On Sat, Jul 25, 2020 at 08:10:10PM +0200, Stefan Sperling wrote:
> Got lacks a built-in command to examine work tree meta data.
> 
> It has always been possible to obtain some information from the files
> in the .got/ directory because some of those files are plain-text:
> 
>  - Get the current branch:
>    $ cat .got/head-ref
> 
>  - Get the current base commit:
>    $ cat .got/base-commit
> 
> But viewing information stored in the file index is difficult. And making
> this information more accessible could help with debugging problems with
> a broken work tree.
> 
> This first implementation of 'got info' is not very flexible. I did not
> add any options yet. The output can be parsed by scripts but my intention
> is to add some options later to make scripting very easy.
> Options which would make 'got info' print individual bits of information,
> such as "show me the repository path for this work tree", and so on.
> Similar functionality is offered by 'svn info --show-item' in Subversion.
> 
> My initial implementation just dumped everything by default but that can be
> too much. The default output implemented in the patch below shows general
> context information only:
> 
> $ got info
> work tree: /usr/src
> work tree base commit: 3ff74ee673f6439bbf658b9cb1b2ff65fbb3ee09
> work tree path prefix: /
> work tree branch reference: refs/heads/iwm-txagg
> work tree UUID: bb7789ce-609a-472c-86e2-a42c1a58caff
> repository: /git/src.git
> $ 
> 
> Information stored in the file index can be appended to this general info
> by specifying path arguments. (File index content would not be useful in
> isolation, so general context info is always displayed.)
> 
> $ got info Makefile
> work tree: /usr/src
> work tree base commit: 3ff74ee673f6439bbf658b9cb1b2ff65fbb3ee09
> work tree path prefix: /
> work tree branch reference: refs/heads/iwm-txagg
> work tree UUID: bb7789ce-609a-472c-86e2-a42c1a58caff
> repository: /git/src.git
> -----------------------------------------------
> file: Makefile
> mode: 644
> timestamp: Mon Jul 13 14:53:54 2020 CEST
> based on blob: 5002ac02445f9ab3b2fc89397a45f00b8b287fd2
> based on commit: 20599cf97eccdb9ab18672df969fe10c00802e7d
> 
> Entire sub trees can also be shown:
> 
> $ got info sbin/ifconfig/
> work tree: /usr/src
> work tree base commit: 3ff74ee673f6439bbf658b9cb1b2ff65fbb3ee09
> work tree path prefix: /
> work tree branch reference: refs/heads/iwm-txagg
> work tree UUID: bb7789ce-609a-472c-86e2-a42c1a58caff
> repository: /git/src.git
> -----------------------------------------------
> file: sbin/ifconfig/Makefile
> mode: 644
> timestamp: Mon Jul 13 14:58:12 2020 CEST
> based on blob: b674f82f00422b46f899c100a77d8ce875a4342b
> based on commit: 20599cf97eccdb9ab18672df969fe10c00802e7d
> -----------------------------------------------
> file: sbin/ifconfig/brconfig.c
> mode: 644
> timestamp: Mon Jul 13 14:58:12 2020 CEST
> based on blob: e33b73ebf2ca9d3cd26e08d63427e45e378c81af
> based on commit: 20599cf97eccdb9ab18672df969fe10c00802e7d
> -----------------------------------------------
> file: sbin/ifconfig/ifconfig.8
> mode: 644
> timestamp: Mon Jul 13 14:58:12 2020 CEST
> based on blob: 10111bf2c454181e92e103a0dd2bbfdbd77e9f90
> based on commit: 20599cf97eccdb9ab18672df969fe10c00802e7d
> -----------------------------------------------
> file: sbin/ifconfig/ifconfig.c
> mode: 644
> timestamp: Mon Jul 13 14:58:12 2020 CEST
> based on blob: cc314b7a44f8cd9a74ec6e32df6ce1cf8183a068
> based on commit: 20599cf97eccdb9ab18672df969fe10c00802e7d
> -----------------------------------------------
> file: sbin/ifconfig/ifconfig.h
> mode: 644
> timestamp: Mon Jul 13 14:58:12 2020 CEST
> based on blob: 1d5b9ccff10dc0a572b16fcff18d9de45bd00fef
> based on commit: 20599cf97eccdb9ab18672df969fe10c00802e7d
> -----------------------------------------------
> file: sbin/ifconfig/sff.c
> mode: 644
> timestamp: Mon Jul 13 14:58:12 2020 CEST
> based on blob: a800a1f55d0ee593deaaf7e091e22c4c4cc5e1b5
> based on commit: 20599cf97eccdb9ab18672df969fe10c00802e7d
> $
> 
> It is also possible to show all files, for example by specifying a dot
> as the path argument: got info .
