Refactored downloading of index files

Author: hfrenzel-noaa

auxil/download_index.m

@@ -0,0 +1,59 @@
+function success = download_index(file_name, file_type)
+% download_index  This function is part of the
+% MATLAB toolbox for accessing BGC Argo float data.
+%
+% USAGE:
+%   success = download_index(file_name, file_type)
+%
+% DESCRIPTION:
+%   This function downloads the specified index file from the GDAC if
+%   necessary. If the gzipped file is downloaded, it will be unzipped.
+%
+% INPUTS:
+%   file_name : name of the index file (without path)
+%   file_type : type of the index file, e.g. 'Sprof', 'prof', or 'meta'
+%
+% OUTPUTS:
+%   success   : 1 if the file was downloaded successfully or did not
+%               need to be downloaded; 0 if it could not be downloaded
+%
+% AUTHORS:
+%   H. Frenzel, J. Sharp, A. Fassbender (NOAA-PMEL), N. Buzby (UW),
+%   J. Plant, T. Maurer, Y. Takeshita (MBARI), D. Nicholson (WHOI),
+%   and A. Gray (UW)
+%
+% CITATION:
+%   H. Frenzel*, J. Sharp*, A. Fassbender, N. Buzby, J. Plant, T. Maurer,
+%   Y. Takeshita, D. Nicholson, A. Gray, 2021. BGC-Argo-Mat: A MATLAB
+%   toolbox for accessing and visualizing Biogeochemical Argo data.
+%   Zenodo. https://doi.org/10.5281/zenodo.4971318.
+%   (*These authors contributed equally to the code.)
+%
+% LICENSE: bgc_argo_mat_license.m
+%
+% DATE: FEBRUARY 22, 2022  (Version 1.2)
+
+global Settings;
+
+dest_path = [Settings.index_dir, file_name];
+
+if do_download(dest_path)
+    if Settings.verbose
+        fprintf('%s index file will now be downloaded.\n', file_type);
+    end
+    if ~try_download(file_name, dest_path)
+        success = 0;
+        return;
+    end
+    if endsWith(file_name, '.gz')
+        gunzip(dest_path)
+    end
+else
+    if endsWith(file_name, '.gz')
+        gunzip_file = strrep(dest_path, '.gz', '');
+        if ~exist(gunzip_file, 'file')
+            gunzip(dest_path)
+        end
+    end
+end
+success = 1;

initialize_argo.m

@@ -145,29 +145,17 @@ function initialize_argo()
     'jma'; 'kma'; 'kordi'; 'meds'};
 
 % Write Sprof index file from GDAC to Index directory
-sprof = 'argo_synthetic-profile_index.txt'; % file used locally
-sprof_gz = [sprof, '.gz']; % file that is available at GDAC
-Settings.dest_path_sprof = [Settings.index_dir, sprof];
-dest_path_sprof_gz = [Settings.index_dir, sprof_gz];
-if do_download(dest_path_sprof_gz)
-    if Settings.verbose
-        disp('Sprof index file will now be downloaded.')
-        disp('Depending on your internet connection, this may take a while.')
-    end
-    if ~try_download(sprof_gz, dest_path_sprof_gz)
-        error('Sprof index file could not be downloaded')
-    end
-    gunzip(dest_path_sprof_gz)
-elseif ~exist(Settings.dest_path_sprof, 'file')
-    gunzip(dest_path_sprof_gz)
+sprof = 'argo_synthetic-profile_index.txt'; % file at GDAC
+if ~download_index([sprof, '.gz'], 'Sprof')
+    error('Sprof index file could not be downloaded')
 end
 
 % Extract information from Sprof index file
 % NOTE that some quantities will be kept per float (struct Float):
 % file_path, file_name, dac, params, wmoid, update
 % Others will be kept per profile (struct Sprof):
 % date, lat, lon, sens(ors), data_mode
-fid = fopen(Settings.dest_path_sprof);
+fid = fopen([Settings.index_dir, sprof]);
 H = textscan(fid,'%s %s %f %f %s %d %s %s %s %s','headerlines',9,...
     'delimiter',',','whitespace','');
 fclose(fid);
@@ -207,39 +195,22 @@ function initialize_argo()
 Sprof.wmo = regexp(sprof_urls,'\d{7}','once','match');
 [uwmo_sprof,ia] = unique(Sprof.wmo,'stable'); % keep list order
 Sprof.wmo = str2double(Sprof.wmo);
-ulist = sprof_urls(ia);
-dacs = regexp(ulist(:,1),'^\w+','once','match');
-Sprof_fnames = regexprep(uwmo_sprof,'\d{7}','$0_Sprof.nc');
-tmp = regexprep(ulist(:,1),'profiles.+','');
-Sprof_fp = strcat(tmp,Sprof_fnames);
 ia(end+1) = length(sprof_urls) + 1;
 bgc_prof_idx1 = ia(1:end-1);
 bgc_prof_idx2 = ia(2:end) - 1;
 
 % Write prof index file from GDAC to Index directory
-prof = 'ar_index_global_prof.txt'; % file used locally
-prof_gz = [prof, '.gz']; % gzipped file that is available at GDAC
-Settings.dest_path_prof = [Settings.index_dir, prof];
-dest_path_prof_gz = [Settings.index_dir, prof_gz];
-if do_download(dest_path_prof_gz)
-    if Settings.verbose
-        disp('prof index file will now be downloaded.')
-        disp('Depending on your internet connection, this may take a while.')
-    end
-    if ~try_download(prof_gz, dest_path_prof_gz)
-        error('prof index file could not be downloaded')
-    end
-    gunzip(dest_path_prof_gz)
-elseif ~exist(Settings.dest_path_prof, 'file')
-    gunzip(dest_path_prof_gz)
+prof = 'ar_index_global_prof.txt'; % file at GDAC
+if ~download_index([prof, '.gz'], 'prof')
+    error('prof index file could not be downloaded')
 end
 
 % Extract information from prof index file
 % NOTE that some quantities will be kept per float (struct Float):
 % file_path, file_name, dac, params, wmoid, update
 % Others will be kept per profile (struct Prof):
 % date, lat, lon, sens(ors), data_mode
-fid = fopen(Settings.dest_path_prof);
+fid = fopen([Settings.index_dir, prof]);
 H = textscan(fid,'%s %s %f %f %s %d %s %s','headerlines',9,...
     'delimiter',',','whitespace','');
 fclose(fid);
@@ -362,20 +333,14 @@ function initialize_argo()
     Sprof.date_update(bgc_prof_idx2(is_true_bgc==1));
 
 % Write meta index file from GDAC to Index directory
-% Since it is rather small, download the uncompressed file
+% Since it is rather small, download the uncompressed file directly
 meta = 'ar_index_global_meta.txt';
-Settings.dest_path_meta = [Settings.index_dir, meta];
-if do_download(Settings.dest_path_meta)
-    if Settings.verbose
-        disp('meta index file will now be downloaded.')
-    end
-    if ~try_download(meta, Settings.dest_path_meta)
-        error('meta index file could not be downloaded')
-    end
+if ~download_index(meta, 'meta')
+    error('meta index file could not be downloaded')
 end
 
 % Extract information from meta index file
-fid = fopen(Settings.dest_path_meta);
+fid = fopen([Settings.index_dir, meta]);
 H = textscan(fid,'%s %s %s %s','headerlines',9,...
     'delimiter',',','whitespace','');
 fclose(fid);