Allow input readers to produce dict doc output

Follows the same practice now allowed with a process_document() function for map module. Dict gets automatically wrapper as internal data for a document object. Bytes obj gets wrapped as raw data for a doc of the appropriate typeAllow input readers to produce dict doc output.

src/python/pimlico/core/modules/inputs.py

@@ -12,6 +12,7 @@
 
 from pimlico.core.modules.base import BaseModuleExecutor
 from pimlico.core.modules.execute import ModuleExecutionError
+from pimlico.core.modules.map import output_to_document
 from pimlico.datatypes import PimlicoDatatype, GroupedCorpus
 from pimlico.datatypes.base import PimlicoDatatypeReaderMeta
 from pimlico.datatypes.corpora import IterableCorpus
@@ -76,6 +77,9 @@ def __iter__(self):
         # of the factory function has made a mistake in building the iter_fn
         it = iter(self.iterate())
         doc_name, doc = next(it)
+        dp_type = self.datatype.data_point_type
+        # Like process_document() on doc map modules, we allow raw data or dicts to be returned
+        doc = output_to_document(doc, dp_type)
         if not self.datatype.data_point_type.is_type_for_doc(doc):
             raise TypeError("data iterator for input reader yielded the wrong type of document. Expected "
                             "a document of data point type {}, but got {}".format(
@@ -85,6 +89,7 @@ def __iter__(self):
         yield doc_name, doc
         # Just iterate over the rest
         for doc_name, doc in it:
+            doc = output_to_document(doc, dp_type)
             yield doc_name, doc
 
     def process_setup(self):

src/python/pimlico/datatypes/corpora/data_points.py

@@ -139,7 +139,15 @@ def is_type_for_doc(self, doc):
         """
         Check whether the given document is of this type, or a subclass of this one.
 
+        If the object is not a document instance (or, more precisely, doesn't have a
+        data_point_type attr), this will always return False.
+
         """
+        if not hasattr(doc, "data_point_type"):
+            # Sometimes things other than document instances will turn up here, e.g. when
+            # a doc map module's process_document() produces a dict or raw data output
+            # That's fine: we return false simply
+            return False
         return isinstance(doc.data_point_type, type(self))
 
     def reader_init(self, reader):