> This is probably not intuitive and I did consider adding an option argument
> for this purpose (such as -a, "show all"). But such an option becomes awkward
> to handle in combination with path arguments so I decided to stick to just
> path arguments for now. The man page explains how it works.
> 
> I am dropping the alias 'got in' (for 'got init').
> This alias is too close to either 'init' or 'info'. Both commands are
> short enough so I think the best approach is to remove this alias.
> 
> Unfortunately, the pledge for 'got info' is quite large:
> 	"stdio rpath wpath flock proc exec sendfd unveil"
> The "proc exec sendfd" pledges are only needed because we are reading the
> repository config file with 'got-read-gitconfig' when opening a work tree.
> This could be improved later.
> 
> Any suggestions or concerns?
> 

Diff looks good to me. It's a good start.

> diff cf07f22bd8beb1ed9e048433d553e394cc323c5f /home/stsp/src/got
> blob - 136fa542011eaf775871dd87e5f0c0a2ba5ef5a4
> file + got/got.1
> --- got/got.1
> +++ got/got.1
> @@ -73,9 +73,6 @@ the
>  command must be used to populate the empty repository before
>  .Cm got checkout
>  can be used.
> -.It Cm in
> -Short alias for
> -.Cm init .
>  .It Cm import Oo Fl b Ar branch Oc Oo Fl m Ar message Oc Oo Fl r Ar repository-path Oc Oo Fl I Ar pattern Oc Ar directory
>  Create an initial commit in a repository from the file hierarchy
>  within the specified
> @@ -1862,6 +1859,23 @@ Interpret all arguments as paths only.
>  This option can be used to resolve ambiguity in cases where paths
>  look like tag names, reference names, or object IDs.
>  .El
> +.It Cm info Op Ar path ...
> +Display meta-data stored in a work tree.
> +See
> +.Xr got-worktree 5
> +for details.
> +.Pp
> +The work tree to use is resolved implicitly by walking upwards from the
> +current working directory.
> +.Pp
> +If one or more
> +.Ar path
> +arguments are specified, show additional per-file information for tracked
> +files located at or within these paths.
> +If a
> +.Ar path
> +argument corresponds to the work tree's root directory, display information
> +for all tracked files.
