blob: 97f387540b17fe233310fa0d17407b6362c82d26 [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 "config.h"
#include "blob.h"
#include "delta.h"
#include "diff.h"
#include "dir.h"
#include "xdiff-interface.h"
#include "ll-merge.h"
#include "lockfile.h"
#include "parse-options.h"
#include "quote.h"
#include "rerere.h"
#include "apply.h"
static void git_apply_config(void)
{
git_config_get_string_const("apply.whitespace", &apply_default_whitespace);
git_config_get_string_const("apply.ignorewhitespace", &apply_default_ignorewhitespace);
git_config(git_default_config, NULL);
}
static int parse_whitespace_option(struct apply_state *state, const char *option)
{
if (!option) {
state->ws_error_action = warn_on_ws_error;
return 0;
}
if (!strcmp(option, "warn")) {
state->ws_error_action = warn_on_ws_error;
return 0;
}
if (!strcmp(option, "nowarn")) {
state->ws_error_action = nowarn_ws_error;
return 0;
}
if (!strcmp(option, "error")) {
state->ws_error_action = die_on_ws_error;
return 0;
}
if (!strcmp(option, "error-all")) {
state->ws_error_action = die_on_ws_error;
state->squelch_whitespace_errors = 0;
return 0;
}
if (!strcmp(option, "strip") || !strcmp(option, "fix")) {
state->ws_error_action = correct_ws_error;
return 0;
}
return error(_("unrecognized whitespace option '%s'"), option);
}
static int parse_ignorewhitespace_option(struct apply_state *state,
const char *option)
{
if (!option || !strcmp(option, "no") ||
!strcmp(option, "false") || !strcmp(option, "never") ||
!strcmp(option, "none")) {
state->ws_ignore_action = ignore_ws_none;
return 0;
}
if (!strcmp(option, "change")) {
state->ws_ignore_action = ignore_ws_change;
return 0;
}
return error(_("unrecognized whitespace ignore option '%s'"), option);
}
int init_apply_state(struct apply_state *state,
const char *prefix,
struct lock_file *lock_file)
{
memset(state, 0, sizeof(*state));
state->prefix = prefix;
state->lock_file = lock_file;
state->newfd = -1;
state->apply = 1;
state->line_termination = '\n';
state->p_value = 1;
state->p_context = UINT_MAX;
state->squelch_whitespace_errors = 5;
state->ws_error_action = warn_on_ws_error;
state->ws_ignore_action = ignore_ws_none;
state->linenr = 1;
string_list_init(&state->fn_table, 0);
string_list_init(&state->limit_by_name, 0);
string_list_init(&state->symlink_changes, 0);
strbuf_init(&state->root, 0);
git_apply_config();
if (apply_default_whitespace && parse_whitespace_option(state, apply_default_whitespace))
return -1;
if (apply_default_ignorewhitespace && parse_ignorewhitespace_option(state, apply_default_ignorewhitespace))
return -1;
return 0;
}
void clear_apply_state(struct apply_state *state)
{
string_list_clear(&state->limit_by_name, 0);
string_list_clear(&state->symlink_changes, 0);
strbuf_release(&state->root);
/* &state->fn_table is cleared at the end of apply_patch() */
}
static void mute_routine(const char *msg, va_list params)
{
/* do nothing */
}
int check_apply_state(struct apply_state *state, int force_apply)
{
int is_not_gitdir = !startup_info->have_repository;
if (state->apply_with_reject && state->threeway)
return error(_("--reject and --3way cannot be used together."));
if (state->cached && state->threeway)
return error(_("--cached and --3way cannot be used together."));
if (state->threeway) {
if (is_not_gitdir)
return error(_("--3way outside a repository"));
state->check_index = 1;
}
if (state->apply_with_reject) {
state->apply = 1;
if (state->apply_verbosity == verbosity_normal)
state->apply_verbosity = verbosity_verbose;
}
if (!force_apply && (state->diffstat || state->numstat || state->summary || state->check || state->fake_ancestor))
state->apply = 0;
if (state->check_index && is_not_gitdir)
return error(_("--index outside a repository"));
if (state->cached) {
if (is_not_gitdir)
return error(_("--cached outside a repository"));
