krh@0: #include <stdlib.h>
krh@0: #include <stdio.h>
krh@0: #include <string.h>
krh@0: #include <sys/types.h>
krh@0: #include <sys/stat.h>
krh@0: #include <sys/mman.h>
krh@0: #include <unistd.h>
krh@0: #include <fcntl.h>
krh@0: 
krh@0: #include <expat.h>
krh@0: #include "sha1.h"
krh@0: 
krh@0: static int
krh@0: write_to_fd(int fd, void *p, size_t size)
krh@0: {
krh@0: 	int rest, len;
krh@0: 
krh@0: 	rest = size;
krh@0: 	while (rest > 0) {
krh@0: 		len = write(fd, p, rest);
krh@0: 		if (len < 0)
krh@0: 			return -1;
krh@0: 		rest -= len;
krh@0: 	}
krh@0: 
krh@0: 	return 0;
krh@0: }
krh@0: 
krh@0: static int
krh@0: write_to_file(const char *filename, void *p, size_t size)
krh@0: {
krh@0: 	int fd, err;
krh@0: 
krh@0: 	fd = open(filename, O_CREAT | O_WRONLY | O_TRUNC, 0666);
krh@0: 	if (fd < 0)
krh@0: 		return -1;
krh@0: 	err = write_to_fd(fd, p, size);
krh@0: 	close(fd);
krh@0: 
krh@0: 	return err;
krh@0: }
krh@0: 
krh@0: static void *
krh@0: zalloc(size_t size)
krh@0: {
krh@0: 	void *p;
krh@0: 
krh@0: 	p = malloc(size);
krh@0: 	memset(p, 0, size);
krh@0: 
krh@0: 	return p;
krh@0: }
krh@0: 
krh@4: struct razor_set_header {
krh@4: 	unsigned int magic;
krh@4: 	unsigned int version;
krh@4: 	struct { unsigned int type, offset; } sections[0];
krh@4: };
krh@4: 
krh@4: #define RAZOR_MAGIC 0x7a7a7a7a
krh@4: #define RAZOR_VERSION 1
krh@4: #define RAZOR_BUCKETS 1
krh@4: #define RAZOR_STRINGS 2
krh@4: #define RAZOR_PACKAGES 3
krh@4: 
krh@4: struct razor_package {
krh@4: 	unsigned long name;
krh@4: 	unsigned long version;
krh@4: };
krh@4: 
krh@4: struct razor_set {
krh@4: 	unsigned long *buckets;
krh@4: 	int bucket_count, bucket_alloc;
krh@4: 	char *string_pool;
krh@4: 	int pool_size, pool_alloc;
krh@4: 	struct razor_set_header *header;
krh@4: 
krh@4: 	struct razor_package *packages;
krh@4: 	int package_count, package_alloc;
krh@4: };
krh@4: 
krh@4: struct razor_set *
krh@4: razor_set_create(void)
krh@0: {
krh@4: 	struct razor_set *set;
krh@0: 
krh@4: 	set = zalloc(sizeof *set);
krh@4: 	set->buckets = zalloc(4096 * sizeof *set->buckets);
krh@4: 	set->bucket_count = 0;
krh@4: 	set->bucket_alloc = 4096;
krh@0: 
krh@4: 	set->string_pool = zalloc(4096);
krh@4: 	set->pool_size = 1;
krh@4: 	set->pool_alloc = 4096;
krh@0: 
krh@4: 	set->packages = zalloc(4096 * sizeof *set->packages);
krh@4: 	set->package_count = 0;
krh@4: 	set->package_alloc = 4096;
krh@3: 
krh@4: 	return set;
krh@0: }
krh@0: 
krh@4: struct razor_set *
krh@4: razor_set_open(const char *filename)
krh@0: {
krh@4: 	struct razor_set *set;
krh@0: 	struct stat stat;
krh@0: 	unsigned int size, offset;
krh@0: 	int fd, i;
krh@0: 
krh@4: 	set = zalloc(sizeof *set);
krh@0: 	fd = open(filename, O_RDONLY);
krh@0: 	if (fstat(fd, &stat) < 0)
krh@0: 		return NULL;
