Check if atom data download exists (tardis-sn#2479)

* Refactor atom data download function and update
configuration file

* Update import statements in util.py and
shell_info.py

* Fix path concatenation in get_data_dir function

* Fix path issue in get_internal_configuration
function

tardis/io/atom_data/atom_web_download.py

@@ -1,4 +1,3 @@
-import os
 import logging
 
 from tardis.io.util import get_internal_data_path, download_from_url
@@ -21,7 +20,7 @@ def get_atomic_repo_config():
     return yaml.load(open(atomic_repo_fname), Loader=yaml.CLoader)
 
 
-def download_atom_data(atomic_data_name=None):
+def download_atom_data(atomic_data_name=None, force_download=False):
     """
     Download the atomic data from the repository
 
@@ -30,6 +29,9 @@ def download_atom_data(atomic_data_name=None):
     atomic_data_name : str
         if None
 
+    force_download : bool
+        if True, force download even if file exists
+
     Returns
     -------
         : None
@@ -42,10 +44,16 @@ def download_atom_data(atomic_data_name=None):
     if atomic_data_name not in atomic_repo:
         raise ValueError(f"Atomic Data name {atomic_data_name} not known")
 
-    dst_dir = os.path.join(get_data_dir(), f"{atomic_data_name}.h5")
+    dst_fname = get_data_dir() / f"{atomic_data_name}.h5"
+
+    if dst_fname.exists() and not force_download:
+        logger.warning(
+            f"Atomic Data {atomic_data_name} already exists in {dst_fname}. Will not download - override with force_download=True."
+        )
+        return
     src_url = atomic_repo[atomic_data_name]["url"]
     mirrors = tuple(atomic_repo[atomic_data_name]["mirrors"])
     checksum = atomic_repo[atomic_data_name]["md5"]
 
-    logger.info(f"Downloading atomic data from {src_url} to {dst_dir}")
-    download_from_url(src_url, dst_dir, checksum, mirrors)
+    logger.info(f"Downloading atomic data from {src_url} to {dst_fname}")
+    download_from_url(src_url, dst_fname, checksum, mirrors)

tardis/io/configuration/config_internal.py

@@ -1,22 +1,23 @@
-from tardis import __path__ as TARDIS_PATH
-import os, logging, shutil
+import logging, shutil
 import yaml
+import os
+from pathlib import Path
 
+from tardis import __path__ as TARDIS_PATH
 from astropy.config import get_config_dir
 
-TARDIS_PATH = TARDIS_PATH[0]
-DEFAULT_CONFIG_PATH = os.path.join(
-    TARDIS_PATH, "data", "default_tardis_internal_config.yml"
-)
-DEFAULT_DATA_DIR = os.path.join(
-    os.path.expanduser("~"), "Downloads", "tardis-data"
+TARDIS_PATH = Path(TARDIS_PATH[0])
+DEFAULT_CONFIG_PATH = (
+    TARDIS_PATH / "data" / "default_tardis_internal_config.yml"
 )
+
+DEFAULT_DATA_DIR = Path(os.path.expanduser("~")) / "Downloads" / "tardis-data"
+
 logger = logging.getLogger(__name__)
 
 
 def get_internal_configuration():
-
-    config_fpath = os.path.join(get_config_dir(), "tardis_internal_config.yml")
+    config_fpath = Path(get_config_dir()) / "tardis_internal_config.yml"
     if not os.path.exists(config_fpath):
         logger.warning(
             f"Configuration File {config_fpath} does not exist - creating new one from default"
@@ -27,21 +28,18 @@ def get_internal_configuration():
 
 
 def get_data_dir():
-
     config = get_internal_configuration()
     data_dir = config.get("data_dir", None)
     if data_dir is None:
-        config_fpath = os.path.join(
-            get_config_dir(), "tardis_internal_config.yml"
-        )
+        config_fpath = Path(get_config_dir()) / "tardis_internal_config.yml"
         logging.critical(
             f"\n{'*' * 80}\n\nTARDIS will download different kinds of data (e.g. atomic) to its data directory {DEFAULT_DATA_DIR}\n\n"
             f"TARDIS DATA DIRECTORY not specified in {config_fpath}:\n\n"
             f"ASSUMING DEFAULT DATA DIRECTORY {DEFAULT_DATA_DIR}\n "
             f"YOU CAN CHANGE THIS AT ANY TIME IN {config_fpath} \n\n"
             f"{'*' * 80} \n\n"
         )
-        if not os.path.exists(DEFAULT_DATA_DIR):
+        if not DEFAULT_DATA_DIR.exists():
             os.makedirs(DEFAULT_DATA_DIR)
         config["data_dir"] = DEFAULT_DATA_DIR
         yaml.dump(config, open(config_fpath, "w"), default_flow_style=False)

tardis/visualization/widgets/shell_info.py

@@ -1,5 +1,5 @@
 from tardis.base import run_tardis
-from tardis.io.atom_data.util import download_atom_data
+from tardis.io.atom_data.atom_web_download import download_atom_data
 from tardis.util.base import (
     atomic_number2element_symbol,
     species_tuple_to_string,