blob: 6696ea4c3ffd4f7af04f3f56609a78a74a0a0b55 [file] [log] [blame]
/*
* apply.c
*
* Copyright (C) Linus Torvalds, 2005
*
* This applies patches on top of some (arbitrary) version of the SCM.
*
*/
#include "cache.h"
#include "lockfile.h"
#include "cache-tree.h"
#include "quote.h"
#include "blob.h"
#include "delta.h"
#include "builtin.h"
#include "string-list.h"
#include "dir.h"
#include "diff.h"
#include "parse-options.h"
#include "xdiff-interface.h"
#include "ll-merge.h"
#include "rerere.h"
/*
* --check turns on checking that the working tree matches the
* files that are being modified, but doesn't apply the patch
* --stat does just a diffstat, and doesn't actually apply
* --numstat does numeric diffstat, and doesn't actually apply
* --index-info shows the old and new index info for paths if available.
* --index updates the cache as well.
* --cached updates only the cache without ever touching the working tree.
*/
static const char *prefix;
static int prefix_length = -1;
static int newfd = -1;
static int unidiff_zero;
static int p_value = 1;
static int p_value_known;
static int check_index;
static int update_index;
static int cached;
static int diffstat;
static int numstat;
static int summary;
static int check;
static int apply = 1;
static int apply_in_reverse;
static int apply_with_reject;
static int apply_verbosely;
static int allow_overlap;
static int no_add;
static int threeway;
static const char *fake_ancestor;
static int line_termination = '\n';
static unsigned int p_context = UINT_MAX;
static const char * const apply_usage[] = {
N_("git apply [options] [<patch>...]"),
NULL
};
static enum ws_error_action {
nowarn_ws_error,
warn_on_ws_error,
die_on_ws_error,
correct_ws_error
} ws_error_action = warn_on_ws_error;
static int whitespace_error;
static int squelch_whitespace_errors = 5;
static int applied_after_fixing_ws;
static enum ws_ignore {
ignore_ws_none,
ignore_ws_change
} ws_ignore_action = ignore_ws_none;
static const char *patch_input_file;
static const char *root;
static int root_len;
static int read_stdin = 1;
static int options;
static void parse_whitespace_option(const char *option)
{
if (!option) {
ws_error_action = warn_on_ws_error;
return;
}
if (!strcmp(option, "warn")) {
ws_error_action = warn_on_ws_error;
return;
}
if (!strcmp(option, "nowarn")) {
ws_error_action = nowarn_ws_error;
return;
}
if (!strcmp(option, "error")) {
ws_error_action = die_on_ws_error;
return;
}
if (!strcmp(option, "error-all")) {
ws_error_action = die_on_ws_error;
squelch_whitespace_errors = 0;
return;
}
if (!strcmp(option, "strip") || !strcmp(option, "fix")) {
ws_error_action = correct_ws_error;
return;
}
die(_("unrecognized whitespace option '%s'"), option);
}
static void parse_ignorewhitespace_option(const char *option)
{
if (!option || !strcmp(option, "no") ||
!strcmp(option, "false") || !strcmp(option, "never") ||
!strcmp(option, "none")) {
ws_ignore_action = ignore_ws_none;
return;
}
if (!strcmp(option, "change")) {
ws_ignore_action = ignore_ws_change;
return;
}
die(_("unrecognized whitespace ignore option '%s'"), option);
}
static void set_default_whitespace_mode(const char *whitespace_option)
{
if (!whitespace_option && !apply_default_whitespace)
ws_error_action = (apply ? warn_on_ws_error : nowarn_ws_error);
}
/*
* For "diff-stat" like behaviour, we keep track of the biggest change
* we've seen, and the longest filename. That allows us to do simple
* scaling.
*/
static int max_change, max_len;
/*
* Various "current state", notably line numbers and what
* file (and how) we're patching right now.. The "is_xxxx"
* things are flags, where -1 means "don't know yet".
*/
static int linenr = 1;
/*
* This represents one "hunk" from a patch, starting with
* "@@ -oldpos,oldlines +newpos,newlines @@" marker. The
* patch text is pointed at by patch, and its byte length
* is stored in size. leading and trailing are the number
* of context lines.
*/
struct fragment {
unsigned long leading, trailing;
unsigned long oldpos, oldlines;
unsigned long newpos, newlines;
/*
* 'patch' is usually borrowed from buf in apply_patch(),
* but some codepaths store an allocated buffer.
*/
const char *patch;
unsigned free_patch:1,
rejected:1;
int size;
int linenr;
struct fragment *next;
};
/*
* When dealing with a binary patch, we reuse "leading" field
* to store the type of the binary hunk, either deflated "delta"
* or deflated "literal".
*/
#define binary_patch_method leading
#define BINARY_DELTA_DEFLATED 1
#define BINARY_LITERAL_DEFLATED 2
/*
* This represents a "patch" to a file, both metainfo changes
* such as creation/deletion, filemode and content changes represented
* as a series of fragments.
*/
struct patch {
char *new_name, *old_name, *def_name;
unsigned int old_mode, new_mode;
int is_new, is_delete; /* -1 = unknown, 0 = false, 1 = true */
int rejected;
unsigned ws_rule;
int lines_added, lines_deleted;
int score;
unsigned int is_toplevel_relative:1;
unsigned int inaccurate_eof:1;
unsigned int is_binary:1;
unsigned int is_copy:1;
unsigned int is_rename:1;
unsigned int recount:1;
unsigned int conflicted_threeway:1;
unsigned int direct_to_threeway:1;
struct fragment *fragments;
char *result;
size_t resultsize;
char old_sha1_prefix[41];
char new_sha1_prefix[41];
struct patch *next;
/* three-way fallback result */
unsigned char threeway_stage[3][20];
};
static void free_fragment_list(struct fragment *list)
{
while (list) {
struct fragment *next = list->next;
if (list->free_patch)
free((char *)list->patch);
free(list);
list = next;
}
}
static void free_patch(struct patch *patch)
{
free_fragment_list(patch->fragments);
free(patch->def_name);
free(patch->old_name);
free(patch->new_name);
free(patch->result);
free(patch);
}
static void free_patch_list(struct patch *list)
{
while (list) {
struct patch *next = list->next;
free_patch(list);
list = next;
}
}
/*
* A line in a file, len-bytes long (includes the terminating LF,
* except for an incomplete line at the end if the file ends with
* one), and its contents hashes to 'hash'.
*/
struct line {
size_t len;
unsigned hash : 24;
unsigned flag : 8;
#define LINE_COMMON 1
#define LINE_PATCHED 2
};
/*
* This represents a "file", which is an array of "lines".
*/
struct image {
char *buf;
size_t len;
size_t nr;
size_t alloc;
struct line *line_allocated;
struct line *line;
};
/*
* Records filenames that have been touched, in order to handle
* the case where more than one patches touch the same file.
*/
static struct string_list fn_table;
static uint32_t hash_line(const char *cp, size_t len)
{
size_t i;
uint32_t h;
for (i = 0, h = 0; i < len; i++) {
if (!isspace(cp[i])) {
h = h * 3 + (cp[i] & 0xff);
}
}
return h;
}
/*
* Compare lines s1 of length n1 and s2 of length n2, ignoring
* whitespace difference. Returns 1 if they match, 0 otherwise
*/
static int fuzzy_matchlines(const char *s1, size_t n1,
const char *s2, size_t n2)
{
const char *last1 = s1 + n1 - 1;
const char *last2 = s2 + n2 - 1;
int result = 0;
/* ignore line endings */
while ((*last1 == '\r') || (*last1 == '\n'))
last1--;
while ((*last2 == '\r') || (*last2 == '\n'))
last2--;
/* skip leading whitespaces, if both begin with whitespace */
if (s1 <= last1 && s2 <= last2 && isspace(*s1) && isspace(*s2)) {
while (isspace(*s1) && (s1 <= last1))
s1++;
while (isspace(*s2) && (s2 <= last2))
s2++;
}
/* early return if both lines are empty */
if ((s1 > last1) && (s2 > last2))
return 1;
while (!result) {
result = *s1++ - *s2++;
/*
* Skip whitespace inside. We check for whitespace on
* both buffers because we don't want "a b" to match
* "ab"
*/
if (isspace(*s1) && isspace(*s2)) {
while (isspace(*s1) && s1 <= last1)
s1++;
while (isspace(*s2) && s2 <= last2)
s2++;
}
/*
* If we reached the end on one side only,
* lines don't match
*/
if (
((s2 > last2) && (s1 <= last1)) ||
((s1 > last1) && (s2 <= last2)))
return 0;
if ((s1 > last1) && (s2 > last2))
break;
}
return !result;
}
static void add_line_info(struct image *img, const char *bol, size_t len, unsigned flag)
{
ALLOC_GROW(img->line_allocated, img->nr + 1, img->alloc);
img->line_allocated[img->nr].len = len;
img->line_allocated[img->nr].hash = hash_line(bol, len);
img->line_allocated[img->nr].flag = flag;
img->nr++;
}
/*
* "buf" has the file contents to be patched (read from various sources).
