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: struct hashtable_header {
krh@0: 	unsigned int magic;
krh@0: 	unsigned int version;
krh@0: 	struct { unsigned int type, offset; } sections[0];
krh@0: };
krh@0: 
krh@0: #define HASHTABLE_MAGIC 0x7a7a7a7a
krh@0: #define HASHTABLE_VERSION 1
krh@0: #define HASHTABLE_BUCKETS 1
krh@0: #define HASHTABLE_STRINGS 2
krh@0: 
krh@0: struct hashtable {
krh@0: 	unsigned long *buckets;
krh@0: 	int bucket_count, bucket_alloc;
krh@0: 	char *string_pool;
krh@0: 	int pool_size, pool_alloc;
krh@0: 	struct hashtable_header *header;
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@0: struct hashtable *
krh@0: hashtable_create(void)
krh@0: {
krh@0: 	struct hashtable *ht;
krh@0: 
krh@0: 	ht = zalloc(sizeof *ht);
krh@0: 	ht->buckets = zalloc(4096 * sizeof *ht->buckets);
krh@0: 	ht->bucket_count = 0;
krh@0: 	ht->bucket_alloc = 4096;
krh@0: 
krh@0: 	ht->string_pool = zalloc(4096);
krh@0: 	ht->pool_size = 1;
krh@0: 	ht->pool_alloc = 4096;
krh@0: 
krh@0: 	return ht;
krh@0: }
krh@0: 
krh@0: struct hashtable *
krh@0: hashtable_create_from_file(const char *filename)
krh@0: {
krh@0: 	struct hashtable *ht;
krh@0: 	struct stat stat;
krh@0: 	unsigned int size, offset;
krh@0: 	int fd, i;
krh@0: 
krh@0: 	ht = zalloc(sizeof *ht);
krh@0: 	fd = open(filename, O_RDONLY);
krh@0: 	if (fstat(fd, &stat) < 0)
krh@0: 		return NULL;
krh@0: 	ht->header = mmap(NULL, stat.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
krh@0: 	if (ht->header == MAP_FAILED) {
krh@0: 		free(ht);
krh@0: 		return NULL;
krh@0: 	}
krh@0: 
krh@0: 	for (i = 0; i < ht->header->sections[i].type; i++) {
krh@0: 		offset = ht->header->sections[i].offset;
krh@0: 		size = ht->header->sections[i + 1].offset - offset;
krh@0: 
krh@0: 		switch (ht->header->sections[i].type) {
krh@0: 		case HASHTABLE_BUCKETS:
krh@0: 			ht->buckets = (void *) ht->header + offset;
krh@0: 			ht->bucket_count = size / sizeof *ht->buckets;
krh@0: 			ht->bucket_alloc = ht->bucket_count;
krh@0: 			break;
krh@0: 		case HASHTABLE_STRINGS:
krh@0: 			ht->string_pool = (void *) ht->header + offset;
krh@0: 			ht->pool_size = size;
krh@0: 			ht->pool_alloc = size;
krh@0: 			break;
krh@0: 		}
krh@0: 	}
krh@0: 	close(fd);
krh@0: 
krh@0: 	return ht;
krh@0: }
krh@0: 
krh@0: void
krh@0: hashtable_destroy(struct hashtable *ht)
krh@0: {
krh@0: 	unsigned int size;
krh@0: 	int i;
krh@0: 
krh@0: 	if (ht->header) {
krh@0: 		for (i = 0; ht->header->sections[i].type; i++)
krh@0: 			;
krh@0: 		size = ht->header->sections[i].type;
krh@0: 		munmap(ht->header, size);
krh@0: 	} else {
krh@0: 		free(ht->buckets);
krh@0: 		free(ht->string_pool);
krh@0: 	}
krh@0: 
krh@0: 	free(ht);
krh@0: }
krh@0: 
krh@0: static int
krh@0: hashtable_write(struct hashtable *ht, const char *filename)
krh@0: {
krh@0: 	int fd;
krh@0: 	char data[4096];
krh@0: 	struct hashtable_header *header = (struct hashtable_header *) data;
krh@0: 
krh@0: 	memset(data, 0, sizeof data);
krh@0: 	header->magic = HASHTABLE_MAGIC;
krh@0: 	header->version = HASHTABLE_VERSION;
krh@0: 
