hydra/src/util.c

/*
 *  Hydra, an http server
 *  Copyright (C) 1995 Paul Phillips <paulp@go2net.com>
 *  Some changes Copyright (C) 1996,97 Larry Doolittle <ldoolitt@boa.org>
 *  Some changes Copyright (C) 1996 Charles F. Randall <crandall@goldsys.com>
 *  Some changes Copyright (C) 1996-99 Jon Nelson <jnelson@boa.org>
 *  Portions Copyright (C) 2002 Nikos Mavroyanopoulos <nmav@gnutls.org>
 *
 *  This program is free software; you can redistribute it and/or modify
 *  it under the terms of the GNU General Public License as published by
 *  the Free Software Foundation; either version 1, or (at your option)
 *  any later version.
 *
 *  This program is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *  GNU General Public License for more details.
 *
 *  You should have received a copy of the GNU General Public License
 *  along with this program; if not, write to the Free Software
 *  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 *
 */

/* $Id: util.c,v 1.16 2002/10/26 12:40:29 nmav Exp $ */

#include "boa.h"

#define HEX_TO_DECIMAL(char1, char2)    \
    (((char1 >= 'A') ? (((char1 & 0xdf) - 'A') + 10) : (char1 - '0')) * 16) + \
    (((char2 >= 'A') ? (((char2 & 0xdf) - 'A') + 10) : (char2 - '0')))

const char month_tab[48] =
    "Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec ";
const char day_tab[] = "Sun,Mon,Tue,Wed,Thu,Fri,Sat,";

/*
 * Name: clean_pathname
 *
 * Description: Replaces unsafe/incorrect instances of:
 *  //[...] with /
 *  /./ with /
 *  /../ with / (technically not what we want, but browsers should deal
 *   with this, not servers)
 */

void clean_pathname(char *pathname)
{
    char *cleanpath, c;

    cleanpath = pathname;
    while ((c = *pathname++)) {
        if (c == '/') {
            while (1) {
                if (*pathname == '/')
                    pathname++;
                else if (*pathname == '.' && *(pathname + 1) == '/')
                    pathname += 2;
                else if (*pathname == '.' && *(pathname + 1) == '.' &&
                         *(pathname + 2) == '/') {
                    pathname += 3;
                } else
                    break;
            }
            c = '/';
        }
        *cleanpath++ = c;
    }

    *cleanpath = '\0';
}

/*
 * Name: get_commonlog_time
 *
 * Description: Returns the current time in common log format in a static
 * char buffer.
 *
 * commonlog time is exactly 25 characters long
 * because this is only used in logging, we add " [" before and "] " after
 * making 29 characters
 * "[27/Feb/1998:20:20:04 +0000] "
 *
 * Constrast with rfc822 time:
 * "Sun, 06 Nov 1994 08:49:37 GMT"
 *
 * Altered 10 Jan 2000 by Jon Nelson ala Drew Streib for non UTC logging
 *
 */

void get_commonlog_time(char buf[30])
{
    struct tm *t;
    char *p;
    unsigned int a;
    int time_offset;

    if (use_localtime) {
        t = localtime(&current_time);
        time_offset = TIMEZONE_OFFSET(t);
    } else {
        t = gmtime(&current_time);
        time_offset = 0;
    }

    p = buf + 29;
    *p-- = '\0';
    *p-- = ' ';
    *p-- = ']';
    a = abs(time_offset / 60);
    *p-- = '0' + a % 10;
    a /= 10;
    *p-- = '0' + a % 6;
    a /= 6;
    *p-- = '0' + a % 10;
    *p-- = '0' + a / 10;
    *p-- = (time_offset >= 0) ? '+' : '-';
    *p-- = ' ';

    a = t->tm_sec;
    *p-- = '0' + a % 10;
    *p-- = '0' + a / 10;
    *p-- = ':';
    a = t->tm_min;
    *p-- = '0' + a % 10;
    *p-- = '0' + a / 10;
    *p-- = ':';
    a = t->tm_hour;
    *p-- = '0' + a % 10;
    *p-- = '0' + a / 10;
    *p-- = ':';
    a = 1900 + t->tm_year;
    while (a) {
        *p-- = '0' + a % 10;
        a /= 10;
    }
    /* p points to an unused spot */
    *p-- = '/';
    p -= 2;
    memcpy(p--, month_tab + 4 * (t->tm_mon), 3);
    *p-- = '/';
    a = t->tm_mday;
    *p-- = '0' + a % 10;
    *p-- = '0' + a / 10;
    *p = '[';
    return;                   /* should be same as returning buf */
}

/*
 * Name: month2int
 *
 * Description: Turns a three letter month into a 0-11 int
 *
 * Note: This function is from wn-v1.07 -- it's clever and fast
 */

int month2int(char *monthname)
{
    switch (*monthname) {
    case 'A':
        return (*++monthname == 'p' ? 3 : 7);
    case 'D':
        return (11);
    case 'F':
        return (1);
    case 'J':
        if (*++monthname == 'a')
            return (0);
        return (*++monthname == 'n' ? 5 : 6);
    case 'M':
        return (*(monthname + 2) == 'r' ? 2 : 4);
    case 'N':
        return (10);
    case 'O':
        return (9);
    case 'S':
        return (8);
    default:
        return (-1);
    }
}

/*
 * Name: modified_since
 * Description: Decides whether a file's mtime is newer than the
 * If-Modified-Since header of a request.
 *

