xref: /freebsd/contrib/openpam/lib/libpam/openpam_readword.c (revision f7e6344d4a3b7072f8b8a1dec8f483a5a3ca50eb)
1*f7e6344dSDag-Erling Smørgrav /*-
2*f7e6344dSDag-Erling Smørgrav  * Copyright (c) 2012 Dag-Erling Smørgrav
3*f7e6344dSDag-Erling Smørgrav  * All rights reserved.
4*f7e6344dSDag-Erling Smørgrav  *
5*f7e6344dSDag-Erling Smørgrav  * Redistribution and use in source and binary forms, with or without
6*f7e6344dSDag-Erling Smørgrav  * modification, are permitted provided that the following conditions
7*f7e6344dSDag-Erling Smørgrav  * are met:
8*f7e6344dSDag-Erling Smørgrav  * 1. Redistributions of source code must retain the above copyright
9*f7e6344dSDag-Erling Smørgrav  *    notice, this list of conditions and the following disclaimer
10*f7e6344dSDag-Erling Smørgrav  *    in this position and unchanged.
11*f7e6344dSDag-Erling Smørgrav  * 2. Redistributions in binary form must reproduce the above copyright
12*f7e6344dSDag-Erling Smørgrav  *    notice, this list of conditions and the following disclaimer in the
13*f7e6344dSDag-Erling Smørgrav  *    documentation and/or other materials provided with the distribution.
14*f7e6344dSDag-Erling Smørgrav  * 3. The name of the author may not be used to endorse or promote
15*f7e6344dSDag-Erling Smørgrav  *    products derived from this software without specific prior written
16*f7e6344dSDag-Erling Smørgrav  *    permission.
17*f7e6344dSDag-Erling Smørgrav  *
18*f7e6344dSDag-Erling Smørgrav  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
19*f7e6344dSDag-Erling Smørgrav  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20*f7e6344dSDag-Erling Smørgrav  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21*f7e6344dSDag-Erling Smørgrav  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
22*f7e6344dSDag-Erling Smørgrav  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23*f7e6344dSDag-Erling Smørgrav  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24*f7e6344dSDag-Erling Smørgrav  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25*f7e6344dSDag-Erling Smørgrav  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26*f7e6344dSDag-Erling Smørgrav  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27*f7e6344dSDag-Erling Smørgrav  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28*f7e6344dSDag-Erling Smørgrav  * SUCH DAMAGE.
29*f7e6344dSDag-Erling Smørgrav  *
30*f7e6344dSDag-Erling Smørgrav  * $Id: openpam_readword.c 588 2012-04-08 11:52:25Z des $
31*f7e6344dSDag-Erling Smørgrav  */
32*f7e6344dSDag-Erling Smørgrav 
33*f7e6344dSDag-Erling Smørgrav #ifdef HAVE_CONFIG_H
34*f7e6344dSDag-Erling Smørgrav # include "config.h"
35*f7e6344dSDag-Erling Smørgrav #endif
36*f7e6344dSDag-Erling Smørgrav 
37*f7e6344dSDag-Erling Smørgrav #include <errno.h>
38*f7e6344dSDag-Erling Smørgrav #include <stdio.h>
39*f7e6344dSDag-Erling Smørgrav #include <stdlib.h>
40*f7e6344dSDag-Erling Smørgrav 
41*f7e6344dSDag-Erling Smørgrav #include <security/pam_appl.h>
42*f7e6344dSDag-Erling Smørgrav 
43*f7e6344dSDag-Erling Smørgrav #include "openpam_impl.h"
44*f7e6344dSDag-Erling Smørgrav #include "openpam_ctype.h"
45*f7e6344dSDag-Erling Smørgrav 
46*f7e6344dSDag-Erling Smørgrav #define MIN_WORD_SIZE	32
47*f7e6344dSDag-Erling Smørgrav 
48*f7e6344dSDag-Erling Smørgrav /*
49*f7e6344dSDag-Erling Smørgrav  * OpenPAM extension
50*f7e6344dSDag-Erling Smørgrav  *
51*f7e6344dSDag-Erling Smørgrav  * Read a word from a file, respecting shell quoting rules.
