skip.c

Go to the documentation of this file.

/*
 *   This library is free software; you can redistribute it and/or
 *   modify it under the terms of the GNU Lesser General Public
 *   License as published by the Free Software Foundation; either
 *   version 2.1 of the License, or (at your option) any later version.
 *
 *   This library 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
 *   Lesser General Public License for more details.
 *
 *   You should have received a copy of the GNU Lesser General Public
 *   License along with this library; if not, write to the Free Software
 *   Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
 */
 
/** Preparse input by skipping known tokens
 *
 * @file src/lib/util/skip.c
 *
 * @copyright 2025 Network RADIUS SAS (legal@networkradius.com)
 */
RCSID("$Id: e9ac32fb36cfcd151d798ac5dc662dd355cb2679 $")
 
#include <freeradius-devel/util/misc.h>
#include <freeradius-devel/util/skip.h>
 
/**  Skip a quoted string.
 *
 *  @param[in] start    start of the string, pointing to the quotation character
 *  @param[in] end      end of the string (or NULL for zero-terminated strings)
 *  @return
 *      >0 length of the string which was parsed
 *      <=0 on error
 */
ssize_t fr_skip_string(char const *start, char const *end)
{
        char const *p = start;
        char quote;
 
        quote = *(p++);
 
        while ((end && (p < end)) || *p) {
                /*
                 *      Stop at the quotation character
                 */
                if (*p == quote) {
                        p++;
                        return p - start;
                }
 
                /*
                 *      Not an escape character: it's OK.
                 */
                if (*p != '\\') {
                        p++;
                        continue;
                }
 
                if (end && ((p + 2) >= end)) {
                fail:
                        fr_strerror_const("Unexpected escape at end of string");
                        return -(p - start);
                }
 
                /*
                 *      Escape at EOL is not allowed.
                 */
                if (p[1] < ' ') goto fail;
 
                /*
                 *      \r or \n, etc.
                 */
                if (!isdigit((uint8_t) p[1])) {
                        p += 2;
                        continue;
                }
 
                /*
                 *      Double-quoted strings use \000
                 *      Regexes use \0
                 */
                if (quote == '/') {
                        p++;
                        continue;
                }
 
                if (end && ((p + 4) >= end)) goto fail;
 
                /*
                 *      Allow for \1f in single quoted strings
                 */
                if ((quote == '\'') && isxdigit((uint8_t) p[1]) && isxdigit((uint8_t) p[2])) {
                        p += 3;
                        continue;
                }
 
                if (!isdigit((uint8_t) p[2]) || !isdigit((uint8_t) p[3])) {
                        fr_strerror_const("Invalid octal escape");
                        return -(p - start);
                }
 
                p += 4;
        }
 
        /*
         *      Unexpected end of string.
         */
        fr_strerror_const("Unexpected end of string");
        return -(p - start);
}
 
/*
 *      Recursion cap shared by fr_skip_brackets and fr_skip_xlat, which
 *      are mutually recursive. Real configs nest far below this; the
 *      cap exists so untrusted input (config-file fuzzer) can't exhaust
 *      the C stack via `((((...` or `${${${...`.
 */
#define SKIP_MAX_DEPTH          64
 
static ssize_t skip_brackets(char const *start, char const *end, char end_quote, unsigned int depth);
static ssize_t skip_xlat(char const *start, char const *end, unsigned int depth);
 
static ssize_t skip_brackets(char const *start, char const *end, char end_quote, unsigned int depth)
{
        ssize_t slen;
        char const *p = start;
 
        if (depth >= SKIP_MAX_DEPTH) {
                fr_strerror_const("Nesting too deep");
                return -(p - start);
        }
 
        while ((end && (p < end)) || *p) {
                if (*p == end_quote) {
                        p++;
                        return p - start;
                }
 
                /*
                 *      Expressions.  Arguably we want to
                 *      differentiate conditions and function
                 *      arguments, but it's not clear how to do that
                 *      in a pre-parsing stage.
                 */
                if (*p == '(') {
                        p++;
                        slen = skip_brackets(p, end, ')', depth + 1);
 
                next:
                        if (slen <= 0) return slen - (p - start);
 
