geoseeq 0.6.3__py3-none-any.whl → 0.6.5__py3-none-any.whl

geoseeq/cli/main.py

@@ -11,7 +11,7 @@ from .upload import cli_upload, cli_upload_advanced
 from .user import cli_user
 from .view import cli_view
 from .search import cli_search
-from geoseeq.vc.cli import cli_vc
+
 from geoseeq.knex import DEFAULT_ENDPOINT
 from geoseeq.utils import set_profile
 from .shared_params.opts_and_args import overwrite_option, yes_option
@@ -54,7 +54,7 @@ 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('0.6.3')  # remember to update setup
+    click.echo('0.6.5')  # remember to update setup
 
 
 @main.group('advanced')
@@ -73,7 +73,13 @@ def cli_experimental():
     """Experimental commands."""
     pass
 
-cli_experimental.add_command(cli_vc)
+
+
+try:
+    from geoseeq.vc.cli import cli_vc
+    cli_experimental.add_command(cli_vc)
+except (ModuleNotFoundError, ImportError):
+    pass
 
 @main.command('config')
 @yes_option

geoseeq/constants.py

@@ -2,7 +2,7 @@ from os import environ
 from os.path import join
 
 FIVE_MB = 5 * (1024 ** 2)
-FASTQ_MODULE_NAMES = ['short_read::paired_end', 'short_read::single_end', 'long_read::nanopore']
+FASTQ_MODULE_NAMES = ['short_read::paired_end', 'short_read::single_end', 'long_read::nanopore', 'raw::raw_reads']
 DEFAULT_ENDPOINT = "https://backend.geoseeq.com"
 
 CONFIG_FOLDER = environ.get("XDG_CONFIG_HOME", join(environ["HOME"], ".config"))

geoseeq/organization.py

@@ -1,6 +1,8 @@
 
-from .remote_object import RemoteObject
+import urllib
+
 from .project import Project
+from .remote_object import RemoteObject
 
 
 class Organization(RemoteObject):
@@ -19,7 +21,8 @@ class Organization(RemoteObject):
         self.name = name
 
     def nested_url(self):
-        return f'nested/{self.name}'
+        escaped_name = urllib.parse.quote(self.name, safe="")
+        return f'nested/{escaped_name}'
 
     def _save(self):
         data = {

geoseeq/project.py

@@ -1,11 +1,13 @@
-from .result import ProjectResultFolder
+import logging
+import urllib
+
+import pandas as pd
+
+from .pipeline import Pipeline
 from .remote_object import RemoteObject
+from .result import ProjectResultFolder
 from .sample import Sample
 from .utils import paginated_iterator
-from .pipeline import Pipeline
-import json
-import pandas as pd
-import logging
 
 logger = logging.getLogger("geoseeq_api")
 
@@ -72,7 +74,8 @@ class Project(RemoteObject):
         return data
 
     def nested_url(self):
-        return self.org.nested_url() + f"/sample_groups/{self.name}"
+        escaped_name = urllib.parse.quote(self.name, safe="")
+        return self.org.nested_url() + f"/sample_groups/{escaped_name}"
 
     def _save_group_obj(self):
         data = self.get_post_data()

geoseeq/result/result_folder.py

@@ -1,9 +1,6 @@
-import json
-import logging
 import os
-import time
-import urllib.request
-from os.path import basename, getsize, join, isfile, isdir, dirname
+import urllib
+from os.path import basename, dirname, getsize, isdir, isfile, join
 from pathlib import Path
 from tempfile import NamedTemporaryFile
 
@@ -155,7 +152,8 @@ class SampleResultFolder(ResultFolder, SampleBioInfoFolder):
         self.is_private = is_private
 
     def nested_url(self):
-        return self.sample.nested_url() + f"/analysis_results/{self.module_name}"
+        escaped_name = urllib.parse.quote(self.module_name, safe="")
+        return self.sample.nested_url() + f"/analysis_results/{escaped_name}"
 
     def _save(self):
         data = {field: getattr(self, field) for field in self.remote_fields if hasattr(self, field)}

geoseeq/sample.py

@@ -1,5 +1,7 @@
-from .result import SampleResultFolder, SampleResultFile
+import urllib
+
 from .remote_object import RemoteObject
+from .result import SampleResultFile, SampleResultFolder
 
