Read Internet Archive catalogs

Internet Archive publish catalog with a different format.

Signed-off-by: Daniel Castelo <dcastelo@plan.ceibal.edu.uy>

3 files changed, 53 insertions, 22 deletions

diff --git a/GetIABooksActivity.py b/GetIABooksActivity.py
index 9dba0f4..371ea79 100644
--- a/GetIABooksActivity.py
+++ b/GetIABooksActivity.py
@@ -81,7 +81,7 @@ class GetIABooksActivity(activity.Activity):
         self.show_images = True
         self.languages = {}
         self._lang_code_handler = languagenames.LanguageNames()
-        self.catalogs = {}
+        self.catalogs_configuration = {}
         self.catalog_history = []
 
         if os.path.exists('/etc/get-books.cfg'):
@@ -221,7 +221,11 @@ class GetIABooksActivity(activity.Activity):
                     catalog_config['name'] = catalog
                     catalog_config['summary_field'] = \
                         source_config['summary_field']
-                    self.catalogs[catalog] = catalog_config
+                    self.catalogs_configuration[catalog] = catalog_config
+
+        self.source = _SOURCES_CONFIG.keys()[0]
+
+        self.filter_catalogs_by_source()
 
         logging.error('languages %s', self.languages)
         logging.error('catalogs %s', self.catalogs)
@@ -495,16 +499,8 @@ class GetIABooksActivity(activity.Activity):
         self.bt_move_up_catalog.hide_image()
         self.treecol.set_widget(self.bt_move_up_catalog)
 
-        if len(self.catalogs) > 0:
-            self.catalog_history.append({'title': _('Catalogs'),
-                    'catalogs': self.catalogs})
-            self.categories = []
-            self.path_iter = {}
-            for key in self.catalogs.keys():
-                self.categories.append({'text': key, 'dentro': []})
-            self.treemodel.clear()
-            for p in self.categories:
-                self.path_iter[p['text']] = self.treemodel.append([p['text']])
+        self.load_source_catalogs()
+
         self.tree_scroller = Gtk.ScrolledWindow(hadjustment=None,
                 vadjustment=None)
         self.tree_scroller.set_policy(Gtk.PolicyType.NEVER,
@@ -597,6 +593,29 @@ class GetIABooksActivity(activity.Activity):
         if len(self.catalogs) > 0:
             self.bt_catalogs.set_active(True)
 
+    def filter_catalogs_by_source(self):
+        self.catalogs = {}
+        for catalog_key in self.catalogs_configuration:
+            catalog = self.catalogs_configuration[catalog_key]
+            if catalog['source'] == self.source:
+                self.catalogs[catalog_key] = catalog
+
+    def load_source_catalogs(self):
+        self.filter_catalogs_by_source()
+
+        if len(self.catalogs) > 0:
+            self.categories = []
+            self.path_iter = {}
+            self.catalog_history = []
+            self.catalog_history.append({'title': _('Catalogs'),
+                'catalogs': self.catalogs})
+            for key in self.catalogs.keys():
+                self.categories.append({'text': key, 'dentro': []})
+            self.treemodel.clear()
+
+            for p in self.categories:
+                self.path_iter[p['text']] = self.treemodel.append([p['text']])
+
     def can_close(self):
         if self.queryresults is not None:
             self.queryresults.cancel()
@@ -882,13 +901,11 @@ class GetIABooksActivity(activity.Activity):
             self.find_books(search_terms)
         # enable/disable catalogs button if configuration is available
         self.source = self._books_toolbar.source_combo.props.value
-        have_catalogs = False
-        for catalog_name in self.catalogs.keys():
-            catalog_config = self.catalogs[catalog_name]
-            if catalog_config['source'] == self.source:
-                have_catalogs = True
-                break
-        if have_catalogs:
+
+        # Get catalogs for this source
+        self.load_source_catalogs()
+
+        if len(self.catalogs) > 0:
             self.bt_catalogs.show()
             self.bt_catalogs.set_active(True)
         else:
diff --git a/get-books.cfg b/get-books.cfg
index dd07716..385403a 100644
--- a/get-books.cfg
+++ b/get-books.cfg
@@ -22,4 +22,6 @@ Public Domain Books = http://www.feedbooks.com/publicdomain/catalog.atom
 name = Internet Archive
 query_uri = http://bookserver.archive.org/catalog/opensearch?q=
 opds_cover = http://opds-spec.org/image
-#opds_cover = http://opds-spec.org/image/thumbnail
+
+[Catalogs_Internet Archive]
+Index = http://bookserver.archive.org/catalog/
diff --git a/opds.py b/opds.py
index ced6964..46158f2 100644
--- a/opds.py
+++ b/opds.py
@@ -38,6 +38,8 @@ _REL_OPDS_ACQUISTION = u'http://opds-spec.org/acquisition'
 _REL_SUBSECTION = 'subsection'
 _REL_OPDS_POPULAR = u'http://opds-spec.org/sort/popular'
 _REL_OPDS_NEW = u'http://opds-spec.org/sort/new'
+_REL_ALTERNATE = 'alternate'
+_REL_CRAWLABLE = 'http://opds-spec.org/crawlable'
 
 GObject.threads_init()
 
@@ -91,10 +93,18 @@ class DownloadThread(threading.Thread):
         else:
             feedobj = feedparser.parse(self.obj._uri)
 
+        # Get catalog Type
+        CATALOG_TYPE = 'COMMON'
+        if 'links' in feedobj['feed']:
+            for link in feedobj['feed']['links']:
+                if link['rel'] == _REL_CRAWLABLE:
+                    CATALOG_TYPE = 'CRAWLABLE'
+                    break
+
         for entry in feedobj['entries']:
-            if entry_type(entry) == 'BOOK':
+            if entry_type(entry) == 'BOOK' and CATALOG_TYPE is not 'CRAWLABLE':
                 self.obj._booklist.append(Book(self.obj._configuration, entry))
-            elif entry_type(entry) == 'CATALOG':
+            elif entry_type(entry) == 'CATALOG' or CATALOG_TYPE == 'CRAWLABLE':
                 self.obj._cataloglist.append( \
                     Book(self.obj._configuration, entry))
 
@@ -147,6 +157,8 @@ class Book(object):
             elif link['rel'] in \
             [_REL_OPDS_POPULAR, _REL_OPDS_NEW, _REL_SUBSECTION]:
                 ret[link['type']] = link['href']
+            elif link['rel'] == _REL_ALTERNATE:
+                ret[link['type']] = link['href']
             else:
                 pass
         return ret