* attach it to "image" and add line-based index to it.
* "image" now owns the "buf".
*/
static void prepare_image(struct image *image, char *buf, size_t len,
int prepare_linetable)
{
const char *cp, *ep;
memset(image, 0, sizeof(*image));
image->buf = buf;
image->len = len;
if (!prepare_linetable)
return;
ep = image->buf + image->len;
cp = image->buf;
while (cp < ep) {
const char *next;
for (next = cp; next < ep && *next != '\n'; next++)
;
if (next < ep)
next++;
add_line_info(image, cp, next - cp, 0);
cp = next;
}
image->line = image->line_allocated;
}
static void clear_image(struct image *image)
{
free(image->buf);
free(image->line_allocated);
memset(image, 0, sizeof(*image));
}
/* fmt must contain _one_ %s and no other substitution */
static void say_patch_name(FILE *output, const char *fmt, struct patch *patch)
{
struct strbuf sb = STRBUF_INIT;
if (patch->old_name && patch->new_name &&
strcmp(patch->old_name, patch->new_name)) {
quote_c_style(patch->old_name, &sb, NULL, 0);
strbuf_addstr(&sb, " => ");
quote_c_style(patch->new_name, &sb, NULL, 0);
} else {
const char *n = patch->new_name;
if (!n)
n = patch->old_name;
quote_c_style(n, &sb, NULL, 0);
}
fprintf(output, fmt, sb.buf);
fputc('\n', output);
strbuf_release(&sb);
}
#define SLOP (16)
static void read_patch_file(struct strbuf *sb, int fd)
{
if (strbuf_read(sb, fd, 0) < 0)
die_errno("git apply: failed to read");
/*
* Make sure that we have some slop in the buffer
* so that we can do speculative "memcmp" etc, and
* see to it that it is NUL-filled.
*/
strbuf_grow(sb, SLOP);
memset(sb->buf + sb->len, 0, SLOP);
}
static unsigned long linelen(const char *buffer, unsigned long size)
{
unsigned long len = 0;
while (size--) {
len++;
if (*buffer++ == '\n')
break;
}
return len;
}
static int is_dev_null(const char *str)
{
return skip_prefix(str, "/dev/null", &str) && isspace(*str);
}
#define TERM_SPACE 1
#define TERM_TAB 2
static int name_terminate(const char *name, int namelen, int c, int terminate)
{
if (c == ' ' && !(terminate & TERM_SPACE))
return 0;
if (c == '\t' && !(terminate & TERM_TAB))
return 0;
return 1;
}
/* remove double slashes to make --index work with such filenames */
static char *squash_slash(char *name)
{
int i = 0, j = 0;
if (!name)
return NULL;
while (name[i]) {
if ((name[j++] = name[i++]) == '/')
while (name[i] == '/')
i++;
}
name[j] = '\0';
return name;
}
static char *find_name_gnu(const char *line, const char *def, int p_value)
{
struct strbuf name = STRBUF_INIT;
char *cp;
/*
* Proposed "new-style" GNU patch/diff format; see
* http://marc.info/?l=git&m=112927316408690&w=2
*/
if (unquote_c_style(&name, line, NULL)) {
strbuf_release(&name);
return NULL;
}
for (cp = name.buf; p_value; p_value--) {
cp = strchr(cp, '/');
if (!cp) {
strbuf_release(&name);
return NULL;
}
cp++;
}
strbuf_remove(&name, 0, cp - name.buf);
if (root)
strbuf_insert(&name, 0, root, root_len);
return squash_slash(strbuf_detach(&name, NULL));
}
static size_t sane_tz_len(const char *line, size_t len)
{
const char *tz, *p;
if (len < strlen(" +0500") || line[len-strlen(" +0500")] != ' ')
return 0;
tz = line + len - strlen(" +0500");
if (tz[1] != '+' && tz[1] != '-')
return 0;
for (p = tz + 2; p != line + len; p++)
if (!isdigit(*p))
return 0;
return line + len - tz;
}
static size_t tz_with_colon_len(const char *line, size_t len)
{
const char *tz, *p;
if (len < strlen(" +08:00") || line[len - strlen(":00")] != ':')
return 0;
tz = line + len - strlen(" +08:00");
if (tz[0] != ' ' || (tz[1] != '+' && tz[1] != '-'))
return 0;
p = tz + 2;
if (!isdigit(*p++) || !isdigit(*p++) || *p++ != ':' ||
!isdigit(*p++) || !isdigit(*p++))
return 0;
return line + len - tz;
}
static size_t date_len(const char *line, size_t len)
{
const char *date, *p;
if (len < strlen("72-02-05") || line[len-strlen("-05")] != '-')
return 0;
p = date = line + len - strlen("72-02-05");
if (!isdigit(*p++) || !isdigit(*p++) || *p++ != '-' ||
!isdigit(*p++) || !isdigit(*p++) || *p++ != '-' ||
!isdigit(*p++) || !isdigit(*p++)) /* Not a date. */
return 0;
if (date - line >= strlen("19") &&
isdigit(date[-1]) && isdigit(date[-2])) /* 4-digit year */
date -= strlen("19");
return line + len - date;
}
static size_t short_time_len(const char *line, size_t len)
{
const char *time, *p;
if (len < strlen(" 07:01:32") || line[len-strlen(":32")] != ':')
return 0;
p = time = line + len - strlen(" 07:01:32");
/* Permit 1-digit hours? */
if (*p++ != ' ' ||
!isdigit(*p++) || !isdigit(*p++) || *p++ != ':' ||
!isdigit(*p++) || !isdigit(*p++) || *p++ != ':' ||
!isdigit(*p++) || !isdigit(*p++)) /* Not a time. */
return 0;
return line + len - time;
}
static size_t fractional_time_len(const char *line, size_t len)
{
const char *p;
size_t n;
/* Expected format: 19:41:17.620000023 */
if (!len || !isdigit(line[len - 1]))
return 0;
p = line + len - 1;
/* Fractional seconds. */
while (p > line && isdigit(*p))
p--;
if (*p != '.')
return 0;
/* Hours, minutes, and whole seconds. */
n = short_time_len(line, p - line);
if (!n)
return 0;
return line + len - p + n;
}
static size_t trailing_spaces_len(const char *line, size_t len)
{
const char *p;
/* Expected format: ' ' x (1 or more) */
if (!len || line[len - 1] != ' ')
return 0;
p = line + len;
while (p != line) {
p--;
if (*p != ' ')
return line + len - (p + 1);
}
/* All spaces! */
return len;
}
static size_t diff_timestamp_len(const char *line, size_t len)
{
const char *end = line + len;
size_t n;
/*
* Posix: 2010-07-05 19:41:17
* GNU: 2010-07-05 19:41:17.620000023 -0500
*/
if (!isdigit(end[-1]))
return 0;
n = sane_tz_len(line, end - line);
if (!n)
n = tz_with_colon_len(line, end - line);
end -= n;
n = short_time_len(line, end - line);
if (!n)
n = fractional_time_len(line, end - line);
end -= n;
n = date_len(line, end - line);
if (!n) /* No date. Too bad. */
return 0;
end -= n;
if (end == line) /* No space before date. */
return 0;
if (end[-1] == '\t') { /* Success! */
end--;
return line + len - end;
}
if (end[-1] != ' ') /* No space before date. */
return 0;
/* Whitespace damage. */
end -= trailing_spaces_len(line, end - line);
return line + len - end;
}
static char *null_strdup(const char *s)
{
return s ? xstrdup(s) : NULL;
}
static char *find_name_common(const char *line, const char *def,
int p_value, const char *end, int terminate)
{
int len;
const char *start = NULL;
if (p_value == 0)
start = line;
while (line != end) {
char c = *line;
if (!end && isspace(c)) {
if (c == '\n')
break;
if (name_terminate(start, line-start, c, terminate))
break;
}
line++;
if (c == '/' && !--p_value)
start = line;
}
if (!start)
return squash_slash(null_strdup(def));
len = line - start;
if (!len)
return squash_slash(null_strdup(def));
/*
* Generally we prefer the shorter name, especially
* if the other one is just a variation of that with
* something else tacked on to the end (ie "file.orig"
* or "file~").
