Merge pull request #499 from vilhelmp/master

Search provider - NASA ADS

Author: keflavich

astroquery/nasa_ads/__init__.py

@@ -0,0 +1,62 @@
+# Licensed under a 3-clause BSD style license - see LICENSE.rst
+"""
+SAO/NASA ADS Query Tool
+-----------------------------------
+
+:Author: Magnus Vilhelm Persson ([email protected])
+
+"""
+
+from astropy import config as _config
+
+
+class Conf(_config.ConfigNamespace):
+    """
+    Configuration parameters for `astroquery.nasa_ads`.
+    """
+   
+    server = _config.ConfigItem(
+            'http://adswww.harvard.edu',
+            'SAO/NASA ADS main server.'
+            )
+    mirrors = _config.ConfigItem(
+            ['http://cdsads.u-strasbg.fr',
+            'http://ukads.nottingham.ac.uk',
+            'http://esoads.eso.org',
+            'http://ads.ari.uni-heidelberg.de',
+            'http://ads.inasan.ru',
+            'http://ads.mao.kiev.ua',
+            'http://ads.astro.puc.cl',
+            'http://ads.nao.ac.jp',
+            'http://ads.bao.ac.cn',
+            'http://ads.iucaa.ernet.in',
+            'http://ads.arsip.lipi.go.id',
+            'http://saaoads.chpc.ac.za',
+            'http://ads.on.br'],
+            'SAO/NASA ADS mirrors around the world'
+            )
+    advanced_path = _config.ConfigItem(
+            '/cgi-bin/nph-abs_connect',
+            'Path for advanced query (unconfirmed)'    
+            )
+    simple_path = _config.ConfigItem(
+            '/cgi-bin/basic_connect',
+            'Path for simple query (return XML)'
+            )
+    timeout = _config.ConfigItem(
+            120,
+            'Time limit for connecting to ADS server'
+    )
+        
+conf = Conf()
+
+
+conf.adsfields = ['bibcode', 'title', 'author', 'affiliation',
+        'journal', 'volume', 'pubdate', 'page', 'lastpage', 'keywords', 'keyword',
+        'origin', 'copyright', 'link', 'name', 'url', 'count', 'score', 'citations',
+        'abstract', 'doi', 'eprintid']
+
+from .core import ADSClass, ADS
+
+__all__ = ['ADSClass', 'ADS',
+        'Conf', 'conf']

astroquery/nasa_ads/core.py

@@ -0,0 +1,102 @@
+# Licensed under a 3-clause BSD style license - see LICENSE.rst
+"""
+Module to search the SAO/NASA Astrophysics Data System
+
+:author: Magnus Persson <[email protected]>
+
+"""
+
+import warnings
+from ..query import BaseQuery
+from ..utils import commons, async_to_sync
+from . import conf
+from astropy.table import Table, Column
+
+from ..utils.class_or_instance import class_or_instance
+from ..utils import commons, async_to_sync
+from .utils import *
+
+from xml.dom import minidom
+
+__all__ = ['ADS', 'ADSClass']
+
+@async_to_sync
+class ADSClass(BaseQuery):
+
+    SERVER = conf.server
+    QUERY_ADVANCED_PATH = conf.advanced_path
+    QUERY_SIMPLE_PATH = conf.simple_path
+    TIMEOUT = conf.timeout
+    
+    QUERY_SIMPLE_URL = SERVER + QUERY_SIMPLE_PATH
+    QUERY_ADVANCED_URL = SERVER + QUERY_ADVANCED_PATH
+
+    def __init__(self, *args):
+        """ set some parameters """
+        pass
+    
+    @class_or_instance
+    def query_simple(self, query_string, get_query_payload=False, get_raw_response=False):
+        self.query_string = query_string
+        request_payload = self._args_to_payload(query_string)
+        
+        response = commons.send_request(self.QUERY_SIMPLE_URL, request_payload, self.TIMEOUT)
+
+        # primarily for debug purposes, but also useful if you want to send
+        # someone a URL linking directly to the data
+        if get_query_payload:
+            return request_payload
+        if get_raw_response:
+            return response
+        # parse the XML response into AstroPy Table
+        resulttable = self._parse_response(response.encode(results.encoding).decode('utf-8'))
+        
+        return resulttable
+    
+    def _parse_response(self, response):
+        xmlrepr = minidom.parseString(response.text)
+        # Check if there are any results!
+        
+        # get the list of hits
+        hitlist = xmlrepr.childNodes[0].childNodes
+        hitlist = hitlist[1::2] # every second hit is a "line break"
+        
+        # Grab the various fields
+        titles = _get_data_from_xml(hitlist, 'title')
+        bibcode = _get_data_from_xml(hitlist, 'bibcode')
+        journal = _get_data_from_xml(hitlist, 'journal')
+        volume = _get_data_from_xml(hitlist, 'volume')
+        pubdate = _get_data_from_xml(hitlist, 'pubdate')
+        page = _get_data_from_xml(hitlist, 'page')
+        score = _get_data_from_xml(hitlist, 'score')
+        citations = _get_data_from_xml(hitlist, 'citations')
+        abstract = _get_data_from_xml(hitlist, 'abstract')
+        doi = _get_data_from_xml(hitlist, 'DOI')
+        eprintid = _get_data_from_xml(hitlist, 'eprintid')
+        authors = _get_data_from_xml(hitlist, 'author')
+        # put into AstroPy Table
+        t = Table()
+        t['title'] = titles
+        t['bibcode'] = bibcode
+        t['journal'] = journal
+        t['volume'] = volume
+        t['pubdate'] = pubdate
+        t['page'] = page
+        t['score'] = score
+        t['citations'] = citations
+        t['abstract'] = abstract 
+        t['doi'] = doi
+        t['eprintid'] = eprintid 
+        t['authors'] = authors
+        
+        return t
+         
+    def _args_to_payload(self, query_string):
+        # convert arguments to a valid requests payload
+        # i.e. a dictionary
+        return {'qsearch' : query_string, 'data_type' : 'XML'}
+
+
+
+ADS = ADSClass()
+

