Merge pull request #3167 from lonvia/explicit-encoding

[nominatim.git] / nominatim / api / v1 / server_glue.py

diff --git a/nominatim/api/v1/server_glue.py b/nominatim/api/v1/server_glue.py
index 865e13318c61732f2f29f04a77fee3f1f4c28344..95484c5bea8bce16ae10c2eda41759fa903c65b2 100644 (file)
--- a/nominatim/api/v1/server_glue.py
+++ b/nominatim/api/v1/server_glue.py
@@ -15,24 +15,28 @@ import dataclasses
 import math
 from urllib.parse import urlencode
 
 import math
 from urllib.parse import urlencode
 

+import sqlalchemy as sa
+

 from nominatim.errors import UsageError
 from nominatim.config import Configuration
 import nominatim.api as napi
 import nominatim.api.logging as loglib
 from nominatim.api.v1.format import dispatch as formatting
 from nominatim.errors import UsageError
 from nominatim.config import Configuration
 import nominatim.api as napi
 import nominatim.api.logging as loglib
 from nominatim.api.v1.format import dispatch as formatting

+from nominatim.api.v1.format import RawDataList

 from nominatim.api.v1 import helpers
 
 from nominatim.api.v1 import helpers
 

-CONTENT_TYPE = {
-  'text': 'text/plain; charset=utf-8',
-  'xml': 'text/xml; charset=utf-8',
-  'debug': 'text/html; charset=utf-8'
-}
+CONTENT_TEXT = 'text/plain; charset=utf-8'
+CONTENT_XML = 'text/xml; charset=utf-8'
+CONTENT_HTML = 'text/html; charset=utf-8'
+CONTENT_JSON = 'application/json; charset=utf-8'
+
+CONTENT_TYPE = {'text': CONTENT_TEXT, 'xml': CONTENT_XML, 'debug': CONTENT_HTML}

 
 class ASGIAdaptor(abc.ABC):
     """ Adapter class for the different ASGI frameworks.
         Wraps functionality over concrete requests and responses.
     """
 
 class ASGIAdaptor(abc.ABC):
     """ Adapter class for the different ASGI frameworks.
         Wraps functionality over concrete requests and responses.
     """

-    content_type: str = 'text/plain; charset=utf-8'
+    content_type: str = CONTENT_TEXT

 
     @abc.abstractmethod
     def get(self, name: str, default: Optional[str] = None) -> Optional[str]:
 
     @abc.abstractmethod
     def get(self, name: str, default: Optional[str] = None) -> Optional[str]:


@@ -55,7 +59,7 @@ class ASGIAdaptor(abc.ABC):
 
 
     @abc.abstractmethod
 
 
     @abc.abstractmethod

-    def create_response(self, status: int, output: str) -> Any:
+    def create_response(self, status: int, output: str, num_results: int) -> Any:

         """ Create a response from the given parameters. The result will
             be returned by the endpoint functions. The adaptor may also
             return None when the response is created internally with some
         """ Create a response from the given parameters. The result will
             be returned by the endpoint functions. The adaptor may also
             return None when the response is created internally with some


@@ -66,6 +70,11 @@ class ASGIAdaptor(abc.ABC):
             body of the response to 'output'.
         """
 
             body of the response to 'output'.
         """
 

+    @abc.abstractmethod
+    def base_uri(self) -> str:
+        """ Return the URI of the original request.
+        """
+

 
     @abc.abstractmethod
     def config(self) -> Configuration:
 
     @abc.abstractmethod
     def config(self) -> Configuration:


@@ -73,19 +82,19 @@ class ASGIAdaptor(abc.ABC):
         """
 
 
         """
 
 

-    def build_response(self, output: str, status: int = 200) -> Any:
+    def build_response(self, output: str, status: int = 200, num_results: int = 0) -> Any:

         """ Create a response from the given output. Wraps a JSONP function
             around the response, if necessary.
         """
         """ Create a response from the given output. Wraps a JSONP function
             around the response, if necessary.
         """