 
 class Sample(RemoteObject):
@@ -33,7 +35,8 @@ class Sample(RemoteObject):
         return f'brn:{self.knex.instance_code()}:sample:{self.uuid}'
 
     def nested_url(self):
-        return self.lib.nested_url() + f"/samples/{self.name}"
+        escaped_name = urllib.parse.quote(self.name, safe="")
+        return self.lib.nested_url() + f"/samples/{escaped_name}"
 
     def change_library(self, new_lib):
         self.new_lib = new_lib
@@ -212,9 +215,19 @@ class Sample(RemoteObject):
                         )
                     else:
                         files[read_type][folder_name].append(
-                            self._grn_to_file(file_grn[0])
+                            self._grn_to_file(file_grn)
                         )
         return files
+    
+    def get_one_fasta(self):
+        """Return a 2-ple, a fasta ResultFile and a string with the read type.
+
+        Does not download the file.
+        """
+        url = f"data/samples/{self.uuid}/one-fasta"
+        blob = self.knex.get(url)
+        file = self._grn_to_file(blob["grn"])
+        return file, blob["read_type"]
 
     def __str__(self):
         return f"<Geoseeq::Sample {self.name} {self.uuid} />"

{geoseeq-0.6.3.dist-info → geoseeq-0.6.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: geoseeq
-Version: 0.6.3
+Version: 0.6.5
 Summary: GeoSeeq command line tools and python API
 Author: David C. Danko
 Author-email: "David C. Danko" <dcdanko@biotia.io>

