git/builtin-grep.c

868 строки
22 KiB
C
Исходник Обычный вид История

/*
* Builtin "git grep"
*
* Copyright (c) 2006 Junio C Hamano
*/
#include "cache.h"
#include "blob.h"
#include "tree.h"
#include "commit.h"
#include "tag.h"
#include "tree-walk.h"
#include "builtin.h"
#include "parse-options.h"
#include "userdiff.h"
#include "grep.h"
#ifndef NO_EXTERNAL_GREP
#ifdef __unix__
#define NO_EXTERNAL_GREP 0
#else
#define NO_EXTERNAL_GREP 1
#endif
#endif
static char const * const grep_usage[] = {
"git grep [options] [-e] <pattern> [<rev>...] [[--] path...]",
NULL
};
static int grep_config(const char *var, const char *value, void *cb)
{
struct grep_opt *opt = cb;
switch (userdiff_config(var, value)) {
case 0: break;
case -1: return -1;
default: return 0;
}
if (!strcmp(var, "color.grep")) {
opt->color = git_config_colorbool(var, value, -1);
return 0;
}
if (!strcmp(var, "color.grep.external"))
return git_config_string(&(opt->color_external), var, value);
if (!strcmp(var, "color.grep.match")) {
if (!value)
return config_error_nonbool(var);
color_parse(value, var, opt->color_match);
return 0;
}
return git_color_default_config(var, value, cb);
}
/*
* git grep pathspecs are somewhat different from diff-tree pathspecs;
* pathname wildcards are allowed.
*/
static int pathspec_matches(const char **paths, const char *name)
{
int namelen, i;
if (!paths || !*paths)
return 1;
namelen = strlen(name);
for (i = 0; paths[i]; i++) {
const char *match = paths[i];
int matchlen = strlen(match);
const char *cp, *meta;
if (!matchlen ||
((matchlen <= namelen) &&
!strncmp(name, match, matchlen) &&
(match[matchlen-1] == '/' ||
name[matchlen] == '\0' || name[matchlen] == '/')))
return 1;
if (!fnmatch(match, name, 0))
return 1;
if (name[namelen-1] != '/')
continue;
/* We are being asked if the directory ("name") is worth
* descending into.
*
* Find the longest leading directory name that does
* not have metacharacter in the pathspec; the name
* we are looking at must overlap with that directory.
*/
for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
char ch = *cp;
if (ch == '*' || ch == '[' || ch == '?') {
meta = cp;
break;
}
}
if (!meta)
meta = cp; /* fully literal */
if (namelen <= meta - match) {
/* Looking at "Documentation/" and
* the pattern says "Documentation/howto/", or
* "Documentation/diff*.txt". The name we
* have should match prefix.
*/
if (!memcmp(match, name, namelen))
return 1;
continue;
}
if (meta - match < namelen) {
/* Looking at "Documentation/howto/" and
* the pattern says "Documentation/h*";
* match up to "Do.../h"; this avoids descending
* into "Documentation/technical/".
*/
if (!memcmp(match, name, meta - match))
return 1;
continue;
}
}
return 0;
}
static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
{
unsigned long size;
char *data;
enum object_type type;
char *to_free = NULL;
int hit;
data = read_sha1_file(sha1, &type, &size);
if (!data) {
error("'%s': unable to read %s", name, sha1_to_hex(sha1));
return 0;
}
if (opt->relative && opt->prefix_length) {
static char name_buf[PATH_MAX];
char *cp;
int name_len = strlen(name) - opt->prefix_length + 1;
if (!tree_name_len)
name += opt->prefix_length;
else {
if (ARRAY_SIZE(name_buf) <= name_len)
cp = to_free = xmalloc(name_len);
else
cp = name_buf;
memcpy(cp, name, tree_name_len);
strcpy(cp + tree_name_len,
name + tree_name_len + opt->prefix_length);
name = cp;
}
}
hit = grep_buffer(opt, name, data, size);
free(data);
free(to_free);
return hit;
}
static int grep_file(struct grep_opt *opt, const char *filename)
{
struct stat st;
int i;
char *data;
size_t sz;
if (lstat(filename, &st) < 0) {
err_ret:
if (errno != ENOENT)
error("'%s': %s", filename, strerror(errno));
return 0;
}
if (!st.st_size)
return 0; /* empty file -- no grep hit */
if (!S_ISREG(st.st_mode))
return 0;
sz = xsize_t(st.st_size);
i = open(filename, O_RDONLY);
if (i < 0)
goto err_ret;
data = xmalloc(sz + 1);
if (st.st_size != read_in_full(i, data, sz)) {
error("'%s': short read %s", filename, strerror(errno));
close(i);
free(data);
return 0;
}
close(i);
if (opt->relative && opt->prefix_length)
filename += opt->prefix_length;
i = grep_buffer(opt, filename, data, sz);
free(data);
return i;
}
#if !NO_EXTERNAL_GREP
static int exec_grep(int argc, const char **argv)
{
pid_t pid;
int status;
argv[argc] = NULL;
pid = fork();
if (pid < 0)
return pid;
if (!pid) {
execvp("grep", (char **) argv);
exit(255);
}
while (waitpid(pid, &status, 0) < 0) {
if (errno == EINTR)
continue;
return -1;
}
if (WIFEXITED(status)) {
if (!WEXITSTATUS(status))
return 1;
return 0;
}
return -1;
}
#define MAXARGS 1000
#define ARGBUF 4096
#define push_arg(a) do { \
if (nr < MAXARGS) argv[nr++] = (a); \
else die("maximum number of args exceeded"); \
} while (0)
/*
* If you send a singleton filename to grep, it does not give
* the name of the file. GNU grep has "-H" but we would want
* that behaviour in a portable way.
