/** BEGIN COPYRIGHT BLOCK
 * Copyright (C) 2009 Red Hat, Inc.
 * All rights reserved.
 *
 * License: GPL (version 3 or any later version).
 * See LICENSE for details.
 * END COPYRIGHT BLOCK **/

#ifdef HAVE_CONFIG_H
#include <config.h>
#endif

/* validate.c - syntax validation helper functions */

#include <stdio.h>
#include <string.h>
#include <sys/types.h>
#include "syntax.h"

/* Helper function for processing a 'keystring'.
 *
 * Returns 0 is the value between begin and end is a valid 'keystring'.
 * Returns non-zero if the value is not a valide 'keystring'.
 */
int
keystring_validate(
    const char *begin,
    const char *end)
{
    int rc = 0; /* assume the value is valid */
    const char *p = begin;

    if ((begin == NULL) || (end == NULL)) {
        rc = 1;
        goto exit;
    }

    /* Per RFC4512:
     *
     *   keystring = leadkeychar *keychar
     */
    if (IS_LEADKEYCHAR(*p)) {
        for (p++; p <= end; p++) {
            if (!IS_KEYCHAR(*p)) {
                rc = 1;
                goto exit;
            }
        }
    } else {
        rc = 1;
        goto exit;
    }

exit:
    return (rc);
}

/* Helper function for processing a 'numericoid'.
 *
 * Returns 0 is the value between begin and end is a valid 'numericoid'.
 * Returns non-zero if the value is not a valide 'numericoid'.
 */
int
numericoid_validate(
    const char *begin,
    const char *end)
{
    int rc = 0; /* assume the value is valid */
    int found_separator = 0;
    const char *p = NULL;

    if ((begin == NULL) || (end == NULL)) {
        rc = 1;
        goto exit;
    }

    /* Per RFC 4512:
     *
     *   numericoid = number 1*( DOT number )
     */

    /* one pass of this loop should process one element of the oid (number DOT) */
    for (p = begin; p <= end; p++) {
        if (IS_LDIGIT(*p)) {
            /* loop until we get to a separator char */
            while (*p != '.') {
                p++;
                if (p > end) {
                    /* ensure we got at least 2 elements */
                    if (!found_separator) {
                        rc = 1;
                        goto exit;
                    } else {
                        /* looks like a valid numericoid */
                        goto exit;
                    }
                } else if (*p == '.') {
                    /* we can not end with a '.' */
                    if (p == end) {
                        rc = 1;
                        goto exit;
                    } else {
                        found_separator = 1;
                    }
                } else if (!isdigit(*p)) {
                    rc = 1;
                    goto exit;
                }
            }
        } else if (*p == '0') {
            p++;
            if (p > end) {
                /* ensure we got at least 2 elements */
                if (!found_separator) {
                    rc = 1;
                    goto exit;
                } else {
                    /* looks like a valid numericoid */
                    goto exit;
                }
            } else if (*p != '.') {
                /* a leading 0 is not allowed unless the entire element is simply 0 */
                rc = 1;
                goto exit;
            }

            /* At this point, *p is '.'.  We can not end with a '.' */
            if (p == end) {
                rc = 1;
                goto exit;
            } else {
                found_separator = 1;
            }
        } else {
            rc = 1;
            goto exit;
        }
    }

exit:
    return (rc);
}

/* Helper to validate a single UTF-8 character.
 * It is assumed that the first byte of the character
 * is pointed to by begin.  This function will not read
 * past the byte pointed to by the end parameter.  The
 * last pointer will be filled in the the address of
 * the last byte of the validated character if the
 * character is valid, or the last byte processed
 * in the invalid case.
 *
 * Returns 0 if it is valid and non-zero otherwise. */
int
utf8char_validate(
    const char *begin,
    const char *end,
    const char **last)
{
    int rc = 0; /* Assume char is valid */
    const char *p = begin;