{geoseeq-0.6.3.dist-info → geoseeq-0.6.5.dist-info}/RECORD

@@ -2,14 +2,14 @@ geoseeq/__init__.py,sha256=4_5QIXLt29gDwxjDLtyJY6y5Yi5p0usvKnAuJkkmECY,946
 geoseeq/app.py,sha256=Y6d1UzxFLfE3RNccATbFCVi6kH3eFmzwoUbeR2Ry09A,2387
 geoseeq/blob_constructors.py,sha256=AkWpDQY0EdGMxF1p6eRspyHKubcUdiW4it-_Q7S2QWk,188
 geoseeq/bulk_creators.py,sha256=pdn-Dv7yv5SFv-PfDuQbuOnw2W4-BfIfRJVRAhM8U6s,2115
-geoseeq/constants.py,sha256=h9RURz4xs2bZyDrSGocej7ANJvRLr_H1H7JRxpNUXJM,431
+geoseeq/constants.py,sha256=z_ninEd7WsS5DaLntdR-sqAFib6Ie22jlhPKzLvLerw,449
 geoseeq/file_system_cache.py,sha256=HzVZWtwLD2fjWWSo_UfWmGeBltm9He4lP_OqzKwNGWg,4138
 geoseeq/knex.py,sha256=SlK3Z9Y51APecIeJep4eNvFqlwKpQzvtokBnKe0L5Oc,7965
-geoseeq/organization.py,sha256=a9xmGDE0tQsjPJfyFkYnWagxZ8xpdeckkwvkhH6LNIk,2462
+geoseeq/organization.py,sha256=bJkYL8_D-k6IYAaii2ZbxjwYnXy6lvu6iLXscxKlA3w,2542
 geoseeq/pipeline.py,sha256=89mhWaecsKnm6tyRkdkaVp4dmZh62_v42Ze0oXf8OTY,9873
-geoseeq/project.py,sha256=-9Y2ik0-BpT3iqh89v8VQBbdadhI58oaUP9oZK8oetc,13741
+geoseeq/project.py,sha256=pVx4etzkYmYAYwcPJsjN9PrI-7GZEkAaz2Q5GFdng1s,13810
 geoseeq/remote_object.py,sha256=Es-JlAz8iLRmCpAzh1MOwUh2MqtbuQM-p8wHIBAqNlQ,7131
-geoseeq/sample.py,sha256=whgEVk6GnDJJLjn5uTOqFqRtVxZD3BgjTo7brAC5noU,7981
+geoseeq/sample.py,sha256=KkN9fUTkjTIvozR5y2pS9c6rt4jwHN2Bap38dU_Lrck,8391
 geoseeq/search.py,sha256=gawad6Cx5FxJBPlYkXWb-UKAO-UC0_yhvyU9Ca1kaNI,3388
 geoseeq/upload_download_manager.py,sha256=FMRqLLg77o1qFbWZc5Yc86a2pjeZrrn1rHJr1iaxKCU,8757
 geoseeq/user.py,sha256=tol8i1UGLRrbMw5jeJDnna1ikRgrCDd50Jxz0a1lSgg,690
@@ -22,7 +22,7 @@ geoseeq/cli/detail.py,sha256=q8Suu-j2k18knfSVFG-SWWGNsKM-n8y9RMA3LcIIi9Y,4132
 geoseeq/cli/download.py,sha256=QTNA7qFjCdRJg2vKbAm5yH8WGlcF5fb5bSjm5QiI4XE,17768
 geoseeq/cli/fastq_utils.py,sha256=-bmeQLaiMBm57zWOF0R5OlWTU0_3sh1JBC1RYw2BOFM,3083
 geoseeq/cli/get_eula.py,sha256=79mbUwyiF7O1r0g6UTxG9kJGQEqKuH805E6eLkPC6Y4,997
-geoseeq/cli/main.py,sha256=9yQRXR8Bs304a0m0eVFnrtmTClChED9S9C3ns2I0szA,3830
+geoseeq/cli/main.py,sha256=6ctfMtfT9-V5rUquEqeY39xyPrIeelCQw_8WmHep18I,3898
 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
@@ -48,9 +48,6 @@ geoseeq/contrib/ncbi/api.py,sha256=WQeLoGA_-Zha-QeSO8_i7HpvXyD8UkV0qc5okm11KiA,1
 geoseeq/contrib/ncbi/bioproject.py,sha256=_oThTd_iLDOC8cLOlJKAatSr362OBYZCEV3YrqodhFg,4341
 geoseeq/contrib/ncbi/cli.py,sha256=j9zEcaZPTryK3a4xluRxigcJKDhRpRxbp3KZSx-Bfhk,2400
 geoseeq/contrib/ncbi/setup_logging.py,sha256=Tp1bY1U0f-o739aHpvVYriG2qdd1lFvCYBXZeXQgt-w,175
