From 369aff8fabdc0168d34c2653b4742893abad79f7 Mon Sep 17 00:00:00 2001 From: MertOkur Date: Sat, 11 Apr 2026 12:10:20 +0200 Subject: [PATCH] Adding update option for PubMed --- src/pybool_ir/cli/__main__.py | 28 +++++++++++++++++++ .../datasets/pubmed/update_pubmed.py | 0 2 files changed, 28 insertions(+) create mode 100644 src/pybool_ir/datasets/pubmed/update_pubmed.py diff --git a/src/pybool_ir/cli/__main__.py b/src/pybool_ir/cli/__main__.py index 24852b1..524cb8d 100644 --- a/src/pybool_ir/cli/__main__.py +++ b/src/pybool_ir/cli/__main__.py @@ -84,6 +84,34 @@ def pubmed_download(baseline_path: Path, limit: int): print(f"Download full baseline to {baseline_path}...") download_baseline(Path(baseline_path)) +@pubmed.command("update") +@click.option( + "-u", + "--update-path", + "update_path", + type=click.Path(), + multiple=False, + required=True, + help="Location to download PubMed update files" +) + +@click.option( + "-l", + "--limit", + "limit", + type=int, + default=0, + help="Number of update documents that should be downloaded" +) + +def pubmed_update(update_path: Path, limit: int): + from pybool_ir.datasets.pubmed.baseline import update_baseline + + print(f"Start fetching PubMed updates to {update_path}...") + + update_baseline(Path(update_path), limit=limit if limit > 0 else None) + print("Updates downloaded successfully!") + @pmc.command("download") @click.option( diff --git a/src/pybool_ir/datasets/pubmed/update_pubmed.py b/src/pybool_ir/datasets/pubmed/update_pubmed.py new file mode 100644 index 0000000..e69de29