geoseeq 0.6.14.dev3__py3-none-any.whl → 0.6.14.dev5__py3-none-any.whl

geoseeq/cli/download.py

@@ -379,22 +379,26 @@ def cli_download_ids(state, cores, target_dir, file_name, yes, download, head, i
 
 
 def _get_sample_result_files_with_names(sample, module_name=None, first=False):
+    result_files_with_names = []
     for read_type, folder in sample.get_all_fastqs().items():
         if module_name and module_name != read_type:
             continue
-        result_files_with_names = []
         for folder_name, result_files in folder.items():
-            for result_file in result_files:
+            for lane_num, result_file in enumerate(result_files):
+                lane_num = lane_num + 1  # 1 indexed
                 if read_type in ["short_read::paired_end"]:
+                    key = (sample, read_type, 1, lane_num)  # sample name, read type, read number, lane number
                     result_files_with_names.append(
-                        (result_file[0], result_file[0].get_referenced_filename())
+                        (result_file[0], result_file[0].get_referenced_filename(), key)
                     )
+                    key = (sample, read_type, 2, lane_num)
                     result_files_with_names.append(
-                        (result_file[1], result_file[1].get_referenced_filename())
+                        (result_file[1], result_file[1].get_referenced_filename(), key)
                     )
                 else:
+                    key = (sample, read_type, 1, lane_num)
                     result_files_with_names.append(
-                        (result_file, result_file.get_referenced_filename())
+                        (result_file, result_file.get_referenced_filename(), key)
                     )
             if first:
                 break
@@ -402,6 +406,43 @@ def _get_sample_result_files_with_names(sample, module_name=None, first=False):
     return result_files_with_names
 
 
+def _make_read_configs(download_results, config_dir="."):
+    """Make JSON config files that look like this.
+    
+    {
+        "sample_name": "small",
+        "reads_1": ["small.fq.gz"],
+        "reads_2": [],
+        "fastq_checksum": "",
+        "data_type": "short-read",
+        "bdx_result_dir": "results",
+        "geoseeq_uuid": "05bf22e9-9d25-42db-af25-31bc538a7006"
+    }
+    """
+    config_blobs = {}  # sample ids -> config_blobs
+    download_results = sorted(download_results, key=lambda x: x[1][3])  # sort by lane number
+    for local_path, (sample, read_type, read_num, lane_num), _ in download_results:
+        if sample.name not in config_blobs:
+            config_blobs[sample.name] = {
+                "sample_name": sample.name,
+                "reads_1": [],
+                "reads_2": [],
+                "fastq_checksum": "",
+                "data_type": "short-read",
+                "bdx_result_dir": "results",
+                "geoseeq_uuid": sample.uuid,
+            }
+        if read_num == 1:
+            config_blobs[sample.name]["reads_1"].append(local_path)  # sorted by lane number
+        else:
+            config_blobs[sample.name]["reads_2"].append(local_path)
+
+    for sample_name, config_blob in config_blobs.items():
+        config_path = join(config_dir, f"{sample_name}.config.json")
+        with open(config_path, "w") as f:
+            json.dump(config_blob, f, indent=4)
+
+
 @cli_download.command("fastqs")
 @use_common_state
 @cores_option
@@ -409,6 +450,7 @@ def _get_sample_result_files_with_names(sample, module_name=None, first=False):
 @yes_option
 @click.option("--first/--all", default=False, help="Download only the first folder of fastq files for each sample.")
 @click.option("--download/--urls-only", default=True, help="Download files or just print urls")
+@click.option("--config-dir", default=None, help="Directory to write read config files. If unset do not write config files.")
 @module_option(FASTQ_MODULE_NAMES, use_default=False)
 @ignore_errors_option
 @alt_id_option
@@ -420,6 +462,7 @@ def cli_download_fastqs(state,
                         yes,
                         first,
                         download,
+                        config_dir,
                         module_name,
                         ignore_errors,
                         alt_sample_id,
@@ -496,8 +539,8 @@ def cli_download_fastqs(state,
         log_level=state.log_level,
         progress_tracker_factory=PBarManager().get_new_bar,
     )
-    for result_file, filename in result_files_with_names:
-        download_manager.add_download(result_file, join(target_dir, filename))
+    for result_file, filename, key in result_files_with_names:
+        download_manager.add_download(result_file, join(target_dir, filename), key=key)
     if not download:
         print(download_manager.get_url_string(), file=state.outfile)
     else:
@@ -505,5 +548,6 @@ def cli_download_fastqs(state,
         if not yes:
             click.confirm('Continue?', abort=True)
         logger.info(f'Downloading {len(download_manager)} files to {target_dir}')
-        download_manager.download_files()
-
+        download_results = download_manager.download_files()
+        if config_dir:
+            _make_read_configs(download_results, config_dir)