-geoseeq/file_system/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-geoseeq/file_system/filesystem_download.py,sha256=8bcnxjWltekmCvb5N0b1guBIjLp4-CL2VtsEok-snv4,16963
-geoseeq/file_system/main.py,sha256=4HgYGq7WhlF96JlVIf16iFBTDujlBpxImmtoh4VCzDA,3627
 geoseeq/id_constructors/__init__.py,sha256=w5E0PNQ9UuAxBeZbDI7KBnUoERd85gGz3nScz45bd2o,126
 geoseeq/id_constructors/from_blobs.py,sha256=aj7M7NRpKGs3u3xUvuFJwmJdFeIcJPmaI2_bhwbFfEs,5702
 geoseeq/id_constructors/from_ids.py,sha256=bbAJX4LnuN70v9bny6N-jAwOudb2-ztHvlMBgRuSDz0,3151
@@ -72,7 +69,7 @@ geoseeq/result/file_chunker.py,sha256=bXq1csuRtqMB5sbH-AfWo6gdPwrivv5DJPuHVj-h08
 geoseeq/result/file_download.py,sha256=KalIkwBbFI8xRpbhToixfd1KMAu_0FYwxdKq146NAHw,7832
 geoseeq/result/file_upload.py,sha256=xs1DrI-h4ZP7xN8HPBc3SFpcPAxR5HAolraP1Zu7tvE,10648
 geoseeq/result/result_file.py,sha256=1Yj9fkZhds3J-tay6eNH2-EHi00MovHGV1M80_ckHD8,8677
-geoseeq/result/result_folder.py,sha256=6porOXPh7Tpxw3oX5yMRPYQzNCGYqszqmFJd3SwQmTc,11122
+geoseeq/result/result_folder.py,sha256=-m1lDVLpNHKy-JUGihboVzvdMJEnHossyRnxmBe1XLo,11140
 geoseeq/result/resumable_download_tracker.py,sha256=YEzqHBBnE7L3XokTvlTAhHZ8TcDTIE_pyTQ7YadOfbU,3667
 geoseeq/result/resumable_upload_tracker.py,sha256=2aI09gYz2yw63jEXqs8lmCRKQ79TIc3YuPETvP0Jeek,3811
 geoseeq/result/utils.py,sha256=C-CxGzB3WddlnRiqFSkrY78I_m0yFgNqsTBRzGU-y8Q,2772
@@ -88,9 +85,9 @@ 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-0.6.3.dist-info/LICENSE,sha256=IuhIl1XCxXLPLJT_coN1CNqQU4Khlq7x4IdW7ioOJD8,1067
-geoseeq-0.6.3.dist-info/METADATA,sha256=IH-VB_1NsbGTLGO_j2OuJunReO7L-FSnougvAgZUQkE,4915
-geoseeq-0.6.3.dist-info/WHEEL,sha256=y4mX-SOX4fYIkonsAGA5N0Oy-8_gI4FXw5HNI1xqvWg,91
-geoseeq-0.6.3.dist-info/entry_points.txt,sha256=yF-6KDM8zXib4Al0qn49TX-qM7PUkWUIcYtsgt36rjM,45
-geoseeq-0.6.3.dist-info/top_level.txt,sha256=zZk7mmeaqAYqFJG8nq2DTgSQPbflRjJwkDIhNURPDEU,14
-geoseeq-0.6.3.dist-info/RECORD,,
+geoseeq-0.6.5.dist-info/LICENSE,sha256=IuhIl1XCxXLPLJT_coN1CNqQU4Khlq7x4IdW7ioOJD8,1067
+geoseeq-0.6.5.dist-info/METADATA,sha256=C8OwRXrQg0qp_R7g13Y4qheE5XaoFE6a_n3X471ui_k,4915
+geoseeq-0.6.5.dist-info/WHEEL,sha256=cVxcB9AmuTcXqmwrtPhNK88dr7IR_b6qagTj0UvIEbY,91
+geoseeq-0.6.5.dist-info/entry_points.txt,sha256=yF-6KDM8zXib4Al0qn49TX-qM7PUkWUIcYtsgt36rjM,45
+geoseeq-0.6.5.dist-info/top_level.txt,sha256=zZk7mmeaqAYqFJG8nq2DTgSQPbflRjJwkDIhNURPDEU,14
+geoseeq-0.6.5.dist-info/RECORD,,

