Merge pull request #105 from wflynny/robust-doi

[Fix #95] robust handling of DOIs

pubs/commands/add_cmd.py

@@ -1,3 +1,4 @@
+import argparse
 from ..uis import get_ui
 from .. import bibstruct
 from .. import content
@@ -7,13 +8,21 @@ from .. import templates
 from .. import apis
 from .. import color
 from .. import pretty
+from .. import utils
+
+
+class ValidateDOI(argparse.Action):
+    def __call__(self, parser, namespace, values, option_string=None):
+        doi = values
+        new_doi = utils.standardize_doi(doi)
+        setattr(namespace, self.dest, new_doi)
 
 
 def parser(subparsers, conf):
     parser = subparsers.add_parser('add', help='add a paper to the repository')
     parser.add_argument('bibfile', nargs='?', default=None,
                         help='bibtex file')
-    parser.add_argument('-D', '--doi', help='doi number to retrieve the bibtex entry, if it is not provided', default=None)
+    parser.add_argument('-D', '--doi', help='doi number to retrieve the bibtex entry, if it is not provided', default=None, action=ValidateDOI)
     parser.add_argument('-I', '--isbn', help='isbn number to retrieve the bibtex entry, if it is not provided', default=None)
     parser.add_argument('-d', '--docfile', help='pdf or ps file', default=None)
     parser.add_argument('-t', '--tags', help='tags associated to the paper, separated by commas',

pubs/utils.py

@@ -1,28 +1,34 @@
 # Function here may belong somewhere else. In the mean time...
+import re
 
 from . import color
 from . import pretty
 
+
 def resolve_citekey(repo, citekey, ui=None, exit_on_fail=True):
-    """Check that a citekey exists, or autocompletes it if not ambiguous."""
+    """Check that a citekey exists, or autocompletes it if not ambiguous.
-    """ :returns found citekey """
+        :returns found citekey
+    """
     # FIXME. Make me optionally non ui interactive/exiting
     citekeys = repo.citekeys_from_prefix(citekey)
     if len(citekeys) == 0:
         if ui is not None:
-            ui.error("No citekey named or beginning with '{}'".format(color.dye_out(citekey, 'citekey')))
+            ui.error("No citekey named or beginning with '{}'".format(
+                     color.dye_out(citekey, 'citekey')))
             if exit_on_fail:
                 ui.exit()
     elif len(citekeys) == 1:
         if citekeys[0] != citekey:
             if ui is not None:
-                ui.info("'{}' has been autocompleted into '{}'.".format(color.dye_out(citekey, 'citekey'), color.dye_out(citekeys[0], 'citekey')))
+                ui.info("'{}' has been autocompleted into '{}'.".format(
+                        color.dye_out(citekey, 'citekey'),
+                        color.dye_out(citekeys[0], 'citekey')))
             citekey = citekeys[0]
     elif citekey not in citekeys:
         if ui is not None:
             citekeys = sorted(citekeys)
-            ui.error("Be more specific; '{}' matches multiples citekeys:".format(
+            ui.error("Be more specific; '{}' matches multiples "
-                     citekey))
+                     "citekeys:".format(citekey))
             for c in citekeys:
                 p = repo.pull_paper(c)
                 ui.message(u'    {}'.format(pretty.paper_oneliner(p)))
@@ -44,3 +50,37 @@ def resolve_citekey_list(repo, citekeys, ui=None, exit_on_fail=True):
         ui.exit()
     else:
         return keys
+
+
+def standardize_doi(doi):
+    """
+    Given a putative doi, attempts to always return it in the form of
+    10.XXXX/...  Specifically designed to handle these cases:
+    -   https://doi.org/<doi>
+    -   http://doi.org/<doi>
+    -   https://dx.doi.org/<doi>
+    -   http://dx.doi.org/<doi>
+    -   dx.doi.org/<doi>
+    -   doi.org/<doi>
+    and attempts to verify doi adherence to DOI handbook standards and
+    crossref.org advice:
+    https://www.doi.org/doi_handbook/2_Numbering.html
+    https://www.crossref.org/blog/dois-and-matching-regular-expressions/
+
+        :returns standardized doi
+    """
+
+    doi_regexes = (
+        '(10\.\d{4,9}/[-._;()/:A-z0-9\>\<]+)',
+        '(10.1002/[^\s]+)',
+        '(10\.\d{4}/\d+-\d+X?(\d+)\d+<[\d\w]+:[\d\w]*>\d+.\d+.\w+;\d)',
+        '(10\.1021/\w\w\d+\+)',
+        '(10\.1207/[\w\d]+\&\d+_\d+)')
+    doi_pattern = re.compile('|'.join(doi_regexes))
+
+    match = doi_pattern.search(doi)
+    if not match:
+        raise ValueError("Not a valid doi: %s", doi)
+    new_doi = match.group(0)
+
+    return new_doi

