octstr.c.debug

The Kannel Open Source WAP and SMS gateway works as both an SMS gateway, for implementing keyword b

/*
 * octstr.c - implementation of Octet strings
 *
 * See octstr.h for explanations of what public functions should do.
 *
 * Lars Wirzenius
 */


#include <ctype.h>
#include <limits.h>
#include <errno.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <netinet/in.h>

#include "gwlib.h"


/***********************************************************************
 * Definitions of data structures. These are not visible to the external
 * world -- they may be accessed only via the functions declared in
 * octstr.h. This ensures they really are abstract.
 */

/*
 * The octet string.
 *
 * `data' is a pointer to dynamically allocated memory are where the 
 * octets in the string. It may be bigger than the actual length of the
 * string.
 *
 * `len' is the length of the string.
 *
 * `size' is the size of the memory area `data' points at.
 *
 * When `size' is greater than zero, it is at least `len+1', and the
 * character at `len' is '\0'. This is so that octstr_get_cstr will
 * always work.
 *
 * `immutable' defines whether the octet string is immutable or not.
 */
struct Octstr
{
    unsigned char *data;
    long len;
    long size;
    int immutable;
};


/**********************************************************************
 * Hash table of immutable octet strings.
 */

#define MAX_IMMUTABLES 1024

static Octstr *immutables[MAX_IMMUTABLES];
static Mutex immutables_mutex;
static int immutables_init = 0;

static char is_safe[UCHAR_MAX + 1];

/*
 * Convert a pointer to a C string literal to a long that can be used
 * for hashing. This is done by converting the pointer into an integer
 * and discarding the lowest to bits to get rid of typical alignment
 * bits.
 */
#define CSTR_TO_LONG(ptr)	(((long) ptr) >> 2)

/***********************************************************************
 * Declarations of internal functions. These are defined at the end of
 * the file.
 */


static void seems_valid_real(Octstr *ostr, const char *filename, long lineno,
                             const char *function);
#ifdef NO_GWASSERT
#define seems_valid(ostr)
#else
#define seems_valid(ostr) \
    (seems_valid_real(ostr, __FILE__, __LINE__, __func__))
#endif


/***********************************************************************
 * Implementations of the functions declared in octstr.h. See the
 * header for explanations of what they should do.
 */


/* Reserve space for at least 'size' octets */
static void octstr_grow(Octstr *ostr, long size)
{
    gw_assert(!ostr->immutable);
    seems_valid(ostr);
    gw_assert(size >= 0);

    size++;   /* make room for the invisible terminating NUL */

    if (size > ostr->size) {
        ostr->data = gw_realloc(ostr->data, size);
        ostr->size = size;
    }
}


/*
 * Fill is_safe table. is_safe[c] means that c can be left as such when
 * url-encoded.
 * RFC 2396 defines the list of characters that need to be encoded.
 * Space is treated as an exception by the encoding routine;
 * it's listed as safe here, but is actually changed to '+'.
 */
static void urlcode_init(void)
{
    int i;

    unsigned char *safe = " 0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
	"abcdefghijklmnopqrstuvwxyz-_.!~*'()";
    for (i = 0; safe[i] != '\0'; ++i)
	is_safe[safe[i]] = 1;
}


void octstr_init(void)
{
    urlcode_init();
    mutex_init_static(&immutables_mutex);
    immutables_init = 1;
}


void octstr_shutdown(void)
{
    long i, n;

    n = 0;
    for (i = 0; i < MAX_IMMUTABLES; ++i) {
        if (immutables[i] != NULL) {
	    gw_free(immutables[i]);
            ++n;
        }
    }
    debug("gwlib.octstr", 0, "Immutable octet strings: %ld.", n);
    mutex_destroy(&immutables_mutex);
}


Octstr *octstr_create_real(const char *cstr)
{
    gw_assert(cstr != NULL);
    return octstr_create_from_data(cstr, strlen(cstr));
}


Octstr *octstr_create_from_data_real(const char *data, long len)
{
    Octstr *ostr;

    gw_assert(len >= 0);
    if (data == NULL)
        gw_assert(len == 0);

    ostr = gw_malloc(sizeof(*ostr));
    if (len == 0) {
        ostr->len = 0;
        ostr->size = 0;
        ostr->data = NULL;
    } else {
        ostr->len = len;
        ostr->size = len + 1;
        ostr->data = gw_malloc(ostr->size);
        memcpy(ostr->data, data, len);
        ostr->data[len] = '\0';
    }
    ostr->immutable = 0;
    seems_valid(ostr);
    return ostr;
}