{geoseeq-0.6.3.dist-info → geoseeq-0.6.5.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (70.2.0)
+Generator: setuptools (74.1.2)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

geoseeq/file_system/filesystem_download.py

@@ -1,434 +0,0 @@
-
-import os
-import json
-from geoseeq import (
-    result_file_from_id,
-    result_folder_from_id,
-    sample_from_id,
-    project_from_id,
-)
-from geoseeq.utils import md5_checksum
-from time import time
-
-FILE_STATUS_MODIFIED_REMOTE = 'MODIFIED_REMOTE'
-FILE_STATUS_MODIFIED_LOCAL = 'MODIFIED_LOCAL'
-FILE_STATUS_NEW_LOCAL = 'NEW_LOCAL'
-FILE_STATUS_NEW_REMOTE = 'NEW_REMOTE'
-FILE_STATUS_IS_LOCAL_STUB = 'IS_LOCAL_STUB'
-
-
-def dedupe_modified_files(modified_files):
-    """Remove duplicates from a list of modified files.
-    
-    This function will remove duplicates from a list of modified files
-    based on the path to the file. The first instance of the file will be
-    kept and all others will be removed.
-    """
-    seen = set()
-    deduped = []
-    for x in modified_files:
-        if x[2] not in seen:
-            deduped.append(x)
-            seen.add(x[2])
-    return deduped
-
-
-class ResultFileOnFilesystem:
-    """
-
-    Note: unlike other filesystem classes the `path` is a file, not
-    a directory. This is because the file is downloaded directly to 
-    the path.
-    """
-
-    def __init__(self, result_file, path, kind):
-        self.result_file = result_file
-        self.path = path
-        self.kind = kind
-
-    @property
-    def info_filepath(self):
-        dirpath = os.path.dirname(self.path)
-        basename = os.path.basename(self.path)
-        return os.path.join(dirpath, f'.gs_result_file__{basename}')
-    
-    @property
-    def is_stub(self):
-        return os.path.exists(self.path) and os.path.getsize(self.path) == 0
-    
-    def file_is_ok(self, stubs_are_ok=False):
-        if self.is_stub:
-            return stubs_are_ok
-        return self.result_file.download_needs_update(self.path)
-
-    def download(self, use_stubs=False, exists_ok=False):
-        if os.path.exists(self.info_filepath):
-            if exists_ok and self.file_is_ok(stubs_are_ok=use_stubs):
-                return
-            elif not exists_ok:
-                raise ValueError('Result file already exists at path: {}'.format(self.info_filepath))
-        
-        # Download the file
-        if use_stubs:
-            open(self.path, 'w').close()
-        else:
-            self.result_file.download(self.path)
-
-        self.write_info_file()
-
-    def local_file_checksum(self):
-        if self.is_stub:
-            return "__STUB__"
-        return md5_checksum(self.path)
-    
-    def locally_modified(self):
-        raise NotImplementedError('This function is not implemented')
-
-    def status_is_ok(self, stubs_are_ok=False):
-        # check for an info file
-        if not os.path.exists(self.info_filepath):
-            return False
-        if stubs_are_ok:
-            return True
-        return not self.result_file.download_needs_update(self.path)
-    
-    def write_info_file(self):
-        result_file_info = {
-            "uuid": self.result_file.uuid,
-            "kind": self.kind,
-            "checksum": self.local_file_checksum(),
-        }
-        with open(self.info_filepath, 'w') as f:
-            json.dump(result_file_info, f)
-
-    @classmethod
-    def from_path(cls, path):
-        obj = cls(None, path, None)
-        try:
-            with open(obj.info_filepath, 'r') as f:
-                result_file_info = json.load(f)
-            obj.result_file = result_file_from_id(result_file_info['uuid'])
-            obj.kind = result_file_info['kind']
-            obj.stored_checksum = result_file_info['checksum']
-        except FileNotFoundError:
-            pass
-        return obj
-    
-    def write_info_file(self):
-        result_file_info = {
-            "uuid": self.result_file.uuid,
-            "kind": self.kind,
-            "checksum": self.local_file_checksum(),
-        }
-        with open(self.info_filepath, 'w') as f:
-            json.dump(result_file_info, f)
-    
-    def list_abnormal_objects(self):
-        """Return a list of files that have been modified.
-        
-        Since this class is a single file the list will either be empty
-        or have one element.
-
-        Note that if a file was modified locally then uploaded to the server
-        the file will be marked as modified remote.
-        """
-        if self.result_file is None:
-            return [('FILE', FILE_STATUS_NEW_LOCAL, self.path, None)]
-        if not os.path.exists(self.path):
-            return [('FILE', FILE_STATUS_NEW_REMOTE, self.path, self.result_file)]
-        if self.is_stub:
-            return [('FILE', FILE_STATUS_IS_LOCAL_STUB, self.path, self.result_file)]
-        if self.result_file and self.result_file.download_needs_update(self.path):
-            return [('FILE', FILE_STATUS_MODIFIED_REMOTE, self.path, self.result_file)]
-        if self.locally_modified():
-            return [('FILE', FILE_STATUS_MODIFIED_LOCAL, self.path, self.result_file)]
-
