extras/errorlog/summarise-log.pl

#!/usr/bin/perl

$verbose=0;
$verbose=1 if($#ARGV==0 && $ARGV[0] eq "-v");

$html=0;
$html=1 if($#ARGV==0 && $ARGV[0] eq "-html");

die "Usage: $0 [-v | -html] < <error-log-file>\n" if($#ARGV>0 || ($#ARGV==0 && !$verbose && !$html));


# Read in each line from the error log and store them

%errors=();
%errorids=();
%errortypes=();

while(<STDIN>)
  {
   s%\r*\n%%;

   undef $errorid;

   if(m%nodes ([0-9]+) and ([0-9]+) in way ([0-9]+)%i) # Special case pair of nodes and a way
     {
      $errorid="($1 $2 $3)";
      $errortype="N2W";
      s%nodes [0-9]+ and [0-9]+ in way [0-9]+%nodes <node-id1> and <node-id2> in way <way-id>%;
     }

   elsif(m%node ([0-9]+) in way ([0-9]+)%i) # Special case node and a way
     {
      $errorid="($1 $2)";
      $errortype="NW";
      s%Node [0-9]+ in way [0-9]+%Node <node-id> in way <way-id>%;
     }

   elsif(m%nodes ([0-9]+) and ([0-9]+)%i) # Special case pair of nodes
     {
      $errorid="($1 $2)";
      $errortype="N2";
      s%nodes [0-9]+ and [0-9]+%nodes <node-id1> and <node-id2>%;
     }

   elsif(m%Segment (contains|connects) node ([0-9]+)%) # Special case node
     {
      $errorid=$2;
      $errortype="N";
      s%node [0-9]+%node <node-id>%;
     }

   elsif(m%Relation ([0-9]+).* contains Node ([0-9]+)%) # Special case relation/node
     {
      $errorid="($1 $2)";
      $errortype="RN";
      s%Relation [0-9]+%Relation <relation-id>%;
      s%Node [0-9]+%node <node-id>%;
     }

   elsif(m%Relation ([0-9]+).* contains Way ([0-9]+)%) # Generic case relation/way
     {
      $errorid="($1 $2)";
      $errortype="RW";
      s%Relation [0-9]+%Relation <relation-id>%;
      s%Way [0-9]+%way <way-id>%;
     }

   elsif(!m%Way ([0-9]+)% && !m%Relation ([0-9]+)% && m%Node ([0-9]+)%) # Generic node
     {
      $errorid=$1;
      $errortype="N";
      s%Node [0-9]+%Node <node-id>%;
     }

   elsif(!m%Node ([0-9]+)% && !m%Relation ([0-9]+)% && m%Way ([0-9]+)%) # Generic way
     {
      $errorid=$1;
      $errortype="W";
      s%Way [0-9]+%Way <way-id>%;
     }

   elsif(!m%Node ([0-9]+)% && !m%Way ([0-9]+)% && m%Relation ([0-9]+)%) # Generic relation
     {
      $errorid=$1;
      $errortype="R";
      s%Relation [0-9]+%Relation <relation-id>%;
     }

   else
     {
      $errorid="ERROR";
      $errortype="E";
      warn "Unrecognised error message '$_'\n";
     }

   $errors{$_}++;

   if($verbose || $html)
     {
      if(defined $errorids{$_})
        {
         push(@{$errorids{$_}},$errorid);
        }
      else
        {
         $errorids{$_}=[$errorid];
        }
     }

   if($html)
     {
      $errortypes{$_}=$errortype;
     }
  }


# Print out the results as text

if( ! $html )
  {

   foreach $error (sort { if ( $errors{$b} == $errors{$a} ) { return $errors{$a} cmp $errors{$b} }
                          else                              { return $errors{$b} <=> $errors{$a} } } (keys %errors))
     {
      printf "%9d : $error\n",$errors{$error};

      if($verbose)
        {
         @ids=sort({ return $a <=> $b } @{$errorids{$error}});

         print "            ".join(",",@ids)."\n";
        }
     }

  }

# Print out the results as HTML

else
  {

   print "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">\n".
         "<HTML>\n".
         "\n".
         "<HEAD>\n".
         "<TITLE>Routino Error Log File Summary</TITLE>\n".
         "<META http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\">\n".
         "<STYLE type=\"text/css\">\n".
         "<!--\n".
         "   body {font-family: sans-serif; font-size: 12px;}\n".
         "   h1   {font-family: sans-serif; font-size: 14px; font-style: bold;}\n".
         "   h2   {font-family: sans-serif; font-size: 13px; font-style: bold;}\n".
         "   h3   {font-family: sans-serif; font-size: 12px; font-style: bold;}\n".
         "-->\n".
         "</STYLE>\n".
         "</HEAD>\n".
         "\n".
         "<BODY>\n".
         "\n".
         "<h1>Routino Error Log File Summary</h1>\n".
         "\n".
         "This HTML file contains a summary of the Routino OSM parser error log file with\n".
         "links to the OSM website that allow browsing each of the nodes, ways or relations\n".
         "that are responsible for the error messages.\n".
         "\n";

