Merge pull request #2192 from lonvia/database-versioning

[nominatim.git] / nominatim / tools / refresh.py

diff --git a/nominatim/tools/refresh.py b/nominatim/tools/refresh.py
index 5fbb07f86f53ece24ec2b5a01daeec816b970010..5cfa1ab00e92a46d941bd1c55a3d7effbf6d38f1 100644 (file)
--- a/nominatim/tools/refresh.py
+++ b/nominatim/tools/refresh.py
@@ -2,23 +2,28 @@
 Functions for bringing auxiliary data in the database up-to-date.
 """
 import json
+import logging
 import re
+from textwrap import dedent
 
 from psycopg2.extras import execute_values
 
 from ..db.utils import execute_file
+from ..version import NOMINATIM_VERSION
 
-def update_postcodes(conn, datadir):
+LOG = logging.getLogger()
+
+def update_postcodes(dsn, sql_dir):
     """ Recalculate postcode centroids and add, remove and update entries in the
         location_postcode table. `conn` is an opne connection to the database.
     """
-    execute_file(conn, datadir / 'sql' / 'update-postcodes.sql')
+    execute_file(dsn, sql_dir / 'update-postcodes.sql')
 
 
-def recompute_word_counts(conn, datadir):
+def recompute_word_counts(dsn, sql_dir):
     """ Compute the frequency of full-word search terms.
     """
-    execute_file(conn, datadir / 'sql' / 'words_from_search_name.sql')
+    execute_file(dsn, sql_dir / 'words_from_search_name.sql')
 
 
 def _add_address_level_rows_from_entry(rows, entry):
@@ -153,12 +158,10 @@ def _get_partition_function_sql(conn, sql_dir):
 
     return replace_partition_string(sql, sorted(partitions))
 
-def create_functions(conn, config, data_dir,
+def create_functions(conn, config, sql_dir,
                      enable_diff_updates=True, enable_debug=False):
     """ (Re)create the PL/pgSQL functions.
     """
-    sql_dir = data_dir / 'sql'
-
     sql = _get_standard_function_sql(conn, config, sql_dir,
                                      enable_diff_updates, enable_debug)
     sql += _get_partition_function_sql(conn, sql_dir)
@@ -167,3 +170,113 @@ def create_functions(conn, config, data_dir,
         cur.execute(sql)
 
     conn.commit()
+
+
+WEBSITE_SCRIPTS = (
+    'deletable.php',
+    'details.php',
+    'lookup.php',
+    'polygons.php',
+    'reverse.php',
+    'search.php',
+    'status.php'
+)
+
+# constants needed by PHP scripts: PHP name, config name, type
+PHP_CONST_DEFS = (
+    ('Database_DSN', 'DATABASE_DSN', str),
+    ('Default_Language', 'DEFAULT_LANGUAGE', str),
+    ('Log_DB', 'LOG_DB', bool),
+    ('Log_File', 'LOG_FILE', str),
+    ('Max_Word_Frequency', 'MAX_WORD_FREQUENCY', int),
+    ('NoAccessControl', 'CORS_NOACCESSCONTROL', bool),
+    ('Places_Max_ID_count', 'LOOKUP_MAX_COUNT', int),
+    ('PolygonOutput_MaximumTypes', 'POLYGON_OUTPUT_MAX_TYPES', int),
+    ('Search_BatchMode', 'SEARCH_BATCH_MODE', bool),
+    ('Search_NameOnlySearchFrequencyThreshold', 'SEARCH_NAME_ONLY_THRESHOLD', str),
+    ('Term_Normalization_Rules', 'TERM_NORMALIZATION', str),
+    ('Use_Aux_Location_data', 'USE_AUX_LOCATION_DATA', bool),
+    ('Use_US_Tiger_Data', 'USE_US_TIGER_DATA', bool),
+    ('MapIcon_URL', 'MAPICON_URL', str),
+)
+
+
+def import_wikipedia_articles(dsn, data_path, ignore_errors=False):
+    """ Replaces the wikipedia importance tables with new data.
+        The import is run in a single transaction so that the new data
+        is replace seemlessly.
+
+        Returns 0 if all was well and 1 if the importance file could not
+        be found. Throws an exception if there was an error reading the file.
+    """
+    datafile = data_path / 'wikimedia-importance.sql.gz'
+
+    if not datafile.exists():
+        return 1
+
+    pre_code = """BEGIN;
+                  DROP TABLE IF EXISTS "wikipedia_article";
+                  DROP TABLE IF EXISTS "wikipedia_redirect"
+               """
+    post_code = "COMMIT"
+    execute_file(dsn, datafile, ignore_errors=ignore_errors,
+                 pre_code=pre_code, post_code=post_code)
+
+    return 0
+
+
+def recompute_importance(conn):
+    """ Recompute wikipedia links and importance for all entries in placex.
+        This is a long-running operations that must not be executed in
+        parallel with updates.
+    """
+    with conn.cursor() as cur:
+        cur.execute('ALTER TABLE placex DISABLE TRIGGER ALL')
+        cur.execute("""
+            UPDATE placex SET (wikipedia, importance) =
+               (SELECT wikipedia, importance
+                FROM compute_importance(extratags, country_code, osm_type, osm_id))
+            """)
+        cur.execute("""
+            UPDATE placex s SET wikipedia = d.wikipedia, importance = d.importance
+             FROM placex d
+             WHERE s.place_id = d.linked_place_id and d.wikipedia is not null
+                   and (s.wikipedia is null or s.importance < d.importance);
+            """)
+
+        cur.execute('ALTER TABLE placex ENABLE TRIGGER ALL')
+    conn.commit()
+
+
+def setup_website(basedir, phplib_dir, config):
+    """ Create the website script stubs.
+    """
+    if not basedir.exists():
+        LOG.info('Creating website directory.')
+        basedir.mkdir()
+
+    template = dedent("""\
+                      <?php
+
+                      @define('CONST_Debug', $_GET['debug'] ?? false);
+                      @define('CONST_LibDir', '{0}');
+                      @define('CONST_NominatimVersion', '{1[0]}.{1[1]}.{1[2]}-{1[3]}');
+
+                      """.format(phplib_dir, NOMINATIM_VERSION))
+
+    for php_name, conf_name, var_type in PHP_CONST_DEFS:
+        if var_type == bool:
+            varout = 'true' if config.get_bool(conf_name) else 'false'
+        elif var_type == int:
+            varout = getattr(config, conf_name)
+        elif not getattr(config, conf_name):
+            varout = 'false'
+        else:
+            varout = "'{}'".format(getattr(config, conf_name).replace("'", "\\'"))
+
+        template += "@define('CONST_{}', {});\n".format(php_name, varout)
+
+    template += "\nrequire_once('{}/website/{{}}');\n".format(phplib_dir)
+
+    for script in WEBSITE_SCRIPTS:
+        (basedir / script).write_text(template.format(script), 'utf-8')