fmt, ...);
static void bufadd(Buf *b, const char *s, size_t n);
static void bufc(Buf *b, int c);
static size_t byid(const char *id);
static char *clean(const char *s, size_t n);
static int cmpdate(const void *a, const void *b);
static int cmpfirst(const void *a, const void *b);
static int cmpid(const void *a, const void *b);
static int cmpmtime(const void *a, const void *b);
static int cmpthread(const void *a, const void *b);
static void decode(const Part *p, Buf *out);
static void decode2047(const char *s, Buf *out);
static void die(const char *msg);
static size_t digits(const char **s, size_t max, long *v);
static const char *encword(const char *p, Buf *out);
static const char *envor(const char *name, const char *fallback);
static char *field(const char *v);
static void freemsg(Msg *m);
static void freemsgs(void);
static void gzblock(int last);
static void gzput(const char *s, size_t n);
static size_t hdrend(const char *s, size_t n, size_t *body);
static char *hdrget(const char *h, size_t n, const char *name);
static int hex(int c);
static void httpheader(int status, const char *type);
static void indexpage(void);
static void indexrow(const char *name);
static int islist(const char *name);
static int isstat(const char *s, const char *end);
static const char *lineclass(const char *l, size_t len, const char *next,
                             const char *end, int *patch);
static void listheader(int addrs);
static void listpage(void);
static int loadmsg(Msg *m);
static void mboxpage(size_t t);
static void msgurl(const char *id, const char *page);
static void multipart(const Part *p, int depth, size_t m);
static char *nextid(const char **sp);
static void notfound(void);
static FILE *openmsg(const char *file);
static void pagebegin(int status, const char *pagetitle);
static void pageend(void);
static int param(const char *v, const char *attr, char *buf, size_t len);
static int parsedate(const char *s, long long *t, int *tz);
static void parsefrom(const char *v, Msg *m);
static void parsepart(const char *s, size_t n, Part *p);
static void printage(long long t);
static void printbody(const char *s, size_t n);
static void printmsg(size_t i, int sel);
static void printtime(long long t, int tz, const char *fmt);
static void qp(const char *s, size_t n, int q, Buf *out);
static void rawpage(size_t t);
static int readfile(const char *file, size_t max, Buf *b, int hdronly);
static void readline(const char *name, const char *file, char *buf,
                     size_t len);
static void renderpart(const char *s, size_t n, int depth, size_t m);
static void route(char *path);
static void sanitize(const char *s, size_t n, Buf *out, int multi);
static void scan(const char *name);
static const char *skipcfws(const char *s);
static void threadmsgs(void);
static void threadpage(size_t t);
static int tou8(const char *cs, const char *s, size_t n, Buf *out);
static void urlencode(const char *s, const char *safe);
static int validid(const char *s, size_t n);
static int validlist(const char *s);
static void xmlencode(const char *s);
static void xmlencoden(const char *s, size_t len);
static char *xstrndup(const char *s, size_t n);

/* variables */
static const char pathsafe[] = "-._~!$()*+,;=:@";
static const char mailsafe[] = "-._~@";
static unsigned char gzbuf[65535];
static size_t gzlen;
static unsigned long gzcrc, gzsize;
static const char *base;
static const char *list;
static Msg *msgs;
static size_t nmsgs;
static size_t *ids;
static size_t left;
static int cut;
static long ofs;

#include "config.h"

/* function implementations */
void
attachment(const Part *p)
{
	char *name;
	size_t i, n = p->len;

	if (!strcmp(p->cte, "base64")) {
		for (i = n = 0; i < p->len; i++)
			if (isalnum((unsigned char)p->body[i])
			|| p->body[i] == '+' || p->body[i] == '/')
				n++;
		n = n * 3 / 4;
	}
	name = field(*p->name ? p->name : NULL);
	if (strlen(name) + 64 > left) {
		cut = 1;
		free(name);
		return;
	}
	left -= strlen(name) + 64;
	fputs("<p class=\"att\">attachment: ", stdout);
	xmlencode(*name ? name : "(unnamed)");
	fputs(" (", stdout);
	xmlencode(p->type);
	printf(", %zuB)</p>\n", n);
	free(name);
}

void
b64(const char *s, size_t n, Buf *out)
{
	unsigned long v = 0;
	size_t i;
	int c, bits = 0;

	for (i = 0; i < n; i++) {
		c = (unsigned char)s[i];
		if (c >= 'A' && c <= 'Z')
			c -= 'A';
		else if (c >= 'a' && c <= 'z')
			c -= 'a' - 26;
		else if (c >= '0' && c <= '9')
			c -= '0' - 52;
		else if (c == '+')
			c = 62;
		else if (c == '/')
			c = 63;
		else {
			if (c == '=')
				bits = 0;
			continue;
		}
		v = (v << 6 | (unsigned long)c) & 0xffffff;
		if ((bits += 6) >= 8) {
			bits -= 8;
			bufc(out, (int)(v >> bits & 0xff));
		}
	}
}

const char *
boundary(const char *s, const char *end, const char *b, size_t blen)
{
	const char *p;

	while (s < end) {
		if ((size_t)(end - s) >= blen + 2 && s[0] == '-' && s[1] == '-'
		&& !memcmp(s + 2, b, blen)
		&& (s + blen + 2 == end
		|| (s[blen + 2] && strchr("\r\n \t-", s[blen + 2]))))
			return s;
		if (!(p = memchr(s, '\n', end - s)))
			break;
		s = p + 1;
	}
	return NULL;
}

int
bprintf(char *buf, size_t len, const char *fmt, ...)
{
	va_list ap;
	int n;

	va_start(ap, fmt);
	n = vsnprintf(buf, len, fmt, ap);
	va_end(ap);
	if (n < 0 || (size_t)n >= len)
		return -1;
	return 0;
}

void
bufadd(Buf *b, const char *s, size_t n)
{
	size_t cap;
	char *p;

	if (b->n + n + 1 > b->cap) {
		for (cap = b->cap ? b->cap : 256; cap < b->n + n + 1; cap *= 2)
			;
		if (!(p = realloc(b->s, cap)))
			die("out of memory");
		b->s = p;
		b->cap = cap;
	}
	if (n)
		memcpy(b->s + b->n, s, n);
	b->n += n;
	b->s[b->n] = '\0';
}

void
bufc(Buf *b, int c)
{
	char ch = (char)c;

	bufadd(b, &ch, 1);
}

size_t
byid(const char *id)
{
	size_t lo = 0, hi = nmsgs, mid;

	while (lo < hi) {
		mid = lo + (hi - lo) / 2;
		if (strcmp(msgs[ids[mid]].id, id) < 0)
			lo = mid + 1;
		else
			hi = mid;
	}
	if (lo < nmsgs && !strcmp(msgs[ids[lo]].id, id))
		return ids[lo];
	return NONE;
}

char *
clean(const char *s, size_t n)
{
	Buf b = { 0 };
	char *r;
	size_t i;

	sanitize(s, n, &b, 0);
	bufadd(&b, "", 0);
	for (i = 0; b.s[i] == ' '; i++)
		;
	for (n = b.n; n > i && b.s[n - 1] == ' '; n--)
		;
	if (n - i > FIELDLEN) {
		for (n = i + FIELDLEN; n > i
		     && ((unsigned char)b.s[n] & 0xc0) == 0x80; n--)
			;
	}
	r = xstrndup(b.s + i, n - i);
	free(b.s);
	return r;
}

