xref: /freebsd/contrib/openpam/lib/libpam/openpam_readword.c (revision 49e56509dbc8fd972cf527b2966bf56ce9c0726e)
1f7e6344dSDag-Erling Smørgrav /*-
2*49e56509SDag-Erling Smørgrav  * Copyright (c) 2012-2017 Dag-Erling Smørgrav
3f7e6344dSDag-Erling Smørgrav  * All rights reserved.
4f7e6344dSDag-Erling Smørgrav  *
5f7e6344dSDag-Erling Smørgrav  * Redistribution and use in source and binary forms, with or without
6f7e6344dSDag-Erling Smørgrav  * modification, are permitted provided that the following conditions
7f7e6344dSDag-Erling Smørgrav  * are met:
8f7e6344dSDag-Erling Smørgrav  * 1. Redistributions of source code must retain the above copyright
9ce77a8d6SDag-Erling Smørgrav  *    notice, this list of conditions and the following disclaimer.
10f7e6344dSDag-Erling Smørgrav  * 2. Redistributions in binary form must reproduce the above copyright
11f7e6344dSDag-Erling Smørgrav  *    notice, this list of conditions and the following disclaimer in the
12f7e6344dSDag-Erling Smørgrav  *    documentation and/or other materials provided with the distribution.
13f7e6344dSDag-Erling Smørgrav  * 3. The name of the author may not be used to endorse or promote
14f7e6344dSDag-Erling Smørgrav  *    products derived from this software without specific prior written
15f7e6344dSDag-Erling Smørgrav  *    permission.
16f7e6344dSDag-Erling Smørgrav  *
17f7e6344dSDag-Erling Smørgrav  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
18f7e6344dSDag-Erling Smørgrav  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19f7e6344dSDag-Erling Smørgrav  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20f7e6344dSDag-Erling Smørgrav  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
21f7e6344dSDag-Erling Smørgrav  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22f7e6344dSDag-Erling Smørgrav  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23f7e6344dSDag-Erling Smørgrav  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24f7e6344dSDag-Erling Smørgrav  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25f7e6344dSDag-Erling Smørgrav  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26f7e6344dSDag-Erling Smørgrav  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27f7e6344dSDag-Erling Smørgrav  * SUCH DAMAGE.
28f7e6344dSDag-Erling Smørgrav  *
29*49e56509SDag-Erling Smørgrav  * $OpenPAM: openpam_readword.c 938 2017-04-30 21:34:42Z des $
30f7e6344dSDag-Erling Smørgrav  */
31f7e6344dSDag-Erling Smørgrav 
32f7e6344dSDag-Erling Smørgrav #ifdef HAVE_CONFIG_H
33f7e6344dSDag-Erling Smørgrav # include "config.h"
34f7e6344dSDag-Erling Smørgrav #endif
35f7e6344dSDag-Erling Smørgrav 
36f7e6344dSDag-Erling Smørgrav #include <errno.h>
37f7e6344dSDag-Erling Smørgrav #include <stdio.h>
38f7e6344dSDag-Erling Smørgrav #include <stdlib.h>
39f7e6344dSDag-Erling Smørgrav 
40f7e6344dSDag-Erling Smørgrav #include <security/pam_appl.h>
41f7e6344dSDag-Erling Smørgrav 
42f7e6344dSDag-Erling Smørgrav #include "openpam_impl.h"
43f7e6344dSDag-Erling Smørgrav #include "openpam_ctype.h"
44f7e6344dSDag-Erling Smørgrav 
45f7e6344dSDag-Erling Smørgrav #define MIN_WORD_SIZE	32
46f7e6344dSDag-Erling Smørgrav 
47f7e6344dSDag-Erling Smørgrav /*
48f7e6344dSDag-Erling Smørgrav  * OpenPAM extension
49f7e6344dSDag-Erling Smørgrav  *
50f7e6344dSDag-Erling Smørgrav  * Read a word from a file, respecting shell quoting rules.
