loop on all results pages

modules/seloger/browser.py

@@ -59,9 +59,17 @@ class SeLogerBrowser(BaseBrowser):
 
         self.location(self.buildurl('http://ws.seloger.com/search.xml', **data))
 
+        while 1:
             assert self.is_on_page(SearchResultsPage)
 
-        return self.page.iter_housings()
+            for housing in self.page.iter_housings():
+                yield housing
+
+            url = self.page.next_page_url()
+            if url is None:
+                return
+
+            self.location(url)
 
     def get_housing(self, id, obj=None):
         self.location(self.buildurl('http://ws.seloger.com/annonceDetail.xml', idAnnonce=id, noAudiotel=1))

modules/seloger/pages.py

@@ -29,6 +29,13 @@ __all__ = ['SearchResultsPage', 'HousingPage']
 
 
 class SearchResultsPage(BasePage):
+    def next_page_url(self):
+        urls = self.document.getroot().xpath('//pagesuivante')
+        if len(urls) == 0:
+            return None
+        else:
+            return urls[0].text
+
     def iter_housings(self):
         for a in self.document.getroot().xpath('//annonce'):
             housing = Housing(a.find('idannonce').text)