geoseeq/cli/main.py

@@ -55,7 +55,7 @@ def version():
     Use of this tool implies acceptance of the GeoSeeq End User License Agreement.
     Run `geoseeq eula show` to view the EULA.
     """
-    click.echo('0.6.14dev3')  # remember to update pyproject.toml
+    click.echo('0.6.14dev5')  # remember to update pyproject.toml
 
 
 @main.group('advanced')

geoseeq/result/file_download.py

@@ -107,7 +107,9 @@ def download_url(url, kind='guess', filename=None, head=None, progress_tracker=N
     elif kind == 'ftp':
         return download_ftp(url, filename, head=head)
     elif kind == 'http':
-        # for http[s] files we care about head is often respected in practice (e.g. by the ENA) 
+        # for http[s] files we care about head is often respected in practice (e.g. by the ENA)
+        if not url.startswith("http"):
+            url = "https://" + url
         return _download_head(url, filename, head=head, progress_tracker=progress_tracker)
     else:
         raise ValueError(f"Unknown download kind: {kind}")
@@ -179,7 +181,7 @@ class ResultFileDownload:
 
         url = self.get_download_url()
         filepath = download_url(
-            url, blob_type, filename,
+            url, kind=blob_type, filename=filename,
             head=head, progress_tracker=progress_tracker,
         )
         if cache and flag_suffix:

{geoseeq-0.6.14.dev3.dist-info → geoseeq-0.6.14.dev5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: geoseeq
-Version: 0.6.14.dev3
+Version: 0.6.14.dev5
 Summary: GeoSeeq command line tools and python API
 Author: David C. Danko
 Author-email: "David C. Danko" <dcdanko@biotia.io>