51f7e6344dSDag-Erling Smørgrav  */
52f7e6344dSDag-Erling Smørgrav 
53f7e6344dSDag-Erling Smørgrav char *
54f7e6344dSDag-Erling Smørgrav openpam_readword(FILE *f, int *lineno, size_t *lenp)
55f7e6344dSDag-Erling Smørgrav {
56f7e6344dSDag-Erling Smørgrav 	char *word;
57f7e6344dSDag-Erling Smørgrav 	size_t size, len;
5849e2762fSDag-Erling Smørgrav 	int ch, escape, quote;
59f7e6344dSDag-Erling Smørgrav 	int serrno;
60f7e6344dSDag-Erling Smørgrav 
61f7e6344dSDag-Erling Smørgrav 	errno = 0;
62f7e6344dSDag-Erling Smørgrav 
63f7e6344dSDag-Erling Smørgrav 	/* skip initial whitespace */
6449e2762fSDag-Erling Smørgrav 	escape = quote = 0;
6549e2762fSDag-Erling Smørgrav 	while ((ch = getc(f)) != EOF) {
6649e2762fSDag-Erling Smørgrav 		if (ch == '\n') {
6749e2762fSDag-Erling Smørgrav 			/* either EOL or line continuation */
6849e2762fSDag-Erling Smørgrav 			if (!escape)
69f7e6344dSDag-Erling Smørgrav 				break;
7049e2762fSDag-Erling Smørgrav 			if (lineno != NULL)
7149e2762fSDag-Erling Smørgrav 				++*lineno;
7249e2762fSDag-Erling Smørgrav 			escape = 0;
7349e2762fSDag-Erling Smørgrav 		} else if (escape) {
7449e2762fSDag-Erling Smørgrav 			/* escaped something else */
7549e2762fSDag-Erling Smørgrav 			break;
7649e2762fSDag-Erling Smørgrav 		} else if (ch == '#') {
7749e2762fSDag-Erling Smørgrav 			/* comment: until EOL, no continuation */
7849e2762fSDag-Erling Smørgrav 			while ((ch = getc(f)) != EOF)
7949e2762fSDag-Erling Smørgrav 				if (ch == '\n')
8049e2762fSDag-Erling Smørgrav 					break;
8149e2762fSDag-Erling Smørgrav 			break;
8249e2762fSDag-Erling Smørgrav 		} else if (ch == '\\') {
8349e2762fSDag-Erling Smørgrav 			escape = 1;
8449e2762fSDag-Erling Smørgrav 		} else if (!is_ws(ch)) {
8549e2762fSDag-Erling Smørgrav 			break;
8649e2762fSDag-Erling Smørgrav 		}
87f7e6344dSDag-Erling Smørgrav 	}
88f7e6344dSDag-Erling Smørgrav 	if (ch == EOF)
89f7e6344dSDag-Erling Smørgrav 		return (NULL);
90f7e6344dSDag-Erling Smørgrav 	ungetc(ch, f);
91f7e6344dSDag-Erling Smørgrav 	if (ch == '\n')
92f7e6344dSDag-Erling Smørgrav 		return (NULL);
93f7e6344dSDag-Erling Smørgrav 
94f7e6344dSDag-Erling Smørgrav 	word = NULL;
95f7e6344dSDag-Erling Smørgrav 	size = len = 0;
96f7e6344dSDag-Erling Smørgrav 	while ((ch = fgetc(f)) != EOF && (!is_ws(ch) || quote || escape)) {
97f7e6344dSDag-Erling Smørgrav 		if (ch == '\\' && !escape && quote != '\'') {
98f7e6344dSDag-Erling Smørgrav 			/* escape next character */
99f7e6344dSDag-Erling Smørgrav 			escape = ch;
100f7e6344dSDag-Erling Smørgrav 		} else if ((ch == '\'' || ch == '"') && !quote && !escape) {
101f7e6344dSDag-Erling Smørgrav 			/* begin quote */
102f7e6344dSDag-Erling Smørgrav 			quote = ch;
103f7e6344dSDag-Erling Smørgrav 			/* edge case: empty quoted string */
104f7e6344dSDag-Erling Smørgrav 			if (openpam_straddch(&word, &size, &len, 0) != 0)
105f7e6344dSDag-Erling Smørgrav 				return (NULL);
106f7e6344dSDag-Erling Smørgrav 		} else if (ch == quote && !escape) {
107f7e6344dSDag-Erling Smørgrav 			/* end quote */
108f7e6344dSDag-Erling Smørgrav 			quote = 0;
10949e2762fSDag-Erling Smørgrav 		} else if (ch == '\n' && escape) {
110f7e6344dSDag-Erling Smørgrav 			/* line continuation */
111f7e6344dSDag-Erling Smørgrav 			escape = 0;
112f7e6344dSDag-Erling Smørgrav 		} else {
113f7e6344dSDag-Erling Smørgrav 			if (escape && quote && ch != '\\' && ch != quote &&
