no upper limit for number of documents

pandora_client/__init__.py

@@ -784,7 +784,20 @@ class Client(object):
                             print(r)
 
     def _get_documents(self):
-        files = self.api.findMedia({
+        query = {
+            'conditions': [
+                {'key': 'filename', 'operator': '', 'value': value}
+                for value in DOCUMENT_FORMATS
+            ],
+            'operator': '|'
+        }
+        n = self.api.findMedia({'query': query})['data']['items']
+        if n:
+            o = 0
+            chunk = 5000
+            files = []
+            while o < n:
+                files += self.api.findMedia({
                     'query': {
                         'conditions': [
                             {'key': 'filename', 'operator': '', 'value': value}
@@ -793,19 +806,26 @@ class Client(object):
                         'operator': '|'
                     },
                     'keys': ['item', 'id', 'extension'],
-            'range': [0, 5000]
+                    'range': [o, o+chunk]
                 })['data']['items']
-        d = self.api.findDocuments({
+                o += chunk
+            d = []
+            o = 0
+            while o < len(files):
+                d += self.api.findDocuments({
                     'query': {
                         'conditions': [
                             {'key': 'oshash', 'operator': '==', 'value': f['id']}
-                    for f in files
+                            for f in files[o:o+chunk]
                         ],
                         'operator': '|'
                     },
                     'keys': ['id', 'oshash', 'extension'],
-            'range': [0, len(files)]
+                    'range': [0, chunk]
                 })['data']['items']
+                o += chunk
+        else:
+            d = []
         available = set(f['oshash']
                         for f in d if f['extension'] in DOCUMENT_FORMATS)
         missing = [(f['id'], f['item']) for f in files