Merge branch 'dev'

naturtag/app/app.py

@@ -62,7 +62,7 @@ def post_init(self):
             root_level=self.settings.log_level_external,
             logfile=self.settings.logfile,
         )
-        self.threadpool = ThreadPool()
+        self.threadpool = ThreadPool(n_worker_threads=self.settings.n_worker_threads)
         self.user_dirs = UserDirs(self.settings)
 
 

naturtag/app/threadpool.py

@@ -26,9 +26,11 @@ class ThreadPool(QThreadPool):
     bar.
     """
 
-    def __init__(self, **kwargs):
+    def __init__(self, n_worker_threads: int = 0, **kwargs):
         super().__init__(**kwargs)
         self.progress = ProgressBar()
+        if n_worker_threads:
+            self.setMaxThreadCount(n_worker_threads)
 
     def schedule(
         self,

naturtag/client.py

@@ -6,13 +6,32 @@
 from time import time
 from typing import TYPE_CHECKING, Optional
 
-from pyinaturalist import ClientSession, Observation, Photo, Taxon, WrapperPaginator, iNatClient
+from pyinaturalist import (
+    ClientSession,
+    Observation,
+    Photo,
+    Taxon,
+    WrapperPaginator,
+    iNatClient,
+)
+from pyinaturalist.constants import MultiInt
 from pyinaturalist.controllers import ObservationController, TaxonController
-from pyinaturalist.converters import format_file_size
-from pyinaturalist_convert.db import get_db_observations, get_db_taxa, save_observations, save_taxa
+from pyinaturalist.converters import ensure_list, format_file_size
+from pyinaturalist_convert.db import (
+    get_db_observations,
+    get_db_taxa,
+    save_observations,
+    save_taxa,
+)
 from requests_cache import SQLiteDict
 
-from naturtag.constants import DB_PATH, DEFAULT_PAGE_SIZE, IMAGE_CACHE, ROOT_TAXON_ID, PathOrStr
+from naturtag.constants import (
+    DB_PATH,
+    DEFAULT_PAGE_SIZE,
+    IMAGE_CACHE,
+    ROOT_TAXON_ID,
+    PathOrStr,
+)
 
 if TYPE_CHECKING:
     from PySide6.QtGui import QPixmap
@@ -30,7 +49,7 @@ def __init__(self, db_path: Path = DB_PATH, **kwargs):
         self.taxa = TaxonDbController(self)
         self.observations = ObservationDbController(self, taxon_controller=self.taxa)
 
-    def from_ids(
+    def from_id(
         self, observation_id: Optional[int] = None, taxon_id: Optional[int] = None
     ) -> Optional[Observation]:
         """Get an iNaturalist observation and/or taxon matching the specified ID(s). If only a taxon ID
@@ -67,11 +86,16 @@ def __init__(self, *args, taxon_controller: 'TaxonDbController', **kwargs):
         self.taxon_controller = taxon_controller
 
     def from_ids(
-        self, *observation_ids, refresh: bool = False, taxonomy: bool = False, **params
+        self,
+        observation_ids: MultiInt,
+        refresh: bool = False,
+        taxonomy: bool = False,
+        **params,
     ) -> WrapperPaginator[Observation]:
         """Get observations by ID; first from the database, then from the API"""
         # Get any observations saved in the database (unless refreshing)
         start = time()
+        observation_ids = ensure_list(observation_ids)
         if refresh:
             observations = []
         else:
@@ -82,7 +106,7 @@ def from_ids(
         remaining_ids = set(observation_ids) - {obs.id for obs in observations}
         if remaining_ids:
             logger.debug(f'Fetching remaining {len(remaining_ids)} observations from API')
-            api_results = super().from_ids(*remaining_ids, **params).all()
+            api_results = super().from_ids(remaining_ids, **params).all()
             observations.extend(api_results)
             self.save(api_results)
 
@@ -155,22 +179,23 @@ def save(self, observations: list[Observation]):
 class TaxonDbController(TaxonController):
     def from_ids(
         self,
-        *taxon_ids: int,
+        taxon_ids: MultiInt,
         accept_partial: bool = False,
         refresh: bool = False,
         **params,
     ) -> WrapperPaginator[Taxon]:
         """Get taxa by ID; first from the database, then from the API"""
         # Get any taxa saved in the database (unless refreshing)
         start = time()
-        taxa = [] if refresh else self._get_db_taxa(list(taxon_ids), accept_partial)
+        taxon_ids = ensure_list(taxon_ids)
+        taxa = [] if refresh else self._get_db_taxa(taxon_ids, accept_partial)
         logger.debug(f'{len(taxa)} taxa found in database')
 
         # Get remaining taxa from the API and save to the database
         remaining_ids = set(taxon_ids) - {taxon.id for taxon in taxa}
         if remaining_ids:
             logger.debug(f'Fetching remaining {len(remaining_ids)} taxa from API')
-            api_results = super().from_ids(*remaining_ids, **params).all() if remaining_ids else []
+            api_results = super().from_ids(remaining_ids, **params).all() if remaining_ids else []
             taxa.extend(api_results)
             save_taxa(api_results, self.client.db_path)
 