   %errortypeorder=(
                    "N"   , 1,
                    "NW"  , 2,
                    "N2W" , 3,
                    "N2"  , 4,
                    "W"   , 5,
                    "R"   , 6,
                    "RN"  , 7,
                    "RW"  , 8,
                    "E"   , 9
                   );

   %errortypelabel=(
                    "N"   , "Nodes",
                    "NW"  , "Node in a Way",
                    "N2W" , "Node Pairs in a Way",
                    "N2"  , "Node Pairs",
                    "W"   , "Ways",
                    "R"   , "Relations",
                    "RN"  , "Relations/Nodes",
                    "RW"  , "Relations/Ways",
                    "E"   , "ERROR"
                   );

   $lasterrortype="";

   foreach $error (sort { if    ( $errortypes{$b} ne $errortypes{$a} ) { return $errortypeorder{$errortypes{$a}} <=> $errortypeorder{$errortypes{$b}} }
                          elsif ( $errors{$b}     == $errors{$a} )     { return $errors{$a} cmp $errors{$b} }
                          else                                         { return $errors{$b} <=> $errors{$a} } } (keys %errors))
     {
      $errorhtml=$error;

      $errorhtml =~ s/&/&amp;/g;
      $errorhtml =~ s/</&lt;/g;
      $errorhtml =~ s/>/&gt;/g;

      if($errortypes{$error} ne $lasterrortype)
        {
         print "<h2>$errortypelabel{$errortypes{$error}}</h2>\n";
         $lasterrortype=$errortypes{$error};
        }

      print "<h3>$errorhtml</h3>\n";

      if($errors{$error}>100)
        {
         print "$errors{$error} occurences (not listed).\n";
        }
      else
        {
         @ids=sort({ return $a <=> $b } @{$errorids{$error}});

         $first=1;

         foreach $id (@ids)
           {
            if($first)
              {
               print "$errortypelabel{$errortypes{$error}}:\n";
              }
            else
              {
               print ",";
              }

            $first=0;

            print "<a href=\"http://www.openstreetmap.org/browse/node/$id\">$id</a>" if($errortypes{$error} eq "N");
            print "<a href=\"http://www.openstreetmap.org/browse/way/$id\">$id</a>" if($errortypes{$error} eq "W");
            print "<a href=\"http://www.openstreetmap.org/browse/relation/$id\">$id</a>" if($errortypes{$error} eq "R");

            if($errortypes{$error} eq "NW" || $errortypes{$error} eq "N2" || $errortypes{$error} eq "RN" || $errortypes{$error} eq "RW")
              {
               $id =~ m%\(([0-9]+) ([0-9]+)\)%;
               print "(<a href=\"http://www.openstreetmap.org/browse/node/$1\">$1</a> <a href=\"http://www.openstreetmap.org/browse/way/$2\">$2</a>)" if($errortypes{$error} eq "NW");
               print "(<a href=\"http://www.openstreetmap.org/browse/node/$1\">$1</a> <a href=\"http://www.openstreetmap.org/browse/node/$2\">$2</a>)" if($errortypes{$error} eq "N2");
               print "(<a href=\"http://www.openstreetmap.org/browse/relation/$1\">$1</a> <a href=\"http://www.openstreetmap.org/browse/node/$2\">$2</a>)" if($errortypes{$error} eq "RN");
               print "(<a href=\"http://www.openstreetmap.org/browse/relation/$1\">$1</a> <a href=\"http://www.openstreetmap.org/browse/way/$2\">$2</a>)" if($errortypes{$error} eq "RW");
              }

            if($errortypes{$error} eq "N2W")
              {
               $id =~ m%\(([0-9]+) ([0-9]+) ([0-9]+)\)%;
               print "(<a href=\"http://www.openstreetmap.org/browse/node/$1\">$1</a> <a href=\"http://www.openstreetmap.org/browse/node/$2\">$2</a> <a href=\"http://www.openstreetmap.org/browse/way/$3\">$3</a>)" if($errortypes{$error} eq "N2W");
              }

            print "\n";
           }
        }
     }

   print "\n".
         "</BODY>\n".
         "\n".
         "</HTML>\n";