 Sun, 06 Nov 1994 08:49:37 GMT    ; RFC 822, updated by RFC 1123
 Sunday, 06-Nov-94 08:49:37 GMT   ; RFC 850, obsoleted by RFC 1036
 Sun Nov  6 08:49:37 1994         ; ANSI C's asctime() format
 31 September 2000 23:59:59 GMT   ; non-standard

 * RETURN VALUES:
 *  0: File has not been modified since specified time.
 *  1: File has been.
 * -1: Error!
 */

int modified_since(time_t mtime, char *if_modified_since)
{
    struct tm *file_gmt;
    char *ims_info;
    char monthname[10 + 1];
    int day, month, year, hour, minute, second;
    int comp;

    ims_info = if_modified_since;
    while (*ims_info != ' ' && *ims_info != '\0')
        ++ims_info;
    if (*ims_info != ' ')
        return -1;

    /* the pre-space in the third scanf skips whitespace for the string */
    if (sscanf(ims_info, "%d %3s %d %d:%d:%d GMT", /* RFC 1123 */
               &day, monthname, &year, &hour, &minute, &second) == 6);
    else if (sscanf(ims_info, "%d-%3s-%d %d:%d:%d GMT", /* RFC 1036 */
                    &day, monthname, &year, &hour, &minute, &second) == 6)
        year += 1900;
    else if (sscanf(ims_info, " %3s %d %d:%d:%d %d", /* asctime() format */
                    monthname, &day, &hour, &minute, &second, &year) == 6);
    /*  allow this non-standard date format: 31 September 2000 23:59:59 GMT */
    /* NOTE: Use if_modified_since here, because the date *starts*
     *       with the day, versus a throwaway item
     */
    else if (sscanf(if_modified_since, "%d %10s %d %d:%d:%d GMT",
                    &day, monthname, &year, &hour, &minute, &second) == 6);
    else {
        log_error_time();
        fprintf(stderr, "Error in %s, line %d: Unable to sscanf \"%s\"\n",
                __FILE__, __LINE__, ims_info);
        return -1;              /* error */
    }

    file_gmt = gmtime(&mtime);
    month = month2int(monthname);

    /* Go through from years to seconds -- if they are ever unequal,
     we know which one is newer and can return */

    if ((comp = 1900 + file_gmt->tm_year - year))
        return (comp > 0);
    if ((comp = file_gmt->tm_mon - month))
        return (comp > 0);
    if ((comp = file_gmt->tm_mday - day))
        return (comp > 0);
    if ((comp = file_gmt->tm_hour - hour))
        return (comp > 0);
    if ((comp = file_gmt->tm_min - minute))
        return (comp > 0);
    if ((comp = file_gmt->tm_sec - second))
        return (comp > 0);

    return 0;                   /* this person must really be into the latest/greatest */
}


/*
 * Name: to_upper
 *
 * Description: Turns a string into all upper case (for HTTP_ header forming)
 * AND changes - into _
 */

char *to_upper(char *str)
{
    char *start = str;

    while (*str) {
        if (*str == '-')
            *str = '_';
        else
            *str = toupper(*str);

        str++;
    }

    return start;
}

/*
 * Name: unescape_uri
 *
 * Description: Decodes a uri, changing %xx encodings with the actual
 * character.  The query_string should already be gone.
 *
 * Return values:
 *  1: success
 *  0: illegal string
 */

int unescape_uri(char *uri, char ** query_string)
{
    char c, d;
    char *uri_old;

    uri_old = uri;

    while ((c = *uri_old)) {
        if (c == '%') {
            uri_old++;
            if ((c = *uri_old++) && (d = *uri_old++))
                *uri++ = HEX_TO_DECIMAL(c, d);
            else
                return 0;       /* NULL in chars to be decoded */
        } else if (c == '?') { /* query string */
            if (query_string)
                *query_string = ++uri_old;
            /* stop here */
            *uri = '\0';
            return(1);
            break;
        } else if (c == '#') { /* fragment */
            /* legal part of URL, but we do *not* care.
             * However, we still have to look for the query string */
            if (query_string) {
                ++uri_old;
                while((c = *uri_old)) {
                    if (c == '?') {
                        *query_string = ++uri_old;
                        break;
                    }
                    ++uri_old;
                }
            }
            break;
        } else {
            *uri++ = c;
            uri_old++;
        }
    }

    *uri = '\0';
    return 1;
}

/* rfc822 (1123) time is exactly 29 characters long
 * "Sun, 06 Nov 1994 08:49:37 GMT"
 */

void rfc822_time_buf(char *buf, time_t s)
{
    struct tm *t;
    char *p;
    unsigned int a;

    if (!s) {
        t = gmtime(&current_time);
    } else
        t = gmtime(&s);

    p = buf + 28;
    /* p points to the last char in the buf */

    p -= 3;
    /* p points to where the ' ' will go */
    memcpy(p--, " GMT", 4);

    a = t->tm_sec;
    *p-- = '0' + a % 10;
    *p-- = '0' + a / 10;
    *p-- = ':';
    a = t->tm_min;
    *p-- = '0' + a % 10;
    *p-- = '0' + a / 10;
    *p-- = ':';
    a = t->tm_hour;
    *p-- = '0' + a % 10;
    *p-- = '0' + a / 10;
    *p-- = ' ';
    a = 1900 + t->tm_year;
    while (a) {
        *p-- = '0' + a % 10;
        a /= 10;
    }
    /* p points to an unused spot to where the space will go */
    p -= 3;
    /* p points to where the first char of the month will go */
    memcpy(p--, month_tab + 4 * (t->tm_mon), 4);
    *p-- = ' ';
    a = t->tm_mday;
    *p-- = '0' + a % 10;
    *p-- = '0' + a / 10;
    *p-- = ' ';
    p -= 3;
    memcpy(p, day_tab + t->tm_wday * 4, 4);
}

