reintroduce cutoffs when searching for very frequent words

[nominatim.git] / nominatim / config.py

diff --git a/nominatim/config.py b/nominatim/config.py
index 1728c291e92fadec129f8a531247165dbea688ee..3344a425a5667d58f6e2cd79e916f0b9bbb8c839 100644 (file)
--- a/nominatim/config.py
+++ b/nominatim/config.py
@@ -17,6 +17,7 @@ import json
 import yaml
 
 from dotenv import dotenv_values
 import yaml
 
 from dotenv import dotenv_values

+from psycopg2.extensions import parse_dsn

 
 from nominatim.typing import StrPath
 from nominatim.errors import UsageError
 
 from nominatim.typing import StrPath
 from nominatim.errors import UsageError


@@ -46,17 +47,15 @@ def flatten_config_list(content: Any, section: str = '') -> List[Any]:
 
 
 class Configuration:
 
 
 class Configuration:

-    """ Load and manage the project configuration.
-
-        Nominatim uses dotenv to configure the software. Configuration options
-        are resolved in the following order:
-
-         * from the OS environment (or the dirctionary given in `environ`
-         * from the .env file in the project directory of the installation
-         * from the default installation in the configuration directory
+    """ This class wraps access to the configuration settings
+        for the Nominatim instance in use.

 
         All Nominatim configuration options are prefixed with 'NOMINATIM_' to
 
         All Nominatim configuration options are prefixed with 'NOMINATIM_' to

-        avoid conflicts with other environment variables.
+        avoid conflicts with other environment variables. All settings can
+        be accessed as properties of the class under the same name as the
+        setting but with the `NOMINATIM_` prefix removed. In addition, there
+        are accessor functions that convert the setting values to types
+        other than string.

     """
 
     def __init__(self, project_dir: Optional[Path],
     """
 
     def __init__(self, project_dir: Optional[Path],


@@ -98,14 +97,29 @@ class Configuration:
 
     def get_bool(self, name: str) -> bool:
         """ Return the given configuration parameter as a boolean.
 
     def get_bool(self, name: str) -> bool:
         """ Return the given configuration parameter as a boolean.

-            Values of '1', 'yes' and 'true' are accepted as truthy values,
-            everything else is interpreted as false.
+
+            Parameters:
+              name: Name of the configuration parameter with the NOMINATIM_
+                prefix removed.
+
+            Returns:
+              `True` for values of '1', 'yes' and 'true', `False` otherwise.

         """
         return getattr(self, name).lower() in ('1', 'yes', 'true')
 
 
     def get_int(self, name: str) -> int:
         """ Return the given configuration parameter as an int.
         """
         return getattr(self, name).lower() in ('1', 'yes', 'true')
 
 
     def get_int(self, name: str) -> int:
         """ Return the given configuration parameter as an int.

+
+            Parameters:
+              name: Name of the configuration parameter with the NOMINATIM_
+                prefix removed.
+
+            Returns:
+              The configuration value converted to int.
+
+            Raises:
+              ValueError: when the value is not a number.

         """
         try:
             return int(getattr(self, name))
         """
         try:
             return int(getattr(self, name))


@@ -117,8 +131,17 @@ class Configuration:
     def get_str_list(self, name: str) -> Optional[List[str]]:
         """ Return the given configuration parameter as a list of strings.
             The values are assumed to be given as a comma-sparated list and
     def get_str_list(self, name: str) -> Optional[List[str]]:
         """ Return the given configuration parameter as a list of strings.
             The values are assumed to be given as a comma-sparated list and

-            will be stripped before returning them. On empty values None
-            is returned.
+            will be stripped before returning them. 
+
+            Parameters:
+              name: Name of the configuration parameter with the NOMINATIM_
+                prefix removed.
+
+            Returns:
+              (List[str]): The comma-split parameter as a list. The
+                elements are stripped of leading and final spaces before
+                being returned.
+              (None): The configuration parameter was unset or empty.

         """
         raw = getattr(self, name)
 
         """
         raw = getattr(self, name)
 


@@ -127,9 +150,16 @@ class Configuration:
 
     def get_path(self, name: str) -> Optional[Path]:
         """ Return the given configuration parameter as a Path.
 
     def get_path(self, name: str) -> Optional[Path]:
         """ Return the given configuration parameter as a Path.

-            If a relative path is configured, then the function converts this
-            into an absolute path with the project directory as root path.
-            If the configuration is unset, None is returned.
+
+            Parameters:
+              name: Name of the configuration parameter with the NOMINATIM_
+                prefix removed.
+
+            Returns:
+              (Path): A Path object of the parameter value.
+                  If a relative path is configured, then the function converts this
+                  into an absolute path with the project directory as root path.
+              (None): The configuration parameter was unset or empty.

         """
         value = getattr(self, name)
         if not value:
         """
         value = getattr(self, name)
         if not value:


@@ -164,6 +194,18 @@ class Configuration:
         return dsn
 
 
         return dsn
 
 

+    def get_database_params(self) -> Mapping[str, str]:
+        """ Get the configured parameters for the database connection
+            as a mapping.
+        """
+        dsn = self.DATABASE_DSN
+
+        if dsn.startswith('pgsql:'):
+            return dict((p.split('=', 1) for p in dsn[6:].split(';')))
+
+        return parse_dsn(dsn)
+
+

     def get_import_style_file(self) -> Path:
         """ Return the import style file as a path object. Translates the
             name of the standard styles automatically into a file in the
     def get_import_style_file(self) -> Path:
         """ Return the import style file as a path object. Translates the
             name of the standard styles automatically into a file in the


@@ -172,7 +214,7 @@ class Configuration:
         style = getattr(self, 'IMPORT_STYLE')
 
         if style in ('admin', 'street', 'address', 'full', 'extratags'):
         style = getattr(self, 'IMPORT_STYLE')
 
         if style in ('admin', 'street', 'address', 'full', 'extratags'):

-            return self.config_dir / f'import-{style}.style'
+            return self.config_dir / f'import-{style}.lua'

 
         return self.find_config_file('', 'IMPORT_STYLE')
 
 
         return self.find_config_file('', 'IMPORT_STYLE')