1/*-
2 * Copyright 1986, Larry Wall
3 *
4 * Redistribution and use in source and binary forms, with or without
5 * modification, are permitted provided that the following condition is met:
6 * 1. Redistributions of source code must retain the above copyright notice,
7 * this condition and the following disclaimer.
8 *
9 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND ANY
10 * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
11 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
12 * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
13 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
14 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
15 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
16 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
17 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
18 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
19 * SUCH DAMAGE.
20 *
21 * patch - a program to apply diffs to original files
22 *
23 * -C option added in 1998, original code by Marc Espie, based on FreeBSD
24 * behaviour
25 *
26 * $OpenBSD: inp.c,v 1.36 2012/04/10 14:46:34 ajacoutot Exp $
27 * $FreeBSD: stable/11/usr.bin/patch/inp.c 328462 2018-01-27 06:20:27Z kevans $
28 */
29
30#include <sys/types.h>
31#include <sys/file.h>
32#include <sys/stat.h>
33#include <sys/mman.h>
34#include <sys/wait.h>
35
36#include <ctype.h>
37#include <errno.h>
38#include <libgen.h>
39#include <paths.h>
40#include <spawn.h>
41#include <stddef.h>
42#include <stdint.h>
43#include <stdio.h>
44#include <stdlib.h>
45#include <string.h>
46#include <unistd.h>
47
48#include "common.h"
49#include "util.h"
50#include "pch.h"
51#include "inp.h"
52
53
54/* Input-file-with-indexable-lines abstract type */
55
56static size_t	i_size;		/* size of the input file */
57static char	*i_womp;	/* plan a buffer for entire file */
58static char	**i_ptr;	/* pointers to lines in i_womp */
59static char	empty_line[] = { '\0' };
60
61static int	tifd = -1;	/* plan b virtual string array */
62static char	*tibuf[2];	/* plan b buffers */
63static LINENUM	tiline[2] = {-1, -1};	/* 1st line in each buffer */
64static size_t	lines_per_buf;	/* how many lines per buffer */
65static size_t	tibuflen;	/* plan b buffer length */
66static size_t	tireclen;	/* length of records in tmp file */
67
68static bool	rev_in_string(const char *);
69static bool	reallocate_lines(size_t *);
70
71/* returns false if insufficient memory */
72static bool	plan_a(const char *);
73
74static void	plan_b(const char *);
75
76/* New patch--prepare to edit another file. */
77
78void
79re_input(void)
80{
81	if (using_plan_a) {
82		free(i_ptr);
83		i_ptr = NULL;
84		if (i_womp != NULL) {
85			munmap(i_womp, i_size);
86			i_womp = NULL;
87		}
88		i_size = 0;
89	} else {
90		using_plan_a = true;	/* maybe the next one is smaller */
91		close(tifd);
92		tifd = -1;
93		free(tibuf[0]);
94		free(tibuf[1]);
95		tibuf[0] = tibuf[1] = NULL;
96		tiline[0] = tiline[1] = -1;
97		tireclen = 0;
98	}
99}
100
101/* Construct the line index, somehow or other. */
102
103void
104scan_input(const char *filename)
105{
106	if (!plan_a(filename))
107		plan_b(filename);
108	if (verbose) {
109		say("Patching file %s using Plan %s...\n", filename,
110		    (using_plan_a ? "A" : "B"));
111	}
112}
113
114static bool
115reallocate_lines(size_t *lines_allocated)
116{
117	char	**p;
118	size_t	new_size;
119
120	new_size = *lines_allocated * 3 / 2;
121	p = realloc(i_ptr, (new_size + 2) * sizeof(char *));
122	if (p == NULL) {	/* shucks, it was a near thing */
123		munmap(i_womp, i_size);
124		i_womp = NULL;
125		free(i_ptr);
126		i_ptr = NULL;
127		*lines_allocated = 0;
128		return false;
129	}
130	*lines_allocated = new_size;
131	i_ptr = p;
132	return true;
133}
134
135/* Try keeping everything in memory. */
136
137static bool
138plan_a(const char *filename)
139{
140	int		ifd, statfailed;
141	char		*p, *s;
142	struct stat	filestat;
143	ptrdiff_t	sz;
144	size_t		i;
145	size_t		iline, lines_allocated;
146
147#ifdef DEBUGGING
148	if (debug & 8)
149		return false;
150#endif
151