int
cmpdate(const void *a, const void *b)
{
	const Msg *x = a, *y = b;

	if (x->date != y->date)
		return x->date < y->date ? -1 : 1;
	if (x->mtime != y->mtime)
		return x->mtime < y->mtime ? -1 : 1;
	return strcmp(x->file, y->file);
}

int
cmpfirst(const void *a, const void *b)
{
	const Msg *x = a, *y = b;
	int c;

	if ((c = strcmp(x->id, y->id)))
		return c;
	if (x->mtime != y->mtime)
		return x->mtime < y->mtime ? -1 : 1;
	return strcmp(x->file, y->file);
}

int
cmpid(const void *a, const void *b)
{
	size_t x = *(const size_t *)a, y = *(const size_t *)b;
	int c;

	if ((c = strcmp(msgs[x].id, msgs[y].id)))
		return c;
	return x < y ? -1 : x > y;
}

int
cmpmtime(const void *a, const void *b)
{
	const Msg *x = a, *y = b;

	if (x->mtime != y->mtime)
		return x->mtime > y->mtime ? -1 : 1;
	return strcmp(x->file, y->file);
}

int
cmpthread(const void *a, const void *b)
{
	const Thread *x = a, *y = b;

	if (x->last != y->last)
		return x->last > y->last ? -1 : 1;
	return strcmp(msgs[x->root].id, msgs[y->root].id);
}

void
decode(const Part *p, Buf *out)
{
	if (!strcmp(p->cte, "base64"))
		b64(p->body, p->len, out);
	else if (!strcmp(p->cte, "quoted-printable"))
		qp(p->body, p->len, 0, out);
	else
		bufadd(out, p->body, p->len);
}

void
decode2047(const char *s, Buf *out)
{
	const char *q, *r;
	int enc = 0;

	while (*s) {
		if (s[0] == '=' && s[1] == '?' && (q = encword(s, out))) {
			s = q;
			enc = 1;
			continue;
		}
		if (enc && isspace((unsigned char)*s)) {
			for (q = s; isspace((unsigned char)*q); q++)
				;
			if (q[0] == '=' && q[1] == '?'
			&& (r = encword(q, out))) {
				s = r;
				continue;
			}
			bufadd(out, s, q - s);
			s = q;
			enc = 0;
			continue;
		}
		bufc(out, *s++);
		enc = 0;
	}
}

void
die(const char *msg)
{
	fprintf(stderr, "herbarium: %s\n", msg);
	exit(1);
}

size_t
digits(const char **s, size_t max, long *v)
{
	size_t n;

	*v = 0;
	for (n = 0; isdigit((unsigned char)(*s)[n]); n++)
		if (n < max)
			*v = *v * 10 + (*s)[n] - '0';
	*s += n;
	return n > max ? 0 : n;
}

const char *
encword(const char *p, Buf *out)
{
	Buf raw = { 0 };
	char cs[64];
	const char *q, *e;
	size_t n;
	int enc, r;

	q = p + 2;
	for (n = 0; q[n] != '?'; n++)
		if (!q[n] || n + 1 >= sizeof(cs)
		|| isspace((unsigned char)q[n]))
			return NULL;
	memcpy(cs, q, n);
	cs[n] = '\0';
	q += n;
	enc = tolower((unsigned char)q[1]);
	if ((enc != 'b' && enc != 'q') || q[2] != '?')
		return NULL;
	q += 3;
	for (e = q; !(e[0] == '?' && e[1] == '='); e++)
		if (!*e || isspace((unsigned char)*e))
			return NULL;
	if (enc == 'b')
		b64(q, e - q, &raw);
	else
		qp(q, e - q, 1, &raw);
	r = tou8(cs, raw.s, raw.n, out);
	free(raw.s);
	return r < 0 ? NULL : e + 2;
}

const char *
envor(const char *name, const char *fallback)
{
	const char *v;

	v = getenv(name);
	return v && *v ? v : fallback;
}

char *
field(const char *v)
{
	Buf b = { 0 };
	char *s, *r;
	size_t n;

	if (!v)
		return xstrndup("", 0);
	n = strlen(v);
	s = xstrndup(v, n > 4 * FIELDLEN ? 4 * FIELDLEN : n);
	decode2047(s, &b);
	r = clean(b.s, b.n);
	free(s);
	free(b.s);
	return r;
}

void
freemsg(Msg *m)
{
	size_t i;

	free(m->file);
	free(m->id);
	free(m->irt);
	for (i = 0; i < m->nrefs; i++)
		free(m->refs[i]);
	free(m->subject);
	free(m->name);
	free(m->addr);
}

void
freemsgs(void)
{
	size_t i;

	for (i = 0; i < nmsgs; i++)
		freemsg(&msgs[i]);
	free(msgs);
	free(ids);
	msgs = NULL;
	ids = NULL;
	nmsgs = 0;
}

void
gzblock(int last)
{
	unsigned char h[5];

	h[0] = last ? 1 : 0;
	h[1] = gzlen & 0xff;
	h[2] = gzlen >> 8 & 0xff;
	h[3] = ~gzlen & 0xff;
	h[4] = ~gzlen >> 8 & 0xff;
	fwrite(h, 1, sizeof(h), stdout);
	fwrite(gzbuf, 1, gzlen, stdout);
	gzlen = 0;
}

void
gzput(const char *s, size_t n)
{
	static unsigned long tab[256];
	unsigned long c;
	size_t i;
	int k;

	if (!tab[1]) {
		for (i = 0; i < LENGTH(tab); i++) {
			for (c = i, k = 0; k < 8; k++)
				c = c & 1 ? 0xedb88320UL ^ c >> 1 : c >> 1;
			tab[i] = c;
		}
	}
	for (i = 0; i < n; i++) {
		gzcrc = tab[(gzcrc ^ (unsigned char)s[i]) & 0xff] ^ gzcrc >> 8;
		gzbuf[gzlen++] = s[i];
		if (gzlen == sizeof(gzbuf))
			gzblock(0);
	}
	gzsize += n;
}

size_t
hdrend(const char *s, size_t n, size_t *body)
{
	const char *p;
	size_t pos = 0;

	while (pos < n) {
		if (s[pos] == '\n') {
			*body = pos + 1;
			return pos;
		}
		if (s[pos] == '\r' && pos + 1 < n && s[pos + 1] == '\n') {
			*body = pos + 2;
			return pos;
		}
		if (!(p = memchr(s + pos, '\n', n - pos)))
			break;
		pos = p - s + 1;
	}
	*body = n;
	return n;
}

char *
hdrget(const char *h, size_t n, const char *name)
{
	Buf b = { 0 };
	const char *p;
	size_t pos, k, len = strlen(name);
	char *r;

	for (pos = 0; pos < n; pos = p ? (size_t)(p - h) + 1 : n) {
		p = memchr(h + pos, '\n', n - pos);
		if (n - pos <= len || strncasecmp(h + pos, name, len))
			continue;
		for (k = pos + len; k < n && (h[k] == ' ' || h[k] == '\t');
		     k++)
			;
		if (k >= n || h[k] != ':')
			continue;
		bufadd(&b, "", 0);
		for (k++; k < n; k++) {
			if (h[k] == '\n') {
				if (k + 1 >= n || (h[k + 1] != ' '
				&& h[k + 1] != '\t'))
					break;
			} else if (h[k] != '\r' && h[k]) {
				bufc(&b, h[k]);
			}
		}
		for (k = 0; isspace((unsigned char)b.s[k]); k++)
			;
		while (b.n > k && isspace((unsigned char)b.s[b.n - 1]))
			b.n--;
		r = xstrndup(b.s + k, b.n - k);
		free(b.s);
		return r;
	}
	return NULL;
}

