web/bin/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 $errorids{$_})
        {
         $errorids{$_}.=",$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 $errorids{$a} cmp $errorids{$b} }
                          else                              { return $errors{$b}   <=> $errors{$a}   } } (keys %errors))
     {
      printf "%9d : $error\n",$errors{$error};

      if($verbose && defined $errorids{$error})
        {
         print "            $errorids{$error}\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 $errorids{$a} cmp $errorids{$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=split(",",$errorids{$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	$errorids{$_}.=",$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 $errorids{$a} cmp $errorids{$b} }
123	else { return $errors{$b} <=> $errors{$a} } } (keys %errors))
124	{
125	printf "%9d : $error\n",$errors{$error};
126
127	if($verbose && defined $errorids{$error})
128	{
129	print " $errorids{$error}\n";
130	}
131	}
132
133	}
134
135	# Print out the results as HTML
136
137	else
138	{
139
140	print "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">\n".
141	"<HTML>\n".
142	"\n".
143	"<HEAD>\n".
144	"<TITLE>Routino Error Log File Summary</TITLE>\n".
145	"<META http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\">\n".
146	"<STYLE type=\"text/css\">\n".
147	"<!--\n".
148	" body {font-family: sans-serif; font-size: 12px;}\n".
149	" h1 {font-family: sans-serif; font-size: 14px; font-style: bold;}\n".
150	" h2 {font-family: sans-serif; font-size: 13px; font-style: bold;}\n".
151	" h3 {font-family: sans-serif; font-size: 12px; font-style: bold;}\n".
152	"-->\n".
153	"</STYLE>\n".
154	"</HEAD>\n".
155	"\n".
156	"<BODY>\n".
157	"\n".
158	"<h1>Routino Error Log File Summary</h1>\n".
159	"\n".
160	"This HTML file contains a summary of the Routino OSM parser error log file with\n".
161	"links to the OSM website that allow browsing each of the nodes, ways or relations\n".
162	"that are responsible for the error messages.\n".
163	"\n";
164
165	%errortypeorder=(
166	"N" , 1,
167	"NW" , 2,
168	"N2W" , 3,
169	"N2" , 4,
170	"W" , 5,
171	"R" , 6,
172	"RN" , 7,
173	"RW" , 8,
174	"E" , 9
175	);
176
177	%errortypelabel=(
178	"N" , "Nodes",
179	"NW" , "Node in a Way",
180	"N2W" , "Node Pairs in a Way",
181	"N2" , "Node Pairs",
182	"W" , "Ways",
183	"R" , "Relations",
184	"RN" , "Relations/Nodes",
185	"RW" , "Relations/Ways",
186	"E" , "ERROR"
187	);
188
189	$lasterrortype="";
190
191	foreach $error (sort { if ( $errortypes{$b} ne $errortypes{$a} ) { return $errortypeorder{$errortypes{$a}} <=> $errortypeorder{$errortypes{$b}} }
192	elsif ( $errors{$b} == $errors{$a} ) { return $errorids{$a} cmp $errorids{$b} }
193	else { return $errors{$b} <=> $errors{$a} } } (keys %errors))
194	{
195	$errorhtml=$error;
196
197	$errorhtml =~ s/&/&/g;
198	$errorhtml =~ s/</</g;
199	$errorhtml =~ s/>/>/g;
200
201	if($errortypes{$error} ne $lasterrortype)
202	{
203	print "<h2>$errortypelabel{$errortypes{$error}}</h2>\n";
204	$lasterrortype=$errortypes{$error};
205	}
206
207	print "<h3>$errorhtml</h3>\n";
208
209	if($errors{$error}>100)
210	{
211	print "$errors{$error} occurences (not listed).\n";
212	}
213	else
214	{
215	@ids=split(",",$errorids{$error});
216
217	$first=1;
218
219	foreach $id (@ids)
220	{
221	if($first)
222	{
223	print "$errortypelabel{$errortypes{$error}}:\n";
224	}
225	else
226	{
227	print ",";
228	}
229
230	$first=0;
231
232	print "<a href=\"http://www.openstreetmap.org/browse/node/$id\">$id</a>" if($errortypes{$error} eq "N");
233	print "<a href=\"http://www.openstreetmap.org/browse/way/$id\">$id</a>" if($errortypes{$error} eq "W");
234	print "<a href=\"http://www.openstreetmap.org/browse/relation/$id\">$id</a>" if($errortypes{$error} eq "R");
235
236	if($errortypes{$error} eq "NW" \|\| $errortypes{$error} eq "N2" \|\| $errortypes{$error} eq "RN" \|\| $errortypes{$error} eq "RW")
237	{
238	$id =~ m%\(([0-9]+) ([0-9]+)\)%;
239	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");
240	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");
241	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");
242	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{$error} eq "RW");
243	}
244
245	if($errortypes{$error} eq "N2W")
246	{
247	$id =~ m%\(([0-9]+) ([0-9]+) ([0-9]+)\)%;
248	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");
249	}
250
251	print "\n";
252	}
253	}
254	}
255
256	print "\n".
257	"</BODY>\n".
258	"\n".
259	"</HTML>\n";
260
261	}