1berry/wireshark
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
wireshark/ui/text_import.c
Guy Harris 67944f32e3 wiretap: add more wtap_setup_XXX_rec() routines. 
For now, they just set the rec_type, butthey may do more in the future.

Also, fix tfshark and a Lua test file reader plugin to use
wtap_setup_packet_rec() or its Lua variant.
2025-05-20 17:35:36 -07:00

2044 lines
72 KiB
C
Raw Blame History

/* text_import.c
* State machine for text import
* November 2010, Jaap Keuter <jaap.keuter@xs4all.nl>
* Modified March 2021, Paul Weiß <paulniklasweiss@gmail.com>
*
* Wireshark - Network traffic analyzer
* By Gerald Combs <gerald@wireshark.org>
* Copyright 1998 Gerald Combs
*
* Based on text2pcap.c by Ashok Narayanan <ashokn@cisco.com>
*
* SPDX-License-Identifier: GPL-2.0-or-later
*/
/*******************************************************************************
*
* This code reads in an ASCII hexdump of this common format:
*
* 00000000 00 E0 1E A7 05 6F 00 10 5A A0 B9 12 08 00 46 00 .....o..Z.....F.
* 00000010 03 68 00 00 00 00 0A 2E EE 33 0F 19 08 7F 0F 19 .h.......3......
* 00000020 03 80 94 04 00 00 10 01 16 A2 0A 00 03 50 00 0C .............P..
* 00000030 01 01 0F 19 03 80 11 01 1E 61 00 0C 03 01 0F 19 .........a......
*
* Each bytestring line consists of an offset, one or more bytes, and
* text at the end. An offset is defined as a hex string of more than
* two characters. A byte is defined as a hex string of exactly two
* characters. The text at the end is ignored, as is any text before
* the offset. Bytes read from a bytestring line are added to the
* current packet only if all the following conditions are satisfied:
*
* - No text appears between the offset and the bytes (any bytes appearing after
* such text would be ignored)
*
* - The offset must be arithmetically correct, i.e. if the offset is 00000020,
* then exactly 32 bytes must have been read into this packet before this.
* If the offset is wrong, the packet is immediately terminated
*
* A packet start is signaled by a zero offset.
*
* Lines starting with #TEXT2PCAP are directives. These allow the user
* to embed instructions into the capture file which allows text2pcap
* to take some actions (e.g. specifying the encapsulation
* etc.). Currently no directives are implemented.
*
* Lines beginning with # which are not directives are ignored as
* comments. Currently all non-hexdump text is ignored by text2pcap;
* in the future, text processing may be added, but lines prefixed
* with '#' will still be ignored.
*
* The output is a libpcap packet containing Ethernet frames by
* default. This program takes options which allow the user to add
* dummy Ethernet, IP and UDP, TCP or SCTP headers to the packets in order
* to allow dumps of L3 or higher protocols to be decoded.
*
* Considerable flexibility is built into this code to read hexdumps
* of slightly different formats. For example, any text prefixing the
* hexdump line is dropped (including mail forwarding '>'). The offset
* can be any hex number of four digits or greater.
*
* This converter cannot read a single packet greater than
* WTAP_MAX_PACKET_SIZE_STANDARD. The snapshot length is automatically
* set to WTAP_MAX_PACKET_SIZE_STANDARD.
*/
/*******************************************************************************
* Alternatively this parses a Textfile based on a prel regex containing named
* capturing groups like so:
* (?<seqno>\d+)\s*(?<dir><|>)\s*(?<time>\d+:\d\d:\d\d.\d+)\s+(?<data>[0-9a-fA-F]+)\\s+
*
* Fields are decoded using a leanient parser, but only one attempt is made.
* Except for in data invalid values will be replaced by default ones.
* data currently only accepts plain HEX, OCT or BIN encoded data.
* common field separators are ignored. Note however that 0x or 0b prefixing is
* not supported and no automatic format detection is attempted.
*/
#include "config.h"
#define WS_LOG_DOMAIN LOG_DOMAIN_MAIN
#include "text_import.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <wsutil/file_util.h>
#include <ws_exit_codes.h>
#include <time.h>
#include <glib.h>
#include <errno.h>
#include <assert.h>
#include <epan/tvbuff.h>
#include <wsutil/crc32.h>
#include <epan/in_cksum.h>
#include <wsutil/report_message.h>
#include <wsutil/exported_pdu_tlvs.h>
#include <wsutil/nstime.h>
#include <wsutil/str_util.h>
#include <wsutil/time_util.h>
#include <wsutil/ws_strptime.h>
#include <wsutil/version_info.h>
#include <wsutil/cpu_info.h>
#include <wsutil/os_version_info.h>
#include "text_import_scanner.h"
#include "text_import_scanner_lex.h"
#include "text_import_regex.h"
/*--- Options --------------------------------------------------------------------*/
/* maximum time precision we can handle = 10^(-SUBSEC_PREC) */
#define SUBSEC_PREC 9
static text_import_info_t *info_p;
/* Dummy Ethernet header */
static bool hdr_ethernet;
static uint8_t hdr_eth_dest_addr[6] = {0x20, 0x52, 0x45, 0x43, 0x56, 0x00};
static uint8_t hdr_eth_src_addr[6] = {0x20, 0x53, 0x45, 0x4E, 0x44, 0x00};
static uint32_t hdr_ethernet_proto;
/* Dummy IP header */
static bool hdr_ip;
static bool hdr_ipv6;
static unsigned hdr_ip_proto;
/* Destination and source addresses for IP header */
static ws_in6_addr NO_IPv6_ADDRESS = {{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}};
/* These IPv6 default addresses are unique local addresses generated using
* the pseudo-random method from Section 3.2.2 of RFC 4193
*/
static ws_in6_addr IPv6_SRC = {{0xfd, 0xce, 0xd8, 0x62, 0x14, 0x1b, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01}};
static ws_in6_addr IPv6_DST = {{0xfd, 0xce, 0xd8, 0x62, 0x14, 0x1b, 0x00, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02}};
/* Dummy UDP header */
static bool hdr_udp;
/* Dummy TCP header */
static bool hdr_tcp;
/* TCP sequence numbers when has_direction is true */
static uint32_t tcp_in_seq_num;
static uint32_t tcp_out_seq_num;
/* Dummy SCTP header */
static bool hdr_sctp;
/* Dummy DATA chunk header */
static bool hdr_data_chunk;
static uint8_t hdr_data_chunk_type;
static uint8_t hdr_data_chunk_bits;
static uint32_t hdr_data_chunk_tsn;
static uint16_t hdr_data_chunk_sid;
static uint16_t hdr_data_chunk_ssn;
/* Dummy ExportPdu header */
static bool hdr_export_pdu;
/* Hex+ASCII text dump identification, to handle an edge case where
* the ASCII representation contains patterns that look like bytes. */
static uint8_t* pkt_lnstart;
static bool has_direction;
static uint32_t direction = PACK_FLAGS_RECEPTION_TYPE_UNSPECIFIED;
static bool has_seqno;
static uint64_t seqno;
/*--- Local data -----------------------------------------------------------------*/
/* This is where we store the packet currently being built */
static uint8_t *packet_buf;
static uint32_t curr_offset;
static uint32_t packet_start;
static bool offset_warned;
static import_status_t start_new_packet(bool);
/* This buffer contains strings present before the packet offset 0 */
#define PACKET_PREAMBLE_MAX_LEN 2048
static uint8_t packet_preamble[PACKET_PREAMBLE_MAX_LEN+1];
static int packet_preamble_len;
/* Time code of packet, derived from packet_preamble */
static time_t ts_sec;
static uint32_t ts_nsec;
static bool ts_fmt_iso;
static struct tm timecode_default;
static bool timecode_warned;
/* The time delta to add to packets without a valid time code.
* This can be no smaller than the time resolution of the dump
* file, so the default is 1000 nanoseconds, or 1 microsecond.
* XXX: We should at least get this from the resolution of the file we're
* writing to, and possibly allow the user to set a different value.