*/
if (def) {
int deflen = strlen(def);
if (deflen < len && !strncmp(start, def, deflen))
return squash_slash(xstrdup(def));
}
if (root) {
char *ret = xmalloc(root_len + len + 1);
strcpy(ret, root);
memcpy(ret + root_len, start, len);
ret[root_len + len] = '\0';
return squash_slash(ret);
}
return squash_slash(xmemdupz(start, len));
}
static char *find_name(const char *line, char *def, int p_value, int terminate)
{
if (*line == '"') {
char *name = find_name_gnu(line, def, p_value);
if (name)
return name;
}
return find_name_common(line, def, p_value, NULL, terminate);
}
static char *find_name_traditional(const char *line, char *def, int p_value)
{
size_t len;
size_t date_len;
if (*line == '"') {
char *name = find_name_gnu(line, def, p_value);
if (name)
return name;
}
len = strchrnul(line, '\n') - line;
date_len = diff_timestamp_len(line, len);
if (!date_len)
return find_name_common(line, def, p_value, NULL, TERM_TAB);
len -= date_len;
return find_name_common(line, def, p_value, line + len, 0);
}
static int count_slashes(const char *cp)
{
int cnt = 0;
char ch;
while ((ch = *cp++))
if (ch == '/')
cnt++;
return cnt;
}
/*
* Given the string after "--- " or "+++ ", guess the appropriate
* p_value for the given patch.
*/
static int guess_p_value(const char *nameline)
{
char *name, *cp;
int val = -1;
if (is_dev_null(nameline))
return -1;
name = find_name_traditional(nameline, NULL, 0);
if (!name)
return -1;
cp = strchr(name, '/');
if (!cp)
val = 0;
else if (prefix) {
/*
* Does it begin with "a/$our-prefix" and such? Then this is
* very likely to apply to our directory.
*/
if (!strncmp(name, prefix, prefix_length))
val = count_slashes(prefix);
else {
cp++;
if (!strncmp(cp, prefix, prefix_length))
val = count_slashes(prefix) + 1;
}
}
free(name);
return val;
}
/*
* Does the ---/+++ line has the POSIX timestamp after the last HT?
* GNU diff puts epoch there to signal a creation/deletion event. Is
* this such a timestamp?
*/
static int has_epoch_timestamp(const char *nameline)
{
/*
* We are only interested in epoch timestamp; any non-zero
* fraction cannot be one, hence "(\.0+)?" in the regexp below.
* For the same reason, the date must be either 1969-12-31 or
* 1970-01-01, and the seconds part must be "00".
*/
const char stamp_regexp[] =
"^(1969-12-31|1970-01-01)"
" "
"[0-2][0-9]:[0-5][0-9]:00(\\.0+)?"
" "
"([-+][0-2][0-9]:?[0-5][0-9])\n";
const char *timestamp = NULL, *cp, *colon;
static regex_t *stamp;
regmatch_t m[10];
int zoneoffset;
int hourminute;
int status;
for (cp = nameline; *cp != '\n'; cp++) {
if (*cp == '\t')
timestamp = cp + 1;
}
if (!timestamp)
return 0;
if (!stamp) {
stamp = xmalloc(sizeof(*stamp));
if (regcomp(stamp, stamp_regexp, REG_EXTENDED)) {
warning(_("Cannot prepare timestamp regexp %s"),
stamp_regexp);
return 0;
}
}
status = regexec(stamp, timestamp, ARRAY_SIZE(m), m, 0);
if (status) {
if (status != REG_NOMATCH)
warning(_("regexec returned %d for input: %s"),
status, timestamp);
return 0;
}
zoneoffset = strtol(timestamp + m[3].rm_so + 1, (char **) &colon, 10);
if (*colon == ':')
zoneoffset = zoneoffset * 60 + strtol(colon + 1, NULL, 10);
else
zoneoffset = (zoneoffset / 100) * 60 + (zoneoffset % 100);
if (timestamp[m[3].rm_so] == '-')
zoneoffset = -zoneoffset;
/*
* YYYY-MM-DD hh:mm:ss must be from either 1969-12-31
* (west of GMT) or 1970-01-01 (east of GMT)
*/
if ((zoneoffset < 0 && memcmp(timestamp, "1969-12-31", 10)) ||
(0 <= zoneoffset && memcmp(timestamp, "1970-01-01", 10)))
return 0;
hourminute = (strtol(timestamp + 11, NULL, 10) * 60 +
strtol(timestamp + 14, NULL, 10) -
zoneoffset);
return ((zoneoffset < 0 && hourminute == 1440) ||
(0 <= zoneoffset && !hourminute));
}
/*
* Get the name etc info from the ---/+++ lines of a traditional patch header
*
* FIXME! The end-of-filename heuristics are kind of screwy. For existing
* files, we can happily check the index for a match, but for creating a
* new file we should try to match whatever "patch" does. I have no idea.
*/
static void parse_traditional_patch(const char *first, const char *second, struct patch *patch)
{
char *name;
first += 4; /* skip "--- " */
second += 4; /* skip "+++ " */
if (!p_value_known) {
int p, q;
p = guess_p_value(first);
q = guess_p_value(second);
if (p < 0) p = q;
if (0 <= p && p == q) {
p_value = p;
p_value_known = 1;
}
}
if (is_dev_null(first)) {
patch->is_new = 1;
patch->is_delete = 0;
name = find_name_traditional(second, NULL, p_value);
patch->new_name = name;
} else if (is_dev_null(second)) {
patch->is_new = 0;
patch->is_delete = 1;
name = find_name_traditional(first, NULL, p_value);
patch->old_name = name;
} else {
char *first_name;
first_name = find_name_traditional(first, NULL, p_value);
name = find_name_traditional(second, first_name, p_value);
free(first_name);
if (has_epoch_timestamp(first)) {
patch->is_new = 1;
patch->is_delete = 0;
patch->new_name = name;
} else if (has_epoch_timestamp(second)) {
patch->is_new = 0;
patch->is_delete = 1;
patch->old_name = name;
} else {
patch->old_name = name;
patch->new_name = null_strdup(name);
}
}
if (!name)
die(_("unable to find filename in patch at line %d"), linenr);
}
static int gitdiff_hdrend(const char *line, struct patch *patch)
{
return -1;
}
/*
* We're anal about diff header consistency, to make
* sure that we don't end up having strange ambiguous
* patches floating around.
*
* As a result, gitdiff_{old|new}name() will check
* their names against any previous information, just
* to make sure..
*/
#define DIFF_OLD_NAME 0
#define DIFF_NEW_NAME 1
static char *gitdiff_verify_name(const char *line, int isnull, char *orig_name, int side)
{
if (!orig_name && !isnull)
return find_name(line, NULL, p_value, TERM_TAB);
if (orig_name) {
int len;
const char *name;
char *another;
name = orig_name;
len = strlen(name);
if (isnull)
die(_("git apply: bad git-diff - expected /dev/null, got %s on line %d"), name, linenr);
another = find_name(line, NULL, p_value, TERM_TAB);
if (!another || memcmp(another, name, len + 1))
die((side == DIFF_NEW_NAME) ?
_("git apply: bad git-diff - inconsistent new filename on line %d") :
_("git apply: bad git-diff - inconsistent old filename on line %d"), linenr);
free(another);
return orig_name;
}
else {
/* expect "/dev/null" */
if (memcmp("/dev/null", line, 9) || line[9] != '\n')
die(_("git apply: bad git-diff - expected /dev/null on line %d"), linenr);
return NULL;
}
}
static int gitdiff_oldname(const char *line, struct patch *patch)
{
char *orig = patch->old_name;
patch->old_name = gitdiff_verify_name(line, patch->is_new, patch->old_name,
DIFF_OLD_NAME);
if (orig != patch->old_name)
free(orig);
return 0;
}
static int gitdiff_newname(const char *line, struct patch *patch)
{
char *orig = patch->new_name;
patch->new_name = gitdiff_verify_name(line, patch->is_delete, patch->new_name,
DIFF_NEW_NAME);
if (orig != patch->new_name)
free(orig);
return 0;
}
static int gitdiff_oldmode(const char *line, struct patch *patch)
{
patch->old_mode = strtoul(line, NULL, 8);
return 0;
}
static int gitdiff_newmode(const char *line, struct patch *patch)
{
patch->new_mode = strtoul(line, NULL, 8);
return 0;
}
static int gitdiff_delete(const char *line, struct patch *patch)
{
patch->is_delete = 1;
free(patch->old_name);
patch->old_name = null_strdup(patch->def_name);
return gitdiff_oldmode(line, patch);
}
static int gitdiff_newfile(const char *line, struct patch *patch)
{
patch->is_new = 1;
free(patch->new_name);
patch->new_name = null_strdup(patch->def_name);
return gitdiff_newmode(line, patch);
}
static int gitdiff_copysrc(const char *line, struct patch *patch)
{
patch->is_copy = 1;
free(patch->old_name);
patch->old_name = find_name(line, NULL, p_value ? p_value - 1 : 0, 0);
return 0;
}
static int gitdiff_copydst(const char *line, struct patch *patch)
{
patch->is_copy = 1;
free(patch->new_name);
patch->new_name = find_name(line, NULL, p_value ? p_value - 1 : 0, 0);
return 0;
}
static int gitdiff_renamesrc(const char *line, struct patch *patch)
{
patch->is_rename = 1;
free(patch->old_name);
patch->old_name = find_name(line, NULL, p_value ? p_value - 1 : 0, 0);
return 0;
}
static int gitdiff_renamedst(const char *line, struct patch *patch)
{
patch->is_rename = 1;
free(patch->new_name);
patch->new_name = find_name(line, NULL, p_value ? p_value - 1 : 0, 0);
return 0;
}
static int gitdiff_similarity(const char *line, struct patch *patch)
{
unsigned long val = strtoul(line, NULL, 10);
if (val <= 100)
patch->score = val;
return 0;
}
static int gitdiff_dissimilarity(const char *line, struct patch *patch)
{
unsigned long val = strtoul(line, NULL, 10);
if (val <= 100)
patch->score = val;
return 0;
}
static int gitdiff_index(const char *line, struct patch *patch)
{
/*
* index line is N hexadecimal, "..", N hexadecimal,
* and optional space with octal mode.
