]> git.openstreetmap.org Git - nominatim.git/blobdiff - nominatim/nominatim.py
Merge pull request #2087 from lonvia/only-one-link-per-node
[nominatim.git] / nominatim / nominatim.py
old mode 100644 (file)
new mode 100755 (executable)
index 6b25cf5..b20673d
@@ -1,4 +1,4 @@
-#! /usr/bin/env python
+#! /usr/bin/env python3
 #-----------------------------------------------------------------------------
 # nominatim - [description]
 #-----------------------------------------------------------------------------
@@ -28,226 +28,186 @@ import sys
 import re
 import getpass
 from datetime import datetime
-import psycopg2
-from psycopg2.extras import wait_select
 import select
 
-log = logging.getLogger()
-
-def make_connection(options, asynchronous=False):
-    return psycopg2.connect(dbname=options.dbname, user=options.user,
-                            password=options.password, host=options.host,
-                            port=options.port, async_=asynchronous)
-
-class IndexingThread(object):
-
-    def __init__(self, thread_num, options):
-        log.debug("Creating thread {}".format(thread_num))
-        self.thread_num = thread_num
-        self.conn = make_connection(options, asynchronous=True)
-        self.wait()
-
-        self.cursor = self.conn.cursor()
-        self.perform("SET lc_messages TO 'C'")
-        self.wait()
-        self.perform(InterpolationRunner.prepare())
-        self.wait()
-        self.perform(RankRunner.prepare())
-        self.wait()
-
-        self.current_query = None
-        self.current_params = None
-
-    def wait(self):
-        wait_select(self.conn)
-        self.current_query = None
-
-    def perform(self, sql, args=None):
-        self.current_query = sql
-        self.current_params = args
-        self.cursor.execute(sql, args)
-
-    def fileno(self):
-        return self.conn.fileno()
-
-    def is_done(self):
-        if self.current_query is None:
-            return True
-
-        try:
-            if self.conn.poll() == psycopg2.extensions.POLL_OK:
-                self.current_query = None
-                return True
-        except psycopg2.extensions.TransactionRollbackError as e:
-            if e.pgcode is None:
-                raise RuntimeError("Postgres exception has no error code")
-            if e.pgcode == '40P01':
-                log.info("Deadlock detected, retry.")
-                self.cursor.execute(self.current_query, self.current_params)
-            else:
-                raise
-
-
+from indexer.progress import ProgressLogger
+from indexer.db import DBConnection, make_connection
 
-class Indexer(object):
-
-    def __init__(self, options):
-        self.options = options
-        self.conn = make_connection(options)
+log = logging.getLogger()
 
-        self.threads = []
-        self.poll = select.poll()
-        for i in range(options.threads):
-            t = IndexingThread(i, options)
-            self.threads.append(t)
-            self.poll.register(t, select.EPOLLIN)
+class RankRunner(object):
+    """ Returns SQL commands for indexing one rank within the placex table.
+    """
 
-    def run(self):
-        log.info("Starting indexing rank ({} to {}) using {} threads".format(
-                 self.options.minrank, self.options.maxrank,
-                 self.options.threads))
+    def __init__(self, rank):
+        self.rank = rank
 
-        for rank in range(self.options.minrank, 30):
-            self.index(RankRunner(rank))
+    def name(self):
+        return "rank {}".format(self.rank)
 
-        if self.options.maxrank >= 30:
-            self.index(InterpolationRunner())
-            self.index(RankRunner(30))
+    def sql_count_objects(self):
+        return """SELECT count(*) FROM placex
+                  WHERE rank_address = {} and indexed_status > 0
+               """.format(self.rank)
 
-    def index(self, obj):
-        log.info("Starting {}".format(obj.name()))
+    def sql_get_objects(self):
+        return """SELECT place_id FROM placex
+                  WHERE indexed_status > 0 and rank_address = {}
+                  ORDER BY geometry_sector""".format(self.rank)
 
-        cur = self.conn.cursor(name="main")
-        cur.execute(obj.sql_index_sectors())
+    def sql_index_place(self, ids):
+        return "UPDATE placex SET indexed_status = 0 WHERE place_id IN ({})"\
+               .format(','.join((str(i) for i in ids)))
 
-        total_tuples = 0
-        for r in cur:
-            total_tuples += r[1]
-        log.debug("Total number of rows; {}".format(total_tuples))
 
-        cur.scroll(0, mode='absolute')
+class InterpolationRunner(object):
+    """ Returns SQL commands for indexing the address interpolation table
+        location_property_osmline.
+    """
 
-        next_thread = self.find_free_thread()
-        done_tuples = 0
-        rank_start_time = datetime.now()
-        for r in cur:
-            sector = r[0]
+    def name(self):
+        return "interpolation lines (location_property_osmline)"
 
