initial working parser for both linux and bsd

jc/parsers/ping_s.py

@@ -56,6 +56,7 @@ Examples:
     ...
 """
 import string
+import ipaddress
 import jc.utils
 from jc.utils import stream_success, stream_error
 
@@ -102,23 +103,12 @@ def _process(proc_data):
     return proc_data
 
 
-def parse(data, raw=False, quiet=False):
+class state:
-    """
+    os_detected = None
-    Main text parsing generator function. Produces an iterable object.
+    linux = None
-
+    bsd = None
-    Parameters:
+    ipv4 = None
-
+    hostname = None
-        data:        (string)  line-based text data to parse
-        raw:         (boolean) output preprocessed JSON if True
-        quiet:       (boolean) suppress warning messages and ignore parsing errors if True
-
-    Yields:
-
-        Dictionary. Raw or processed structured data.
-    """
-    if not quiet:
-        jc.utils.compatibility(__name__, info.compatible)
-
     destination_ip = None
     sent_bytes = None
     pattern = None
@@ -128,88 +118,269 @@ def parse(data, raw=False, quiet=False):
     packet_loss_percent = None
     time_ms = None
     duplicates = None
+    ping_error = None
 
-    for line in data:
+
+def _ipv6_in(line):
+    line_list = line.replace('(', ' ').replace(')', ' ').replace(',', ' ').replace('%', ' ').split()
+    ipv6 = False
+    for item in line_list:
         try:
+            _ = ipaddress.IPv6Address(item)
+            ipv6 = True
+        except Exception:
+            pass
+    return ipv6
+
+
+def _error_type(line):
+    # from https://github.com/dgibson/iputils/blob/master/ping.c
+    # https://android.googlesource.com/platform/external/ping/+/8fc3c91cf9e7f87bc20b9e6d3ea2982d87b70d9a/ping.c
+    # https://opensource.apple.com/source/network_cmds/network_cmds-328/ping.tproj/ping.c
+    type_map = {
+        'Destination Net Unreachable': 'destination_net_unreachable',
+        'Destination Host Unreachable': 'destination_host_unreachable',
+        'Destination Protocol Unreachable': 'destination_protocol_unreachable',
+        'Destination Port Unreachable': 'destination_port_unreachable',
+        'Frag needed and DF set': 'frag_needed_and_df_set',
+        'Source Route Failed': 'source_route_failed',
+        'Destination Net Unknown': 'destination_net_unknown',
+        'Destination Host Unknown': 'destination_host_unknown',
+        'Source Host Isolated': 'source_host_isolated',
+        'Destination Net Prohibited': 'destination_net_prohibited',
+        'Destination Host Prohibited': 'destination_host_prohibited',
+        'Destination Net Unreachable for Type of Service': 'destination_net_unreachable_for_type_of_service',
+        'Destination Host Unreachable for Type of Service': 'destination_host_unreachable_for_type_of_service',
+        'Packet filtered': 'packet_filtered',
+        'Precedence Violation': 'precedence_violation',
+        'Precedence Cutoff': 'precedence_cutoff',
+        'Dest Unreachable, Bad Code': 'dest_unreachable_bad_code',
+        'Redirect Network': 'redirect_network',
+        'Redirect Host': 'redirect_host',
+        'Redirect Type of Service and Network': 'redirect_type_of_service_and_network',
+        'Redirect, Bad Code': 'redirect_bad_code',
+        'Time to live exceeded': 'time_to_live_exceeded',
+        'Frag reassembly time exceeded': 'frag_reassembly_time_exceeded',
+        'Time exceeded, Bad Code': 'time_exceeded_bad_code'
+    }
+
+    for err_type, code in type_map.items():
+        if err_type in line:
+            return code
+
+
+def _bsd_parse(line, s):
     output_line = {}
 