52*f7e6344dSDag-Erling Smørgrav  */
53*f7e6344dSDag-Erling Smørgrav 
54*f7e6344dSDag-Erling Smørgrav char *
55*f7e6344dSDag-Erling Smørgrav openpam_readword(FILE *f, int *lineno, size_t *lenp)
56*f7e6344dSDag-Erling Smørgrav {
57*f7e6344dSDag-Erling Smørgrav 	char *word;
58*f7e6344dSDag-Erling Smørgrav 	size_t size, len;
59*f7e6344dSDag-Erling Smørgrav 	int ch, comment, escape, quote;
60*f7e6344dSDag-Erling Smørgrav 	int serrno;
61*f7e6344dSDag-Erling Smørgrav 
62*f7e6344dSDag-Erling Smørgrav 	errno = 0;
63*f7e6344dSDag-Erling Smørgrav 
64*f7e6344dSDag-Erling Smørgrav 	/* skip initial whitespace */
65*f7e6344dSDag-Erling Smørgrav 	comment = 0;
66*f7e6344dSDag-Erling Smørgrav 	while ((ch = getc(f)) != EOF && ch != '\n') {
67*f7e6344dSDag-Erling Smørgrav 		if (ch == '#')
68*f7e6344dSDag-Erling Smørgrav 			comment = 1;
69*f7e6344dSDag-Erling Smørgrav 		if (!is_lws(ch) && !comment)
70*f7e6344dSDag-Erling Smørgrav 			break;
71*f7e6344dSDag-Erling Smørgrav 	}
72*f7e6344dSDag-Erling Smørgrav 	if (ch == EOF)
73*f7e6344dSDag-Erling Smørgrav 		return (NULL);
74*f7e6344dSDag-Erling Smørgrav 	ungetc(ch, f);
75*f7e6344dSDag-Erling Smørgrav 	if (ch == '\n')
76*f7e6344dSDag-Erling Smørgrav 		return (NULL);
77*f7e6344dSDag-Erling Smørgrav 
78*f7e6344dSDag-Erling Smørgrav 	word = NULL;
79*f7e6344dSDag-Erling Smørgrav 	size = len = 0;
80*f7e6344dSDag-Erling Smørgrav 	escape = quote = 0;
81*f7e6344dSDag-Erling Smørgrav 	while ((ch = fgetc(f)) != EOF && (!is_ws(ch) || quote || escape)) {
82*f7e6344dSDag-Erling Smørgrav 		if (ch == '\\' && !escape && quote != '\'') {
83*f7e6344dSDag-Erling Smørgrav 			/* escape next character */
84*f7e6344dSDag-Erling Smørgrav 			escape = ch;
85*f7e6344dSDag-Erling Smørgrav 		} else if ((ch == '\'' || ch == '"') && !quote && !escape) {
86*f7e6344dSDag-Erling Smørgrav 			/* begin quote */
87*f7e6344dSDag-Erling Smørgrav 			quote = ch;
88*f7e6344dSDag-Erling Smørgrav 			/* edge case: empty quoted string */
89*f7e6344dSDag-Erling Smørgrav 			if (openpam_straddch(&word, &size, &len, 0) != 0)
90*f7e6344dSDag-Erling Smørgrav 				return (NULL);
91*f7e6344dSDag-Erling Smørgrav 		} else if (ch == quote && !escape) {
92*f7e6344dSDag-Erling Smørgrav 			/* end quote */
93*f7e6344dSDag-Erling Smørgrav 			quote = 0;
94*f7e6344dSDag-Erling Smørgrav 		} else if (ch == '\n' && escape && quote != '\'') {
95*f7e6344dSDag-Erling Smørgrav 			/* line continuation */
96*f7e6344dSDag-Erling Smørgrav 			escape = 0;
97*f7e6344dSDag-Erling Smørgrav 		} else {
98*f7e6344dSDag-Erling Smørgrav 			if (escape && quote && ch != '\\' && ch != quote &&
99*f7e6344dSDag-Erling Smørgrav 			    openpam_straddch(&word, &size, &len, '\\') != 0) {
100*f7e6344dSDag-Erling Smørgrav 				free(word);
101*f7e6344dSDag-Erling Smørgrav 				errno = ENOMEM;
102*f7e6344dSDag-Erling Smørgrav 				return (NULL);
103*f7e6344dSDag-Erling Smørgrav 			}
104*f7e6344dSDag-Erling Smørgrav 			if (openpam_straddch(&word, &size, &len, ch) != 0) {
105*f7e6344dSDag-Erling Smørgrav 				free(word);
106*f7e6344dSDag-Erling Smørgrav 				errno = ENOMEM;