*/
const char *ptr, *eol;
int len;
ptr = strchr(line, '.');
if (!ptr || ptr[1] != '.' || 40 < ptr - line)
return 0;
len = ptr - line;
memcpy(patch->old_sha1_prefix, line, len);
patch->old_sha1_prefix[len] = 0;
line = ptr + 2;
ptr = strchr(line, ' ');
eol = strchrnul(line, '\n');
if (!ptr || eol < ptr)
ptr = eol;
len = ptr - line;
if (40 < len)
return 0;
memcpy(patch->new_sha1_prefix, line, len);
patch->new_sha1_prefix[len] = 0;
if (*ptr == ' ')
patch->old_mode = strtoul(ptr+1, NULL, 8);
return 0;
}
/*
* This is normal for a diff that doesn't change anything: we'll fall through
* into the next diff. Tell the parser to break out.
*/
static int gitdiff_unrecognized(const char *line, struct patch *patch)
{
return -1;
}
/*
* Skip p_value leading components from "line"; as we do not accept
* absolute paths, return NULL in that case.
*/
static const char *skip_tree_prefix(const char *line, int llen)
{
int nslash;
int i;
if (!p_value)
return (llen && line[0] == '/') ? NULL : line;
nslash = p_value;
for (i = 0; i < llen; i++) {
int ch = line[i];
if (ch == '/' && --nslash <= 0)
return (i == 0) ? NULL : &line[i + 1];
}
return NULL;
}
/*
* This is to extract the same name that appears on "diff --git"
* line. We do not find and return anything if it is a rename
* patch, and it is OK because we will find the name elsewhere.
* We need to reliably find name only when it is mode-change only,
* creation or deletion of an empty file. In any of these cases,
* both sides are the same name under a/ and b/ respectively.
*/
static char *git_header_name(const char *line, int llen)
{
const char *name;
const char *second = NULL;
size_t len, line_len;
line += strlen("diff --git ");
llen -= strlen("diff --git ");
if (*line == '"') {
const char *cp;
struct strbuf first = STRBUF_INIT;
struct strbuf sp = STRBUF_INIT;
if (unquote_c_style(&first, line, &second))
goto free_and_fail1;
/* strip the a/b prefix including trailing slash */
cp = skip_tree_prefix(first.buf, first.len);
if (!cp)
goto free_and_fail1;
strbuf_remove(&first, 0, cp - first.buf);
/*
* second points at one past closing dq of name.
* find the second name.
*/
while ((second < line + llen) && isspace(*second))
second++;
if (line + llen <= second)
goto free_and_fail1;
if (*second == '"') {
if (unquote_c_style(&sp, second, NULL))
goto free_and_fail1;
cp = skip_tree_prefix(sp.buf, sp.len);
if (!cp)
goto free_and_fail1;
/* They must match, otherwise ignore */
if (strcmp(cp, first.buf))
goto free_and_fail1;
strbuf_release(&sp);
return strbuf_detach(&first, NULL);
}
/* unquoted second */
cp = skip_tree_prefix(second, line + llen - second);
if (!cp)
goto free_and_fail1;
if (line + llen - cp != first.len ||
memcmp(first.buf, cp, first.len))
goto free_and_fail1;
return strbuf_detach(&first, NULL);
free_and_fail1:
strbuf_release(&first);
strbuf_release(&sp);
return NULL;
}
/* unquoted first name */
name = skip_tree_prefix(line, llen);
if (!name)
return NULL;
/*
* since the first name is unquoted, a dq if exists must be
* the beginning of the second name.
*/
for (second = name; second < line + llen; second++) {
if (*second == '"') {
struct strbuf sp = STRBUF_INIT;
const char *np;
if (unquote_c_style(&sp, second, NULL))
goto free_and_fail2;
np = skip_tree_prefix(sp.buf, sp.len);
if (!np)
goto free_and_fail2;
len = sp.buf + sp.len - np;
if (len < second - name &&
!strncmp(np, name, len) &&
isspace(name[len])) {
/* Good */
strbuf_remove(&sp, 0, np - sp.buf);
return strbuf_detach(&sp, NULL);
}
free_and_fail2:
strbuf_release(&sp);
return NULL;
}
}
/*
* Accept a name only if it shows up twice, exactly the same
* form.
*/
second = strchr(name, '\n');
if (!second)
return NULL;
line_len = second - name;
for (len = 0 ; ; len++) {
switch (name[len]) {
default:
continue;
case '\n':
return NULL;
case '\t': case ' ':
/*
* Is this the separator between the preimage
* and the postimage pathname? Again, we are
* only interested in the case where there is
* no rename, as this is only to set def_name
* and a rename patch has the names elsewhere
* in an unambiguous form.
*/
if (!name[len + 1])
return NULL; /* no postimage name */
second = skip_tree_prefix(name + len + 1,
line_len - (len + 1));
if (!second)
return NULL;
/*
* Does len bytes starting at "name" and "second"
* (that are separated by one HT or SP we just
* found) exactly match?
*/
if (second[len] == '\n' && !strncmp(name, second, len))
return xmemdupz(name, len);
}
}
}
/* Verify that we recognize the lines following a git header */
static int parse_git_header(const char *line, int len, unsigned int size, struct patch *patch)
{
unsigned long offset;
/* A git diff has explicit new/delete information, so we don't guess */
patch->is_new = 0;
patch->is_delete = 0;
/*
* Some things may not have the old name in the
* rest of the headers anywhere (pure mode changes,
* or removing or adding empty files), so we get
* the default name from the header.
*/
patch->def_name = git_header_name(line, len);
if (patch->def_name && root) {
char *s = xstrfmt("%s%s", root, patch->def_name);
free(patch->def_name);
patch->def_name = s;
}
line += len;
size -= len;
linenr++;
for (offset = len ; size > 0 ; offset += len, size -= len, line += len, linenr++) {
static const struct opentry {
const char *str;
int (*fn)(const char *, struct patch *);
} optable[] = {
{ "@@ -", gitdiff_hdrend },
{ "--- ", gitdiff_oldname },
{ "+++ ", gitdiff_newname },
{ "old mode ", gitdiff_oldmode },
{ "new mode ", gitdiff_newmode },
{ "deleted file mode ", gitdiff_delete },
{ "new file mode ", gitdiff_newfile },
{ "copy from ", gitdiff_copysrc },
{ "copy to ", gitdiff_copydst },
{ "rename old ", gitdiff_renamesrc },
{ "rename new ", gitdiff_renamedst },
{ "rename from ", gitdiff_renamesrc },
{ "rename to ", gitdiff_renamedst },
{ "similarity index ", gitdiff_similarity },
{ "dissimilarity index ", gitdiff_dissimilarity },
{ "index ", gitdiff_index },
{ "", gitdiff_unrecognized },
};
int i;
len = linelen(line, size);
if (!len || line[len-1] != '\n')
break;
for (i = 0; i < ARRAY_SIZE(optable); i++) {
const struct opentry *p = optable + i;
int oplen = strlen(p->str);
if (len < oplen || memcmp(p->str, line, oplen))
continue;
if (p->fn(line + oplen, patch) < 0)
return offset;
break;
}
}
return offset;
}
static int parse_num(const char *line, unsigned long *p)
{
char *ptr;
if (!isdigit(*line))
return 0;
*p = strtoul(line, &ptr, 10);
return ptr - line;
}
static int parse_range(const char *line, int len, int offset, const char *expect,
unsigned long *p1, unsigned long *p2)
{
int digits, ex;
if (offset < 0 || offset >= len)
return -1;
line += offset;
len -= offset;
digits = parse_num(line, p1);
if (!digits)
return -1;
offset += digits;
line += digits;
len -= digits;
*p2 = 1;
if (*line == ',') {
digits = parse_num(line+1, p2);
if (!digits)
return -1;
offset += digits+1;
line += digits+1;
len -= digits+1;
}
ex = strlen(expect);
if (ex > len)
return -1;
if (memcmp(line, expect, ex))
return -1;
return offset + ex;
}
static void recount_diff(const char *line, int size, struct fragment *fragment)
{
int oldlines = 0, newlines = 0, ret = 0;
if (size < 1) {
warning("recount: ignore empty hunk");
return;
}
for (;;) {
int len = linelen(line, size);
size -= len;
line += len;
if (size < 1)
break;
switch (*line) {
case ' ': case '\n':
newlines++;
/* fall through */
case '-':
oldlines++;
continue;
case '+':
newlines++;
continue;
case '\\':
continue;
case '@':
ret = size < 3 || !starts_with(line, "@@ ");
break;
case 'd':
ret = size < 5 || !starts_with(line, "diff ");
break;
default:
ret = -1;
break;
}
if (ret) {
warning(_("recount: unexpected line: %.*s"),
(int)linelen(line, size), line);
return;
}
break;
}
fragment->oldlines = oldlines;
fragment->newlines = newlines;
}
/*
* Parse a unified diff fragment header of the
* form "@@ -a,b +c,d @@"
*/
static int parse_fragment_header(const char *line, int len, struct fragment *fragment)
{
int offset;
if (!len || line[len-1] != '\n')
return -1;
/* Figure out the number of lines in a fragment */
offset = parse_range(line, len, 4, " +", &fragment->oldpos, &fragment->oldlines);
offset = parse_range(line, len, offset, " @@", &fragment->newpos, &fragment->newlines);
return offset;
}
static int find_header(const char *line, unsigned long size, int *hdrsize, struct patch *patch)
{
unsigned long offset, len;
patch->is_toplevel_relative = 0;
patch->is_rename = patch->is_copy = 0;
patch->is_new = patch->is_delete = -1;
patch->old_mode = patch->new_mode = 0;
patch->old_name = patch->new_name = NULL;
for (offset = 0; size > 0; offset += len, size -= len, line += len, linenr++) {
unsigned long nextlen;
len = linelen(line, size);
if (!len)
break;
/* Testing this early allows us to take a few shortcuts.. */
if (len < 6)
continue;
/*
* Make sure we don't find any unconnected patch fragments.
