send kindle from s3

add file archiving code, because we needed it for push distribution.

Also use archive for kindle. tweak send-to-kindle docs.

core/models.py

@@ -1001,8 +1001,8 @@ class Campaign(models.Model):
         self.work.make_ebooks_from_ebfs(add_ask=True)
         
     def make_unglued_ebf(self, format, watermarked):
-        ebf=EbookFile.objects.create(edition=self.work.preferred_edition, format=format)
         r=urllib2.urlopen(watermarked.download_link(format))
+        ebf=EbookFile.objects.create(edition=self.work.preferred_edition, format=format)
         ebf.file.save(path_for_file(ebf,None),ContentFile(r.read()))
         ebf.file.close()
         ebf.save()
@@ -2005,7 +2005,38 @@ class Ebook(models.Model):
             return True
         else:
             return False
-            
+    
+    def get_archive(self): # returns an archived file
+        if self.edition.ebook_files.filter(format=self.format).count()==0:
+            if self.provider is not 'Unglue.it':
+                try:
+                    r=urllib2.urlopen(self.url)
+                    try:
+                        self.filesize = int(r.info().getheaders("Content-Length")[0])
+                        if self.save:
+                            self.filesize =  self.filesize if self.filesize < 2147483647 else 2147483647  # largest safe positive integer
+                            self.save()
+                        ebf=EbookFile.objects.create(edition=self.edition, format=self.format)
+                        ebf.file.save(path_for_file(ebf,None),ContentFile(r.read()))
+                        ebf.file.close()
+                        ebf.save()
+                        return ebf.file.open()
+                    except IndexError:
+                        # response has no Content-Length header probably a bad link
+                        logging.error( 'Bad link error: {}'.format(ebook.url) )
+                except IOError:
+                    logger.error(u'could not open {}'.format(self.url) )
+            else:
+                # this shouldn't happen, except in testing perhaps
+                logger.error(u'couldn\'t find ebookfile for {}'.format(self.url) )
+                # try the url instead
+                f = urllib.urlopen(self.url)
+                return f
+        else:
+            f= self.edition.ebook_files.filter(format=self.format).order_by('-created')[0].file
+            f.open()
+            return f
+        
     def set_provider(self):
         self.provider=Ebook.infer_provider(self.url)
         return self.provider

core/tests.py

@@ -87,11 +87,18 @@ class BookLoaderTests(TestCase):
         huck = models.Work.objects.get(id=huck_id)
         self.assertTrue( huck.ebooks().count()>1)
         
+        
     def test_add_by_yaml(self):  
         space_id = bookloader.load_from_yaml('https://github.com/gitenberg-dev/metadata/raw/master/samples/pandata.yaml')
         huck_id = bookloader.load_from_yaml('https://github.com/GITenberg/Adventures-of-Huckleberry-Finn_76/raw/master/metadata.yaml')
         space = models.Work.objects.get(id=space_id)
         huck = models.Work.objects.get(id=huck_id)
+
+        #test ebook archiving
+        num_ebf= EbookFile.objects.all().count()
+        for ebook in huck.ebooks().all():
+            f = ebook.get_archive()
+        self.assertTrue(EbookFile.objects.all().count()>num_ebf)
         
     def test_valid_subject(self):
         self.assertTrue(bookloader.valid_subject('A, valid, suj\xc3t'))
@@ -455,7 +462,10 @@ class BookLoaderTests(TestCase):
         ebook = bookloader.load_gutenberg_edition(title, gutenberg_etext_id, ol_work_id, seed_isbn, epub_url, format, license, lang, publication_date)
         self.assertEqual(ebook.url, epub_url)
         
-
+    def tearDown(self):
+        for ebf in EbookFile.objects.all():
+            ebf.file.delete()
+            
 class SearchTests(TestCase):
 
     def test_basic_search(self):

frontend/templates/kindle_config.html

@@ -62,11 +62,11 @@
             document.getElementById('id_kindle_email').focus()
         </script>
     </div>
-    https://www.amazon.com/gp/digital/fiona/manage#manageDevices
+
     <div>
         <p>Don't know the email address for your device or reading app?  <a href="https://www.amazon.com/myk#manageDevices">Find it here</a>. (If you're not logged in to Amazon, you need to click "Manage Your Devices" in the "Your Kindle Account" section on the lower left side of the page.)</p>
                 
-        <p>Once we have your Kindle email, you'll be able to send unglued ebooks to your Kindle device or app with one click from any Unglue.it download page.</p>
+        <p>Once we have your Kindle email, you'll be able to send unglued ebooks to your Kindle device or app with one click from any Unglue.it download page. Note: ebooks that we send will appear in the "Docs" tab, not "Books".</p>
     </div>
 {% endwith %}
 {% endblock %}

frontend/views.py

@@ -3323,21 +3323,13 @@ def send_to_kindle(request, work_id, javascript='0'):
     This won't perfectly measure size of email, but should be safe, and is much faster than doing the check after download.
     """
     try:
-        filehandle = urllib.urlopen(ebook.url)
+        filehandle = ebook.get_archive()
     except IOError:
         # problems connection to the ebook source
         logger.error("couldn't connect error: %s", ebook.url)
         return local_response(request, javascript,  context, 5)
     if not ebook.filesize:
-        try:
-            ebook.filesize = int(filehandle.info().getheaders("Content-Length")[0])
-            if ebook.save:
-                ebook.filesize =  ebook.filesize if ebook.filesize < 2147483647 else 2147483647  # largest safe positive integer
-                ebook.save()
-        except IndexError:
-            # response has no Content-Length header probably a bad link
-            logger.error('Bad link error: %s', ebook.url)
-            return local_response(request, javascript,  context, 4)
+        return local_response(request, javascript,  context, 4)
     if ebook.filesize > models.send_to_kindle_limit:
         logger.info('ebook %s is too large to be emailed' % work.id)
         return local_response(request, javascript,  context, 0)