-        if self.content_type == 'application/json' and status == 200:
+        if self.content_type == CONTENT_JSON and status == 200:

             jsonp = self.get('json_callback')
             if jsonp is not None:
                 if any(not part.isidentifier() for part in jsonp.split('.')):
                     self.raise_error('Invalid json_callback value')
                 output = f"{jsonp}({output})"
             jsonp = self.get('json_callback')
             if jsonp is not None:
                 if any(not part.isidentifier() for part in jsonp.split('.')):
                     self.raise_error('Invalid json_callback value')
                 output = f"{jsonp}({output})"

-                self.content_type = 'application/javascript'
+                self.content_type = 'application/javascript; charset=utf-8'

 
 

-        return self.create_response(status, output)
+        return self.create_response(status, output, num_results)

 
 
     def raise_error(self, msg: str, status: int = 400) -> NoReturn:
 
 
     def raise_error(self, msg: str, status: int = 400) -> NoReturn:


@@ -93,16 +102,16 @@ class ASGIAdaptor(abc.ABC):
             message. The message will be formatted according to the
             output format chosen by the request.
         """
             message. The message will be formatted according to the
             output format chosen by the request.
         """

-        if self.content_type == 'text/xml; charset=utf-8':
+        if self.content_type == CONTENT_XML:

             msg = f"""<?xml version="1.0" encoding="UTF-8" ?>
                       <error>
                         <code>{status}</code>
                         <message>{msg}</message>
                       </error>
                    """
             msg = f"""<?xml version="1.0" encoding="UTF-8" ?>
                       <error>
                         <code>{status}</code>
                         <message>{msg}</message>
                       </error>
                    """

-        elif self.content_type == 'application/json':
+        elif self.content_type == CONTENT_JSON:

             msg = f"""{{"error":{{"code":{status},"message":"{msg}"}}}}"""
             msg = f"""{{"error":{{"code":{status},"message":"{msg}"}}}}"""

-        elif self.content_type == 'text/html; charset=utf-8':
+        elif self.content_type == CONTENT_HTML:

             loglib.log().section('Execution error')
             loglib.log().var_dump('Status', status)
             loglib.log().var_dump('Message', msg)
             loglib.log().section('Execution error')
             loglib.log().var_dump('Status', status)
             loglib.log().var_dump('Message', msg)


@@ -196,7 +205,7 @@ class ASGIAdaptor(abc.ABC):
         """
         if self.get_bool('debug', False):
             loglib.set_log_output('html')
         """
         if self.get_bool('debug', False):
             loglib.set_log_output('html')

-            self.content_type = 'text/html; charset=utf-8'
+            self.content_type = CONTENT_HTML

             return True
 
         return False
             return True
 
         return False


@@ -226,7 +235,7 @@ class ASGIAdaptor(abc.ABC):
             self.raise_error("Parameter 'format' must be one of: " +
                               ', '.join(formatting.list_formats(result_type)))
 
             self.raise_error("Parameter 'format' must be one of: " +
                               ', '.join(formatting.list_formats(result_type)))
 

-        self.content_type = CONTENT_TYPE.get(fmt, 'application/json')
+        self.content_type = CONTENT_TYPE.get(fmt, CONTENT_JSON)

         return fmt
 
 
         return fmt
 
 


@@ -250,7 +259,7 @@ class ASGIAdaptor(abc.ABC):
                 numgeoms += 1
 
         if numgeoms > self.config().get_int('POLYGON_OUTPUT_MAX_TYPES'):
                 numgeoms += 1
 
         if numgeoms > self.config().get_int('POLYGON_OUTPUT_MAX_TYPES'):

-            self.raise_error('Too many polgyon output options selected.')
+            self.raise_error('Too many polygon output options selected.')

 
         return {'address_details': True,
                 'geometry_simplification': self.get_float('polygon_threshold', 0.0),
 
         return {'address_details': True,
                 'geometry_simplification': self.get_float('polygon_threshold', 0.0),


@@ -294,7 +303,7 @@ async def details_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) ->
 
     result = await api.details(place,
                                address_details=params.get_bool('addressdetails', False),
 
     result = await api.details(place,
                                address_details=params.get_bool('addressdetails', False),

-                               linked_places=params.get_bool('linkedplaces', False),
+                               linked_places=params.get_bool('linkedplaces', True),

                                parented_places=params.get_bool('hierarchy', False),
                                keywords=params.get_bool('keywords', False),
                                geometry_output = napi.GeometryFormat.GEOJSON
                                parented_places=params.get_bool('hierarchy', False),
                                keywords=params.get_bool('keywords', False),
                                geometry_output = napi.GeometryFormat.GEOJSON