    if ((begin == NULL) || (end == NULL)) {
        rc = 1;
        goto exit;
    }

    /* Per RFC 4512:
     *
     *   UTF8  = UTF1 / UTFMB
     *   UTFMB = UTF2 / UTF3 / UTF4
     *   UTF0  = %x80-BF
     *   UTF1  = %x00-7F
     *   UTF2  = %xC2-DF UTF0
     *   UTF3  = %xE0 %xA0-BF UTF0 / %xE1-EC 2(UTF0) /
     *           %xED %x80-9F UTF0 / %xEE-EF 2(UTF0)
     *   UTF4  = %xF0 %x90-BF 2(UTF0) / %xF1-F3 3(UTF0) /
     *           %xF4 %x80-8F 2(UTF0)
     */

    /* If we have a single byte (ASCII) character, we
     * don't really have any work to do. */
    if (IS_UTF1(*p)) {
        goto exit;
    } else if (IS_UTF2(*p)) {
        /* Ensure that there is another byte
         * and that is is 'UTF0'. */
        if ((p == end) || !IS_UTF0(*(p + 1))) {
            rc = 1;
            goto exit;
        }

        /* Advance p so last is set correctly */
        p++;
    } else if (IS_UTF3(*p)) {
        /* Ensure that there are at least 2 more bytes. */
        if (end - p < 2) {
            rc = 1;
            goto exit;
        }

        /* The first byte determines what is legal for
         * the second byte. */
        if (*p == '\xE0') {
            /* The next byte must be %xA0-BF. */
            p++;
            if (((unsigned char)*p < (unsigned char)'\xA0') || ((unsigned char)*p > (unsigned char)'\xBF')) {
                rc = 1;
                goto exit;
            }
        } else if (*p == '\xED') {
            /* The next byte must be %x80-9F. */
            p++;
            if (((unsigned char)*p < (unsigned char)'\x80') || ((unsigned char)*p > (unsigned char)'\x9F')) {
                rc = 1;
                goto exit;
            }
        } else {
            /* The next byte must each be 'UTF0'. */
            p++;
            if (!IS_UTF0(*p)) {
                rc = 1;
                goto exit;
            }
        }

        /* The last byte must be 'UTF0'. */
        p++;
        if (!IS_UTF0(*p)) {
            rc = 1;
            goto exit;
        }
    } else if (IS_UTF4(*p)) {
        /* Ensure that there are at least 3 more bytes. */
        if (end - p < 3) {
            rc = 1;
            goto exit;
        }

        /* The first byte determines what is legal for
         * the second byte. */
        if (*p == '\xF0') {
            /* The next byte must be %x90-BF. */
            p++;
            if (((unsigned char)*p < (unsigned char)'\x90') || ((unsigned char)*p > (unsigned char)'\xBF')) {
                rc = 1;
                goto exit;
            }
        } else if (*p == '\xF4') {
            /* The next byte must be %x80-BF. */
            p++;
            if (((unsigned char)*p < (unsigned char)'\x80') || ((unsigned char)*p > (unsigned char)'\xBF')) {
                rc = 1;
                goto exit;
            }
        } else {
            /* The next byte must each be 'UTF0'. */
            p++;
            if (!IS_UTF0(*p)) {
                rc = 1;
                goto exit;
            }
        }

        /* The last 2 bytes must be 'UTF0'. */
        p++;
        if (!IS_UTF0(*p) || !IS_UTF0(*(p + 1))) {
            rc = 1;
            goto exit;
        }

        /* Advance the pointer so last is set correctly
         * when we return. */
        p++;
    } else {
        /* We found an illegal first byte. */
        rc = 1;
        goto exit;
    }

exit:
    if (last) {
        *last = (const char *)p;
    }
    return (rc);
}

/* Validates that a non '\0' terminated string is UTF8.  This
 * function will not read past the byte pointed to by the end
 * parameter.  The last pointer will be filled in to point to
 * the address of the last byte of the last validated character
 * if the string is valid, or the last byte processed in the
 * invalid case.
 *
 * Returns 0 if it is valid and non-zero otherwise. */
int
utf8string_validate(
    const char *begin,
    const char *end,
    const char **last)
{
    int rc = 0; /* Assume string is valid */
    const char *p = NULL;

    if ((begin == NULL) || (end == NULL)) {
        rc = 1;
        goto exit;
    }

    for (p = begin; p <= end; p++) {
        if ((rc = utf8char_validate(p, end, &p)) != 0) {
            goto exit;
        }
    }