107*f7e6344dSDag-Erling Smørgrav 				return (NULL);
108*f7e6344dSDag-Erling Smørgrav 			}
109*f7e6344dSDag-Erling Smørgrav 			escape = 0;
110*f7e6344dSDag-Erling Smørgrav 		}
111*f7e6344dSDag-Erling Smørgrav 		if (lineno != NULL && ch == '\n')
112*f7e6344dSDag-Erling Smørgrav 			++*lineno;
113*f7e6344dSDag-Erling Smørgrav 	}
114*f7e6344dSDag-Erling Smørgrav 	if (ch == EOF && ferror(f)) {
115*f7e6344dSDag-Erling Smørgrav 		serrno = errno;
116*f7e6344dSDag-Erling Smørgrav 		free(word);
117*f7e6344dSDag-Erling Smørgrav 		errno = serrno;
118*f7e6344dSDag-Erling Smørgrav 		return (NULL);
119*f7e6344dSDag-Erling Smørgrav 	}
120*f7e6344dSDag-Erling Smørgrav 	if (ch == EOF && (escape || quote)) {
121*f7e6344dSDag-Erling Smørgrav 		/* Missing escaped character or closing quote. */
122*f7e6344dSDag-Erling Smørgrav 		openpam_log(PAM_LOG_ERROR, "unexpected end of file");
123*f7e6344dSDag-Erling Smørgrav 		free(word);
124*f7e6344dSDag-Erling Smørgrav 		errno = EINVAL;
125*f7e6344dSDag-Erling Smørgrav 		return (NULL);
126*f7e6344dSDag-Erling Smørgrav 	}
127*f7e6344dSDag-Erling Smørgrav 	ungetc(ch, f);
128*f7e6344dSDag-Erling Smørgrav 	if (lenp != NULL)
129*f7e6344dSDag-Erling Smørgrav 		*lenp = len;
130*f7e6344dSDag-Erling Smørgrav 	return (word);
131*f7e6344dSDag-Erling Smørgrav }
132*f7e6344dSDag-Erling Smørgrav 
133*f7e6344dSDag-Erling Smørgrav /**
134*f7e6344dSDag-Erling Smørgrav  * The =openpam_readword function reads the next word from a file, and
135*f7e6344dSDag-Erling Smørgrav  * returns it in a NUL-terminated buffer allocated with =!malloc.
136*f7e6344dSDag-Erling Smørgrav  *
137*f7e6344dSDag-Erling Smørgrav  * A word is a sequence of non-whitespace characters.
138*f7e6344dSDag-Erling Smørgrav  * However, whitespace characters can be included in a word if quoted or
139*f7e6344dSDag-Erling Smørgrav  * escaped according to the following rules:
140*f7e6344dSDag-Erling Smørgrav  *
141*f7e6344dSDag-Erling Smørgrav  *  - An unescaped single or double quote introduces a quoted string,
142*f7e6344dSDag-Erling Smørgrav  *    which ends when the same quote character is encountered a second
143*f7e6344dSDag-Erling Smørgrav  *    time.
144*f7e6344dSDag-Erling Smørgrav  *    The quotes themselves are stripped.
145*f7e6344dSDag-Erling Smørgrav  *
146*f7e6344dSDag-Erling Smørgrav  *  - Within a single- or double-quoted string, all whitespace characters,
147*f7e6344dSDag-Erling Smørgrav  *    including the newline character, are preserved as-is.
148*f7e6344dSDag-Erling Smørgrav  *
149*f7e6344dSDag-Erling Smørgrav  *  - Outside a quoted string, a backslash escapes the next character,
150*f7e6344dSDag-Erling Smørgrav  *    which is preserved as-is, unless that character is a newline, in
151*f7e6344dSDag-Erling Smørgrav  *    which case it is discarded and reading continues at the beginning of
152*f7e6344dSDag-Erling Smørgrav  *    the next line as if the backslash and newline had not been there.
153*f7e6344dSDag-Erling Smørgrav  *    In all cases, the backslash itself is discarded.