*
* So we keep two pathnames in argv buffer unsent to grep in
* the main loop if we need to do more than one grep.
*/
static int flush_grep(struct grep_opt *opt,
int argc, int arg0, const char **argv, int *kept)
{
int status;
int count = argc - arg0;
const char *kept_0 = NULL;
if (count <= 2) {
/*
* Because we keep at least 2 paths in the call from
* the main loop (i.e. kept != NULL), and MAXARGS is
* far greater than 2, this usually is a call to
* conclude the grep. However, the user could attempt
* to overflow the argv buffer by giving too many
* options to leave very small number of real
* arguments even for the call in the main loop.
*/
if (kept)
die("insanely many options to grep");
/*
* If we have two or more paths, we do not have to do
* anything special, but we need to push /dev/null to
* get "-H" behaviour of GNU grep portably but when we
* are not doing "-l" nor "-L" nor "-c".
*/
if (count == 1 &&
!opt->name_only &&
!opt->unmatch_name_only &&
!opt->count) {
argv[argc++] = "/dev/null";
argv[argc] = NULL;
}
}
else if (kept) {
/*
* Called because we found many paths and haven't finished
* iterating over the cache yet. We keep two paths
* for the concluding call. argv[argc-2] and argv[argc-1]
* has the last two paths, so save the first one away,
* replace it with NULL while sending the list to grep,
* and recover them after we are done.
*/
*kept = 2;
kept_0 = argv[argc-2];
argv[argc-2] = NULL;
argc -= 2;
}
if (opt->pre_context || opt->post_context) {
/*
* grep handles hunk marks between files, but we need to
* do that ourselves between multiple calls.
*/
if (opt->show_hunk_mark)
write_or_die(1, "--\n", 3);
else
opt->show_hunk_mark = 1;
}
status = exec_grep(argc, argv);
if (kept_0) {
/*
* Then recover them. Now the last arg is beyond the
* terminating NULL which is at argc, and the second
* from the last is what we saved away in kept_0
*/
argv[arg0++] = kept_0;
argv[arg0] = argv[argc+1];
}
return status;
}
static void grep_add_color(struct strbuf *sb, const char *escape_seq)
{
size_t orig_len = sb->len;
while (*escape_seq) {
if (*escape_seq == 'm')
strbuf_addch(sb, ';');
else if (*escape_seq != '\033' && *escape_seq != '[')
strbuf_addch(sb, *escape_seq);
escape_seq++;
}
if (sb->len > orig_len && sb->buf[sb->len - 1] == ';')
strbuf_setlen(sb, sb->len - 1);
}
static int external_grep(struct grep_opt *opt, const char **paths, int cached)
{
int i, nr, argc, hit, len, status;
const char *argv[MAXARGS+1];
char randarg[ARGBUF];
char *argptr = randarg;
struct grep_pat *p;
if (opt->extended || (opt->relative && opt->prefix_length))
return -1;
len = nr = 0;
push_arg("grep");
if (opt->fixed)
push_arg("-F");
if (opt->linenum)
push_arg("-n");
if (!opt->pathname)
push_arg("-h");
if (opt->regflags & REG_EXTENDED)
push_arg("-E");
if (opt->regflags & REG_ICASE)
push_arg("-i");
if (opt->binary == GREP_BINARY_NOMATCH)
push_arg("-I");
if (opt->word_regexp)
push_arg("-w");
if (opt->name_only)
push_arg("-l");
if (opt->unmatch_name_only)
push_arg("-L");
if (opt->null_following_name)
/* in GNU grep git's "-z" translates to "-Z" */
push_arg("-Z");
if (opt->count)
push_arg("-c");
if (opt->post_context || opt->pre_context) {
if (opt->post_context != opt->pre_context) {
if (opt->pre_context) {
push_arg("-B");
len += snprintf(argptr, sizeof(randarg)-len,
"%u", opt->pre_context) + 1;
if (sizeof(randarg) <= len)
die("maximum length of args exceeded");
push_arg(argptr);
argptr += len;
}
if (opt->post_context) {
push_arg("-A");
len += snprintf(argptr, sizeof(randarg)-len,
"%u", opt->post_context) + 1;
if (sizeof(randarg) <= len)
die("maximum length of args exceeded");
push_arg(argptr);
argptr += len;
}
}
else {
push_arg("-C");
len += snprintf(argptr, sizeof(randarg)-len,
"%u", opt->post_context) + 1;
if (sizeof(randarg) <= len)
die("maximum length of args exceeded");
push_arg(argptr);
argptr += len;
}
}
for (p = opt->pattern_list; p; p = p->next) {
push_arg("-e");
push_arg(p->pattern);
}
if (opt->color) {
struct strbuf sb = STRBUF_INIT;
grep_add_color(&sb, opt->color_match);