state->check_index = 1;
}
if (state->check_index)
state->unsafe_paths = 0;
if (!state->lock_file)
return error("BUG: state->lock_file should not be NULL");
if (state->apply_verbosity <= verbosity_silent) {
state->saved_error_routine = get_error_routine();
state->saved_warn_routine = get_warn_routine();
set_error_routine(mute_routine);
set_warn_routine(mute_routine);
}
return 0;
}
static void set_default_whitespace_mode(struct apply_state *state)
{
if (!state->whitespace_option && !apply_default_whitespace)
state->ws_error_action = (state->apply ? warn_on_ws_error : nowarn_ws_error);
}
/*
* 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;
int extension_linenr; /* first line specifying delete/new/rename/copy */
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;
unsigned int crlf_in_old: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 */
struct object_id threeway_stage[3];
};
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;
};
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 *end1 = s1 + n1;
const char *end2 = s2 + n2;
/* ignore line endings */
while (s1 < end1 && (end1[-1] == '\r' || end1[-1] == '\n'))
end1--;
while (s2 < end2 && (end2[-1] == '\r' || end2[-1] == '\n'))
end2--;
while (s1 < end1 && s2 < end2) {
if (isspace(*s1)) {
/*
* Skip whitespace. We check on both buffers
* because we don't want "a b" to match "ab".
*/
if (!isspace(*s2))
return 0;
while (s1 < end1 && isspace(*s1))
s1++;
while (s2 < end2 && isspace(*s2))
s2++;
} else if (*s1++ != *s2++)
return 0;
}
/* If we reached the end on one side only, lines don't match. */
return s1 == end1 && s2 == end2;
}
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 int read_patch_file(struct strbuf *sb, int fd)
{
if (strbuf_read(sb, fd, 0) < 0)
return error_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);
return 0;
}
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(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(struct apply_state *state,
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 (state->root.len)
strbuf_insert(&name, 0, state->root.buf, state->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 *find_name_common(struct apply_state *state,
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(c, terminate))
break;
}
line++;
if (c == '/' && !--p_value)
start = line;
}
if (!start)
return squash_slash(xstrdup_or_null(def));
len = line - start;
if (!len)
return squash_slash(xstrdup_or_null(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 (state->root.len) {
char *ret = xstrfmt("%s%.*s", state->root.buf, len, start);
return squash_slash(ret);
}
return squash_slash(xmemdupz(start, len));
}
static char *find_name(struct apply_state *state,
const char *line,
char *def,
int p_value,
int terminate)
{
if (*line == '"') {
char *name = find_name_gnu(state, line, def, p_value);
if (name)
return name;
}
return find_name_common(state, line, def, p_value, NULL, terminate);
}
static char *find_name_traditional(struct apply_state *state,
const char *line,
char *def,
int p_value)
{
size_t len;
size_t date_len;
if (*line == '"') {
char *name = find_name_gnu(state, 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(state, line, def, p_value, NULL, TERM_TAB);
len -= date_len;
return find_name_common(state, line, def, p_value, line + len, 0);
}
/*
* Given the string after "--- " or "+++ ", guess the appropriate
* p_value for the given patch.
*/
static int guess_p_value(struct apply_state *state, const char *nameline)
{
char *name, *cp;
int val = -1;
if (is_dev_null(nameline))
return -1;
name = find_name_traditional(state, nameline, NULL, 0);
if (!name)
return -1;
cp = strchr(name, '/');
if (!cp)
val = 0;
else if (state->prefix) {
/*
* Does it begin with "a/$our-prefix" and such? Then this is
* very likely to apply to our directory.
