Merge pull request #71 from robertatakenaka/melhora_get_doc_records

Cria id2json3.get_doc_records

scielo_classic_website/iid2json/id2json3.py

@@ -60,6 +60,25 @@ def pids_and_their_records(id_file_path, db_type):
     return _get_id_and_json_records(records, id_function)
 
 
+def get_doc_records(id_file_path):
+    issues = {}
+    for item_id, records in pids_and_their_records(id_file_path, "artigo"):
+        record_type = None
+        if records:
+            record_type = _get_value(records[0], "v706")
+
+        if record_type == "i":
+            issues[item_id] = records[0]
+        elif record_type == "o":
+            if len(item_id) == 23:
+                i_id = item_id[1:18]
+                yield {"issue_id": i_id, "doc_id": item_id, "issue": issues.get(i_id), "article": records}
+            else:
+                yield {"invalid_records": True, "item_id": item_id, "records": records}
+        else:
+            yield {"invalid_records": True, "item_id": item_id, "records": records}
+
+
 def _get_value(data, tag):
     """
     Returns first value of field `tag`