Non-standard string functions used throughout Tor. More...

#include "lib/string/util_string.h"
#include "lib/string/compat_ctype.h"
#include "lib/err/torerr.h"
#include "lib/ctime/di_ops.h"
#include "lib/defs/digest_sizes.h"
#include <string.h>
#include <stdlib.h>

Macros
#define	TOP_BITS(x) ((uint8_t)(0xFF << (8 - (x))))

#define	LOW_BITS(x) ((uint8_t)(0xFF >> (8 - (x))))

Functions
const void *	tor_memmem (const void _haystack, size_t hlen, const void _needle, size_t nlen)

const void *	tor_memstr (const void haystack, size_t hlen, const char needle)

int	fast_mem_is_zero (const char *mem, size_t len)

int	tor_digest_is_zero (const char *digest)

int	tor_digest256_is_zero (const char *digest)

void	tor_strstrip (char s, const char strip)

void	tor_strlower (char *s)

void	tor_strupper (char *s)

void	tor_strreplacechar (char *s, char find, char replacement)

int	tor_strisprint (const char *s)

int	tor_strisnonupper (const char *s)

int	tor_strisspace (const char *s)

int	strcmp_opt (const char s1, const char s2)

int	strcmpstart (const char s1, const char s2)

int	strcasecmpstart (const char s1, const char s2)

int	fast_memcmpstart (const void mem, size_t memlen, const char prefix)

int	strcmpend (const char s1, const char s2)

int	strcasecmpend (const char s1, const char s2)

const char *	eat_whitespace (const char *s)

const char *	eat_whitespace_eos (const char s, const char eos)

const char *	eat_whitespace_no_nl (const char *s)

const char *	eat_whitespace_eos_no_nl (const char s, const char eos)

const char *	find_whitespace (const char *s)

const char *	find_whitespace_eos (const char s, const char eos)

const char *	find_str_at_start_of_line (const char haystack, const char needle)

int	string_is_C_identifier (const char *string)

static uint8_t	bytes_in_char (uint8_t b)

static bool	is_continuation_byte (uint8_t b)

static bool	validate_char (const uint8_t *c, uint8_t len)

int	string_is_utf8 (const char *str, size_t len)

int	string_is_utf8_no_bom (const char *str, size_t len)

Detailed Description

Non-standard string functions used throughout Tor.

Definition in file util_string.c.

Macro Definition Documentation

◆ LOW_BITS

#define LOW_BITS ( x ) ((uint8_t)(0xFF >> (8 - (x))))

A byte with the lowest x bits set.

Definition at line 449 of file util_string.c.

◆ TOP_BITS

#define TOP_BITS ( x ) ((uint8_t)(0xFF << (8 - (x))))

A byte with the top x bits set.

Definition at line 447 of file util_string.c.

Function Documentation

◆ bytes_in_char()

static uint8_t bytes_in_char ( uint8_t b )

static

Given the leading byte b, return the total number of bytes in the UTF-8 character. Returns 0 if it's an invalid leading byte.

Definition at line 455 of file util_string.c.

Referenced by string_is_utf8().

◆ eat_whitespace()

const char* eat_whitespace ( const char * s )

Return a pointer to the first char of s that is not whitespace and not a comment, or to the terminating NUL if no such character exists.

Definition at line 277 of file util_string.c.

Referenced by parse_log_severity_config(), port_cfg_line_extract_addrport(), router_parse_addr_policy_item_from_string(), router_parse_addr_policy_private(), tor_addr_port_lookup(), and tor_version_parse_platform().

◆ eat_whitespace_eos()

const char* eat_whitespace_eos	(	const char *	s,
		const char *	eos
	)

Return a pointer to the first char of s that is not whitespace and not a comment, or to the terminating NUL if no such character exists.

Definition at line 304 of file util_string.c.

Referenced by authority_cert_parse_from_string(), find_start_of_next_microdesc(), find_start_of_next_router_or_extrainfo(), microdescs_parse_from_string(), and pem_decode().

◆ eat_whitespace_eos_no_nl()

const char* eat_whitespace_eos_no_nl	(	const char *	s,
		const char *	eos
	)

As eat_whitespace_no_nl, but stop at eos whether we have found a non-whitespace character or not.

Definition at line 342 of file util_string.c.

◆ eat_whitespace_no_nl()

const char* eat_whitespace_no_nl ( const char * s )

Return a pointer to the first char of s that is not a space or a tab or a \r, or to the terminating NUL if no such character exists.

Definition at line 332 of file util_string.c.

Referenced by parse_http_command().

◆ fast_mem_is_zero()

int fast_mem_is_zero	(	const char *	mem,
		size_t	len
	)

Return true iff the 'len' bytes at 'mem' are all zero.

Definition at line 74 of file util_string.c.