/* Returns the number of digits */
int simple_itoa(large_int i, char buf[22])
{
    /* 21 digits plus null terminator, good for 64-bit or smaller ints
     * for bigger ints, use a bigger buffer!
     *
     * 4294967295 is, incidentally, MAX_UINT (on 32bit systems at this time)
     * and is 10 bytes long
     */
    char *p = &buf[21];
    int digits = 1; /* include null char */

    *p-- = '\0';
    do {
        digits++;
        *p-- = '0' + i % 10;
        i /= 10;
    } while (i > 0);

    p++;
    if (p!=buf) memmove( buf, p, digits);

    return digits - 1;
}

/* Generates an Etag, by using the file size, and the modification time
 */
int create_etag(unsigned long int size, unsigned long int mod_time, 
   char etag[MAX_ETAG_LENGTH])
{
char buf[22];
int len, len2;

    etag[0] = '\"'; 
    len = 1;

    len2 = simple_itoa( size % 100000, buf); /* up to 5 digits */
    memcpy( &etag[len], buf, len2);
    len += len2;

    etag[len++] = '-';
    len2 = simple_itoa( mod_time % 100000, buf); /* also 5 digits */
    memcpy( &etag[len], buf, len2);
    len += len2;

    etag[len++] = '\"';
    etag[len] = 0; /* etag is null terminated */

    return len;
}


/* I don't "do" negative conversions
 * Therefore, -1 indicates error
 */

int boa_atoi(const char *s)
{
    int retval;
    char reconv[22];

    if (!isdigit(*s))
        return -1;

    retval = atoi( s);
    if (retval < 0)
        return -1;

    simple_itoa(retval, reconv);
    if (memcmp(s,reconv,strlen(s)) != 0) {
        return -1;
    }
    return retval;
}

large_int boa_atoll(const char *s)
{
    long int retval;
    char reconv[22];

    if (!isdigit(*s))
        return -1;

#ifdef HAVE_ATOLL
    retval = atoll( s);
#else
    retval = atol( s);
#endif
    if (retval < 0)
        return -1;

    simple_itoa(retval, reconv);
    if (memcmp(s,reconv,strlen(s)) != 0) {
        return -1;
    }
    return retval;
}

#define TEMP_FILE_TEMPLATE "/hydra.temp.XXXXXX"
#define TEMP_FILE_TEMPLATE_LEN sizeof(TEMP_FILE_TEMPLATE)-1

void close_tmp_fd( tmp_fd* fds) 
{
   if (fds->pipe) {
      close( fds->fds[0]);
      close( fds->fds[1]);
   } else close( fds->fds[0]);
   
   fds->fds[0] = fds->fds[1] = -1;
   
}

const static tmp_fd EMPTY_FDS = { {-1, -1}, 0};

/* returns -1 on error */
/* size holds the number of data that will be written to
 * the temporary file.
 */
tmp_fd create_temporary_file(short want_unlink, int size)
{
    char boa_tempfile[MAX_PATH_LENGTH + 1];
    tmp_fd fd;
    int total_len;

    fd.pipe = 0;

    if (size > 0 && size < PIPE_BUF)
        if (pipe( fd.fds) != -1) {
           fd.pipe = 1;
           return fd;
        }

    total_len = tempdir_len + TEMP_FILE_TEMPLATE_LEN;
    if (total_len > MAX_PATH_LENGTH) {
        log_error_time();
        fprintf(stderr, "Temporary file length (%d) is too long\n", total_len);
        return EMPTY_FDS;
    }

    memcpy( boa_tempfile, tempdir, tempdir_len);
    memcpy( &boa_tempfile[tempdir_len], TEMP_FILE_TEMPLATE, TEMP_FILE_TEMPLATE_LEN);
    boa_tempfile[total_len] = 0; /* null terminated */

    fd.fds[0] = fd.fds[1] = -1;
    /* open temp file 
     */
    fd.fds[0] = mkstemp(boa_tempfile);
    if (fd.fds[0] == -1) {
        log_error_time();
        perror("mkstemp");
        return EMPTY_FDS;
    }

    if (want_unlink) {
        if (unlink(boa_tempfile) == -1) {
            close(fd.fds[0]);
            fd.fds[0] = -1;
            log_error_time();
            fprintf(stderr, "unlink temp file\n");
        }
    }

    fd.fds[1] = fd.fds[0];
    return (fd);
}

int set_block_fd(int fd)
{
    int flags;

    flags = fcntl(fd, F_GETFL);
    if (flags == -1)
        return -1;

    flags &= ~O_NONBLOCK;
    flags = fcntl(fd, F_SETFL, flags);
    return flags;
}

int set_nonblock_fd(int fd)
{
    int flags;

    flags = fcntl(fd, F_GETFL);
    if (flags == -1)
        return -1;

    flags |= O_NONBLOCK;
    flags = fcntl(fd, F_SETFL, flags);
    return flags;
}

int set_cloexec_fd(int fd)
{
    int flags;

    flags = fcntl(fd, F_GETFL);
    if (flags == -1)
        return -1;

    flags |= FD_CLOEXEC;
    flags = fcntl(fd, F_SETFL, flags);
    return flags;
}

void create_url( char * buffer, int buffer_size, int secure, 
   const char* hostname, int port, const char* request_uri)
{
char * proto;
char str_port[23];
int do_port = 0;