*/
static uint32_t ts_tick = 1000;
/* HDR_ETH Offset base to parse */
static uint32_t offset_base = 16;
/* ----- State machine -----------------------------------------------------------*/
/* Current state of parser */
typedef enum {
INIT, /* Waiting for start of new packet */
START_OF_LINE, /* Starting from beginning of line */
READ_OFFSET, /* Just read the offset */
READ_BYTE, /* Just read a byte */
READ_TEXT /* Just read text - ignore until EOL */
} parser_state_t;
static parser_state_t state = INIT;
static const char *state_str[] = {"Init",
"Start-of-line",
"Offset",
"Byte",
"Text"
};
static const char *token_str[] = {"",
"Byte",
"Bytes",
"Offset",
"Directive",
"Text",
"End-of-line",
"End-of-file"
};
/* ----- Skeleton Packet Headers --------------------------------------------------*/
typedef struct {
uint8_t dest_addr[6];
uint8_t src_addr[6];
uint16_t ethertype;
} hdr_ethernet_t;
static hdr_ethernet_t HDR_ETHERNET;
typedef struct {
uint8_t ver_hdrlen;
uint8_t dscp;
uint16_t packet_length;
uint16_t identification;
uint8_t flags;
uint8_t fragment;
uint8_t ttl;
uint8_t protocol;
uint16_t hdr_checksum;
uint32_t src_addr;
uint32_t dest_addr;
} hdr_ip_t;
/* Default IPv4 addresses if none supplied */
#if G_BYTE_ORDER == G_BIG_ENDIAN
#define IP_ID 0x1234
#define IP_SRC 0x0a010101
#define IP_DST 0x0a020202
#else
#define IP_ID 0x3412
#define IP_SRC 0x0101010a
#define IP_DST 0x0202020a
#endif
static hdr_ip_t HDR_IP =
{0x45, 0, 0, IP_ID, 0, 0, 0xff, 0, 0, IP_SRC, IP_DST};
static struct { /* pseudo header for checksum calculation */
uint32_t src_addr;
uint32_t dest_addr;
uint8_t zero;
uint8_t protocol;
uint16_t length;
} pseudoh;
/* headers taken from glibc */
typedef struct {
union {
struct ip6_hdrctl {
uint32_t ip6_un1_flow; /* 24 bits of flow-ID */
uint16_t ip6_un1_plen; /* payload length */
uint8_t ip6_un1_nxt; /* next header */
uint8_t ip6_un1_hlim; /* hop limit */
} ip6_un1;
uint8_t ip6_un2_vfc; /* 4 bits version, 4 bits priority */
} ip6_ctlun;
ws_in6_addr ip6_src; /* source address */
ws_in6_addr ip6_dst; /* destination address */
} hdr_ipv6_t;
static hdr_ipv6_t HDR_IPv6;
/* https://tools.ietf.org/html/rfc2460#section-8.1 */
static struct { /* pseudo header ipv6 for checksum calculation */
struct e_in6_addr src_addr6;
struct e_in6_addr dst_addr6;
uint32_t length;
uint8_t zero[3];
uint8_t next_header;
} pseudoh6;
typedef struct {
uint16_t source_port;
uint16_t dest_port;
uint16_t length;
uint16_t checksum;
} hdr_udp_t;
static hdr_udp_t HDR_UDP = {0, 0, 0, 0};
typedef struct {
uint16_t source_port;
uint16_t dest_port;
uint32_t seq_num;
uint32_t ack_num;
uint8_t hdr_length;
uint8_t flags;
uint16_t window;
uint16_t checksum;
uint16_t urg;
} hdr_tcp_t;
static hdr_tcp_t HDR_TCP = {0, 0, 0, 0, 0x50, 0, 0, 0, 0};
typedef struct {
uint16_t src_port;
uint16_t dest_port;
uint32_t tag;
uint32_t checksum;
} hdr_sctp_t;
static hdr_sctp_t HDR_SCTP = {0, 0, 0, 0};
typedef struct {
uint8_t type;
uint8_t bits;
uint16_t length;
uint32_t tsn;
uint16_t sid;
uint16_t ssn;
uint32_t ppid;
} hdr_data_chunk_t;
static hdr_data_chunk_t HDR_DATA_CHUNK = {0, 0, 0, 0, 0, 0, 0};
typedef struct {
uint16_t tag_type;
uint16_t payload_len;
} hdr_export_pdu_t;
static hdr_export_pdu_t HDR_EXPORT_PDU = {0, 0};
#define EXPORT_PDU_END_OF_OPTIONS_SIZE 4
/*----------------------------------------------------------------------
* Parse a single hex number
* Will abort the program if it can't parse the number
* Pass in true if this is an offset, false if not
*/
static import_status_t
parse_num(const char *str, int offset, uint32_t* num)
{
char *c;
if (str == NULL) {
report_failure("FATAL ERROR: str is NULL");
return IMPORT_FAILURE;
}
errno = 0;
unsigned long ulnum = strtoul(str, &c, offset ? offset_base : 16);
if (errno != 0) {
report_failure("Unable to convert %s to base %u: %s", str,
offset ? offset_base : 16, g_strerror(errno));
return IMPORT_FAILURE;
}
if (c == str) {
report_failure("Unable to convert %s to base %u", str,
offset ? offset_base : 16);
return IMPORT_FAILURE;
}
if (ulnum > UINT32_MAX) {
report_failure("%s too large", str);
return IMPORT_FAILURE;
}
*num = (uint32_t) ulnum;
return IMPORT_SUCCESS;
}
/*----------------------------------------------------------------------
* Write this byte into current packet
*/
static import_status_t
write_byte(const char *str)
{
uint32_t num;
if (parse_num(str, false, &num) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
packet_buf[curr_offset] = (uint8_t) num;
curr_offset++;
if (curr_offset >= info_p->max_frame_length) /* packet full */
if (start_new_packet(true) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
return IMPORT_SUCCESS;
}
static import_status_t
write_bytes(const char *str)
{
uint32_t num;
if (parse_num(str, false, &num) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
int len = (int)strlen(str);
/* There's always extra room in the packet_buf for the dummy headers
* compared to max_frame_length, so we could copy all the bytes at
* once and then check for overflow afterwards (copying it off).
* That would be moderately faster due to optimization on most
* compilers (and we could use the routines from wsutil/pint.h)
*/
if (info_p->hexdump.little_endian) {
for (int i = 0; i < 4*len; i += 8) {
packet_buf[curr_offset] = (uint8_t)(num >> i);
curr_offset++;
if (curr_offset >= info_p->max_frame_length) /* packet full */
if (start_new_packet(true) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
}
} else {
for (int i = 4*len - 8; i >= 0; i -= 8) {
packet_buf[curr_offset] = (uint8_t)(num >> i);
curr_offset++;
if (curr_offset >= info_p->max_frame_length) /* packet full */
if (start_new_packet(true) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
}
}
return IMPORT_SUCCESS;
}
/*----------------------------------------------------------------------
* Remove bytes from the current packet
*/
static void
unwrite_bytes (uint32_t nbytes)
{
curr_offset -= nbytes;
}
/*----------------------------------------------------------------------
* Determine SCTP chunk padding length
*/
static uint32_t
number_of_padding_bytes (uint32_t length)
{
uint32_t remainder;
remainder = length % 4;
if (remainder == 0)
return 0;
else
return 4 - remainder;
}
/*----------------------------------------------------------------------
* Write current packet out
*
* @param cont [IN] true if a packet is being written because the max frame
* length was reached, and the original packet from the input file is
* continued in a later frame. Used to set fragmentation fields in dummy
* headers (currently only implemented for SCTP; IPv4 could be added later.)
*/
static import_status_t
write_current_packet(bool cont)
{
int prefix_length = 0;
int proto_length = 0;
int ip_length = 0;
int eth_trailer_length = 0;
int prefix_index = 0;
int i, padding_length;
if (curr_offset > 0) {
/* Write the packet */
/* Is direction indication on with an inbound packet? */
bool isOutbound = has_direction && (direction == PACK_FLAGS_DIRECTION_OUTBOUND);
/* Compute packet length */
prefix_length = 0;
if (hdr_export_pdu) {
prefix_length += (int)sizeof(HDR_EXPORT_PDU) + (int)strlen(info_p->payload) + EXPORT_PDU_END_OF_OPTIONS_SIZE;
proto_length = prefix_length + curr_offset;
}
if (hdr_data_chunk) { prefix_length += (int)sizeof(HDR_DATA_CHUNK); }
if (hdr_sctp) { prefix_length += (int)sizeof(HDR_SCTP); }
if (hdr_udp) { prefix_length += (int)sizeof(HDR_UDP); proto_length = prefix_length + curr_offset; }
if (hdr_tcp) { prefix_length += (int)sizeof(HDR_TCP); proto_length = prefix_length + curr_offset; }
if (hdr_ip) {
prefix_length += (int)sizeof(HDR_IP);
ip_length = prefix_length + curr_offset + ((hdr_data_chunk) ? number_of_padding_bytes(curr_offset) : 0);
} else if (hdr_ipv6) {
ip_length = prefix_length + curr_offset + ((hdr_data_chunk) ? number_of_padding_bytes(curr_offset) : 0);
/* IPv6 payload length field does not include the header itself.
* It does include extension headers, but we don't put any
* (if we later do fragments, that would change.)