setenv("GREP_COLOR", sb.buf, 1);
strbuf_reset(&sb);
strbuf_addstr(&sb, "mt=");
grep_add_color(&sb, opt->color_match);
strbuf_addstr(&sb, ":sl=:cx=:fn=:ln=:bn=:se=");
setenv("GREP_COLORS", sb.buf, 1);
strbuf_release(&sb);
if (opt->color_external && strlen(opt->color_external) > 0)
push_arg(opt->color_external);
}
hit = 0;
argc = nr;
for (i = 0; i < active_nr; i++) {
struct cache_entry *ce = active_cache[i];
char *name;
int kept;
if (!S_ISREG(ce->ce_mode))
continue;
if (!pathspec_matches(paths, ce->name))
continue;
name = ce->name;
if (name[0] == '-') {
int len = ce_namelen(ce);
name = xmalloc(len + 3);
memcpy(name, "./", 2);
memcpy(name + 2, ce->name, len + 1);
}
argv[argc++] = name;
if (MAXARGS <= argc) {
status = flush_grep(opt, argc, nr, argv, &kept);
if (0 < status)
hit = 1;
argc = nr + kept;
}
if (ce_stage(ce)) {
do {
i++;
} while (i < active_nr &&
!strcmp(ce->name, active_cache[i]->name));
i--; /* compensate for loop control */
}
}
if (argc > nr) {
status = flush_grep(opt, argc, nr, argv, NULL);
if (0 < status)
hit = 1;
}
return hit;
}
#endif
static int grep_cache(struct grep_opt *opt, const char **paths, int cached,
int external_grep_allowed)
{
int hit = 0;
int nr;
read_cache();
#if !NO_EXTERNAL_GREP
/*
* Use the external "grep" command for the case where
* we grep through the checked-out files. It tends to
* be a lot more optimized
*/
if (!cached && external_grep_allowed) {
hit = external_grep(opt, paths, cached);
if (hit >= 0)
return hit;
}
#endif
for (nr = 0; nr < active_nr; nr++) {
struct cache_entry *ce = active_cache[nr];
if (!S_ISREG(ce->ce_mode))
continue;
if (!pathspec_matches(paths, ce->name))
continue;
/*
* If CE_VALID is on, we assume worktree file and its cache entry
* are identical, even if worktree file has been modified, so use
* cache version instead
*/
if (cached || (ce->ce_flags & CE_VALID)) {
if (ce_stage(ce))
continue;
hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
}
else
hit |= grep_file(opt, ce->name);
if (ce_stage(ce)) {
do {
nr++;
} while (nr < active_nr &&
!strcmp(ce->name, active_cache[nr]->name));
nr--; /* compensate for loop control */
}
}
free_grep_patterns(opt);
return hit;
}
static int grep_tree(struct grep_opt *opt, const char **paths,
struct tree_desc *tree,
const char *tree_name, const char *base)
{
int len;
int hit = 0;
tree_entry(): new tree-walking helper function This adds a "tree_entry()" function that combines the common operation of doing a "tree_entry_extract()" + "update_tree_entry()". It also has a simplified calling convention, designed for simple loops that traverse over a whole tree: the arguments are pointers to the tree descriptor and a name_entry structure to fill in, and it returns a boolean "true" if there was an entry left to be gotten in the tree. This allows tree traversal with struct tree_desc desc; struct name_entry entry; desc.buf = tree->buffer; desc.size = tree->size; while (tree_entry(&desc, &entry) { ... use "entry.{path, sha1, mode, pathlen}" ... } which is not only shorter than writing it out in full, it's hopefully less error prone too. [ It's actually a tad faster too - we don't need to recalculate the entry pathlength in both extract and update, but need to do it only once. Also, some callers can avoid doing a "strlen()" on the result, since it's returned as part of the name_entry structure. However, by now we're talking just 1% speedup on "git-rev-list --objects --all", and we're definitely at the point where tree walking is no longer the issue any more. ] NOTE! Not everybody wants to use this new helper function, since some of the tree walkers very much on purpose do the descriptor update separately from the entry extraction. So the "extract + update" sequence still remains as the core sequence, this is just a simplified interface. We should probably add a silly two-line inline helper function for initializing the descriptor from the "struct tree" too, just to cut down on the noise from that common "desc" initializer. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-05-30 20:45:45 +04:00
