]> git.openstreetmap.org Git - nominatim.git/commitdiff
Merge remote-tracking branch 'upstream/master'
authorSarah Hoffmann <lonvia@denofr.de>
Thu, 26 Nov 2020 13:41:16 +0000 (14:41 +0100)
committerSarah Hoffmann <lonvia@denofr.de>
Thu, 26 Nov 2020 13:41:16 +0000 (14:41 +0100)
16 files changed:
CMakeLists.txt
lib/Geocode.php
munin/nominatim_query_speed_querylog [new file with mode: 0755]
munin/nominatim_requests_querylog [new file with mode: 0755]
munin/nominatim_throttled_ips [new file with mode: 0755]
sql/functions/address_lookup.sql
sql/tiger_import_finish.sql
utils/cron_ipanalyse.py [new file with mode: 0755]
utils/update.php
website/403.html [new file with mode: 0644]
website/509.html [new file with mode: 0644]
website/crossdomain.xml [new file with mode: 0644]
website/favicon.ico [new file with mode: 0644]
website/nominatim.xml [new file with mode: 0644]
website/robots.txt [new file with mode: 0644]
website/taginfo.json [new file with mode: 0644]

index cb1373419987d4f79e228f935857b3438ab9b860..816ce647613c4a996a6ee6c7ed980607f6b584fc 100644 (file)
@@ -128,6 +128,16 @@ if (BUILD_API)
        website/status.php
    )
 
+   set(WEBSITEFILES
+       403.html
+       509.html
+       crossdomain.xml
+       favicon.ico
+       nominatim.xml
+       robots.txt
+       taginfo.json
+   )
+
    foreach (script_source ${WEBSITESCRIPTS})
        configure_file(${PROJECT_SOURCE_DIR}/cmake/website.tmpl
                       ${PROJECT_BINARY_DIR}/${script_source})
@@ -135,6 +145,11 @@ if (BUILD_API)
 
    set(WEBPATHS css images js)
 
+   foreach (webfile ${WEBSITEFILES})
+       configure_file(${PROJECT_SOURCE_DIR}/website/${webfile}
+                      ${PROJECT_BINARY_DIR}/website/${webfile})
+   endforeach()
+
    foreach (wp ${WEBPATHS})
        execute_process(
            COMMAND ln -sf ${PROJECT_SOURCE_DIR}/website/${wp} ${PROJECT_BINARY_DIR}/website/
index ed02848eac4c2c846ea1a655089caa75aca879cd..6589f75485dceb98f37ad248e1ff97e8338e9f99 100644 (file)
@@ -18,7 +18,7 @@ class Geocode
     protected $aLangPrefOrder = array();
 
     protected $aExcludePlaceIDs = array();
-    protected $bReverseInPlan = false;
+    protected $bReverseInPlan = true;
 
     protected $iLimit = 20;
     protected $iFinalLimit = 10;
diff --git a/munin/nominatim_query_speed_querylog b/munin/nominatim_query_speed_querylog
new file mode 100755 (executable)
index 0000000..f35793f
--- /dev/null
@@ -0,0 +1,163 @@
+#!/usr/bin/python3
+#
+# Plugin to monitor the types of requsts made to the API
+#
+# Uses the query log.
+#
+# Parameters:
+#
+#       config   (required)
+#       autoconf (optional - used by munin-config)
+#
+
+import re
+import os
+import sys
+from datetime import datetime, timedelta
+
+CONFIG="""graph_title Total Nominatim response time
+graph_vlabel Time to response
+graph_category Nominatim 
+graph_period minute
+graph_args --base 1000
+
+avgs.label Average search time
+avgs.draw LINE
+avgs.type GAUGE
+avgs.min 0
+avgs.info Moving 5 minute average time to perform search
+
+avgr.label Average reverse time
+avgr.draw LINE
+avgr.type GAUGE
+avgr.min 0
+avgr.info Moving 5 minute average time to perform reverse
+
+max.label Slowest time to response (1/100)
+max.draw LINE
+max.type GAUGE
+max.min 0
+max.info Slowest query in last 5 minutes (unit: 100s)"""
+
+ENTRY_REGEX = re.compile(r'\[[^]]+\] (?P<dur>[0-9.]+) (?P<numres>\d+) (?P<type>[a-z]+) ')
+TIME_REGEX = re.compile(r'\[(?P<t_year>\d\d\d\d)-(?P<t_month>\d\d)-(?P<t_day>\d\d) (?P<t_hour>\d\d):(?P<t_min>\d\d):(?P<t_sec>\d\d)[0-9.]*\] ')
+
+
+class LogFile:
+    """ A query log file, unpacked. """
+
+    def __init__(self, filename):
+        self.fd = open(filename, encoding='utf-8', errors='replace')
+        self.len = os.path.getsize(filename)
+
+    def __del__(self):
+        self.fd.close()
+
+    def seek_next(self, abstime):
+        self.fd.seek(abstime)
+        self.fd.readline()
+        l = self.fd.readline()
+        e = TIME_REGEX.match(l)
+        if e is None:
+            return None
+        e = e.groupdict()
+        return datetime(int(e['t_year']), int(e['t_month']), int(e['t_day']),
+                             int(e['t_hour']), int(e['t_min']), int(e['t_sec']))
+
+    def seek_to_date(self, target):
+        # start position for binary search
+        fromseek = 0
+        fromdate = self.seek_next(0)
+        if fromdate > target:
+            return True
+        # end position for binary search
+        toseek = -100
+        while -toseek < self.len:
+            todate = self.seek_next(self.len + toseek)
+            if todate is not None:
+                break
+            toseek -= 100
+        if todate is None or todate < target:
+            return False
+        toseek = self.len + toseek
+
+
+        while True:
+            bps = (toseek - fromseek) / (todate - fromdate).total_seconds()
+            newseek = fromseek + int((target - fromdate).total_seconds() * bps)
+            newdate = self.seek_next(newseek)
+            if newdate is None:
+                return False;
+            error = abs((target - newdate).total_seconds())
+            if error < 1:
+                return True
+            if newdate > target:
+                toseek = newseek