@@ -315,7 +324,7 @@ async def details_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) ->
                   'group_hierarchy': params.get_bool('group_hierarchy', False),
                   'icon_base_url': params.config().MAPICON_URL})
 
                   'group_hierarchy': params.get_bool('group_hierarchy', False),
                   'icon_base_url': params.config().MAPICON_URL})
 

-    return params.build_response(output)
+    return params.build_response(output, num_results=1)

 
 
 async def reverse_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) -> Any:
 
 
 async def reverse_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) -> Any:


@@ -332,7 +341,7 @@ async def reverse_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) ->
     result = await api.reverse(coord, **details)
 
     if debug:
     result = await api.reverse(coord, **details)
 
     if debug:

-        return params.build_response(loglib.get_and_disable())
+        return params.build_response(loglib.get_and_disable(), num_results=1 if result else 0)

 
     if fmt == 'xml':
         queryparts = {'lat': str(coord.lat), 'lon': str(coord.lon), 'format': 'xml'}
 
     if fmt == 'xml':
         queryparts = {'lat': str(coord.lat), 'lon': str(coord.lon), 'format': 'xml'}


@@ -354,7 +363,7 @@ async def reverse_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) ->
     output = formatting.format_result(napi.ReverseResults([result] if result else []),
                                       fmt, fmt_options)
 
     output = formatting.format_result(napi.ReverseResults([result] if result else []),
                                       fmt, fmt_options)
 

-    return params.build_response(output)
+    return params.build_response(output, num_results=1 if result else 0)

 
 
 async def lookup_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) -> Any:
 
 
 async def lookup_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) -> Any:


@@ -379,7 +388,7 @@ async def lookup_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) -> A
         results = napi.SearchResults()
 
     if debug:
         results = napi.SearchResults()
 
     if debug:

-        return params.build_response(loglib.get_and_disable())
+        return params.build_response(loglib.get_and_disable(), num_results=len(results))

 
     fmt_options = {'extratags': params.get_bool('extratags', False),
                    'namedetails': params.get_bool('namedetails', False),
 
     fmt_options = {'extratags': params.get_bool('extratags', False),
                    'namedetails': params.get_bool('namedetails', False),


@@ -389,7 +398,7 @@ async def lookup_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) -> A
 
     output = formatting.format_result(results, fmt, fmt_options)
 
 
     output = formatting.format_result(results, fmt, fmt_options)
 

-    return params.build_response(output)
+    return params.build_response(output, num_results=len(results))

 
 
 async def _unstructured_search(query: str, api: napi.NominatimAPIAsync,
 
 
 async def _unstructured_search(query: str, api: napi.NominatimAPIAsync,


@@ -445,17 +454,24 @@ async def search_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) -> A
     if params.get('featureType', None) is not None:
         details['layers'] = napi.DataLayer.ADDRESS
 
     if params.get('featureType', None) is not None:
         details['layers'] = napi.DataLayer.ADDRESS
 

+    # unstructured query parameters

     query = params.get('q', None)
     query = params.get('q', None)

+    # structured query parameters

     queryparts = {}
     queryparts = {}

+    for key in ('amenity', 'street', 'city', 'county', 'state', 'postalcode', 'country'):
+        details[key] = params.get(key, None)
+        if details[key]:
+            queryparts[key] = details[key]
+

     try:
         if query is not None:
     try:
         if query is not None:

+            if queryparts:
+                params.raise_error("Structured query parameters"
+                                   "(amenity, street, city, county, state, postalcode, country)"
+                                   " cannot be used together with 'q' parameter.")

             queryparts['q'] = query
             results = await _unstructured_search(query, api, details)
         else:
             queryparts['q'] = query
             results = await _unstructured_search(query, api, details)
         else:

-            for key in ('amenity', 'street', 'city', 'county', 'state', 'postalcode', 'country'):
-                details[key] = params.get(key, None)
-                if details[key]:
-                    queryparts[key] = details[key]

             query = ', '.join(queryparts.values())
 
             results = await api.search_address(**details)
             query = ', '.join(queryparts.values())
 
             results = await api.search_address(**details)