* That's a sign that we didn't find a header, and that a
* patch has become corrupted/broken up.
*/
if (!memcmp("@@ -", line, 4)) {
struct fragment dummy;
if (parse_fragment_header(line, len, &dummy) < 0)
continue;
die(_("patch fragment without header at line %d: %.*s"),
linenr, (int)len-1, line);
}
if (size < len + 6)
break;
/*
* Git patch? It might not have a real patch, just a rename
* or mode change, so we handle that specially
*/
if (!memcmp("diff --git ", line, 11)) {
int git_hdr_len = parse_git_header(line, len, size, patch);
if (git_hdr_len <= len)
continue;
if (!patch->old_name && !patch->new_name) {
if (!patch->def_name)
die(Q_("git diff header lacks filename information when removing "
"%d leading pathname component (line %d)",
"git diff header lacks filename information when removing "
"%d leading pathname components (line %d)",
p_value),
p_value, linenr);
patch->old_name = xstrdup(patch->def_name);
patch->new_name = xstrdup(patch->def_name);
}
if (!patch->is_delete && !patch->new_name)
die("git diff header lacks filename information "
"(line %d)", linenr);
patch->is_toplevel_relative = 1;
*hdrsize = git_hdr_len;
return offset;
}
/* --- followed by +++ ? */
if (memcmp("--- ", line, 4) || memcmp("+++ ", line + len, 4))
continue;
/*
* We only accept unified patches, so we want it to
* at least have "@@ -a,b +c,d @@\n", which is 14 chars
* minimum ("@@ -0,0 +1 @@\n" is the shortest).
*/
nextlen = linelen(line + len, size - len);
if (size < nextlen + 14 || memcmp("@@ -", line + len + nextlen, 4))
continue;
/* Ok, we'll consider it a patch */
parse_traditional_patch(line, line+len, patch);
*hdrsize = len + nextlen;
linenr += 2;
return offset;
}
return -1;
}
static void record_ws_error(unsigned result, const char *line, int len, int linenr)
{
char *err;
if (!result)
return;
whitespace_error++;
if (squelch_whitespace_errors &&
squelch_whitespace_errors < whitespace_error)
return;
err = whitespace_error_string(result);
fprintf(stderr, "%s:%d: %s.\n%.*s\n",
patch_input_file, linenr, err, len, line);
free(err);
}
static void check_whitespace(const char *line, int len, unsigned ws_rule)
{
unsigned result = ws_check(line + 1, len - 1, ws_rule);
record_ws_error(result, line + 1, len - 2, linenr);
}
/*
* Parse a unified diff. Note that this really needs to parse each
* fragment separately, since the only way to know the difference
* between a "---" that is part of a patch, and a "---" that starts
* the next patch is to look at the line counts..
*/
static int parse_fragment(const char *line, unsigned long size,
struct patch *patch, struct fragment *fragment)
{
int added, deleted;
int len = linelen(line, size), offset;
unsigned long oldlines, newlines;
unsigned long leading, trailing;
offset = parse_fragment_header(line, len, fragment);
if (offset < 0)
return -1;
if (offset > 0 && patch->recount)
recount_diff(line + offset, size - offset, fragment);
oldlines = fragment->oldlines;
newlines = fragment->newlines;
leading = 0;
trailing = 0;
/* Parse the thing.. */
line += len;
size -= len;
linenr++;
added = deleted = 0;
for (offset = len;
0 < size;
offset += len, size -= len, line += len, linenr++) {
if (!oldlines && !newlines)
break;
len = linelen(line, size);
if (!len || line[len-1] != '\n')
return -1;
switch (*line) {
default:
return -1;
case '\n': /* newer GNU diff, an empty context line */
case ' ':
oldlines--;
newlines--;
if (!deleted && !added)
leading++;
trailing++;
break;
case '-':
if (apply_in_reverse &&
ws_error_action != nowarn_ws_error)
check_whitespace(line, len, patch->ws_rule);
deleted++;
oldlines--;
trailing = 0;
break;
case '+':
if (!apply_in_reverse &&
ws_error_action != nowarn_ws_error)
check_whitespace(line, len, patch->ws_rule);
added++;
newlines--;
trailing = 0;
break;
/*
* We allow "\ No newline at end of file". Depending
* on locale settings when the patch was produced we
* don't know what this line looks like. The only
* thing we do know is that it begins with "\ ".
* Checking for 12 is just for sanity check -- any
* l10n of "\ No newline..." is at least that long.
*/
case '\\':
if (len < 12 || memcmp(line, "\\ ", 2))
return -1;
break;
}
}
if (oldlines || newlines)
return -1;
fragment->leading = leading;
fragment->trailing = trailing;
/*
* If a fragment ends with an incomplete line, we failed to include
* it in the above loop because we hit oldlines == newlines == 0
* before seeing it.
*/
if (12 < size && !memcmp(line, "\\ ", 2))
offset += linelen(line, size);
patch->lines_added += added;
patch->lines_deleted += deleted;
if (0 < patch->is_new && oldlines)
return error(_("new file depends on old contents"));
if (0 < patch->is_delete && newlines)
return error(_("deleted file still has contents"));
return offset;
}
/*
* We have seen "diff --git a/... b/..." header (or a traditional patch
* header). Read hunks that belong to this patch into fragments and hang
* them to the given patch structure.
*
* The (fragment->patch, fragment->size) pair points into the memory given
* by the caller, not a copy, when we return.
*/
static int parse_single_patch(const char *line, unsigned long size, struct patch *patch)
{
unsigned long offset = 0;
unsigned long oldlines = 0, newlines = 0, context = 0;
struct fragment **fragp = &patch->fragments;
while (size > 4 && !memcmp(line, "@@ -", 4)) {
struct fragment *fragment;
int len;
fragment = xcalloc(1, sizeof(*fragment));
fragment->linenr = linenr;
len = parse_fragment(line, size, patch, fragment);
if (len <= 0)
die(_("corrupt patch at line %d"), linenr);
fragment->patch = line;
fragment->size = len;
oldlines += fragment->oldlines;
newlines += fragment->newlines;
context += fragment->leading + fragment->trailing;
*fragp = fragment;
fragp = &fragment->next;
offset += len;
line += len;
size -= len;
}
/*
* If something was removed (i.e. we have old-lines) it cannot
* be creation, and if something was added it cannot be
* deletion. However, the reverse is not true; --unified=0
* patches that only add are not necessarily creation even
* though they do not have any old lines, and ones that only
* delete are not necessarily deletion.
*
* Unfortunately, a real creation/deletion patch do _not_ have
* any context line by definition, so we cannot safely tell it
* apart with --unified=0 insanity. At least if the patch has
* more than one hunk it is not creation or deletion.