>  .El
>  .Sh ENVIRONMENT
>  .Bl -tag -width GOT_AUTHOR
> @@ -1996,6 +2010,11 @@ working directory, so this command will log the subdir
>  And this command has the same effect:
>  .Pp
>  .Dl $ cd sys/dev/usb && got log ../../uvm
> +.Pp
> +And this command displays work tree meta-data about all tracked files:
> +.Pp
> +.Dl $ cd /usr/src
> +.Dl $ got info\ . | less
>  .Pp
>  Add new files and remove obsolete files in a work tree directory:
>  .Pp
> blob - fb2d55929f89f48f959e0ddfdc2fa639fe2c64d0
> file + got/got.c
> --- got/got.c
> +++ got/got.c
> @@ -109,6 +109,7 @@ __dead static void	usage_integrate(void);
>  __dead static void	usage_stage(void);
>  __dead static void	usage_unstage(void);
>  __dead static void	usage_cat(void);
> +__dead static void	usage_info(void);
>  
>  static const struct got_error*		cmd_init(int, char *[]);
>  static const struct got_error*		cmd_import(int, char *[]);
> @@ -136,9 +137,10 @@ static const struct got_error*		cmd_integrate(int, cha
>  static const struct got_error*		cmd_stage(int, char *[]);
>  static const struct got_error*		cmd_unstage(int, char *[]);
>  static const struct got_error*		cmd_cat(int, char *[]);
> +static const struct got_error*		cmd_info(int, char *[]);
>  
>  static struct got_cmd got_commands[] = {
> -	{ "init",	cmd_init,	usage_init,	"in" },
> +	{ "init",	cmd_init,	usage_init,	"" },
>  	{ "import",	cmd_import,	usage_import,	"im" },
>  	{ "clone",	cmd_clone,	usage_clone,	"cl" },
>  	{ "fetch",	cmd_fetch,	usage_fetch,	"fe" },
> @@ -164,6 +166,7 @@ static struct got_cmd got_commands[] = {
>  	{ "stage",	cmd_stage,	usage_stage,	"sg" },
>  	{ "unstage",	cmd_unstage,	usage_unstage,	"ug" },
>  	{ "cat",	cmd_cat,	usage_cat,	"" },
> +	{ "info",	cmd_info,	usage_info,	"" },
>  };
>  
>  static void
> @@ -9315,5 +9318,186 @@ done:
>  		if (error == NULL)
>  			error = repo_error;
>  	}
> +	return error;
> +}
> +
> +__dead static void
> +usage_info(void)
> +{
> +	fprintf(stderr, "usage: %s info [path ...]\n",
> +	    getprogname());
> +	exit(1);
> +}
> +
> +static const struct got_error *
> +print_path_info(void *arg, const char *path, mode_t mode, time_t mtime,
> +    struct got_object_id *blob_id, struct got_object_id *staged_blob_id,
> +    struct got_object_id *commit_id)
> +{
> +	const struct got_error *err = NULL;
> +	char *id_str = NULL;
> +	char datebuf[128];
> +	struct tm mytm, *tm;
> +	struct got_pathlist_head *paths = arg;
> +	struct got_pathlist_entry *pe;
> +
> +	/*
> +	 * Clear error indication from any of the path arguments which
> +	 * would cause this file index entry to be displayed.
> +	 */
> +	TAILQ_FOREACH(pe, paths, entry) {
> +		if (got_path_cmp(path, pe->path, strlen(path),
> +		    pe->path_len) == 0 ||
> +		    got_path_is_child(path, pe->path, pe->path_len))
> +			pe->data = NULL; /* no error */
> +	}
> +
> +	printf(GOT_COMMIT_SEP_STR);
> +	if (S_ISLNK(mode))
> +		printf("symlink: %s\n", path);
> +	else if (S_ISREG(mode)) {
> +		printf("file: %s\n", path);
> +		printf("mode: %o\n", mode & (S_IRWXU | S_IRWXG | S_IRWXO));
> +	} else if (S_ISDIR(mode))