int
hex(int c)
{
	if (c >= '0' && c <= '9')
		return c - '0';
	if (c >= 'a' && c <= 'f')
		return c - 'a' + 10;
	if (c >= 'A' && c <= 'F')
		return c - 'A' + 10;
	return -1;
}

void
httpheader(int status, const char *type)
{
	if (status == 404)
		fputs("Status: 404 Not Found\r\n", stdout);
	printf("Content-Type: %s\r\n", type);
	fputs("Content-Security-Policy: default-src 'none'; "
		"style-src 'self'; img-src 'self'\r\n"
		"X-Content-Type-Options: nosniff\r\n\r\n", stdout);
}

void
indexpage(void)
{
	struct dirent **ents;
	int i, n;

	pagebegin(200, title);
	fputs("<main>\n<table class=\"lists\">\n<thead><tr><th>Name</th>"
		"<th class=\"sm\">Description</th>"
		"<th class=\"sm\">Address</th>"
		"<th class=\"r\">Threads</th><th class=\"r\">Updated</th>"
		"</tr></thead>\n<tbody>\n", stdout);
	if ((n = scandir(root, &ents, NULL, alphasort)) >= 0) {
		for (i = 0; i < n; i++) {
			if (islist(ents[i]->d_name))
				indexrow(ents[i]->d_name);
			free(ents[i]);
		}
		free(ents);
	}
	fputs("</tbody>\n</table>\n", stdout);
	pageend();
}

void
indexrow(const char *name)
{
	char desc[256];
	long long last = -1;
	size_t i, threads = 0;

	list = name;
	scan(name);
	for (i = 0; i < nmsgs; i++) {
		if (msgs[i].root == i)
			threads++;
		if (msgs[i].date > last)
			last = msgs[i].date;
	}
	readline(name, "description", desc, sizeof(desc));
	fputs("<tr><td class=\"name\"><a href=\"", stdout);
	xmlencode(base);
	putchar('/');
	urlencode(name, pathsafe);
	fputs("/\">", stdout);
	xmlencode(name);
	fputs("</a></td><td class=\"desc sm\">", stdout);
	xmlencode(desc);
	fputs("</td><td class=\"sm\"><a href=\"mailto:", stdout);
	urlencode(name, mailsafe);
	putchar('@');
	xmlencode(domain);
	fputs("\">", stdout);
	xmlencode(name);
	putchar('@');
	xmlencode(domain);
	printf("</a></td><td class=\"r num\">%zu</td>", threads);
	printage(last);
	fputs("</tr>\n", stdout);
	freemsgs();
}

int
islist(const char *name)
{
	char path[PATH_MAX];
	struct stat st;

	return validlist(name)
	    && bprintf(path, sizeof(path), "%s/%s", root, name) == 0
	    && stat(path, &st) == 0 && S_ISDIR(st.st_mode)
	    && bprintf(path, sizeof(path), "%s/%s/public", root, name) == 0
	    && stat(path, &st) == 0;
}

int
isstat(const char *s, const char *end)
{
	const char *nl;
	size_t i, len;

	nl = memchr(s, '\n', end - s);
	len = nl ? (size_t)(nl - s) : (size_t)(end - s);
	if (!len || *s != ' ')
		return 0;
	for (i = 0; i + 3 <= len; i++)
		if (!memcmp(s + i, " | ", 3))
			return 1;
	return 0;
}

const char *
lineclass(const char *l, size_t len, const char *next, const char *end,
          int *patch)
{
	if (len >= 11 && !memcmp(l, "diff --git ", 11)) {
		*patch = 1;
		return "f";
	}
	if (!*patch) {
		if (len == 3 && !memcmp(l, "---", 3) && next
		&& isstat(next, end)) {
			*patch = 1;
			return "f";
		}
		return len && *l == '>' ? "q" : NULL;
	}
	if (len == 3 && !memcmp(l, "-- ", 3)) {
		*patch = 0;
		return NULL;
	}
	if ((len == 3 && !memcmp(l, "---", 3))
	|| (len >= 4 && (!memcmp(l, "--- ", 4) || !memcmp(l, "+++ ", 4))))
		return "f";
	if (len >= 2 && !memcmp(l, "@@", 2))
		return "h";
	if (len && *l == '+')
		return "i";
	if (len && *l == '-')
		return "d";
	if (len && isalpha((unsigned char)*l))
		return "f";
	return NULL;
}

void
listheader(int addrs)
{
	static const char *cmds[] = { "", "subscribe", "unsubscribe", "help" };
	char desc[256];
	size_t i;

	readline(list, "description", desc, sizeof(desc));
	fputs("<header>\n<h1><a href=\"", stdout);
	xmlencode(base);
	putchar('/');
	urlencode(list, pathsafe);
	fputs("/\">", stdout);
	xmlencode(list);
	fputs("</a></h1>\n", stdout);
	if (*desc) {
		fputs("<p class=\"desc\">", stdout);
		xmlencode(desc);
		fputs("</p>\n", stdout);
	}
	if (addrs) {
		fputs("<table class=\"addrs\">\n", stdout);
		for (i = 0; i < LENGTH(cmds); i++) {
			printf("<tr><td>%s</td><td><a href=\"mailto:", *cmds[i]
				? cmds[i] : "post");
			urlencode(list, mailsafe);
			printf("%s%s@", *cmds[i] ? "+" : "", cmds[i]);
			xmlencode(domain);
			fputs("\">", stdout);
			xmlencode(list);
			printf("%s%s@", *cmds[i] ? "+" : "", cmds[i]);
			xmlencode(domain);
			fputs("</a></td></tr>\n", stdout);
		}
		fputs("</table>\n", stdout);
	}
	fputs("</header>\n<main>\n", stdout);
}

void
listpage(void)
{
	Thread *t;
	const Msg *m;
	size_t *slot, i, r, n = 0, shown = 0;

	scan(list);
	if (!(t = calloc(nmsgs + 1, sizeof(*t)))
	|| !(slot = calloc(nmsgs + 1, sizeof(*slot))))
		die("out of memory");
	for (i = 0; i < nmsgs; i++)
		slot[i] = NONE;
	for (i = 0; i < nmsgs; i++) {
		r = msgs[i].root;
		if (slot[r] == NONE) {
			slot[r] = n;
			t[n].root = r;
			t[n].last = msgs[i].date;
			n++;
		}
		t[slot[r]].n++;
		if (msgs[i].date > t[slot[r]].last)
			t[slot[r]].last = msgs[i].date;
	}
	qsort(t, n, sizeof(*t), cmpthread);