+                todate = newdate
+                oldfromseek = fromseek
+                fromseek = toseek - error * bps
+                while True:
+                    if fromseek <= oldfromseek:
+                        fromseek = oldfromseek
+                        fromdate = self.seek_next(fromseek)
+                        break
+                    fromdate = self.seek_next(fromseek)
+                    if fromdate < target:
+                        break;
+                    bps *=2
+                    fromseek -= error * bps
+            else:
+                fromseek = newseek
+                fromdate = newdate
+                oldtoseek = toseek
+                toseek = fromseek + error * bps
+                while True:
+                    if toseek > oldtoseek:
+                        toseek = oldtoseek
+                        todate = self.seek_next(toseek)
+                        break
+                    todate = self.seek_next(toseek)
+                    if todate > target:
+                        break
+                    bps *=2
+                    toseek += error * bps
+            if toseek - fromseek < 500:
+                return True
+
+
+    def loglines(self):
+        for l in self.fd:
+            e = ENTRY_REGEX.match(l)
+            if e is not None:
+                yield e.groupdict()
+
+
+if __name__ == '__main__':
+
+    if len(sys.argv) > 1 and sys.argv[1] == 'config':
+        print(CONFIG)
+        sys.exit(0)
+
+    sumrev = 0
+    numrev = 0
+    sumsearch = 0
+    numsearch = 0
+    maxres = 0
+    if 'NOMINATIM_QUERYLOG' in os.environ:
+        lf = LogFile(os.environ['NOMINATIM_QUERYLOG'])
+        if lf.seek_to_date(datetime.now() - timedelta(minutes=5)):
+            for l in lf.loglines():
+                dur = float(l['dur'])
+                if l['type'] == 'reverse':
+                    numrev += 1
+                    sumrev += dur
+                elif  l['type'] == 'search':
+                    numsearch += 1
+                    sumsearch += dur
+                if dur > maxres:
+                    maxres = dur
+
+
+    print('avgs.value', 0 if numsearch == 0 else sumsearch/numsearch)
+    print('avgr.value', 0 if numrev == 0 else sumrev/numrev)
+    print('max.value', maxres/100.0)
diff --git a/munin/nominatim_requests_querylog b/munin/nominatim_requests_querylog
new file mode 100755 (executable)
index 0000000..8a103cf
--- /dev/null
@@ -0,0 +1,163 @@
+#!/usr/bin/python3
+#
+# Plugin to monitor the types of requsts made to the API
+#
+# Uses the query log.
+#
+# Parameters: 
+#
+#       config   (required)
+#       autoconf (optional - used by munin-config)
+#
+
+import re
+import os
+import sys
+from datetime import datetime, timedelta
+
+CONFIG="""graph_title Requests by API call
+graph_args --base 1000 -l 0
+graph_vlabel requests per minute
+graph_category nominatim
+z1.label reverse
+z1.draw AREA
+z1.type GAUGE
+z2.label search (successful)
+z2.draw STACK
+z2.type GAUGE
+z3.label search (no result)
+z3.draw STACK
+z3.type GAUGE
+z4.label lookup
+z4.draw STACK
+z4.type GAUGE
+z4.label details
+z4.draw STACK
+z4.type GAUGE"""
+
+ENTRY_REGEX = re.compile(r'\[[^]]+\] (?P<dur>[0-9.]+) (?P<numres>\d+) (?P<type>[a-z]+) ')
+TIME_REGEX = re.compile(r'\[(?P<t_year>\d\d\d\d)-(?P<t_month>\d\d)-(?P<t_day>\d\d) (?P<t_hour>\d\d):(?P<t_min>\d\d):(?P<t_sec>\d\d)[0-9.]*\] ')
+
+
+class LogFile:
+    """ A query log file, unpacked. """
+
+    def __init__(self, filename):
+        self.fd = open(filename, encoding='utf-8', errors='replace')
+        self.len = os.path.getsize(filename)
+
+    def __del__(self):
+        self.fd.close()
+
+    def seek_next(self, abstime):
+        self.fd.seek(abstime)
+        self.fd.readline()
+        l = self.fd.readline()
+        e = TIME_REGEX.match(l)
+        if e is None:
+            return None
+        e = e.groupdict()
+        return datetime(int(e['t_year']), int(e['t_month']), int(e['t_day']),
+                             int(e['t_hour']), int(e['t_min']), int(e['t_sec']))
+
+    def seek_to_date(self, target):
+        # start position for binary search
+        fromseek = 0
+        fromdate = self.seek_next(0)
+        if fromdate > target:
+            return True
+        # end position for binary search
+        toseek = -100
+        while -toseek < self.len:
+            todate = self.seek_next(self.len + toseek)
+            if todate is not None:
+                break
+            toseek -= 100
+        if todate is None or todate < target:
+            return False
+        toseek = self.len + toseek
+
+
+        while True:
+            bps = (toseek - fromseek) / (todate - fromdate).total_seconds()
+            newseek = fromseek + int((target - fromdate).total_seconds() * bps)
+            newdate = self.seek_next(newseek)
+            if newdate is None:
+                return False;
+            error = abs((target - newdate).total_seconds())
+            if error < 1:
+                return True
+            if newdate > target:
+                toseek = newseek
+                todate = newdate
+                oldfromseek = fromseek
+                fromseek = toseek - error * bps