Referenced by build_service_desc_keys(), build_service_desc_plaintext(), check_sockaddr(), commit_has_reveal_value(), construct_ntor_key_map(), crypto_read_tagged_contents_from_file(), decrypt_descriptor_cookie(), dirauth_get_voting_schedule(), disk_state_put_commit_line(), hs_build_blinded_keypair(), hs_build_blinded_pubkey(), hs_desc_build_authorized_client(), microdesc_has_curve25519_onion_key(), and routerinfo_incompatible_with_extrainfo().

◆ fast_memcmpstart()

int fast_memcmpstart	(	const void *	mem,
		size_t	memlen,
		const char *	prefix
	)

Compare the value of the string prefix with the start of the memlen-byte memory chunk at mem. Return as for strcmp.

[As fast_memcmp(mem, prefix, strlen(prefix)) but returns -1 if memlen is less than strlen(prefix).]

Definition at line 238 of file util_string.c.

◆ find_str_at_start_of_line()

const char* find_str_at_start_of_line	(	const char *	haystack,
		const char *	needle
	)

Return the first occurrence of needle in haystack that occurs at the start of a line (that is, at the beginning of haystack or immediately after a newline). Return NULL if no such string is found.

Definition at line 400 of file util_string.c.

Referenced by get_total_system_memory_impl().

◆ find_whitespace()

const char* find_whitespace ( const char * s )

Return a pointer to the first char of s that is whitespace or #, or to the terminating NUL if no such character exists.

Definition at line 353 of file util_string.c.

Referenced by parse_http_command(), tor_version_parse_platform(), and version_from_platform().

◆ find_whitespace_eos()

const char* find_whitespace_eos	(	const char *	s,
		const char *	eos
	)

As find_whitespace, but stop at eos whether we have found a whitespace or not.

Definition at line 375 of file util_string.c.

◆ is_continuation_byte()

static bool is_continuation_byte ( uint8_t b )

static

Returns true iff b is a UTF-8 continuation byte.

Definition at line 472 of file util_string.c.

Referenced by validate_char().

◆ strcasecmpend()

int strcasecmpend	(	const char *	s1,
		const char *	s2
	)

Compares the last strlen(s2) characters of s1 with s2. Returns as for strcasecmp.

Definition at line 264 of file util_string.c.

Referenced by address_is_in_virtual_range(), addressmap_address_should_automap(), hostname_in_track_host_exits(), tor_addr_hostname_is_local(), and tor_addr_parse_PTR_name().

◆ strcasecmpstart()

int strcasecmpstart	(	const char *	s1,
		const char *	s2
	)

Compares the first strlen(s2) characters of s1 with s2. Returns as for strcasecmp.

Definition at line 225 of file util_string.c.

Referenced by circuit_purpose_from_string(), config_lines_dup_and_filter(), handle_control_hspost(), http_get_header(), parse_dirauth_dirport(), and tor_version_parse().

◆ strcmp_opt()

int strcmp_opt	(	const char *	s1,
		const char *	s2
	)

As strcmp, except that either string may be NULL. The NULL string is considered to be before any non-NULL string.

Definition at line 197 of file util_string.c.

Referenced by bridge_resolve_conflicts(), compare_by_valid_after_(), connection_dir_list_by_purpose_and_resource(), connection_dir_list_by_purpose_resource_and_state(), opt_streq(), and proxy_needs_restart().

◆ strcmpend()

int strcmpend	(	const char *	s1,
		const char *	s2
	)

Compares the last strlen(s2) characters of s1 with s2. Returns as for strcmp.

Definition at line 251 of file util_string.c.

Referenced by auth_key_filename_is_valid(), client_filename_is_valid(), crypto_read_tagged_contents_from_file(), get_onion_public_key(), and storage_dir_clean_tmpfiles().

◆ strcmpstart()

int strcmpstart	(	const char *	s1,
		const char *	s2
	)

Compares the first strlen(s2) characters of s1 with s2. Returns as for strcmp.

Definition at line 215 of file util_string.c.

Referenced by address_is_invalid_mapaddress_target(), body_is_plausible(), connection_dir_bridge_routerdesc_failed(), controller_get_routerdescs(), crypto_read_tagged_contents_from_file(), decode_hashed_passwords(), dirvote_get_intermediate_param_value(), evdns_log_cb(), extract_param_buggy(), find_start_of_next_router_or_extrainfo(), get_net_param_from_list(), get_transport_bindaddr(), getinfo_helper_dir(), getinfo_helper_downloads(), getinfo_helper_downloads_cert(), getinfo_helper_geoip(), handle_get_descriptor(), handle_get_hs_descriptor_v3(), handle_get_keys(), handle_getinfo_helper(), handle_response_fetch_desc(), list_pending_fpsk_downloads(), metrics_connection_process_inbuf(), parse_dirauth_dirport(), parse_http_command(), parse_http_url(), parse_openssl_version_str(), port_cfg_line_extract_addrport(), router_parse_addr_policy_private(), string_is_utf8_no_bom(), summarize_protover_flags(), tor_version_parse_platform(), and version_from_platform().