*/
if (patch->is_new < 0 &&
(oldlines || (patch->fragments && patch->fragments->next)))
patch->is_new = 0;
if (patch->is_delete < 0 &&
(newlines || (patch->fragments && patch->fragments->next)))
patch->is_delete = 0;
if (0 < patch->is_new && oldlines)
die(_("new file %s depends on old contents"), patch->new_name);
if (0 < patch->is_delete && newlines)
die(_("deleted file %s still has contents"), patch->old_name);
if (!patch->is_delete && !newlines && context)
fprintf_ln(stderr,
_("** warning: "
"file %s becomes empty but is not deleted"),
patch->new_name);
return offset;
}
static inline int metadata_changes(struct patch *patch)
{
return patch->is_rename > 0 ||
patch->is_copy > 0 ||
patch->is_new > 0 ||
patch->is_delete ||
(patch->old_mode && patch->new_mode &&
patch->old_mode != patch->new_mode);
}
static char *inflate_it(const void *data, unsigned long size,
unsigned long inflated_size)
{
git_zstream stream;
void *out;
int st;
memset(&stream, 0, sizeof(stream));
stream.next_in = (unsigned char *)data;
stream.avail_in = size;
stream.next_out = out = xmalloc(inflated_size);
stream.avail_out = inflated_size;
git_inflate_init(&stream);
st = git_inflate(&stream, Z_FINISH);
git_inflate_end(&stream);
if ((st != Z_STREAM_END) || stream.total_out != inflated_size) {
free(out);
return NULL;
}
return out;
}
/*
* Read a binary hunk and return a new fragment; fragment->patch
* points at an allocated memory that the caller must free, so
* it is marked as "->free_patch = 1".
*/
static struct fragment *parse_binary_hunk(char **buf_p,
unsigned long *sz_p,
int *status_p,
int *used_p)
{
/*
* Expect a line that begins with binary patch method ("literal"
* or "delta"), followed by the length of data before deflating.
* a sequence of 'length-byte' followed by base-85 encoded data
* should follow, terminated by a newline.
*
* Each 5-byte sequence of base-85 encodes up to 4 bytes,
* and we would limit the patch line to 66 characters,
* so one line can fit up to 13 groups that would decode
* to 52 bytes max. The length byte 'A'-'Z' corresponds
* to 1-26 bytes, and 'a'-'z' corresponds to 27-52 bytes.
*/
int llen, used;
unsigned long size = *sz_p;
char *buffer = *buf_p;
int patch_method;
unsigned long origlen;
char *data = NULL;
int hunk_size = 0;
struct fragment *frag;
llen = linelen(buffer, size);
used = llen;
*status_p = 0;
if (starts_with(buffer, "delta ")) {
patch_method = BINARY_DELTA_DEFLATED;
origlen = strtoul(buffer + 6, NULL, 10);
}
else if (starts_with(buffer, "literal ")) {
patch_method = BINARY_LITERAL_DEFLATED;
origlen = strtoul(buffer + 8, NULL, 10);
}
else
return NULL;
linenr++;
buffer += llen;
while (1) {
int byte_length, max_byte_length, newsize;
llen = linelen(buffer, size);
used += llen;
linenr++;
if (llen == 1) {
/* consume the blank line */
buffer++;
size--;
break;
}
/*
* Minimum line is "A00000\n" which is 7-byte long,
* and the line length must be multiple of 5 plus 2.
*/
if ((llen < 7) || (llen-2) % 5)
goto corrupt;
max_byte_length = (llen - 2) / 5 * 4;
byte_length = *buffer;
if ('A' <= byte_length && byte_length <= 'Z')
byte_length = byte_length - 'A' + 1;
else if ('a' <= byte_length && byte_length <= 'z')
byte_length = byte_length - 'a' + 27;
else
goto corrupt;
/* if the input length was not multiple of 4, we would
* have filler at the end but the filler should never
* exceed 3 bytes
*/
if (max_byte_length < byte_length ||
byte_length <= max_byte_length - 4)
goto corrupt;
newsize = hunk_size + byte_length;
data = xrealloc(data, newsize);
if (decode_85(data + hunk_size, buffer + 1, byte_length))
goto corrupt;
hunk_size = newsize;
buffer += llen;
size -= llen;
}
frag = xcalloc(1, sizeof(*frag));
frag->patch = inflate_it(data, hunk_size, origlen);
frag->free_patch = 1;
if (!frag->patch)
goto corrupt;
free(data);
frag->size = origlen;
*buf_p = buffer;
*sz_p = size;
*used_p = used;
frag->binary_patch_method = patch_method;
return frag;
corrupt:
free(data);
*status_p = -1;
error(_("corrupt binary patch at line %d: %.*s"),
linenr-1, llen-1, buffer);
return NULL;
}
static int parse_binary(char *buffer, unsigned long size, struct patch *patch)
{
/*
* We have read "GIT binary patch\n"; what follows is a line
* that says the patch method (currently, either "literal" or
* "delta") and the length of data before deflating; a
* sequence of 'length-byte' followed by base-85 encoded data
* follows.
*
* When a binary patch is reversible, there is another binary
* hunk in the same format, starting with patch method (either
* "literal" or "delta") with the length of data, and a sequence
* of length-byte + base-85 encoded data, terminated with another
* empty line. This data, when applied to the postimage, produces
* the preimage.
*/
struct fragment *forward;
struct fragment *reverse;
int status;
int used, used_1;
forward = parse_binary_hunk(&buffer, &size, &status, &used);
if (!forward && !status)
/* there has to be one hunk (forward hunk) */
return error(_("unrecognized binary patch at line %d"), linenr-1);
if (status)
/* otherwise we already gave an error message */
return status;
reverse = parse_binary_hunk(&buffer, &size, &status, &used_1);
if (reverse)
used += used_1;
else if (status) {
/*
* Not having reverse hunk is not an error, but having
* a corrupt reverse hunk is.
*/
free((void*) forward->patch);
free(forward);
return status;
}
forward->next = reverse;
patch->fragments = forward;
patch->is_binary = 1;
return used;
}
static void prefix_one(char **name)
{
char *old_name = *name;
if (!old_name)
return;
*name = xstrdup(prefix_filename(prefix, prefix_length, *name));
free(old_name);
}
static void prefix_patch(struct patch *p)
{
if (!prefix || p->is_toplevel_relative)
return;
prefix_one(&p->new_name);
prefix_one(&p->old_name);
}
/*
* include/exclude
*/
static struct string_list limit_by_name;
static int has_include;
static void add_name_limit(const char *name, int exclude)
{
struct string_list_item *it;
it = string_list_append(&limit_by_name, name);
it->util = exclude ? NULL : (void *) 1;
}
static int use_patch(struct patch *p)
{
const char *pathname = p->new_name ? p->new_name : p->old_name;
int i;
/* Paths outside are not touched regardless of "--include" */
if (0 < prefix_length) {
int pathlen = strlen(pathname);
if (pathlen <= prefix_length ||
memcmp(prefix, pathname, prefix_length))
return 0;
}
/* See if it matches any of exclude/include rule */
for (i = 0; i < limit_by_name.nr; i++) {
struct string_list_item *it = &limit_by_name.items[i];
if (!wildmatch(it->string, pathname, 0, NULL))
return (it->util != NULL);
}
/*
* If we had any include, a path that does not match any rule is
* not used. Otherwise, we saw bunch of exclude rules (or none)
* and such a path is used.
*/
return !has_include;
}
/*
* Read the patch text in "buffer" that extends for "size" bytes; stop
* reading after seeing a single patch (i.e. changes to a single file).
* Create fragments (i.e. patch hunks) and hang them to the given patch.
* Return the number of bytes consumed, so that the caller can call us
* again for the next patch.
*/
static int parse_chunk(char *buffer, unsigned long size, struct patch *patch)
{
int hdrsize, patchsize;
int offset = find_header(buffer, size, &hdrsize, patch);
if (offset < 0)
return offset;
prefix_patch(patch);
if (!use_patch(patch))
patch->ws_rule = 0;
else
patch->ws_rule = whitespace_rule(patch->new_name
? patch->new_name
: patch->old_name);
patchsize = parse_single_patch(buffer + offset + hdrsize,
size - offset - hdrsize, patch);
if (!patchsize) {
static const char git_binary[] = "GIT binary patch\n";
int hd = hdrsize + offset;
unsigned long llen = linelen(buffer + hd, size - hd);
if (llen == sizeof(git_binary) - 1 &&
!memcmp(git_binary, buffer + hd, llen)) {
int used;
linenr++;
used = parse_binary(buffer + hd + llen,
size - hd - llen, patch);
if (used)
patchsize = used + llen;
else
patchsize = 0;
}
else if (!memcmp(" differ\n", buffer + hd + llen - 8, 8)) {
static const char *binhdr[] = {
"Binary files ",
"Files ",
NULL,
};
int i;
for (i = 0; binhdr[i]; i++) {
int len = strlen(binhdr[i]);
if (len < size - hd &&
!memcmp(binhdr[i], buffer + hd, len)) {
linenr++;
patch->is_binary = 1;
patchsize = llen;
break;
}
}
}
/* Empty patch cannot be applied if it is a text patch
* without metadata change. A binary patch appears
* empty to us here.