    /* Adjust the pointer so last is set correctly for caller. */
    p--;

exit:
    if (last) {
        *last = p;
    }
    return (rc);
}

/*
 * Validates a distinguishedName as degined in RFC 4514.  Returns
 * 0 if the value from begin to end is a valid distinguishedName.
 * Returns 1 otherwise.
 */
int
distinguishedname_validate(
    const char *begin,
    const char *end)
{
    int rc = 0; /* Assume value is valid */
    const char *p = begin;
    const char *last = NULL;

    /* Per RFC 4514:
     *
     * distinguishedName = [ relativeDistinguishedName
     *     *( COMMA relativeDistinguishedName ) ]
     * relativeDistinguishedName = attributeTypeAndValue
     *     *( PLUS attributeTypeAndValue )
     * attributeTypeAndValue = attribyteType EQUALS attributeValue
     * attributeType = descr / numericoid
     * attributeValue = string / hexstring
     */

    /* Validate one RDN at a time in a loop. */
    while (p <= end) {
        if ((rc = rdn_validate(p, end, &last)) != 0) {
            goto exit;
        }
        p = last + 1;

        /* p should be pointing at a comma, or one past
         * the end of the entire dn value.  If we have
         * not reached the end, ensure that the next
         * character is a comma and that there is at
         * least another character after the comma. */
        if ((p <= end) && ((p == end) || (*p != ','))) {
            rc = 1;
            goto exit;
        }

        /* Advance the pointer past the comma so it
         * points at the beginning of the next RDN
         * (if there is one). */
        p++;
    }

exit:
    return rc;
}

/*
 * Helper function for validating a DN.  This function will validate
 * a single RDN.  If the RDN is valid, 0 will be returned, otherwise
 * non-zero will be returned. A pointer to the last character processed
 * will be set in the "last parameter.  This will be the end of the RDN
 * in the valid case, and the illegal character in the invalid case.
 */
int
rdn_validate(const char *begin, const char *end, const char **last)
{
    int rc = 0; /* Assume RDN is valid */
    int numericform = 0;
    char *separator = NULL;
    const char *p = begin;

    /* Find the '=', then use the helpers for descr and numericoid */
    if ((separator = PL_strnchr(p, '=', end - begin + 1)) == NULL) {
        rc = 1;
        goto exit;
    }

    /* Process an attribute type. The 'descr'
     * form must start with a 'leadkeychar'. */
    if (IS_LEADKEYCHAR(*p)) {
        if ((rc = keystring_validate(p, separator - 1))) {
            goto exit;
        }
        /* See if the 'numericoid' form is being used */
    } else if (isdigit(*p)) {
        numericform = 1;
        if ((rc = numericoid_validate(p, separator - 1))) {
            goto exit;
        }
    } else {
        rc = 1;
        goto exit;
    }

    /* Advance the pointer past the '=' and make sure
     * we're not past the end of the string. */
    p = separator + 1;
    if (p > end) {
        rc = 1;
        goto exit;
    }