*/
if (starts_with(name, state->prefix))
val = count_slashes(state->prefix);
else {
cp++;
if (starts_with(cp, state->prefix))
val = count_slashes(state->prefix) + 1;
}
}
free(name);
return val;
}
/*
* Does the ---/+++ line have 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[] =
"^[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, epoch_hour, hour, minute;
int status;
for (cp = nameline; *cp != '\n'; cp++) {
if (*cp == '\t')
timestamp = cp + 1;
}
if (!timestamp)
return 0;
/*
* YYYY-MM-DD hh:mm:ss must be from either 1969-12-31
* (west of GMT) or 1970-01-01 (east of GMT)
*/
if (skip_prefix(timestamp, "1969-12-31 ", &timestamp))
epoch_hour = 24;
else if (skip_prefix(timestamp, "1970-01-01 ", &timestamp))
epoch_hour = 0;
else
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;
}
hour = strtol(timestamp, NULL, 10);
minute = strtol(timestamp + m[1].rm_so, NULL, 10);
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;
return hour * 60 + minute - zoneoffset == epoch_hour * 60;
}
/*
* 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 int parse_traditional_patch(struct apply_state *state,
const char *first,
const char *second,
struct patch *patch)
{
char *name;
first += 4; /* skip "--- " */
second += 4; /* skip "+++ " */
if (!state->p_value_known) {
int p, q;
p = guess_p_value(state, first);
q = guess_p_value(state, second);
if (p < 0) p = q;
if (0 <= p && p == q) {
state->p_value = p;
state->p_value_known = 1;
}
}
if (is_dev_null(first)) {
patch->is_new = 1;
patch->is_delete = 0;
name = find_name_traditional(state, second, NULL, state->p_value);
patch->new_name = name;
} else if (is_dev_null(second)) {
patch->is_new = 0;
patch->is_delete = 1;
name = find_name_traditional(state, first, NULL, state->p_value);
patch->old_name = name;
} else {
char *first_name;
first_name = find_name_traditional(state, first, NULL, state->p_value);
name = find_name_traditional(state, second, first_name, state->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 = xstrdup_or_null(name);
}
}
if (!name)
return error(_("unable to find filename in patch at line %d"), state->linenr);
return 0;
}
static int gitdiff_hdrend(struct apply_state *state,
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 int gitdiff_verify_name(struct apply_state *state,
const char *line,
int isnull,
char **name,
int side)
{
if (!*name && !isnull) {
*name = find_name(state, line, NULL, state->p_value, TERM_TAB);
return 0;
}
if (*name) {
char *another;
if (isnull)
return error(_("git apply: bad git-diff - expected /dev/null, got %s on line %d"),
*name, state->linenr);
another = find_name(state, line, NULL, state->p_value, TERM_TAB);
if (!another || strcmp(another, *name)) {
free(another);
return error((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"), state->linenr);
}
free(another);
} else {
if (!starts_with(line, "/dev/null\n"))
return error(_("git apply: bad git-diff - expected /dev/null on line %d"), state->linenr);
}
return 0;
}
static int gitdiff_oldname(struct apply_state *state,
const char *line,
struct patch *patch)
{
return gitdiff_verify_name(state, line,
patch->is_new, &patch->old_name,
DIFF_OLD_NAME);
}
static int gitdiff_newname(struct apply_state *state,
const char *line,
struct patch *patch)
{
return gitdiff_verify_name(state, line,
patch->is_delete, &patch->new_name,
DIFF_NEW_NAME);
}
static int parse_mode_line(const char *line, int linenr, unsigned int *mode)
{
char *end;
*mode = strtoul(line, &end, 8);
if (end == line || !isspace(*end))
return error(_("invalid mode on line %d: %s"), linenr, line);
return 0;
}
static int gitdiff_oldmode(struct apply_state *state,
const char *line,
struct patch *patch)
{
return parse_mode_line(line, state->linenr, &patch->old_mode);
}
static int gitdiff_newmode(struct apply_state *state,
const char *line,
struct patch *patch)
{
return parse_mode_line(line, state->linenr, &patch->new_mode);
}
static int gitdiff_delete(struct apply_state *state,
const char *line,
struct patch *patch)
{
patch->is_delete = 1;
free(patch->old_name);
patch->old_name = xstrdup_or_null(patch->def_name);
return gitdiff_oldmode(state, line, patch);
}
static int gitdiff_newfile(struct apply_state *state,
const char *line,
struct patch *patch)
{
patch->is_new = 1;
free(patch->new_name);
patch->new_name = xstrdup_or_null(patch->def_name);
return gitdiff_newmode(state, line, patch);
}
static int gitdiff_copysrc(struct apply_state *state,
const char *line,
struct patch *patch)
{
patch->is_copy = 1;
free(patch->old_name);
patch->old_name = find_name(state, line, NULL, state->p_value ? state->p_value - 1 : 0, 0);
return 0;
}
static int gitdiff_copydst(struct apply_state *state,
const char *line,
struct patch *patch)
{
patch->is_copy = 1;
free(patch->new_name);
patch->new_name = find_name(state, line, NULL, state->p_value ? state->p_value - 1 : 0, 0);
return 0;
}
static int gitdiff_renamesrc(struct apply_state *state,
