Fix Cypher query returns duplicate authors

AtomikLabs · May 24, 2024 · dbe99a9 · dbe99a9
1 parent ad56f44
commit dbe99a9
Show file tree

Hide file tree

Showing 2 changed files with 29 additions and 7 deletions.
diff --git a/infra/core/dev.tfvars b/infra/core/dev.tfvars
@@ -3,7 +3,7 @@
 # **********************************************************
 
 app_name = "atomiklabs"
-app_version = "0.3.6-alpha"
+app_version = "0.3.7-alpha"
 availability_zones = ["us-east-1a", "us-east-1b", "us-east-1c"]
 aws_region = "us-east-1"
 backend_dynamodb_table = "terraform-state-locks"
@@ -63,9 +63,9 @@ arxiv_sets = ["cs"]
 default_lambda_runtime = "python3.10"
 pods_prefix = "pods"
 
-create_pod_task_version = "0.1.0"
+create_pod_task_version = "0.1.1"
 fetch_from_arxiv_task_version = "0.1.0"
 most_recent_research_records_version = "0.0.2"
 parse_summaries_task_version = "0.1.0"
-persist_summaries_task_version = "0.1.0"
+persist_summaries_task_version = "0.1.1"
 save_summaries_to_datalake_task_version = "0.0.1"
diff --git a/...rflow/dags/publishing/tasks/create_pod.py → .../dags/publishing/tasks/create_pod_task.py b/...rflow/dags/publishing/tasks/create_pod.py → .../dags/publishing/tasks/create_pod_task.py
@@ -30,6 +30,7 @@
     PUBLISHES,
     RECORDS_PREFIX,
     RETRIEVAL_ERRORS,
+    SUMMARIZES,
 )
 from shared.utils.utils import get_config
 
@@ -73,10 +74,31 @@ def run(
                     logger.info("No summaries for date", set=arxiv_set, category=category, date=pod_date)
                     continue
                 pod_summaries = get_pod_summaries(context, config, summaries)
-                scripts = write_pod_script(config, pod_summaries, arxiv_set, category, pod_date)
+                scripts = write_pod_script(
+                    config=config,
+                    pod_summaries=pod_summaries,
+                    arxiv_set=arxiv_set,
+                    category=category,
+                    episode_date=pod_date,
+                )
                 for key, script, part_record_ids in scripts:
-                    audio_key = create_audio(config, arxiv_set, category, pod_date, script, key)
-                    create_pod_node(config, arxiv_set, category, pod_date, key, audio_key, part_record_ids)
+                    audio_key = create_audio(
+                        config=config,
+                        arxiv_set=arxiv_set,
+                        category=category,
+                        episode_date=pod_date,
+                        script_text=script,
+                        key=key,
+                    )
+                    create_pod_node(
+                        config=config,
+                        arxiv_set=arxiv_set,
+                        category=category,
+                        episode_date=pod_date,
+                        key=key,
+                        audio_key=audio_key,
+                        part_record_ids=part_record_ids,
+                    )
             except Exception as e:
                 logger.error("Error creating pod", set=arxiv_set, category=category, date=pod_date, error=e)
                 continue
@@ -123,7 +145,7 @@ def get_summaries(config: dict, arxiv_set: str, category: str, episode_date: dat
             query = (
                 f"MATCH (s:ArxivSet {{code: $arxiv_set}}) "
                 f"-[:{CATEGORIZED_BY}]->(c:ArxivCategory {{code: $category}}) "
-                f"<-[:{PRIMARILY_CATEGORIZED_BY}]-(a:ArxivRecord)--(b:Abstract) "
+                f"<-[:{PRIMARILY_CATEGORIZED_BY}]-(a:ArxivRecord)<-[:{SUMMARIZES}]-(b:Abstract) "
                 "MATCH (a)-[:AUTHORED_BY]->(author:Author)"
                 "WHERE a.date = $date "
                 "RETURN {record: a, abstract: b, authors: collect({first_name: author.first_name, last_name: author.last_name})} AS result"