+                while True:
+                    if fromseek <= oldfromseek:
+                        fromseek = oldfromseek
+                        fromdate = self.seek_next(fromseek)
+                        break
+                    fromdate = self.seek_next(fromseek)
+                    if fromdate < target:
+                        break;
+                    bps *=2
+                    fromseek -= error * bps
+            else:
+                fromseek = newseek
+                fromdate = newdate
+                oldtoseek = toseek
+                toseek = fromseek + error * bps
+                while True:
+                    if toseek > oldtoseek:
+                        toseek = oldtoseek
+                        todate = self.seek_next(toseek)
+                        break
+                    todate = self.seek_next(toseek)
+                    if todate > target:
+                        break
+                    bps *=2
+                    toseek += error * bps
+            if toseek - fromseek < 500:
+                return True
+
+
+    def loglines(self):
+        for l in self.fd:
+            e = ENTRY_REGEX.match(l)
+            if e is not None:
+                yield e.groupdict()
+
+
+if __name__ == '__main__':
+
+    if len(sys.argv) > 1 and sys.argv[1] == 'config':
+        print(CONFIG)
+        sys.exit(0)
+
+    reverse = 0
+    searchy = 0
+    searchn = 0
+    lookup = 0
+    details = 0
+    if 'NOMINATIM_QUERYLOG' in os.environ:
+        lf = LogFile(os.environ['NOMINATIM_QUERYLOG'])
+        if lf.seek_to_date(datetime.now() - timedelta(minutes=5)):
+            for l in lf.loglines():
+                if l['type'] == 'reverse':
+                    reverse += 1
+                elif  l['type'] == 'search':
+                    if l['numres'] == '0':
+                        searchn += 1
+                    else:
+                        searchy += 1
+                elif  l['type'] == 'place':
+                    lookup +=1
+                else:
+                    details += 1
+
+
+    print('z1.value', reverse/5)
+    print('z2.value', searchy/5)
+    print('z3.value', searchn/5)
+    print('z4.value', lookup/5)
+    print('z4.value', details/5)
diff --git a/munin/nominatim_throttled_ips b/munin/nominatim_throttled_ips
new file mode 100755 (executable)
index 0000000..a56ff31
--- /dev/null
@@ -0,0 +1,28 @@
+#!/bin/sh
+#
+# Plugin to monitor the number of IPs in special pools
+#
+# Parameters: 
+#
+#       config   (required)
+#       autoconf (optional - used by munin-config)
+#
+if [ "$1" = "config" ]; then
+        echo 'graph_title Restricted IPs' 
+        echo 'graph_args -l 0'
+        echo 'graph_vlabel number of IPs'
+        echo 'graph_category nominatim'
+        echo 'bulk.label bulk'
+        echo 'bulk.draw AREA'
+        echo 'bulk.type GAUGE'
+        echo 'block.label blocked'
+        echo 'block.draw STACK'
+        echo 'block.type GAUGE'
+        exit 0
+fi
+BASEDIR="$(dirname "$(readlink -f "$0")")"
+
+cut -f 2 -d ' ' $BASEDIR/../../bin/settings/ip_blocks.map | sort | uniq -c | sed 's:[[:space:]]*\([0-9]\+\) \(.*\):\2.value \1:'
index 6ee1f048600b005071d2a546f4aaa470d23be396..2426a698f03c4dc3cb373ff793fa9e1eb364e6e7 100644 (file)
@@ -214,7 +214,7 @@ BEGIN
   FOR location IN
     SELECT placex.place_id, osm_type, osm_id, name, class, type,
            coalesce(extratags->'linked_place', extratags->'place') as place_type,
-           admin_level, fromarea, isaddress,
+           admin_level, fromarea, isaddress and linked_place_id is NULL as isaddress,
            CASE WHEN rank_address = 11 THEN 5 ELSE rank_address END as rank_address,
            distance, country_code, postcode
       FROM place_addressline join placex on (address_place_id = placex.place_id)
index 374c00b39bcaa5bfa37aa1a66b703c89e61583f1..09942bac48cffacf6dad971ac09ac3f023029602 100644 (file)
@@ -4,10 +4,10 @@ CREATE UNIQUE INDEX idx_location_property_tiger_place_id_imp ON location_propert
 
 GRANT SELECT ON location_property_tiger_import TO "{www-user}";
 
-DROP TABLE IF EXISTS location_property_tiger;
-ALTER TABLE location_property_tiger_import RENAME TO location_property_tiger;
+--DROP TABLE IF EXISTS location_property_tiger;
+--ALTER TABLE location_property_tiger_import RENAME TO location_property_tiger;
 
-ALTER INDEX idx_location_property_tiger_parent_place_id_imp RENAME TO idx_location_property_tiger_housenumber_parent_place_id;
-ALTER INDEX idx_location_property_tiger_place_id_imp RENAME TO idx_location_property_tiger_place_id;
+--ALTER INDEX idx_location_property_tiger_parent_place_id_imp RENAME TO idx_location_property_tiger_housenumber_parent_place_id;
+--ALTER INDEX idx_location_property_tiger_place_id_imp RENAME TO idx_location_property_tiger_place_id;
 
 DROP FUNCTION tiger_line_import (linegeo geometry, in_startnumber integer, in_endnumber integer, interpolationtype text, in_street text, in_isin text, in_postcode text);
diff --git a/utils/cron_ipanalyse.py b/utils/cron_ipanalyse.py
new file mode 100755 (executable)
index 0000000..97bad8d
--- /dev/null
@@ -0,0 +1,402 @@
+#!/usr/bin/python3
+#
+# Search apache logs for high-bandwith users and create a list of suspicious IPs.