*/
prefix_length += (int)sizeof(HDR_IPv6);
}
if (hdr_ethernet) { prefix_length += (int)sizeof(HDR_ETHERNET); }
/* Make room for dummy header */
memmove(&packet_buf[prefix_length], packet_buf, curr_offset);
if (hdr_ethernet) {
/* Pad trailer */
if (prefix_length + curr_offset < 60) {
eth_trailer_length = 60 - (prefix_length + curr_offset);
}
}
/* Write Ethernet header */
if (hdr_ethernet) {
if (isOutbound)
{
memcpy(HDR_ETHERNET.dest_addr, hdr_eth_src_addr, 6);
memcpy(HDR_ETHERNET.src_addr, hdr_eth_dest_addr, 6);
} else {
memcpy(HDR_ETHERNET.dest_addr, hdr_eth_dest_addr, 6);
memcpy(HDR_ETHERNET.src_addr, hdr_eth_src_addr, 6);
}
HDR_ETHERNET.ethertype = g_htons(hdr_ethernet_proto);
memcpy(&packet_buf[prefix_index], &HDR_ETHERNET, sizeof(HDR_ETHERNET));
prefix_index += (int)sizeof(HDR_ETHERNET);
}
/* Write IP header */
if (hdr_ip) {
vec_t cksum_vector[1];
if (isOutbound) {
HDR_IP.src_addr = info_p->ip_dest_addr.ipv4 ? info_p->ip_dest_addr.ipv4 : IP_DST;
HDR_IP.dest_addr = info_p->ip_src_addr.ipv4 ? info_p->ip_src_addr.ipv4 : IP_SRC;
}
else {
HDR_IP.src_addr = info_p->ip_src_addr.ipv4 ? info_p->ip_src_addr.ipv4 : IP_SRC;
HDR_IP.dest_addr = info_p->ip_dest_addr.ipv4 ? info_p->ip_dest_addr.ipv4 : IP_DST;
}
HDR_IP.packet_length = g_htons(ip_length);
HDR_IP.protocol = (uint8_t) hdr_ip_proto;
HDR_IP.hdr_checksum = 0;
cksum_vector[0].ptr = (uint8_t *)&HDR_IP; cksum_vector[0].len = sizeof(HDR_IP);
HDR_IP.hdr_checksum = in_cksum(cksum_vector, 1);
memcpy(&packet_buf[prefix_index], &HDR_IP, sizeof(HDR_IP));
prefix_index += (int)sizeof(HDR_IP);
/* initialize pseudo header for checksum calculation */
pseudoh.src_addr = HDR_IP.src_addr;
pseudoh.dest_addr = HDR_IP.dest_addr;
pseudoh.zero = 0;
pseudoh.protocol = (uint8_t) hdr_ip_proto;
pseudoh.length = g_htons(proto_length);
} else if (hdr_ipv6) {
if (memcmp(&info_p->ip_dest_addr.ipv6, &NO_IPv6_ADDRESS, sizeof(ws_in6_addr))) {
memcpy(isOutbound ? &HDR_IPv6.ip6_src : &HDR_IPv6.ip6_dst, &info_p->ip_dest_addr.ipv6, sizeof(ws_in6_addr));
} else {
memcpy(isOutbound ? &HDR_IPv6.ip6_src : &HDR_IPv6.ip6_dst, &IPv6_DST, sizeof(ws_in6_addr));
}
if (memcmp(&info_p->ip_src_addr.ipv6, &NO_IPv6_ADDRESS, sizeof(ws_in6_addr))) {
memcpy(isOutbound ? &HDR_IPv6.ip6_dst : &HDR_IPv6.ip6_src, &info_p->ip_src_addr.ipv6, sizeof(ws_in6_addr));
} else {
memcpy(isOutbound ? &HDR_IPv6.ip6_dst : &HDR_IPv6.ip6_src, &IPv6_SRC, sizeof(ws_in6_addr));
}
HDR_IPv6.ip6_ctlun.ip6_un2_vfc &= 0x0F;
HDR_IPv6.ip6_ctlun.ip6_un2_vfc |= (6<< 4);
HDR_IPv6.ip6_ctlun.ip6_un1.ip6_un1_plen = g_htons(ip_length);
HDR_IPv6.ip6_ctlun.ip6_un1.ip6_un1_nxt = (uint8_t) hdr_ip_proto;
HDR_IPv6.ip6_ctlun.ip6_un1.ip6_un1_hlim = 32;
memcpy(&packet_buf[prefix_index], &HDR_IPv6, sizeof(HDR_IPv6));
prefix_index += (int)sizeof(HDR_IPv6);
/* initialize pseudo ipv6 header for checksum calculation */
pseudoh6.src_addr6 = HDR_IPv6.ip6_src;
pseudoh6.dst_addr6 = HDR_IPv6.ip6_dst;
memset(pseudoh6.zero, 0, sizeof(pseudoh6.zero));
pseudoh6.next_header = (uint8_t) hdr_ip_proto;
pseudoh6.length = g_htons(proto_length);
}
/* Write UDP header */
if (hdr_udp) {
vec_t cksum_vector[3];
HDR_UDP.source_port = isOutbound ? g_htons(info_p->dst_port): g_htons(info_p->src_port);
HDR_UDP.dest_port = isOutbound ? g_htons(info_p->src_port) : g_htons(info_p->dst_port);
HDR_UDP.length = g_htons(proto_length);
HDR_UDP.checksum = 0;
if (hdr_ipv6) {
cksum_vector[0].ptr = (uint8_t *)&pseudoh6; cksum_vector[0].len = sizeof(pseudoh6);
} else {
cksum_vector[0].ptr = (uint8_t *)&pseudoh; cksum_vector[0].len = sizeof(pseudoh);
}
cksum_vector[1].ptr = (uint8_t *)&HDR_UDP; cksum_vector[1].len = sizeof(HDR_UDP);
cksum_vector[2].ptr = &packet_buf[prefix_length]; cksum_vector[2].len = curr_offset;
HDR_UDP.checksum = in_cksum(cksum_vector, 3);
memcpy(&packet_buf[prefix_index], &HDR_UDP, sizeof(HDR_UDP));
prefix_index += (int)sizeof(HDR_UDP);
}
/* Write TCP header */
if (hdr_tcp) {
vec_t cksum_vector[3];
HDR_TCP.source_port = isOutbound ? g_htons(info_p->dst_port): g_htons(info_p->src_port);
HDR_TCP.dest_port = isOutbound ? g_htons(info_p->src_port) : g_htons(info_p->dst_port);
/* set ack number if we have direction */
if (has_direction) {
HDR_TCP.flags = 0x10;
HDR_TCP.ack_num = g_ntohl(isOutbound ? tcp_out_seq_num : tcp_in_seq_num);
HDR_TCP.ack_num = g_htonl(HDR_TCP.ack_num);
}
else {
HDR_TCP.flags = 0;
HDR_TCP.ack_num = 0;
}
HDR_TCP.seq_num = isOutbound ? tcp_in_seq_num : tcp_out_seq_num;
HDR_TCP.window = g_htons(0x2000);
HDR_TCP.checksum = 0;
if (hdr_ipv6) {
cksum_vector[0].ptr = (uint8_t *)&pseudoh6; cksum_vector[0].len = sizeof(pseudoh6);
} else {
cksum_vector[0].ptr = (uint8_t *)&pseudoh; cksum_vector[0].len = sizeof(pseudoh);
}
cksum_vector[1].ptr = (uint8_t *)&HDR_TCP; cksum_vector[1].len = sizeof(HDR_TCP);
cksum_vector[2].ptr = &packet_buf[prefix_length]; cksum_vector[2].len = curr_offset;
HDR_TCP.checksum = in_cksum(cksum_vector, 3);
memcpy(&packet_buf[prefix_index], &HDR_TCP, sizeof(HDR_TCP));
prefix_index += (int)sizeof(HDR_TCP);
if (isOutbound) {
tcp_in_seq_num = g_ntohl(tcp_in_seq_num) + curr_offset;
tcp_in_seq_num = g_htonl(tcp_in_seq_num);
}
else {
tcp_out_seq_num = g_ntohl(tcp_out_seq_num) + curr_offset;
tcp_out_seq_num = g_htonl(tcp_out_seq_num);
}
}
/* Compute DATA chunk header and append padding */
if (hdr_data_chunk) {
hdr_data_chunk_bits = 0;
if (packet_start == 0) {
hdr_data_chunk_bits |= 0x02;
}
if (!cont) {
hdr_data_chunk_bits |= 0x01;
}
HDR_DATA_CHUNK.type = hdr_data_chunk_type;
HDR_DATA_CHUNK.bits = hdr_data_chunk_bits;
HDR_DATA_CHUNK.length = g_htons(curr_offset + sizeof(HDR_DATA_CHUNK));
HDR_DATA_CHUNK.tsn = g_htonl(hdr_data_chunk_tsn);
HDR_DATA_CHUNK.sid = g_htons(hdr_data_chunk_sid);
HDR_DATA_CHUNK.ssn = g_htons(hdr_data_chunk_ssn);
HDR_DATA_CHUNK.ppid = g_htonl(info_p->ppi);
hdr_data_chunk_tsn++;
if (!cont) {
hdr_data_chunk_ssn++;
}
padding_length = number_of_padding_bytes(curr_offset);
for (i=0; i<padding_length; i++)
packet_buf[prefix_length+curr_offset+i] = 0;
curr_offset += padding_length;
}
/* Write SCTP header */
if (hdr_sctp) {
HDR_SCTP.src_port = isOutbound ? g_htons(info_p->dst_port): g_htons(info_p->src_port);
HDR_SCTP.dest_port = isOutbound ? g_htons(info_p->src_port) : g_htons(info_p->dst_port);