-        return []
-
-    
-class ResultFolderOnFilesystem:
-
-    def __init__(self, result_folder, path, kind):
-        self.result_folder = result_folder
-        self.path = path
-        self.kind = kind
-
-    @property
-    def info_filepath(self):
-        return os.path.join(self.path, '.gs_result_folder')
-
-    def download(self, use_stubs=False, exists_ok=False):
-        if os.path.exists(self.info_filepath) and not exists_ok:
-            raise ValueError('Result folder already exists at path: {}'.format(self.info_filepath))
-        
-        # Download the files in the result folder
-        for result_file in self.result_folder.get_fields():
-            result_file_local_path = os.path.join(self.path, result_file.name)
-            os.makedirs(os.path.dirname(result_file_local_path), exist_ok=True)
-            ResultFileOnFilesystem(result_file, result_file_local_path, self.kind)\
-                .download(use_stubs=use_stubs, exists_ok=exists_ok)
-        
-        # Write the result folder data
-        result_folder_info = {
-            "uuid": self.result_folder.uuid,
-            "kind": self.kind
-        }
-        with open(self.info_filepath, 'w') as f:
-            json.dump(result_folder_info, f)
-
-    def status_is_ok(self):
-        # check for an info file
-        if not os.path.exists(self.info_filepath):
-            return False
-        
-        # check that all files are downloaded
-        for result_file in self.result_folder.get_files():
-            result_file_path = os.path.join(self.path, result_file.name)
-            if not os.path.exists(result_file_path):
-                return False
-        
-        return True
-    
-    @classmethod
-    def from_path(cls, path):
-        obj = cls(None, path, None)
-        try:
-            with open(os.path.join(path, '.gs_result_folder'), 'r') as f:
-                result_folder_info = json.load(f)
-            obj.result_folder = result_folder_from_id(result_folder_info['uuid'])
-            obj.kind = result_folder_info['kind']
-        except FileNotFoundError:
-            pass
-        return obj
-    
-    def list_abnormal_objects(self):
-        """Return a list of files that have been modified.
-        
-        This function will return a list of tuples where the first element
-        is the status of the file and the second element is the path to the file.
-        """
-        modified_files = []
-        if not self.result_folder:
-            modified_files.append(('FOLDER', FILE_STATUS_NEW_LOCAL, self.path, None))
-        if not os.path.exists(self.path):
-            modified_files.append(('FOLDER', FILE_STATUS_NEW_REMOTE, self.path, self.result_folder))
-
-        # list local files
-        if os.path.exists(self.path):
-            for local_file in os.listdir(self.path):
-                if local_file.startswith('.gs_'):
-                    continue
-                local_file_path = os.path.join(self.path, local_file)
-                result_file_on_fs = ResultFileOnFilesystem.from_path(local_file_path)
-                modified_files.extend(result_file_on_fs.list_abnormal_objects())
-
-        # list remote files
-        if self.result_folder:
-            for result_file in self.result_folder.get_fields():
-                result_file_path = os.path.join(self.path, result_file.name)
-                result_file_on_fs = ResultFileOnFilesystem(result_file, result_file_path, self.kind)
-                modified_files.extend(result_file_on_fs.list_abnormal_objects())
-    
-        return dedupe_modified_files(modified_files)
-    
-
-class SampleOnFilesystem:
-
-    def __init__(self, sample, path):
-        self.sample = sample
-        self.path = path if path[-1] != '/' else path[:-1]  # remove trailing slash
-
-    @property
-    def info_filepath(self):
-        return os.path.join(self.path, '.gs_sample')
-
-    def download(self, use_stubs=False, exists_ok=False):
-        if os.path.exists(self.info_filepath) and not exists_ok:
-            raise ValueError('Sample already exists at path: {}'.format(self.info_filepath))
-        
-        # download result folders
-        for result_folder in self.sample.get_result_folders():
-            result_folder_local_path = os.path.join(self.path, result_folder.name)
-            os.makedirs(result_folder_local_path, exist_ok=True)
-            ResultFolderOnFilesystem(result_folder, result_folder_local_path, "sample")\
-                .download(use_stubs=use_stubs, exists_ok=exists_ok)
-
-        # Write the sample data
-        sample_info = {
-            "uuid": self.sample.uuid
-        }
-        with open(self.info_filepath, 'w') as f:
-            json.dump(sample_info, f)
-
-    def status_is_ok(self):
-        # check for an info file
-        if not os.path.exists(self.info_filepath):
-            return False
-        
-        # check that all result folders are downloaded
-        for result_folder in self.sample.get_result_folders():
-            result_folder_local_path = os.path.join(self.path, result_folder.name)
-            result_folder_on_fs = ResultFolderOnFilesystem.from_path(result_folder_local_path, "sample")