+# There are three states: bulk, block, ban. The first are bulk requesters
+# that need throtteling, the second bulk requesters that have overdone it
+# and the last manually banned IPs.
+#
+
+import re
+import os
+import sys
+import subprocess
+from datetime import datetime, timedelta
+from collections import defaultdict
+
+#
+# DEFAULT SETTINGS
+#
+# Copy into settings/ip_blcoks.conf and adapt as required.
+#
+BASEDIR = os.path.normpath(os.path.join(os.path.realpath(__file__), '../..'))
+BLOCKEDFILE= BASEDIR + '/settings/ip_blocks.map'
+LOGFILE= BASEDIR + '/log/restricted_ip.log'
+
+# space-separated list of IPs that are never banned
+WHITELIST = ''
+# space-separated list of IPs manually blocked
+BLACKLIST = ''
+# user-agents that should be blocked from bulk mode
+# (matched with startswith)
+UA_BLOCKLIST = ()
+
+# time before a automatically blocked IP is allowed back
+BLOCKCOOLOFF_DELTA=timedelta(hours=1)
+# quiet time before an IP is released from the bulk pool
+BULKCOOLOFF_DELTA=timedelta(minutes=15)
+# time to check if new accesses appear despite being blocked
+BLOCKCHECK_DELTA=timedelta(minutes=1)
+
+BULKLONG_LIMIT=8000
+BULKSHORT_LIMIT=2000
+BLOCK_UPPER=19000
+BLOCK_LOWER=4000
+BLOCK_LOADFAC=380
+BULK_LOADFAC=160
+BULK_LOWER=1500
+MAX_BULK_IPS=85
+
+#
+# END OF DEFAULT SETTINGS
+#
+
+try:
+    with open(BASEDIR + "/settings/ip_blocks.conf") as f:
+        code = compile(f.read(), BASEDIR + "/settings/ip_blocks.conf", 'exec')
+        exec(code)
+except IOError:
+    pass
+
+BLOCK_LIMIT = BLOCK_LOWER
+
+time_regex = r'(?P<t_day>\d\d)/(?P<t_month>[A-Za-z]+)/(?P<t_year>\d\d\d\d):(?P<t_hour>\d\d):(?P<t_min>\d\d):(?P<t_sec>\d\d) [+-]\d\d\d\d'
+
+format_pat= re.compile(r'(?P<ip>[a-f\d\.:]+) - - \['+ time_regex + r'] "(?P<query>.*?)" (?P<return>\d+) (?P<bytes>\d+) "(?P<referer>.*?)" "(?P<ua>.*?)"')
+time_pat= re.compile(r'[a-f\d:\.]+ - - \[' + time_regex + '\] ')
+
+logtime_pat = "%d/%b/%Y:%H:%M:%S %z"
+
+MONTHS = { 'Jan' : 1, 'Feb' : 2, 'Mar' : 3, 'Apr' : 4, 'May' : 5, 'Jun' : 6,
+           'Jul' : 7, 'Aug' : 8, 'Sep' : 9, 'Oct' : 10, 'Nov' : 11, 'Dec' : 12 }
+
+class LogEntry:
+    def __init__(self, logline):
+        e = format_pat.match(logline)
+        if e is None:
+            raise ValueError("Invalid log line:", logline)
+        e = e.groupdict()
+        self.ip = e['ip']
+        self.date = datetime(int(e['t_year']), MONTHS[e['t_month']], int(e['t_day']),
+                             int(e['t_hour']), int(e['t_min']), int(e['t_sec']))
+        qp = e['query'].split(' ', 2) 
+        if len(qp) < 2:
+            self.request = None
+            self.query = None
+        else:
+            self.query = qp[1]
+            if qp[0] == 'OPTIONS':
+                self.request = None
+            else:
+                if '/?' in qp[1]:
+                    self.request = 'S'
+                elif '/search' in qp[1]:
+                    self.request = 'S'
+                elif '/reverse' in qp[1]:
+                    self.request = 'R'
+                elif '/details' in qp[1]:
+                    self.request = 'D'
+                elif '/lookup' in qp[1]:
+                    self.request = 'L'
+                else:
+                    self.request = None
+        self.query = e['query']
+        self.retcode = int(e['return'])
+        self.referer = e['referer'] if e['referer'] != '-' else None
+        self.ua = e['ua'] if e['ua'] != '-' else None
+
+    def get_log_time(logline):
+        e = format_pat.match(logline)
+        if e is None:
+            return None
+        e = e.groupdict()
+        #return datetime.strptime(e['time'], logtime_pat).replace(tzinfo=None)
+        return datetime(int(e['t_year']), MONTHS[e['t_month']], int(e['t_day']),
+                             int(e['t_hour']), int(e['t_min']), int(e['t_sec']))
+
+
+class LogFile:
+    """ An apache log file, unpacked. """
+
+    def __init__(self, filename):
+        self.fd = open(filename)
+        self.len = os.path.getsize(filename)
+
+    def __del__(self):
+        self.fd.close()
+
+    def seek_next(self, abstime):
+        self.fd.seek(abstime)
+        self.fd.readline()
+        l = self.fd.readline()
+        return LogEntry.get_log_time(l) if l is not None else None
+
+    def seek_to_date(self, target):
+        # start position for binary search
+        fromseek = 0
+        fromdate = self.seek_next(0)
+        if fromdate > target:
+            return True
+        # end position for binary search
+        toseek = -100
+        while -toseek < self.len:
+            todate = self.seek_next(self.len + toseek)
+            if todate is not None:
+                break
+            toseek -= 100
+        if todate is None or todate < target:
+            return False
+        toseek = self.len + toseek
+
+
+        while True:
+            bps = (toseek - fromseek) / (todate - fromdate).total_seconds()
+            newseek = fromseek + int((target - fromdate).total_seconds() * bps)
+            newdate = self.seek_next(newseek)
+            if newdate is None:
+                return False;
+            error = abs((target - newdate).total_seconds())
+            if error < 1:
+                return True
+            if newdate > target:
+                toseek = newseek
+                todate = newdate
+                oldfromseek = fromseek
+                fromseek = toseek - error * bps
+                while True:
+                    if fromseek <= oldfromseek:
