Import csv

backend/app/services/parse_csv.py

@@ -0,0 +1,18 @@
+from app.models import Document, Corpus
+
+def parse_csv(csv_reader, title="new_data"):
+    doc_list = []
+    for data in csv_reader:
+        new_doc = None
+        new_doc = Document.objects.create_document(title=data['title'], text=data['transcript'], author=data['speaker_1'], year=data['recorded_date'][:4])
+        # The line above creates a new Document instance (You should probably find some way of getting the user to assign columns to keys.For now, assume these keys match whatever dataset we're using.)
+        new_doc.save()
+        doc_list.append(new_doc)
+
+    new_corpus = Corpus(title=f"{title} Corpus")
+    new_corpus.save()
+
+    new_corpus.documents.set(doc_list)
+    new_corpus.save()
+
+    return new_corpus

backend/app/views.py

@@ -21,6 +21,8 @@
 }
 """
 import json
+import csv
+
 
 from rest_framework.decorators import api_view
 from rest_framework.response import Response
@@ -36,7 +38,7 @@
     GenderSerializer,
     CorpusSerializer
 )
-
+from app.services.parse_csv import parse_csv
 
 @api_view(['GET'])
 def get_example(request, example_id):
@@ -102,6 +104,7 @@ def add_document(request):
     """
     API endpoint for adding a piece of document
     """
+    print("we are in add_document")
     attributes = request.data
     new_attributes = {}
     for attribute in attributes['newAttributes']:
@@ -129,6 +132,17 @@ def all_documents(request):
     serializer = SimpleDocumentSerializer(doc_objs, many=True)
     return Response(serializer.data)
 
+@api_view(['POST'])
+def upload_document(request):
+    """
+    API endpoint for uploading csv files that are to be converted to an instance of csv_reader
+    """
+    file = request.data["file"]
+    content = file.read().decode('utf-8').splitlines()
+    csv_reader = csv.DictReader(content)
+    new_corpus = parse_csv(csv_reader)
+    serializer = CorpusSerializer(new_corpus)
+    return Response(serializer.data)
 
 @api_view(['GET'])
 def get_document(request, doc_id):

backend/config/urls.py

@@ -35,6 +35,7 @@
     path('api/update_corpus_docs', views.update_corpus_docs),
     path('api/delete_corpus', views.delete_corpus),
     path('api/corpus/<int:corpus_id>', views.get_corpus),
+    path('api/upload_document', views.upload_document),
 
     # View paths
     path('', views.index, name='index'),
@@ -43,5 +44,5 @@
     path('documents', views.documents, name='documents'),
     path('document/<int:doc_id>', views.single_document, name='document'),
     path('corpora', views.corpora, name='corpora'),
-    path('corpus/<int:corpus_id>', views.corpus, name='corpus')
+    path('corpus/<int:corpus_id>', views.corpus, name='corpus'),
 ]