-            # check for PATTERN
-            if line.startswith('PATTERN: '):
-                pattern = line.strip().split(': ')[1]
-                continue
     
     if line.startswith('PING '):
-                ipv4 = True if 'bytes of data' in line else False
+        s.destination_ip = line.split()[2].lstrip('(').rstrip(':').rstrip(')')
+        s.sent_bytes = line.split()[3]
+        return None
+
+    if line.startswith('PING6('):
+        line = line.replace('(', ' ').replace(')', ' ').replace('=', ' ')
+        s.source_ip = line.split()[4]
+        s.destination_ip = line.split()[6]
+        s.sent_bytes = line.split()[1]
+        return None
+
+    if line.startswith('---'):
+        s.footer = True
+        return None
+
+    if s.footer:
+        if 'packets transmitted' in line:
+            if ' duplicates,' in line:
+                s.packets_transmitted = line.split()[0]
+                s.packets_received = line.split()[3]
+                s.packet_loss_percent = line.split()[8].rstrip('%')
+                s.duplicates = line.split()[6].lstrip('+')
+                return None
             
-                if ipv4 and line[5] not in string.digits:
-                    hostname = True
-                elif ipv4 and line[5] in string.digits:
-                    hostname = False
-                elif not ipv4 and ' (' in line:
-                    hostname = True
             else:
-                    hostname = False
+                s.packets_transmitted = line.split()[0]
+                s.packets_received = line.split()[3]
+                s.packet_loss_percent = line.split()[6].rstrip('%')
+                s.duplicates = '0'
+                return None
 
-                if ipv4 and not hostname:
+        else:
+            split_line = line.split(' = ')[1]
+            split_line = split_line.split('/')
+
+            output_line = {
+                'type': 'summary',
+                'destination_ip': s.destination_ip or None,
+                'sent_bytes': s.sent_bytes or None,
+                'pattern': s.pattern or None,
+                'packets_transmitted': s.packets_transmitted or None,
+                'packets_received': s.packets_received or None,
+                'packet_loss_percent': s.packet_loss_percent or None,
+                'duplicates': s.duplicates or None,
+                'round_trip_ms_min': split_line[0],
+                'round_trip_ms_avg': split_line[1],
+                'round_trip_ms_max': split_line[2],
+                'round_trip_ms_stddev': split_line[3].replace(' ms', '')
+            }
+
+            return output_line
+
+    # ping response lines
+    else:
+        # ipv4 lines
+        if not _ipv6_in(line):
+
+            # request timeout
+            if line.startswith('Request timeout for '):
+                output_line = {
+                    'type': 'timeout',
+                    'destination_ip': s.destination_ip or None,
+                    'sent_bytes': s.sent_bytes or None,
+                    'pattern': s.pattern or None,
+                    'icmp_seq': line.split()[4]
+                }
+                
+                return output_line
+
+            # catch error responses
+            err = _error_type(line)
+            if err:
+                output_line = {
+                    'type': err
+                }
+
+                try:
+                    output_line['bytes'] = line.split()[0]
+                    output_line['response_ip'] = line.split()[4].strip(':').strip('(').strip(')')
+                except Exception:
+                    pass
+
+                return output_line
+
+            # normal response
+            elif ' bytes from ' in line:
+                line = line.replace(':', ' ').replace('=', ' ')
+
+                output_line = {
+                    'type': 'reply',
+                    'destination_ip': s.destination_ip or None,
+                    'sent_bytes': s.sent_bytes or None,
+                    'pattern': s.pattern or None,
+                    'response_bytes': line.split()[0],
+                    'response_ip': line.split()[3],
+                    'icmp_seq': line.split()[5],
+                    'ttl': line.split()[7],
+                    'time_ms': line.split()[9]
+                }
+
+                return output_line
+
+        # ipv6 lines
+        elif ' bytes from ' in line:
+            line = line.replace(',', ' ').replace('=', ' ')
+            
+            output_line = {
+                'type': 'reply',
+                'destination_ip': s.destination_ip or None,
+                'sent_bytes': s.sent_bytes or None,
+                'pattern': s.pattern or None,
+                'bytes': line.split()[0],
+                'response_ip': line.split()[3],
+                'icmp_seq': line.split()[5],
+                'ttl': line.split()[7],
+                'time_ms': line.split()[9]
+            }
+
+            return output_line
+
+
+def _linux_parse(line, s):
+    """
+    Linux ping line parsing function.
+
+    Parameters:
+
+        line:        (string)        line of text data to parse
+        s:           (state object)  global state
+
+    Returns:
+
+        Dictionary. Raw structured data.
+    """
+    output_line = {}
+
+    if line.startswith('PING '):
+        s.ipv4 = True if 'bytes of data' in line else False
+
+        if s.ipv4 and line[5] not in string.digits:
+            s.hostname = True
+        elif s.ipv4 and line[5] in string.digits:
+            s.hostname = False
+        elif not s.ipv4 and ' (' in line:
+            s.hostname = True
+        else:
+            s.hostname = False
+
+        if s.ipv4 and not s.hostname:
             dst_ip, dta_byts = (2, 3)
-                elif ipv4 and hostname:
+        elif s.ipv4 and s.hostname:
             dst_ip, dta_byts = (2, 3)
-                elif not ipv4 and not hostname:
+        elif not s.ipv4 and not s.hostname:
             dst_ip, dta_byts = (2, 3)
         else:
             dst_ip, dta_byts = (3, 4)
 
