ec3/src/capture.c

#include "bin.h"
#include "chunk-table.h"
#include "commands.h"
#include "fs-tree.h"
#include "image.h"
#include "misc.h"
#include "status.h"
#include "string-table.h"
#include "volume.h"

#include <blue/cmd.h>
#include <blue/io/directory.h>
#include <blue/object/list.h>
#include <blue/term/print.h>
#include <errno.h>
#include <stdio.h>
#include <stdlib.h>

#define BUFFER_SIZE 65536

enum {
	OPT_OUTPATH,
	OPT_OUTPATH_PATH,

	OPT_IDENT,
	OPT_IDENT_VAL,

	OPT_DIRECTORY,
	OPT_DIRECTORY_PATH,

	OPT_TAGGED_DIRECTORY,
	OPT_TAGGED_DIRECTORY_TAG,
	OPT_TAGGED_DIRECTORY_PATH,

	OPT_VERBOSE,
};

struct capture_ctx {
	struct fs_tree ctx_tree;
	struct ec3_image_ioctx *ctx_image;
	struct chunk_table ctx_chunks;
	struct string_table ctx_strings;
	struct ec3_volume *ctx_volume;
};

static enum ec3_status capture_file(
	struct capture_ctx *ctx,
	struct ec3_volume *vol,
	b_directory *dir,
	const char *filename,
	const b_path *filepath,
	ec3_chunk_id out_chunk)
{
	const struct ec3_image_info *image_info
		= ec3_image_ioctx_get_info(ctx->ctx_image);
	size_t key = string_table_get(&ctx->ctx_strings, filename);
	size_t buf_len = image_info->img_cluster_size;
	char *buf = malloc(buf_len);

	if (!buf) {
		return EC3_ERR_NO_MEMORY;
	}

	b_file *src = NULL;
	b_status status = b_file_open(
		dir,
		filepath,
		B_FILE_READ_ONLY | B_FILE_BINARY,
		&src);
	if (!B_OK(status)) {
		free(buf);
		return ec3_status_from_b_status(status, EC3_ERR_NO_ENTRY);
	}

	enum ec3_status s2 = EC3_SUCCESS;
	chunk_table_begin_chunk(&ctx->ctx_chunks);

	size_t chunk_size = 0;

	while (1) {
		size_t nr_read = 0;
		status = b_file_read(
			src,
			B_OFFSET_CURRENT,
			buf_len,
			buf,
			&nr_read);

		if (!B_OK(status)) {
			s2 = ec3_status_from_b_status(
				status,
				EC3_ERR_IO_FAILURE);
			break;
		}

		enum ec3_status s2
			= chunk_table_put(&ctx->ctx_chunks, buf, nr_read);

		if (s2 != EC3_SUCCESS) {
			break;
		}

		chunk_size += nr_read;

		if (nr_read < buf_len) {
			break;
		}
	}

	s2 = chunk_table_end_chunk(&ctx->ctx_chunks, out_chunk);

	if (s2 != EC3_SUCCESS) {
		return s2;
	}

	char id_str[128];
	ec3_chunk_id_to_string(out_chunk, id_str, sizeof id_str);
	// printf("wrote %zu byte chunk %s\n", chunk_size, id_str);

	free(buf);
	b_file_release(src);
	return s2;
}

static int print_fs_tree_node(
	struct fs_tree *tree,
	struct fs_tree_node *node,
	unsigned int depth,
	int direction,
	void *arg)
{
	if (direction == ITERATE_POSTORDER) {
		depth = 0;
	}

	for (unsigned int i = 0; i < depth; i++) {
		fputs("    ", stdout);
	}

	printf("%s ", node->n_name);
	char id_str[128];

	switch (node->n_type) {
	case FS_TREE_FILE:
		ec3_chunk_id_to_string(node->n_chunk, id_str, sizeof id_str);
		printf("[%s]", id_str);
		break;
	case FS_TREE_DIR:
		printf("[d]");
		break;
	default:
		break;
	}

	printf("\n");

	return 0;
}

struct capture_directory_structure_args {
	b_list *args_stack;
	struct capture_ctx *args_ctx;
};

static int capture_directory_structure_callback(
	struct fs_tree *tree,
	struct fs_tree_node *node,
	unsigned int depth,
	int direction,
	void *arg)
{
	if (direction != ITERATE_POSTORDER) {
		return 0;
	}

	struct capture_directory_structure_args *args = arg;
	b_list *stack = args->args_stack;
	struct capture_ctx *ctx = args->args_ctx;

	if (node->n_type == FS_TREE_FILE) {
		b_list_push_back(stack, node);
		return 0;
	}