   buffer[0] = 0; /* in case we fail */
   if (hostname == NULL) {
        hostname = "(null)";
   }

   if (request_uri == NULL) {
        request_uri = "(null)";
   }

   if (secure) {
      proto = "https";
      if (port != 443)
         do_port = 1;
   } else {
      proto = "http";
      if (port!=80)
         do_port = 1;
   }

   if (do_port) {
      str_port[0] = ':';
      simple_itoa( port, &str_port[1]);
   }

   if ((strlen(proto) + strlen(str_port) + strlen(hostname) +
       strlen(request_uri) + 5) > buffer_size) {
       /* This is more than impossible. The buffer is long enough */
       log_error_time();
       fprintf(stderr, "Could not create URL. Buffer was not enough.\n");
       return;
   }

   sprintf( buffer, "%s://%s%s%s/", proto, hostname, str_port, request_uri);

   return;
}

/* Breaks a list of "xxx", "yyy", to a character array, of
 * MAX_COMMA_SEP_ELEMENTS size; Note that the given string is modified.
 */
void break_comma_list(char *etag,
                 char *broken_etag[MAX_COMMA_SEP_ELEMENTS],
                         int *elements)
{
   char *p = etag;

   *elements = 0;

   do {
      broken_etag[*elements] = p;

      (*elements)++;

      p = strchr(p, ',');
      if (p) {
         *p = 0;
         p++;                   /* move to next entry and skip white
                                 * space.
                                 */
         while (*p == ' ')
            p++;
      }
   } while (p != NULL && *elements < MAX_COMMA_SEP_ELEMENTS);
}

/* Quoting from rfc1034:

<domain> ::= <subdomain> | " "

<subdomain> ::= <label> | <subdomain> "." <label>

<label> ::= <letter> [ [ <ldh-str> ] <let-dig> ]

<ldh-str> ::= <let-dig-hyp> | <let-dig-hyp> <ldh-str>

<let-dig-hyp> ::= <let-dig> | "-"

<let-dig> ::= <letter> | <digit>

<letter> ::= any one of the 52 alphabetic characters A through Z in
upper case and a through z in lower case

<digit> ::= any one of the ten digits 0 through 9

and

The labels must follow the rules for ARPANET host names.  They must
start with a letter, end with a letter or digit, and have as interior
characters only letters, digits, and hyphen.  There are also some
restrictions on the length.  Labels must be 63 characters or less.

*/

int check_host(char *r)
{
    /* a hostname can only consist of
     * chars and numbers, and sep. by only
     * one period.
     * It may not end with a period, and must
     * not start with a number.
     *
     * >0: correct
     * -1: error
     *  0: not returned
     *
     */
    char *c;
    short period_ok = 0;
    short len = 0;

    c = r;
    if (c == NULL) {
        return -1;
    }

    /* must start with a letter */
    if (!isalpha(*c))
        return -1;

    len = 1;
    while (*(++c) != '\0') {
        /* interior letters may be alphanumeric, '-', or '.' */
        /* '.' may not follow '.' */
        if (isalnum(*c) || *c == '-')
            period_ok = 1;
        else if (*c == '.' && period_ok)
            period_ok = 0;
        else
            return -1;
        ++len;
    }
    /* c points to '\0' */
    --c;
    /* must end with a letter or digit */
    if (!isalnum(*c))
        return -1;
    return len;
}


