jamofetch 3.7.7__tar.gz → 3.7.8__tar.gz

{jamofetch-3.7.7 → jamofetch-3.7.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: jamofetch
-Version: 3.7.7
+Version: 3.7.8
 Summary: A thin wrapper to retrieve sequence from JAMO at NERSC and on Dori.
 Author: Duncan Scott
 Requires-Python: >=3.9
@@ -89,15 +89,15 @@ options:
 (venv) [dnscott@ln005 jamofetch]$ jamofetch -d data -l NPUNN -l NOOHG -l HOGH -w --max -1
 fetching sequence:
 
-apptainer --silent run docker://doejgi/jamo-dori jamo link -s dori library NPUNN
+apptainer --silent exec docker://doejgi/jamo-dori jamo link -s dori custom '{"metadata.library_name":"NPUNN","metadata.fastq_type":"sdm_normal","group":"sdm"}'
 NPUNN /global/dna/dm_archive/sdm/pacbio/00/27/47/pbio-2747.27352.bc1001_BAK8A_OA--bc1001_BAK8A_OA.ccs.fastq.gz BACKUP_COMPLETE 6391936239a7711d789a9380
 NPUNN /clusterfs/jgi/groups/dsi/homes/dnscott/git/jamofetch/data/NPUNN.pbio-2747.27352.bc1001_BAK8A_OA--bc1001_BAK8A_OA.ccs.fastq.gz
 
-apptainer --silent run docker://doejgi/jamo-dori jamo link -s dori library NOOHG
+apptainer --silent exec docker://doejgi/jamo-dori jamo link -s dori custom '{"metadata.library_name":"NOOHG","metadata.fastq_type":"sdm_normal","group":"sdm"}'
 NOOHG /global/dna/dm_archive/sdm/pacbio/00/26/91/pbio-2691.26653.bc1001_BAK8A_OA--bc1001_BAK8A_OA.ccs.fastq.gz RESTORED 6347dbb35bc59487d7e768d6
 NOOHG /clusterfs/jgi/groups/dsi/homes/dnscott/git/jamofetch/data/NOOHG.pbio-2691.26653.bc1001_BAK8A_OA--bc1001_BAK8A_OA.ccs.fastq.gz
 
-apptainer --silent run docker://doejgi/jamo-dori jamo link -s dori library HOGH
+apptainer --silent exec docker://doejgi/jamo-dori jamo link -s dori custom '{"metadata.library_name":"HOGH","metadata.fastq_type":"sdm_normal","group":"sdm"}'
 HOGH /global/dna/dm_archive/sdm/illumina/00/63/97/6397.2.44053.GGCTAC.fastq.gz RESTORED 51d52a82067c014cd6ef4f6f
 HOGH /clusterfs/jgi/groups/dsi/homes/dnscott/git/jamofetch/data/HOGH.6397.2.44053.GGCTAC.fastq.gz
 

{jamofetch-3.7.7 → jamofetch-3.7.8}/README.md

@@ -78,15 +78,15 @@ options:
 (venv) [dnscott@ln005 jamofetch]$ jamofetch -d data -l NPUNN -l NOOHG -l HOGH -w --max -1
 fetching sequence:
 
-apptainer --silent run docker://doejgi/jamo-dori jamo link -s dori library NPUNN
+apptainer --silent exec docker://doejgi/jamo-dori jamo link -s dori custom '{"metadata.library_name":"NPUNN","metadata.fastq_type":"sdm_normal","group":"sdm"}'
 NPUNN /global/dna/dm_archive/sdm/pacbio/00/27/47/pbio-2747.27352.bc1001_BAK8A_OA--bc1001_BAK8A_OA.ccs.fastq.gz BACKUP_COMPLETE 6391936239a7711d789a9380
 NPUNN /clusterfs/jgi/groups/dsi/homes/dnscott/git/jamofetch/data/NPUNN.pbio-2747.27352.bc1001_BAK8A_OA--bc1001_BAK8A_OA.ccs.fastq.gz
 
-apptainer --silent run docker://doejgi/jamo-dori jamo link -s dori library NOOHG
+apptainer --silent exec docker://doejgi/jamo-dori jamo link -s dori custom '{"metadata.library_name":"NOOHG","metadata.fastq_type":"sdm_normal","group":"sdm"}'
 NOOHG /global/dna/dm_archive/sdm/pacbio/00/26/91/pbio-2691.26653.bc1001_BAK8A_OA--bc1001_BAK8A_OA.ccs.fastq.gz RESTORED 6347dbb35bc59487d7e768d6
 NOOHG /clusterfs/jgi/groups/dsi/homes/dnscott/git/jamofetch/data/NOOHG.pbio-2691.26653.bc1001_BAK8A_OA--bc1001_BAK8A_OA.ccs.fastq.gz
 
-apptainer --silent run docker://doejgi/jamo-dori jamo link -s dori library HOGH
+apptainer --silent exec docker://doejgi/jamo-dori jamo link -s dori custom '{"metadata.library_name":"HOGH","metadata.fastq_type":"sdm_normal","group":"sdm"}'
 HOGH /global/dna/dm_archive/sdm/illumina/00/63/97/6397.2.44053.GGCTAC.fastq.gz RESTORED 51d52a82067c014cd6ef4f6f
 HOGH /clusterfs/jgi/groups/dsi/homes/dnscott/git/jamofetch/data/HOGH.6397.2.44053.GGCTAC.fastq.gz
 