krh@4: 	set->header = mmap(NULL, stat.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
krh@4: 	if (set->header == MAP_FAILED) {
krh@4: 		free(set);
krh@0: 		return NULL;
krh@0: 	}
krh@0: 
krh@4: 	for (i = 0; i < set->header->sections[i].type; i++) {
krh@4: 		offset = set->header->sections[i].offset;
krh@4: 		size = set->header->sections[i + 1].offset - offset;
krh@0: 
krh@4: 		switch (set->header->sections[i].type) {
krh@4: 		case RAZOR_BUCKETS:
krh@4: 			set->buckets = (void *) set->header + offset;
krh@4: 			set->bucket_count = size / sizeof *set->buckets;
krh@4: 			set->bucket_alloc = set->bucket_count;
krh@0: 			break;
krh@4: 		case RAZOR_STRINGS:
krh@4: 			set->string_pool = (void *) set->header + offset;
krh@4: 			set->pool_size = size;
krh@4: 			set->pool_alloc = size;
krh@0: 			break;
krh@4: 		case RAZOR_PACKAGES:
krh@4: 			set->packages = (void *) set->header + offset;
krh@4: 			set->package_count = size / sizeof *set->packages;
krh@4: 			set->package_alloc = size / sizeof *set->packages;
krh@3: 			break;
krh@0: 		}
krh@0: 	}
krh@0: 	close(fd);
krh@0: 
krh@4: 	return set;
krh@0: }
krh@0: 
krh@0: void
krh@4: razor_set_destroy(struct razor_set *set)
krh@0: {
krh@0: 	unsigned int size;
krh@0: 	int i;
krh@0: 
krh@4: 	if (set->header) {
krh@4: 		for (i = 0; set->header->sections[i].type; i++)
krh@0: 			;
krh@4: 		size = set->header->sections[i].type;
krh@4: 		munmap(set->header, size);
krh@0: 	} else {
krh@4: 		free(set->buckets);
krh@4: 		free(set->string_pool);
krh@0: 	}
krh@0: 
krh@4: 	free(set);
krh@0: }
krh@0: 
krh@0: static int
krh@4: razor_set_write(struct razor_set *set, const char *filename)
krh@0: {
krh@0: 	char data[4096];
krh@4: 	struct razor_set_header *header = (struct razor_set_header *) data;
krh@3: 	int fd, pool_size, package_size;
krh@3: 
krh@3: 	/* Align these to pages sizes */
krh@4: 	pool_size = (set->pool_size + 4095) & ~4095;
krh@3: 	package_size =
krh@4: 		(set->package_alloc * sizeof *set->packages + 4095) & ~4095;
krh@0: 
krh@0: 	memset(data, 0, sizeof data);
krh@4: 	header->magic = RAZOR_MAGIC;
krh@4: 	header->version = RAZOR_VERSION;
krh@0: 
krh@4: 	header->sections[0].type = RAZOR_BUCKETS;
krh@0: 	header->sections[0].offset = sizeof data;
krh@0: 
krh@4: 	header->sections[1].type = RAZOR_STRINGS;
krh@3: 	header->sections[1].offset = header->sections[0].offset +
krh@4: 		set->bucket_alloc * sizeof *set->buckets;
krh@0: 
krh@4: 	header->sections[2].type = RAZOR_PACKAGES;
krh@3: 	header->sections[2].offset = header->sections[1].offset + pool_size;
krh@3: 
krh@3: 	header->sections[3].type = 0;
krh@3: 	header->sections[3].offset = header->sections[2].offset + package_size;
krh@0: 
krh@0: 	fd = open(filename, O_CREAT | O_WRONLY | O_TRUNC, 0666);
krh@0: 	if (fd < 0)
krh@0: 		return -1;
krh@0: 
krh@0: 	write_to_fd(fd, data, sizeof data);