@@ -468,7 +484,7 @@ async def search_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) -> A
         results = helpers.deduplicate_results(results, max_results)
 
     if debug:
         results = helpers.deduplicate_results(results, max_results)
 
     if debug:

-        return params.build_response(loglib.get_and_disable())
+        return params.build_response(loglib.get_and_disable(), num_results=len(results))

 
     if fmt == 'xml':
         helpers.extend_query_parts(queryparts, details,
 
     if fmt == 'xml':
         helpers.extend_query_parts(queryparts, details,


@@ -478,7 +494,7 @@ async def search_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) -> A
                                    (str(r.place_id) for r in results if r.place_id))
         queryparts['format'] = fmt
 
                                    (str(r.place_id) for r in results if r.place_id))
         queryparts['format'] = fmt
 

-        moreurl = urlencode(queryparts)
+        moreurl = params.base_uri() + '/search?' + urlencode(queryparts)

     else:
         moreurl = ''
 
     else:
         moreurl = ''
 


@@ -491,7 +507,59 @@ async def search_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) -> A
 
     output = formatting.format_result(results, fmt, fmt_options)
 
 
     output = formatting.format_result(results, fmt, fmt_options)
 

-    return params.build_response(output)
+    return params.build_response(output, num_results=len(results))
+
+
+async def deletable_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) -> Any:
+    """ Server glue for /deletable endpoint.
+        This is a special endpoint that shows polygons that have been
+        deleted or are broken in the OSM data but are kept in the
+        Nominatim database to minimize disruption.
+    """
+    fmt = params.parse_format(RawDataList, 'json')
+
+    async with api.begin() as conn:
+        sql = sa.text(""" SELECT p.place_id, country_code,
+                                 name->'name' as name, i.*
+                          FROM placex p, import_polygon_delete i
+                          WHERE p.osm_id = i.osm_id AND p.osm_type = i.osm_type
+                                AND p.class = i.class AND p.type = i.type
+                      """)
+        results = RawDataList(r._asdict() for r in await conn.execute(sql))
+
+    return params.build_response(formatting.format_result(results, fmt, {}))
+
+
+async def polygons_endpoint(api: napi.NominatimAPIAsync, params: ASGIAdaptor) -> Any:
+    """ Server glue for /polygons endpoint.
+        This is a special endpoint that shows polygons that have changed
+        thier size but are kept in the Nominatim database with their
+        old area to minimize disruption.
+    """
+    fmt = params.parse_format(RawDataList, 'json')
+    sql_params: Dict[str, Any] = {
+        'days': params.get_int('days', -1),
+        'cls': params.get('class')
+    }
+    reduced = params.get_bool('reduced', False)
+
+    async with api.begin() as conn:
+        sql = sa.select(sa.text("""osm_type, osm_id, class, type,
+                                   name->'name' as name,
+                                   country_code, errormessage, updated"""))\
+                .select_from(sa.text('import_polygon_error'))
+        if sql_params['days'] > 0:
+            sql = sql.where(sa.text("updated > 'now'::timestamp - make_interval(days => :days)"))
+        if reduced:
+            sql = sql.where(sa.text("errormessage like 'Area reduced%'"))
+        if sql_params['cls'] is not None:
+            sql = sql.where(sa.text("class = :cls"))
+
+        sql = sql.order_by(sa.literal_column('updated').desc()).limit(1000)
+
+        results = RawDataList(r._asdict() for r in await conn.execute(sql, sql_params))
+
+    return params.build_response(formatting.format_result(results, fmt, {}))

 
 
 EndpointFunc = Callable[[napi.NominatimAPIAsync, ASGIAdaptor], Any]
 
 
 EndpointFunc = Callable[[napi.NominatimAPIAsync, ASGIAdaptor], Any]


@@ -501,5 +569,7 @@ ROUTES = [
     ('details', details_endpoint),
     ('reverse', reverse_endpoint),
     ('lookup', lookup_endpoint),
     ('details', details_endpoint),
     ('reverse', reverse_endpoint),
     ('lookup', lookup_endpoint),

-    ('search', search_endpoint)
+    ('search', search_endpoint),
+    ('deletable', deletable_endpoint),
+    ('polygons', polygons_endpoint),

 ]
 ]