add analyzer

supercoderhawk · Jan 9, 2020 · 2dfc35b · 2dfc35b
1 parent 5c6905d
commit 2dfc35b
Show file tree

Hide file tree

Showing 3 changed files with 29 additions and 0 deletions.
diff --git a/wsdm_digg/analyzer/__init__.py b/wsdm_digg/analyzer/__init__.py
@@ -0,0 +1 @@
+# -*- coding: UTF-8 -*-
diff --git a/wsdm_digg/analyzer/data.py b/wsdm_digg/analyzer/data.py
@@ -0,0 +1,24 @@
+# -*- coding: UTF-8 -*-
+from pysenal import read_jsonline_lazy
+from wsdm_digg.constants import *
+from wsdm_digg.elasticsearch.data import get_paper
+
+
+class DataViewer(object):
+    def __init__(self):
+        pass
+
+    def view(self):
+        for idx, item in enumerate(read_jsonline_lazy(DATA_DIR + 'test.jsonl')):
+            if idx > 20:
+                break
+            paper = get_paper(item['paper_id'])
+            print('============')
+            print(item['description_text'])
+            print('-------------')
+            print(paper['title'])
+            print(paper['abstract'])
+
+
+if __name__ == '__main__':
+    DataViewer().view()
diff --git a/wsdm_digg/analyzer/search_result.py b/wsdm_digg/analyzer/search_result.py
@@ -0,0 +1,4 @@
+# -*- coding: UTF-8 -*-
+class SearchResultAnalyzer(object):
+    def __init__(self):
+        pass