154*f7e6344dSDag-Erling Smørgrav  *
155*f7e6344dSDag-Erling Smørgrav  *  - Within a single-quoted string, double quotes and backslashes are
156*f7e6344dSDag-Erling Smørgrav  *    preserved as-is.
157*f7e6344dSDag-Erling Smørgrav  *
158*f7e6344dSDag-Erling Smørgrav  *  - Within a double-quoted string, a single quote is preserved as-is,
159*f7e6344dSDag-Erling Smørgrav  *    and a backslash is preserved as-is unless used to escape a double
160*f7e6344dSDag-Erling Smørgrav  *    quote.
161*f7e6344dSDag-Erling Smørgrav  *
162*f7e6344dSDag-Erling Smørgrav  * In addition, if the first non-whitespace character on the line is a
163*f7e6344dSDag-Erling Smørgrav  * hash character (#), the rest of the line is discarded.
164*f7e6344dSDag-Erling Smørgrav  * If a hash character occurs within a word, however, it is preserved
165*f7e6344dSDag-Erling Smørgrav  * as-is.
166*f7e6344dSDag-Erling Smørgrav  * A backslash at the end of a comment does cause line continuation.
167*f7e6344dSDag-Erling Smørgrav  *
168*f7e6344dSDag-Erling Smørgrav  * If =lineno is not =NULL, the integer variable it points to is
169*f7e6344dSDag-Erling Smørgrav  * incremented every time a quoted or escaped newline character is read.
170*f7e6344dSDag-Erling Smørgrav  *
171*f7e6344dSDag-Erling Smørgrav  * If =lenp is not =NULL, the length of the word (after quotes and
172*f7e6344dSDag-Erling Smørgrav  * backslashes have been removed) is stored in the variable it points to.
173*f7e6344dSDag-Erling Smørgrav  *
174*f7e6344dSDag-Erling Smørgrav  * RETURN VALUES
175*f7e6344dSDag-Erling Smørgrav  *
176*f7e6344dSDag-Erling Smørgrav  * If successful, the =openpam_readword function returns a pointer to a
177*f7e6344dSDag-Erling Smørgrav  * dynamically allocated NUL-terminated string containing the first word
178*f7e6344dSDag-Erling Smørgrav  * encountered on the line.
179*f7e6344dSDag-Erling Smørgrav  *
180*f7e6344dSDag-Erling Smørgrav  * The caller is responsible for releasing the returned buffer by passing
181*f7e6344dSDag-Erling Smørgrav  * it to =!free.
182*f7e6344dSDag-Erling Smørgrav  *
183*f7e6344dSDag-Erling Smørgrav  * If =openpam_readword reaches the end of the line or file before any
184*f7e6344dSDag-Erling Smørgrav  * characters are copied to the word, it returns =NULL.  In the former
185*f7e6344dSDag-Erling Smørgrav  * case, the newline is pushed back to the file.
186*f7e6344dSDag-Erling Smørgrav  *
187*f7e6344dSDag-Erling Smørgrav  * If =openpam_readword reaches the end of the file while a quote or
188*f7e6344dSDag-Erling Smørgrav  * backslash escape is in effect, it sets :errno to =EINVAL and returns
189*f7e6344dSDag-Erling Smørgrav  * =NULL.
190*f7e6344dSDag-Erling Smørgrav  *
191*f7e6344dSDag-Erling Smørgrav  * IMPLEMENTATION NOTES
192*f7e6344dSDag-Erling Smørgrav  *
193*f7e6344dSDag-Erling Smørgrav  * The parsing rules are intended to be equivalent to the normal POSIX
194*f7e6344dSDag-Erling Smørgrav  * shell quoting rules.
195*f7e6344dSDag-Erling Smørgrav  * Any discrepancy is a bug and should be reported to the author along
196*f7e6344dSDag-Erling Smørgrav  * with sample input that can be used to reproduce the error.
197*f7e6344dSDag-Erling Smørgrav  *
198*f7e6344dSDag-Erling Smørgrav  * >openpam_readline
199*f7e6344dSDag-Erling Smørgrav  * >openpam_readlinev
200*f7e6344dSDag-Erling Smørgrav  *
201*f7e6344dSDag-Erling Smørgrav  * AUTHOR DES
202*f7e6344dSDag-Erling Smørgrav  */
203