-            # Should we do the remaining ones together?
-            do_all = total_tuples - done_tuples < len(self.threads) * 1000
+    def sql_count_objects(self):
+        return """SELECT count(*) FROM location_property_osmline
+                  WHERE indexed_status > 0"""
 
-            pcur = self.conn.cursor(name='places')
+    def sql_get_objects(self):
+        return """SELECT place_id FROM location_property_osmline
+                  WHERE indexed_status > 0
+                  ORDER BY geometry_sector"""
 
-            if do_all:
-                pcur.execute(obj.sql_nosector_places())
-            else:
-                pcur.execute(obj.sql_sector_places(), (sector, ))
+    def sql_index_place(self, ids):
+        return """UPDATE location_property_osmline
+                  SET indexed_status = 0 WHERE place_id IN ({})"""\
+               .format(','.join((str(i) for i in ids)))
 
-            for place in pcur:
-                place_id = place[0]
-                log.debug("Processing place {}".format(place_id))
-                thread = next(next_thread)
+class BoundaryRunner(object):
+    """ Returns SQL commands for indexing the administrative boundaries
+        of a certain rank.
+    """
 
-                thread.perform(obj.sql_index_place(), (place_id,))
-                done_tuples += 1
+    def __init__(self, rank):
+        self.rank = rank
 
-            pcur.close()
+    def name(self):
+        return "boundaries rank {}".format(self.rank)
 
-            if do_all:
-                break
+    def sql_count_objects(self):
+        return """SELECT count(*) FROM placex
+                  WHERE indexed_status > 0
+                    AND rank_search = {}
+                    AND class = 'boundary' and type = 'administrative'""".format(self.rank)
 
-        cur.close()
+    def sql_get_objects(self):
+        return """SELECT place_id FROM placex
+                  WHERE indexed_status > 0 and rank_search = {}
+                        and class = 'boundary' and type = 'administrative'
+                  ORDER BY partition, admin_level""".format(self.rank)
 
-        for t in self.threads:
-            t.wait()
+    def sql_index_place(self, ids):
+        return "UPDATE placex SET indexed_status = 0 WHERE place_id IN ({})"\
+               .format(','.join((str(i) for i in ids)))
 
-        rank_end_time = datetime.now()
-        diff_seconds = (rank_end_time-rank_start_time).total_seconds()
+class Indexer(object):
+    """ Main indexing routine.
+    """
 
-        log.info("Done {} in {} @ {} per second - FINISHED {}\n".format(
-                 done_tuples, int(diff_seconds),
-                 done_tuples/diff_seconds, obj.name()))
+    def __init__(self, options):
+        self.minrank = max(1, options.minrank)
+        self.maxrank = min(30, options.maxrank)
+        self.conn = make_connection(options)
+        self.threads = [DBConnection(options) for i in range(options.threads)]
 
-    def find_free_thread(self):
-        thread_lookup = { t.fileno() : t for t in self.threads}
+    def index_boundaries(self):
+        log.warning("Starting indexing boundaries using {} threads".format(
+                      len(self.threads)))
 
-        done_fids = [ t.fileno() for t in self.threads ]
+        for rank in range(max(self.minrank, 5), min(self.maxrank, 26)):
+            self.index(BoundaryRunner(rank))
 
-        while True:
-            for fid in done_fids:
-                thread = thread_lookup[fid]
-                if thread.is_done():
-                    yield thread
-                else:
-                    print("not good", fid)
+    def index_by_rank(self):
+        """ Run classic indexing by rank.
+        """
+        log.warning("Starting indexing rank ({} to {}) using {} threads".format(
+                 self.minrank, self.maxrank, len(self.threads)))
 
-            done_fids = [ x[0] for x in self.poll.poll()]
+        for rank in range(max(1, self.minrank), self.maxrank):
+            self.index(RankRunner(rank))
 
-        assert(False, "Unreachable code")
+        if self.maxrank == 30:
+            self.index(RankRunner(0))
+            self.index(InterpolationRunner(), 20)
+            self.index(RankRunner(self.maxrank), 20)
+        else:
+            self.index(RankRunner(self.maxrank))
 
-class RankRunner(object):
+    def index(self, obj, batch=1):
+        """ Index a single rank or table. `obj` describes the SQL to use
+            for indexing. `batch` describes the number of objects that
+            should be processed with a single SQL statement
+        """
+        log.warning("Starting %s (using batch size %s)", obj.name(), batch)
 
-    def __init__(self, rank):
-        self.rank = rank
+        cur = self.conn.cursor()
+        cur.execute(obj.sql_count_objects())
 