HDR_SCTP.tag = g_htonl(info_p->tag);
HDR_SCTP.checksum = g_htonl(0);
HDR_SCTP.checksum = crc32c_calculate(&HDR_SCTP, sizeof(HDR_SCTP), CRC32C_PRELOAD);
if (hdr_data_chunk)
HDR_SCTP.checksum = crc32c_calculate(&HDR_DATA_CHUNK, sizeof(HDR_DATA_CHUNK), HDR_SCTP.checksum);
HDR_SCTP.checksum = g_htonl(~crc32c_calculate(&packet_buf[prefix_length], curr_offset, HDR_SCTP.checksum));
memcpy(&packet_buf[prefix_index], &HDR_SCTP, sizeof(HDR_SCTP));
prefix_index += (int)sizeof(HDR_SCTP);
}
/* Write DATA chunk header */
if (hdr_data_chunk) {
memcpy(&packet_buf[prefix_index], &HDR_DATA_CHUNK, sizeof(HDR_DATA_CHUNK));
/*prefix_index += (int)sizeof(HDR_DATA_CHUNK);*/
}
/* Write ExportPDU header */
if (hdr_export_pdu) {
unsigned payload_len = (unsigned)strlen(info_p->payload);
HDR_EXPORT_PDU.tag_type = g_htons(EXP_PDU_TAG_DISSECTOR_NAME);
HDR_EXPORT_PDU.payload_len = g_htons(payload_len);
memcpy(&packet_buf[prefix_index], &HDR_EXPORT_PDU, sizeof(HDR_EXPORT_PDU));
prefix_index += sizeof(HDR_EXPORT_PDU);
memcpy(&packet_buf[prefix_index], info_p->payload, payload_len);
prefix_index += payload_len;
/* Add end-of-options tag */
memset(&packet_buf[prefix_index], 0x00, 4);
}
/* Write Ethernet trailer */
if (hdr_ethernet && eth_trailer_length > 0) {
memset(&packet_buf[prefix_length+curr_offset], 0, eth_trailer_length);
}
HDR_TCP.seq_num = g_ntohl(HDR_TCP.seq_num) + curr_offset;
HDR_TCP.seq_num = g_htonl(HDR_TCP.seq_num);
/* Write the record */
int data_length;
wtap_rec rec;
int err;
char *err_info;
data_length = prefix_length + curr_offset + eth_trailer_length;
wtap_rec_init(&rec, data_length);
if (info_p->encapsulation == WTAP_ENCAP_SYSTEMD_JOURNAL) {
wtap_setup_systemd_journal_export_rec(&rec);
rec.block = wtap_block_create(WTAP_BLOCK_SYSTEMD_JOURNAL_EXPORT);
rec.rec_header.systemd_journal_export_header.record_len = data_length;
rec.presence_flags = WTAP_HAS_CAP_LEN|WTAP_HAS_TS;
/* XXX: Ignore our direction, packet id, and timestamp. For a
* systemd Journal Export Block the timestamp comes from the
* __REALTIME_TIMESTAMP= field. We don't check to see if that
* field is there (it MUST be, but we don't check whether our
* input is malformed in general), but since the presence flags
* aren't really used when writing, it doesn't matter.
*/
} else {
wtap_setup_packet_rec(&rec, info_p->encapsulation);
rec.block = wtap_block_create(WTAP_BLOCK_PACKET);
rec.rec_header.packet_header.caplen = rec.rec_header.packet_header.len = data_length;
rec.ts.secs = ts_sec;
rec.ts.nsecs = ts_nsec;
rec.presence_flags = WTAP_HAS_CAP_LEN|WTAP_HAS_INTERFACE_ID|WTAP_HAS_TS;
if (has_direction) {
wtap_block_add_uint32_option(rec.block, OPT_PKT_FLAGS, direction);
}
if (has_seqno) {
wtap_block_add_uint64_option(rec.block, OPT_PKT_PACKETID, seqno);
}
}
ws_buffer_append(&rec.data, packet_buf, data_length);
if (!wtap_dump(info_p->wdh, &rec, &err, &err_info)) {
report_cfile_write_failure(info_p->import_text_filename,
info_p->output_filename, err, err_info,
info_p->num_packets_read,
wtap_dump_file_type_subtype(info_p->wdh));
wtap_rec_cleanup(&rec);
return IMPORT_FAILURE;
}
wtap_rec_cleanup(&rec);
info_p->num_packets_written++;
}
packet_start += curr_offset;
curr_offset = 0;
return IMPORT_SUCCESS;
}
/*----------------------------------------------------------------------
* Append a token to the packet preamble.
*/
static import_status_t
append_to_preamble(char *str)
{
size_t toklen;
if (packet_preamble_len != 0) {
if (packet_preamble_len == PACKET_PREAMBLE_MAX_LEN)
return IMPORT_SUCCESS; /* no room to add more preamble */
/* XXX: Just keep going? This is probably not a problem, unless
* someone had >2000 bytes of whitespace before the timestamp... */
/* Add a blank separator between the previous token and this token. */
packet_preamble[packet_preamble_len++] = ' ';
}
if(str == NULL){
report_failure("FATAL ERROR: str is NULL");
return IMPORT_FAILURE;
}
toklen = strlen(str);
if (toklen != 0) {
if (packet_preamble_len + toklen > PACKET_PREAMBLE_MAX_LEN)
return IMPORT_SUCCESS; /* no room to add token to the preamble */
/* XXX: Just keep going? This is probably not a problem, as above.*/
(void) g_strlcpy(&packet_preamble[packet_preamble_len], str, PACKET_PREAMBLE_MAX_LEN);
packet_preamble_len += (int) toklen;
if (ws_log_get_level() >= LOG_LEVEL_NOISY) {
char *c;
char xs[PACKET_PREAMBLE_MAX_LEN];
(void) g_strlcpy(xs, packet_preamble, PACKET_PREAMBLE_MAX_LEN);
while ((c = strchr(xs, '\r')) != NULL) *c=' ';
ws_noisy("[[append_to_preamble: \"%s\"]]", xs);
}
}
return IMPORT_SUCCESS;
}
#define INVALID_VALUE (-1)
/*
* This includes not just true whitespace values, but also any other
* character silently ignored if part of the data group match, such as
* byte separators or the '=' used for padding at the end of Base64.
*/
#define WHITESPACE_VALUE (-2)
/*
* Information on how to parse any plainly encoded binary data
*
* one Unit is least_common_multiple(bits_per_char, 8) bits.
*/
struct plain_decoding_data {
const char* name;
unsigned chars_per_unit;
unsigned bytes_per_unit : 3; /* Internally a uint64_t is used to hold units */
unsigned bits_per_char : 6;
int8_t table[256];
};
#define _INVALID_INIT2 INVALID_VALUE, INVALID_VALUE
#define _INVALID_INIT4 _INVALID_INIT2, _INVALID_INIT2
#define _INVALID_INIT8 _INVALID_INIT4, _INVALID_INIT4
#define _INVALID_INIT16 _INVALID_INIT8, _INVALID_INIT8
#define _INVALID_INIT32 _INVALID_INIT16, _INVALID_INIT16
#define _INVALID_INIT64 _INVALID_INIT32, _INVALID_INIT32
#define _INVALID_INIT128 _INVALID_INIT64, _INVALID_INIT64
#define _INVALID_INIT256 _INVALID_INIT128, _INVALID_INIT128
#define INVALID_INIT _INVALID_INIT256
// this is a gcc/clang extension:
// [0 ... 255] = INVALID_VALUE
#define WHITESPACE_INIT \
[' '] = WHITESPACE_VALUE, \
['\t'] = WHITESPACE_VALUE, \
['\n'] = WHITESPACE_VALUE, \
['\v'] = WHITESPACE_VALUE, \
['\f'] = WHITESPACE_VALUE, \
['\r'] = WHITESPACE_VALUE
/*
* Some compilers warn about initializing the same subobject
* more than once with designated initializers.
*
* We're doing that - INVALID_INIT iniitalizes everything to
* INVALID_VALUE, but then we override selected elements -
* but we know what we're doing, so just suppress that
* warning.