	size_t nr_children = fs_tree_node_get_nr_children(node);
	struct ec3_directory_entry dent;
	enum ec3_status status = EC3_SUCCESS;

	chunk_table_begin_chunk(&ctx->ctx_chunks);
	for (size_t i = 0; i < nr_children; i++) {
		struct fs_tree_node *child = b_list_pop_back(stack);
		if (!child) {
			return EC3_ERR_INTERNAL_FAILURE;
		}

		dent.d_name = b_i32_htob(
			string_table_get(&ctx->ctx_strings, child->n_name));
		dent.d_vnode = b_i32_htob(child->n_id);

		status = chunk_table_put(&ctx->ctx_chunks, &dent, sizeof dent);
	}

	status = chunk_table_end_chunk(&ctx->ctx_chunks, node->n_chunk);
	if (status != EC3_SUCCESS) {
		return status;
	}

	char id_str[128];
	ec3_chunk_id_to_string(node->n_chunk, id_str, sizeof id_str);
	printf("[d %s / %zu] wrote %zu byte chunk %s\n",
	       node->n_name,
	       node->n_id,
	       nr_children * sizeof dent,
	       id_str);

	struct ec3_vnode vnode = {
		.v_id = node->n_id,
		.v_mode = EC3_V_DIR,
	};

	memcpy(vnode.v_data, node->n_chunk, sizeof vnode.v_data);

	ec3_volume_put_vnode(ctx->ctx_volume, &vnode);

	b_list_push_back(stack, node);

	return 0;
}

static enum ec3_status capture_directory_structure(struct capture_ctx *ctx)
{
	b_list *stack = b_list_create();

	struct capture_directory_structure_args args = {
		.args_stack = stack,
		.args_ctx = ctx,
	};

	fs_tree_iterate(
		&ctx->ctx_tree,
		capture_directory_structure_callback,
		&args);

	return EC3_SUCCESS;
}

static enum ec3_status capture_directory(
	struct capture_ctx *ctx,
	uint64_t id,
	const char *cpath)
{
	b_path *path = b_path_create_from_cstr(cpath);
	b_directory *dir;
	b_status status = b_directory_open(NULL, path, &dir);
	b_path_release(path);

	if (!B_OK(status)) {
		return EC3_ERR_NO_ENTRY;
	}

	struct ec3_tag_ioctx *volu;
	enum ec3_status s2 = ec3_image_ioctx_create_tag(
		ctx->ctx_image,
		EC3_TAG_VOLU,
		id,
		EC3_TAG_IO_READ | EC3_TAG_IO_WRITE,
		&volu);
	if (s2 != EC3_SUCCESS) {
		b_directory_release(dir);
		return s2;
	}

	struct ec3_volume *volume;
	s2 = ec3_volume_create(ctx->ctx_image, volu, &volume);

	if (s2 != EC3_SUCCESS) {
		ec3_tag_ioctx_close(volu);
		b_directory_release(dir);
		return s2;
	}

	ctx->ctx_volume = volume;

	fs_tree_init(&ctx->ctx_tree);

	struct ec3_vnode vnode = {0};
	vnode.v_id = ctx->ctx_tree.fs_root->n_id;
	ec3_volume_put_vnode(volume, &vnode);

	enum ec3_status status2 = EC3_SUCCESS;
	b_directory_iterator it;
	b_directory_iterator_begin(dir, &it, B_DIRECTORY_ITERATE_PARENT_LAST);
	while (b_directory_iterator_is_valid(&it)) {
		if (!b_directory_path_is_file(dir, it.filepath)) {
			printf("dir: %s\n", b_path_ptr(it.filepath));
			b_directory_iterator_next(&it);
			continue;
		}

		ec3_vnode_from_file_info(&it.info, &vnode);

		printf("file: %s\n", b_path_ptr(it.filepath));

		status2 = capture_file(
			ctx,
			volume,
			dir,
			it.filename,
			it.filepath,
			vnode.v_data);

		if (status2 != EC3_SUCCESS) {
			b_err("failed to capture file '%s'",
			      b_path_ptr(it.filepath));
			b_i("error code: %s", ec3_status_to_string(status2));
			break;
		}

		fs_tree_put(
			&ctx->ctx_tree,
			b_path_ptr(it.filepath),
			&vnode.v_id,
			vnode.v_data);

		ec3_volume_put_vnode(volume, &vnode);

		b_directory_iterator_next(&it);
	}

	// fs_tree_iterate(&ctx->ctx_tree, print_fs_tree_node, NULL);

	capture_directory_structure(ctx);
	fs_tree_fini(&ctx->ctx_tree);