	pagebegin(200, list);
	listheader(1);
	fputs("<table class=\"threads\">\n<thead><tr><th>Subject</th>"
		"<th class=\"sm\">From</th><th class=\"r\">Msgs</th>"
		"<th class=\"r\">Last</th></tr></thead>\n<tbody>\n", stdout);
	for (i = (size_t)ofs; i < n && shown < (size_t)perpage; i++, shown++) {
		m = &msgs[t[i].root];
		fputs("<tr><td class=\"msg\"><a href=\"", stdout);
		msgurl(m->id, "");
		fputs("\">", stdout);
		xmlencode(*m->subject ? m->subject : "(no subject)");
		fputs("</a></td><td class=\"sm\">", stdout);
		xmlencode(*m->name ? m->name : m->addr);
		printf("</td><td class=\"r num\">%zu</td><td class=\"r date\" "
			"title=\"", t[i].n);
		printtime(t[i].last, 0, "%Y-%m-%d %H:%M UTC");
		fputs("\">", stdout);
		printtime(t[i].last, 0, "%Y-%m-%d");
		fputs("</td></tr>\n", stdout);
	}
	fputs("</tbody>\n</table>\n", stdout);
	if (!shown) {
		fputs("<p>No messages.</p>\n", stdout);
	} else if (ofs || i < n) {
		fputs("<nav class=\"pager\">", stdout);
		if (ofs) {
			fputs("<a href=\"", stdout);
			xmlencode(base);
			putchar('/');
			urlencode(list, pathsafe);
			putchar('/');
			if (ofs > perpage)
				printf("?ofs=%ld", ofs - perpage);
			fputs("\">&larr; newer</a>", stdout);
		}
		printf("<span>%ld&ndash;%ld</span>", ofs + 1,
			ofs + (long)shown);
		if (i < n) {
			fputs("<a href=\"", stdout);
			xmlencode(base);
			putchar('/');
			urlencode(list, pathsafe);
			printf("/?ofs=%zu\">older &rarr;</a>", i);
		}
		fputs("</nav>\n", stdout);
	}
	pageend();
	free(slot);
	free(t);
}

int
loadmsg(Msg *m)
{
	Buf b = { 0 };
	const char *p;
	char *v;
	size_t h, body;

	if (readfile(m->file, (size_t)maxhdr, &b, 1) < 0)
		return -1;
	h = hdrend(b.s, b.n, &body);
	if ((v = hdrget(b.s, h, "Message-ID"))) {
		p = v;
		if (!(m->id = nextid(&p)) && validid(v, strlen(v)))
			m->id = xstrndup(v, strlen(v));
		free(v);
	}
	if (!m->id) {
		free(b.s);
		return -1;
	}
	if ((v = hdrget(b.s, h, "In-Reply-To"))) {
		p = v;
		m->irt = nextid(&p);
		free(v);
	}
	if ((v = hdrget(b.s, h, "References"))) {
		for (p = v; *p;) {
			if (!(m->refs[m->nrefs] = nextid(&p)))
				break;
			if (++m->nrefs == MAXREFS) {
				free(m->refs[0]);
				memmove(m->refs, m->refs + 1,
					(MAXREFS - 1) * sizeof(*m->refs));
				m->nrefs--;
			}
		}
		free(v);
	}
	v = hdrget(b.s, h, "Subject");
	m->subject = field(v);
	free(v);
	v = hdrget(b.s, h, "From");
	parsefrom(v ? v : "", m);
	free(v);
	v = hdrget(b.s, h, "Date");
	if (!v || parsedate(v, &m->date, &m->tz) < 0
	|| m->date > m->mtime + DAY) {
		m->date = m->mtime;
		m->tz = 0;
	}
	free(v);
	free(b.s);
	return 0;
}

void
mboxpage(size_t t)
{
	char *line = NULL;
	size_t i, k, cap = 0, r = msgs[t].root;
	ssize_t len;
	FILE *fp;
	int last;

	httpheader(200, "application/gzip");
	fwrite("\x1f\x8b\x08\0\0\0\0\0\0\xff", 1, 10, stdout);
	gzcrc = 0xffffffffUL;
	for (i = 0; i < nmsgs; i++) {
		if (msgs[i].root != r || !(fp = openmsg(msgs[i].file)))
			continue;
		gzput("From mboxrd@z Thu Jan  1 00:00:00 1970\n", 39);
		last = '\n';
		while ((len = getline(&line, &cap, fp)) > 0) {
			for (k = 0; k < (size_t)len && line[k] == '>'; k++)
				;
			if ((size_t)len - k >= 5
			&& !memcmp(line + k, "From ", 5))
				gzput(">", 1);
			gzput(line, len);
			last = line[len - 1];
		}
		if (last != '\n')
			gzput("\n", 1);
		gzput("\n", 1);
		fclose(fp);
	}
	free(line);
	gzblock(1);
	gzcrc ^= 0xffffffffUL;
	for (k = 0; k < 4; k++)
		putchar((int)(gzcrc >> 8 * k & 0xff));
	for (k = 0; k < 4; k++)
		putchar((int)(gzsize >> 8 * k & 0xff));
}

void
msgurl(const char *id, const char *page)
{
	xmlencode(base);
	putchar('/');
	urlencode(list, pathsafe);
	putchar('/');
	urlencode(id, pathsafe);
	putchar('/');
	fputs(page, stdout);
}

void
multipart(const Part *p, int depth, size_t m)
{
	const char *s, *end, *next, *pick = NULL;
	size_t blen, len, plen = 0;
	int alt, rank, best = 0;
	Part sub;

	end = p->body + p->len;
	blen = strlen(p->boundary);
	alt = !strcmp(p->type, "multipart/alternative");
	for (s = boundary(p->body, end, p->boundary, blen); s && !cut;
	     s = next) {
		s += blen + 2;
		if (end - s >= 2 && s[0] == '-' && s[1] == '-')
			break;
		if (!(s = memchr(s, '\n', end - s)))
			break;
		s++;
		next = boundary(s, end, p->boundary, blen);
		len = (next ? next : end) - s;
		if (len && s[len - 1] == '\n')
			len--;
		if (len && s[len - 1] == '\r')
			len--;
		if (!alt) {
			renderpart(s, len, depth + 1, m);
			continue;
		}
		parsepart(s, len, &sub);
		if (!strcmp(sub.type, "text/plain"))
			rank = 3;
		else if (!strncmp(sub.type, "multipart/", 10))
			rank = 2;
		else
			rank = 1;
		if (rank > best || (rank == 1 && best == 1)) {
			best = rank;
			pick = s;
			plen = len;
		}
	}
	if (pick)
		renderpart(pick, plen, depth + 1, m);
}

char *
nextid(const char **sp)
{
	const char *a, *b;

	for (a = strchr(*sp, '<'); a; a = b) {
		if (!(b = strpbrk(a + 1, "<>")))
			break;
		if (*b == '<')
			continue;
		*sp = b + 1;
		if (validid(a + 1, b - a - 1))
			return xstrndup(a + 1, b - a - 1);
		b = strchr(b + 1, '<');
	}
	*sp += strlen(*sp);
	return NULL;
}

void
notfound(void)
{
	pagebegin(404, "Not found");
	fputs("<main>\n<p>Not found.</p>\n", stdout);
	pageend();
}