*/
DIAG_OFF_INIT_TWICE
const struct plain_decoding_data hex_decode_info = {
.chars_per_unit = 2,
.bytes_per_unit = 1,
.bits_per_char = 4,
.table = {
INVALID_INIT,
WHITESPACE_INIT,
[':'] = WHITESPACE_VALUE,
['0'] = 0,1,2,3,4,5,6,7,8,9,
['A'] = 10,11,12,13,14,15,
['a'] = 10,11,12,13,14,15
}
};
const struct plain_decoding_data bin_decode_info = {
.chars_per_unit = 8,
.bytes_per_unit = 1,
.bits_per_char = 1,
.table = {
INVALID_INIT,
WHITESPACE_INIT,
['0'] = 0, 1
}
};
const struct plain_decoding_data oct_decode_info = {
.chars_per_unit = 8,
.bytes_per_unit = 3,
.bits_per_char = 3,
.table = {
INVALID_INIT,
WHITESPACE_INIT,
['0'] = 0,1,2,3,4,5,6,7
}
};
const struct plain_decoding_data base64_decode_info = {
.chars_per_unit = 4,
.bytes_per_unit = 3,
.bits_per_char = 6,
.table = {
INVALID_INIT,
WHITESPACE_INIT,
['A'] = 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,
['a'] = 26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,
['0'] = 52,53,54,55,56,57,58,59,60,61,
['+'] = 62,
['/'] = 63,
['='] = WHITESPACE_VALUE /* padding at the end, the decoder doesn't need this, so just ignores it */
}
};
DIAG_ON_INIT_TWICE
/*******************************************************************************
* The modularized part of this mess, used by the wrapper around the regex
* engine in text_import_regex.c to hook into this state-machine backend.
*
* Should the rest be modularized as well? Maybe, but then start with pcap2text.c
*/
/**
* This function parses encoded data according to <encoding> into binary data.
* It will continue until one of the following conditions is met:
* - src is depleted
* - dest cannot hold another full unit of data
* - an invalid character is read
* When this happens any complete bytes will be recovered from the remaining
* possibly incomplete unit and stored to dest (there will be no incomplete unit
* if dest is full). Any remaining bits will be discarded.
* src and dest will be advanced to where parsing including this last incomplete
* unit stopped.
* If you want to continue parsing (meaning incomplete units were due to call
* fragmentation and not actually due to EOT) you have to resume the parser at
* *src_last_unit and dest - result % bytes_per_unit
*/
static int parse_plain_data(unsigned char** src, const unsigned char* src_end,
uint8_t** dest, const uint8_t* dest_end, const struct plain_decoding_data* encoding,
unsigned char** src_last_unit) {
int status = 1;
int units = 0;
/* unit buffer */
uint64_t c_val = 0;
unsigned c_chars = 0;
/**
* Src data |- - -|- - -|- - -|- - -|- - -|- - -|- - -|- - -|
* Bytes |- - - - - - - -|- - - - - - - -|- - - - - - - -|
* Units |- - - - - - - - - - - - - - - - - - - - - - - -|
*/
uint64_t val;
int j;
if (ws_log_get_level() >= LOG_LEVEL_NOISY) {
char* debug_str = wmem_strndup(NULL, *src, (src_end-*src));
ws_noisy("parsing data: %s", debug_str);
wmem_free(NULL, debug_str);
}
while (*src < src_end && *dest + encoding->bytes_per_unit <= dest_end) {
/*
* XXX - This reads one octet at a time. If the GRegex compile flag
* G_REGEX_RAW wasn't given, the match was done using UTF-8 characters.
* That's useful, I suppose, if the input file has UTF-8 text (comments,
* etc.) that ought to be ignored - but it means that character classes
* like \d and \s can match Unicode characters that might (but probably
* aren't) in the data. Those will cause an invalid value warning and
* abort, which is easier than adding full UTF-8 support here.
*/
val = encoding->table[**src];
switch (val) {
case INVALID_VALUE:
report_failure("Unexpected char %d in data", **src);
status = -1;
goto remainder;
case WHITESPACE_VALUE:
break;
default:
c_val = c_val << encoding->bits_per_char | val;
++c_chars;
/* another full unit */
if (c_chars == encoding->chars_per_unit) {
++units;
if (src_last_unit)
*src_last_unit = *src;
c_chars = 0;
for (j = encoding->bytes_per_unit; j > 0; --j) {
**dest = (char) (c_val >> (j * 8 - 8));
*dest += 1;
}
}
}
*src += 1;
}
remainder:
for (j = c_chars * encoding->bits_per_char; j >= 8; j -= 8) {
**dest = (char) (c_val >> (j - 8));
*dest += 1;
}
return status * units;
}
void parse_data(unsigned char* start_field, unsigned char* end_field, enum data_encoding encoding) {
uint8_t* dest = &packet_buf[curr_offset];
uint8_t* dest_end = &packet_buf[info_p->max_frame_length];
const struct plain_decoding_data* table; /* should be further down */
switch (encoding) {
case ENCODING_PLAIN_HEX:
case ENCODING_PLAIN_OCT:
case ENCODING_PLAIN_BIN:
case ENCODING_BASE64:
/* const struct plain_decoding_data* table; // This can't be here because gcc says no */
switch (encoding) {
case ENCODING_PLAIN_HEX:
table = &hex_decode_info;
break;
case ENCODING_PLAIN_OCT:
table = &oct_decode_info;
break;
case ENCODING_PLAIN_BIN:
table = &bin_decode_info;
break;
case ENCODING_BASE64:
table = &base64_decode_info;
break;
default:
return;
}
info_p->num_packets_read++;
while (1) {
parse_plain_data(&start_field, end_field, &dest, dest_end, table, NULL);
curr_offset = (int) (dest - packet_buf);
if (curr_offset == info_p->max_frame_length) {
write_current_packet(true);
dest = &packet_buf[curr_offset];
} else
break;
}
break;
default:
ws_critical("not implemented/invalid encoding type");
return;
}
}
#define setFlags(VAL, MASK, FLAGS) \
((VAL) & ~(MASK)) | ((FLAGS) & (MASK))
static void _parse_dir(const unsigned char* start_field, const unsigned char* end_field _U_, const char* in_indicator, const char* out_indicator, uint32_t* dir) {
for (; *in_indicator && *start_field != *in_indicator; ++in_indicator);
if (*in_indicator) {
*dir = setFlags(*dir, PACK_FLAGS_DIRECTION_MASK << PACK_FLAGS_DIRECTION_SHIFT, PACK_FLAGS_DIRECTION_INBOUND);
return;
}
for (; *out_indicator && *start_field != *out_indicator; ++out_indicator);
if (*out_indicator) {
*dir = setFlags(*dir, PACK_FLAGS_DIRECTION_MASK << PACK_FLAGS_DIRECTION_SHIFT, PACK_FLAGS_DIRECTION_OUTBOUND);
return;
}
*dir = setFlags(*dir, PACK_FLAGS_DIRECTION_MASK << PACK_FLAGS_DIRECTION_SHIFT, PACK_FLAGS_DIRECTION_UNKNOWN);
}
void parse_dir(const unsigned char* start_field, const unsigned char* end_field, const char* in_indicator, const char* out_indicator) {
_parse_dir(start_field, end_field, in_indicator, out_indicator, &direction);
}
#define PARSE_BUF 64
/* Attempt to parse a time according to the given format. If the conversion
* succeeds, set sec and nsec appropriately and return true. If it fails,
* leave sec and nsec unchanged and return false.
*/
static bool
_parse_time(const unsigned char* start_field, const unsigned char* end_field, const char* _format, time_t* sec, int* nsec) {
struct tm timecode;
time_t sec_buf;
int nsec_buf = 0;
char field[PARSE_BUF];
char format[PARSE_BUF];
char* subsecs_fmt;
int subseclen = -1;
char *cursor;
char *p;
int i;
(void) g_strlcpy(field, start_field, MIN(end_field - start_field + 1, PARSE_BUF));
if (ts_fmt_iso) {
nstime_t ts_iso;
if (!iso8601_to_nstime(&ts_iso, field, ISO8601_DATETIME_AUTO)) {
return false;
}
*sec = ts_iso.secs;
*nsec = ts_iso.nsecs;
} else {
(void) g_strlcpy(format, _format, PARSE_BUF);
/*
* Initialize to today, local time, just in case not all fields
* of the date and time are specified.
*/
timecode = timecode_default;
cursor = &field[0];
/*
* %f is for fractions of seconds not supported by strptime
*/
subsecs_fmt = g_strrstr(format, "%f");
if (subsecs_fmt) {
*subsecs_fmt = 0;
}
cursor = ws_strptime_p(cursor, format, &timecode);
if (cursor == NULL) {
return false;
}
if (subsecs_fmt != NULL) {
/*
* Parse subsecs and any following format
*/
nsec_buf = (unsigned) strtol(cursor, &p, 10);
if (p == cursor) {
return false;
}
subseclen = (int) (p - cursor);
cursor = p;
cursor = ws_strptime_p(cursor, subsecs_fmt + 2, &timecode);
if (cursor == NULL) {
return false;
}
}
if (subseclen > 0) {
/*
* Convert that number to a number
* of nanoseconds; if it's N digits
* long, it's in units of 10^(-N) seconds,
* so, to convert it to units of
* 10^-9 seconds, we multiply by
* 10^(9-N).
*/
if (subseclen > SUBSEC_PREC) {
/*
* *More* than 9 digits; 9-N is
* negative, so we divide by
* 10^(N-9).
