geoseeq 0.5.6a8__tar.gz → 0.5.6a10__tar.gz

{geoseeq-0.5.6a8 → geoseeq-0.5.6a10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: geoseeq
-Version: 0.5.6a8
+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-0.5.6a8 → 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-0.5.6a8 → 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('0.5.6a8')  # remember to update setup
+    click.echo('0.5.6a10')  # remember to update setup
 
 
 @main.group('advanced')

{geoseeq-0.5.6a8 → geoseeq-0.5.6a10}/geoseeq/cli/shared_params/opts_and_args.py

@@ -2,7 +2,7 @@ import click
 
 dryrun_option = click.option('--dryrun/--wetrun', default=False, help='Print what will be created without actually creating it')
 overwrite_option = click.option('--overwrite/--no-overwrite', default=False, help='Overwrite existing samples, files, and data')
-
+no_new_versions_option = click.option('--no-new-versions/--new-versions', default=False, help='Do not create new versions of the data')
 def module_option(options, use_default=True, default=None):
     if use_default:
         default = default or options[0]

{geoseeq-0.5.6a8 → geoseeq-0.5.6a10}/geoseeq/cli/upload/upload.py

@@ -24,6 +24,7 @@ from geoseeq.cli.shared_params import (
     handle_project_id,
     project_or_sample_id_arg,
     handle_project_or_sample_id,
+    no_new_versions_option,
 )
 from geoseeq.upload_download_manager import GeoSeeqUploadManager
 
@@ -41,11 +42,12 @@ hidden_option = click.option('--hidden/--no-hidden', default=False, help='Upload
 @link_option
 @recursive_option
 @hidden_option
+@no_new_versions_option
 @click.option('-n', '--geoseeq-file-name', default=None, multiple=True,
               help='Specify a different name for the file on GeoSeeq than the local file name.')
 @folder_id_arg
 @click.argument('file_paths', type=click.Path(exists=True), nargs=-1)
-def cli_upload_file(state, cores, yes, private, link_type, recursive, hidden, geoseeq_file_name, folder_id, file_paths):
+def cli_upload_file(state, cores, yes, private, link_type, recursive, hidden, no_new_versions, geoseeq_file_name, folder_id, file_paths):
     """Upload files to GeoSeeq.
 
     This command uploads files to either a sample or project on GeoSeeq. It can be used to upload
@@ -107,7 +109,7 @@ def cli_upload_file(state, cores, yes, private, link_type, recursive, hidden, ge
         link_type=link_type,
         progress_tracker_factory=PBarManager().get_new_bar,
         log_level=state.log_level,
-        overwrite=True,
+        no_new_versions=no_new_versions,
         use_cache=state.use_cache,
     )
     for geoseeq_file_name, file_path in name_pairs:
@@ -131,9 +133,10 @@ def cli_upload_file(state, cores, yes, private, link_type, recursive, hidden, ge
 @private_option
 @recursive_option
 @hidden_option
+@no_new_versions_option
 @project_or_sample_id_arg
 @click.argument('folder_names', type=click.Path(exists=True), nargs=-1)
-def cli_upload_folder(state, cores, yes, private, recursive, hidden, project_or_sample_id, folder_names):
+def cli_upload_folder(state, cores, yes, private, recursive, hidden, no_new_versions, project_or_sample_id, folder_names):
     knex = state.get_knex()
     root_obj = handle_project_or_sample_id(knex, project_or_sample_id, yes=yes, private=private)
     upload_manager = GeoSeeqUploadManager(
@@ -143,6 +146,7 @@ def cli_upload_folder(state, cores, yes, private, recursive, hidden, project_or_
         log_level=logging.INFO,
         overwrite=True,
         use_cache=state.use_cache,
+        no_new_versions=no_new_versions,
     )
     for folder_name in folder_names:
         result_folder = root_obj.result_folder(folder_name).idem()

{geoseeq-0.5.6a8 → geoseeq-0.5.6a10}/geoseeq/cli/upload/upload_reads.py

@@ -15,6 +15,7 @@ from geoseeq.cli.shared_params import (
     overwrite_option,
     yes_option,
     use_common_state,
+    no_new_versions_option
 )
 from geoseeq.upload_download_manager import GeoSeeqUploadManager
 
@@ -85,7 +86,7 @@ def _group_files(knex, filepaths, module_name, regex, yes):
     return groups
 
 
-def _do_upload(groups, module_name, link_type, lib, filepaths, overwrite, cores, state):
+def _do_upload(groups, module_name, link_type, lib, filepaths, overwrite, no_new_versions, cores, state):
 
     with requests.Session() as session:
         upload_manager = GeoSeeqUploadManager(
@@ -96,6 +97,7 @@ def _do_upload(groups, module_name, link_type, lib, filepaths, overwrite, cores,
             overwrite=overwrite,
             progress_tracker_factory=PBarManager().get_new_bar,
             use_cache=state.use_cache,
+            no_new_versions=no_new_versions,
         )
         for group in groups:
             sample = lib.sample(group['sample_name']).idem()
@@ -139,10 +141,11 @@ def flatten_list_of_fastqs(filepaths):
 @click.option('--regex', default=None, help='An optional regex to use to extract sample names from the file names')
 @private_option
 @link_option
+@no_new_versions_option
 @module_option(FASTQ_MODULE_NAMES)
 @project_id_arg
 @click.argument('fastq_files', type=click.Path(exists=True), nargs=-1)
-def cli_upload_reads_wizard(state, cores, overwrite, yes, regex, private, link_type, module_name, project_id, fastq_files):
+def cli_upload_reads_wizard(state, cores, overwrite, yes, regex, private, link_type, no_new_versions, module_name, project_id, fastq_files):
     """Upload fastq read files to GeoSeeq.
 
