Add provider for SagePub

5 files changed, 72 insertions, 1 deletions

diff --git a/README.md b/README.md
index 7108c3e..62c2b0b 100644
--- a/README.md
+++ b/README.md
@@ -36,6 +36,7 @@ reMarkable from any of the following sources:
 * [OpenReview](https://openreview.net/)
 * [PMLR](http://proceedings.mlr.press/)
 * [PubMed Central](https://www.ncbi.nlm.nih.gov/pmc/)
+* [SagePub](https://journals.sagepub.com/)
 * [SemanticScholar](https://www.semanticscholar.org/)
 * [SpringerLink](https://link.springer.com/)
 * A generic URL to a PDF file
diff --git a/paper2remarkable/providers/__init__.py b/paper2remarkable/providers/__init__.py
index c4e3eb5..e3075f0 100644
--- a/paper2remarkable/providers/__init__.py
+++ b/paper2remarkable/providers/__init__.py
@@ -12,6 +12,7 @@ from .openreview import OpenReview
 from .pdf_url import PdfUrl
 from .pmlr import PMLR
 from .pubmed import PubMed
+from .sagepub import SagePub
 from .springer import Springer
 from .semantic_scholar import SemanticScholar
 
@@ -26,6 +27,7 @@ providers = [
     OpenReview,
     PMLR,
     PubMed,
+    SagePub,
     Springer,
     SemanticScholar,
     LocalFile,
diff --git a/paper2remarkable/providers/sagepub.py b/paper2remarkable/providers/sagepub.py
new file mode 100644
index 0000000..7e76df8
--- /dev/null
+++ b/paper2remarkable/providers/sagepub.py
@@ -0,0 +1,52 @@
+# -*- coding: utf-8 -*-
+
+"""Provider for SagePub
+
+Author: G.J.J. van den Burg
+License: See LICENSE file
+Copyright: 2020, G.J.J. van den Burg
+
+"""
+
+import re
+
+from ._base import Provider
+from ._info import Informer
+from ..exceptions import URLResolutionError
+
+
+class SagePubInformer(Informer):
+
+    meta_author_key = "dc.Creator"
+    meta_title_key = "dc.Title"
+    meta_date_key = "dc.Date"
+
+    def _format_authors(self, soup_authors):
+        return super()._format_authors(soup_authors, sep=" ", idx=-1)
+
+    def _format_year(self, soup_date):
+        return soup_date.split("-")[0]
+
+
+class SagePub(Provider):
+
+    re_abs = "https?:\/\/journals\.sagepub\.com\/doi\/full\/\d{2}\.\d{4}\/\d+"
+    re_pdf = "https?:\/\/journals\.sagepub\.com\/doi\/pdf\/\d{2}\.\d{4}\/\d+"
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.informer = SagePubInformer()
+
+    def get_abs_pdf_urls(self, url):
+        if re.match(self.re_abs, url):
+            abs_url = url
+            pdf_url = url.replace("full", "pdf")
+        elif re.match(self.re_pdf, url):
+            pdf_url = url
+            abs_url = url.replace("pdf", "full")
+        else:
+            raise URLResolutionError("SagePub", url)
+        return abs_url, pdf_url
+
+    def validate(src):
+        return re.match(SagePub.re_abs, src) or re.match(SagePub.re_pdf, src)
diff --git a/paper2remarkable/utils.py b/paper2remarkable/utils.py
index c2917d5..07b1524 100644
--- a/paper2remarkable/utils.py
+++ b/paper2remarkable/utils.py
@@ -38,6 +38,7 @@ def clean_string(s):
     cleaned = "".join(c if c in allowed else "_" for c in normalized)
     while "__" in cleaned:
         cleaned = cleaned.replace("__", "_")
+    cleaned = cleaned.strip('_')
     return cleaned
 
 
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 5c8a8e4..ba1cc3a 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -25,8 +25,9 @@ from paper2remarkable.providers import (
     PMLR,
     PdfUrl,
     PubMed,
+    SagePub,
     Springer,
-    SemanticScholar
+    SemanticScholar,
 )
 
 VERBOSE = False
@@ -290,6 +291,20 @@ class TestProviders(unittest.TestCase):
         filename = prov.run(url)
         self.assertEqual(exp, os.path.basename(filename))
 
+    def test_sagepub_1(self):
+        prov = SagePub(upload=False, verbose=VERBOSE)
+        url = "https://journals.sagepub.com/doi/full/10.1177/0306312714535679"
+        exp = "Rekdal_-_Academic_Urban_Legends_2014.pdf"
+        filename = prov.run(url)
+        self.assertEqual(exp, os.path.basename(filename))
+
+    def test_sagepub_2(self):
+        prov = SagePub(upload=False, verbose=VERBOSE)
+        url = "https://journals.sagepub.com/doi/pdf/10.1177/1352458517694432"
+        exp = "Kobelt_et_al_-_New_Insights_Into_the_Burden_and_Costs_of_Multiple_Sclerosis_in_Europe_2017.pdf"
+        filename = prov.run(url)
+        self.assertEqual(exp, os.path.basename(filename))
+
 
 if __name__ == "__main__":
     unittest.main()