*/
for (i = subseclen - SUBSEC_PREC; i != 0; i--)
nsec_buf /= 10;
} else if (subseclen < SUBSEC_PREC) {
for (i = SUBSEC_PREC - subseclen; i != 0; i--)
nsec_buf *= 10;
}
}
if ( -1 == (sec_buf = mktime(&timecode)) ) {
return false;
}
*sec = sec_buf;
*nsec = nsec_buf;
}
ws_noisy("parsed time %s Format(%s), time(%u), subsecs(%u)\n", field, _format, (uint32_t)*sec, (uint32_t)*nsec);
return true;
}
void parse_time(const unsigned char* start_field, const unsigned char* end_field, const char* format) {
if (format == NULL || !_parse_time(start_field, end_field, format, &ts_sec, &ts_nsec)) {
ts_nsec += ts_tick;
}
}
void parse_seqno(const unsigned char* start_field, const unsigned char* end_field) {
char* buf = (char*) g_alloca(end_field - start_field + 1);
(void) g_strlcpy(buf, start_field, end_field - start_field + 1);
seqno = g_ascii_strtoull(buf, NULL, 10);
}
void flush_packet(void) {
write_current_packet(false);
}
/*----------------------------------------------------------------------
* Parse the preamble to get the timecode.
*/
static void
parse_preamble (void)
{
int i;
bool got_time = false;
/*
* Null-terminate the preamble.
*/
packet_preamble[packet_preamble_len] = '\0';
if (has_direction) {
_parse_dir(&packet_preamble[0], &packet_preamble[1], "iI", "oO", &direction);
i = (direction == PACK_FLAGS_DIRECTION_UNKNOWN) ? 0 : 1;
while (packet_preamble[i] == ' ' ||
packet_preamble[i] == '\r' ||
packet_preamble[i] == '\t') {
i++;
}
packet_preamble_len -= i;
/* Also move the trailing '\0'. */
memmove(packet_preamble, packet_preamble + i, packet_preamble_len + 1);
}
/*
* If no time stamp format was specified, don't attempt to parse
* the packet preamble to extract a time stamp.
*/
/* Ensure preamble has more than two chars before attempting to parse.
* This should cover line breaks etc that get counted.
*/
if ( info_p->timestamp_format != NULL && strlen(packet_preamble) > 2 ) {
got_time = _parse_time(packet_preamble, packet_preamble + strlen(packet_preamble), info_p->timestamp_format, &ts_sec, &ts_nsec);
if (!got_time) {
/* Let's only have a possible GUI popup once, other messages to log
*/
if (!timecode_warned) {
report_warning("Time conversions (%s) failed, advancing time by %d ns from previous packet on failure. First failure was for %s on input packet %d.", info_p->timestamp_format, ts_tick, packet_preamble, info_p->num_packets_read);
timecode_warned = true;
}
ws_warning("Time conversion (%s) failed for %s on input packet %d.", info_p->timestamp_format, packet_preamble, info_p->num_packets_read);
}
}
if (ws_log_get_level() >= LOG_LEVEL_NOISY) {
char *c;
while ((c = strchr(packet_preamble, '\r')) != NULL) *c=' ';
ws_noisy("[[parse_preamble: \"%s\"]]", packet_preamble);
ws_noisy("Format(%s), time(%u), subsecs(%u)", info_p->timestamp_format, (uint32_t)ts_sec, ts_nsec);
}
if (!got_time) {
ts_nsec += ts_tick;
}
/* Clear Preamble */
packet_preamble_len = 0;
}
/*----------------------------------------------------------------------
* Start a new packet
*
* @param cont [IN] true if a new packet is starting because the max frame
* length was reached on the current packet, and the original packet from the
* input file is continued in a later frame. Passed to write_current_packet,
* where it is used to set fragmentation fields in dummy headers (currently
* only implemented for SCTP; IPv4 could be added later.)
*/
static import_status_t
start_new_packet(bool cont)
{
ws_debug("Start new packet (cont = %s).", cont ? "true" : "false");
/* Write out the current packet, if required */
if (write_current_packet(cont) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
info_p->num_packets_read++;
/* Ensure we parse the packet preamble as it may contain the time */
/* THIS IMPLIES A STATE TRANSITION OUTSIDE THE STATE MACHINE */
parse_preamble();
return IMPORT_SUCCESS;
}
/*----------------------------------------------------------------------
* Process a directive
*/
static void
process_directive (char *str _U_)
{
char **tokens;
tokens = g_strsplit_set(str+10, "\r\n", 2);
ws_message("--- Directive [%s] currently unsupported ---", tokens[0]);
g_strfreev(tokens);
}
/*----------------------------------------------------------------------
* If this is a hex+ASCII dump, after finishing a line of packet bytes,
* test if the beginning of the hex portion would produce ASCII that
* would be treated as byte tokens (e.g., "61 62 20 ... ab "), and if so,
* look for the bytes and remove them from the output.
*/
static void
process_rollback(bool by_eol)
{
int rollback = 0;
int pending_rollback = 0;
int line_size;
int i;
GString *s2;
char tmp_str[3];
/* Here a line of pkt bytes reading is finished */
rollback = 0;
/* s2 is the ASCII string, s1 is the HEX string, e.g, when
s2 = "ab ", s1 = "616220"
we should find out the largest tail of s1 matches the head
of s2, it means the matched part in tail is the ASCII dump
of the head byte. Those matched should be rolled back. */
line_size = curr_offset-(int)(pkt_lnstart-packet_buf);
s2 = g_string_new(NULL);
/* gather the possible pattern */
/* This is a simplified version of the lexical analysis done by the scanner
* and has to produce the same tokens. (Simpler because we only match bytes,
* space, and everything else.) Perhaps we should do lexical analysis with
* yy_scan_bytes. */
/* Note we have the bytes after any swapping of little endian byte groups
* has been done. That's what we want; various hexdump programs (hexdump,
* od, xxd) always output the ASCII dump in the natural order even if
* byte groups are little endian. */
i = 0;
while (i + 1 + rollback < line_size) {
if (pkt_lnstart[i] == ' ') {
/* Skip any spaces. We don't skip '\t', because in the ASCII dump
* those are converted to '.' and so wouldn't create extra byte
* tokens. */
i++;
continue;
}
tmp_str[0] = pkt_lnstart[i];
tmp_str[1] = pkt_lnstart[i+1];
tmp_str[2] = '\0';
/* it is a valid convertible string */
if (!g_ascii_isxdigit(tmp_str[0]) || !g_ascii_isxdigit(tmp_str[1])) {
break;
}
g_string_append_c(s2, (char)strtoul(tmp_str, (char **)NULL, 16));
i += 2;
if (by_eol) {
rollback++;
} else {
pending_rollback++;
/* If we had a text token before EOL, then without ' ' after the
* byte it won't parse as a byte token in the ASCII dump.
* If we transitioned straight from T_BYTE to T_EOL, then we already
* know the entire ASCII dump parsed as bytes. */
if (!(i + rollback < line_size)) {
break;
}
if (pkt_lnstart[i] == ' ') {
rollback += pending_rollback;
pending_rollback = 0;
i++;
} else if (pending_rollback >= 4) {
break;
}
}
}
/* If the packet line start contains a possible byte pattern, the line end
should contain the matched pattern if the user gave the -a flag.
The line will be considered invalid if the byte pattern cannot find
a matched one in the line of packet buffer.
XXX - We could instead do nothing (or warn) if the pattern isn't found.
That's probably better than dropping the entire line if the -a flag was
given despite there being no ASCII dump or if there is an ASCII dump but
separated by a delimiter like '|' that would prevent detecting it as
bytes. It's still better to avoid passing the flag if there is no ASCII
dump to avoid the extremely rare false detection where the actual bytes
look like a matching ASCII dump.
*/
if (rollback > 0) {
if (strncmp(pkt_lnstart+line_size-rollback, s2->str, rollback) == 0) {
unwrite_bytes(rollback);
} else {
/* Not matched. This line contains invalid packet bytes, so
discard the whole line */
/* The GUI only reports identical warnings once to save lots of pop-ups.
* Keep the unique information in a console message. */
report_warning("Expected ASCII rollback not found. Was ASCII identification enabled unnecessarily?");
ws_message("Expected %i byte%s to rollback at the end of line offset 0x%0X in packet %u.", rollback, plurality(rollback, "", "s"), curr_offset - line_size, info_p->num_packets_read);
unwrite_bytes(line_size);
}
}
g_string_free(s2, TRUE);
}
/*----------------------------------------------------------------------
* Parse a single token (called from the scanner)
*/
import_status_t
parse_token(token_t token, char *str)
{
uint32_t num;
char **tokens;
/*
* This is implemented as a simple state machine of five states.
* State transitions are caused by tokens being received from the
* scanner. The code should be self_documenting.