         line = line.replace('(', ' ').replace(')', ' ')
-                destination_ip = line.split()[dst_ip].lstrip('(').rstrip(')')
+        s.destination_ip = line.split()[dst_ip].lstrip('(').rstrip(')')
-                sent_bytes = line.split()[dta_byts]
+        s.sent_bytes = line.split()[dta_byts]
 
-                continue
+        return None
 
     if line.startswith('---'):
-                footer = True
+        s.footer = True
-                continue
+        return None
 
-            if footer:
+    if s.footer:
         if 'packets transmitted' in line:
             if ' duplicates,' in line:
-                        packets_transmitted = line.split()[0]
+                s.packets_transmitted = line.split()[0]
-                        packets_received = line.split()[3]
+                s.packets_received = line.split()[3]
-                        packet_loss_percent = line.split()[7].rstrip('%')
+                s.packet_loss_percent = line.split()[7].rstrip('%')
-                        duplicates = line.split()[5].lstrip('+')
+                s.duplicates = line.split()[5].lstrip('+')
-                        time_ms = line.split()[11].replace('ms', '')
+                s.time_ms = line.split()[11].replace('ms', '')
 
-                        continue
+                return None
 
             else:
-                        packets_transmitted = line.split()[0]
+                s.packets_transmitted = line.split()[0]
-                        packets_received = line.split()[3]
+                s.packets_received = line.split()[3]
-                        packet_loss_percent = line.split()[5].rstrip('%')
+                s.packet_loss_percent = line.split()[5].rstrip('%')
-                        duplicates = '0'
+                s.duplicates = '0'
-                        time_ms = line.split()[9].replace('ms', '')
+                s.time_ms = line.split()[9].replace('ms', '')
                 
-                        continue
+                return None
 
         else:
             split_line = line.split(' = ')[1]
             split_line = split_line.split('/')
             output_line = {
                 'type': 'summary',
-                        'destination_ip': destination_ip or None,
+                'destination_ip': s.destination_ip or None,
-                        'sent_bytes': sent_bytes or None,
+                'sent_bytes': s.sent_bytes or None,
-                        'pattern': pattern or None,
+                'pattern': s.pattern or None,
-                        'packets_transmitted': packets_transmitted or None,
+                'packets_transmitted': s.packets_transmitted or None,
-                        'packets_received': packets_received or None,
+                'packets_received': s.packets_received or None,
-                        'packet_loss_percent': packet_loss_percent or None,
+                'packet_loss_percent': s.packet_loss_percent or None,
-                        'duplicates': duplicates or None,
+                'duplicates': s.duplicates or None,
-                        'time_ms': time_ms or None,
+                'time_ms': s.time_ms or None,
                 'round_trip_ms_min': split_line[0],
                 'round_trip_ms_avg': split_line[1],
                 'round_trip_ms_max': split_line[2],
                 'round_trip_ms_stddev': split_line[3].split()[0]
             }
 
-                    yield stream_success(output_line, quiet) if raw else stream_success(_process(output_line), quiet)
+            return output_line
-                    continue
 