const char *line,
struct patch *patch)
{
patch->is_rename = 1;
free(patch->old_name);
patch->old_name = find_name(state, line, NULL, state->p_value ? state->p_value - 1 : 0, 0);
return 0;
}
static int gitdiff_renamedst(struct apply_state *state,
const char *line,
struct patch *patch)
{
patch->is_rename = 1;
free(patch->new_name);
patch->new_name = find_name(state, line, NULL, state->p_value ? state->p_value - 1 : 0, 0);
return 0;
}
static int gitdiff_similarity(struct apply_state *state,
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(struct apply_state *state,
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(struct apply_state *state,
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 == ' ')
return gitdiff_oldmode(state, ptr + 1, patch);
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(struct apply_state *state,
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(struct apply_state *state,
const char *line,
int llen)
{
int nslash;
int i;
if (!state->p_value)
return (llen && line[0] == '/') ? NULL : line;
nslash = state->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(struct apply_state *state,
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(state, 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(state, 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(state, 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(state, 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(state, 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(state, 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);
}
}
}
static int check_header_line(struct apply_state *state, struct patch *patch)
{
int extensions = (patch->is_delete == 1) + (patch->is_new == 1) +
(patch->is_rename == 1) + (patch->is_copy == 1);
if (extensions > 1)
return error(_("inconsistent header lines %d and %d"),
patch->extension_linenr, state->linenr);
if (extensions && !patch->extension_linenr)
patch->extension_linenr = state->linenr;
return 0;
}
/* Verify that we recognize the lines following a git header */
static int parse_git_header(struct apply_state *state,
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(state, line, len);
if (patch->def_name && state->root.len) {
char *s = xstrfmt("%s%s", state->root.buf, patch->def_name);
free(patch->def_name);
patch->def_name = s;
}
line += len;
size -= len;
state->linenr++;
for (offset = len ; size > 0 ; offset += len, size -= len, line += len, state->linenr++) {
static const struct opentry {
const char *str;
int (*fn)(struct apply_state *, 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);
int res;
if (len < oplen || memcmp(p->str, line, oplen))
continue;
res = p->fn(state, line + oplen, patch);
if (res < 0)
return -1;
if (check_header_line(state, patch))
return -1;
if (res > 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;
}
/*
* Find file diff header
*
* Returns:
* -1 if no header was found
* -128 in case of error
* the size of the header in bytes (called "offset") otherwise
*/
static int find_header(struct apply_state *state,
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, state->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;
error(_("patch fragment without header at line %d: %.*s"),
state->linenr, (int)len-1, line);
return -128;
}
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(state, line, len, size, patch);
if (git_hdr_len < 0)
return -128;
if (git_hdr_len <= len)
continue;
if (!patch->old_name && !patch->new_name) {
if (!patch->def_name) {
error(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)",
state->p_value),
state->p_value, state->linenr);
return -128;
}
patch->old_name = xstrdup(patch->def_name);
patch->new_name = xstrdup(patch->def_name);
}
if ((!patch->new_name && !patch->is_delete) ||
(!patch->old_name && !patch->is_new)) {
error(_("git diff header lacks filename information "
"(line %d)"), state->linenr);
return -128;
}
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 */
if (parse_traditional_patch(state, line, line+len, patch))
return -128;
*hdrsize = len + nextlen;
state->linenr += 2;
return offset;
}
return -1;
}
static void record_ws_error(struct apply_state *state,
unsigned result,
const char *line,
int len,
int linenr)
{
char *err;
if (!result)
return;
state->whitespace_error++;
if (state->squelch_whitespace_errors &&
state->squelch_whitespace_errors < state->whitespace_error)
return;
err = whitespace_error_string(result);
if (state->apply_verbosity > verbosity_silent)
fprintf(stderr, "%s:%d: %s.\n%.*s\n",
state->patch_input_file, linenr, err, len, line);
free(err);
}
static void check_whitespace(struct apply_state *state,
const char *line,
int len,
unsigned ws_rule)
{
unsigned result = ws_check(line + 1, len - 1, ws_rule);
record_ws_error(state, result, line + 1, len - 2, state->linenr);
}
/*
* Check if the patch has context lines with CRLF or
* the patch wants to remove lines with CRLF.