FILE *
openmsg(const char *file)
{
	struct stat st;
	FILE *fp;
	int fd;

	if ((fd = open(file, O_RDONLY | O_NOFOLLOW | O_NONBLOCK)) < 0)
		return NULL;
	if (fstat(fd, &st) < 0 || !S_ISREG(st.st_mode)
	|| !(fp = fdopen(fd, "r"))) {
		close(fd);
		return NULL;
	}
	return fp;
}

void
pagebegin(int status, const char *pagetitle)
{
	size_t i;

	httpheader(status, "text/html; charset=utf-8");
	fputs("<!DOCTYPE html>\n<html>\n<head>\n<meta charset=\"utf-8\">\n"
		"<meta name=\"viewport\" "
		"content=\"width=device-width, initial-scale=1\">\n<title>",
		stdout);
	xmlencode(pagetitle);
	fputs("</title>\n<link rel=\"stylesheet\" href=\"", stdout);
	xmlencode(css);
	fputs("\">\n<link rel=\"icon\" href=\"", stdout);
	xmlencode(icon);
	fputs("\">\n</head>\n<body>\n<div id=\"top\"><a class=\"brand\" "
		"href=\"", stdout);
	xmlencode(base);
	fputs("/\"><img src=\"", stdout);
	xmlencode(icon);
	fputs("\" alt=\"\"> ", stdout);
	xmlencode(title);
	fputs("</a><nav>", stdout);
	for (i = 0; i < LENGTH(services); i++) {
		if (!strcmp(services[i].name, service)) {
			fputs("<a class=\"on\" href=\"", stdout);
			xmlencode(base);
			putchar('/');
		} else {
			fputs("<a href=\"", stdout);
			xmlencode(services[i].url);
		}
		fputs("\">", stdout);
		xmlencode(services[i].name);
		fputs("</a>", stdout);
	}
	fputs("</nav></div>\n", stdout);
}

void
pageend(void)
{
	fputs("</main>\n</body>\n</html>\n", stdout);
}

int
param(const char *v, const char *attr, char *buf, size_t len)
{
	Buf raw = { 0 }, d = { 0 }, u = { 0 };
	const Buf *src;
	const char *p = v, *q;
	char *a, *l;
	size_t n, alen = strlen(attr);
	int quote, ext, ret = -1;

	while (ret < 0 && *p) {
		for (quote = 0; *p && (quote || *p != ';'); p++) {
			if (quote && *p == '\\' && p[1])
				p++;
			else if (*p == '"')
				quote = !quote;
		}
		if (!*p)
			break;
		for (p++; isspace((unsigned char)*p); p++)
			;
		for (q = p; *q && *q != '=' && *q != ';'
		     && !isspace((unsigned char)*q); q++)
			;
		n = q - p;
		ext = n == alen + 1 && p[alen] == '*';
		if ((n != alen && !ext) || strncasecmp(p, attr, alen))
			continue;
		for (; isspace((unsigned char)*q); q++)
			;
		if (*q != '=')
			continue;
		for (q++; isspace((unsigned char)*q); q++)
			;
		bufadd(&raw, "", 0);
		if (*q == '"') {
			for (q++; *q && *q != '"'; q++) {
				if (*q == '\\' && q[1])
					q++;
				bufc(&raw, *q);
			}
		} else {
			for (; *q && *q != ';'
			     && !isspace((unsigned char)*q); q++)
				bufc(&raw, *q);
		}
		src = &raw;
		if (ext && (a = strchr(raw.s, '\''))
		&& (l = strchr(a + 1, '\''))) {
			for (q = l + 1; *q; q++) {
				if (*q == '%' && hex(q[1]) >= 0
				&& hex(q[2]) >= 0) {
					bufc(&d, hex(q[1]) << 4 | hex(q[2]));
					q += 2;
				} else {
					bufc(&d, *q);
				}
			}
			*a = '\0';
			if (tou8(raw.s, d.s, d.n, &u) < 0)
				bufadd(&u, d.s, d.n);
			bufadd(&u, "", 0);
			src = &u;
		}
		n = src->n < len - 1 ? src->n : len - 1;
		memcpy(buf, src->s, n);
		buf[n] = '\0';
		ret = 0;
	}
	free(raw.s);
	free(d.s);
	free(u.s);
	return ret;
}

int
parsedate(const char *s, long long *t, int *tz)
{
	static const char months[] = "janfebmaraprmayjunjulaugsepoctnovdec";
	static const int mdays[] = {
		31, 29, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31
	};
	static const struct {
		const char *name;
		int off;
	} zones[] = {
		{ "edt", -240 }, { "est", -300 }, { "cdt", -300 },
		{ "cst", -360 }, { "mdt", -360 }, { "mst", -420 },
		{ "pdt", -420 }, { "pst", -480 },
	};
	long day, mon, year, hh, mm, ss = 0, z = 0, y, era, yoe, doy;
	const char *p;
	size_t n, i;
	int neg;

	s = skipcfws(s);
	if (isalpha((unsigned char)*s)) {
		while (isalpha((unsigned char)*s))
			s++;
		s = skipcfws(s);
	}
	if (!digits(&s, 2, &day))
		return -1;
	s = skipcfws(s);
	for (n = 0; isalpha((unsigned char)s[n]); n++)
		;
	if (n < 3)
		return -1;
	for (mon = 0; mon < 12 && strncasecmp(s, months + 3 * mon, 3); mon++)
		;
	if (mon == 12)
		return -1;
	s = skipcfws(s + n);
	if ((n = digits(&s, 4, &year)) < 2)
		return -1;
	if (n == 2)
		year += year < 50 ? 2000 : 1900;
	else if (n == 3)
		year += 1900;
	s = skipcfws(s);
	if (!digits(&s, 2, &hh))
		return -1;
	if (*(s = skipcfws(s)) != ':')
		return -1;
	s = skipcfws(s + 1);
	if (!digits(&s, 2, &mm))
		return -1;
	if (*(p = skipcfws(s)) == ':') {
		s = skipcfws(p + 1);
		if (!digits(&s, 2, &ss))
			return -1;
	}
	s = skipcfws(s);
	if (*s == '+' || *s == '-') {
		neg = *s++ == '-';
		if (digits(&s, 4, &z) != 4 || z / 100 > 23 || z % 100 > 59)
			return -1;
		z = z / 100 * 60 + z % 100;
		if (neg)
			z = -z;
	} else if (isalpha((unsigned char)*s)) {
		for (n = 0; isalpha((unsigned char)s[n]); n++)
			;
		for (i = 0; i < LENGTH(zones); i++)
			if (n == 3 && !strncasecmp(s, zones[i].name, 3))
				z = zones[i].off;
	}
	if (year < 1900 || year > 9999 || day < 1 || day > mdays[mon]
	|| (mon == 1 && day == 29 && (year % 4 || (!(year % 100)
	&& year % 400))) || hh > 23 || mm > 59 || ss > 60)
		return -1;
	y = year - (mon < 2);
	era = y / 400;
	yoe = y - era * 400;
	doy = (153 * (mon + (mon > 1 ? -2 : 10)) + 2) / 5 + day - 1;
	*t = (long long)(era * 146097 + yoe * 365 + yoe / 4 - yoe / 100 + doy
		- 719468) * 86400 + hh * 3600 + mm * 60 + ss - z * 60;
	*tz = (int)z;
	return 0;
}

void
parsefrom(const char *v, Msg *m)
{
	Buf ph = { 0 }, cm = { 0 }, ad = { 0 };
	const char *p;
	int depth = 0, quote = 0, angle = 0, done = 0;

	for (p = v; *p; p++) {
		if (depth) {
			if (*p == '\\' && p[1])
				p++;
			else if (*p == '(')
				depth++;
			else if (*p == ')' && !--depth)
				continue;
			bufc(&cm, *p);
		} else if (quote) {
			if (*p == '\\' && p[1])
				p++;
			else if (*p == '"') {
				quote = 0;
				continue;
			}
			bufc(&ph, *p);
		} else if (angle) {
			if (*p == '>') {
				angle = 0;
				done = 1;
			} else {
				bufc(&ad, *p);
			}
		} else if (*p == '(') {
			depth = 1;
		} else if (*p == '"') {
			quote = 1;
		} else if (*p == '<' && !done) {
			angle = 1;
		} else if (!done) {
			bufc(&ph, *p);
		}
	}
	bufadd(&ph, "", 0);
	bufadd(&cm, "", 0);
	if (ad.n || done) {
		m->addr = clean(ad.s, ad.n);
		m->name = field(*ph.s ? ph.s : cm.s);
	} else {
		m->addr = clean(ph.s, ph.n);
		m->name = field(cm.s);
	}
	free(ph.s);
	free(cm.s);
	free(ad.s);
}