krh@4: 	write_to_fd(fd, set->buckets, set->bucket_alloc * sizeof *set->buckets);
krh@4: 	write_to_fd(fd, set->string_pool, pool_size);
krh@4: 	write_to_fd(fd, set->packages, package_size);
krh@0: 
krh@0: 	return 0;
krh@0: }
krh@0: 
krh@0: static unsigned int
krh@0: hash_string(const char *key)
krh@0: {
krh@0: 	const char *p;
krh@0: 	unsigned int hash = 0;
krh@0: 
krh@0: 	for (p = key; *p; p++)
krh@0: 		hash = (hash << 2) ^ *p;
krh@0: 
krh@0: 	return hash;
krh@0: }
krh@0: 
krh@0: unsigned long
krh@4: razor_set_lookup(struct razor_set *set, const char *key)
krh@0: {
krh@0: 	unsigned int start;
krh@0: 	unsigned int mask;
krh@0: 	unsigned long value;
krh@0: 	int i;
krh@0: 
krh@4: 	mask = set->bucket_alloc - 1;
krh@0: 	start = hash_string(key) & mask;
krh@0: 	i = start;
krh@0: 	do {
krh@4: 		value = set->buckets[i];
krh@0: 
krh@0: 		if (value == 0)
krh@0: 			return 0;
krh@0: 
krh@4: 		if (strcmp(key, &set->string_pool[value]) == 0)
krh@0: 			return value;
krh@0: 
krh@0: 		i = (i + 1) & mask;
krh@0: 	} while (i != start);
krh@0: 
krh@0: 	return 0;
krh@0: }
krh@0: 
krh@0: static unsigned long
krh@4: add_to_string_pool(struct razor_set *set, const char *key)
krh@0: {
krh@0: 	int len, alloc;
krh@0: 	char *pool;
krh@0: 	unsigned long value;
krh@0: 
krh@0: 	len = strlen(key) + 1;
krh@4: 	alloc = set->pool_alloc;
krh@4: 	while (alloc < set->pool_size + len)
krh@0: 		alloc *= 2;
krh@4: 	if (set->pool_alloc < alloc) {
krh@4: 		pool = realloc(set->string_pool, alloc);
krh@0: 		if (pool == NULL)
krh@0: 			return 0;
krh@4: 		set->string_pool = pool;
krh@4: 		set->pool_alloc = alloc;
krh@0: 	}
krh@0: 
krh@4: 	memcpy(set->string_pool + set->pool_size, key, len);
krh@4: 	value = set->pool_size;
krh@4: 	set->pool_size += len;
krh@0: 
krh@0: 	return value;
krh@0: }
krh@0: 
krh@0: static void
krh@4: do_insert(struct razor_set *set, unsigned long value)
krh@0: {
krh@0: 	unsigned int mask;
krh@0: 	const char *key;
krh@0: 	int i, start;
krh@0: 
krh@4: 	key = &set->string_pool[value];
krh@4: 	mask = set->bucket_alloc - 1;
krh@0: 	start = hash_string(key) & mask;
krh@0: 	i = start;
krh@0: 	do {
krh@4: 		if (set->buckets[i] == 0) {
krh@4: 			set->buckets[i] = value;
krh@0: 			break;
krh@0: 		}
krh@0: 		i = (i + 1) & mask;
krh@0: 	} while (i != start);
krh@0: }
krh@0: 
krh@0: unsigned long
krh@4: razor_set_insert(struct razor_set *set, const char *key)
krh@0: {
krh@0: 	unsigned long value, *buckets, *old_buckets;
krh@0: 	int i, alloc, old_alloc;
krh@0: 
krh@4: 	alloc = set->bucket_alloc;
krh@4: 	while (alloc < 4 * set->bucket_count)
krh@0: 		alloc *= 2;
krh@0: 
krh@4: 	if (alloc != set->bucket_alloc) {
krh@4: 		buckets = zalloc(alloc * sizeof *set->buckets);
krh@0: 		if (buckets == NULL)
krh@0: 			return 0;
krh@4: 		old_buckets = set->buckets;
krh@4: 		set->buckets = buckets;