> +		printf("directory: %s\n", path);
> +	else
> +		printf("something: %s\n", path);
> +
> +	tm = localtime_r(&mtime, &mytm);
> +	if (tm == NULL)
> +		return NULL;
> +	if (strftime(datebuf, sizeof(datebuf), "%c %Z", tm) >= sizeof(datebuf))
> +		return got_error(GOT_ERR_NO_SPACE);
> +	printf("timestamp: %s\n", datebuf);
> +
> +	if (blob_id) {
> +		err = got_object_id_str(&id_str, blob_id);
> +		if (err)
> +			return err;
> +		printf("based on blob: %s\n", id_str);
> +		free(id_str);
> +	}
> +
> +	if (staged_blob_id) {
> +		err = got_object_id_str(&id_str, staged_blob_id);
> +		if (err)
> +			return err;
> +		printf("based on staged blob: %s\n", id_str);
> +		free(id_str);
> +	}
> +
> +	if (commit_id) {
> +		err = got_object_id_str(&id_str, commit_id);
> +		if (err)
> +			return err;
> +		printf("based on commit: %s\n", id_str);
> +		free(id_str);
> +	}
> +
> +	return NULL;
> +}
> +
> +static const struct got_error *
> +cmd_info(int argc, char *argv[])
> +{
> +	const struct got_error *error = NULL;
> +	struct got_worktree *worktree = NULL;
> +	char *cwd = NULL, *id_str = NULL;
> +	struct got_pathlist_head paths;
> +	struct got_pathlist_entry *pe;
> +	char *uuidstr = NULL;
> +	int ch, show_files = 0;
> +
> +	TAILQ_INIT(&paths);
> +
> +	while ((ch = getopt(argc, argv, "")) != -1) {
> +		switch (ch) {
> +		default:
> +			usage_info();
> +			/* NOTREACHED */
> +		}
> +	}
> +
> +	argc -= optind;
> +	argv += optind;
> +
> +#ifndef PROFILE
> +	if (pledge("stdio rpath wpath flock proc exec sendfd unveil",
> +	    NULL) == -1)
> +		err(1, "pledge");
> +#endif
> +	cwd = getcwd(NULL, 0);
> +	if (cwd == NULL) {
> +		error = got_error_from_errno("getcwd");
> +		goto done;
> +	}
> +
> +	error = got_worktree_open(&worktree, cwd);
> +	if (error) {
> +		if (error->code == GOT_ERR_NOT_WORKTREE)
> +			error = wrap_not_worktree_error(error, "status", cwd);
> +		goto done;
> +	}
> +
> +	error = apply_unveil(NULL, 0, got_worktree_get_root_path(worktree));
> +	if (error)
> +		goto done;
> +
> +	if (argc >= 1) {
> +		error = get_worktree_paths_from_argv(&paths, argc, argv,
> +		    worktree);
> +		if (error)
> +			goto done;
> +		show_files = 1;
> +	}
> +
> +	error = got_object_id_str(&id_str,
> +	    got_worktree_get_base_commit_id(worktree));
> +	if (error)
> +		goto done;
> +
> +	error = got_worktree_get_uuid(&uuidstr, worktree);
> +	if (error)
> +		goto done;
> +
> +	printf("work tree: %s\n", got_worktree_get_root_path(worktree));
> +	printf("work tree base commit: %s\n", id_str);
> +	printf("work tree path prefix: %s\n",
> +	    got_worktree_get_path_prefix(worktree));
> +	printf("work tree branch reference: %s\n",
> +	    got_worktree_get_head_ref_name(worktree));
> +	printf("work tree UUID: %s\n", uuidstr);
> +	printf("repository: %s\n", got_worktree_get_repo_path(worktree));
> +
> +	if (show_files) {
> +		struct got_pathlist_entry *pe;
> +		TAILQ_FOREACH(pe, &paths, entry) {
> +			if (pe->path_len == 0)
> +				continue;
> +			/*
> +			 * Assume this path will fail. This will be corrected
> +			 * in print_path_info() in case the path does suceeed.