+                        fromseek = oldfromseek
+                        fromdate = self.seek_next(fromseek)
+                        break
+                    fromdate = self.seek_next(fromseek)
+                    if fromdate < target:
+                        break;
+                    bps *=2
+                    fromseek -= error * bps
+            else:
+                fromseek = newseek
+                fromdate = newdate
+                oldtoseek = toseek
+                toseek = fromseek + error * bps
+                while True:
+                    if toseek > oldtoseek:
+                        toseek = oldtoseek
+                        todate = self.seek_next(toseek)
+                        break
+                    todate = self.seek_next(toseek)
+                    if todate > target:
+                        break
+                    bps *=2
+                    toseek += error * bps
+            if toseek - fromseek < 500:
+                return True
+
+
+    def loglines(self):
+        for l in self.fd:
+            try:
+                yield LogEntry(l)
+            except ValueError:
+                pass # ignore invalid lines
+
+class BlockList:
+
+    def __init__(self):
+        self.whitelist = set(WHITELIST.split()) if WHITELIST else set()
+        self.blacklist = set(BLACKLIST.split()) if BLACKLIST else set()
+        self.prevblocks = set()
+        self.prevbulks = set()
+
+        try:
+            fd = open(BLOCKEDFILE)
+            for line in fd:
+                ip, typ = line.strip().split(' ')
+                if ip not in self.blacklist:
+                    if typ == 'block':
+                        self.prevblocks.add(ip)
+                    elif typ == 'bulk':
+                        self.prevbulks.add(ip)
+            fd.close()
+        except IOError:
+            pass #ignore non-existing file
+
+
+class IPstats:
+
+    def __init__(self):
+        self.redirected = 0
+        self.short_total = 0
+        self.short_api = 0
+        self.long_total = 0
+        self.long_api = 0
+        self.block_total = 0
+        self.bad_ua = False
+
+    def add_long(self, logentry):
+        self.long_total += 1
+        if logentry.retcode == 301:
+            return
+        if logentry.request is not None:
+            self.long_api += 1
+        if not self.bad_ua:
+            if logentry.ua is None:
+                self.bad_ua = True
+
+    def add_short(self, logentry):
+        self.short_total += 1
+        if logentry.retcode == 301:
+            self.redirected += 1
+            return
+        if logentry.request is not None:
+            self.short_api += 1
+        self.add_long(logentry)
+
+    def add_block(self, logentry):
+        self.block_total += 1
+
+    def ignores_warnings(self, wasblocked):
+        return self.block_total > 5 or (wasblocked and self.redirected > 5)
+
+    def new_state(self, was_blocked, was_bulked):
+        if was_blocked:
+            # deblock only if the IP has been really quiet
+            # (properly catches the ones that simply ignore the HTTP error)
+            return None if self.long_total < 20 else 'block'
+        if self.long_api > BLOCK_UPPER \
+            or self.short_api > BLOCK_UPPER / 3 \
+            or (self.redirected > 100 and self.short_total == self.redirected):
+                # client totally overdoing it
+                return 'block'
+        if was_bulked:
+            if self.short_total < 20:
+                # client has stopped, debulk
+                return None
+            if self.long_api > BLOCK_LIMIT or self.short_api > BLOCK_LIMIT / 3:
+                # client is still hammering us, block
+                return 'emblock'
+            return 'bulk'
+
+        if self.long_api > BULKLONG_LIMIT or self.short_api > BULKSHORT_LIMIT:
+            #if self.bad_ua:
+            #    return 'uablock' # bad useragent
+            return 'bulk'
+
+        return None
+
+
+
+if __name__ == '__main__':
+    if len(sys.argv) < 2:
+        print("Usage: %s logfile startdate" % sys.argv[0])
+        sys.exit(-1)
+
+    if len(sys.argv) == 2:
+        dt = datetime.now() - BLOCKCOOLOFF_DELTA
+    else:
+        dt = datetime.strptime(sys.argv[2], "%Y-%m-%d %H:%M:%S")
+
+    if os.path.getsize(sys.argv[1]) < 2*1030*1024:
+        sys.exit(0) # not enough data
+
+    lf = LogFile(sys.argv[1])
+    if not lf.seek_to_date(dt):
+        sys.exit(0)
+
+    bl = BlockList()
+
+    shortstart = dt + BLOCKCOOLOFF_DELTA - BULKCOOLOFF_DELTA
+    blockstart = dt + BLOCKCOOLOFF_DELTA - BLOCKCHECK_DELTA
+    notlogged = bl.whitelist | bl.blacklist
+
+    stats = defaultdict(IPstats)
+
+    for l in lf.loglines():
+        if l.ip not in notlogged:
+            stats[l.ip].add_long(l)
+        if l.date > shortstart:
+            break
+
+    total200 = 0
+    for l in lf.loglines():
+        if l.ip not in notlogged:
+            stats[l.ip].add_short(l)
+        if l.request is not None and l.retcode == 200:
+            total200 += 1
+        if l.date > blockstart and l.retcode in (403, 429):
+            stats[l.ip].add_block(l)
+
+    # adapt limits according to CPU and DB load
+    fd = open("/proc/loadavg")
+    cpuload = int(float(fd.readline().split()[2]))
+    fd.close()
+    # check the number of excess connections to apache
+    dbcons = int(subprocess.check_output("netstat -s | grep 'connections established' | sed 's:^\s*::;s: .*::'", shell=True))