-    def name(self):
-        return "rank {}".format(self.rank)
+        total_tuples = cur.fetchone()[0]
+        log.debug("Total number of rows: {}".format(total_tuples))
 
-    @classmethod
-    def prepare(cls):
-        return """PREPARE rnk_index AS
-                  UPDATE placex
-                  SET indexed_status = 0 WHERE place_id = $1"""
+        cur.close()
 
-    def sql_index_sectors(self):
-        return """SELECT geometry_sector, count(*) FROM placex
-                  WHERE rank_search = {} and indexed_status > 0
-                  GROUP BY geometry_sector
-                  ORDER BY geometry_sector""".format(self.rank)
+        progress = ProgressLogger(obj.name(), total_tuples)
 
-    def sql_nosector_places(self):
-        return """SELECT place_id FROM placex
-                  WHERE indexed_status > 0 and rank_search = {}
-                  ORDER BY geometry_sector""".format(self.rank)
+        if total_tuples > 0:
+            cur = self.conn.cursor(name='places')
+            cur.execute(obj.sql_get_objects())
 
-    def sql_sector_places(self):
-        return """SELECT place_id FROM placex
-                  WHERE indexed_status > 0 and geometry_sector = %s
-                  ORDER BY geometry_sector"""
+            next_thread = self.find_free_thread()
+            while True:
+                places = [p[0] for p in cur.fetchmany(batch)]
+                if len(places) == 0:
+                    break
 
-    def sql_index_place(self):
-        return "EXECUTE rnk_index(%s)"
+                log.debug("Processing places: {}".format(places))
+                thread = next(next_thread)
 
+                thread.perform(obj.sql_index_place(places))
+                progress.add(len(places))
 
-class InterpolationRunner(object):
+            cur.close()
 
-    def name(self):
-        return "interpolation lines (location_property_osmline)"
+            for t in self.threads:
+                t.wait()
 
-    @classmethod
-    def prepare(cls):
-        return """PREPARE ipl_index AS
-                  UPDATE location_property_osmline
-                  SET indexed_status = 0 WHERE place_id = $1"""
+        progress.done()
 
-    def sql_index_sectors(self):
-        return """SELECT geometry_sector, count(*) FROM location_property_osmline
-                  WHERE indexed_status > 0
-                  GROUP BY geometry_sector
-                  ORDER BY geometry_sector"""
+    def find_free_thread(self):
+        """ Generator that returns the next connection that is free for
+            sending a query.
+        """
+        ready = self.threads
+        command_stat = 0
 
-    def sql_nosector_places(self):
-        return """SELECT place_id FROM location_property_osmline
-                  WHERE indexed_status > 0
-                  ORDER BY geometry_sector"""
+        while True:
+            for thread in ready:
+                if thread.is_done():
+                    command_stat += 1
+                    yield thread
 
-    def sql_sector_places(self):
-        return """SELECT place_id FROM location_property_osmline
-                  WHERE indexed_status > 0 and geometry_sector = %s
-                  ORDER BY geometry_sector"""
+            # refresh the connections occasionaly to avoid potential
+            # memory leaks in Postgresql.
+            if command_stat > 100000:
+                for t in self.threads:
+                    while not t.is_done():
+                        t.wait()
+                    t.connect()
+                command_stat = 0
+                ready = self.threads
+            else:
+                ready, _, _ = select.select(self.threads, [], [])
 
-    def sql_index_place(self):
-        return "EXECUTE ipl_index(%s)"
+        assert False, "Unreachable code"
 
 
 def nominatim_arg_parser():
@@ -256,7 +216,7 @@ def nominatim_arg_parser():
     def h(s):
         return re.sub("\s\s+" , " ", s)
 
-    p = ArgumentParser(description=__doc__,
+    p = ArgumentParser(description="Indexing tool for Nominatim.",
                        formatter_class=RawDescriptionHelpFormatter)
 
     p.add_argument('-d', '--database',
@@ -274,6 +234,9 @@ def nominatim_arg_parser():
     p.add_argument('-P', '--port',
                    dest='port', action='store',
                    help='PostgreSQL server port')
+    p.add_argument('-b', '--boundary-only',
+                   dest='boundary_only', action='store_true',
+                   help='Only index administrative boundaries (ignores min/maxrank).')
     p.add_argument('-r', '--minrank',
                    dest='minrank', type=int, metavar='RANK', default=0,
                    help='Minimum/starting rank.')
@@ -301,4 +264,7 @@ if __name__ == '__main__':
         password = getpass.getpass("Database password: ")
         options.password = password
 
-    Indexer(options).run()
+    if options.boundary_only:
+        Indexer(options).index_boundaries()
+    else:
+        Indexer(options).index_by_rank()