1/* Licensed to the Apache Software Foundation (ASF) under one or more
2 * contributor license agreements.  See the NOTICE file distributed with
3 * this work for additional information regarding copyright ownership.
4 * The ASF licenses this file to You under the Apache License, Version 2.0
5 * (the "License"); you may not use this file except in compliance with
6 * the License.  You may obtain a copy of the License at
7 *
8 *     http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17#include "apr.h"
18#include "apr_strings.h"
19#include "apr_private.h"
20#include "apr_lib.h"
21
22#if APR_HAVE_SYS_TYPES_H
23#include <sys/types.h>
24#endif
25#if APR_HAVE_STRING_H
26#include <string.h>
27#endif
28#if APR_HAVE_CTYPE_H
29#include <ctype.h>
30#endif
31
32/*
33 * Apache's "replacement" for the strncpy() function. We roll our
34 * own to implement these specific changes:
35 *   (1) strncpy() doesn't always null terminate and we want it to.
36 *   (2) strncpy() null fills, which is bogus, esp. when copy 8byte
37 *       strings into 8k blocks.
38 *   (3) Instead of returning the pointer to the beginning of
39 *       the destination string, we return a pointer to the
40 *       terminating '\0' to allow us to "check" for truncation
41 *
42 * apr_cpystrn() follows the same call structure as strncpy().
43 */
44
45APR_DECLARE(char *) apr_cpystrn(char *dst, const char *src, apr_size_t dst_size)
46{
47
48    char *d, *end;
49
50    if (dst_size == 0) {
51        return (dst);
52    }
53
54    d = dst;
55    end = dst + dst_size - 1;
56
57    for (; d < end; ++d, ++src) {
58	if (!(*d = *src)) {
59	    return (d);
60	}
61    }
62
63    *d = '\0';	/* always null terminate */
64
65    return (d);
66}
67
68
69/*
70 * This function provides a way to parse a generic argument string
71 * into a standard argv[] form of argument list. It respects the
72 * usual "whitespace" and quoteing rules. In the future this could
73 * be expanded to include support for the apr_call_exec command line
74 * string processing (including converting '+' to ' ' and doing the
75 * url processing. It does not currently support this function.
76 *
77 *    token_context: Context from which pool allocations will occur.
78 *    arg_str:       Input argument string for conversion to argv[].
79 *    argv_out:      Output location. This is a pointer to an array
80 *                   of pointers to strings (ie. &(char *argv[]).
81 *                   This value will be allocated from the contexts
82 *                   pool and filled in with copies of the tokens
83 *                   found during parsing of the arg_str.
84 */
85APR_DECLARE(apr_status_t) apr_tokenize_to_argv(const char *arg_str,
86                                            char ***argv_out,
87                                            apr_pool_t *token_context)
88{
89    const char *cp;
90    const char *ct;
91    char *cleaned, *dirty;
92    int escaped;
93    int isquoted, numargs = 0, argnum;
94
95#define SKIP_WHITESPACE(cp) \
96    for ( ; *cp == ' ' || *cp == '\t'; ) { \
97        cp++; \
98    };
99
100#define CHECK_QUOTATION(cp,isquoted) \
101    isquoted = 0; \
102    if (*cp == '"') { \
103        isquoted = 1; \
104        cp++; \
105    } \
106    else if (*cp == '\'') { \
107        isquoted = 2; \
108        cp++; \
109    }
110
111/* DETERMINE_NEXTSTRING:
112 * At exit, cp will point to one of the following:  NULL, SPACE, TAB or QUOTE.
113 * NULL implies the argument string has been fully traversed.
114 */
115#define DETERMINE_NEXTSTRING(cp,isquoted) \
116    for ( ; *cp != '\0'; cp++) { \
117        if (   (*cp == '\\' && (*(cp+1) == ' ' || *(cp+1) == '\t' || \
118                                *(cp+1) == '"' || *(cp+1) == '\''))) { \
119            cp++; \
120            continue; \
121        } \
122        if (   (!isquoted && (*cp == ' ' || *cp == '\t')) \
123            || (isquoted == 1 && *cp == '"') \
124            || (isquoted == 2 && *cp == '\'')                 ) { \
125            break; \
126        } \
127    }
128
129/* REMOVE_ESCAPE_CHARS:
130 * Compresses the arg string to remove all of the '\' escape chars.
131 * The final argv strings should not have any extra escape chars in it.
132 */
133#define REMOVE_ESCAPE_CHARS(cleaned, dirty, escaped) \
134    escaped = 0; \
135    while(*dirty) { \
136        if (!escaped && *dirty == '\\') { \
137            escaped = 1; \
138        } \
139        else { \
140            escaped = 0; \
141            *cleaned++ = *dirty; \
142        } \
143        ++dirty; \
144    } \
145    *cleaned = 0;        /* last line of macro... */
146
147    cp = arg_str;
148    SKIP_WHITESPACE(cp);
149    ct = cp;
150
151    /* This is ugly and expensive, but if anyone wants to figure a
152     * way to support any number of args without counting and
153     * allocating, please go ahead and change the code.