+    fpms = int(subprocess.check_output('ps -Af | grep php-fpm | wc -l', shell=True))
+    dbload = max(0, dbcons - fpms)
+
+    numbulks = len(bl.prevbulks)
+    BLOCK_LIMIT = max(BLOCK_LIMIT, BLOCK_UPPER - BLOCK_LOADFAC * dbload)
+    BULKLONG_LIMIT = max(BULK_LOWER, BULKLONG_LIMIT - BULK_LOADFAC * cpuload)
+    if numbulks > MAX_BULK_IPS:
+        BLOCK_LIMIT = max(3600, BLOCK_LOWER - (numbulks - MAX_BULK_IPS)*10)
+    # if the bulk pool is still empty, clients will be faster, avoid having
+    # them blocked in this case
+    if numbulks < 10:
+        BLOCK_UPPER *= 2
+        BLOCK_LIMIT = BLOCK_UPPER
+
+
+    # collecting statistics
+    unblocked = []
+    debulked = []
+    bulked = []
+    blocked = []
+    uablocked = []
+    emblocked = []
+    # write out new state file
+    fd = open(BLOCKEDFILE, 'w')
+    for k,v in stats.items():
+        wasblocked = k in bl.prevblocks
+        wasbulked = k in bl.prevbulks
+        state = v.new_state(wasblocked, wasbulked)
+        if state is not None:
+            if state == 'uablock':
+                uablocked.append(k)
+                state = 'block'
+            elif state == 'emblock':
+                emblocked.append(k)
+                state = 'block'
+            elif state == 'block':
+                if not wasblocked:
+                    blocked.append(k)
+            elif state == 'bulk':
+                if not wasbulked:
+                    bulked.append(k)
+            fd.write("%s %s\n" % (k, state))
+        else:
+            if wasblocked:
+                unblocked.append(k)
+            elif wasbulked:
+                debulked.append(k)
+    for i in bl.blacklist:
+        fd.write("%s ban\n" % i)
+    fd.close()
+
+    # TODO write logs (need to collect some statistics)
+    logstr = datetime.now().strftime('%d/%b/%Y:%H:%M:%S') + ' %s %s\n'
+    fd = open(LOGFILE, 'a')
+    if unblocked:
+        fd.write(logstr % ('unblocked:', ', '.join(unblocked)))
+    if debulked:
+        fd.write(logstr % (' debulked:', ', '.join(debulked)))
+    if bulked:
+        fd.write(logstr % ('new bulks:', ', '.join(bulked)))
+    if emblocked:
+        fd.write(logstr % ('dir.block:', ', '.join(emblocked)))
+    if uablocked:
+        fd.write(logstr % (' ua block:', ', '.join(uablocked)))
+    if blocked:
+        fd.write(logstr % ('new block:', ', '.join(blocked)))
+    #for k,v in stats.items():
+    #    if v.ignores_warnings(k in bl.prevblocks) and k not in notlogged and ':' not in k:
+    #        fd.write(logstr % ('Warning ignored:', k))
+    fd.close()
index 7e90016c3c01018cae76afe02231f8976d2f3565..87a5a294221e5fed07f83e42af1e602fb8e1c49f 100644 (file)
@@ -47,6 +47,7 @@ $aCMDOptions
 getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
 
 if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
+
 if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;
 
 date_default_timezone_set('Etc/UTC');
diff --git a/website/403.html b/website/403.html
new file mode 100644 (file)
index 0000000..8d8e323
--- /dev/null
@@ -0,0 +1,23 @@
+<html>
+<head>
+<title>Access blocked</title>
+</head>
+<body>
+<h1>Access blocked</h1>
+
+<p>You have been blocked because you have violated the
+<a href="https://operations.osmfoundation.org/policies/nominatim/">usage policy</a>
+of OSM's Nominatim geocoding service. Please be aware that OSM's resources are
+limited and shared between many users. The usage policy is there to ensure that
+the service remains usable for everybody.</p>
+
+<p>Please review the terms and make sure that your
+software adheres to the terms. You should in particular verify that you have set a
+<b>custom HTTP referrer or HTTP user agent</b> that identifies your application, and
+that you are not overusing the service with massive bulk requests.</p>
+
+<p>If you feel that this block is unjustified or remains after you have adopted
+your usage, you may contact the Nominatim system administrator at
+nominatim@openstreetmap.org to have this block lifted.</p>
+</body>
+</head>
diff --git a/website/509.html b/website/509.html
new file mode 100644 (file)
index 0000000..628c53b
--- /dev/null
@@ -0,0 +1,12 @@
+<html>
+<head>
+<title>Bandwidth limit exceeded</title>
+</head>
+<body>
+<h1>Bandwidth limit exceeded</h1>
+
+<p>You have been temporarily blocked because you have been overusing OSM's geocoding service or because you have not provided sufficient identification of your application. This block will be automatically lifted after a while. Please take the time and adapt your scripts to reduce the number of requests and make sure that you send a valid UserAgent or Referer.</p>
+
+<p>For more information, consult the <a href="https://operations.osmfoundation.org/policies/nominatim/">usage policy</a> for the OSM Nominatim server.</p>
+</body>
+</html>
diff --git a/website/crossdomain.xml b/website/crossdomain.xml
new file mode 100644 (file)
index 0000000..963a682
--- /dev/null
@@ -0,0 +1,5 @@
+<?xml version="1.0"?>
+           <!DOCTYPE cross-domain-policy SYSTEM "http://www.macromedia.com/xml/dtds/cross-domain-policy.dtd">
+           <cross-domain-policy>
+           <allow-access-from domain="*" />
+           </cross-domain-policy> 
diff --git a/website/favicon.ico b/website/favicon.ico
new file mode 100644 (file)
index 0000000..0157ea0
Binary files /dev/null and b/website/favicon.ico differ
diff --git a/website/nominatim.xml b/website/nominatim.xml
new file mode 100644 (file)
index 0000000..28684b1
