From 1ca86cd01d841dba9e5a71e47bc8878bfe1988f1 Mon Sep 17 00:00:00 2001
From: Adam Theisen <atheisen@anl.gov>
Date: Wed, 12 Apr 2023 10:34:01 -0500
Subject: [PATCH] Doi (#657)

* ENH: AirNow API must have updating some algorithms.  What it's displaying now is what I would have assumed would be the closed reported station.

* ENH: Adding feature for users to be able to get the citation for a datastream.  Also adds text after data are done downloading from ARM to please reference the citation of the datastream

* ENH: Changing doi.org to just doi
---
 act/discovery/__init__.py     |  2 +-
 act/discovery/get_armfiles.py | 56 +++++++++++++++++++++++++++++++++++
 act/tests/test_discovery.py   | 15 ++++++++++
 3 files changed, 72 insertions(+), 1 deletion(-)

diff --git a/act/discovery/__init__.py b/act/discovery/__init__.py
index 9268ec1faa..f3318ab0ce 100644
--- a/act/discovery/__init__.py
+++ b/act/discovery/__init__.py
@@ -10,7 +10,7 @@
     __name__,
     submodules=['get_armfiles', 'get_cropscape', 'get_airnow', 'get_noaa_psl', 'get_neon'],
     submod_attrs={
-        'get_armfiles': ['download_data'],
+        'get_armfiles': ['download_data', 'get_arm_doi'],
         'get_asos': ['get_asos'],
         'get_airnow': ['get_airnow_bounded_obs', 'get_airnow_obs', 'get_airnow_forecast'],
         'get_cropscape': ['croptype'],
diff --git a/act/discovery/get_armfiles.py b/act/discovery/get_armfiles.py
index 1bbae4c373..3d730212e1 100644
--- a/act/discovery/get_armfiles.py
+++ b/act/discovery/get_armfiles.py
@@ -8,6 +8,8 @@
 import os
 import sys
 from datetime import timedelta
+import requests
+import textwrap
 
 try:
     from urllib.request import urlopen
@@ -166,4 +168,58 @@ def download_data(username, token, datastream, startdate, enddate, time=None, ou
             'No files returned or url status error.\n' 'Check datastream name, start, and end date.'
         )
 
+    # Get ARM DOI and print it out
+    doi = get_arm_doi(datastream, start_datetime.strftime('%Y-%m-%d'), end_datetime.strftime('%Y-%m-%d'))
+    print('\nIf you use these data to prepare a publication, please cite:\n')
+    print(textwrap.fill(doi, width=80))
+    print('')
+
     return file_names
+
+
+def get_arm_doi(datastream, startdate, enddate):
+    """
+    This function will return a citation with DOI, if available, for specified
+    datastream and date range
+
+    Parameters
+    ----------
+    datastream : str
+        The name of the datastream to get a DOI for.  This must be ARM standard names
+    startdate : str
+        Start date for the citation in the format YY-MM-DD
+    enddate : str
+        End date for the citation in the format YY-MM-DD
+
+    Returns
+    -------
+    doi : str
+        Returns the citation as a string
+
+    """
+
+    site = datastream[0:3]
+    level = datastream.split('.')[-1]
+
+    # Get the instrument class code from the datastream name
+    metadata_url = 'https://adc.arm.gov/solr8/metadata/select?q=datastream%3A' + datastream
+    r = requests.get(url=metadata_url)
+    response = r.json()['response']
+    if len(response['docs']) == 0:
+        raise ValueError('Check parameters')
+    response = response['docs'][0]
+    inst_class = response['instrument_class_code']
+
+    # Get the DOI information
+    doi_url = 'https://adc.arm.gov/citationservice/citation/inst-class?id=' + inst_class + '&citationType=apa'
+    doi_url += '&site=' + site
+    doi_url += '&dataLevel=' + level
+    doi_url += '&startDate=' + startdate
+    doi_url += '&endDate=' + enddate
+    doi = requests.get(url=doi_url)
+    if len(doi.text) > 0:
+        doi = doi.json()['citation']
+    else:
+        doi = 'N/A'
+
+    return doi
diff --git a/act/tests/test_discovery.py b/act/tests/test_discovery.py
index c945839ca5..51fe97de1a 100644
--- a/act/tests/test_discovery.py
+++ b/act/tests/test_discovery.py
@@ -273,3 +273,18 @@ def test_neon():
     assert len(result) == 40
     assert any('readme' in r for r in result)
     assert any('sensor_position' in r for r in result)
+
+
+def test_arm_doi():
+    datastream = 'sgpmetE13.b1'
+    startdate = '2022-01-01'
+    enddate = '2022-12-31'
+    doi = act.discovery.get_arm_doi(datastream, startdate, enddate)
+
+    assert len(doi) > 10
+    assert isinstance(doi, str)
+    assert 'doi' in doi
+    assert 'Kyrouac' in doi
+
+    with np.testing.assert_raises(ValueError):
+        doi = act.discovery.get_arm_doi('test', startdate, enddate)