114f7e6344dSDag-Erling Smørgrav 			    openpam_straddch(&word, &size, &len, '\\') != 0) {
115f7e6344dSDag-Erling Smørgrav 				free(word);
116f7e6344dSDag-Erling Smørgrav 				errno = ENOMEM;
117f7e6344dSDag-Erling Smørgrav 				return (NULL);
118f7e6344dSDag-Erling Smørgrav 			}
119f7e6344dSDag-Erling Smørgrav 			if (openpam_straddch(&word, &size, &len, ch) != 0) {
120f7e6344dSDag-Erling Smørgrav 				free(word);
121f7e6344dSDag-Erling Smørgrav 				errno = ENOMEM;
122f7e6344dSDag-Erling Smørgrav 				return (NULL);
123f7e6344dSDag-Erling Smørgrav 			}
124f7e6344dSDag-Erling Smørgrav 			escape = 0;
125f7e6344dSDag-Erling Smørgrav 		}
126f7e6344dSDag-Erling Smørgrav 		if (lineno != NULL && ch == '\n')
127f7e6344dSDag-Erling Smørgrav 			++*lineno;
128f7e6344dSDag-Erling Smørgrav 	}
129f7e6344dSDag-Erling Smørgrav 	if (ch == EOF && ferror(f)) {
130f7e6344dSDag-Erling Smørgrav 		serrno = errno;
131f7e6344dSDag-Erling Smørgrav 		free(word);
132f7e6344dSDag-Erling Smørgrav 		errno = serrno;
133f7e6344dSDag-Erling Smørgrav 		return (NULL);
134f7e6344dSDag-Erling Smørgrav 	}
135f7e6344dSDag-Erling Smørgrav 	if (ch == EOF && (escape || quote)) {
136f7e6344dSDag-Erling Smørgrav 		/* Missing escaped character or closing quote. */
137f3b0ac34SDag-Erling Smørgrav 		openpam_log(PAM_LOG_DEBUG, "unexpected end of file");
138f7e6344dSDag-Erling Smørgrav 		free(word);
139f7e6344dSDag-Erling Smørgrav 		errno = EINVAL;
140f7e6344dSDag-Erling Smørgrav 		return (NULL);
141f7e6344dSDag-Erling Smørgrav 	}
142f7e6344dSDag-Erling Smørgrav 	ungetc(ch, f);
143f7e6344dSDag-Erling Smørgrav 	if (lenp != NULL)
144f7e6344dSDag-Erling Smørgrav 		*lenp = len;
145f7e6344dSDag-Erling Smørgrav 	return (word);
146f7e6344dSDag-Erling Smørgrav }
147f7e6344dSDag-Erling Smørgrav 
148f7e6344dSDag-Erling Smørgrav /**
149f7e6344dSDag-Erling Smørgrav  * The =openpam_readword function reads the next word from a file, and
150f7e6344dSDag-Erling Smørgrav  * returns it in a NUL-terminated buffer allocated with =!malloc.
151f7e6344dSDag-Erling Smørgrav  *
152f7e6344dSDag-Erling Smørgrav  * A word is a sequence of non-whitespace characters.
153f7e6344dSDag-Erling Smørgrav  * However, whitespace characters can be included in a word if quoted or
154f7e6344dSDag-Erling Smørgrav  * escaped according to the following rules:
155f7e6344dSDag-Erling Smørgrav  *
156f7e6344dSDag-Erling Smørgrav  *  - An unescaped single or double quote introduces a quoted string,
157f7e6344dSDag-Erling Smørgrav  *    which ends when the same quote character is encountered a second
158f7e6344dSDag-Erling Smørgrav  *    time.
159f7e6344dSDag-Erling Smørgrav  *    The quotes themselves are stripped.
160f7e6344dSDag-Erling Smørgrav  *
161f7e6344dSDag-Erling Smørgrav  *  - Within a single- or double-quoted string, all whitespace characters,
162f7e6344dSDag-Erling Smørgrav  *    including the newline character, are preserved as-is.
163f7e6344dSDag-Erling Smørgrav  *
164f7e6344dSDag-Erling Smørgrav  *  - Outside a quoted string, a backslash escapes the next character,
165f7e6344dSDag-Erling Smørgrav  *    which is preserved as-is, unless that character is a newline, in
166f7e6344dSDag-Erling Smørgrav  *    which case it is discarded and reading continues at the beginning of
167f7e6344dSDag-Erling Smørgrav  *    the next line as if the backslash and newline had not been there.