154     *
155     * Must account for the trailing NULL arg.
156     */
157    numargs = 1;
158    while (*ct != '\0') {
159        CHECK_QUOTATION(ct, isquoted);
160        DETERMINE_NEXTSTRING(ct, isquoted);
161        if (*ct != '\0') {
162            ct++;
163        }
164        numargs++;
165        SKIP_WHITESPACE(ct);
166    }
167    *argv_out = apr_palloc(token_context, numargs * sizeof(char*));
168
169    /*  determine first argument */
170    for (argnum = 0; argnum < (numargs-1); argnum++) {
171        SKIP_WHITESPACE(cp);
172        CHECK_QUOTATION(cp, isquoted);
173        ct = cp;
174        DETERMINE_NEXTSTRING(cp, isquoted);
175        cp++;
176        (*argv_out)[argnum] = apr_palloc(token_context, cp - ct);
177        apr_cpystrn((*argv_out)[argnum], ct, cp - ct);
178        cleaned = dirty = (*argv_out)[argnum];
179        REMOVE_ESCAPE_CHARS(cleaned, dirty, escaped);
180    }
181    (*argv_out)[argnum] = NULL;
182
183    return APR_SUCCESS;
184}
185
186/* Filepath_name_get returns the final element of the pathname.
187 * Using the current platform's filename syntax.
188 *   "/foo/bar/gum" -> "gum"
189 *   "/foo/bar/gum/" -> ""
190 *   "gum" -> "gum"
191 *   "wi\\n32\\stuff" -> "stuff
192 *
193 * Corrected Win32 to accept "a/b\\stuff", "a:stuff"
194 */
195
196APR_DECLARE(const char *) apr_filepath_name_get(const char *pathname)
197{
198    const char path_separator = '/';
199    const char *s = strrchr(pathname, path_separator);
200
201#ifdef WIN32
202    const char path_separator_win = '\\';
203    const char drive_separator_win = ':';
204    const char *s2 = strrchr(pathname, path_separator_win);
205
206    if (s2 > s) s = s2;
207
208    if (!s) s = strrchr(pathname, drive_separator_win);
209#endif
210
211    return s ? ++s : pathname;
212}
213
214/* length of dest assumed >= length of src
215 * collapse in place (src == dest) is legal.
216 * returns terminating null ptr to dest string.
217 */
218APR_DECLARE(char *) apr_collapse_spaces(char *dest, const char *src)
219{
220    while (*src) {
221        if (!apr_isspace(*src))
222            *dest++ = *src;
223        ++src;
224    }
225    *dest = 0;
226    return (dest);
227}
228
229#if !APR_HAVE_STRDUP
230char *strdup(const char *str)
231{
232    char *sdup;
233    size_t len = strlen(str) + 1;
234
235    sdup = (char *) malloc(len);
236    memcpy(sdup, str, len);
237
238    return sdup;
239}
240#endif
241
242/* The following two routines were donated for SVR4 by Andreas Vogel */
243#if (!APR_HAVE_STRCASECMP && !APR_HAVE_STRICMP)
244int strcasecmp(const char *a, const char *b)
245{
246    const char *p = a;
247    const char *q = b;
248    for (p = a, q = b; *p && *q; p++, q++) {
249        int diff = apr_tolower(*p) - apr_tolower(*q);
250        if (diff)
251            return diff;
252    }
253    if (*p)
254        return 1;               /* p was longer than q */
255    if (*q)
256        return -1;              /* p was shorter than q */
257    return 0;                   /* Exact match */
258}
259
260#endif
261
262#if (!APR_HAVE_STRNCASECMP && !APR_HAVE_STRNICMP)
263int strncasecmp(const char *a, const char *b, size_t n)
264{
265    const char *p = a;
266    const char *q = b;
267
268    for (p = a, q = b; /*NOTHING */ ; p++, q++) {
269        int diff;
270        if (p == a + n)
271            return 0;           /*   Match up to n characters */
272        if (!(*p && *q))
273            return *p - *q;
274        diff = apr_tolower(*p) - apr_tolower(*q);
275        if (diff)
276            return diff;
277    }
278    /*NOTREACHED */
279}
280#endif
281
282/* The following routine was donated for UTS21 by dwd@bell-labs.com */
283#if (!APR_HAVE_STRSTR)
284char *strstr(char *s1, char *s2)
285{
286    char *p1, *p2;
287    if (*s2 == '\0') {
288        /* an empty s2 */
289        return(s1);
290    }
291    while((s1 = strchr(s1, *s2)) != NULL) {
292        /* found first character of s2, see if the rest matches */
293        p1 = s1;
294        p2 = s2;
295        while (*++p1 == *++p2) {
296            if (*p1 == '\0') {
297                /* both strings ended together */
298                return(s1);
299            }
300        }
301        if (*p2 == '\0') {
302            /* second string ended, a match */
303            break;
304        }
305        /* didn't find a match here, try starting at next character in s1 */
306        s1++;
307    }
308    return(s1);
309}
310#endif
311
312