152	if (filename == NULL || *filename == '\0')
153		return false;
154
155	statfailed = stat(filename, &filestat);
156	if (statfailed && ok_to_create_file) {
157		if (verbose)
158			say("(Creating file %s...)\n", filename);
159
160		/*
161		 * in check_patch case, we still display `Creating file' even
162		 * though we're not. The rule is that -C should be as similar
163		 * to normal patch behavior as possible
164		 */
165		if (check_only)
166			return true;
167		makedirs(filename, true);
168		close(creat(filename, 0666));
169		statfailed = stat(filename, &filestat);
170	}
171	if (statfailed)
172		fatal("can't find %s\n", filename);
173	filemode = filestat.st_mode;
174	if (!S_ISREG(filemode))
175		fatal("%s is not a normal file--can't patch\n", filename);
176	if ((uint64_t)filestat.st_size > SIZE_MAX) {
177		say("block too large to mmap\n");
178		return false;
179	}
180	i_size = (size_t)filestat.st_size;
181	if (out_of_mem) {
182		set_hunkmax();	/* make sure dynamic arrays are allocated */
183		out_of_mem = false;
184		return false;	/* force plan b because plan a bombed */
185	}
186	if ((ifd = open(filename, O_RDONLY)) < 0)
187		pfatal("can't open file %s", filename);
188
189	if (i_size) {
190		i_womp = mmap(NULL, i_size, PROT_READ, MAP_PRIVATE, ifd, 0);
191		if (i_womp == MAP_FAILED) {
192			perror("mmap failed");
193			i_womp = NULL;
194			close(ifd);
195			return false;
196		}
197	} else {
198		i_womp = NULL;
199	}
200
201	close(ifd);
202	if (i_size)
203		madvise(i_womp, i_size, MADV_SEQUENTIAL);
204
205	/* estimate the number of lines */
206	lines_allocated = i_size / 25;
207	if (lines_allocated < 100)
208		lines_allocated = 100;
209
210	if (!reallocate_lines(&lines_allocated))
211		return false;
212
213	/* now scan the buffer and build pointer array */
214	iline = 1;
215	i_ptr[iline] = i_womp;
216	/*
217	 * Testing for NUL here actively breaks files that innocently use NUL
218	 * for other reasons. mmap(2) succeeded, just scan the whole buffer.
219	 */
220	for (s = i_womp, i = 0; i < i_size; s++, i++) {
221		if (*s == '\n') {
222			if (iline == lines_allocated) {
223				if (!reallocate_lines(&lines_allocated))
224					return false;
225			}
226			/* these are NOT NUL terminated */
227			i_ptr[++iline] = s + 1;
228		}
229	}
230	/* if the last line contains no EOL, append one */
231	if (i_size > 0 && i_womp[i_size - 1] != '\n') {
232		last_line_missing_eol = true;
233		/* fix last line */
234		sz = s - i_ptr[iline];
235		p = malloc(sz + 1);
236		if (p == NULL) {
237			free(i_ptr);
238			i_ptr = NULL;
239			munmap(i_womp, i_size);
240			i_womp = NULL;
241			return false;
242		}
243
244		memcpy(p, i_ptr[iline], sz);
245		p[sz] = '\n';
246		i_ptr[iline] = p;
247		/* count the extra line and make it point to some valid mem */
248		i_ptr[++iline] = empty_line;
249	} else
250		last_line_missing_eol = false;
251
252	input_lines = iline - 1;
253
254	/* now check for revision, if any */
255
256	if (revision != NULL) {
257		if (i_womp == NULL || !rev_in_string(i_womp)) {
258			if (force) {
259				if (verbose)
260					say("Warning: this file doesn't appear "
261					    "to be the %s version--patching anyway.\n",
262					    revision);
263			} else if (batch) {
264				fatal("this file doesn't appear to be the "
265				    "%s version--aborting.\n",
266				    revision);
267			} else {
268				ask("This file doesn't appear to be the "
269				    "%s version--patch anyway? [n] ",
270				    revision);
271				if (*buf != 'y')
272					fatal("aborted\n");
273			}
274		} else if (verbose)
275			say("Good.  This file appears to be the %s version.\n",
276			    revision);
277	}
278	return true;		/* plan a will work */
279}
280
281/* Keep (virtually) nothing in memory. */
282
283static void
284plan_b(const char *filename)
285{
286	FILE	*ifp;
287	size_t	i = 0, j, len, maxlen = 1;
288	char	*lbuf = NULL, *p;
289	bool	found_revision = (revision == NULL);
290
291	using_plan_a = false;
292	if ((ifp = fopen(filename, "r")) == NULL)
293		pfatal("can't open file %s", filename);
294	unlink(TMPINNAME);
295	if ((tifd = open(TMPINNAME, O_EXCL | O_CREAT | O_WRONLY, 0666)) < 0)