void
parsepart(const char *s, size_t n, Part *p)
{
	char *ct, *cte, *cd;
	size_t h, b, i;

	h = hdrend(s, n, &b);
	p->body = s + b;
	p->len = n - b;
	ct = hdrget(s, h, "Content-Type");
	cte = hdrget(s, h, "Content-Transfer-Encoding");
	cd = hdrget(s, h, "Content-Disposition");
	for (i = 0; ct && ct[i] && i + 1 < sizeof(p->type)
	     && (isalnum((unsigned char)ct[i]) || strchr("/-+._", ct[i])); i++)
		p->type[i] = tolower((unsigned char)ct[i]);
	p->type[i] = '\0';
	if (!strchr(p->type, '/'))
		strcpy(p->type, "text/plain");
	for (i = 0; cte && cte[i] && i + 1 < sizeof(p->cte)
	     && (isalnum((unsigned char)cte[i]) || cte[i] == '-'); i++)
		p->cte[i] = tolower((unsigned char)cte[i]);
	p->cte[i] = '\0';
	if (!ct || param(ct, "charset", p->charset, sizeof(p->charset)) < 0)
		p->charset[0] = '\0';
	if (!ct || param(ct, "boundary", p->boundary,
	    sizeof(p->boundary)) < 0)
		p->boundary[0] = '\0';
	if ((!cd || param(cd, "filename", p->name, sizeof(p->name)) < 0)
	&& (!ct || param(ct, "name", p->name, sizeof(p->name)) < 0))
		p->name[0] = '\0';
	free(ct);
	free(cte);
	free(cd);
}

void
printage(long long t)
{
	static const char *units[] = {
		"minute", "hour", "day", "week", "month", "year"
	};
	static const long secs[] = {
		60, HOUR, DAY, 7 * DAY, 30 * DAY, 365 * DAY
	};
	static const long upto[] = {
		HOUR, DAY, 14 * DAY, 60 * DAY, 730 * DAY, LONG_MAX
	};
	static const long fade[] = { 7 * DAY, 30 * DAY, 180 * DAY, 730 * DAY };
	long d, n;
	size_t i, level;

	if (t < 0 || t > MAXTIME) {
		fputs("<td class=\"r age a4\">&mdash;</td>", stdout);
		return;
	}
	if ((d = (long)((long long)time(NULL) - t)) < 0)
		d = 0;
	for (i = 0; d >= upto[i]; i++)
		;
	if ((n = d / secs[i]) < 1)
		n = 1;
	for (level = 0; level < LENGTH(fade) && d >= fade[level]; level++)
		;
	printf("<td class=\"r age a%zu\" title=\"", level);
	printtime(t, 0, "%Y-%m-%d %H:%M UTC");
	printf("\">%ld %s%s</td>", n, units[i], n == 1 ? "" : "s");
}

void
printbody(const char *s, size_t n)
{
	const char *l, *nl, *cls;
	size_t len;
	int patch = 0;

	if (n > left) {
		for (n = left; n && ((unsigned char)s[n] & 0xc0) == 0x80; n--)
			;
		cut = 1;
	}
	left -= n;
	fputs("<pre class=\"body\">", stdout);
	for (l = s; l < s + n; l = nl + 1) {
		nl = memchr(l, '\n', s + n - l);
		len = nl ? (size_t)(nl - l) : (size_t)(s + n - l);
		cls = lineclass(l, len, nl ? nl + 1 : NULL, s + n, &patch);
		if (cls)
			printf("<span class=\"%s\">", cls);
		xmlencoden(l, len);
		if (cls)
			fputs("</span>", stdout);
		if (!nl)
			break;
		putchar('\n');
	}
	fputs("</pre>\n", stdout);
}

void
printmsg(size_t i, int sel)
{
	const Msg *m = &msgs[i];
	Buf b = { 0 };
	int trunc;

	printf("<article class=\"mail%s\" id=\"", sel ? " sel" : "");
	xmlencode(m->id);
	fputs("\">\n<table class=\"meta\">\n<tr><td>From</td><td>", stdout);
	if (*m->name) {
		xmlencode(m->name);
		fputs(" &lt;", stdout);
		xmlencode(m->addr);
		fputs("&gt;", stdout);
	} else {
		xmlencode(m->addr);
	}
	fputs("</td></tr>\n<tr><td>Date</td><td>", stdout);
	printtime(m->date, m->tz, "%Y-%m-%d %H:%M");
	printf(" %c%02d%02d</td></tr>\n<tr><td>Subject</td><td>",
		m->tz < 0 ? '-' : '+', abs(m->tz) / 60, abs(m->tz) % 60);
	xmlencode(*m->subject ? m->subject : "(no subject)");
	fputs("</td></tr>\n</table>\n<p class=\"links\"><a href=\"", stdout);
	msgurl(m->id, "");
	fputs("\">permalink</a><a href=\"", stdout);
	msgurl(m->id, "raw");
	fputs("\">raw</a><a href=\"mailto:", stdout);
	urlencode(list, mailsafe);
	putchar('@');
	xmlencode(domain);
	fputs("?In-Reply-To=%3C", stdout);
	urlencode(m->id, mailsafe);
	fputs("%3E&amp;subject=", stdout);
	if (strncasecmp(m->subject, "re:", 3))
		fputs("Re%3A%20", stdout);
	urlencode(m->subject, mailsafe);
	fputs("\">reply</a></p>\n", stdout);

	left = (size_t)maxbody;
	cut = 0;
	if ((trunc = readfile(m->file, (size_t)maxhdr + 2 * (size_t)maxbody,
	    &b, 0)) < 0) {
		fputs("<p class=\"att\">Message unreadable.</p>\n", stdout);
	} else {
		renderpart(b.s, b.n, 0, i);
		if (cut || trunc) {
			fputs("<p class=\"att\">truncated, see <a href=\"",
				stdout);
			msgurl(m->id, "raw");
			fputs("\">raw</a></p>\n", stdout);
		}
	}
	fputs("</article>\n", stdout);
	free(b.s);
}

void
printtime(long long t, int tz, const char *fmt)
{
	struct tm tm;
	time_t secs;
	char buf[64];

	if (t < 0 || t > MAXTIME)
		return;
	secs = (time_t)(t + tz * 60LL);
	if (!gmtime_r(&secs, &tm))
		return;
	if (strftime(buf, sizeof(buf), fmt, &tm))
		fputs(buf, stdout);
}

void
qp(const char *s, size_t n, int q, Buf *out)
{
	size_t i, j;