                        fr_assert((size_t) slen <= (size_t) (end - p));
                        p += slen;
                        continue;
                }
 
                /*
                 *      A quoted string.
                 */
                if ((*p == '"') || (*p == '\'') || (*p == '`')) {
                        slen = fr_skip_string(p, end);
                        goto next;
                }
 
                /*
                 *      Nested expansion.
                 */
                if ((p[0] == '$') || (p[0] == '%')) {
                        if (end && (p + 2) >= end) break;
 
                        /*
                         *      %% inside of an xlat
                         */
                        if ((p[0] == '%') && (p[1] == '%')) {
                                p += 2;
                                continue;
                        }
 
                        if ((p[1] == '{') || (p[1] == '(')) {
                                slen = skip_xlat(p, end, depth + 1);
                                goto next;
                        }
 
                        /*
                         *      Bare $ or %, just leave it alone.
                         */
                        p++;
                        continue;
                }
 
                /*
                 *      Escapes are special.
                 */
                if (*p != '\\') {
                        p++;
                        continue;
                }
 
                if (end && ((p + 2) >= end)) break;
 
                /*
                 *      Escapes here are only one-character escapes.
                 */
                if (p[1] < ' ') break;
                p += 2;
        }
 
        /*
         *      Unexpected end of xlat
         */
        fr_strerror_const("Unexpected end of expansion");
        return -(p - start);
}
 
static ssize_t skip_xlat(char const *start, char const *end, unsigned int depth)
{
        ssize_t slen;
        char const *p = start;
 
        if (depth >= SKIP_MAX_DEPTH) {
                fr_strerror_const("Nesting too deep");
                return -(p - start);
        }
 
        /*
         *      At least %{1} or $(.)
         */
        if (end && ((end - start) < 4)) {
                fr_strerror_const("Invalid expansion");
                return 0;
        }
 
        if (!((memcmp(p, "%{", 2) == 0) || /* xlat */
              (memcmp(p, "${", 2) == 0) || /* config file macro */
              (memcmp(p, "$(", 2) == 0))) {  /* shell expansion in an back-ticks argument */
                fr_strerror_const("Invalid expansion");
                return 0;
        }
        p++;
 
        if (*p == '(') {
                p++;            /* skip the '(' */
                slen = skip_brackets(p, end, ')', depth + 1);
 
        } else if (*p == '{') {
                p++;            /* skip the '{' */
                slen = skip_brackets(p, end, '}', depth + 1);
 
        } else {
                char const *q = p;
 
                /*
                 *      New xlat syntax: %foo(...)
                 */
                while (isalnum((int) *q) || (*q == '.') || (*q == '_') || (*q == '-')) {
                        q++;
                }
 
                if (*q != '(') {
                        fr_strerror_const("Invalid character after '%'");
                        return -(p - start);
                }
 
                p = q + 1;
 
                slen = skip_brackets(p, end, ')', depth + 1);
        }
 
        if (slen <= 0) return slen - (p - start);
        return slen + (p - start);
}
 
/** Skip a generic {...} or (...) arguments
 *
 */
ssize_t fr_skip_brackets(char const *start, char const *end, char end_quote)
{
        return skip_brackets(start, end, end_quote, 0);
}
 
/**  Skip an xlat expression.
 *
 *  This is a simple "peek ahead" parser which tries to not be wrong.  It may accept
 *  some things which will later parse as invalid (e.g. unknown attributes, etc.)
 *  But it also rejects all malformed expressions.
 *
 *  It's used as a quick hack because the full parser isn't always available.
 *
 *  @param[in] start    start of the expression, MUST point to the "%{" or "%("
 *  @param[in] end      end of the string (or NULL for zero-terminated strings)
 *  @return
 *      >0 length of the string which was parsed
 *      <=0 on error
 */
ssize_t fr_skip_xlat(char const *start, char const *end)
{
        return skip_xlat(start, end, 0);
}
 