{jamofetch-3.7.7 → jamofetch-3.7.8}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "flit_core.buildapi"
 
 [project]
 name = "jamofetch"
-version = "3.7.7"
+version = "3.7.8"
 description = "A thin wrapper to retrieve sequence from JAMO at NERSC and on Dori."
 authors = [{ name = "Duncan Scott" }]
 readme = "README.md"

{jamofetch-3.7.7 → jamofetch-3.7.8}/src/jamofetch/jamofetch.py

@@ -1,5 +1,6 @@
 #!/usr/bin/env python3
 
+import json
 import logging
 import os
 import pathlib
@@ -10,17 +11,43 @@ import time
 from argparse import ArgumentParser
 from stat import ST_CTIME
 
-JAMO_CMD_NERSC = 'module load jamo; jamo link library'
-JAMO_CMD_DORI = 'apptainer --silent run docker://doejgi/jamo-dori jamo link -s dori library'
+# Base commands; get_cmd() appends the query.
+#
+# Dori uses `apptainer exec`, not `run`: the jamo-dori image's runscript strips the
+# double quotes out of the JSON query, and jamo then fails with a JSONDecodeError.
+JAMO_CMD_NERSC = 'module load jamo; jamo link'
+JAMO_CMD_DORI = 'apptainer --silent exec docker://doejgi/jamo-dori jamo link -s dori'
 
 TWO_HOURS = 7200  # seconds
 ONE_MINUTE = 60  # seconds
 
 
 def get_cmd(clean_lib_name) -> str:
-    if os.getenv('SLURM_PARTITION') == 'dori':
-        return f"{JAMO_CMD_DORI} {clean_lib_name}"
-    return f"{JAMO_CMD_NERSC} {clean_lib_name}"
+    """
+    Build the site-appropriate `jamo link` command for a library's SDM fastq.
+
+    Uses a custom query rather than `jamo link library <name>`. That form applies
+    jamo's default `raw_normal` filter, {'metadata.fastq_type': 'sdm_normal',
+    'user': 'sdm'}, and files from SDM's newer pipeline are owned by user
+    'sdm_pipeline' (group still 'sdm'), so it matched nothing for them. Filtering
+    on group instead of user matches both.
+
+    Key order matters: jamo names each symlink '<first string-valued query key's
+    value>.<file_name>', so metadata.library_name must stay first or
+    _find_fastq_path() will not find the link.
+
+    A library sequenced on more than one run matches one fastq per run; nothing
+    here chooses between them.
+    """
+    # Validated because the name is interpolated into a shell command.
+    lib_name = _clean_library_name(clean_lib_name)
+    query = json.dumps({
+        "metadata.library_name": lib_name,
+        "metadata.fastq_type": "sdm_normal",
+        "group": "sdm",
+    }, separators=(',', ':'))
+    base = JAMO_CMD_DORI if os.getenv('SLURM_PARTITION') == 'dori' else JAMO_CMD_NERSC
+    return f"{base} custom '{query}'"
 
 
 def _clean_library_name(library_name):

jamofetch-3.7.8/tests/test_jamofetch.py

@@ -0,0 +1,47 @@
+import json
+import shlex
+
+import pytest
+
+from jamofetch import jamofetch
+
+
+def _query(cmd):
+    # The query is the single-quoted final argument; shlex undoes the shell quoting.
+    return json.loads(shlex.split(cmd)[-1])
+
+
+def test_nersc_cmd(monkeypatch):
+    monkeypatch.delenv('SLURM_PARTITION', raising=False)
+    cmd = jamofetch.get_cmd('LBBDHFZ')
+    assert cmd.startswith('module load jamo; jamo link custom ')
+    assert _query(cmd) == {
+        "metadata.library_name": "LBBDHFZ",
+        "metadata.fastq_type": "sdm_normal",
+        "group": "sdm",
+    }
+
+
+def test_dori_cmd_uses_exec(monkeypatch):
+    monkeypatch.setenv('SLURM_PARTITION', 'dori')
+    cmd = jamofetch.get_cmd('LBBDHFZ')
+    assert cmd.startswith('apptainer --silent exec docker://doejgi/jamo-dori jamo link -s dori custom ')
+    assert _query(cmd)["metadata.library_name"] == "LBBDHFZ"
+
+
+def test_library_name_is_first_query_key(monkeypatch):
+    # jamo names the symlink after the first string-valued key; _find_fastq_path
+    # relies on that being the library name.
+    monkeypatch.delenv('SLURM_PARTITION', raising=False)
+    assert next(iter(_query(jamofetch.get_cmd('LBBDHFZ')))) == "metadata.library_name"
+
+
+def test_query_does_not_filter_on_user(monkeypatch):
+    monkeypatch.delenv('SLURM_PARTITION', raising=False)
+    assert "user" not in _query(jamofetch.get_cmd('LBBDHFZ'))
+
+
+@pytest.mark.parametrize('bad', ["LBB'DHFZ", 'lbbdhfz', 'LBB DHFZ', 'LBB;rm'])
+def test_invalid_library_name_rejected(bad):
+    with pytest.raises(ValueError):
+        jamofetch.get_cmd(bad)

jamofetch-3.7.7/tests/test_jamofetch.py

@@ -1,2 +0,0 @@
-from jamofetch import jamofetch
-