1	nmav	1.1	/*
2	nmav	1.7	* Hydra, an http server
3	nmav	1.1	* Copyright (C) 1995 Paul Phillips <paulp@go2net.com>
4			* Some changes Copyright (C) 1996,97 Larry Doolittle <ldoolitt@boa.org>
5			* Some changes Copyright (C) 1996 Charles F. Randall <crandall@goldsys.com>
6			* Some changes Copyright (C) 1996-99 Jon Nelson <jnelson@boa.org>
7	nmav	1.9	* Portions Copyright (C) 2002 Nikos Mavroyanopoulos <nmav@gnutls.org>
8	nmav	1.1	*
9			* This program is free software; you can redistribute it and/or modify
10			* it under the terms of the GNU General Public License as published by
11			* the Free Software Foundation; either version 1, or (at your option)
12			* any later version.
13			*
14			* This program is distributed in the hope that it will be useful,
15			* but WITHOUT ANY WARRANTY; without even the implied warranty of
16			* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17			* GNU General Public License for more details.
18			*
19			* You should have received a copy of the GNU General Public License
20			* along with this program; if not, write to the Free Software
21			* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
22			*
23			*/
24
25	nmav	1.17	/* $Id: util.c,v 1.16 2002/10/26 12:40:29 nmav Exp $ */
26	nmav	1.1
27			#include "boa.h"
28
29			#define HEX_TO_DECIMAL(char1, char2) \
30			(((char1 >= 'A') ? (((char1 & 0xdf) - 'A') + 10) : (char1 - '0')) * 16) + \
31			(((char2 >= 'A') ? (((char2 & 0xdf) - 'A') + 10) : (char2 - '0')))
32
33			const char month_tab[48] =
34			"Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec ";
35			const char day_tab[] = "Sun,Mon,Tue,Wed,Thu,Fri,Sat,";
36
37			/*
38			* Name: clean_pathname
39			*
40			* Description: Replaces unsafe/incorrect instances of:
41			* //[...] with /
42			* /./ with /
43			* /../ with / (technically not what we want, but browsers should deal
44			* with this, not servers)
45			*/
46
47			void clean_pathname(char *pathname)
48			{
49			char *cleanpath, c;
50
51			cleanpath = pathname;
52			while ((c = *pathname++)) {
53			if (c == '/') {
54			while (1) {
55			if (*pathname == '/')
56			pathname++;
57			else if (pathname == '.' && (pathname + 1) == '/')
58			pathname += 2;
59			else if (pathname == '.' && (pathname + 1) == '.' &&
60			*(pathname + 2) == '/') {
61			pathname += 3;
62			} else
63			break;
64			}
65			c = '/';
66			}
67			*cleanpath++ = c;
68			}
69
70			*cleanpath = '\0';
71			}
72
73			/*
74			* Name: get_commonlog_time
75			*
76			* Description: Returns the current time in common log format in a static
77			* char buffer.
78			*
79			* commonlog time is exactly 25 characters long
80			* because this is only used in logging, we add " [" before and "] " after
81			* making 29 characters
82			* "[27/Feb/1998:20:20:04 +0000] "
83			*
84			* Constrast with rfc822 time:
85			* "Sun, 06 Nov 1994 08:49:37 GMT"
86			*
87			* Altered 10 Jan 2000 by Jon Nelson ala Drew Streib for non UTC logging
88			*
89			*/
90
91	nmav	1.2	void get_commonlog_time(char buf[30])
92	nmav	1.1	{
93			struct tm *t;
94			char *p;
95			unsigned int a;
96			int time_offset;
97
98			if (use_localtime) {
99			t = localtime(&current_time);
100			time_offset = TIMEZONE_OFFSET(t);
101			} else {
102			t = gmtime(&current_time);
103			time_offset = 0;
104			}
105
106			p = buf + 29;
107			*p-- = '\0';
108			*p-- = ' ';
109			*p-- = ']';
110			a = abs(time_offset / 60);
111			*p-- = '0' + a % 10;
112			a /= 10;
113			*p-- = '0' + a % 6;
114			a /= 6;
115			*p-- = '0' + a % 10;
116			*p-- = '0' + a / 10;
117			*p-- = (time_offset >= 0) ? '+' : '-';
118			*p-- = ' ';
119
120			a = t->tm_sec;
121			*p-- = '0' + a % 10;
122			*p-- = '0' + a / 10;
123			*p-- = ':';
124			a = t->tm_min;
125			*p-- = '0' + a % 10;
126			*p-- = '0' + a / 10;
127			*p-- = ':';
128			a = t->tm_hour;
129			*p-- = '0' + a % 10;
130			*p-- = '0' + a / 10;
131			*p-- = ':';
132			a = 1900 + t->tm_year;
133			while (a) {
134			*p-- = '0' + a % 10;
135			a /= 10;
136			}
137			/* p points to an unused spot */
138			*p-- = '/';
139			p -= 2;
140			memcpy(p--, month_tab + 4 * (t->tm_mon), 3);
141			*p-- = '/';
142			a = t->tm_mday;
143			*p-- = '0' + a % 10;
144			*p-- = '0' + a / 10;
145			*p = '[';
146	nmav	1.2	return; /* should be same as returning buf */
147	nmav	1.1	}
148
149			/*
150			* Name: month2int
151			*
152			* Description: Turns a three letter month into a 0-11 int
153			*
154			* Note: This function is from wn-v1.07 -- it's clever and fast
155			*/
156
157			int month2int(char *monthname)
158			{
159			switch (*monthname) {
160			case 'A':
161			return (*++monthname == 'p' ? 3 : 7);
162			case 'D':
163			return (11);
164			case 'F':
165			return (1);
166			case 'J':
167			if (*++monthname == 'a')
168			return (0);
169			return (*++monthname == 'n' ? 5 : 6);
170			case 'M':
171			return (*(monthname + 2) == 'r' ? 2 : 4);