struct name_entry entry;
char *down;
int tn_len = strlen(tree_name);
struct strbuf pathbuf;
strbuf_init(&pathbuf, PATH_MAX + tn_len);
if (tn_len) {
strbuf_add(&pathbuf, tree_name, tn_len);
strbuf_addch(&pathbuf, ':');
tn_len = pathbuf.len;
}
strbuf_addstr(&pathbuf, base);
len = pathbuf.len;
tree_entry(): new tree-walking helper function This adds a "tree_entry()" function that combines the common operation of doing a "tree_entry_extract()" + "update_tree_entry()". It also has a simplified calling convention, designed for simple loops that traverse over a whole tree: the arguments are pointers to the tree descriptor and a name_entry structure to fill in, and it returns a boolean "true" if there was an entry left to be gotten in the tree. This allows tree traversal with struct tree_desc desc; struct name_entry entry; desc.buf = tree->buffer; desc.size = tree->size; while (tree_entry(&desc, &entry) { ... use "entry.{path, sha1, mode, pathlen}" ... } which is not only shorter than writing it out in full, it's hopefully less error prone too. [ It's actually a tad faster too - we don't need to recalculate the entry pathlength in both extract and update, but need to do it only once. Also, some callers can avoid doing a "strlen()" on the result, since it's returned as part of the name_entry structure. However, by now we're talking just 1% speedup on "git-rev-list --objects --all", and we're definitely at the point where tree walking is no longer the issue any more. ] NOTE! Not everybody wants to use this new helper function, since some of the tree walkers very much on purpose do the descriptor update separately from the entry extraction. So the "extract + update" sequence still remains as the core sequence, this is just a simplified interface. We should probably add a silly two-line inline helper function for initializing the descriptor from the "struct tree" too, just to cut down on the noise from that common "desc" initializer. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-05-30 20:45:45 +04:00
while (tree_entry(tree, &entry)) {
int te_len = tree_entry_len(entry.path, entry.sha1);
pathbuf.len = len;
strbuf_add(&pathbuf, entry.path, te_len);
tree_entry(): new tree-walking helper function This adds a "tree_entry()" function that combines the common operation of doing a "tree_entry_extract()" + "update_tree_entry()". It also has a simplified calling convention, designed for simple loops that traverse over a whole tree: the arguments are pointers to the tree descriptor and a name_entry structure to fill in, and it returns a boolean "true" if there was an entry left to be gotten in the tree. This allows tree traversal with struct tree_desc desc; struct name_entry entry; desc.buf = tree->buffer; desc.size = tree->size; while (tree_entry(&desc, &entry) { ... use "entry.{path, sha1, mode, pathlen}" ... } which is not only shorter than writing it out in full, it's hopefully less error prone too. [ It's actually a tad faster too - we don't need to recalculate the entry pathlength in both extract and update, but need to do it only once. Also, some callers can avoid doing a "strlen()" on the result, since it's returned as part of the name_entry structure. However, by now we're talking just 1% speedup on "git-rev-list --objects --all", and we're definitely at the point where tree walking is no longer the issue any more. ] NOTE! Not everybody wants to use this new helper function, since some of the tree walkers very much on purpose do the descriptor update separately from the entry extraction. So the "extract + update" sequence still remains as the core sequence, this is just a simplified interface. We should probably add a silly two-line inline helper function for initializing the descriptor from the "struct tree" too, just to cut down on the noise from that common "desc" initializer. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-05-30 20:45:45 +04:00
if (S_ISDIR(entry.mode))
/* Match "abc/" against pathspec to
* decide if we want to descend into "abc"
* directory.