/**  Skip a conditional expression.
 *
 *  This is a simple "peek ahead" parser which tries to not be wrong.  It may accept
 *  some things which will later parse as invalid (e.g. unknown attributes, etc.)
 *  But it also rejects all malformed expressions.
 *
 *  It's used as a quick hack because the full parser isn't always available.
 *
 *  @param[in] start    start of the condition.
 *  @param[in] end      end of the string (or NULL for zero-terminated strings)
 *  @param[in] terminal terminal character(s)
 *  @param[out] eol     did the parse error happen at eol?
 *  @return
 *      >0 length of the string which was parsed.  *eol is false.
 *      <=0 on error, *eol may be set.
 */
ssize_t fr_skip_condition(char const *start, char const *end, bool const terminal[static SBUFF_CHAR_CLASS], bool *eol)
{
        char const *p = start;
        bool was_regex = false;
        int depth = 0;
        ssize_t slen;
 
        if (eol) *eol = false;
 
        /*
         *      Keep parsing the condition until we hit EOS or EOL.
         */
        while ((end && (p < end)) || *p) {
                if (isspace((uint8_t) *p)) {
                        p++;
                        continue;
                }
 
                /*
                 *      In the configuration files, conditions end with ") {" or just "{"
                 */
                if ((depth == 0) && terminal[(uint8_t) *p]) {
                        return p - start;
                }
 
                /*
                 *      "recurse" to get more conditions.
                 */
                if (*p == '(') {
                        p++;
                        depth++;
                        was_regex = false;
                        continue;
                }
 
                if (*p == ')') {
                        if (!depth) {
                                fr_strerror_const("Too many ')'");
                                return -(p - start);
                        }
 
                        p++;
                        depth--;
                        was_regex = false;
                        continue;
                }
 
                /*
                 *      Parse xlats.  They cannot span EOL.
                 */
                if ((*p == '$') || (*p == '%')) {
                        if (end && ((p + 2) >= end)) {
                                fr_strerror_const("Expansions cannot extend across end of line");
                                return -(p - start);
                        }
 
                        if ((p[1] == '{') || ((p[0] == '$') && (p[1] == '('))) {
                                slen = fr_skip_xlat(p, end);
 
                        check:
                                if (slen <= 0) return -(p - start) + slen;
 
                                p += slen;
                                continue;
                        }
 
                        /*
                         *      Bare $ or %, just leave it alone.
                         */
                        p++;
                        was_regex = false;
                        continue;
                }
 
                /*
                 *      Parse quoted strings.  They cannot span EOL.
                 */
                if ((*p == '"') || (*p == '\'') || (*p == '`') || (was_regex && (*p == '/'))) {
                        was_regex = false;
 
                        slen = fr_skip_string((char const *) p, end);
                        goto check;
                }
 
                /*
                 *      192.168/16 is a netmask.  So we only
                 *      allow regex after a regex operator.
                 *
                 *      This isn't perfect, but is good enough
                 *      for most purposes.
                 */
                if ((p[0] == '=') || (p[0] == '!')) {
                        if (end && ((p + 2) >= end)) {
                                fr_strerror_const("Operators cannot extend across end of line");
                                return -(p - start);
                        }
 
                        if (p[1] == '~') {
                                was_regex = true;
                                p += 2;
                                continue;
                        }
 
                        /*
                         *      Some other '==' or '!=', just leave it alone.
                         */
                        p++;
                        was_regex = false;
                        continue;
                }
 
                /*
                 *      Any control characters (other than \t) cause an error.
                 */
                if (*p < ' ') break;
 
                was_regex = false;
 
                /*
                 *      Normal characters just get skipped.
                 */
                if (*p != '\\') {
                        p++;
                        continue;
                }
 
                /*
                 *      Backslashes at EOL are ignored.
                 */
                if (end && ((p + 2) >= end)) break;
 
                /*
                 *      Escapes here are only one-character escapes.
                 */
                if (p[1] < ' ') break;
                p += 2;
        }
 
        /*
         *      We've fallen off of the end of a string.  It may be OK?
         */
        if (eol) *eol = (depth > 0);
 
        if (terminal[(uint8_t) *p]) return p - start;
 
        fr_strerror_const("Unexpected end of condition");
        return -(p - start);
}