--- /dev/null
@@ -0,0 +1,15 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<OpenSearchDescription xmlns="http://a9.com/-/spec/opensearch/1.1/"
+                       xmlns:moz="http://www.mozilla.org/2006/browser/search/">
+       <ShortName>Nominatim</ShortName>
+       <LongName>Nominatim OSM Search</LongName>
+       <Description>Search for a place in OpenStreetMap Nominatim</Description>
+       <InputEncoding>UTF-8</InputEncoding>
+       <OutputEncoding>UTF-8</OutputEncoding>
+       <Url type="text/html" method="get" template="http://nominatim.openstreetmap.org/search/?q={searchTerms}" />
+       <Query role="example" searchTerms="Reigate" />
+       <Developer>Brian Quinion</Developer>
+       <AdultContent>false</AdultContent>
+       <Attribution>Data &amp;copy; OpenStreetMap contributors, Some Rights Reserved. ODbL, http://www.osm.org/copyright.</Attribution>
+</OpenSearchDescription>
+
diff --git a/website/robots.txt b/website/robots.txt
new file mode 100644 (file)
index 0000000..9624d97
--- /dev/null
@@ -0,0 +1,14 @@
+User-agent: ia_archiver
+Allow: /
+
+User-agent: *
+Disallow: /search.php
+Disallow: /search
+Disallow: /details.php
+Disallow: /details
+Disallow: /reverse.php
+Disallow: /reverse
+Disallow: /hierarchy
+Disallow: /hierarchy.php
+Disallow: /lookup
+Disallow: /lookup.php
diff --git a/website/taginfo.json b/website/taginfo.json
new file mode 100644 (file)
index 0000000..98f8b97
--- /dev/null
@@ -0,0 +1,112 @@
+{
+    "data_format": 1,
+    "data_url": "http://nominatim.openstreetmap.org/taginfo.json",
+    "project": {
+        "name": "Nominatim",
+        "description": "OSM search engine.",
+        "project_url": "http://nominatim.openstreetmap.org",
+        "doc_url": "http://wiki.osm.org/wiki/Nominatim",
+        "contact_name": "Sarah Hoffmann",
+        "contact_email": "lonvia@denofr.de"
+    },
+    "tags": [
+      { "key" : "ref", "description": "Searchable name of the place."},
+      { "key" : "int_ref", "description": "Searchable name of the place."},
+      { "key" : "nat_ref", "description": "Searchable name of the place."},
+      { "key" : "reg_ref", "description": "Searchable name of the place."},
+      { "key" : "loc_ref", "description": "Searchable name of the place."},
+      { "key" : "old_ref", "description": "Searchable name of the place."},
+      { "key" : "iata", "description": "Searchable name of the place."},
+      { "key" : "icao", "description": "Searchable name of the place."},
+      { "key" : "pcode", "description": "Searchable name of the place."},
+      { "key" : "name", "description": "Searchable name of the place."},
+      { "key" : "int_name", "description": "Searchable name of the place."},
+      { "key" : "nat_name", "description": "Searchable name of the place."},
+      { "key" : "reg_name", "description": "Searchable name of the place."},
+      { "key" : "loc_name", "description": "Searchable name of the place."},
+      { "key" : "old_name", "description": "Searchable name of the place."},
+      { "key" : "alt_name", "description": "Searchable name of the place."},
+      { "key" : "official_name", "description": "Searchable name of the place."},
+      { "key" : "place_name", "description": "Searchable name of the place."},
+      { "key" : "short_name", "description": "Searchable name of the place."},
+      { "key" : "addr:housename", "description": "Searchable name of the place."},
+      { "key" : "operator", "description": "Searchable name for amenities and shops." },
+      { "key" : "brand", "description": "Searchable name of POI places."},
+      { "key" : "bridge:name", "description" : "Searchable name for bridges."},
+      { "key" : "tunnel:name", "description" : "Searchable name for tunnels."},
+      { "key" : "emergency", "description": "POI in the search database." },
+      { "key" : "tourism", "description": "POI in the search database." },
+      { "key" : "historic", "description": "POI in the search database." },
+      { "key" : "military", "description": "POI in the search database." },
+      { "key" : "natural", "description": "POI in the search database." },
+      { "key" : "man_made", "description": "POI in the search database." },
+      { "key" : "mountain_pass", "description": "POI in the search database." },
+      { "key" : "highway", "description": "POI or street in the search database (not added are: 'no', 'turning_circle', 'traffic_signals', 'mini_roundabout', 'crossing' and traffic signs)." },
+      { "key" : "aerialway", "description": "POI in the search database (unless value is 'no', 'pylon')." },
+      { "key" : "aeroway", "description": "POI in the search database (unless value is 'no')." },
+      { "key" : "amenity", "description": "POI in the search database (unless value is 'no')." },
+      { "key" : "boundary", "description": "Area in the search database (used to compute addresses of other places)." },
+      { "key" : "bridge", "description": "POI in the search database (unless value is 'no')." },
+      { "key" : "craft", "description": "POI in the search database (unless value is 'no')." },
+      { "key" : "leisure", "description": "POI in the search database (unless value is 'no')." },
+      { "key" : "office", "description": "POI in the search database (unless value is 'no')." },
+      { "key" : "railway", "description": "Geographic feature in the search database (unless value is 'no')." },