*/
strbuf_addch(&pathbuf, '/');
down = pathbuf.buf + tn_len;
if (!pathspec_matches(paths, down))
;
tree_entry(): new tree-walking helper function This adds a "tree_entry()" function that combines the common operation of doing a "tree_entry_extract()" + "update_tree_entry()". It also has a simplified calling convention, designed for simple loops that traverse over a whole tree: the arguments are pointers to the tree descriptor and a name_entry structure to fill in, and it returns a boolean "true" if there was an entry left to be gotten in the tree. This allows tree traversal with struct tree_desc desc; struct name_entry entry; desc.buf = tree->buffer; desc.size = tree->size; while (tree_entry(&desc, &entry) { ... use "entry.{path, sha1, mode, pathlen}" ... } which is not only shorter than writing it out in full, it's hopefully less error prone too. [ It's actually a tad faster too - we don't need to recalculate the entry pathlength in both extract and update, but need to do it only once. Also, some callers can avoid doing a "strlen()" on the result, since it's returned as part of the name_entry structure. However, by now we're talking just 1% speedup on "git-rev-list --objects --all", and we're definitely at the point where tree walking is no longer the issue any more. ] NOTE! Not everybody wants to use this new helper function, since some of the tree walkers very much on purpose do the descriptor update separately from the entry extraction. So the "extract + update" sequence still remains as the core sequence, this is just a simplified interface. We should probably add a silly two-line inline helper function for initializing the descriptor from the "struct tree" too, just to cut down on the noise from that common "desc" initializer. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-05-30 20:45:45 +04:00
else if (S_ISREG(entry.mode))
hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len);
tree_entry(): new tree-walking helper function This adds a "tree_entry()" function that combines the common operation of doing a "tree_entry_extract()" + "update_tree_entry()". It also has a simplified calling convention, designed for simple loops that traverse over a whole tree: the arguments are pointers to the tree descriptor and a name_entry structure to fill in, and it returns a boolean "true" if there was an entry left to be gotten in the tree. This allows tree traversal with struct tree_desc desc; struct name_entry entry; desc.buf = tree->buffer; desc.size = tree->size; while (tree_entry(&desc, &entry) { ... use "entry.{path, sha1, mode, pathlen}" ... } which is not only shorter than writing it out in full, it's hopefully less error prone too. [ It's actually a tad faster too - we don't need to recalculate the entry pathlength in both extract and update, but need to do it only once. Also, some callers can avoid doing a "strlen()" on the result, since it's returned as part of the name_entry structure. However, by now we're talking just 1% speedup on "git-rev-list --objects --all", and we're definitely at the point where tree walking is no longer the issue any more. ] NOTE! Not everybody wants to use this new helper function, since some of the tree walkers very much on purpose do the descriptor update separately from the entry extraction. So the "extract + update" sequence still remains as the core sequence, this is just a simplified interface. We should probably add a silly two-line inline helper function for initializing the descriptor from the "struct tree" too, just to cut down on the noise from that common "desc" initializer. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-05-30 20:45:45 +04:00
else if (S_ISDIR(entry.mode)) {
enum object_type type;
struct tree_desc sub;
void *data;
unsigned long size;
data = read_sha1_file(entry.sha1, &type, &size);
if (!data)
die("unable to read tree (%s)",
tree_entry(): new tree-walking helper function This adds a "tree_entry()" function that combines the common operation of doing a "tree_entry_extract()" + "update_tree_entry()". It also has a simplified calling convention, designed for simple loops that traverse over a whole tree: the arguments are pointers to the tree descriptor and a name_entry structure to fill in, and it returns a boolean "true" if there was an entry left to be gotten in the tree. This allows tree traversal with struct tree_desc desc; struct name_entry entry; desc.buf = tree->buffer; desc.size = tree->size; while (tree_entry(&desc, &entry) { ... use "entry.{path, sha1, mode, pathlen}" ... } which is not only shorter than writing it out in full, it's hopefully less error prone too. [ It's actually a tad faster too - we don't need to recalculate the entry pathlength in both extract and update, but need to do it only once. Also, some callers can avoid doing a "strlen()" on the result, since it's returned as part of the name_entry structure. However, by now we're talking just 1% speedup on "git-rev-list --objects --all", and we're definitely at the point where tree walking is no longer the issue any more. ] NOTE! Not everybody wants to use this new helper function, since some of the tree walkers very much on purpose do the descriptor update separately from the entry extraction. So the "extract + update" sequence still remains as the core sequence, this is just a simplified interface. We should probably add a silly two-line inline helper function for initializing the descriptor from the "struct tree" too, just to cut down on the noise from that common "desc" initializer. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-05-30 20:45:45 +04:00