krh@4: 		old_alloc = set->bucket_alloc;
krh@4: 		set->bucket_alloc = alloc;
krh@0: 		
krh@0: 		for (i = 0; i < old_alloc; i++) {
krh@0: 			value = old_buckets[i];
krh@0: 			if (value != 0)
krh@4: 				do_insert(set, value);
krh@0: 		}
krh@0: 		free(old_buckets);
krh@0: 	}
krh@0: 
krh@4: 	value = add_to_string_pool(set, key);
krh@4: 	do_insert (set, value);
krh@4: 	set->bucket_count++;
krh@0: 
krh@0: 	return value;
krh@0: }
krh@0: 
krh@3: static unsigned long
krh@4: razor_set_add_package(struct razor_set *set,
krh@3: 		      unsigned long name, unsigned long version)
krh@3: {
krh@4: 	struct razor_package *packages;
krh@3: 	int alloc;
krh@3: 
krh@4: 	/* FIXME: make 0 an illegal pkgs number. */
krh@4: 	alloc = set->package_alloc;
krh@4: 	while (alloc < set->package_count + 1)
krh@3: 		alloc *= 2;
krh@4: 	if (set->package_alloc < alloc) {
krh@4: 		packages = realloc(set->packages, alloc * sizeof set->packages);
krh@3: 		if (packages == NULL)
krh@3: 			return 0;
krh@4: 		set->packages = packages;
krh@4: 		set->package_alloc = alloc;
krh@3: 	}
krh@3: 
krh@4: 	set->packages[set->package_count].name = name;
krh@4: 	set->packages[set->package_count].version = version;
krh@3: 
krh@4: 	return set->package_count++;
krh@0: }
krh@0: 
krh@0: unsigned long
krh@4: razor_set_tokenize(struct razor_set *set, const char *string)
krh@0: {
krh@0: 	unsigned long token;
krh@0: 
krh@4: 	token = razor_set_lookup(set, string);
krh@0: 	if (token != 0)
krh@0: 		return token;
krh@0: 
krh@4: 	return razor_set_insert(set, string);
krh@0: }
krh@0: 
krh@4: static struct razor_set *qsort_set;
krh@3: 
krh@3: static int
krh@3: compare_packages(const void *p1, const void *p2)
krh@3: {
krh@4: 	const struct razor_package *pkg1 = p1, *pkg2 = p2;
krh@3: 
krh@4: 	return strcmp(&qsort_set->string_pool[pkg1->name],
krh@4: 		      &qsort_set->string_pool[pkg2->name]);
krh@3: }
krh@3: 
krh@3: static void
krh@4: razor_set_sort(struct razor_set *set)
krh@3: {
krh@4: 	qsort_set = set;
krh@4: 	qsort(set->packages, set->package_count, sizeof *set->packages,
krh@3: 	      compare_packages);
krh@3: }
krh@3: 
krh@0: struct parsing_context {
krh@4: 	struct razor_set *set;
krh@4: 	int pkg_id;
krh@0: };
krh@0: 
krh@0: static void
krh@3: parse_package(struct parsing_context *ctx, const char **atts)
krh@3: {
krh@3: 	unsigned long name, version;
krh@3: 	int i;
krh@3: 
krh@3: 	for (i = 0; atts[i]; i += 2) {
krh@3: 		if (strcmp(atts[i], "name") == 0)
krh@4: 			name = razor_set_tokenize(ctx->set, atts[i + 1]);
krh@3: 		else if (strcmp(atts[i], "version") == 0)
krh@4: 			version = razor_set_tokenize(ctx->set, atts[i + 1]);
krh@3: 	}
krh@3: 
krh@3: 	if (name == 0 || version == 0) {
krh@3: 		fprintf(stderr, "invalid package tag, "
krh@3: 			"missing name or version attributes\n");
krh@3: 		return;