	ec3_tag_ioctx_close(volu);

	b_directory_release(dir);
	return status2;
}

static int capture(
	const b_command *self,
	const b_arglist *opt,
	const b_array *args)
{
	const char *out_path = NULL;
	b_arglist_get_string(opt, OPT_OUTPATH, OPT_OUTPATH_PATH, 0, &out_path);

	enum ec3_status status = EC3_SUCCESS;

	uint64_t ident = 0;
	const char *ident_str;
	b_arglist_get_string(opt, OPT_IDENT, OPT_IDENT_VAL, 0, &ident_str);
	if (ident_str) {
		status = ec3_identifier_from_string(ident_str, &ident);
	}

	if (status != EC3_SUCCESS) {
		b_err("'%s' is not a valid container identifier", ident_str);
		return -1;
	}

	struct ec3_image_ioctx *image = NULL;
	struct ec3_parameters param = {
		.p_cluster_size = EC3_CLUSTER_16K,
		.p_compression_func = EC3_COMPRESSION_ZSTD,
		.p_ident = ident,
	};
	status = ec3_image_ioctx_open(
		out_path,
		&param,
		EC3_IMAGE_IO_WRITE | EC3_IMAGE_IO_TRUNCATE,
		&image);

	if (status != EC3_SUCCESS) {
		b_err("cannot initialise EC3 writer");
		return -1;
	}

	uint64_t ctab_id, cdat_id, stab_id;
	ec3_identifier_from_string("_CHKTAB0", &ctab_id);
	ec3_identifier_from_string("_CHKDAT0", &cdat_id);
	ec3_identifier_from_string("_VOLSTR0", &stab_id);

	struct ec3_tag_ioctx *ctab, *cdat, *stab;
	status = ec3_image_ioctx_create_tag(
		image,
		EC3_TAG_CTAB,
		ctab_id,
		EC3_TAG_IO_READ | EC3_TAG_IO_WRITE,
		&ctab);

	if (status != EC3_SUCCESS) {
		b_err("cannot create chunk table tag");
		b_i("error code: %s", ec3_status_to_string(status));
		return -1;
	}

	status = ec3_image_ioctx_create_tag(
		image,
		EC3_TAG_CDAT,
		cdat_id,
		EC3_TAG_IO_READ | EC3_TAG_IO_WRITE,
		&cdat);

	if (status != EC3_SUCCESS) {
		b_err("cannot create chunk data tag");
		b_i("error code: %s", ec3_status_to_string(status));
		return -1;
	}

	status = ec3_image_ioctx_create_tag(
		image,
		EC3_TAG_STAB,
		stab_id,
		EC3_TAG_IO_WRITE | EC3_TAG_IO_SEQUENTIAL,
		&stab);

	if (status != EC3_SUCCESS) {
		b_err("cannot create string table tag");
		b_i("error code: %s", ec3_status_to_string(status));
		return -1;
	}

	struct capture_ctx ctx = {0};
	ctx.ctx_image = image;
	chunk_table_init(ctab, cdat, param.p_cluster_size, &ctx.ctx_chunks);
	chunk_table_init_empty_table(&ctx.ctx_chunks);
	string_table_init(&ctx.ctx_strings);

	uint64_t next_auto_id = 0;

	b_arglist_iterator it = {0};
	b_arglist_foreach_filtered(&it, opt, OPT_DIRECTORY, OPT_DIRECTORY_PATH)
	{
		printf("%s\n", it.value->val_str);

		status = capture_directory(
			&ctx,
			next_auto_id,
			it.value->val_str);
		next_auto_id++;

		if (status != EC3_SUCCESS) {
			b_err("an error occurred while writing to the "
			      "container");
			b_i("error: %s", ec3_status_to_string(status));
			return -1;
		}
	}

	for (size_t i = 0;; i++) {
		b_arglist_option *option = NULL;
		b_status err = b_arglist_get_option(
			opt,
			OPT_TAGGED_DIRECTORY,
			i,
			&option);
		if (!option) {
			break;
		}

		b_arglist_value *tag = NULL, *path = NULL;
		err = b_arglist_option_get_value(
			option,
			OPT_TAGGED_DIRECTORY_TAG,
			0,
			&tag);
		err = b_arglist_option_get_value(
			option,
			OPT_TAGGED_DIRECTORY_PATH,
			0,
			&path);

		printf("%s:%s\n", tag->val_str, path->val_str);

		uint64_t id = 0;
		status = ec3_identifier_from_string(tag->val_str, &id);

		if (status != EC3_SUCCESS) {
			b_err("'%s' is not a valid tag identifier", id);
			return -1;
		}