*/
if (ws_log_get_level() >= LOG_LEVEL_NOISY) {
/* Sanitize - remove all '\r' */
char *c;
if (str!=NULL) { while ((c = strchr(str, '\r')) != NULL) *c=' '; }
ws_noisy("(%s, %s \"%s\") -> (",
state_str[state], token_str[token], str ? str : "");
}
switch(state) {
/* ----- Waiting for new packet -------------------------------------------*/
case INIT:
switch(token) {
case T_TEXT:
append_to_preamble(str);
break;
case T_DIRECTIVE:
process_directive(str);
break;
case T_OFFSET:
case T_BYTES:
if (offset_base == 0) {
append_to_preamble(str);
/* If we're still in the INIT state, maybe there's something
* odd like a time format with no separators. That wouldn't
* work in a mode with an offset, but give it a try.
*/
tokens = g_strsplit_set(str, ": \t\r\n", 2);
if (!offset_warned) {
report_warning("Running in no offset mode but read offset (%s) at start of file, treating as preamble", tokens[0]);
offset_warned = true;
}
ws_warning("Running in no offset mode but read offset (%s) at start of file, treating as preamble", tokens[0]);
g_strfreev(tokens);
break;
}
if (parse_num(str, true, &num) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
if (num == 0) {
/* New packet starts here */
/* XXX - This could just be any other run of three or more
* zeros in a preamble. That only creates a problem if followed
* by something that looks like bytes (otherwise a zero length
* packet is ignored.)
*/
if (start_new_packet(false) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
packet_start = 0;
state = READ_OFFSET;
pkt_lnstart = packet_buf + num;
} else {
/* Not a new packet; assume it's part of a preamble that
* looks like an offset token. (#17140)
*/
append_to_preamble(str);
}
break;
case T_BYTE:
if (offset_base == 0) {
/* In no offset mode, assume this starts our packet */
if (start_new_packet(false) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
if (write_byte(str) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
state = READ_BYTE;
pkt_lnstart = packet_buf;
break;
}
/* Otherwise, this is probably text that looks like a byte,
* e.g. a day or month with certain timestamp formats.
*/
append_to_preamble(str);
break;
case T_EOF:
if (write_current_packet(false) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
break;
default:
break;
}
break;
/* ----- Processing packet, start of new line -----------------------------*/
case START_OF_LINE:
switch(token) {
case T_TEXT:
/* If the only text allowed on the same line before an offset (other
* than the zero offset that begins a packet) were mailfwd ('>') and
* blanks, we could ignore those, and and switch back to INIT state
* here to decrease ambiguity. (That is, assume that other text at
* line start indicates the end of an old packet and start of new.)
*/
if (offset_base != 0) {
append_to_preamble(str);
}
break;
case T_DIRECTIVE:
process_directive(str);
break;
case T_OFFSET:
case T_BYTES:
if (offset_base == 0) {
/* After starting the packet there's no point adding it to
* the preamble in this mode (we only do one packet.)
* Use a generic warning message to suppress the many
* expected duplicates. */
tokens = g_strsplit_set(str, ": \t\r\n", 2);
if (!offset_warned) {
report_warning("Running in no offset mode but read offset (%s) at start of line, ignoring", tokens[0]);
offset_warned = true;
}
ws_warning("Running in no offset mode but read offset (%s) at start of line, ignoring.", tokens[0]);
g_strfreev(tokens);
break;
}
if (parse_num(str, true, &num) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
if (num == 0) {
/* New packet starts here */
/* XXX - This could just be any other run of three or more
* zeros in a preamble. That only creates a problem if followed
* by something that looks like bytes (otherwise a zero length
* packet is ignored.) */
if (start_new_packet(false) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
packet_start = 0;
state = READ_OFFSET;
} else if ((num - packet_start) != curr_offset) {
/*
* The offset we read isn't the one we expected.
* This may only mean that we mistakenly interpreted
* some text as byte values (e.g., if the text dump
* of packet data included a number with spaces around
* it). If the offset is less than what we expected,
* assume that's the problem, and throw away the putative
* extra byte values. (If identify_ascii is true, then
* process_rollback does something similar, except that
* it actually looks at the relevant bytes of the last
* line and works if the mistaken text is on the last
* line with an offset.)
*
* hexdump and od (but not xxd or tcpdump) print an extra
* offset with the total length followed by no bytes. This
* is necessary when those programs print multiple byte
* units instead of individual bytes. They will zero pad
* the last byte group, and the offset on the extra line
* is used to remove the excess null bytes. This also
* works for that case.
*
* In both of the above cases, the true offset should be
* between the previous line offset start and the current
* offset, as we only add bytes when a line started with
* the correct offset (except in no offset mode, where we
* don't get here.)
*
* XXX - It could be part of a preamble. There is a narrow
* window for false detection here, but it's possible if the
* line has, e.g., the packet length for the next packet.
*/
if (num < curr_offset && (num >= (uint32_t)(pkt_lnstart - packet_buf))) {
unwrite_bytes(curr_offset - num);
state = READ_OFFSET;
} else {
/* Bad offset; switch to INIT state */
report_warning("Inconsistent offset. Ending current packet.");
ws_message("Inconsistent offset. Expecting %0X, got %0X. Ending current packet (%i).",
curr_offset, num, info_p->num_packets_read);
if (write_current_packet(false) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
/* Instead of being an offset, it might be part of the preamble
* that looks like an offset, e.g. a year. (#17140)
*/
append_to_preamble(str);
state = INIT;
}
} else {
/* Continues the previous packet at the correct offset */
/* Clear Preamble (text in the middle of a packet isn't preamble) */
packet_preamble_len = 0;
state = READ_OFFSET;
}
pkt_lnstart = packet_buf + num;
break;
case T_BYTE:
if (offset_base == 0) {
if (write_byte(str) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
state = READ_BYTE;
pkt_lnstart = packet_buf;
break;
}
/* Since we expect an offset, assume this is text between packets
* that looks like a byte, e.g. a month or day.
*/
append_to_preamble(str);
break;
case T_EOF:
if (write_current_packet(false) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
break;
default:
break;
}
break;
/* ----- Processing packet, read offset -----------------------------------*/
case READ_OFFSET:
switch(token) {
case T_BYTE:
/* Record the byte */
state = READ_BYTE;
if (write_byte(str) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
break;
case T_BYTES:
state = READ_BYTE;
if (write_bytes(str) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
break;
case T_TEXT:
case T_DIRECTIVE:
case T_OFFSET:
state = READ_TEXT;
break;
case T_EOL:
state = START_OF_LINE;
break;
case T_EOF:
if (write_current_packet(false) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
break;
default:
break;
}
break;
/* ----- Processing packet, read byte -------------------------------------*/
case READ_BYTE:
switch(token) {
case T_BYTE:
/* Record the byte */
if (write_byte(str) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
break;
case T_BYTES:
if (write_bytes(str) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
break;
case T_TEXT:
case T_DIRECTIVE:
case T_OFFSET:
state = READ_TEXT;
if (info_p->hexdump.identify_ascii) {
process_rollback(false);
}
break;
case T_EOL:
state = START_OF_LINE;
if (info_p->hexdump.identify_ascii) {
process_rollback(true);
}
break;
case T_EOF:
if (write_current_packet(false) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
break;
default:
break;
}
break;
/* ----- Processing packet, read text -------------------------------------*/
case READ_TEXT:
switch(token) {
case T_EOL:
state = START_OF_LINE;
break;
case T_EOF:
if (write_current_packet(false) != IMPORT_SUCCESS)
return IMPORT_FAILURE;
break;
default:
break;
}
break;
default:
report_failure("FATAL ERROR: Bad state (%d)", state);
return IMPORT_FAILURE;
}
ws_noisy(", %s)", state_str[state]);
return IMPORT_SUCCESS;
}
/*----------------------------------------------------------------------
* Import a text file.
*/
int
text_import(text_import_info_t * const info)
{
import_status_t status;
int ret;
struct tm *now_tm;
/* Lets start from the beginning */
state = INIT;
curr_offset = 0;
packet_start = 0;
packet_preamble_len = 0;
direction = PACK_FLAGS_DIRECTION_UNKNOWN;
ts_sec = time(0); /* initialize to current time */
now_tm = localtime(&ts_sec);
if (now_tm == NULL) {
/*
* This shouldn't happen - on UN*X, this should Just Work, and
* on 32 bit Windows built with 32 bit time_t, it won't work if ts_sec
* is before the Epoch, but it's long after 1970 (and even 32 bit
* Windows builds with 64 bit time_t by default now), so....