172			case 'N':
173			return (10);
174			case 'O':
175			return (9);
176			case 'S':
177			return (8);
178			default:
179			return (-1);
180			}
181			}
182
183			/*
184			* Name: modified_since
185			* Description: Decides whether a file's mtime is newer than the
186			* If-Modified-Since header of a request.
187			*
188
189			Sun, 06 Nov 1994 08:49:37 GMT ; RFC 822, updated by RFC 1123
190			Sunday, 06-Nov-94 08:49:37 GMT ; RFC 850, obsoleted by RFC 1036
191			Sun Nov 6 08:49:37 1994 ; ANSI C's asctime() format
192			31 September 2000 23:59:59 GMT ; non-standard
193
194			* RETURN VALUES:
195			* 0: File has not been modified since specified time.
196			* 1: File has been.
197			* -1: Error!
198			*/
199
200	nmav	1.10	int modified_since(time_t mtime, char *if_modified_since)
201	nmav	1.1	{
202			struct tm *file_gmt;
203			char *ims_info;
204			char monthname[10 + 1];
205			int day, month, year, hour, minute, second;
206			int comp;
207
208			ims_info = if_modified_since;
209			while (ims_info != ' ' && ims_info != '\0')
210			++ims_info;
211			if (*ims_info != ' ')
212			return -1;
213
214			/* the pre-space in the third scanf skips whitespace for the string */
215			if (sscanf(ims_info, "%d %3s %d %d:%d:%d GMT", /* RFC 1123 */
216			&day, monthname, &year, &hour, &minute, &second) == 6);
217			else if (sscanf(ims_info, "%d-%3s-%d %d:%d:%d GMT", /* RFC 1036 */
218			&day, monthname, &year, &hour, &minute, &second) == 6)
219			year += 1900;
220			else if (sscanf(ims_info, " %3s %d %d:%d:%d %d", /* asctime() format */
221			monthname, &day, &hour, &minute, &second, &year) == 6);
222			/* allow this non-standard date format: 31 September 2000 23:59:59 GMT */
223			/* NOTE: Use if_modified_since here, because the date starts
224			* with the day, versus a throwaway item
225			*/
226			else if (sscanf(if_modified_since, "%d %10s %d %d:%d:%d GMT",
227			&day, monthname, &year, &hour, &minute, &second) == 6);
228			else {
229			log_error_time();
230			fprintf(stderr, "Error in %s, line %d: Unable to sscanf \"%s\"\n",
231			__FILE__, __LINE__, ims_info);
232			return -1; /* error */
233			}
234
235	nmav	1.10	file_gmt = gmtime(&mtime);
236	nmav	1.1	month = month2int(monthname);
237
238			/* Go through from years to seconds -- if they are ever unequal,
239			we know which one is newer and can return */
240
241			if ((comp = 1900 + file_gmt->tm_year - year))
242			return (comp > 0);
243			if ((comp = file_gmt->tm_mon - month))
244			return (comp > 0);
245			if ((comp = file_gmt->tm_mday - day))
246			return (comp > 0);
247			if ((comp = file_gmt->tm_hour - hour))
248			return (comp > 0);
249			if ((comp = file_gmt->tm_min - minute))
250			return (comp > 0);
251			if ((comp = file_gmt->tm_sec - second))
252			return (comp > 0);
253
254			return 0; /* this person must really be into the latest/greatest */
255			}
256
257
258			/*
259			* Name: to_upper
260			*
261			* Description: Turns a string into all upper case (for HTTP_ header forming)
262			* AND changes - into _
263			*/
264
265			char to_upper(char str)
266			{
267			char *start = str;
268
269			while (*str) {
270			if (*str == '-')
271			*str = '_';
272			else
273			str = toupper(str);
274
275			str++;
276			}
277
278			return start;
279			}
280
281			/*
282			* Name: unescape_uri
283			*
284			* Description: Decodes a uri, changing %xx encodings with the actual
285			* character. The query_string should already be gone.
286			*
287			* Return values:
288			* 1: success
289			* 0: illegal string
290			*/
291
292			int unescape_uri(char uri, char * query_string)
293			{
294			char c, d;
295			char *uri_old;
296
297			uri_old = uri;
298
299			while ((c = *uri_old)) {
300			if (c == '%') {
301			uri_old++;
302			if ((c = uri_old++) && (d = uri_old++))
303			*uri++ = HEX_TO_DECIMAL(c, d);
304			else
305			return 0; /* NULL in chars to be decoded */
306			} else if (c == '?') { /* query string */
307			if (query_string)
308			*query_string = ++uri_old;
309			/* stop here */
310			*uri = '\0';
311			return(1);
312			break;
313			} else if (c == '#') { /* fragment */
314			/* legal part of URL, but we do not care.
315			* However, we still have to look for the query string */
316			if (query_string) {
317			++uri_old;
318			while((c = *uri_old)) {
319			if (c == '?') {
320			*query_string = ++uri_old;
321			break;
322			}
323			++uri_old;
324			}
325			}
326			break;
327			} else {
328			*uri++ = c;
329			uri_old++;
330			}
331			}
332
333			*uri = '\0';
334			return 1;
335			}
336
337			/* rfc822 (1123) time is exactly 29 characters long
338			* "Sun, 06 Nov 1994 08:49:37 GMT"
339			*/
340
341			void rfc822_time_buf(char *buf, time_t s)
342			{
343			struct tm *t;
344			char *p;
345			unsigned int a;
346
347			if (!s) {
348			t = gmtime(&current_time);
349			} else
350			t = gmtime(&s);
351
352			p = buf + 28;