     # ping response lines
     else:
@@ -225,15 +396,14 @@ def parse(data, raw=False, quiet=False):
 
             output_line = {
                 'type': 'timeout',
-                        'destination_ip': destination_ip or None,
+                'destination_ip': s.destination_ip or None,
-                        'sent_bytes': sent_bytes or None,
+                'sent_bytes': s.sent_bytes or None,
-                        'pattern': pattern or None,
+                'pattern': s.pattern or None,
                 'timestamp': line.split()[0].lstrip('[').rstrip(']') if timestamp else None,
                 'icmp_seq': line.replace('=', ' ').split()[isequence]
             }
             
-                    yield stream_success(output_line, quiet) if raw else stream_success(_process(output_line), quiet)
+            return output_line
-                    continue
 
         # normal responses
         elif ' bytes from ' in line:
@@ -241,13 +411,13 @@ def parse(data, raw=False, quiet=False):
             line = line.replace('(', ' ').replace(')', ' ').replace('=', ' ')
 
             # positions of items depend on whether ipv4/ipv6 and/or ip/hostname is used
-                    if ipv4 and not hostname:
+            if s.ipv4 and not s.hostname:
                 bts, rip, iseq, t2l, tms = (0, 3, 5, 7, 9)
-                    elif ipv4 and hostname:
+            elif s.ipv4 and s.hostname:
                 bts, rip, iseq, t2l, tms = (0, 4, 7, 9, 11)
-                    elif not ipv4 and not hostname:
+            elif not s.ipv4 and not s.hostname:
                 bts, rip, iseq, t2l, tms = (0, 3, 5, 7, 9)
-                    elif not ipv4 and hostname:
+            elif not s.ipv4 and s.hostname:
                 bts, rip, iseq, t2l, tms = (0, 4, 7, 9, 11)
 
             # if timestamp option is specified, then shift everything right by one
@@ -258,9 +428,9 @@ def parse(data, raw=False, quiet=False):
 
             output_line = {
                 'type': 'reply',
-                        'destination_ip': destination_ip or None,
+                'destination_ip': s.destination_ip or None,
-                        'sent_bytes': sent_bytes or None,
+                'sent_bytes': s.sent_bytes or None,
-                        'pattern': pattern or None,
+                'pattern': s.pattern or None,
                 'timestamp': line.split()[0].lstrip('[').rstrip(']') if timestamp else None,
                 'response_bytes': line.split()[bts],
                 'response_ip': line.split()[rip].rstrip(':'),
@@ -270,7 +440,67 @@ def parse(data, raw=False, quiet=False):
                 'duplicate': True if 'DUP!' in line else False
             }
 
+            return output_line
+
+
+def parse(data, raw=False, quiet=False):
+    """
+    Main text parsing generator function. Produces an iterable object.
+
+    Parameters:
+
+        data:        (string)  line-based text data to parse
+        raw:         (boolean) output preprocessed JSON if True
+        quiet:       (boolean) suppress warning messages and ignore parsing errors if True
+
+    Yields:
+
+        Dictionary. Raw or processed structured data.
+    """
+    s = state()
+
+    if not quiet:
+        jc.utils.compatibility(__name__, info.compatible)
+
+    for line in data:
+
+        output_line = {}
+
+        try:
+            # check for PATTERN
+            if line.startswith('PATTERN: '):
+                s.pattern = line.strip().split(': ')[1]
+                continue
+
+            # detect Linux vs. BSD ping
+            if not s.os_detected and line.strip().endswith('bytes of data.'):
+                s.os_detected = True
+                s.linux = True
+
+            if not s.os_detected and '-->' in line:
+                s.os_detected = True
+                s.bsd = True
+
+            if not s.os_detected and _ipv6_in(line) and line.strip().endswith('data bytes'):
+                s.os_detected = True
+                s.linux = True
+
+            if not s.os_detected:
+                s.os_detected = True
+                s.bsd = True
+
+            # parse the data
+            if s.os_detected and s.linux:
+                output_line = _linux_parse(line, s)
+
+            if s.os_detected and s.bsd:
+                output_line = _bsd_parse(line, s)
+
+            # yield the output line if it has data
+            if output_line:
                 yield stream_success(output_line, quiet) if raw else stream_success(_process(output_line), quiet)
+            else:
+                continue
 
         except Exception as e:
             yield stream_error(e, quiet, line)