     This command automatically groups files by their sample name, lane number
@@ -196,4 +199,4 @@ def cli_upload_reads_wizard(state, cores, overwrite, yes, regex, private, link_t
     click.echo(f'Found {len(filepaths)} files to upload.', err=True)
     regex = _get_regex(knex, filepaths, module_name, proj, regex)
     groups = _group_files(knex, filepaths, module_name, regex, yes)
-    _do_upload(groups, module_name, link_type, proj, filepaths, overwrite, cores, state)
+    _do_upload(groups, module_name, link_type, proj, filepaths, overwrite, no_new_versions, cores, state)

{geoseeq-0.5.6a8 → 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:
@@ -260,10 +266,12 @@ class ResultFileUpload:
         logger.info(f'Finished Upload for "{filepath}"')
         return self
 
-    def upload_file(self, filepath, multipart_thresh=FIVE_MB, overwrite=True, **kwargs):
+    def upload_file(self, filepath, multipart_thresh=FIVE_MB, overwrite=True, no_new_versions=False, **kwargs):
         if self.exists() and not overwrite:  
             raise GeoseeqGeneralError(f"Overwrite is set to False and file {self.uuid} already exists.")
         self.idem()
+        if no_new_versions and self.has_downloadable_file():
+            raise GeoseeqGeneralError(f"File {self} already has a downloadable file. Not uploading a new version.")
         resolved_path = Path(filepath).resolve()
         file_size = getsize(resolved_path)
         return self.multipart_upload_file(filepath, file_size, **kwargs)

{geoseeq-0.5.6a8 → geoseeq-0.5.6a10}/geoseeq/result/result_file.py

@@ -53,6 +53,14 @@ class ResultFile(RemoteObject, ResultFileUpload, ResultFileDownload):
         obj_type = "sample" if self.canon_url() == "sample_ar_fields" else "project"
         brn = f"brn:{self.knex.instance_code()}:{obj_type}_result_field:{self.uuid}"
 
+    def has_downloadable_file(self):
+        """Return True if this field has a downloadable file."""
+        try:
+            self.download(head=10, cache=False)
+            return True
+        except Exception as e:
+            return False
+
     def nested_url(self):
         escaped_name = urllib.parse.quote(self.name, safe="")
         return self.parent.nested_url() + f"/fields/{escaped_name}"

{geoseeq-0.5.6a8 → geoseeq-0.5.6a10}/geoseeq/upload_download_manager.py

@@ -19,12 +19,18 @@ def _make_in_process_logger(log_level):
 
 
 def _upload_one_file(args):
-    result_file, filepath, session, progress_tracker, link_type, overwrite, log_level, parallel_uploads, use_cache = args
+    (result_file, filepath, session, progress_tracker,
+     link_type, overwrite, log_level, parallel_uploads,
+     use_cache, no_new_versions) = args
     if parallel_uploads:
         _make_in_process_logger(log_level)
     if link_type == 'upload':
         # TODO: check checksums to see if the file is the same
-        result_file.upload_file(filepath, session=session, overwrite=overwrite, progress_tracker=progress_tracker, threads=4, use_cache=use_cache)
+        result_file.upload_file(
+            filepath,
+            session=session, overwrite=overwrite, progress_tracker=progress_tracker,
+            threads=4, use_cache=use_cache, no_new_versions=no_new_versions
+        )
     else:
         result_file.link_file(link_type, filepath)
     return result_file
@@ -39,6 +45,7 @@ class GeoSeeqUploadManager:
                  progress_tracker_factory=None,
                  log_level=logging.WARNING,
                  overwrite=True,
+                 no_new_versions=False,
                  use_cache=True):
         self.session = session
         self.n_parallel_uploads = n_parallel_uploads
@@ -47,6 +54,7 @@ class GeoSeeqUploadManager:
         self.link_type = link_type
         self.overwrite = overwrite
         self._result_files = []
+        self.no_new_versions = no_new_versions
         self.use_cache = use_cache
 
     def add_result_file(self, result_file, local_path):
@@ -72,7 +80,7 @@ class GeoSeeqUploadManager:
                 result_file, local_path,
                 self.session, self.progress_tracker_factory(local_path),
                 self.link_type, self.overwrite, self.log_level,
-                self.n_parallel_uploads > 1, self.use_cache
+                self.n_parallel_uploads > 1, self.use_cache, self.no_new_versions
             ) for result_file, local_path in self._result_files
         ]
         out = []

{geoseeq-0.5.6a8 → geoseeq-0.5.6a10}/geoseeq.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: geoseeq
-Version: 0.5.6a8
+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-0.5.6a8 → geoseeq-0.5.6a10}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "geoseeq"
-version = "0.5.6a8"
+version = "0.5.6a10"
 authors = [
   { name="David C. Danko", email="dcdanko@biotia.io" },
 ]

{geoseeq-0.5.6a8 → geoseeq-0.5.6a10}/setup.py

@@ -5,7 +5,7 @@ import setuptools
 
 setuptools.setup(
     name='geoseeq',
-    version='0.5.6a7',  # remember to update version string in CLI as well
+    version='0.5.6a7',  # DEPRECATED see pyproject.toml remember to update version string in CLI as well
     author="David C. Danko",
     author_email='dcdanko@biotia.io',
     description=open('README.md').read(),