Fix interruped downloads with --latest-stamps (#1219)

The most recent post is cached in NodeIterator (and
saved to the disk), and its timestamp is used, instead of the
timestamp instaloader was run.

This way, even in later resuming runs the timestamp stored is the same
that would have been stored if the first run.

Fixes #1206.

instaloader/nodeiterator.py

@@ -17,7 +17,8 @@ FrozenNodeIterator = NamedTuple('FrozenNodeIterator',
                                  ('context_username', Optional[str]),
                                  ('total_index', int),
                                  ('best_before', Optional[float]),
-                                 ('remaining_data', Optional[Dict])])
+                                 ('remaining_data', Optional[Dict]),
+                                 ('first_node', Optional[Dict])])
 FrozenNodeIterator.query_hash.__doc__ = """The GraphQL ``query_hash`` parameter."""
 FrozenNodeIterator.query_variables.__doc__ = """The GraphQL ``query_variables`` parameter."""
 FrozenNodeIterator.query_referer.__doc__ = """The HTTP referer used for the GraphQL query."""
@@ -26,7 +27,7 @@ FrozenNodeIterator.total_index.__doc__ = """Number of items that have already be
 FrozenNodeIterator.best_before.__doc__ = """Date when parts of the stored nodes might have expired."""
 FrozenNodeIterator.remaining_data.__doc__ = \
     """The already-retrieved, yet-unprocessed ``edges`` and the ``page_info`` at time of freezing."""
-
+FrozenNodeIterator.first_node.__doc__ = """Node data of the first item, if an item has been produced."""
 
 T = TypeVar('T')
 
@@ -89,6 +90,7 @@ class NodeIterator(Iterator[T]):
             self._best_before = datetime.now() + NodeIterator._shelf_life
         else:
             self._data = self._query()
+        self._first_node: Optional[Dict] = None
 
     def _query(self, after: Optional[str] = None) -> Dict:
         pagination_variables = {'first': NodeIterator._graphql_page_length}  # type: Dict[str, Any]
@@ -125,7 +127,10 @@ class NodeIterator(Iterator[T]):
             except KeyboardInterrupt:
                 self._page_index, self._total_index = page_index, total_index
                 raise
-            return self._node_wrapper(node)
+            item = self._node_wrapper(node)
+            if self._first_node is None:
+                self._first_node = node
+            return item
         if self._data['page_info']['has_next_page']:
             query_response = self._query(self._data['page_info']['end_cursor'])
             page_index, data = self._page_index, self._data
@@ -157,6 +162,15 @@ class NodeIterator(Iterator[T]):
         ).encode())
         return base64.urlsafe_b64encode(magic_hash.digest()).decode()
 
+    @property
+    def first_item(self) -> Optional[T]:
+        """
+        If this iterator has produced any items, returns the first item produced.
+
+        .. versionadded:: 4.8
+        """
+        return self._node_wrapper(self._first_node) if self._first_node is not None else None
+
     def freeze(self) -> FrozenNodeIterator:
         """Freeze the iterator for later resuming."""
         remaining_data = None
@@ -171,6 +185,7 @@ class NodeIterator(Iterator[T]):
             total_index=max(self.total_index - 1, 0),
             best_before=self._best_before.timestamp() if self._best_before else None,
             remaining_data=remaining_data,
+            first_node=self._first_node,
         )
 
     def thaw(self, frozen: FrozenNodeIterator) -> None:
@@ -197,6 +212,8 @@ class NodeIterator(Iterator[T]):
         self._total_index = frozen.total_index
         self._best_before = datetime.fromtimestamp(frozen.best_before)
         self._data = frozen.remaining_data
+        if frozen.first_node is not None:
+            self._first_node = frozen.first_node
 
 
 @contextmanager