}
1	#!/usr/bin/perl
2
3	$verbose=0;
4	$verbose=1 if($#ARGV==0 && $ARGV[0] eq "-v");
5
6	$html=0;
7	$html=1 if($#ARGV==0 && $ARGV[0] eq "-html");
8
9	die "Usage: $0 [-v \| -html] < <error-log-file>\n" if($#ARGV>0 \|\| ($#ARGV==0 && !$verbose && !$html));
10
11
12	# Read in each line from the error log and store them
13
14	%errors=();
15	%errorids=();
16	%errortypes=();
17
18	while(<STDIN>)
19	{
20	s%\r*\n%%;
21
22	undef $errorid;
23
24	if(m%nodes ([0-9]+) and ([0-9]+) in way ([0-9]+)%i) # Special case pair of nodes and a way
25	{
26	$errorid="($1 $2 $3)";
27	$errortype="N2W";
28	s%nodes [0-9]+ and [0-9]+ in way [0-9]+%nodes <node-id1> and <node-id2> in way <way-id>%;
29	}
30
31	elsif(m%node ([0-9]+) in way ([0-9]+)%i) # Special case node and a way
32	{
33	$errorid="($1 $2)";
34	$errortype="NW";
35	s%Node [0-9]+ in way [0-9]+%Node <node-id> in way <way-id>%;
36	}
37
38	elsif(m%nodes ([0-9]+) and ([0-9]+)%i) # Special case pair of nodes
39	{
40	$errorid="($1 $2)";
41	$errortype="N2";
42	s%nodes [0-9]+ and [0-9]+%nodes <node-id1> and <node-id2>%;
43	}
44
45	elsif(m%Segment (contains\|connects) node ([0-9]+)%) # Special case node
46	{
47	$errorid=$2;
48	$errortype="N";
49	s%node [0-9]+%node <node-id>%;
50	}
51
52	elsif(m%Relation ([0-9]+).* contains Node ([0-9]+)%) # Special case relation/node
53	{
54	$errorid="($1 $2)";
55	$errortype="RN";
56	s%Relation [0-9]+%Relation <relation-id>%;
57	s%Node [0-9]+%node <node-id>%;
58	}
59
60	elsif(m%Relation ([0-9]+).* contains Way ([0-9]+)%) # Generic case relation/way
61	{
62	$errorid="($1 $2)";
63	$errortype="RW";
64	s%Relation [0-9]+%Relation <relation-id>%;
65	s%Way [0-9]+%way <way-id>%;
66	}
67
68	elsif(!m%Way ([0-9]+)% && !m%Relation ([0-9]+)% && m%Node ([0-9]+)%) # Generic node
69	{
70	$errorid=$1;
71	$errortype="N";
72	s%Node [0-9]+%Node <node-id>%;
73	}
74
75	elsif(!m%Node ([0-9]+)% && !m%Relation ([0-9]+)% && m%Way ([0-9]+)%) # Generic way
76	{
77	$errorid=$1;
78	$errortype="W";
79	s%Way [0-9]+%Way <way-id>%;
80	}
81
82	elsif(!m%Node ([0-9]+)% && !m%Way ([0-9]+)% && m%Relation ([0-9]+)%) # Generic relation
83	{
84	$errorid=$1;
85	$errortype="R";
86	s%Relation [0-9]+%Relation <relation-id>%;
87	}
88
89	else
90	{
91	$errorid="ERROR";
92	$errortype="E";
93	warn "Unrecognised error message '$_'\n";
94	}
95
96	$errors{$_}++;
97
98	if($verbose \|\| $html)
99	{
100	if(defined $errorids{$_})
101	{
102	push(@{$errorids{$_}},$errorid);
103	}
104	else
105	{
106	$errorids{$_}=[$errorid];
107	}
108	}
109
110	if($html)
111	{
112	$errortypes{$_}=$errortype;
113	}
114	}
115
116
117	# Print out the results as text
118
119	if( ! $html )
120	{
121
122	foreach $error (sort { if ( $errors{$b} == $errors{$a} ) { return $errors{$a} cmp $errors{$b} }
123	else { return $errors{$b} <=> $errors{$a} } } (keys %errors))
124	{
125	printf "%9d : $error\n",$errors{$error};
126
127	if($verbose)
128	{
129	@ids=sort({ return $a <=> $b } @{$errorids{$error}});
130
131	print " ".join(",",@ids)."\n";
132	}
133	}
134
135	}
136
137	# Print out the results as HTML
138
139	else
140	{
141
142	print "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">\n".
143	"<HTML>\n".
144	"\n".
145	"<HEAD>\n".
146	"<TITLE>Routino Error Log File Summary</TITLE>\n".
147	"<META http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\">\n".
148	"<STYLE type=\"text/css\">\n".
149	"<!--\n".
150	" body {font-family: sans-serif; font-size: 12px;}\n".
151	" h1 {font-family: sans-serif; font-size: 14px; font-style: bold;}\n".
152	" h2 {font-family: sans-serif; font-size: 13px; font-style: bold;}\n".