*/
static void check_old_for_crlf(struct patch *patch, const char *line, int len)
{
if (len >= 2 && line[len-1] == '\n' && line[len-2] == '\r') {
patch->ws_rule |= WS_CR_AT_EOL;
patch->crlf_in_old = 1;
}
}
/*
* 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(struct apply_state *state,
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;
state->linenr++;
added = deleted = 0;
for (offset = len;
0 < size;
offset += len, size -= len, line += len, state->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++;
check_old_for_crlf(patch, line, len);
if (!state->apply_in_reverse &&
state->ws_error_action == correct_ws_error)
check_whitespace(state, line, len, patch->ws_rule);
break;
case '-':
if (!state->apply_in_reverse)
check_old_for_crlf(patch, line, len);
if (state->apply_in_reverse &&
state->ws_error_action != nowarn_ws_error)
check_whitespace(state, line, len, patch->ws_rule);
deleted++;
oldlines--;
trailing = 0;
break;
case '+':
if (state->apply_in_reverse)
check_old_for_crlf(patch, line, len);
if (!state->apply_in_reverse &&
state->ws_error_action != nowarn_ws_error)
check_whitespace(state, 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;
if (!deleted && !added)
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.
*
* Returns:
* -1 in case of error,
* the number of bytes in the patch otherwise.
*/
static int parse_single_patch(struct apply_state *state,
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 = state->linenr;
len = parse_fragment(state, line, size, patch, fragment);
if (len <= 0) {
free(fragment);
return error(_("corrupt patch at line %d"), state->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)
return error(_("new file %s depends on old contents"), patch->new_name);
if (0 < patch->is_delete && newlines)
return error(_("deleted file %s still has contents"), patch->old_name);
if (!patch->is_delete && !newlines && context && state->apply_verbosity > verbosity_silent)
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(struct apply_state *state,
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;
state->linenr++;
buffer += llen;
while (1) {
int byte_length, max_byte_length, newsize;
llen = linelen(buffer, size);
used += llen;
state->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"),
state->linenr-1, llen-1, buffer);
return NULL;
}
/*
* Returns:
* -1 in case of error,
* the length of the parsed binary patch otherwise
*/
static int parse_binary(struct apply_state *state,
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(state, &buffer, &size, &status, &used);
if (!forward && !status)
/* there has to be one hunk (forward hunk) */
return error(_("unrecognized binary patch at line %d"), state->linenr-1);
if (status)
/* otherwise we already gave an error message */
return status;
reverse = parse_binary_hunk(state, &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(struct apply_state *state, char **name)
{
char *old_name = *name;
if (!old_name)
return;
*name = prefix_filename(state->prefix, *name);
free(old_name);
}
static void prefix_patch(struct apply_state *state, struct patch *p)
{
if (!state->prefix || p->is_toplevel_relative)
return;
prefix_one(state, &p->new_name);
prefix_one(state, &p->old_name);
}
/*
* include/exclude
*/
static void add_name_limit(struct apply_state *state,
const char *name,
int exclude)
{
struct string_list_item *it;
it = string_list_append(&state->limit_by_name, name);
it->util = exclude ? NULL : (void *) 1;
}
static int use_patch(struct apply_state *state, 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 (state->prefix && *state->prefix) {
const char *rest;
if (!skip_prefix(pathname, state->prefix, &rest) || !*rest)
return 0;
}
/* See if it matches any of exclude/include rule */
for (i = 0; i < state->limit_by_name.nr; i++) {
struct string_list_item *it = &state->limit_by_name.items[i];
if (!wildmatch(it->string, pathname, 0))
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 !state->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.
*
* Returns:
* -1 if no header was found or parse_binary() failed,
* -128 on another error,
* the number of bytes consumed otherwise,
* so that the caller can call us again for the next patch.