353			/* p points to the last char in the buf */
354
355			p -= 3;
356			/* p points to where the ' ' will go */
357			memcpy(p--, " GMT", 4);
358
359			a = t->tm_sec;
360			*p-- = '0' + a % 10;
361			*p-- = '0' + a / 10;
362			*p-- = ':';
363			a = t->tm_min;
364			*p-- = '0' + a % 10;
365			*p-- = '0' + a / 10;
366			*p-- = ':';
367			a = t->tm_hour;
368			*p-- = '0' + a % 10;
369			*p-- = '0' + a / 10;
370			*p-- = ' ';
371			a = 1900 + t->tm_year;
372			while (a) {
373			*p-- = '0' + a % 10;
374			a /= 10;
375			}
376			/* p points to an unused spot to where the space will go */
377			p -= 3;
378			/* p points to where the first char of the month will go */
379			memcpy(p--, month_tab + 4 * (t->tm_mon), 4);
380			*p-- = ' ';
381			a = t->tm_mday;
382			*p-- = '0' + a % 10;
383			*p-- = '0' + a / 10;
384			*p-- = ' ';
385			p -= 3;
386			memcpy(p, day_tab + t->tm_wday * 4, 4);
387			}
388
389	nmav	1.11	/* Returns the number of digits */
390	nmav	1.17	int simple_itoa(large_int i, char buf[22])
391	nmav	1.1	{
392			/* 21 digits plus null terminator, good for 64-bit or smaller ints
393			* for bigger ints, use a bigger buffer!
394			*
395			* 4294967295 is, incidentally, MAX_UINT (on 32bit systems at this time)
396			* and is 10 bytes long
397			*/
398	nmav	1.2	char *p = &buf[21];
399	nmav	1.3	int digits = 1; /* include null char */
400
401	nmav	1.1	*p-- = '\0';
402			do {
403	nmav	1.3	digits++;
404	nmav	1.1	*p-- = '0' + i % 10;
405			i /= 10;
406			} while (i > 0);
407	nmav	1.2
408	nmav	1.3	p++;
409			if (p!=buf) memmove( buf, p, digits);
410
411	nmav	1.11	return digits - 1;
412	nmav	1.1	}
413	nmav	1.11
414			/* Generates an Etag, by using the file size, and the modification time
415			*/
416			int create_etag(unsigned long int size, unsigned long int mod_time,
417			char etag[MAX_ETAG_LENGTH])
418			{
419			char buf[22];
420			int len, len2;
421
422			etag[0] = '\"';
423			len = 1;
424
425	nmav	1.12	len2 = simple_itoa( size % 100000, buf); /* up to 5 digits */
426	nmav	1.11	memcpy( &etag[len], buf, len2);
427			len += len2;
428
429			etag[len++] = '-';
430	nmav	1.12	len2 = simple_itoa( mod_time % 100000, buf); /* also 5 digits */
431	nmav	1.11	memcpy( &etag[len], buf, len2);
432			len += len2;
433
434			etag[len++] = '\"';
435			etag[len] = 0; /* etag is null terminated */
436
437			return len;
438			}
439
440	nmav	1.1
441			/* I don't "do" negative conversions
442			* Therefore, -1 indicates error
443			*/
444
445	nmav	1.3	int boa_atoi(const char *s)
446	nmav	1.1	{
447			int retval;
448	nmav	1.2	char reconv[22];
449	nmav	1.1
450			if (!isdigit(*s))
451			return -1;
452
453	nmav	1.3	retval = atoi( s);
454	nmav	1.6	if (retval < 0)
455			return -1;
456
457			simple_itoa(retval, reconv);
458			if (memcmp(s,reconv,strlen(s)) != 0) {
459			return -1;
460			}
461			return retval;
462			}
463
464	nmav	1.17	large_int boa_atoll(const char *s)
465	nmav	1.6	{
466			long int retval;
467			char reconv[22];
468
469			if (!isdigit(*s))
470			return -1;
471
472	nmav	1.13	#ifdef HAVE_ATOLL
473			retval = atoll( s);
474			#else
475	nmav	1.6	retval = atol( s);
476	nmav	1.13	#endif
477	nmav	1.1	if (retval < 0)
478			return -1;
479
480	nmav	1.2	simple_itoa(retval, reconv);
481	nmav	1.1	if (memcmp(s,reconv,strlen(s)) != 0) {
482			return -1;
483			}
484			return retval;
485			}
486
487	nmav	1.8	#define TEMP_FILE_TEMPLATE "/hydra.temp.XXXXXX"
488			#define TEMP_FILE_TEMPLATE_LEN sizeof(TEMP_FILE_TEMPLATE)-1
489
490	nmav	1.17	void close_tmp_fd( tmp_fd* fds)
491			{
492			if (fds->pipe) {
493			close( fds->fds[0]);
494			close( fds->fds[1]);
495			} else close( fds->fds[0]);
496
497			fds->fds[0] = fds->fds[1] = -1;
498
499			}
500
501			const static tmp_fd EMPTY_FDS = { {-1, -1}, 0};
502
503	nmav	1.5	/* returns -1 on error */
504	nmav	1.17	/* size holds the number of data that will be written to
505			* the temporary file.
506			*/
507			tmp_fd create_temporary_file(short want_unlink, int size)
508	nmav	1.1	{
509	nmav	1.2	char boa_tempfile[MAX_PATH_LENGTH + 1];
510	nmav	1.17	tmp_fd fd;
511			int total_len;
512
513			fd.pipe = 0;
514
515			if (size > 0 && size < PIPE_BUF)
516			if (pipe( fd.fds) != -1) {
517			fd.pipe = 1;
518			return fd;
519			}
520	nmav	1.1
521	nmav	1.8	total_len = tempdir_len + TEMP_FILE_TEMPLATE_LEN;
522			if (total_len > MAX_PATH_LENGTH) {
523			log_error_time();
524			fprintf(stderr, "Temporary file length (%d) is too long\n", total_len);
525	nmav	1.17	return EMPTY_FDS;
526	nmav	1.8	}
527	nmav	1.1
528	nmav	1.8	memcpy( boa_tempfile, tempdir, tempdir_len);
529			memcpy( &boa_tempfile[tempdir_len], TEMP_FILE_TEMPLATE, TEMP_FILE_TEMPLATE_LEN);
530			boa_tempfile[total_len] = 0; /* null terminated */
531
532	nmav	1.17	fd.fds[0] = fd.fds[1] = -1;
533	nmav	1.8	/* open temp file
534			*/
535	nmav	1.17	fd.fds[0] = mkstemp(boa_tempfile);
536			if (fd.fds[0] == -1) {