Octstr *octstr_imm(const char *cstr)
{
    Octstr *os;
    long i, index;
    unsigned char *data;

    gw_assert(immutables_init);
    gw_assert(cstr != NULL);

    index = CSTR_TO_LONG(cstr) % MAX_IMMUTABLES;
    data = (unsigned char *) cstr;

    mutex_lock(&immutables_mutex);
    i = index;
    for (; ; ) {
	if (immutables[i] == NULL || immutables[i]->data == data)
            break;
        i = (i + 1) % MAX_IMMUTABLES;
        if (i == index)
            panic(0, "Too many immutable strings.");
    }
    os = immutables[i];
    if (os == NULL) {
	/*
	 * Can't use octstr_create() because it copies the string,
	 * which would break our hashing.
	 */
	os = gw_malloc(sizeof(*os));
        os->data = data;
        os->len = strlen(data);
        os->size = os->len + 1;
        os->immutable = 1;
	immutables[i] = os;
	seems_valid(os);
    }
    mutex_unlock(&immutables_mutex);

    return os;
}


void octstr_destroy_impl(Octstr *ostr, const char *file,
			 long line, const char *func)
{
    if (ostr != NULL) {
	seems_valid_real(ostr, file, line, func);
	if (!ostr->immutable) {
            gw_free(ostr->data);
            gw_free(ostr);
        }
    }
}


void octstr_destroy_item(void *os)
{
    octstr_destroy(os);
}


long octstr_len(Octstr *ostr)
{
    if (ostr == NULL)
        return 0;
    seems_valid(ostr);
    return ostr->len;
}


Octstr *octstr_copy_real(Octstr *ostr, long from, long len)
{
    seems_valid(ostr);
    gw_assert(from >= 0);
    gw_assert(len >= 0);

    if (from >= ostr->len)
        return octstr_create("");

    if (len > ostr->len - from)
        len = ostr->len - from;

    return octstr_create_from_data(ostr->data + from, len);
}



Octstr *octstr_duplicate_real(Octstr *ostr)
{
    if (ostr == NULL)
        return NULL;
    seems_valid(ostr);
    return octstr_create_from_data(ostr->data, ostr->len);
}


Octstr *octstr_cat(Octstr *ostr1, Octstr *ostr2)
{
    Octstr *ostr;

    seems_valid(ostr1);
    seems_valid(ostr2);
    gw_assert(!ostr1->immutable);

    ostr = octstr_create("");
    ostr->len = ostr1->len + ostr2->len;
    ostr->size = ostr->len + 1;
    ostr->data = gw_malloc(ostr->size);

    if (ostr1->len > 0)
        memcpy(ostr->data, ostr1->data, ostr1->len);
    if (ostr2->len > 0)
        memcpy(ostr->data + ostr1->len, ostr2->data, ostr2->len);
    ostr->data[ostr->len] = '\0';

    seems_valid(ostr);
    return ostr;
}


int octstr_get_char(Octstr *ostr, long pos)
{
    seems_valid(ostr);
    if (pos >= ostr->len || pos < 0)
        return -1;
    return ostr->data[pos];
}


void octstr_set_char(Octstr *ostr, long pos, int ch)
{
    seems_valid(ostr);
    gw_assert(!ostr->immutable);
    if (pos < ostr->len)
        ostr->data[pos] = ch;
    seems_valid(ostr);
}


void octstr_get_many_chars(char *buf, Octstr *ostr, long pos, long len)
{
    gw_assert(buf != NULL);
    seems_valid(ostr);

    if (pos >= ostr->len)
        return;
    if (pos + len > ostr->len)
        len = ostr->len - pos;
    if (len > 0)
        memcpy(buf, ostr->data + pos, len);
}


char *octstr_get_cstr_real(Octstr *ostr, const char *file, long line, 
    	    	    	   const char *func)
{
    seems_valid_real(ostr, file, line, func);
    if (ostr->len == 0)
        return "";
    return ostr->data;
}


void octstr_append_from_hex(Octstr *ostr, char *hex)
{
    Octstr *output;
	
    seems_valid(ostr);
    gw_assert(!ostr->immutable);
	
    output = octstr_create(hex);
    octstr_hex_to_binary(output);
    octstr_append(ostr, output);
    octstr_destroy(output);
}


void octstr_binary_to_hex(Octstr *ostr, int uppercase)
{
    unsigned char *hexits;
    long i;

    seems_valid(ostr);
    gw_assert(!ostr->immutable);
    if (ostr->len == 0)
        return;

    hexits = uppercase ? "0123456789ABCDEF" : "0123456789abcdef";
    octstr_grow(ostr, ostr->len * 2);

    /* In-place modification must be done back-to-front to avoid
     * overwriting the data while we read it.  Even the order of
     * the two assignments is important, to get i == 0 right. */
    for (i = ostr->len - 1; i >= 0; i--) {
        ostr->data[i * 2 + 1] = hexits[ostr->data[i] % 16];
        ostr->data[i * 2] = hexits[(ostr->data[i] / 16) & 0xf];
    }

    ostr->len = ostr->len * 2;
    ostr->data[ostr->len] = '\0';

    seems_valid(ostr);
}


int octstr_hex_to_binary(Octstr *ostr)
{
    long len, i;
    unsigned char *p;

    seems_valid(ostr);
    gw_assert(!ostr->immutable);

    if (ostr->len == 0)
        return 0;