◆ string_is_C_identifier()

int string_is_C_identifier ( const char * string )

Returns true if string could be a C identifier. A C identifier must begin with a letter or an underscore and the rest of its characters can be letters, numbers or underscores. No length limit is imposed.

Definition at line 423 of file util_string.c.

◆ string_is_utf8()

int string_is_utf8	(	const char *	str,
		size_t	len
	)

Returns true iff the first len bytes in str are a valid UTF-8 string.

Definition at line 516 of file util_string.c.

Referenced by string_is_utf8_no_bom().

◆ string_is_utf8_no_bom()

int string_is_utf8_no_bom	(	const char *	str,
		size_t	len
	)

As string_is_utf8(), but returns false if the string begins with a UTF-8 byte order mark (BOM).

Definition at line 557 of file util_string.c.

◆ tor_digest256_is_zero()

int tor_digest256_is_zero ( const char * digest )

Return true iff the DIGEST256_LEN bytes in digest are all zero.

Definition at line 103 of file util_string.c.

◆ tor_digest_is_zero()

int tor_digest_is_zero ( const char * digest )

Return true iff the DIGEST_LEN bytes in digest are all zero.

Definition at line 96 of file util_string.c.

Referenced by bridge_get_rsa_id_digest(), bridge_has_digest(), bridge_resolve_conflicts(), channel_dump_statistics(), circuit_get_all_pending_on_channel(), connection_ap_fail_onehop(), connection_or_client_learned_peer_id(), connection_unlink(), directory_request_dir_contact_info_specified(), entry_guard_learned_bridge_identity(), get_bridge_info_for_guard(), get_configured_bridge_by_addr_port_digest(), get_configured_bridge_by_orports_digest(), get_or_history(), learned_bridge_descriptor(), learned_router_identity(), lookup_cached_dir_by_fp(), and orconn_target_get_name().

◆ tor_memmem()

const void* tor_memmem	(	const void *	_haystack,
		size_t	hlen,
		const void *	_needle,
		size_t	nlen
	)

Given hlen bytes at haystack and nlen bytes at needle, return a pointer to the first occurrence of the needle within the haystack, or NULL if there is no such occurrence.

This function is not timing-safe.

Requires that nlen be greater than zero.

Definition at line 29 of file util_string.c.

◆ tor_strisnonupper()

int tor_strisnonupper ( const char * s )

Return 1 if no character in s is uppercase, else return 0.

Definition at line 171 of file util_string.c.

Referenced by assert_resolve_ok().

◆ tor_strisprint()

int tor_strisprint ( const char * s )

Return 1 if every character in s is printable, else return 0.

Definition at line 158 of file util_string.c.

◆ tor_strisspace()

int tor_strisspace ( const char * s )

Return true iff every character in s is whitespace space; else return false.

Definition at line 184 of file util_string.c.

◆ tor_strlower()

void tor_strlower ( char * s )

Convert all alphabetic characters in the nul-terminated string s to lowercase.

Definition at line 127 of file util_string.c.

Referenced by clear_trackexithost_mappings().

◆ tor_strreplacechar()

void tor_strreplacechar	(	char *	s,
		char	find,
		char	replacement
	)

Replaces old with replacement in s

Definition at line 148 of file util_string.c.

◆ tor_strstrip()

void tor_strstrip	(	char *	s,
		const char *	strip
	)

Remove from the string s every character which appears in strip.

Definition at line 111 of file util_string.c.

◆ tor_strupper()

void tor_strupper ( char * s )

Convert all alphabetic characters in the nul-terminated string s to lowercase.

Definition at line 138 of file util_string.c.

◆ validate_char()

static bool validate_char	(	const uint8_t *	c,
		uint8_t	len
	)

static

Returns true iff the len bytes in c are a valid UTF-8 character.

Definition at line 482 of file util_string.c.

Referenced by string_is_utf8().

Macros

Functions

Detailed Description

Macro Definition Documentation

◆ LOW_BITS

◆ TOP_BITS

Function Documentation

◆ bytes_in_char()

◆ eat_whitespace()

◆ eat_whitespace_eos()

◆ eat_whitespace_eos_no_nl()

◆ eat_whitespace_no_nl()

◆ fast_mem_is_zero()

◆ fast_memcmpstart()

◆ find_str_at_start_of_line()

◆ find_whitespace()

◆ find_whitespace_eos()

◆ is_continuation_byte()

◆ strcasecmpend()

◆ strcasecmpstart()

◆ strcmp_opt()

◆ strcmpend()

◆ strcmpstart()

◆ string_is_C_identifier()

◆ string_is_utf8()

◆ string_is_utf8_no_bom()

◆ tor_digest256_is_zero()

◆ tor_digest_is_zero()

◆ tor_memmem()

◆ tor_strisnonupper()

◆ tor_strisprint()

◆ tor_strisspace()

◆ tor_strlower()

◆ tor_strreplacechar()

◆ tor_strstrip()

◆ tor_strupper()

◆ validate_char()