296		pfatal("can't open file %s", TMPINNAME);
297	while ((p = fgetln(ifp, &len)) != NULL) {
298		if (p[len - 1] == '\n')
299			p[len - 1] = '\0';
300		else {
301			/* EOF without EOL, copy and add the NUL */
302			if ((lbuf = malloc(len + 1)) == NULL)
303				fatal("out of memory\n");
304			memcpy(lbuf, p, len);
305			lbuf[len] = '\0';
306			p = lbuf;
307
308			last_line_missing_eol = true;
309			len++;
310		}
311		if (revision != NULL && !found_revision && rev_in_string(p))
312			found_revision = true;
313		if (len > maxlen)
314			maxlen = len;   /* find longest line */
315	}
316	free(lbuf);
317	if (ferror(ifp))
318		pfatal("can't read file %s", filename);
319
320	if (revision != NULL) {
321		if (!found_revision) {
322			if (force) {
323				if (verbose)
324					say("Warning: this file doesn't appear "
325					    "to be the %s version--patching anyway.\n",
326					    revision);
327			} else if (batch) {
328				fatal("this file doesn't appear to be the "
329				    "%s version--aborting.\n",
330				    revision);
331			} else {
332				ask("This file doesn't appear to be the %s "
333				    "version--patch anyway? [n] ",
334				    revision);
335				if (*buf != 'y')
336					fatal("aborted\n");
337			}
338		} else if (verbose)
339			say("Good.  This file appears to be the %s version.\n",
340			    revision);
341	}
342	fseek(ifp, 0L, SEEK_SET);	/* rewind file */
343	tireclen = maxlen;
344	tibuflen = maxlen > BUFFERSIZE ? maxlen : BUFFERSIZE;
345	lines_per_buf = tibuflen / maxlen;
346	tibuf[0] = malloc(tibuflen + 1);
347	if (tibuf[0] == NULL)
348		fatal("out of memory\n");
349	tibuf[1] = malloc(tibuflen + 1);
350	if (tibuf[1] == NULL)
351		fatal("out of memory\n");
352	for (i = 1;; i++) {
353		p = tibuf[0] + maxlen * (i % lines_per_buf);
354		if (i % lines_per_buf == 0)	/* new block */
355			if (write(tifd, tibuf[0], tibuflen) !=
356			    (ssize_t) tibuflen)
357				pfatal("can't write temp file");
358		if (fgets(p, maxlen + 1, ifp) == NULL) {
359			input_lines = i - 1;
360			if (i % lines_per_buf != 0)
361				if (write(tifd, tibuf[0], tibuflen) !=
362				    (ssize_t) tibuflen)
363					pfatal("can't write temp file");
364			break;
365		}
366		j = strlen(p);
367		/* These are '\n' terminated strings, so no need to add a NUL */
368		if (j == 0 || p[j - 1] != '\n')
369			p[j] = '\n';
370	}
371	fclose(ifp);
372	close(tifd);
373	if ((tifd = open(TMPINNAME, O_RDONLY)) < 0)
374		pfatal("can't reopen file %s", TMPINNAME);
375}
376
377/*
378 * Fetch a line from the input file, \n terminated, not necessarily \0.
379 */
380char *
381ifetch(LINENUM line, int whichbuf)
382{
383	if (line < 1 || line > input_lines) {
384		if (warn_on_invalid_line) {
385			say("No such line %ld in input file, ignoring\n", line);
386			warn_on_invalid_line = false;
387		}
388		return NULL;
389	}
390	if (using_plan_a)
391		return i_ptr[line];
392	else {
393		LINENUM	offline = line % lines_per_buf;
394		LINENUM	baseline = line - offline;
395
396		if (tiline[0] == baseline)
397			whichbuf = 0;
398		else if (tiline[1] == baseline)
399			whichbuf = 1;
400		else {
401			tiline[whichbuf] = baseline;
402
403			if (lseek(tifd, (off_t) (baseline / lines_per_buf *
404			    tibuflen), SEEK_SET) < 0)
405				pfatal("cannot seek in the temporary input file");
406
407			if (read(tifd, tibuf[whichbuf], tibuflen) !=
408			    (ssize_t) tibuflen)
409				pfatal("error reading tmp file %s", TMPINNAME);
410		}
411		return tibuf[whichbuf] + (tireclen * offline);
412	}
413}
414
415/*
416 * True if the string argument contains the revision number we want.
417 */
418static bool
419rev_in_string(const char *string)
420{
421	const char	*s;
422	size_t		patlen;
423
424	if (revision == NULL)
425		return true;
426	patlen = strlen(revision);
427	if (strnEQ(string, revision, patlen) && isspace((unsigned char)string[patlen]))
428		return true;
429	for (s = string; *s; s++) {
430		if (isspace((unsigned char)*s) && strnEQ(s + 1, revision, patlen) &&
431		    isspace((unsigned char)s[patlen + 1])) {
432			return true;
433		}
434	}
435	return false;
436}
437