tests/test_doi.py

@@ -0,0 +1,80 @@
+# coding: utf8
+
+from __future__ import unicode_literals
+import unittest
+
+from pubs.utils import standardize_doi
+
+
+class TestDOIStandardization(unittest.TestCase):
+
+    def setUp(self):
+        # some of these come from
+        # https://stackoverflow.com/questions/27910/finding-a-doi-in-a-document-or-page
+        self.crossref_dois = (
+            '10.2310/JIM.0b013e31820bab4c',
+            '10.1007/978-3-642-28108-2_19',
+            '10.1016/S0735-1097(98)00347-7',
+        )
+
+        self.hard_dois = (
+            '10.1175/1520-0485(2002)032<0870:CT>2.0.CO;2',
+            '10.1002/(SICI)1522-2594(199911)42:5<952::AID-MRM16>3.0.CO;2-S',
+            '10.1579/0044-7447(2006)35\[89:RDUICP\]2.0.CO;2',
+        )
+
+        self.currently_not_supported = (
+            '10.1007.10/978-3-642-28108-2_19',
+            '10.1000.10/123456',
+            '10.1016.12.31/nature.S0735-1097(98)2000/12/31/34:7-7',
+        )
+
+    def test_http_dxdoi_org(self):
+        doi = 'http://dx.doi.org/10.1109/5.771073'
+        sdoi = standardize_doi(doi)
+        self.assertEqual(sdoi, '10.1109/5.771073')
+
+    def test_https_dxdoi_org(self):
+        doi = 'https://dx.doi.org/10.1109/5.771073'
+        sdoi = standardize_doi(doi)
+        self.assertEqual(sdoi, '10.1109/5.771073')
+
+    def test_http_doi_org(self):
+        doi = 'http://doi.org/10.1109/5.771073'
+        sdoi = standardize_doi(doi)
+        self.assertEqual(sdoi, '10.1109/5.771073')
+
+    def test_https_doi_org(self):
+        doi = 'https://doi.org/10.1109/5.771073'
+        sdoi = standardize_doi(doi)
+        self.assertEqual(sdoi, '10.1109/5.771073')
+
+    def test_doi_org(self):
+        doi = 'doi.org/10.1109/5.771073'
+        sdoi = standardize_doi(doi)
+        self.assertEqual(sdoi, '10.1109/5.771073')
+
+    def test_dxdoi_org(self):
+        doi = 'dx.doi.org/10.1109/5.771073'
+        sdoi = standardize_doi(doi)
+        self.assertEqual(sdoi, '10.1109/5.771073')
+
+    def test_doi_colon(self):
+        doi = 'doi:10.1109/5.771073'
+        sdoi = standardize_doi(doi)
+        self.assertEqual(sdoi, '10.1109/5.771073')
+
+    def test_crossref_dois(self):
+        for doi in self.crossref_dois:
+            sdoi = standardize_doi(doi)
+            self.assertEqual(doi, sdoi)
+
+    def test_hard_dois(self):
+        for doi in self.hard_dois:
+            sdoi = standardize_doi(doi)
+            self.assertEqual(doi, sdoi)
+
+    def test_currently_not_supported(self):
+        for doi in self.currently_not_supported:
+            with self.assertRaises(ValueError):
+                standardize_doi(doi)