krh@0: 	header->sections[0].type = HASHTABLE_BUCKETS;
krh@0: 	header->sections[0].offset = sizeof data;
krh@0: 
krh@0: 	header->sections[1].type = HASHTABLE_STRINGS;
krh@0: 	header->sections[1].offset =
krh@0: 		sizeof data + ht->bucket_alloc * sizeof *ht->buckets;
krh@0: 
krh@0: 	header->sections[2].type = 0;
krh@0: 	header->sections[2].offset =
krh@0: 		header->sections[1].offset + ht->pool_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@0: 	write_to_fd(fd, ht->buckets, ht->bucket_alloc * sizeof *ht->buckets);
krh@0: 	write_to_fd(fd, ht->string_pool, ht->pool_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@0: hashtable_lookup(struct hashtable *ht, 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@0: 	mask = ht->bucket_alloc - 1;
krh@0: 	start = hash_string(key) & mask;
krh@0: 	i = start;
krh@0: 	do {
krh@0: 		value = ht->buckets[i];
krh@0: 
krh@0: 		if (value == 0)
krh@0: 			return 0;
krh@0: 
krh@0: 		if (strcmp(key, &ht->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@0: add_to_string_pool(struct hashtable *ht, 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@0: 	alloc = ht->pool_alloc;
krh@0: 	while (alloc < ht->pool_size + len)
krh@0: 		alloc *= 2;
krh@0: 	if (ht->pool_alloc < alloc) {
krh@0: 		pool = realloc(ht->string_pool, alloc);
krh@0: 		if (pool == NULL)
krh@0: 			return 0;
krh@0: 		ht->string_pool = pool;
krh@0: 		ht->pool_alloc = alloc;
krh@0: 	}
krh@0: 
krh@0: 	memcpy(ht->string_pool + ht->pool_size, key, len);
krh@0: 	value = ht->pool_size;
krh@0: 	ht->pool_size += len;
krh@0: 
krh@0: 	return value;
krh@0: }
krh@0: 
krh@0: static void
krh@0: do_insert(struct hashtable *ht, unsigned long value)
krh@0: {
krh@0: 	unsigned int mask;
krh@0: 	const char *key;
krh@0: 	int i, start;
krh@0: 
krh@0: 	key = &ht->string_pool[value];
krh@0: 	mask = ht->bucket_alloc - 1;
krh@0: 	start = hash_string(key) & mask;
krh@0: 	i = start;
krh@0: 	do {
krh@0: 		if (ht->buckets[i] == 0) {
krh@0: 			ht->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@0: hashtable_insert(struct hashtable *ht, const char *key)
krh@0: {
krh@0: 	unsigned long value, *buckets, *old_buckets;
krh@0: 	int i, alloc, old_alloc;
krh@0: 
krh@0: 	alloc = ht->bucket_alloc;
krh@0: 	while (alloc < 4 * ht->bucket_count)
krh@0: 		alloc *= 2;
krh@0: 
krh@0: 	if (alloc != ht->bucket_alloc) {
krh@0: 		buckets = zalloc(alloc * sizeof *ht->buckets);
krh@0: 		if (buckets == NULL)
krh@0: 			return 0;
krh@0: 		old_buckets = ht->buckets;
krh@0: 		ht->buckets = buckets;
krh@0: 		old_alloc = ht->bucket_alloc;
krh@0: 		ht->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@0: 				do_insert(ht, value);
krh@0: 		}
krh@0: 		free(old_buckets);
krh@0: 	}
krh@0: 
krh@0: 	value = add_to_string_pool(ht, key);
krh@0: 	do_insert (ht, value);
krh@0: 	ht->bucket_count++;
krh@0: 
krh@0: 	return value;
krh@0: }
krh@0: 
krh@0: struct razor_context {
krh@0: 	struct hashtable *global_ht;
krh@0: };
krh@0: 
krh@0: struct razor_context *
krh@0: razor_context_create (void)
krh@0: {
krh@0: 	struct razor_context *ctx;
krh@0: 
krh@0: 	ctx = malloc(sizeof *ctx);