*/
if ((apply || check) &&
(!patch->is_binary && !metadata_changes(patch)))
die(_("patch with only garbage at line %d"), linenr);
}
return offset + hdrsize + patchsize;
}
#define swap(a,b) myswap((a),(b),sizeof(a))
#define myswap(a, b, size) do { \
unsigned char mytmp[size]; \
memcpy(mytmp, &a, size); \
memcpy(&a, &b, size); \
memcpy(&b, mytmp, size); \
} while (0)
static void reverse_patches(struct patch *p)
{
for (; p; p = p->next) {
struct fragment *frag = p->fragments;
swap(p->new_name, p->old_name);
swap(p->new_mode, p->old_mode);
swap(p->is_new, p->is_delete);
swap(p->lines_added, p->lines_deleted);
swap(p->old_sha1_prefix, p->new_sha1_prefix);
for (; frag; frag = frag->next) {
swap(frag->newpos, frag->oldpos);
swap(frag->newlines, frag->oldlines);
}
}
}
static const char pluses[] =
"++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++";
static const char minuses[]=
"----------------------------------------------------------------------";
static void show_stats(struct patch *patch)
{
struct strbuf qname = STRBUF_INIT;
char *cp = patch->new_name ? patch->new_name : patch->old_name;
int max, add, del;
quote_c_style(cp, &qname, NULL, 0);
/*
* "scale" the filename
*/
max = max_len;
if (max > 50)
max = 50;
if (qname.len > max) {
cp = strchr(qname.buf + qname.len + 3 - max, '/');
if (!cp)
cp = qname.buf + qname.len + 3 - max;
strbuf_splice(&qname, 0, cp - qname.buf, "...", 3);
}
if (patch->is_binary) {
printf(" %-*s | Bin\n", max, qname.buf);
strbuf_release(&qname);
return;
}
printf(" %-*s |", max, qname.buf);
strbuf_release(&qname);
/*
* scale the add/delete
*/
max = max + max_change > 70 ? 70 - max : max_change;
add = patch->lines_added;
del = patch->lines_deleted;
if (max_change > 0) {
int total = ((add + del) * max + max_change / 2) / max_change;
add = (add * max + max_change / 2) / max_change;
del = total - add;
}
printf("%5d %.*s%.*s\n", patch->lines_added + patch->lines_deleted,
add, pluses, del, minuses);
}
static int read_old_data(struct stat *st, const char *path, struct strbuf *buf)
{
switch (st->st_mode & S_IFMT) {
case S_IFLNK:
if (strbuf_readlink(buf, path, st->st_size) < 0)
return error(_("unable to read symlink %s"), path);
return 0;
case S_IFREG:
if (strbuf_read_file(buf, path, st->st_size) != st->st_size)
return error(_("unable to open or read %s"), path);
convert_to_git(path, buf->buf, buf->len, buf, 0);
return 0;
default:
return -1;
}
}
/*
* Update the preimage, and the common lines in postimage,
* from buffer buf of length len. If postlen is 0 the postimage
* is updated in place, otherwise it's updated on a new buffer
* of length postlen
*/
static void update_pre_post_images(struct image *preimage,
struct image *postimage,
char *buf,
size_t len, size_t postlen)
{
int i, ctx, reduced;
char *new, *old, *fixed;
struct image fixed_preimage;
/*
* Update the preimage with whitespace fixes. Note that we
* are not losing preimage->buf -- apply_one_fragment() will
* free "oldlines".
*/
prepare_image(&fixed_preimage, buf, len, 1);
assert(postlen
? fixed_preimage.nr == preimage->nr
: fixed_preimage.nr <= preimage->nr);
for (i = 0; i < fixed_preimage.nr; i++)
fixed_preimage.line[i].flag = preimage->line[i].flag;
free(preimage->line_allocated);
*preimage = fixed_preimage;
/*
* Adjust the common context lines in postimage. This can be
* done in-place when we are shrinking it with whitespace
* fixing, but needs a new buffer when ignoring whitespace or
* expanding leading tabs to spaces.
*
* We trust the caller to tell us if the update can be done
* in place (postlen==0) or not.
*/
old = postimage->buf;
if (postlen)
new = postimage->buf = xmalloc(postlen);
else
new = old;
fixed = preimage->buf;
for (i = reduced = ctx = 0; i < postimage->nr; i++) {
size_t len = postimage->line[i].len;
if (!(postimage->line[i].flag & LINE_COMMON)) {
/* an added line -- no counterparts in preimage */
memmove(new, old, len);
old += len;
new += len;
continue;
}
/* a common context -- skip it in the original postimage */
old += len;
/* and find the corresponding one in the fixed preimage */
while (ctx < preimage->nr &&
!(preimage->line[ctx].flag & LINE_COMMON)) {
fixed += preimage->line[ctx].len;
ctx++;
}
/*
* preimage is expected to run out, if the caller
* fixed addition of trailing blank lines.
*/
if (preimage->nr <= ctx) {
reduced++;
continue;
}
/* and copy it in, while fixing the line length */
len = preimage->line[ctx].len;
memcpy(new, fixed, len);
new += len;
fixed += len;
postimage->line[i].len = len;
ctx++;
}
/* Fix the length of the whole thing */
postimage->len = new - postimage->buf;
postimage->nr -= reduced;
}
static int match_fragment(struct image *img,
struct image *preimage,
struct image *postimage,
unsigned long try,
int try_lno,
unsigned ws_rule,
int match_beginning, int match_end)
{
int i;
char *fixed_buf, *buf, *orig, *target;
struct strbuf fixed;
size_t fixed_len, postlen;
int preimage_limit;
if (preimage->nr + try_lno <= img->nr) {
/*
* The hunk falls within the boundaries of img.
*/
preimage_limit = preimage->nr;
if (match_end && (preimage->nr + try_lno != img->nr))
return 0;
} else if (ws_error_action == correct_ws_error &&
(ws_rule & WS_BLANK_AT_EOF)) {
/*
* This hunk extends beyond the end of img, and we are
* removing blank lines at the end of the file. This
* many lines from the beginning of the preimage must
* match with img, and the remainder of the preimage
* must be blank.
*/
preimage_limit = img->nr - try_lno;
} else {
/*
* The hunk extends beyond the end of the img and
* we are not removing blanks at the end, so we
* should reject the hunk at this position.
*/
return 0;
}
if (match_beginning && try_lno)
return 0;
/* Quick hash check */
for (i = 0; i < preimage_limit; i++)
if ((img->line[try_lno + i].flag & LINE_PATCHED) ||
(preimage->line[i].hash != img->line[try_lno + i].hash))
return 0;
if (preimage_limit == preimage->nr) {
/*
* Do we have an exact match? If we were told to match
* at the end, size must be exactly at try+fragsize,
* otherwise try+fragsize must be still within the preimage,
* and either case, the old piece should match the preimage
* exactly.
*/
if ((match_end
? (try + preimage->len == img->len)
: (try + preimage->len <= img->len)) &&
!memcmp(img->buf + try, preimage->buf, preimage->len))
return 1;
} else {
/*
* The preimage extends beyond the end of img, so
* there cannot be an exact match.
*
* There must be one non-blank context line that match
* a line before the end of img.
*/
char *buf_end;
buf = preimage->buf;
buf_end = buf;
for (i = 0; i < preimage_limit; i++)
buf_end += preimage->line[i].len;
for ( ; buf < buf_end; buf++)
if (!isspace(*buf))
break;
if (buf == buf_end)
return 0;
}
/*
* No exact match. If we are ignoring whitespace, run a line-by-line
* fuzzy matching. We collect all the line length information because
* we need it to adjust whitespace if we match.
*/
if (ws_ignore_action == ignore_ws_change) {
size_t imgoff = 0;
size_t preoff = 0;
size_t postlen = postimage->len;
size_t extra_chars;
char *preimage_eof;
char *preimage_end;
for (i = 0; i < preimage_limit; i++) {
size_t prelen = preimage->line[i].len;
size_t imglen = img->line[try_lno+i].len;
if (!fuzzy_matchlines(img->buf + try + imgoff, imglen,
preimage->buf + preoff, prelen))
return 0;
if (preimage->line[i].flag & LINE_COMMON)
postlen += imglen - prelen;
imgoff += imglen;
preoff += prelen;
}
/*
* Ok, the preimage matches with whitespace fuzz.
*
* imgoff now holds the true length of the target that
* matches the preimage before the end of the file.
*
* Count the number of characters in the preimage that fall
* beyond the end of the file and make sure that all of them
* are whitespace characters. (This can only happen if
* we are removing blank lines at the end of the file.)
