geoseeq 0.5.6a9__tar.gz → 0.5.6a10__tar.gz

{geoseeq-0.5.6a9 → geoseeq-0.5.6a10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: geoseeq
-Version: 0.5.6a9
+Version: 0.5.6a10
 Summary: GeoSeeq command line tools and python API
 Author: David C. Danko
 Author-email: "David C. Danko" <dcdanko@biotia.io>

{geoseeq-0.5.6a9 → geoseeq-0.5.6a10}/geoseeq/cli/download.py

@@ -97,13 +97,14 @@ def cli_download_metadata(state, sample_ids):
 
 
 cores_option = click.option('--cores', default=1, help='Number of downloads to run in parallel')
-
+head_option = click.option('--head', default=None, type=int, help='Download the first N bytes of each file')
 
 @cli_download.command("files")
 @use_common_state
 @cores_option
 @click.option("--target-dir", default=".")
 @yes_option
+@head_option
 @click.option("--download/--urls-only", default=True, help="Download files or just print urls")
 @click.option("--folder-type", type=click.Choice(['all', 'sample', 'project'], case_sensitive=False), default="all", help='Download files from sample folders, project folders, or both')
 @click.option("--folder-name", multiple=True, help='Filter folders for names that include this string. Case insensitive.')
@@ -120,6 +121,7 @@ def cli_download_files(
     sample_name_includes,
     target_dir,
     yes,
+    head,
     folder_type,
     folder_name,
     file_name,
@@ -213,6 +215,7 @@ def cli_download_files(
             ignore_errors=ignore_errors,
             log_level=state.log_level,
             progress_tracker_factory=PBarManager().get_new_bar,
+            head=head,
         )
         for fname, url in response["links"].items():
             download_manager.add_download(url, join(target_dir, fname))
@@ -230,11 +233,12 @@ def cli_download_files(
 @cores_option
 @click.option("-t", "--target-dir", default=".")
 @yes_option
+@head_option
 @click.option("--download/--urls-only", default=True, help="Download files or just print urls")
 @ignore_errors_option
 @click.option('--hidden/--no-hidden', default=True, help='Download hidden files in folder')
 @folder_ids_arg
-def cli_download_folders(state, cores, target_dir, yes, download, ignore_errors, hidden, folder_ids):
+def cli_download_folders(state, cores, target_dir, yes, head, download, ignore_errors, hidden, folder_ids):
     """Download entire folders from GeoSeeq.
     
     This command downloads folders directly based on their ID. This is used for "manual"
@@ -267,6 +271,7 @@ def cli_download_folders(state, cores, target_dir, yes, download, ignore_errors,
         ignore_errors=ignore_errors,
         log_level=state.log_level,
         progress_tracker_factory=PBarManager().get_new_bar,
+        head=head,
     )
     for result_folder in result_folders:
         download_manager.add_result_folder_download(
@@ -286,7 +291,7 @@ def cli_download_folders(state, cores, target_dir, yes, download, ignore_errors,
 @click.option("-n", "--file-name", multiple=True, help="File name to use for downloaded files. If set you must specify once per ID.")
 @yes_option
 @click.option("--download/--urls-only", default=True, help="Download files or just print urls")
-@click.option('--head', default=None, type=int, help='Download the first N bytes of each file')
+@head_option
 @ignore_errors_option
 @click.argument("ids", nargs=-1)
 def cli_download_ids(state, cores, target_dir, file_name, yes, download, head, ignore_errors, ids):

{geoseeq-0.5.6a9 → geoseeq-0.5.6a10}/geoseeq/cli/main.py

@@ -53,7 +53,7 @@ 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('0.5.6a9')  # remember to update setup
+    click.echo('0.5.6a10')  # remember to update setup
 
 
 @main.group('advanced')

{geoseeq-0.5.6a9 → geoseeq-0.5.6a10}/geoseeq/result/file_upload.py

@@ -2,7 +2,7 @@
 import time
 import json
 import os
-from os.path import basename, getsize, join, dirname, isfile
+from os.path import basename, getsize, join, dirname, isfile, getctime
 from pathlib import Path
 
 import requests
@@ -49,7 +49,7 @@ class ResumableUploadTracker:
         self.filepath = filepath
         self.tracker_file = join(
             GEOSEEQ_CACHE_DIR, 'upload',
-            tracker_file_prefix + f".{chunk_size}." + basename(filepath)
+            tracker_file_prefix + f".{chunk_size}.{getsize(filepath)}." + basename(filepath)
         )
         try:
             os.makedirs(dirname(self.tracker_file), exist_ok=True)
@@ -64,7 +64,7 @@ class ResumableUploadTracker:
             return
         if self.upload_started:
             raise GeoseeqGeneralError("Upload has already started.")
-        blob = dict(upload_id=upload_id, urls=urls)
+        blob = dict(upload_id=upload_id, urls=urls, start_time=time.time())
         serialized = json.dumps(blob)
         with open(self.tracker_file, "w") as f:
             f.write(serialized + "\n")
@@ -89,6 +89,11 @@ class ResumableUploadTracker:
         with open(self.tracker_file, "r") as f:
             header_blob = json.loads(f.readline())
             self.upload_id, self.urls = header_blob["upload_id"], header_blob["urls"]
+            start_time = header_blob["start_time"]
+            if (time.time() - start_time) > (60 * 60 * 23):
+                logger.warning(f"Tracker file {self.tracker_file} is too old. Deleting.")
+                os.remove(self.tracker_file)
+                return
             self.upload_started = True
             for line in f:
                 blob = json.loads(line)
@@ -161,14 +166,15 @@ class ResultFileUpload:
                 http_response.raise_for_status()
                 logger.debug(f"Upload for part {num + 1} succeeded.")
                 break
-            except requests.exceptions.HTTPError:
-                logger.warn(
+            except (requests.exceptions.HTTPError, requests.exceptions.SSLError, requests.exceptions.ConnectionError) as e:
+                logger.debug(
                     f"Upload for part {num + 1} failed. Attempt {attempts + 1} of {max_retries}."
                 )
                 attempts += 1
                 if attempts == max_retries:
-                    raise
+                    raise e
                 time.sleep(10**attempts)  # exponential backoff, (10 ** 2)s default max
+            
         etag = http_response.headers["ETag"].replace('"', "")
         blob = {"ETag": etag, "PartNumber": num + 1}
         if resumable_upload_tracker:

{geoseeq-0.5.6a9 → geoseeq-0.5.6a10}/geoseeq.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: geoseeq
-Version: 0.5.6a9
+Version: 0.5.6a10
 Summary: GeoSeeq command line tools and python API
 Author: David C. Danko
 Author-email: "David C. Danko" <dcdanko@biotia.io>

{geoseeq-0.5.6a9 → geoseeq-0.5.6a10}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "geoseeq"
-version = "0.5.6a9"
+version = "0.5.6a10"
 authors = [
   { name="David C. Danko", email="dcdanko@biotia.io" },
 ]