		status = capture_directory(&ctx, id, path->val_str);

		if (status != EC3_SUCCESS) {
			b_err("an error occurred while writing to the "
			      "container");
			return -1;
		}
	}

	b_btree_iterator s_it;
	b_btree_iterator_begin(&ctx.ctx_strings.s_offset_tree, &s_it);
	while (b_btree_iterator_is_valid(&s_it)) {
		struct string_table_entry *entry = b_unbox(
			struct string_table_entry,
			s_it.node,
			e_offset_node);

		size_t len = strlen(entry->e_str) + 1;
		size_t nr_written = 0;
		status = ec3_tag_ioctx_write(
			stab,
			entry->e_str,
			len,
			&nr_written);
		b_btree_iterator_next(&s_it);
	}

	string_table_finish(&ctx.ctx_strings);
	chunk_table_finish(&ctx.ctx_chunks);

	ec3_tag_ioctx_close(stab);
	ec3_tag_ioctx_close(ctab);
	ec3_tag_ioctx_close(cdat);

	ec3_image_ioctx_close(image);

	return 0;
}

B_COMMAND(CMD_CAPTURE, CMD_ROOT)
{
	B_COMMAND_NAME("capture");
	B_COMMAND_SHORT_NAME('Z');
	B_COMMAND_DESC(
		"capture one or more directories into an ec3 container. each "
		"directory specified will be stored in a separate volume "
		"within "
		"the created container.");
	B_COMMAND_FLAGS(B_COMMAND_SHOW_HELP_BY_DEFAULT);
	B_COMMAND_FUNCTION(capture);

	B_COMMAND_HELP_OPTION();

	B_COMMAND_OPTION(OPT_IDENT)
	{
		B_OPTION_SHORT_NAME('I');
		B_OPTION_LONG_NAME("ident");
		B_OPTION_DESC(
			"the string or number to use as the container "
			"identifier");

		B_OPTION_ARG(OPT_IDENT_VAL)
		{
			B_ARG_NAME("value");
			B_ARG_NR_VALUES(1);
		}
	}

	B_COMMAND_OPTION(OPT_OUTPATH)
	{
		B_OPTION_SHORT_NAME('o');
		B_OPTION_LONG_NAME("out");
		B_OPTION_DESC("the path to save the new file to.");

		B_OPTION_ARG(OPT_OUTPATH_PATH)
		{
			B_ARG_NAME("path");
			B_ARG_NR_VALUES(1);
		}
	}

	B_COMMAND_OPTION(OPT_DIRECTORY)
	{
		B_OPTION_SHORT_NAME('d');
		B_OPTION_LONG_NAME("directory");
		B_OPTION_DESC(
			"a directory to add to the container. a volume "
			"will be created "
			"within the container to store the specified "
			"directory.");

		B_OPTION_ARG(OPT_DIRECTORY_PATH)
		{
			B_ARG_NAME("path");
			B_ARG_NR_VALUES(1);
		}
	}

	B_COMMAND_OPTION(OPT_TAGGED_DIRECTORY)
	{
		B_OPTION_SHORT_NAME('D');
		B_OPTION_LONG_NAME("tagged-directory");
		B_OPTION_DESC(
			"a file to add to the container, with an associated "
			"tag. a disk "
			"image will be created within the container to store "
			"the specified "
			"directory. the tag must be either: (a) a 64-bit "
			"hexadecimal "
			"number; or (b) a string of no more than 8 "
			"characters.");

		B_OPTION_ARG(OPT_TAGGED_DIRECTORY_TAG)
		{
			B_ARG_NAME("tag");
			B_ARG_NR_VALUES(1);
		}

		B_OPTION_ARG(OPT_TAGGED_DIRECTORY_PATH)
		{
			B_ARG_NAME("path");
			B_ARG_NR_VALUES(1);
		}
	}

	B_COMMAND_OPTION(OPT_VERBOSE)
	{
		B_OPTION_SHORT_NAME('v');
		B_OPTION_LONG_NAME("verbose");
		B_OPTION_DESC(
			"show detailed output logs. this option can be "
			"specified multiple "
			"times to increase the level of output.");
	}

	B_COMMAND_USAGE()
	{
		B_COMMAND_USAGE_OPT(OPT_OUTPATH);
		B_COMMAND_USAGE_OPT(OPT_DIRECTORY);
	}

	B_COMMAND_USAGE()
	{
		B_COMMAND_USAGE_OPT(OPT_OUTPATH);
		B_COMMAND_USAGE_OPT(OPT_TAGGED_DIRECTORY);
	}
}