	for (i = 0; i < n; i++) {
		if (q && s[i] == '_') {
			bufc(out, ' ');
		} else if (s[i] != '=') {
			bufc(out, s[i]);
		} else if (i + 2 < n && hex(s[i + 1]) >= 0
		&& hex(s[i + 2]) >= 0) {
			bufc(out, hex(s[i + 1]) << 4 | hex(s[i + 2]));
			i += 2;
		} else {
			for (j = i + 1; j < n && (s[j] == ' ' || s[j] == '\t');
			     j++)
				;
			if (j < n && s[j] == '\n')
				i = j;
			else if (j + 1 < n && s[j] == '\r' && s[j + 1] == '\n')
				i = j + 1;
			else
				bufc(out, '=');
		}
	}
}

void
rawpage(size_t t)
{
	char buf[8192];
	size_t n;
	FILE *fp;

	if (!(fp = openmsg(msgs[t].file))) {
		notfound();
		return;
	}
	httpheader(200, "text/plain; charset=utf-8");
	while ((n = fread(buf, 1, sizeof(buf), fp)))
		fwrite(buf, 1, n, stdout);
	fclose(fp);
}

int
readfile(const char *file, size_t max, Buf *b, int hdronly)
{
	char chunk[4096];
	size_t want, r, i;
	FILE *fp;
	int trunc = 0;

	bufadd(b, "", 0);
	if (!(fp = openmsg(file)))
		return -1;
	for (;;) {
		if (!(want = max - b->n)) {
			trunc = fgetc(fp) != EOF;
			break;
		}
		if (want > sizeof(chunk))
			want = sizeof(chunk);
		if (!(r = fread(chunk, 1, want, fp)))
			break;
		i = b->n > 2 ? b->n - 2 : 0;
		bufadd(b, chunk, r);
		for (; hdronly && i + 1 < b->n; i++)
			if (b->s[i] == '\n' && (b->s[i + 1] == '\n'
			|| (b->s[i + 1] == '\r' && i + 2 < b->n
			&& b->s[i + 2] == '\n')))
				hdronly = 2;
		if (hdronly == 2)
			break;
	}
	fclose(fp);
	return trunc;
}

void
readline(const char *name, const char *file, char *buf, size_t len)
{
	char path[PATH_MAX];
	FILE *fp;

	buf[0] = '\0';
	if (bprintf(path, sizeof(path), "%s/%s/%s", root, name, file) < 0
	|| !(fp = fopen(path, "r")))
		return;
	if (fgets(buf, (int)len, fp))
		buf[strcspn(buf, "\r\n")] = '\0';
	fclose(fp);
}

void
renderpart(const char *s, size_t n, int depth, size_t m)
{
	Buf raw = { 0 }, u = { 0 }, t = { 0 };
	Part p;

	if (cut)
		return;
	parsepart(s, n, &p);
	if (!strncmp(p.type, "multipart/", 10) && *p.boundary
	&& depth < maxdepth) {
		multipart(&p, depth, m);
	} else if (!strcmp(p.type, "text/plain")
	|| !strcmp(p.type, "text/x-diff") || !strcmp(p.type, "text/x-patch")) {
		decode(&p, &raw);
		if (tou8(p.charset, raw.s, raw.n, &u) < 0)
			bufadd(&u, raw.s, raw.n);
		sanitize(u.s, u.n, &t, 1);
		printbody(t.s, t.n);
	} else if (!strcmp(p.type, "text/html")) {
		fputs("<p class=\"att\">HTML part not shown, see <a href=\"",
			stdout);
		msgurl(msgs[m].id, "raw");
		fputs("\">raw</a></p>\n", stdout);
	} else {
		attachment(&p);
	}
	free(raw.s);
	free(u.s);
	free(t.s);
}

void
route(char *path)
{
	char *rest;
	size_t n, t;
	int page = 0;

	while (*path == '/')
		path++;
	if (!*path) {
		indexpage();
		return;
	}
	if ((rest = strchr(path, '/')))
		*rest++ = '\0';
	if (!islist(path)) {
		notfound();
		return;
	}
	list = path;
	if (!rest || !*rest) {
		listpage();
		return;
	}
	n = strlen(rest);
	if (rest[n - 1] == '/') {
		rest[--n] = '\0';
	} else if (n > 4 && !strcmp(rest + n - 4, "/raw")) {
		rest[n -= 4] = '\0';
		page = 1;
	} else if (n > 10 && !strcmp(rest + n - 10, "/t.mbox.gz")) {
		rest[n -= 10] = '\0';
		page = 2;
	}
	if (!validid(rest, n)) {
		notfound();
		return;
	}
	scan(list);
	if ((t = byid(rest)) == NONE)
		notfound();
	else if (page == 1)
		rawpage(t);
	else if (page == 2)
		mboxpage(t);
	else
		threadpage(t);
}

void
sanitize(const char *s, size_t n, Buf *out, int multi)
{
	const unsigned char *p = (const unsigned char *)s, *e = p + n;
	unsigned long c;
	size_t i, len;

	while (p < e) {
		c = *p;
		if (c < 0x80) {
			if (c == '\n' || c == '\t' || (c == '\r' && !multi))
				bufc(out, multi ? (int)c : ' ');
			else if ((c < 0x20 && c != '\r') || c == 0x7f)
				bufadd(out, REPL, 3);
			else if (c != '\r')
				bufc(out, (int)c);
			p++;
			continue;
		}
		if (c >= 0xc2 && c <= 0xdf) {
			len = 2;
			c &= 0x1f;
		} else if (c >= 0xe0 && c <= 0xef) {
			len = 3;
			c &= 0x0f;
		} else if (c >= 0xf0 && c <= 0xf4) {
			len = 4;
			c &= 0x07;
		} else {
			len = 0;
		}
		for (i = 1; i < len; i++) {
			if (p + i >= e || (p[i] & 0xc0) != 0x80)
				len = 0;
			else
				c = c << 6 | (p[i] & 0x3f);
		}
		if (!len || (len == 3 && c < 0x800)
		|| (len == 4 && (c < 0x10000 || c > 0x10ffff))
		|| (c >= 0xd800 && c <= 0xdfff)) {
			bufadd(out, REPL, 3);
			p++;
			continue;
		}
		if (c < 0xa0 || (c >= 0x202a && c <= 0x202e)
		|| (c >= 0x2066 && c <= 0x2069))
			bufadd(out, REPL, 3);
		else
			bufadd(out, (const char *)p, len);
		p += len;
	}
}

void
scan(const char *name)
{
	static const char *sub[] = { "new", "cur" };
	char dir[PATH_MAX], path[PATH_MAX];
	struct dirent *e;
	struct stat st;
	Msg *v;
	DIR *dp;
	size_t i, j, n = 0, cap = 0;

	freemsgs();
	for (i = 0; i < LENGTH(sub); i++) {
		if (bprintf(dir, sizeof(dir), "%s/%s/archive/%s", root, name,
		    sub[i]) < 0 || !(dp = opendir(dir)))
			continue;
		while ((e = readdir(dp))) {
			if (e->d_name[0] == '.'
			|| bprintf(path, sizeof(path), "%s/%s", dir,
			   e->d_name) < 0
			|| lstat(path, &st) < 0 || !S_ISREG(st.st_mode))
				continue;
			if (n == cap) {
				cap = cap ? cap * 2 : 256;
				if (!(v = realloc(msgs, cap * sizeof(*v))))
					die("out of memory");
				msgs = v;
			}
			memset(&msgs[n], 0, sizeof(*msgs));
			msgs[n].file = xstrndup(path, strlen(path));
			msgs[n].mtime = (long long)st.st_mtime;
			n++;
		}
		closedir(dp);
	}
	if (n)
		qsort(msgs, n, sizeof(*msgs), cmpmtime);
	for (i = j = 0; i < n; i++) {
		if (i < (size_t)maxmsgs && loadmsg(&msgs[i]) >= 0)
			msgs[j++] = msgs[i];
		else
			freemsg(&msgs[i]);
	}
	nmsgs = j;
	if (nmsgs)
		qsort(msgs, nmsgs, sizeof(*msgs), cmpfirst);
	for (i = j = 0; i < nmsgs; i++) {
		if (j && !strcmp(msgs[j - 1].id, msgs[i].id))
			freemsg(&msgs[i]);
		else
			msgs[j++] = msgs[i];
	}
	nmsgs = j;
	if (nmsgs)
		qsort(msgs, nmsgs, sizeof(*msgs), cmpdate);
	if (!(ids = calloc(nmsgs + 1, sizeof(*ids))))
		die("out of memory");
	for (i = 0; i < nmsgs; i++)
		ids[i] = i;
	if (nmsgs)
		qsort(ids, nmsgs, sizeof(*ids), cmpid);
	threadmsgs();
}