+      { "key" : "landuse", "description": "Geographic feature in the search database (unless value is 'no')." },
+      { "key" : "shop", "description": "POI in the search database (unless value is 'no')." },
+      { "key" : "tunnel", "description": "POI in the search database (unless value is 'no')." },
+      { "key" : "waterway", "description": "Geographic feature in the search database (unless value is 'riverbank')."},
+      { "key" : "place", "description": "Settlement on the search database (used to compute addresses of other places)." },
+      { "key" : "postal_code", "description": "Postcode in search database (used to compute postcodes of places around)." },
+      { "key" : "postcode", "description": "Postcode in search database (used to compute postcodes of places around)." },
+      { "key" : "addr:postcode", "description": "Postcode in search database (used to compute postcodes of places around)." },
+      { "key" : "tiger:zip_left", "description": "Postcode in search database (used to compute postcodes of places around)." },
+      { "key" : "tiger:zip_right", "description": "Postcode in search database (used to compute postcodes of places around)." },
+      { "key" : "addr:street", "description": "Used to determine the street of a house or POI. Note that a street with the same name must exist for the tag to be effective."},
+      { "key" : "addr:place", "description": "Used to determine the settlement of a house or POI with a street-less address."},
+      { "key" : "country_code", "description": "Used to determine the country a place is in."},
+      { "key" : "ISO3166-1", "description": "Used to determine the country a place is in."},
+      { "key" : "is_in:country_code", "description": "Used to determine the country a place is in."},
+      { "key" : "is_in:country", "description": "Used to determine the country a place is in."},
+      { "key" : "addr:country", "description": "Used to determine the country a place is in."},
+      { "key" : "addr:country_code", "description": "Used to determine the country a place is in."},
+      { "key" : "addr:housenumber", "description": "House number of the place (no ranges)."},
+      { "key" : "addr:conscriptionnumber", "description": "House number of the place (Eastern European system)."},
+      { "key" : "addr:streetnumber", "description": "House number of the place (Eastern European system)."},
+      { "key" : "addr:interpolation", "description": "Way along which house numbers are interpolated."} ,
+      { "key" : "tiger:county", "description": "Used to determine the address in the US (needs a place with the same name and a county suffix)."},
+      { "key" : "addr:suburb", "description": "Used to determine the address of a place."},
+      { "key" : "addr:city", "description": "Used to determine the address of a place."},
+      { "key" : "addr:state_code", "description": "Used to determine the address of a place."},
+      { "key" : "addr:state", "description": "Used to determine the address of a place."},
+      { "key" : "addr:province", "description": "Used to determine the address of a place."},
+      { "key" : "addr:district", "description": "Used to determine the address of a place."},
+      { "key" : "addr:region", "description": "Used to determine the address of a place."},
+      { "key" : "addr:county", "description": "Used to determine the address of a place."},
+      { "key" : "addr:municipality", "description": "Used to determine the address of a place."},
+      { "key" : "addr:hamlet", "description": "Used to determine the address of a place."},
+      { "key" : "addr:village", "description": "Used to determine the address of a place."},
+      { "key" : "addr:subdistrict", "description": "Used to determine the address of a place."},
+      { "key" : "addr:town", "description": "Used to determine the address of a place."},
+      { "key" : "addr:neighbourhood", "description": "Used to determine the address of a place."},
+      { "key" : "addr:quarter", "description": "Used to determine the address of a place."},
+      { "key" : "addr:parish", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:suburb", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:city", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:state_code", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:state", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:province", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:district", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:region", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:county", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:municipality", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:hamlet", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:village", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:subdistrict", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:town", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:neighbourhood", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:quarter", "description": "Used to determine the address of a place."},
+      { "key" : "is_in:parish", "description": "Used to determine the address of a place."},
+      { "key" : "admin_level", "description": "Determines the hierarchy for administrative boundaries."},
+      { "key" : "wikipedia", "description": "Linking to the right wikipedia article helps to guess the importance of a place, which determines how far up in the search results it appears."}
+   ]
+}