sha1_to_hex(entry.sha1));
init_tree_desc(&sub, data, size);
hit |= grep_tree(opt, paths, &sub, tree_name, down);
free(data);
}
}
strbuf_release(&pathbuf);
return hit;
}
static int grep_object(struct grep_opt *opt, const char **paths,
struct object *obj, const char *name)
{
if (obj->type == OBJ_BLOB)
return grep_sha1(opt, obj->sha1, name, 0);
if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
struct tree_desc tree;
void *data;
unsigned long size;
int hit;
data = read_object_with_reference(obj->sha1, tree_type,
&size, NULL);
if (!data)
die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
init_tree_desc(&tree, data, size);
hit = grep_tree(opt, paths, &tree, name, "");
free(data);
return hit;
}
die("unable to grep from object of type %s", typename(obj->type));
}
static int context_callback(const struct option *opt, const char *arg,
int unset)
{
struct grep_opt *grep_opt = opt->value;
int value;
const char *endp;
if (unset) {
grep_opt->pre_context = grep_opt->post_context = 0;
return 0;
}
value = strtol(arg, (char **)&endp, 10);
if (*endp) {
return error("switch `%c' expects a numerical value",
opt->short_name);
}
grep_opt->pre_context = grep_opt->post_context = value;
return 0;
}
static int file_callback(const struct option *opt, const char *arg, int unset)
{
struct grep_opt *grep_opt = opt->value;
FILE *patterns;
int lno = 0;
struct strbuf sb;
patterns = fopen(arg, "r");
if (!patterns)
die_errno("cannot open '%s'", arg);
while (strbuf_getline(&sb, patterns, '\n') == 0) {
/* ignore empty line like grep does */
if (sb.len == 0)
continue;
append_grep_pattern(grep_opt, strbuf_detach(&sb, NULL), arg,
++lno, GREP_PATTERN);
}
fclose(patterns);
strbuf_release(&sb);
return 0;
}
static int not_callback(const struct option *opt, const char *arg, int unset)
{
struct grep_opt *grep_opt = opt->value;
append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
return 0;
}
static int and_callback(const struct option *opt, const char *arg, int unset)
{
struct grep_opt *grep_opt = opt->value;
append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
return 0;
}
static int open_callback(const struct option *opt, const char *arg, int unset)
{
struct grep_opt *grep_opt = opt->value;
append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
return 0;
}
static int close_callback(const struct option *opt, const char *arg, int unset)
{
struct grep_opt *grep_opt = opt->value;
append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
return 0;
}
static int pattern_callback(const struct option *opt, const char *arg,
int unset)
{
struct grep_opt *grep_opt = opt->value;
append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
return 0;
}
static int help_callback(const struct option *opt, const char *arg, int unset)
{
return -1;
}
int cmd_grep(int argc, const char **argv, const char *prefix)
{
int hit = 0;
int cached = 0;
int external_grep_allowed = 1;
int seen_dashdash = 0;
struct grep_opt opt;
Add "named object array" concept We've had this notion of a "object_list" for a long time, which eventually grew a "name" member because some users (notably git-rev-list) wanted to name each object as it is generated. That object_list is great for some things, but it isn't all that wonderful for others, and the "name" member is generally not used by everybody. This patch splits the users of the object_list array up into two: the traditional list users, who want the list-like format, and who don't actually use or want the name. And another class of users that really used the list as an extensible array, and generally wanted to name the objects. The patch is fairly straightforward, but it's also biggish. Most of it really just cleans things up: switching the revision parsing and listing over to the array makes things like the builtin-diff usage much simpler (we now see exactly how many members the array has, and we don't get the objects reversed from the order they were on the command line). One of the main reasons for doing this at all is that the malloc overhead of the simple object list was actually pretty high, and the array is just a lot denser. So this patch brings down memory usage by git-rev-list by just under 3% (on top of all the other memory use optimizations) on the mozilla archive. It does add more lines than it removes, and more importantly, it adds a whole new infrastructure for maintaining lists of objects, but on the other hand, the new dynamic array code is pretty obvious. The change to builtin-diff-tree.c shows a fairly good example of why an array interface is sometimes more natural, and just much simpler for everybody. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-06-20 04:42:35 +04:00
struct object_array list = { 0, 0, NULL };
const char **paths = NULL;
int i;
int dummy;
struct option options[] = {
OPT_BOOLEAN(0, "cached", &cached,
"search in index instead of in the work tree"),
OPT_GROUP(""),
OPT_BOOLEAN('v', "invert-match", &opt.invert,
"show non-matching lines"),
OPT_BIT('i', "ignore-case", &opt.regflags,
"case insensitive matching", REG_ICASE),
OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