    /* Check if it's in the right format */
    if (!octstr_check_range(ostr, 0, ostr->len, gw_isxdigit))
        return -1;

    len = ostr->len;

    /* Convert ascii data to binary values */
    for (i = 0, p = ostr->data; i < len; i++, p++) {
        if (*p >= '0' && *p <= '9')
            *p -= '0';
        else if (*p >= 'a' && *p <= 'f')
            *p = *p - 'a' + 10;
        else if (*p >= 'A' && *p <= 'F')
            *p = *p - 'A' + 10;
        else {
            /* isxdigit checked the whole string, so we should
             * not be able to get here. */
            gw_assert(0);
            *p = 0;
        }
    }

    /* De-hexing will compress data by factor of 2 */
    len = ostr->len / 2;

    for (i = 0; i < len; i++) {
        ostr->data[i] = ostr->data[i * 2] * 16 | ostr->data[i * 2 + 1];
    }

    ostr->len = len;
    ostr->data[len] = '\0';

    seems_valid(ostr);
    return 0;
}


void octstr_binary_to_base64(Octstr *ostr)
{
    static const unsigned char base64[64] =
        "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
    long triplets;
    long lines;
    long orig_len;
    unsigned char *data;
    long from, to;
    int left_on_line;

    seems_valid(ostr);
    gw_assert(!ostr->immutable);

    if (ostr->len == 0) {
        /* Always terminate with CR LF */
        octstr_insert(ostr, octstr_imm("\015\012"), 0);
        return;
    }

    /* The lines must be 76 characters each (or less), and each
     * triplet will expand to 4 characters, so we can fit 19
     * triplets on one line.  We need a CR LF after each line,
     * which will add 2 octets per 19 triplets (rounded up). */
    triplets = (ostr->len + 2) / 3;   /* round up */
    lines = (triplets + 18) / 19;

    octstr_grow(ostr, triplets * 4 + lines * 2);
    orig_len = ostr->len;
    data = ostr->data;

    ostr->len = triplets * 4 + lines * 2;
    data[ostr->len] = '\0';

    /* This function works back-to-front, so that encoded data will
     * not overwrite source data.
     * from points to the start of the last triplet (which may be
     * an odd-sized one), and to points to the start of where the
     * last quad should go.  */
    from = (triplets - 1) * 3;
    to = (triplets - 1) * 4 + (lines - 1) * 2;

    /* First write the CR LF after the last quad */
    data[to + 5] = 10;   /* LF */
    data[to + 4] = 13;   /* CR */
    left_on_line = triplets - ((lines - 1) * 19);

    /* base64 encoding is in 3-octet units.  To handle leftover
     * octets, conceptually we have to zero-pad up to the next
     * 6-bit unit, and pad with '=' characters for missing 6-bit
     * units.
     * We do it by first completing the first triplet with 
     * zero-octets, and after the loop replacing some of the
     * result characters with '=' characters.
     * There is enough room for this, because even with a 1 or 2
     * octet source string, space for four octets of output
     * will be reserved.
     */
    switch (orig_len % 3) {
    case 0:
        break;
    case 1:
        data[orig_len] = 0;
        data[orig_len + 1] = 0;
        break;
    case 2:
        data[orig_len + 1] = 0;
        break;
    }

    /* Now we only have perfect triplets. */
    while (from >= 0) {
        long whole_triplet;

        /* Add a newline, if necessary */
        if (left_on_line == 0) {
            to -= 2;
            data[to + 5] = 10;  /* LF */
            data[to + 4] = 13;  /* CR */
            left_on_line = 19;
        }

        whole_triplet = (data[from] << 16) |
                        (data[from + 1] << 8) |
                        data[from + 2];
        data[to + 3] = base64[whole_triplet % 64];
        data[to + 2] = base64[(whole_triplet >> 6) % 64];
        data[to + 1] = base64[(whole_triplet >> 12) % 64];
        data[to] = base64[(whole_triplet >> 18) % 64];

        to -= 4;
        from -= 3;
        left_on_line--;
    }

    gw_assert(left_on_line == 0);
    gw_assert(from == -3);
    gw_assert(to == -4);

    /* Insert padding characters in the last quad.  Remember that
     * there is a CR LF between the last quad and the end of the
     * string. */
    switch (orig_len % 3) {
    case 0:
        break;
    case 1:
        gw_assert(data[ostr->len - 3] == 'A');
        gw_assert(data[ostr->len - 4] == 'A');
        data[ostr->len - 3] = '=';
        data[ostr->len - 4] = '=';
        break;
    case 2:
        gw_assert(data[ostr->len - 3] == 'A');
        data[ostr->len - 3] = '=';
        break;
    }

    seems_valid(ostr);
}