> +			 */
> +			pe->data = (void *)got_error_path(pe->path,
> +			    GOT_ERR_BAD_PATH);
> +		}
> +		error = got_worktree_path_info(worktree, &paths,
> +		    print_path_info, &paths, check_cancelled, NULL);
> +		if (error)
> +			goto done;
> +		TAILQ_FOREACH(pe, &paths, entry) {
> +			if (pe->data != NULL) {
> +				error = pe->data; /* bad path */
> +				break;
> +			}
> +		}
> +	}
> +done:
> +	TAILQ_FOREACH(pe, &paths, entry)
> +		free((char *)pe->path);
> +	got_pathlist_free(&paths);
> +	free(cwd);
> +	free(id_str);
> +	free(uuidstr);
>  	return error;
>  }
> blob - 277c66b0096dcee60d7ae5305877516328376f34
> file + include/got_worktree.h
> --- include/got_worktree.h
> +++ include/got_worktree.h
> @@ -75,6 +75,12 @@ const char *got_worktree_get_repo_path(struct got_work
>  const char *got_worktree_get_path_prefix(struct got_worktree *);
>  
>  /*
> + * Get the UUID of a work tree as a string.
> + * The caller must dispose of the returned UUID string with free(3).
> + */
> +const struct got_error *got_worktree_get_uuid(char **, struct got_worktree *);
> +
> +/*
>   * Check if a user-provided path prefix matches that of the worktree.
>   */
>  const struct got_error *got_worktree_match_path_prefix(int *,
> @@ -442,3 +448,18 @@ const struct got_error *got_worktree_stage(struct got_
>  const struct got_error *got_worktree_unstage(struct got_worktree *,
>      struct got_pathlist_head *, got_worktree_checkout_cb, void *,
>      got_worktree_patch_cb, void *, struct got_repository *);
> +
> +/* A callback function which is invoked with per-path info. */
> +typedef const struct got_error *(*got_worktree_path_info_cb)(void *,
> +    const char *path, mode_t mode, time_t mtime,
> +    struct got_object_id *blob_id, struct got_object_id *staged_blob_id,
> +    struct got_object_id *commit_id);
> +
> +/* 
> + * Report the status of paths in the work tree.
> + * The info callback will be invoked with the provided void * argument,
> + * a path, and a corresponding status code.
> + */
> +const struct got_error *
> +got_worktree_path_info(struct got_worktree *, struct got_pathlist_head *,
> +    got_worktree_path_info_cb, void *, got_cancel_cb , void *);
> blob - 0a072932c20e3ee4dec70243cb7cdc3739d3c8ba
> file + lib/worktree.c
> --- lib/worktree.c
> +++ lib/worktree.c
> @@ -2143,21 +2143,33 @@ diff_new(void *arg, struct got_tree_entry *te, const c
>  	return err;
>  }
>  
> +const struct got_error *
> +got_worktree_get_uuid(char **uuidstr, struct got_worktree *worktree)
> +{
> +	uint32_t uuid_status;
> +
> +	uuid_to_string(&worktree->uuid, uuidstr, &uuid_status);
> +	if (uuid_status != uuid_s_ok) {
> +		*uuidstr = NULL;
> +		return got_error_uuid(uuid_status, "uuid_to_string");
> +	}
> +
> +	return NULL;
> +}
> +
>  static const struct got_error *
>  get_ref_name(char **refname, struct got_worktree *worktree, const char *prefix)
>  {
>  	const struct got_error *err = NULL;
>  	char *uuidstr = NULL;
> -	uint32_t uuid_status;
>  
>  	*refname = NULL;
>  
> -	uuid_to_string(&worktree->uuid, &uuidstr, &uuid_status);
> -	if (uuid_status != uuid_s_ok)
> -		return got_error_uuid(uuid_status, "uuid_to_string");
> +	err = got_worktree_get_uuid(&uuidstr, worktree);