const char *
skipcfws(const char *s)
{
	int depth = 0;

	for (; *s; s++) {
		if (depth && *s == '\\' && s[1])
			s++;
		else if (*s == '(')
			depth++;
		else if (depth && *s == ')')
			depth--;
		else if (!depth && !isspace((unsigned char)*s) && *s != ',')
			break;
	}
	return s;
}

void
threadmsgs(void)
{
	size_t *stack, i, j, k, n, p, r;

	for (i = 0; i < nmsgs; i++) {
		msgs[i].parent = msgs[i].root = NONE;
		if (msgs[i].irt && (p = byid(msgs[i].irt)) != NONE && p != i)
			msgs[i].parent = p;
		for (k = msgs[i].nrefs; msgs[i].parent == NONE && k; k--)
			if ((p = byid(msgs[i].refs[k - 1])) != NONE && p != i)
				msgs[i].parent = p;
	}
	if (!(stack = calloc(nmsgs + 1, sizeof(*stack))))
		die("out of memory");
	for (i = 0; i < nmsgs; i++) {
		n = 0;
		for (j = i; msgs[j].root == NONE && msgs[j].seen != i + 1;
		     j = msgs[j].parent) {
			msgs[j].seen = i + 1;
			stack[n++] = j;
			if (msgs[j].parent == NONE)
				break;
		}
		r = msgs[j].root != NONE ? msgs[j].root : j;
		for (k = 0; k < n; k++)
			msgs[stack[k]].root = r;
	}
	free(stack);
}

void
threadpage(size_t t)
{
	const Msg *r = &msgs[msgs[t].root];
	size_t i, n = 0;

	if (msgs[t].root != t) {
		printf("Status: 302 Found\r\nLocation: %s/", base);
		urlencode(list, pathsafe);
		putchar('/');
		urlencode(r->id, pathsafe);
		fputs("/#", stdout);
		urlencode(msgs[t].id, pathsafe);
		fputs("\r\n", stdout);
		httpheader(302, "text/plain; charset=utf-8");
		return;
	}
	for (i = 0; i < nmsgs; i++)
		if (msgs[i].root == t)
			n++;
	pagebegin(200, *r->subject ? r->subject : "(no subject)");
	listheader(0);
	fputs("<h2 class=\"subject\">", stdout);
	xmlencode(*r->subject ? r->subject : "(no subject)");
	printf("</h2>\n<p class=\"info\">%zu message%s &middot; <a href=\"",
		n, n == 1 ? "" : "s");
	msgurl(r->id, "t.mbox.gz");
	fputs("\">t.mbox.gz</a></p>\n", stdout);
	for (i = 0; i < nmsgs; i++)
		if (msgs[i].root == t)
			printmsg(i, i == t);
	pageend();
}

int
tou8(const char *cs, const char *s, size_t n, Buf *out)
{
	char name[64], tmp[1024], *in, *o;
	size_t i, il, ol, r;
	iconv_t cd;

	for (i = 0; cs[i] && cs[i] != '*'; i++) {
		if (i + 1 >= sizeof(name) || !(isalnum((unsigned char)cs[i])
		|| strchr("-_.:", cs[i])))
			return -1;
		name[i] = cs[i];
	}
	name[i] = '\0';
	if (!*name || !strcasecmp(name, "utf-8") || !strcasecmp(name, "utf8")
	|| !strcasecmp(name, "us-ascii") || !strcasecmp(name, "ascii")) {
		bufadd(out, s, n);
		return 0;
	}
	if ((cd = iconv_open("UTF-8", name)) == (iconv_t)-1)
		return -1;
	in = (char *)s;
	il = n;
	while (il) {
		o = tmp;
		ol = sizeof(tmp);
		r = iconv(cd, &in, &il, &o, &ol);
		bufadd(out, tmp, o - tmp);
		if (r != (size_t)-1 || errno == E2BIG)
			continue;
		bufadd(out, REPL, 3);
		if (errno != EILSEQ)
			break;
		in++;
		il--;
	}
	o = tmp;
	ol = sizeof(tmp);
	iconv(cd, NULL, NULL, &o, &ol);
	bufadd(out, tmp, o - tmp);
	iconv_close(cd);
	return 0;
}

void
urlencode(const char *s, const char *safe)
{
	for (; *s; s++) {
		if (isalnum((unsigned char)*s) || strchr(safe, *s))
			putchar(*s);
		else
			printf("%%%02X", (unsigned char)*s);
	}
}

int
validid(const char *s, size_t n)
{
	size_t i;

	if (!n || n > FIELDLEN)
		return 0;
	for (i = 0; i < n; i++)
		if ((unsigned char)s[i] <= ' ' || s[i] == 0x7f || s[i] == '<'
		|| s[i] == '>')
			return 0;
	return 1;
}

int
validlist(const char *s)
{
	size_t n;

	for (n = 0; s[n]; n++)
		if (!islower((unsigned char)s[n])
		&& !isdigit((unsigned char)s[n]) && !(n && s[n] == '-'))
			return 0;
	return n && n <= 64;
}

void
xmlencode(const char *s)
{
	if (s)
		xmlencoden(s, strlen(s));
}

void
xmlencoden(const char *s, size_t len)
{
	for (; len; s++, len--) {
		switch (*s) {
		case '<':  fputs("&lt;", stdout);   break;
		case '>':  fputs("&gt;", stdout);   break;
		case '&':  fputs("&amp;", stdout);  break;
		case '"':  fputs("&quot;", stdout); break;
		case '\'': fputs("&#39;", stdout);  break;
		default:   putchar(*s);
		}
	}
}

char *
xstrndup(const char *s, size_t n)
{
	char *r;

	if (!(r = malloc(n + 1)))
		die("out of memory");
	if (n)
		memcpy(r, s, n);
	r[n] = '\0';
	return r;
}

/* main */
int
main(int argc, char *argv[])
{
	char path[2048];
	const char *q;

	if (argc == 2 && !strcmp(argv[1], "-v")) {
		fputs("herbarium-"VERSION"\n", stderr);
		return 1;
	}
	base = envor("SCRIPT_NAME", "");
	q = envor("QUERY_STRING", "");
	if (!strncmp(q, "ofs=", 4) && (ofs = strtol(q + 4, NULL, 10)) < 0)
		ofs = 0;

	if (bprintf(path, sizeof(path), "%s", envor("PATH_INFO", "")) < 0)
		notfound();
	else
		route(path);
	return 0;
}