krh@3: 	}
krh@3: 
krh@4: 	ctx->pkg_id = razor_set_add_package(ctx->set, name, version);
krh@4: 
krh@4: 	return;
krh@3: }
krh@3: 
krh@3: static void
krh@0: start_element(void *data, const char *name, const char **atts)
krh@0: {
krh@0: 	struct parsing_context *ctx = data;
krh@0: 	int i;
krh@0: 
krh@3: 	if (strcmp(name, "package") == 0)
krh@3: 		parse_package(ctx, atts);
krh@3: 
krh@0: 	for (i = 0; atts[i]; i += 2)
krh@4: 		razor_set_tokenize(ctx->set, atts[i + 1]);
krh@0: }
krh@0: 
krh@0: static void
krh@0: end_element (void *data, const char *name)
krh@0: {
krh@4: 	struct parsing_context *ctx = data;
krh@4: 
krh@4: 	if (strcmp(name, "package") == 0)
krh@4: 		ctx->pkg_id = 0;
krh@0: }
krh@0: 
krh@0: static char *
krh@0: sha1_to_hex(const unsigned char *sha1)
krh@0: {
krh@0: 	static int bufno;
krh@0: 	static char hexbuffer[4][50];
krh@0: 	static const char hex[] = "0123456789abcdef";
krh@0: 	char *buffer = hexbuffer[3 & ++bufno], *buf = buffer;
krh@0: 	int i;
krh@0: 
krh@0: 	for (i = 0; i < 20; i++) {
krh@0: 		unsigned int val = *sha1++;
krh@0: 		*buf++ = hex[val >> 4];
krh@0: 		*buf++ = hex[val & 0xf];
krh@0: 	}
krh@0: 	*buf = '\0';
krh@0: 
krh@0: 	return buffer;
krh@0: }
krh@0: 
krh@0: static int
krh@4: razor_set_import(struct razor_set *set, const char *filename)
krh@0: {
krh@0: 	SHA_CTX sha1;
krh@0: 	XML_Parser parser;
krh@4: 	struct parsing_context ctx;
krh@0: 	int fd;
krh@0: 	void *p;
krh@0: 	struct stat stat;
krh@0: 	char buf[128];
krh@0: 	unsigned char hash[20];
krh@0: 
krh@0: 	fd = open(filename, O_RDONLY);
krh@0: 	if (fstat(fd, &stat) < 0)
krh@0: 		return -1;
krh@0: 	p = mmap(NULL, stat.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
krh@0: 	if (p == MAP_FAILED)
krh@0: 		return -1;
krh@0: 
krh@0: 	parser = XML_ParserCreate(NULL);
krh@4: 	ctx.set = set;
krh@4: 	XML_SetUserData(parser, &ctx);
krh@0: 	XML_SetElementHandler(parser, start_element, end_element);
krh@0: 	if (XML_Parse(parser, p, stat.st_size, 1) == XML_STATUS_ERROR) {
krh@0: 		fprintf(stderr,
krh@0: 			"%s at line %d, %s\n",
krh@0: 			XML_ErrorString(XML_GetErrorCode(parser)),
krh@0: 			XML_GetCurrentLineNumber(parser),
krh@0: 			filename);
krh@0: 		return 1;
krh@0: 	}
krh@0: 
krh@0: 	XML_ParserFree(parser);
krh@0: 
krh@0: 	SHA1_Init(&sha1);
krh@0: 	SHA1_Update(&sha1, p, stat.st_size);
krh@0: 	SHA1_Final(hash, &sha1);
krh@0: 
krh@0: 	close(fd);
krh@0: 
krh@0: 	snprintf(buf, sizeof buf, "set/%s", sha1_to_hex(hash));
krh@0: 	if (write_to_file(buf, p, stat.st_size) < 0)
krh@0: 		return -1;
krh@0: 	munmap(p, stat.st_size);
krh@0: 
krh@0: 	return 0;
krh@0: }
krh@0: 
krh@0: void
krh@4: razor_set_list(struct razor_set *set)
krh@3: {
krh@3: 	int i;
krh@4: 	struct razor_package *p;
krh@3: 
krh@4: 	p = set->packages;
krh@4: 	for (i = 0; i < set->package_count && p->name; i++, p++) {