*/
buf = preimage_eof = preimage->buf + preoff;
for ( ; i < preimage->nr; i++)
preoff += preimage->line[i].len;
preimage_end = preimage->buf + preoff;
for ( ; buf < preimage_end; buf++)
if (!isspace(*buf))
return 0;
/*
* Update the preimage and the common postimage context
* lines to use the same whitespace as the target.
* If whitespace is missing in the target (i.e.
* if the preimage extends beyond the end of the file),
* use the whitespace from the preimage.
*/
extra_chars = preimage_end - preimage_eof;
strbuf_init(&fixed, imgoff + extra_chars);
strbuf_add(&fixed, img->buf + try, imgoff);
strbuf_add(&fixed, preimage_eof, extra_chars);
fixed_buf = strbuf_detach(&fixed, &fixed_len);
update_pre_post_images(preimage, postimage,
fixed_buf, fixed_len, postlen);
return 1;
}
if (ws_error_action != correct_ws_error)
return 0;
/*
* The hunk does not apply byte-by-byte, but the hash says
* it might with whitespace fuzz. We haven't been asked to
* ignore whitespace, we were asked to correct whitespace
* errors, so let's try matching after whitespace correction.
*
* The preimage may extend beyond the end of the file,
* but in this loop we will only handle the part of the
* preimage that falls within the file.
*/
strbuf_init(&fixed, preimage->len + 1);
orig = preimage->buf;
target = img->buf + try;
postlen = 0;
for (i = 0; i < preimage_limit; i++) {
size_t oldlen = preimage->line[i].len;
size_t tgtlen = img->line[try_lno + i].len;
size_t fixstart = fixed.len;
struct strbuf tgtfix;
int match;
/* Try fixing the line in the preimage */
ws_fix_copy(&fixed, orig, oldlen, ws_rule, NULL);
/* Try fixing the line in the target */
strbuf_init(&tgtfix, tgtlen);
ws_fix_copy(&tgtfix, target, tgtlen, ws_rule, NULL);
/*
* If they match, either the preimage was based on
* a version before our tree fixed whitespace breakage,
* or we are lacking a whitespace-fix patch the tree
* the preimage was based on already had (i.e. target
* has whitespace breakage, the preimage doesn't).
* In either case, we are fixing the whitespace breakages
* so we might as well take the fix together with their
* real change.
*/
match = (tgtfix.len == fixed.len - fixstart &&
!memcmp(tgtfix.buf, fixed.buf + fixstart,
fixed.len - fixstart));
postlen += tgtfix.len;
strbuf_release(&tgtfix);
if (!match)
goto unmatch_exit;
orig += oldlen;
target += tgtlen;
}
/*
* Now handle the lines in the preimage that falls beyond the
* end of the file (if any). They will only match if they are
* empty or only contain whitespace (if WS_BLANK_AT_EOL is
* false).
*/
for ( ; i < preimage->nr; i++) {
size_t fixstart = fixed.len; /* start of the fixed preimage */
size_t oldlen = preimage->line[i].len;
int j;
/* Try fixing the line in the preimage */
ws_fix_copy(&fixed, orig, oldlen, ws_rule, NULL);
for (j = fixstart; j < fixed.len; j++)
if (!isspace(fixed.buf[j]))
goto unmatch_exit;
orig += oldlen;
}
/*
* Yes, the preimage is based on an older version that still
* has whitespace breakages unfixed, and fixing them makes the
* hunk match. Update the context lines in the postimage.
*/
fixed_buf = strbuf_detach(&fixed, &fixed_len);
if (postlen < postimage->len)
postlen = 0;
update_pre_post_images(preimage, postimage,
fixed_buf, fixed_len, postlen);
return 1;
unmatch_exit:
strbuf_release(&fixed);
return 0;
}
static int find_pos(struct image *img,
struct image *preimage,
struct image *postimage,
int line,
unsigned ws_rule,
int match_beginning, int match_end)
{
int i;
unsigned long backwards, forwards, try;
int backwards_lno, forwards_lno, try_lno;
/*
* If match_beginning or match_end is specified, there is no
* point starting from a wrong line that will never match and
* wander around and wait for a match at the specified end.
*/
if (match_beginning)
line = 0;
else if (match_end)
line = img->nr - preimage->nr;
/*
* Because the comparison is unsigned, the following test
* will also take care of a negative line number that can
* result when match_end and preimage is larger than the target.
*/
if ((size_t) line > img->nr)
line = img->nr;
try = 0;
for (i = 0; i < line; i++)
try += img->line[i].len;
/*
* There's probably some smart way to do this, but I'll leave
* that to the smart and beautiful people. I'm simple and stupid.
*/
backwards = try;
backwards_lno = line;
forwards = try;
forwards_lno = line;
try_lno = line;
for (i = 0; ; i++) {
if (match_fragment(img, preimage, postimage,
try, try_lno, ws_rule,
match_beginning, match_end))
return try_lno;
again:
if (backwards_lno == 0 && forwards_lno == img->nr)
break;
if (i & 1) {
if (backwards_lno == 0) {
i++;
goto again;
}
backwards_lno--;
backwards -= img->line[backwards_lno].len;
try = backwards;
try_lno = backwards_lno;
} else {
if (forwards_lno == img->nr) {
i++;
goto again;
}
forwards += img->line[forwards_lno].len;
forwards_lno++;
try = forwards;
try_lno = forwards_lno;
}
}
return -1;
}
static void remove_first_line(struct image *img)
{
img->buf += img->line[0].len;
img->len -= img->line[0].len;
img->line++;
img->nr--;
}
static void remove_last_line(struct image *img)
{
img->len -= img->line[--img->nr].len;
}
/*
* The change from "preimage" and "postimage" has been found to
* apply at applied_pos (counts in line numbers) in "img".
* Update "img" to remove "preimage" and replace it with "postimage".
*/
static void update_image(struct image *img,
int applied_pos,
struct image *preimage,
struct image *postimage)
{
/*
* remove the copy of preimage at offset in img
* and replace it with postimage
*/
int i, nr;
size_t remove_count, insert_count, applied_at = 0;
char *result;
int preimage_limit;
/*
* If we are removing blank lines at the end of img,
* the preimage may extend beyond the end.
* If that is the case, we must be careful only to
* remove the part of the preimage that falls within
* the boundaries of img. Initialize preimage_limit
* to the number of lines in the preimage that falls
* within the boundaries.
*/
preimage_limit = preimage->nr;
if (preimage_limit > img->nr - applied_pos)
preimage_limit = img->nr - applied_pos;
for (i = 0; i < applied_pos; i++)
applied_at += img->line[i].len;
remove_count = 0;
for (i = 0; i < preimage_limit; i++)
remove_count += img->line[applied_pos + i].len;
insert_count = postimage->len;
/* Adjust the contents */
result = xmalloc(img->len + insert_count - remove_count + 1);
memcpy(result, img->buf, applied_at);
memcpy(result + applied_at, postimage->buf, postimage->len);
memcpy(result + applied_at + postimage->len,
img->buf + (applied_at + remove_count),
img->len - (applied_at + remove_count));
free(img->buf);
img->buf = result;
img->len += insert_count - remove_count;
result[img->len] = '\0';
/* Adjust the line table */
nr = img->nr + postimage->nr - preimage_limit;
if (preimage_limit < postimage->nr) {
/*
* NOTE: this knows that we never call remove_first_line()
* on anything other than pre/post image.
*/
REALLOC_ARRAY(img->line, nr);
img->line_allocated = img->line;
}
if (preimage_limit != postimage->nr)
memmove(img->line + applied_pos + postimage->nr,
img->line + applied_pos + preimage_limit,
(img->nr - (applied_pos + preimage_limit)) *
sizeof(*img->line));
memcpy(img->line + applied_pos,
postimage->line,
postimage->nr * sizeof(*img->line));
if (!allow_overlap)
for (i = 0; i < postimage->nr; i++)
img->line[applied_pos + i].flag |= LINE_PATCHED;
img->nr = nr;
}
/*
* Use the patch-hunk text in "frag" to prepare two images (preimage and
* postimage) for the hunk. Find lines that match "preimage" in "img" and
* replace the part of "img" with "postimage" text.
*/
static int apply_one_fragment(struct image *img, struct fragment *frag,
int inaccurate_eof, unsigned ws_rule,
int nth_fragment)
{
int match_beginning, match_end;
const char *patch = frag->patch;
int size = frag->size;
char *old, *oldlines;
struct strbuf newlines;
int new_blank_lines_at_end = 0;
int found_new_blank_lines_at_end = 0;
int hunk_linenr = frag->linenr;
unsigned long leading, trailing;
int pos, applied_pos;
struct image preimage;
struct image postimage;
memset(&preimage, 0, sizeof(preimage));
memset(&postimage, 0, sizeof(postimage));
oldlines = xmalloc(size);
strbuf_init(&newlines, size);
old = oldlines;
while (size > 0) {
char first;
int len = linelen(patch, size);
int plen;
int added_blank_line = 0