feat: add topical._hashtag_latlng and social.community base

lbsntransform/config/config.py

@@ -379,6 +379,7 @@ def parse_args(self):
                                    '* hashtag  '
                                    '* emoji  '
                                    '* term  '
+                                   '* _hashtag_latlng  '
                                    '* _term_latlng  '
                                    '* _emoji_latlng  '
                                    '* monthofyear  '

lbsntransform/output/hll/base/social.py

@@ -3,7 +3,7 @@
 """Hll bases for social facet
 """
 
-from typing import Tuple
+from typing import Tuple, Union
 
 from lbsnstructure import lbsnstructure_pb2 as lbsn
 from lbsntransform.output.hll import hll_bases as hll
@@ -61,15 +61,18 @@ class CommunityBase(SocialBase):
     """Extends Social Base Class"""
     NAME = hll.HllBaseRef(facet=FACET, base='community')
 
-    def __init__(self, record: lbsn.Origin = None):
+    def __init__(self, record: Union[lbsn.Origin, lbsn.Post] = None):
         super().__init__()
         self.attrs['name'] = None
         self.key['id'] = None
         if record is None:
             # init empty
             return
-        self.attrs['name'] = record.DESCRIPTOR.name
-        self.key['id'] = record.origin_id
+        if hasattr(record, "origin_id"):
+            self.key['id'] = record.origin_id
+            self.attrs['name'] = record.DESCRIPTOR.name
+        else:
+            self.key['id'] = record.pkey.origin.origin_id
 
 
 class CultureBase(SocialBase):

lbsntransform/output/hll/base/topical.py

@@ -194,6 +194,45 @@ def __init__(self, record: lbsn.Post = None, term: str = None):
                 "from lbsn.Post")
 
 
+class HashtagLatLngBase(hll.HllBase):
+    """Composite Base (c-base) that extends from HLL base Class
+
+    Note: To distinguish c-bases which are composite bases combining
+    aspects from multiple facets, they're termed with a leading underscore
+    """
+    NAME = hll.HllBaseRef(facet=FACET, base='_hashtag_latlng')
+
+    def __init__(self, record: lbsn.Post = None, hashtag: str = None):
+        super().__init__()
+        self.key['latitude'] = None
+        self.key['longitude'] = None
+        self.key['hashtag'] = None
+        self.attrs['latlng_geom'] = None
+        self.metrics['date_hll'] = set()
+        if hashtag is None:
+            # init empty
+            return
+        self.key['hashtag'] = hashtag.lower()
+        if record is None:
+            # init empty
+            return
+        if isinstance(record, lbsn.Post):
+            coordinates_geom = record.post_latlng
+            coordinates = HF.get_coordinates_from_ewkt(
+                coordinates_geom
+            )
+            self.key['latitude'] = coordinates.lat
+            self.key['longitude'] = coordinates.lng
+            # additional (optional) attributes
+            # formatted ready for sql upsert
+            self.attrs['latlng_geom'] = HF.return_ewkb_from_geotext(
+                coordinates_geom)
+        else:
+            raise ValueError(
+                "Parsing of LatLngBase only supported "
+                "from lbsn.Post")
+
+
 class EmojiLatLngBase(hll.HllBase):
     """Composite Base (c-base) that extends from HLL base Class
 

lbsntransform/output/hll/hll_bases.py

@@ -164,6 +164,12 @@ def base_factory(facet=None, base=None, record: lbsn.Post = None):
             # create base for each term
             records.append(
                 base_structure(record=record, term=term))
+    elif base == '_hashtag_latlng':
+        # any hashtag explicitly used
+        tag_terms = HF.filter_terms(record.hashtags)
+        for tag in tag_terms:
+            records.append(
+                base_structure(record=record, hashtag=tag))
     elif base == '_emoji_latlng':
         # any term mentioned in title,
         # body or hashtag

lbsntransform/output/hll/shared_structure_proto_hlldb.py

@@ -20,7 +20,8 @@
     temporal.DateBase, temporal.MonthBase,
     temporal.YearBase, topical.TermBase,
     topical.EmojiBase, topical.TermLatLngBase,
-    topical.EmojiLatLngBase]
+    topical.HashtagLatLngBase, topical.EmojiLatLngBase,
+    social.CommunityBase]
 
 
 class ProtoHLLMapping():
@@ -144,7 +145,8 @@ def extract_hll_bases(
                 base_list.extend(base_records)
             # Topical Facet
             topical_bases = [
-                'hashtag', 'emoji', 'term', '_term_latlng', '_emoji_latlng']
+                'hashtag', 'emoji', 'term', '_term_latlng', '_emoji_latlng',
+                '_hashtag_latlng']
             topical_bases = self.filter_bases(
                 topical_bases, include_lbsn_bases)
             base_records = self.make_bases(
@@ -153,6 +155,15 @@ def extract_hll_bases(
                 record=record)
             if base_records:
                 base_list.extend(base_records)
+            social_bases = ['community']
+            social_bases = self.filter_bases(
+                social_bases, include_lbsn_bases)
+            base_records = self.make_bases(
+                facet='social',
+                bases=social_bases,
+                record=record)
+            if base_records:
+                base_list.extend(base_records)
 
         # Places
         if record_type == lbsn.Place.DESCRIPTOR.name:
@@ -238,9 +249,7 @@ def get_hll_metrics(cls, record) -> hll.HllMetrics:
     @staticmethod
     def get_origin_metrics(record) -> hll.HllMetrics:
         """Get hll metrics from lbsn.Origin record"""
-        post_hll = HLF.hll_concat_origin_guid(record)
-        hll_metrics = hll.HllMetrics(post_hll=post_hll)
-        return hll_metrics
+        return
 
     @staticmethod
     def get_country_metrics(record) -> hll.HllMetrics:

lbsntransform/output/submit_data.py

@@ -84,7 +84,9 @@ def __init__(self, db_cursor=None,
             topical.HashtagBase.NAME: dict(),
             topical.EmojiBase.NAME: dict(),
             topical.TermLatLngBase.NAME: dict(),
+            topical.HashtagLatLngBase.NAME: dict(),
             topical.EmojiLatLngBase.NAME: dict(),
+            social.CommunityBase.NAME: dict()
         }
         self.count_round = 0
         if dbformat_output == "lbsn":

lbsntransform/tools/helper_functions.py

@@ -99,7 +99,7 @@ def sanitize_string(str_text: str):
     def format_base_repr(base):
         """Return formatted string of base"""
         return (
-            f'{base.__name__}\nFacet: {base.facet}, '
+            f'{base.NAME.base}\nFacet: {base.NAME.facet}, '
             f'Key: {base.get_key_value()}, '
             f'Metrics: \n'
             f'{[":".join([k, str(len(v))]) for k, v in base.metrics.items()]}')