*/
report_failure("localtime(right now) failed");
return WS_EXIT_INIT_FAILED;
}
timecode_default = *now_tm;
timecode_default.tm_isdst = -1; /* Unknown for now, depends on time given to the strptime() function */
ts_nsec = 0;
/* Get input parameters. */
info_p = info;
/* Dummy headers */
hdr_ethernet = false;
hdr_ip = false;
hdr_udp = false;
hdr_tcp = false;
hdr_sctp = false;
hdr_data_chunk = false;
hdr_export_pdu = false;
if (info->mode == TEXT_IMPORT_HEXDUMP) {
switch (info->hexdump.offset_type)
{
case OFFSET_NONE:
offset_base = 0;
break;
case OFFSET_HEX:
offset_base = 16;
break;
case OFFSET_OCT:
offset_base = 8;
break;
case OFFSET_DEC:
offset_base = 10;
break;
}
has_direction = info->hexdump.has_direction;
} else if (info->mode == TEXT_IMPORT_REGEX) {
has_direction = g_regex_get_string_number(info->regex.format, "dir") >= 0;
has_seqno = g_regex_get_string_number(info->regex.format, "seqno") >= 0;
}
if (info->timestamp_format == NULL || g_ascii_strcasecmp(info->timestamp_format, "ISO")) {
ts_fmt_iso = false;
} else {
ts_fmt_iso = true;
}
offset_warned = false;
timecode_warned = false;
/* XXX: It would be good to know the time precision of the file,
* to use for the time delta for packets without timestamps. (ts_tick)
* That could either be added to text_import_info_t or a method
* added to get it from wtap_dumper (which is opaque.)
*/
switch (info->dummy_header_type)
{
case HEADER_ETH:
hdr_ethernet = true;
hdr_ethernet_proto = info->pid;
break;
case HEADER_IPV4:
hdr_ip = true;
hdr_ip_proto = info->protocol;
break;
case HEADER_UDP:
hdr_udp = true;
hdr_tcp = false;
hdr_ip = true;
hdr_ip_proto = 17;
break;
case HEADER_TCP:
hdr_tcp = true;
hdr_udp = false;
hdr_ip = true;
hdr_ip_proto = 6;
break;
case HEADER_SCTP:
hdr_sctp = true;
hdr_ip = true;
hdr_ip_proto = 132;
break;
case HEADER_SCTP_DATA:
hdr_sctp = true;
hdr_data_chunk = true;
hdr_ip = true;
hdr_ip_proto = 132;
break;
case HEADER_EXPORT_PDU:
hdr_export_pdu = true;
break;
default:
break;
}
if (hdr_ip) {
if (info->ipv6) {
hdr_ipv6 = true;
hdr_ip = false;
hdr_ethernet_proto = 0x86DD;
} else {
hdr_ethernet_proto = 0x0800;
}
switch (info->encapsulation) {
case (WTAP_ENCAP_ETHERNET):
hdr_ethernet = true;
break;
case (WTAP_ENCAP_RAW_IP):
break;
case (WTAP_ENCAP_RAW_IP4):
if (info->ipv6) {
report_failure("Encapsulation %s only supports IPv4 headers, not IPv6", wtap_encap_name(info->encapsulation));
return WS_EXIT_INVALID_OPTION;
}
break;
case (WTAP_ENCAP_RAW_IP6):
if (!info->ipv6) {
report_failure("Encapsulation %s only supports IPv6 headers, not IPv4", wtap_encap_name(info->encapsulation));
return WS_EXIT_INVALID_OPTION;
}
break;
default:
report_failure("Dummy IP header not supported with encapsulation: %s (%s)", wtap_encap_name(info->encapsulation), wtap_encap_description(info->encapsulation));
return WS_EXIT_INVALID_OPTION;
}
}
info->num_packets_read = 0;
info->num_packets_written = 0;
packet_buf = (uint8_t *)g_malloc(sizeof(HDR_ETHERNET) + sizeof(HDR_IP) +
sizeof(HDR_SCTP) + sizeof(HDR_DATA_CHUNK) +
sizeof(HDR_EXPORT_PDU) + WTAP_MAX_PACKET_SIZE_STANDARD);
if (!packet_buf)
{
/* XXX: This doesn't happen, because g_malloc aborts the program on
* error, unlike malloc or g_try_malloc.
*/
report_failure("FATAL ERROR: no memory for packet buffer");
return WS_EXIT_INIT_FAILED;
}
if (info->mode == TEXT_IMPORT_HEXDUMP) {
status = text_import_scan(info->hexdump.import_text_FILE);
switch(status) {
case (IMPORT_SUCCESS):
ret = 0;
break;
case (IMPORT_FAILURE):
ret = WS_EXIT_INVALID_FILE;
break;
case (IMPORT_INIT_FAILED):
report_failure("Can't initialize scanner: %s", g_strerror(errno));
ret = WS_EXIT_INIT_FAILED;
break;
default:
ret = 0;
}
} else if (info->mode == TEXT_IMPORT_REGEX) {
ret = text_import_regex(info);
if (ret > 0) {
info->num_packets_read = ret;
ret = 0;
} else if (ret < 0) {
ret = WS_EXIT_INVALID_FILE;
}
} else {
ret = WS_EXIT_INVALID_OPTION;
}
g_free(packet_buf);
return ret;
}
/* Write the SHB and IDB to the wtap_dump_params before opening the wtap dump
* file. While dummy headers can be written automatically, this writes out
* some extra information including an optional interface name.
*/
int
text_import_pre_open(wtap_dump_params * const params, int file_type_subtype, const char* const input_filename, const char* const interface_name)
{
wtap_block_t shb_hdr;
wtap_block_t int_data;
wtapng_if_descr_mandatory_t *int_data_mand;
char *comment;
GString *info_str;
if (wtap_file_type_subtype_supports_block(file_type_subtype, WTAP_BLOCK_SECTION) != BLOCK_NOT_SUPPORTED &&
wtap_file_type_subtype_supports_option(file_type_subtype, WTAP_BLOCK_SECTION, OPT_COMMENT) != OPTION_NOT_SUPPORTED) {
shb_hdr = wtap_block_create(WTAP_BLOCK_SECTION);
comment = ws_strdup_printf("Generated from input file %s.", input_filename);
wtap_block_add_string_option(shb_hdr, OPT_COMMENT, comment, strlen(comment));
g_free(comment);
info_str = g_string_new("");
get_cpu_info(info_str);
if (info_str->str) {
wtap_block_add_string_option(shb_hdr, OPT_SHB_HARDWARE, info_str->str, info_str->len);
}
g_string_free(info_str, TRUE);
info_str = g_string_new("");
get_os_version_info(info_str);
if (info_str->str) {
wtap_block_add_string_option(shb_hdr, OPT_SHB_OS, info_str->str, info_str->len);
}
g_string_free(info_str, TRUE);
wtap_block_add_string_option_format(shb_hdr, OPT_SHB_USERAPPL, "%s", get_appname_and_version());
params->shb_hdrs = g_array_new(false, false, sizeof(wtap_block_t));
g_array_append_val(params->shb_hdrs, shb_hdr);
}
/* wtap_dump_init_dumper() will create a interface block if the file type
* supports it and one isn't created already, but since we have the
* option of including the interface name, create it ourself.
*
* (XXX: IDBs should be optional for wtap_dump_init_dumper(), e.g. if
* the encap type is WTAP_ENCAP_SYSTEMD_JOURNAL, which doesn't use
* interfaces. But it's not, so always create it here.)
*/
if (wtap_file_type_subtype_supports_block(file_type_subtype, WTAP_BLOCK_IF_ID_AND_INFO) != BLOCK_NOT_SUPPORTED) {
int_data = wtap_block_create(WTAP_BLOCK_IF_ID_AND_INFO);
int_data_mand = (wtapng_if_descr_mandatory_t*)wtap_block_get_mandatory_data(int_data);
int_data_mand->wtap_encap = params->encap;
int_data_mand->time_units_per_second = 1000000000;
int_data_mand->snap_len = params->snaplen;
if (interface_name != NULL) {
wtap_block_add_string_option(int_data, OPT_IDB_NAME, interface_name, strlen(interface_name));
} else {
wtap_block_add_string_option(int_data, OPT_IDB_NAME, "Fake IF, text2pcap", strlen("Fake IF, text2pcap"));
}
if (params->tsprec >= 0 && params->tsprec <= WS_TSPREC_MAX) {
/*
* This is a valid time precision.
*/
/*
* Compute 10^{params->tsprec}.
*/
int_data_mand->time_units_per_second = 1;
for (int i = 0; i < params->tsprec; i++)
int_data_mand->time_units_per_second *= 10;
if (params->tsprec != WTAP_TSPREC_USEC) {
/*
* Microsecond precision is the default, so we only
* add an option if the precision isn't microsecond
* precision.
*/
wtap_block_add_uint8_option(int_data, OPT_IDB_TSRESOL, params->tsprec);
}
} else {
/*
* Either WTAP_TSPREC_PER_PACKET, WTAP_TSPREC_UNKNOWN,
* or not a valid precision.
*
* Don't do this.
*/
ws_assert_not_reached();
}
params->idb_inf = g_new(wtapng_iface_descriptions_t,1);
params->idb_inf->interface_data = g_array_new(false, false, sizeof(wtap_block_t));
g_array_append_val(params->idb_inf->interface_data, int_data);
}
return EXIT_SUCCESS;
}
Reference in New Issue View Git Blame Copy Permalink