@@ -193,7 +218,7 @@ def _get_taxonomy(self, taxa: list[Taxon]) -> list[Taxon]:
         Besides, some may be missing and need to be fetched from the API.
         """
         fetch_ids = chain.from_iterable([t.ancestor_ids + t.child_ids for t in taxa])
-        extended_taxa = {t.id: t for t in self.from_ids(*set(fetch_ids), accept_partial=True)}
+        extended_taxa = {t.id: t for t in self.from_ids(set(fetch_ids), accept_partial=True)}
 
         for taxon in taxa:
             # Depending on data source, the taxon itself may have already been added to ancestry

naturtag/controllers/taxon_controller.py

@@ -169,7 +169,7 @@ def load_user_taxa(self):
         def get_recent_taxa():
             logger.info(f'Loading {len(display_ids)} user taxa')
             return client.taxa.from_ids(
-                *display_ids, locale=app.settings.locale, accept_partial=True
+                display_ids, locale=app.settings.locale, accept_partial=True
             ).all()
 
         future = app.threadpool.schedule(get_recent_taxa, priority=QThread.LowPriority)

naturtag/metadata/inat_metadata.py

@@ -59,7 +59,7 @@ def tag_images(
     settings = settings or Settings.read()
     client = client or iNatDbClient(settings.db_path)
 
-    observation = client.from_ids(observation_id, taxon_id)
+    observation = client.from_id(observation_id, taxon_id)
     if not observation:
         return []
 
@@ -141,7 +141,7 @@ def _refresh_tags(
 
     logger.debug(f'Refreshing tags for {metadata.image_path}')
     settings = settings or Settings.read()
-    observation = client.from_ids(metadata.observation_id, metadata.taxon_id)
+    observation = client.from_id(metadata.observation_id, metadata.taxon_id)
     metadata = observation_to_metadata(
         observation,
         common_names=settings.common_names,

naturtag/settings.py

@@ -122,14 +122,17 @@ class Settings(YamlMixin):
 
     # iNaturalist
     all_ranks: bool = doc_field(
-        default=False, doc='Show all available taxonomic rank filters on taxon search page'
+        default=False,
+        doc='Show all available taxonomic rank filters on taxon search page',
     )
     casual_observations: bool = doc_field(
         default=True, doc='Include taxa from casual observations in user taxa list'
     )
     locale: str = doc_field(default='en', doc='Locale preference for species common names')
     preferred_place_id: int = doc_field(
-        default=1, converter=int, doc='Place preference for regional species common names'
+        default=1,
+        converter=int,
+        doc='Place preference for regional species common names',
     )
     search_locale: bool = doc_field(
         default=True, doc='Search common names for only your selected locale'
@@ -154,9 +157,11 @@ class Settings(YamlMixin):
     recent_image_dirs: list[Path] = field(factory=list)
     favorite_image_dirs: list[Path] = field(factory=list)
 
+    # Internal
     debug: bool = field(default=False)
     setup_complete: bool = field(default=False)
     last_obs_check: Optional[datetime] = field(default=None)
+    n_worker_threads: int = field(default=1)
 
     @classmethod
     def read(cls, path: Path = CONFIG_PATH) -> 'Settings':  # type: ignore

packaging/README.md

@@ -9,7 +9,10 @@ This directory contains scripts and config for building Naturtag packages in the
 * Taxonomy full text search database (optional download)
 
 # Release steps
-* Locally build full taxonomy FTS database (See: https://pyinaturalist-convert.readthedocs.io/en/stable/modules/fts.html)
+* Locally build full taxonomy FTS database with `build_taxon_db.py`. For reference, see:
+  * [pyinaturalist_convert.dwca](https://pyinaturalist-convert.readthedocs.io/en/stable/modules/dwca.html)
+  * [pyinaturalist_convert.taxonomy](https://pyinaturalist-convert.readthedocs.io/en/stable/modules/taxonomy.html)
+  * [pyinaturalist_convert.fts](https://pyinaturalist-convert.readthedocs.io/en/stable/modules/fts.html)
 * Export subset of db (English language, common species only) with `export_taxa.sh`
   * Commit to repo under `assets/data/taxonomy.tar.gz`
 * Update version in `pyproject.toml` and Actual Installer config (`naturtag.aip`)

packaging/build_taxon_db.py

@@ -0,0 +1,12 @@
+#!/usr/bin/env python
+from pyinaturalist_convert import (
+    aggregate_taxon_db,
+    enable_logging,
+    load_dwca_tables,
+    load_fts_taxa,
+)
+
+enable_logging('DEBUG')
+load_dwca_tables()
+aggregate_taxon_db()
+load_fts_taxa(language='all')