168f7e6344dSDag-Erling Smørgrav  *    In all cases, the backslash itself is discarded.
169f7e6344dSDag-Erling Smørgrav  *
170f7e6344dSDag-Erling Smørgrav  *  - Within a single-quoted string, double quotes and backslashes are
171f7e6344dSDag-Erling Smørgrav  *    preserved as-is.
172f7e6344dSDag-Erling Smørgrav  *
173f7e6344dSDag-Erling Smørgrav  *  - Within a double-quoted string, a single quote is preserved as-is,
174f7e6344dSDag-Erling Smørgrav  *    and a backslash is preserved as-is unless used to escape a double
175f7e6344dSDag-Erling Smørgrav  *    quote.
176f7e6344dSDag-Erling Smørgrav  *
177f7e6344dSDag-Erling Smørgrav  * In addition, if the first non-whitespace character on the line is a
178f7e6344dSDag-Erling Smørgrav  * hash character (#), the rest of the line is discarded.
179f7e6344dSDag-Erling Smørgrav  * If a hash character occurs within a word, however, it is preserved
180f7e6344dSDag-Erling Smørgrav  * as-is.
181f7e6344dSDag-Erling Smørgrav  * A backslash at the end of a comment does cause line continuation.
182f7e6344dSDag-Erling Smørgrav  *
183f7e6344dSDag-Erling Smørgrav  * If =lineno is not =NULL, the integer variable it points to is
184f7e6344dSDag-Erling Smørgrav  * incremented every time a quoted or escaped newline character is read.
185f7e6344dSDag-Erling Smørgrav  *
186f7e6344dSDag-Erling Smørgrav  * If =lenp is not =NULL, the length of the word (after quotes and
187f7e6344dSDag-Erling Smørgrav  * backslashes have been removed) is stored in the variable it points to.
188f7e6344dSDag-Erling Smørgrav  *
189f7e6344dSDag-Erling Smørgrav  * RETURN VALUES
190f7e6344dSDag-Erling Smørgrav  *
191f7e6344dSDag-Erling Smørgrav  * If successful, the =openpam_readword function returns a pointer to a
192f7e6344dSDag-Erling Smørgrav  * dynamically allocated NUL-terminated string containing the first word
193f7e6344dSDag-Erling Smørgrav  * encountered on the line.
194f7e6344dSDag-Erling Smørgrav  *
195f7e6344dSDag-Erling Smørgrav  * The caller is responsible for releasing the returned buffer by passing
196f7e6344dSDag-Erling Smørgrav  * it to =!free.
197f7e6344dSDag-Erling Smørgrav  *
198f7e6344dSDag-Erling Smørgrav  * If =openpam_readword reaches the end of the line or file before any
199f7e6344dSDag-Erling Smørgrav  * characters are copied to the word, it returns =NULL.  In the former
200f7e6344dSDag-Erling Smørgrav  * case, the newline is pushed back to the file.
201f7e6344dSDag-Erling Smørgrav  *
202f7e6344dSDag-Erling Smørgrav  * If =openpam_readword reaches the end of the file while a quote or
203f7e6344dSDag-Erling Smørgrav  * backslash escape is in effect, it sets :errno to =EINVAL and returns
204f7e6344dSDag-Erling Smørgrav  * =NULL.
205f7e6344dSDag-Erling Smørgrav  *
206f7e6344dSDag-Erling Smørgrav  * IMPLEMENTATION NOTES
207f7e6344dSDag-Erling Smørgrav  *
208f7e6344dSDag-Erling Smørgrav  * The parsing rules are intended to be equivalent to the normal POSIX
209f7e6344dSDag-Erling Smørgrav  * shell quoting rules.
210f7e6344dSDag-Erling Smørgrav  * Any discrepancy is a bug and should be reported to the author along
211f7e6344dSDag-Erling Smørgrav  * with sample input that can be used to reproduce the error.
212f7e6344dSDag-Erling Smørgrav  *
213f7e6344dSDag-Erling Smørgrav  * >openpam_readline
214f7e6344dSDag-Erling Smørgrav  * >openpam_readlinev
215f7e6344dSDag-Erling Smørgrav  *
216f7e6344dSDag-Erling Smørgrav  * AUTHOR DES
217f7e6344dSDag-Erling Smørgrav  */
218