krh@0: 	ctx->global_ht = hashtable_create();
krh@0: 
krh@0: 	return ctx;
krh@0: }
krh@0: 
krh@0: struct razor_context *
krh@0: razor_context_create_from_file (const char *filename)
krh@0: {
krh@0: 	struct razor_context *ctx;
krh@0: 
krh@0: 	ctx = malloc(sizeof *ctx);
krh@0: 	ctx->global_ht = hashtable_create_from_file(filename);
krh@0: 
krh@0: 	return ctx;
krh@0: }
krh@0: 
krh@0: unsigned long
krh@0: razor_context_tokenize(struct razor_context *ctx, const char *string)
krh@0: {
krh@0: 	unsigned long token;
krh@0: 
krh@0: 	token = hashtable_lookup(ctx->global_ht, string);
krh@0: 	if (token != 0)
krh@0: 		return token;
krh@0: 
krh@0: 	return hashtable_insert(ctx->global_ht, string);
krh@0: }
krh@0: 
krh@0: struct razor_set {
krh@0: 	struct razor_context *ctx;
krh@0: };
krh@0: 
krh@0: struct parsing_context {
krh@0: 	struct razor_context *ctx;
krh@0: };
krh@0: 
krh@0: 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@0: 	for (i = 0; atts[i]; i += 2)
krh@0: 		razor_context_tokenize(ctx->ctx, atts[i + 1]);
krh@0: }
krh@0: 
krh@0: static void
krh@0: end_element (void *data, const char *name)
krh@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@0: razor_context_read_file(struct razor_context *ctx, const char *filename)
krh@0: {
krh@0: 	SHA_CTX sha1;
krh@0: 	XML_Parser parser;
krh@0: 	struct parsing_context pctx;
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@0: 	pctx.ctx = ctx;
krh@0: 	XML_SetUserData(parser, &pctx);
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: int
krh@0: razor_context_write(struct razor_context *ctx, const char *filename)
krh@0: {
krh@0: 	return hashtable_write(ctx->global_ht, filename);
krh@0: }
krh@0: 
krh@0: void
krh@0: razor_context_destroy(struct razor_context *ctx)
krh@0: {
krh@0: 	hashtable_destroy(ctx->global_ht);
krh@0: 	free(ctx);
krh@0: }
krh@0: 
krh@0: static int
krh@0: usage(void)
krh@0: {
krh@0: 	printf("usage: razor [ import FILES | lookup <key> ]\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@0: 	struct razor_context *ctx;
krh@0: 	struct stat statbuf;
krh@0: 
krh@0: 	if (argc < 3) {
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@0: 		ctx = razor_context_create();
krh@0: 
krh@0: 		for (i = 2; i < argc; i++) {
krh@0: 			if (razor_context_read_file(ctx, 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@0: 		printf("number of buckets: %d\n",
krh@0: 		       ctx->global_ht->bucket_count);
krh@0: 		printf("bucket allocation: %d\n",
krh@0: 		       ctx->global_ht->bucket_alloc);
krh@0: 		printf("pool size: %d\n", ctx->global_ht->pool_size);
krh@0: 		printf("pool allocation: %d\n", ctx->global_ht->pool_alloc);
krh@0: 
krh@0: 		razor_context_write(ctx, repo_filename);
krh@0: 
krh@0: 		razor_context_destroy(ctx);
krh@0: 	} else if (strcmp(argv[1], "lookup") == 0) {
krh@0: 		ctx = razor_context_create_from_file(repo_filename);
krh@0: 		printf("%s is %lu\n", argv[2],
krh@0: 		       hashtable_lookup(ctx->global_ht, argv[2]));
krh@0: 		razor_context_destroy(ctx);
krh@0: 	} else {
krh@0: 		usage();
krh@0: 	}
krh@0: 
krh@0: 	return 0;
krh@0: }