"match patterns only at word boundaries"),
OPT_SET_INT('a', "text", &opt.binary,
"process binary files as text", GREP_BINARY_TEXT),
OPT_SET_INT('I', NULL, &opt.binary,
"don't match patterns in binary files",
GREP_BINARY_NOMATCH),
OPT_GROUP(""),
OPT_BIT('E', "extended-regexp", &opt.regflags,
"use extended POSIX regular expressions", REG_EXTENDED),
OPT_NEGBIT('G', "basic-regexp", &opt.regflags,
"use basic POSIX regular expressions (default)",
REG_EXTENDED),
OPT_BOOLEAN('F', "fixed-strings", &opt.fixed,
"interpret patterns as fixed strings"),
OPT_GROUP(""),
OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"),
OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
OPT_NEGBIT(0, "full-name", &opt.relative,
"show filenames relative to top directory", 1),
OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
"show only filenames instead of matching lines"),
OPT_BOOLEAN(0, "name-only", &opt.name_only,
"synonym for --files-with-matches"),
OPT_BOOLEAN('L', "files-without-match",
&opt.unmatch_name_only,
"show only the names of files without match"),
OPT_BOOLEAN('z', "null", &opt.null_following_name,
"print NUL after filenames"),
OPT_BOOLEAN('c', "count", &opt.count,
"show the number of matches instead of matching lines"),
OPT_SET_INT(0, "color", &opt.color, "highlight matches", 1),
OPT_GROUP(""),
OPT_CALLBACK('C', NULL, &opt, "n",
"show <n> context lines before and after matches",
context_callback),
OPT_INTEGER('B', NULL, &opt.pre_context,
"show <n> context lines before matches"),
OPT_INTEGER('A', NULL, &opt.post_context,
"show <n> context lines after matches"),
OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
context_callback),
OPT_BOOLEAN('p', "show-function", &opt.funcname,
"show a line with the function name before matches"),
OPT_GROUP(""),
OPT_CALLBACK('f', NULL, &opt, "file",
"read patterns from file", file_callback),
{ OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
"match <pattern>", PARSE_OPT_NONEG, pattern_callback },
{ OPTION_CALLBACK, 0, "and", &opt, NULL,
"combine patterns specified with -e",
PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
OPT_BOOLEAN(0, "or", &dummy, ""),
{ OPTION_CALLBACK, 0, "not", &opt, NULL, "",
PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
{ OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
open_callback },
{ OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
close_callback },
OPT_BOOLEAN(0, "all-match", &opt.all_match,
"show only matches from files that match all patterns"),
OPT_GROUP(""),
#if NO_EXTERNAL_GREP
OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed,
"allow calling of grep(1) (ignored by this build)"),
#else
OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed,
"allow calling of grep(1) (default)"),
#endif
{ OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
OPT_END()
};
memset(&opt, 0, sizeof(opt));
opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
opt.relative = 1;
opt.pathname = 1;
opt.pattern_tail = &opt.pattern_list;
opt.regflags = REG_NEWLINE;
strcpy(opt.color_match, GIT_COLOR_RED GIT_COLOR_BOLD);
opt.color = -1;
git_config(grep_config, &opt);
if (opt.color == -1)
opt.color = git_use_color_default;
/*
* If there is no -- then the paths must exist in the working
* tree. If there is no explicit pattern specified with -e or
* -f, we take the first unrecognized non option to be the
* pattern, but then what follows it must be zero or more
* valid refs up to the -- (if exists), and then existing
* paths. If there is an explicit pattern, then the first
* unrecognized non option is the beginning of the refs list
* that continues up to the -- (if exists), and then paths.
*/
argc = parse_options(argc, argv, prefix, options, grep_usage,
PARSE_OPT_KEEP_DASHDASH |
PARSE_OPT_STOP_AT_NON_OPTION |
PARSE_OPT_NO_INTERNAL_HELP);
/* First unrecognized non-option token */
if (argc > 0 && !opt.pattern_list) {
append_grep_pattern(&opt, argv[0], "command line", 0,
GREP_PATTERN);
argv++;
argc--;
}
if ((opt.color && !opt.color_external) || opt.funcname)
external_grep_allowed = 0;
if (!opt.pattern_list)
die("no pattern given.");
if ((opt.regflags != REG_NEWLINE) && opt.fixed)
die("cannot mix --fixed-strings and regexp");
compile_grep_patterns(&opt);
/* Check revs and then paths */
for (i = 0; i < argc; i++) {
const char *arg = argv[i];
unsigned char sha1[20];
/* Is it a rev? */
if (!get_sha1(arg, sha1)) {
struct object *object = parse_object(sha1);
if (!object)
die("bad object %s", arg);
Add "named object array" concept We've had this notion of a "object_list" for a long time, which eventually grew a "name" member because some users (notably git-rev-list) wanted to name each object as it is generated. That object_list is great for some things, but it isn't all that wonderful for others, and the "name" member is generally not used by everybody. This patch splits the users of the object_list array up into two: the traditional list users, who want the list-like format, and who don't actually use or want the name. And another class of users that really used the list as an extensible array, and generally wanted to name the objects. The patch is fairly straightforward, but it's also biggish. Most of it really just cleans things up: switching the revision parsing and listing over to the array makes things like the builtin-diff usage much simpler (we now see exactly how many members the array has, and we don't get the objects reversed from the order they were on the command line). One of the main reasons for doing this at all is that the malloc overhead of the simple object list was actually pretty high, and the array is just a lot denser. So this patch brings down memory usage by git-rev-list by just under 3% (on top of all the other memory use optimizations) on the mozilla archive. It does add more lines than it removes, and more importantly, it adds a whole new infrastructure for maintaining lists of objects, but on the other hand, the new dynamic array code is pretty obvious. The change to builtin-diff-tree.c shows a fairly good example of why an array interface is sometimes more natural, and just much simpler for everybody. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-06-20 04:42:35 +04:00
add_object_array(object, arg, &list);
continue;
}
if (!strcmp(arg, "--")) {
i++;
seen_dashdash = 1;
}
break;
}
/* The rest are paths */
if (!seen_dashdash) {
int j;
for (j = i; j < argc; j++)
verify_filename(prefix, argv[j]);
}
if (i < argc) {
paths = get_pathspec(prefix, argv + i);
if (opt.prefix_length && opt.relative) {
/* Make sure we do not get outside of paths */
for (i = 0; paths[i]; i++)
if (strncmp(prefix, paths[i], opt.prefix_length))
die("git grep: cannot generate relative filenames containing '..'");
}
}
else if (prefix) {
paths = xcalloc(2, sizeof(const char *));
paths[0] = prefix;
paths[1] = NULL;
}
if (!list.nr) {
if (!cached)
setup_work_tree();
return !grep_cache(&opt, paths, cached, external_grep_allowed);
}
if (cached)
die("both --cached and trees are given.");
Add "named object array" concept We've had this notion of a "object_list" for a long time, which eventually grew a "name" member because some users (notably git-rev-list) wanted to name each object as it is generated. That object_list is great for some things, but it isn't all that wonderful for others, and the "name" member is generally not used by everybody. This patch splits the users of the object_list array up into two: the traditional list users, who want the list-like format, and who don't actually use or want the name. And another class of users that really used the list as an extensible array, and generally wanted to name the objects. The patch is fairly straightforward, but it's also biggish. Most of it really just cleans things up: switching the revision parsing and listing over to the array makes things like the builtin-diff usage much simpler (we now see exactly how many members the array has, and we don't get the objects reversed from the order they were on the command line). One of the main reasons for doing this at all is that the malloc overhead of the simple object list was actually pretty high, and the array is just a lot denser. So this patch brings down memory usage by git-rev-list by just under 3% (on top of all the other memory use optimizations) on the mozilla archive. It does add more lines than it removes, and more importantly, it adds a whole new infrastructure for maintaining lists of objects, but on the other hand, the new dynamic array code is pretty obvious. The change to builtin-diff-tree.c shows a fairly good example of why an array interface is sometimes more natural, and just much simpler for everybody. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-06-20 04:42:35 +04:00
for (i = 0; i < list.nr; i++) {
struct object *real_obj;
Add "named object array" concept We've had this notion of a "object_list" for a long time, which eventually grew a "name" member because some users (notably git-rev-list) wanted to name each object as it is generated. That object_list is great for some things, but it isn't all that wonderful for others, and the "name" member is generally not used by everybody. This patch splits the users of the object_list array up into two: the traditional list users, who want the list-like format, and who don't actually use or want the name. And another class of users that really used the list as an extensible array, and generally wanted to name the objects. The patch is fairly straightforward, but it's also biggish. Most of it really just cleans things up: switching the revision parsing and listing over to the array makes things like the builtin-diff usage much simpler (we now see exactly how many members the array has, and we don't get the objects reversed from the order they were on the command line). One of the main reasons for doing this at all is that the malloc overhead of the simple object list was actually pretty high, and the array is just a lot denser. So this patch brings down memory usage by git-rev-list by just under 3% (on top of all the other memory use optimizations) on the mozilla archive. It does add more lines than it removes, and more importantly, it adds a whole new infrastructure for maintaining lists of objects, but on the other hand, the new dynamic array code is pretty obvious. The change to builtin-diff-tree.c shows a fairly good example of why an array interface is sometimes more natural, and just much simpler for everybody. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-06-20 04:42:35 +04:00
real_obj = deref_tag(list.objects[i].item, NULL, 0);
if (grep_object(&opt, paths, real_obj, list.objects[i].name))
hit = 1;
}
free_grep_patterns(&opt);
return !hit;
}