> +	if (err)
> +		return err;
>  
> -	if (asprintf(refname, "%s-%s", prefix, uuidstr)
> -	    == -1) {
> +	if (asprintf(refname, "%s-%s", prefix, uuidstr) == -1) {
>  		err = got_error_from_errno("asprintf");
>  		*refname = NULL;
>  	}
> @@ -7667,6 +7679,95 @@ done:
>  	if (fileindex)
>  		got_fileindex_free(fileindex);
>  	unlockerr = lock_worktree(worktree, LOCK_SH);
> +	if (unlockerr && err == NULL)
> +		err = unlockerr;
> +	return err;
> +}
> +
> +struct report_file_info_arg {
> +	struct got_worktree *worktree;
> +	got_worktree_path_info_cb info_cb;
> +	void *info_arg;
> +	struct got_pathlist_head *paths;
> +	got_cancel_cb cancel_cb;
> +	void *cancel_arg;
> +};
> +
> +static const struct got_error *
> +report_file_info(void *arg, struct got_fileindex_entry *ie)
> +{
> +	struct report_file_info_arg *a = arg;
> +	struct got_pathlist_entry *pe;
> +	struct got_object_id blob_id, staged_blob_id, commit_id;
> +	struct got_object_id *blob_idp = NULL, *staged_blob_idp = NULL;
> +	struct got_object_id *commit_idp = NULL;
> +	int stage;
> +
> +	if (a->cancel_cb && a->cancel_cb(a->cancel_arg))
> +		return got_error(GOT_ERR_CANCELLED);
> +
> +	TAILQ_FOREACH(pe, a->paths, entry) {
> +		if (pe->path_len == 0 || strcmp(pe->path, ie->path) == 0 ||
> +		    got_path_is_child(ie->path, pe->path, pe->path_len))
> +			break;
> +	}
> +	if (pe == NULL) /* not found */
> +		return NULL;
> +
> +	if (got_fileindex_entry_has_blob(ie)) {
> +		memcpy(blob_id.sha1, ie->blob_sha1, SHA1_DIGEST_LENGTH);
> +		blob_idp = &blob_id;
> +	}
> +	stage = got_fileindex_entry_stage_get(ie);
> +	if (stage == GOT_FILEIDX_STAGE_MODIFY ||
> +	    stage == GOT_FILEIDX_STAGE_ADD) {
> +		memcpy(staged_blob_id.sha1, ie->staged_blob_sha1,
> +		    SHA1_DIGEST_LENGTH);
> +		staged_blob_idp = &staged_blob_id;
> +	}
> +
> +	if (got_fileindex_entry_has_commit(ie)) {
> +		memcpy(commit_id.sha1, ie->commit_sha1, SHA1_DIGEST_LENGTH);
> +		commit_idp = &commit_id;
> +	}
> +
> +	return a->info_cb(a->info_arg, ie->path, got_fileindex_perms_to_st(ie),
> +	    (time_t)ie->mtime_sec, blob_idp, staged_blob_idp, commit_idp);
> +}
> +
> +const struct got_error *
> +got_worktree_path_info(struct got_worktree *worktree,
> +    struct got_pathlist_head *paths,
> +    got_worktree_path_info_cb info_cb, void *info_arg,
> +    got_cancel_cb cancel_cb, void *cancel_arg)
> +
> +{
> +	const struct got_error *err = NULL, *unlockerr;
> +	struct got_fileindex *fileindex = NULL;
> +	char *fileindex_path = NULL;
> +	struct report_file_info_arg arg;
> +
> +	err = lock_worktree(worktree, LOCK_SH);
> +	if (err)
> +		return err;
> +
> +	err = open_fileindex(&fileindex, &fileindex_path, worktree);
> +	if (err)
> +		goto done;
> +
> +	arg.worktree = worktree;
> +	arg.info_cb = info_cb;
> +	arg.info_arg = info_arg;
> +	arg.paths = paths;
> +	arg.cancel_cb = cancel_cb;
> +	arg.cancel_arg = cancel_arg;
> +	err = got_fileindex_for_each_entry_safe(fileindex, report_file_info,
> +	    &arg);
> +done:
> +	free(fileindex_path);
> +	if (fileindex)
> +		got_fileindex_free(fileindex);
> +	unlockerr = lock_worktree(worktree, LOCK_UN);
>  	if (unlockerr && err == NULL)
>  		err = unlockerr;
>  	return err;

-- 

Tracey Emery