537	nmav	1.1	log_error_time();
538			perror("mkstemp");
539	nmav	1.17	return EMPTY_FDS;
540	nmav	1.1	}
541
542			if (want_unlink) {
543			if (unlink(boa_tempfile) == -1) {
544	nmav	1.17	close(fd.fds[0]);
545			fd.fds[0] = -1;
546	nmav	1.1	log_error_time();
547			fprintf(stderr, "unlink temp file\n");
548			}
549			}
550
551	nmav	1.17	fd.fds[1] = fd.fds[0];
552	nmav	1.1	return (fd);
553			}
554
555	nmav	1.15	int set_block_fd(int fd)
556	nmav	1.1	{
557	nmav	1.15	int flags;
558	nmav	1.1
559	nmav	1.15	flags = fcntl(fd, F_GETFL);
560			if (flags == -1)
561			return -1;
562	nmav	1.1
563	nmav	1.15	flags &= ~O_NONBLOCK;
564			flags = fcntl(fd, F_SETFL, flags);
565			return flags;
566	nmav	1.1	}
567
568	nmav	1.15	int set_nonblock_fd(int fd)
569	nmav	1.1	{
570			int flags;
571
572			flags = fcntl(fd, F_GETFL);
573			if (flags == -1)
574			return -1;
575
576	nmav	1.15	flags \|= O_NONBLOCK;
577	nmav	1.1	flags = fcntl(fd, F_SETFL, flags);
578			return flags;
579			}
580
581	nmav	1.15	int set_cloexec_fd(int fd)
582	nmav	1.1	{
583			int flags;
584
585			flags = fcntl(fd, F_GETFL);
586			if (flags == -1)
587			return -1;
588
589	nmav	1.15	flags \|= FD_CLOEXEC;
590	nmav	1.1	flags = fcntl(fd, F_SETFL, flags);
591			return flags;
592	nmav	1.9	}
593
594			void create_url( char * buffer, int buffer_size, int secure,
595			const char* hostname, int port, const char* request_uri)
596			{
597			char * proto;
598			char str_port[23];
599			int do_port = 0;
600
601			buffer[0] = 0; /* in case we fail */
602	nmav	1.16	if (hostname == NULL) {
603			hostname = "(null)";
604			}
605
606			if (request_uri == NULL) {
607			request_uri = "(null)";
608			}
609	nmav	1.9
610			if (secure) {
611			proto = "https";
612			if (port != 443)
613			do_port = 1;
614			} else {
615			proto = "http";
616			if (port!=80)
617			do_port = 1;
618			}
619
620			if (do_port) {
621			str_port[0] = ':';
622			simple_itoa( port, &str_port[1]);
623			}
624
625			if ((strlen(proto) + strlen(str_port) + strlen(hostname) +
626			strlen(request_uri) + 5) > buffer_size) {
627			/* This is more than impossible. The buffer is long enough */
628			log_error_time();
629			fprintf(stderr, "Could not create URL. Buffer was not enough.\n");
630			return;
631			}
632
633			sprintf( buffer, "%s://%s%s%s/", proto, hostname, str_port, request_uri);
634
635			return;
636	nmav	1.14	}
637
638			/* Breaks a list of "xxx", "yyy", to a character array, of
639			* MAX_COMMA_SEP_ELEMENTS size; Note that the given string is modified.
640			*/
641			void break_comma_list(char *etag,
642			char *broken_etag[MAX_COMMA_SEP_ELEMENTS],
643			int *elements)
644			{
645			char *p = etag;
646
647			*elements = 0;
648
649			do {
650			broken_etag[*elements] = p;
651
652			(*elements)++;
653
654			p = strchr(p, ',');
655			if (p) {
656			*p = 0;
657			p++; /* move to next entry and skip white
658			* space.
659			*/
660			while (*p == ' ')
661			p++;
662			}
663			} while (p != NULL && *elements < MAX_COMMA_SEP_ELEMENTS);
664	nmav	1.1	}
665	nmav	1.15
666			/* Quoting from rfc1034:
667
668			<domain> ::= <subdomain> \| " "
669
670			<subdomain> ::= <label> \| <subdomain> "." <label>
671
672			<label> ::= <letter> [ [ <ldh-str> ] <let-dig> ]
673
674			<ldh-str> ::= <let-dig-hyp> \| <let-dig-hyp> <ldh-str>
675
676			<let-dig-hyp> ::= <let-dig> \| "-"
677
678			<let-dig> ::= <letter> \| <digit>
679
680			<letter> ::= any one of the 52 alphabetic characters A through Z in
681			upper case and a through z in lower case
682
683			<digit> ::= any one of the ten digits 0 through 9
684
685			and
686
687			The labels must follow the rules for ARPANET host names. They must
688			start with a letter, end with a letter or digit, and have as interior
689			characters only letters, digits, and hyphen. There are also some
690			restrictions on the length. Labels must be 63 characters or less.
691
692			*/
693
694			int check_host(char *r)
695			{
696			/* a hostname can only consist of
697			* chars and numbers, and sep. by only
698			* one period.
699			* It may not end with a period, and must
700			* not start with a number.
701			*
702			* >0: correct
703			* -1: error
704			* 0: not returned
705			*
706			*/
707			char *c;
708			short period_ok = 0;
709			short len = 0;
710
711			c = r;
712			if (c == NULL) {
713			return -1;
714			}
715
716			/* must start with a letter */
717			if (!isalpha(*c))
718			return -1;
719
720			len = 1;
721			while (*(++c) != '\0') {
722			/* interior letters may be alphanumeric, '-', or '.' */
723			/* '.' may not follow '.' */
724			if (isalnum(c) \|\| c == '-')
725			period_ok = 1;
726			else if (*c == '.' && period_ok)
727			period_ok = 0;
728			else
729			return -1;
730			++len;
731			}
732			/* c points to '\0' */
733			--c;
734			/* must end with a letter or digit */
735			if (!isalnum(*c))
736			return -1;
737			return len;
738			}
739
740