reintroduce cutoffs when searching for very frequent words

[nominatim.git] / nominatim / config.py

diff --git a/nominatim/config.py b/nominatim/config.py
index adc022e2e84b2d319b9b9d4d6c43099765d1be5b..3344a425a5667d58f6e2cd79e916f0b9bbb8c839 100644 (file)
--- a/nominatim/config.py
+++ b/nominatim/config.py
@@ -8,16 +8,20 @@
 Nominatim configuration accessor.
 """
 from typing import Dict, Any, List, Mapping, Optional
 Nominatim configuration accessor.
 """
 from typing import Dict, Any, List, Mapping, Optional

+import importlib.util

 import logging
 import os
 import logging
 import os

+import sys

 from pathlib import Path
 import json
 import yaml
 
 from dotenv import dotenv_values
 from pathlib import Path
 import json
 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

+import nominatim.paths

 
 LOG = logging.getLogger()
 CONFIG_CACHE : Dict[str, Any] = {}
 
 LOG = logging.getLogger()
 CONFIG_CACHE : Dict[str, Any] = {}


@@ -43,39 +47,43 @@ 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: Path, config_dir: Path,
+    def __init__(self, project_dir: Optional[Path],

                  environ: Optional[Mapping[str, str]] = None) -> None:
         self.environ = environ or os.environ
         self.project_dir = project_dir
                  environ: Optional[Mapping[str, str]] = None) -> None:
         self.environ = environ or os.environ
         self.project_dir = project_dir

-        self.config_dir = config_dir
-        self._config = dotenv_values(str((config_dir / 'env.defaults').resolve()))
-        if project_dir is not None and (project_dir / '.env').is_file():
-            self._config.update(dotenv_values(str((project_dir / '.env').resolve())))
+        self.config_dir = nominatim.paths.CONFIG_DIR
+        self._config = dotenv_values(str(self.config_dir / 'env.defaults'))
+        if self.project_dir is not None and (self.project_dir / '.env').is_file():
+            self.project_dir = self.project_dir.resolve()
+            self._config.update(dotenv_values(str(self.project_dir / '.env')))

 
         class _LibDirs:
 
         class _LibDirs:

-            pass
+            module: Path
+            osm2pgsql: Path
+            php = nominatim.paths.PHPLIB_DIR
+            sql = nominatim.paths.SQLLIB_DIR
+            data = nominatim.paths.DATA_DIR

 
         self.lib_dir = _LibDirs()
 
         self.lib_dir = _LibDirs()

+        self._private_plugins: Dict[str, object] = {}

 
 
     def set_libdirs(self, **kwargs: StrPath) -> None:
         """ Set paths to library functions and data.
         """
         for key, value in kwargs.items():
 
 
     def set_libdirs(self, **kwargs: StrPath) -> None:
         """ Set paths to library functions and data.
         """
         for key, value in kwargs.items():

-            setattr(self.lib_dir, key, Path(value).resolve())
+            setattr(self.lib_dir, key, Path(value))

 
 
     def __getattr__(self, name: str) -> str:
 
 
     def __getattr__(self, name: str) -> str:


@@ -89,14 +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))


@@ -108,8 +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)
 


@@ -118,9 +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:


@@ -129,6 +168,7 @@ class Configuration:
         cfgpath = Path(value)
 
         if not cfgpath.is_absolute():
         cfgpath = Path(value)
 
         if not cfgpath.is_absolute():

+            assert self.project_dir is not None

             cfgpath = self.project_dir / cfgpath
 
         return cfgpath.resolve()
             cfgpath = self.project_dir / cfgpath
 
         return cfgpath.resolve()


@@ -154,6 +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


@@ -162,16 +214,16 @@ 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')
 
 

-    def get_os_env(self) -> Dict[str, Optional[str]]:
+    def get_os_env(self) -> Dict[str, str]:

         """ Return a copy of the OS environment with the Nominatim configuration
             merged in.
         """
         """ Return a copy of the OS environment with the Nominatim configuration
             merged in.
         """

-        env = dict(self._config)
+        env = {k: v for k, v in self._config.items() if v is not None}

         env.update(self.environ)
 
         return env
         env.update(self.environ)
 
         return env


@@ -181,7 +233,7 @@ class Configuration:
                                config: Optional[str] = None) -> Any:
         """ Load additional configuration from a file. `filename` is the name
             of the configuration file. The file is first searched in the
                                config: Optional[str] = None) -> Any:
         """ Load additional configuration from a file. `filename` is the name
             of the configuration file. The file is first searched in the

-            project directory and then in the global settings dirctory.
+            project directory and then in the global settings directory.

 
             If `config` is set, then the name of the configuration file can
             be additionally given through a .env configuration option. When
 
             If `config` is set, then the name of the configuration file can
             be additionally given through a .env configuration option. When


@@ -215,6 +267,49 @@ class Configuration:
         return result
 
 
         return result
 
 

+    def load_plugin_module(self, module_name: str, internal_path: str) -> Any:
+        """ Load a Python module as a plugin.
+
+            The module_name may have three variants:
+
+            * A name without any '.' is assumed to be an internal module
+              and will be searched relative to `internal_path`.
+            * If the name ends in `.py`, module_name is assumed to be a
+              file name relative to the project directory.
+            * Any other name is assumed to be an absolute module name.
+
+            In either of the variants the module name must start with a letter.
+        """
+        if not module_name or not module_name[0].isidentifier():
+            raise UsageError(f'Invalid module name {module_name}')
+
+        if '.' not in module_name:
+            module_name = module_name.replace('-', '_')
+            full_module = f'{internal_path}.{module_name}'
+            return sys.modules.get(full_module) or importlib.import_module(full_module)
+
+        if module_name.endswith('.py'):
+            if self.project_dir is None or not (self.project_dir / module_name).exists():
+                raise UsageError(f"Cannot find module '{module_name}' in project directory.")
+
+            if module_name in self._private_plugins:
+                return self._private_plugins[module_name]
+
+            file_path = str(self.project_dir / module_name)
+            spec = importlib.util.spec_from_file_location(module_name, file_path)
+            if spec:
+                module = importlib.util.module_from_spec(spec)
+                # Do not add to global modules because there is no standard
+                # module name that Python can resolve.
+                self._private_plugins[module_name] = module
+                assert spec.loader is not None
+                spec.loader.exec_module(module)
+
+                return module
+
+        return sys.modules.get(module_name) or importlib.import_module(module_name)
+
+

     def find_config_file(self, filename: StrPath,
                          config: Optional[str] = None) -> Path:
         """ Resolve the location of a configuration file given a filename and
     def find_config_file(self, filename: StrPath,
                          config: Optional[str] = None) -> Path:
         """ Resolve the location of a configuration file given a filename and


@@ -262,7 +357,7 @@ class Configuration:
         """ Handler for the '!include' operator in YAML files.
 
             When the filename is relative, then the file is first searched in the
         """ Handler for the '!include' operator in YAML files.
 
             When the filename is relative, then the file is first searched in the

-            project directory and then in the global settings dirctory.
+            project directory and then in the global settings directory.

         """
         fname = loader.construct_scalar(node)
 
         """
         fname = loader.construct_scalar(node)