astroquery/nasa_ads/tests/__init__.py

@@ -0,0 +1,7 @@
+from ... import nasa_ads
+from astropy.tests.helper import remote_data
+
+@remote_data
+def test_simple():
+    x = nasa_ads.ADS.query_simple('^Persson Origin of water around deeply embedded low-mass protostars')
+    assert x[-1]['authors'][0] == 'Persson, M. V.'

astroquery/nasa_ads/tests/test_nasaads.py

@@ -0,0 +1,18 @@
+
+
+def _get_data_from_xml(doclist, fieldname, nohitreturn=None):
+    """Get the fieldname (i.e. author, title etc)
+    from minidom.parseString().childNodes[0].childNodes list
+    """
+    result = []
+    for element in doclist:
+        fieldlist = element.getElementsByTagName(fieldname)
+        try:
+            tmp = fieldlist[0]
+        except IndexError:
+            fields = [nohitreturn]
+        fields = []
+        for field in fieldlist: # this is useful for e.g. author field
+            fields.append(field.childNodes[0].data)
+        result.append(fields)
+    return result

astroquery/nasa_ads/utils.py

@@ -171,7 +171,7 @@ Directory Structure::
             if get_query_payload:
                 return request_payload
 
-            return result
+            return response
 
         @class_or_instance
         def get_images_async(self, *args):

docs/api.rst

@@ -132,6 +132,7 @@ The following modules have been completed using a common API:
   atomic/atomic.rst
   alma/alma.rst
   skyview/skyview.rst
+  nasa_ads/nasa_ads.rst
 
 These others are functional, but do not follow a common & consistent API:
 
@@ -225,6 +226,7 @@ above categories.
   lamda/lamda.rst
   nist/nist.rst
   splatalogue/splatalogue.rst
+  nasa_ads/nasa_ads.rst
 
 
 Developer documentation

docs/index.rst

@@ -0,0 +1,65 @@
+.. doctest-skip-all
+
+.. _astroquery.nasa_ads:
+
+****************************************
+NASA ADS Queries (`astroquery.nasa_ads`)
+****************************************
+
+Getting Started
+===============
+
+This module provides an interface to the online `SAO/NASA Astrophysics Data System`_.
+At the moment only the "simple search", i.e. omni-box search is available, and only 
+a subset of the results are accessible.
+
+Examples
+========
+
+Search works by specific identifier
+-----------------------------------
+.. code-block:: python
+
+    from astroquery import nasa_ads as na
+    # the "^" makes ADS to return only papers where Persson 
+    # is first author
+    results = na.ADS.query_simple('^Persson Origin of water\
+	 around deeply embedded low-mass protostars') results[0].title 
+    
+    # to sort after publication date
+    results.sort(['pubdate']) 
+    
+    #  get the title of the last hit
+    title = results[-1]['title'][0] 
+    
+    # printout the authors of the last hit
+    print results[-1]['authors']
+
+
+Get links 
+---------
+Not yet implemented.
+
+Download publisher/ArXiv PDF
+----------------------------
+Not yet implemented.
+
+Get Bibtex
+----------
+Not yet implemented.
+
+
+
+
+
+
+Reference/API
+=============
+
+#.. automodapi:: astroquery.nasa_ads:no-inheritance-diagram:
+
+.. _nasa_ads: http://adsabs.harvard.edu/
+.. _SAO/NASA Astrophysics Data System: http://adsabs.harvard.edu/
+
+
+