153	" h3 {font-family: sans-serif; font-size: 12px; font-style: bold;}\n".
154	"-->\n".
155	"</STYLE>\n".
156	"</HEAD>\n".
157	"\n".
158	"<BODY>\n".
159	"\n".
160	"<h1>Routino Error Log File Summary</h1>\n".
161	"\n".
162	"This HTML file contains a summary of the Routino OSM parser error log file with\n".
163	"links to the OSM website that allow browsing each of the nodes, ways or relations\n".
164	"that are responsible for the error messages.\n".
165	"\n";
166
167	%errortypeorder=(
168	"N" , 1,
169	"NW" , 2,
170	"N2W" , 3,
171	"N2" , 4,
172	"W" , 5,
173	"R" , 6,
174	"RN" , 7,
175	"RW" , 8,
176	"E" , 9
177	);
178
179	%errortypelabel=(
180	"N" , "Nodes",
181	"NW" , "Node in a Way",
182	"N2W" , "Node Pairs in a Way",
183	"N2" , "Node Pairs",
184	"W" , "Ways",
185	"R" , "Relations",
186	"RN" , "Relations/Nodes",
187	"RW" , "Relations/Ways",
188	"E" , "ERROR"
189	);
190
191	$lasterrortype="";
192
193	foreach $error (sort { if ( $errortypes{$b} ne $errortypes{$a} ) { return $errortypeorder{$errortypes{$a}} <=> $errortypeorder{$errortypes{$b}} }
194	elsif ( $errors{$b} == $errors{$a} ) { return $errors{$a} cmp $errors{$b} }
195	else { return $errors{$b} <=> $errors{$a} } } (keys %errors))
196	{
197	$errorhtml=$error;
198
199	$errorhtml =~ s/&/&/g;
200	$errorhtml =~ s/</</g;
201	$errorhtml =~ s/>/>/g;
202
203	if($errortypes{$error} ne $lasterrortype)
204	{
205	print "<h2>$errortypelabel{$errortypes{$error}}</h2>\n";
206	$lasterrortype=$errortypes{$error};
207	}
208
209	print "<h3>$errorhtml</h3>\n";
210
211	if($errors{$error}>100)
212	{
213	print "$errors{$error} occurences (not listed).\n";
214	}
215	else
216	{
217	@ids=sort({ return $a <=> $b } @{$errorids{$error}});
218
219	$first=1;
220
221	foreach $id (@ids)
222	{
223	if($first)
224	{
225	print "$errortypelabel{$errortypes{$error}}:\n";
226	}
227	else
228	{
229	print ",";
230	}
231
232	$first=0;
233
234	print "<a href=\"http://www.openstreetmap.org/browse/node/$id\">$id</a>" if($errortypes{$error} eq "N");
235	print "<a href=\"http://www.openstreetmap.org/browse/way/$id\">$id</a>" if($errortypes{$error} eq "W");
236	print "<a href=\"http://www.openstreetmap.org/browse/relation/$id\">$id</a>" if($errortypes{$error} eq "R");
237
238	if($errortypes{$error} eq "NW" \|\| $errortypes{$error} eq "N2" \|\| $errortypes{$error} eq "RN" \|\| $errortypes{$error} eq "RW")
239	{
240	$id =~ m%\(([0-9]+) ([0-9]+)\)%;
241	print "(<a href=\"http://www.openstreetmap.org/browse/node/$1\">$1</a> <a href=\"http://www.openstreetmap.org/browse/way/$2\">$2</a>)" if($errortypes{$error} eq "NW");
242	print "(<a href=\"http://www.openstreetmap.org/browse/node/$1\">$1</a> <a href=\"http://www.openstreetmap.org/browse/node/$2\">$2</a>)" if($errortypes{$error} eq "N2");
243	print "(<a href=\"http://www.openstreetmap.org/browse/relation/$1\">$1</a> <a href=\"http://www.openstreetmap.org/browse/node/$2\">$2</a>)" if($errortypes{$error} eq "RN");
244	print "(<a href=\"http://www.openstreetmap.org/browse/relation/$1\">$1</a> <a href=\"http://www.openstreetmap.org/browse/way/$2\">$2</a>)" if($errortypes{$error} eq "RW");
245	}
246
247	if($errortypes{$error} eq "N2W")
248	{
249	$id =~ m%\(([0-9]+) ([0-9]+) ([0-9]+)\)%;
250	print "(<a href=\"http://www.openstreetmap.org/browse/node/$1\">$1</a> <a href=\"http://www.openstreetmap.org/browse/node/$2\">$2</a> <a href=\"http://www.openstreetmap.org/browse/way/$3\">$3</a>)" if($errortypes{$error} eq "N2W");
251	}
252
253	print "\n";
254	}
255	}
256	}
257
258	print "\n".
259	"</BODY>\n".
260	"\n".
261	"</HTML>\n";
262
263	}