    /* The value must be a 'hexstring' if the 'numericoid'
     * form of 'attributeType' is used.  Per RFC 4514:
     *
     *   hexstring = SHARP 1*hexpair
     *   hexpair = HEX HEX
     */
    if (numericform) {
        if ((p == end) || !IS_SHARP(*p)) {
            rc = 1;
            goto exit;
        }
        p++;
        /* The value must be a 'string' when the 'descr' form
     * of 'attributeType' is used.  Per RFC 4514:
     *
     *   string = [ ( leadchar / pair ) [ *( stringchar / pair )
     *      ( trailchar / pair ) ] ]
     *
     *   leadchar   = LUTF1 / UTFMB
     *   trailchar  = TUTF1 / UTFMB
     *   stringchar = SUTF1 / UTFMB
     *
     *   pair = ESC (ESC / special / hexpair )
     *   special = escaped / SPACE / SHARP / EQUALS
     *   escaped = DQUOTE / PLUS / COMMA / SEMI / LANGLE / RANGLE
     *   hexpair = HEX HEX
     */
    } else {
        /* Check the leadchar to see if anything illegal
         * is there.  We need to allow a 'pair' to get
         * through, so we'll assume that a '\' is the
         * start of a 'pair' for now. */
        if (IS_UTF1(*p) && !IS_ESC(*p) && !IS_LUTF1(*p)) {
            rc = 1;
            goto exit;
        }
    }

    /* Loop through string until we find the ',' separator, a '+'
     * char indicating a multi-value RDN, or we reach the end.  */
    while ((p <= end) && (*p != ',') && (*p != '+')) {
        if (numericform) {
            /* Process a single 'hexpair' */
            if ((p == end) || !isxdigit(*p) || !isxdigit(*p + 1)) {
                rc = 1;
                goto exit;
            }
            p = p + 2;
        } else {
            /* Check for a valid 'stringchar'.  We handle
             * multi-byte characters separately. */
            if (IS_UTF1(*p)) {
                /* If we're at the end, check if we have
                 * a valid 'trailchar'. */
                if ((p == end) && !IS_TUTF1(*p)) {
                    rc = 1;
                    goto exit;
                    /* Check for a 'pair'. */
                } else if (IS_ESC(*p)) {
                    /* We're guaranteed to still have at
                     * least one more character, so lets
                     * take a look at it. */
                    p++;
                    if (!IS_ESC(*p) && !IS_SPECIAL(*p)) {
                        /* The only thing valid now
                         * is a 'hexpair'. */
                        if ((p == end) || !isxdigit(*p) || !isxdigit(*p + 1)) {
                            rc = 1;
                            goto exit;
                        }
                        p++;
                    }
                    /* Only allow 'SUTF1' chars now. */
                } else if (!IS_SUTF1(*p)) {
                    rc = 1;
                    goto exit;
                }

                p++;
            } else {
                /* Validate a single 'UTFMB' (multi-byte) character. */
                if (utf8char_validate(p, end, &p) != 0) {
                    rc = 1;
                    goto exit;
                }

                /* Advance the pointer past the multi-byte char. */
                p++;
            }
        }
    }

    /* We'll end up either at the comma, a '+', or one past end.
     * If we are processing a multi-valued RDN, we recurse to
     * process the next 'attributeTypeAndValue'. */
    if ((p <= end) && (*p == '+')) {
        /* Make sure that there is something after the '+'. */
        if (p == end) {
            rc = 1;
            goto exit;
        }
        p++;

        /* Recurse to process the next value.  We need to reset p to
         * ensure that last is set correctly for the original caller. */
        rc = rdn_validate(p, end, last);
        p = *last + 1;
    }

exit:
    *last = p - 1;
    return rc;
}

int
bitstring_validate_internal(const char *begin, const char *end)
{
    int rc = 0; /* assume the value is valid */
    const char *p = NULL;

    /* Per RFC4517:
         *
         * BitString    = SQUOTE *binary-digit SQUOTE "B"
         * binary-digit = "0" / "1"
         */

    /* Check that the value starts with a SQUOTE and
         * ends with SQUOTE "B". */
    if (!IS_SQUOTE(*begin) || (*end != 'B') ||
        !IS_SQUOTE(*(end - 1))) {
        rc = 1;
        goto exit;
    }

    /* Ensure that only '0' and '1' are between the SQUOTE chars. */
    for (p = begin + 1; p <= end - 2; p++) {
        if ((*p != '0') && (*p != '1')) {
            rc = 1;
            goto exit;
        }
    }

exit:
    return rc;
}