*/
static int parse_chunk(struct apply_state *state, char *buffer, unsigned long size, struct patch *patch)
{
int hdrsize, patchsize;
int offset = find_header(state, buffer, size, &hdrsize, patch);
if (offset < 0)
return offset;
prefix_patch(state, patch);
if (!use_patch(state, patch))
patch->ws_rule = 0;
else
patch->ws_rule = whitespace_rule(patch->new_name
? patch->new_name
: patch->old_name);
patchsize = parse_single_patch(state,
buffer + offset + hdrsize,
size - offset - hdrsize,
patch);
if (patchsize < 0)
return -128;
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;
state->linenr++;
used = parse_binary(state, buffer + hd + llen,
size - hd - llen, patch);
if (used < 0)
return -1;
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)) {
state->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 ((state->apply || state->check) &&
(!patch->is_binary && !metadata_changes(patch))) {
error(_("patch with only garbage at line %d"), state->linenr);
return -128;
}
}
return offset + hdrsize + patchsize;
}
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 apply_state *state, 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 = state->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 + state->max_change > 70 ? 70 - max : state->max_change;
add = patch->lines_added;
del = patch->lines_deleted;
if (state->max_change > 0) {
int total = ((add + del) * max + state->max_change / 2) / state->max_change;
add = (add * max + state->max_change / 2) / state->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, struct patch *patch,
const char *path, struct strbuf *buf)
{
enum safe_crlf safe_crlf = patch->crlf_in_old ?
SAFE_CRLF_KEEP_CRLF : SAFE_CRLF_RENORMALIZE;
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);
/*
* "git apply" without "--index/--cached" should never look
* at the index; the target file may not have been added to
* the index yet, and we may not even be in any Git repository.
* Pass NULL to convert_to_git() to stress this; the function
* should never look at the index when explicit crlf option
* is given.
*/
convert_to_git(NULL, path, buf->buf, buf->len, buf, safe_crlf);
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 l_len = postimage->line[i].len;
if (!(postimage->line[i].flag & LINE_COMMON)) {
/* an added line -- no counterparts in preimage */
memmove(new, old, l_len);
old += l_len;
new += l_len;
continue;
}
/* a common context -- skip it in the original postimage */
old += l_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 */
l_len = preimage->line[ctx].len;
memcpy(new, fixed, l_len);
new += l_len;
fixed += l_len;
postimage->line[i].len = l_len;
ctx++;
}
if (postlen
? postlen < new - postimage->buf
: postimage->len < new - postimage->buf)
die("BUG: caller miscounted postlen: asked %d, orig = %d, used = %d",
(int)postlen, (int) postimage->len, (int)(new - postimage->buf));
/* Fix the length of the whole thing */
postimage->len = new - postimage->buf;
postimage->nr -= reduced;
}
static int line_by_line_fuzzy_match(struct image *img,
struct image *preimage,
struct image *postimage,
unsigned long try,
int try_lno,
int preimage_limit)
{
int i;
size_t imgoff = 0;
size_t preoff = 0;
size_t postlen = postimage->len;
size_t extra_chars;
char *buf;
char *preimage_eof;
char *preimage_end;
struct strbuf fixed;
char *fixed_buf;
size_t fixed_len;
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;
}
static int match_fragment(struct apply_state *state,
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 (state->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 (state->ws_ignore_action == ignore_ws_change)
return line_by_line_fuzzy_match(img, preimage, postimage,
try, try_lno, preimage_limit);
if (state->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 weren't asked to
* ignore whitespace, we were asked to correct whitespace
* errors, so let's try matching after whitespace correction.
*
* While checking the preimage against the target, whitespace
* errors in both fixed, we count how large the corresponding
* postimage needs to be. The postimage prepared by
* apply_one_fragment() has whitespace errors fixed on added
* lines already, but the common lines were propagated as-is,
* which may become longer when their whitespace errors are
* fixed.
*/
/* First count added lines in postimage */
postlen = 0;
for (i = 0; i < postimage->nr; i++) {
if (!(postimage->line[i].flag & LINE_COMMON))
postlen += postimage->line[i].len;
}
/*
* 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;
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));
/* Add the length if this is common with the postimage */
if (preimage->line[i].flag & LINE_COMMON)
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;