krh@3: 		printf("%s %s\n",
krh@4: 		       &set->string_pool[p->name],
krh@4: 		       &set->string_pool[p->version]);
krh@3: 	}
krh@3: }
krh@3: 
krh@3: void
krh@4: razor_set_info(struct razor_set *set)
krh@3: {
krh@3: 	unsigned int offset, size;
krh@3: 	int i;
krh@3: 
krh@4: 	for (i = 0; i < set->header->sections[i].type; i++) {
krh@4: 		offset = set->header->sections[i].offset;
krh@4: 		size = set->header->sections[i + 1].offset - offset;
krh@3: 
krh@4: 		switch (set->header->sections[i].type) {
krh@4: 		case RAZOR_BUCKETS:
krh@3: 			printf("bucket section:\t\t%dkb\n", size / 1024);
krh@3: 			break;
krh@4: 		case RAZOR_STRINGS:
krh@3: 			printf("string pool:\t\t%dkb\n", size / 1024);
krh@3: 			break;
krh@4: 		case RAZOR_PACKAGES:
krh@3: 			printf("package section:\t%dkb\n", size / 1024);
krh@3: 			break;
krh@3: 		}
krh@3: 	}
krh@0: }
krh@0: 
krh@0: static int
krh@0: usage(void)
krh@0: {
krh@3: 	printf("usage: razor [ import FILES | lookup <key> | list | info ]\n");
krh@0: 	exit(1);
krh@0: }
krh@0: 
krh@0: static const char repo_filename[] = "system.repo";
krh@0: 
krh@0: int
krh@0: main(int argc, char *argv[])
krh@0: {
krh@0: 	int i;
krh@4: 	struct razor_set *set;
krh@0: 	struct stat statbuf;
krh@0: 
krh@3: 	if (argc < 2) {
krh@0: 		usage();
krh@0: 	} else if (strcmp(argv[1], "import") == 0) {
krh@0: 		if (stat("set", &statbuf) && mkdir("set", 0777)) {
krh@0: 			fprintf(stderr, "could not create directory 'set'\n");
krh@0: 			exit(-1);
krh@0: 		}
krh@0: 			
krh@4: 		set = razor_set_create();
krh@0: 
krh@0: 		for (i = 2; i < argc; i++) {
krh@4: 			if (razor_set_import(set, argv[i]) < 0) {
krh@0: 				fprintf(stderr, "failed to import %s\n",
krh@0: 					argv[i]);
krh@0: 				exit(-1);
krh@0: 			}
krh@0: 		}
krh@0: 
krh@4: 		razor_set_sort(set);
krh@3: 
krh@0: 		printf("number of buckets: %d\n",
krh@4: 		       set->bucket_count);
krh@0: 		printf("bucket allocation: %d\n",
krh@4: 		       set->bucket_alloc);
krh@4: 		printf("pool size: %d\n", set->pool_size);
krh@4: 		printf("pool allocation: %d\n", set->pool_alloc);
krh@0: 
krh@4: 		razor_set_write(set, repo_filename);
krh@0: 
krh@4: 		razor_set_destroy(set);
krh@0: 	} else if (strcmp(argv[1], "lookup") == 0) {
krh@4: 		set = razor_set_open(repo_filename);
krh@0: 		printf("%s is %lu\n", argv[2],
krh@4: 		       razor_set_lookup(set, argv[2]));
krh@4: 		razor_set_destroy(set);
krh@3: 	} else if (strcmp(argv[1], "list") == 0) {
krh@4: 		set = razor_set_open(repo_filename);
krh@4: 		razor_set_list(set);
krh@4: 		razor_set_destroy(set);
krh@3: 	} else if (strcmp(argv[1], "info") == 0) {
krh@4: 		set = razor_set_open(repo_filename);
krh@4: 		razor_set_info(set);
krh@4: 		razor_set_destroy(set);
krh@0: 	} else {
krh@0: 		usage();
krh@0: 	}
krh@0: 
krh@0: 	return 0;
krh@0: }