{geoseeq-0.6.14.dev3.dist-info → geoseeq-0.6.14.dev5.dist-info}/RECORD

@@ -20,11 +20,11 @@ geoseeq/cli/__init__.py,sha256=4WnK87K5seRK3SGJAxNWnQTqyg5uBhdhrOrzB1D4b3M,24
 geoseeq/cli/constants.py,sha256=Do5AUf9lMO9_P8KpFJ3XwwFBAWsxSjZ6sx9_QEGyC_c,176
 geoseeq/cli/copy.py,sha256=02U9kdrAIbbM8MlRMLL6p-LMYFSuRObE3h5jyvcL__M,2275
 geoseeq/cli/detail.py,sha256=q8Suu-j2k18knfSVFG-SWWGNsKM-n8y9RMA3LcIIi9Y,4132
-geoseeq/cli/download.py,sha256=WBStk-K2pPzM2rqjIxwEhl1jAPn6UZwUDg7Dc3R2ahM,19295
+geoseeq/cli/download.py,sha256=cOLIVzAfxpOJnSd4WKWR-HCkOrKjPMbgYaNXDSd9i90,21305
 geoseeq/cli/fastq_utils.py,sha256=-bmeQLaiMBm57zWOF0R5OlWTU0_3sh1JBC1RYw2BOFM,3083
 geoseeq/cli/find_grn.py,sha256=oMDxkzGQBQb2_cCuvmwoeHOsFHqyO9RLeJzrB6bAe5M,439
 geoseeq/cli/get_eula.py,sha256=79mbUwyiF7O1r0g6UTxG9kJGQEqKuH805E6eLkPC6Y4,997
-geoseeq/cli/main.py,sha256=Ue0W3uOnGF2qhrNS-yWthKrGO_pbK9iyT0mrnbObJuA,3987
+geoseeq/cli/main.py,sha256=EL5VL2rB3e6TNi8ZaHH1adGV_9SOE-cUKty1i6hK1W8,3987
 geoseeq/cli/manage.py,sha256=wGXAcVaXqE5JQEU8Jh6OlHr02nB396bpS_SFcOZdrEo,5929
 geoseeq/cli/progress_bar.py,sha256=p1Xl01nkYxSBZCB30ue2verIIi22W93m3ZAMAxipD0g,738
 geoseeq/cli/project.py,sha256=V5SdXm2Hwo2lxrkpwRDedw-mAE4XnM2uwT-Gj1D90VQ,3030
@@ -68,7 +68,7 @@ geoseeq/plotting/map/overlay.py,sha256=4VmxqOESTQra9tPr8b8OLEUhJSit9lNipabeSznEY
 geoseeq/result/__init__.py,sha256=IFHIyRV8ZzuKIfwfze1SXgcKwNMcSgMAknLHMkwjXIU,356
 geoseeq/result/bioinfo.py,sha256=QQtbyogrdro9avJSN0713sxLVnVeA24mFw3hWtKDKyw,1782
 geoseeq/result/file_chunker.py,sha256=bXq1csuRtqMB5sbH-AfWo6gdPwrivv5DJPuHVj-h08w,1758
-geoseeq/result/file_download.py,sha256=F0qSlZyrzh7G2R-qxmGwRicTJiojBUbcBN_SWF1sZac,8123
+geoseeq/result/file_download.py,sha256=5IXg_dIWlrRHBJQssO42da5_bIJOyH0_b8K2KWVAFBE,8210
 geoseeq/result/file_upload.py,sha256=xs1DrI-h4ZP7xN8HPBc3SFpcPAxR5HAolraP1Zu7tvE,10648
 geoseeq/result/result_file.py,sha256=A2PIdkuNY0czXLXPenSRdUPdCFdjJGGqs3nBOflNwnA,9099
 geoseeq/result/result_folder.py,sha256=-m1lDVLpNHKy-JUGihboVzvdMJEnHossyRnxmBe1XLo,11140
@@ -87,9 +87,9 @@ geoseeq/vc/vc_stub.py,sha256=IQr8dI0zsWKVAeY_5ybDD6n49_3othcgfHS3P0O9tuY,3110
 tests/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 tests/test_api_client.py,sha256=TS5njc5pcPP_Ycy-ljcfPVT1hQRBsFVdQ0lCqBmoesU,12810
 tests/test_plotting.py,sha256=TcTu-2ARr8sxZJ7wPQxmbs3-gHw7uRvsgrhhhg0qKik,784
-geoseeq-0.6.14.dev3.dist-info/LICENSE,sha256=IuhIl1XCxXLPLJT_coN1CNqQU4Khlq7x4IdW7ioOJD8,1067
-geoseeq-0.6.14.dev3.dist-info/METADATA,sha256=vgvY3SCXNFYwWspRIDAt-W9I2bk3q_dSeWSHJEFGVOA,4937
-geoseeq-0.6.14.dev3.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-geoseeq-0.6.14.dev3.dist-info/entry_points.txt,sha256=yF-6KDM8zXib4Al0qn49TX-qM7PUkWUIcYtsgt36rjM,45
-geoseeq-0.6.14.dev3.dist-info/top_level.txt,sha256=zZk7mmeaqAYqFJG8nq2DTgSQPbflRjJwkDIhNURPDEU,14
-geoseeq-0.6.14.dev3.dist-info/RECORD,,
+geoseeq-0.6.14.dev5.dist-info/LICENSE,sha256=IuhIl1XCxXLPLJT_coN1CNqQU4Khlq7x4IdW7ioOJD8,1067
+geoseeq-0.6.14.dev5.dist-info/METADATA,sha256=SCO_zG5oJC5KNM_nAOb1p0cQgSrSWWIBs5FXpxYDbpI,4937
+geoseeq-0.6.14.dev5.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+geoseeq-0.6.14.dev5.dist-info/entry_points.txt,sha256=yF-6KDM8zXib4Al0qn49TX-qM7PUkWUIcYtsgt36rjM,45
+geoseeq-0.6.14.dev5.dist-info/top_level.txt,sha256=zZk7mmeaqAYqFJG8nq2DTgSQPbflRjJwkDIhNURPDEU,14
+geoseeq-0.6.14.dev5.dist-info/RECORD,,