Merge pull request #42 from ecohealthalliance/to_json-fix

Replace broken doc.to_json method with doc.to_dict

epitator/annodoc.py

@@ -3,7 +3,6 @@
 """Annotator"""
 from __future__ import absolute_import
 from __future__ import print_function
-import json
 from . import maximum_weight_interval_set as mwis
 import six
 import re
@@ -24,7 +23,6 @@ def __init__(self, text=None, date=None):
         else:
             raise TypeError("text must be string or unicode")
         self.tiers = {}
-        self.properties = {}
         self.date = date
 
     def add_tier(self, annotator, **kwargs):
@@ -69,21 +67,38 @@ def create_regex_tier(self, regex, label=None):
                     match.group(0))], label))
         return AnnoTier(spans, presorted=True)
 
-    def to_json(self):
-        json_obj = {'text': self.text,
-                    'properties': self.properties}
+    def to_dict(self):
+        """
+        Convert the document into a json serializable dictionary.
+        This does not store all the document's data. For a complete
+        serialization use pickle.
 
+        >>> from .annospan import AnnoSpan
+        >>> from .annotier import AnnoTier
+        >>> import datetime
+        >>> doc = AnnoDoc('one two three', date=datetime.datetime(2011, 11, 11))
+        >>> doc.tiers = {
+        ...     'test': AnnoTier([AnnoSpan(0, 3, doc), AnnoSpan(4, 7, doc)])}
+        >>> d = doc.to_dict()
+        >>> str(d['text'])
+        'one two three'
+        >>> str(d['date'])
+        '2011-11-11T00:00:00Z'
+        >>> sorted(d['tiers']['test'][0].items())
+        [('label', None), ('textOffsets', [[0, 3]])]
+        >>> sorted(d['tiers']['test'][1].items())
+        [('label', None), ('textOffsets', [[4, 7]])]
+        """
+        json_obj = {
+            'text': self.text
+        }
         if self.date:
             json_obj['date'] = self.date.strftime("%Y-%m-%dT%H:%M:%S") + 'Z'
-
-        if self.properties:
-            json_obj['properties'] = self.properties
-
         json_obj['tiers'] = {}
         for name, tier in self.tiers.items():
-            json_obj['tiers'][name] = tier.to_json()
-
-        return json.dumps(json_obj)
+            json_obj['tiers'][name] = [
+                span.to_dict() for span in tier]
+        return json_obj
 
     def filter_overlapping_spans(self, tiers=None, tier_names=None, score_func=None):
         """Remove the smaller of any overlapping spans."""

epitator/annospan.py

@@ -139,13 +139,17 @@ def groupdict(self):
         >>> number_span_g = SpanGroup([AnnoSpan(0, 3, doc, 'number'),
         ...                            AnnoSpan(4, 7, doc, 'number'),
         ...                            AnnoSpan(8, 12, doc, 'animal')])
-        >>> number_span_g.groupdict()
-        {'number': [AnnoSpan(0-3, number), AnnoSpan(4-7, number)], 'animal': [AnnoSpan(8-12, animal)]}
+        >>> number_span_g.groupdict()['number']
+        [AnnoSpan(0-3, number), AnnoSpan(4-7, number)]
+        >>> number_span_g.groupdict()['animal']
+        [AnnoSpan(8-12, animal)]
         """
         out = {}
         for base_span in self.base_spans:
             for key, values in base_span.groupdict().items():
                 out[key] = out.get(key, []) + values
+        for values in out.values():
+            values.sort()
         if self.label:
             out[self.label] = [self]
         return out

epitator/annotier.py

@@ -1,7 +1,6 @@
 #!/usr/bin/env python
 # coding=utf8
 from __future__ import absolute_import
-import json
 import re
 from .annospan import SpanGroup, AnnoSpan
 from . import maximum_weight_interval_set as mwis
@@ -39,14 +38,6 @@ def __iter__(self):
     def __getitem__(self, idx):
         return self.spans[idx]
 
-    def to_json(self):
-        docless_spans = []
-        for span in self.spans:
-            span_dict = span.__dict__.copy()
-            del span_dict['doc']
-            docless_spans.append(span_dict)
-        return json.dumps(docless_spans)
-
     def group_spans_by_containing_span(self,
                                        other_tier,
                                        allow_partial_containment=False):