-            if not result_folder_on_fs.status_is_ok():
-                return False
-        
-        return True
-
-    @classmethod
-    def from_path(cls, path):
-        obj = cls(None, path)
-        try:
-            with open(os.path.join(path, '.gs_sample'), 'r') as f:
-                sample_info = json.load(f)
-            obj.sample = sample_from_id(sample_info['uuid'])
-        except FileNotFoundError:
-            pass
-        return obj
-    
-    def list_abnormal_objects(self):
-        """Return a list of files that have been modified.
-        
-        This function will return a list of tuples where the first element
-        is the status of the file and the second element is the path to the file.
-        """
-        modified_files = []
-        if not self.sample:
-            modified_files.append(('SAMPLE', FILE_STATUS_NEW_LOCAL, self.path, None))
-        if not os.path.exists(self.path):
-            modified_files.append(('SAMPLE', FILE_STATUS_NEW_REMOTE, self.path, self.sample))
-
-        # list local folders
-        if os.path.exists(self.path):
-            for local_folder in os.listdir(self.path):
-                local_folder_path = os.path.join(self.path, local_folder)
-                if not os.path.isdir(local_folder_path):
-                    continue
-                result_folder_on_fs = ResultFolderOnFilesystem.from_path(local_folder_path)
-                modified_files.extend(result_folder_on_fs.list_abnormal_objects())
-
-        # list remote folders
-        if self.sample:
-            for result_folder in self.sample.get_result_folders():
-                result_folder_path = os.path.join(self.path, result_folder.name)
-                result_folder_on_fs = ResultFolderOnFilesystem(result_folder, result_folder_path, "sample")
-                modified_files.extend(result_folder_on_fs.list_abnormal_objects())
-
-        return dedupe_modified_files(modified_files)
-
-
-class ProjectOnFilesystem:
-
-    def __init__(self, project, path):
-        self.project = project
-        self.path = path
-
-    @property
-    def info_filepath(self):
-        return os.path.join(self.path, '.gs_project')
-
-    def download(self, use_stubs=False, exists_ok=False):
-        if os.path.exists(self.info_filepath) and not exists_ok:
-            raise ValueError('Project already exists at path: {}'.format(self.info_filepath))
-        
-        # download samples
-        for sample in self.project.get_samples():
-            sample_local_path = os.path.join(self.path, "sample_results", sample.name)
-            os.makedirs(sample_local_path, exist_ok=True)
-            SampleOnFilesystem(sample, sample_local_path)\
-                .download(use_stubs=use_stubs, exists_ok=exists_ok)
-        
-        # download project result folders
-        for result_folder in self.project.get_result_folders():
-            result_folder_local_path = os.path.join(self.path, "project_results", result_folder.name)
-            os.makedirs(result_folder_local_path, exist_ok=True)
-            ResultFolderOnFilesystem(result_folder, result_folder_local_path, "project")\
-                .download(use_stubs=use_stubs, exists_ok=exists_ok)
-
-        # Write the project data
-        project_info = {
-            "uuid": self.project.uuid
-        }
-        with open(self.info_filepath, 'w') as f:
-            json.dump(project_info, f)
-
-    def status_is_ok(self):
-        # check for an info file
-        if not os.path.exists(self.info_filepath):
-            return False
-        
-        # check that all samples are downloaded
-        for sample in self.project.get_samples():
-            sample_local_path = os.path.join(self.path, "sample_results", sample.name)
-            sample_on_fs = SampleOnFilesystem.from_path(sample_local_path)
-            if not sample_on_fs.status_is_ok():
-                return False
-        
-        # check that all project result folders are downloaded
-        for result_folder in self.project.get_result_folders():
-            result_folder_local_path = os.path.join(self.path, "project_results", result_folder.name)
-            result_folder_on_fs = ResultFolderOnFilesystem.from_path(result_folder_local_path, "project")
-            if not result_folder_on_fs.status_is_ok():
-                return False
-        
-        return True
-
-    @classmethod
-    def from_path(cls, path, recursive=False):
-        try:
-            with open(os.path.join(path, '.gs_project'), 'r') as f:
-                project_info = json.load(f)
-            project = project_from_id(project_info['uuid'])
-            return cls(project, path)
-        except FileNotFoundError:
-            if not recursive:
-                raise ValueError('No project found in path or parent directories')
-            updir = os.path.dirname(os.path.abspath(path))
-            if updir == path:
-                raise ValueError('No project found in path or parent directories')
-            return cls.from_path(updir, recursive=recursive)
-        
-    def path_from_project_root(self, path):
-        if path[0] == "/":
-            return path.replace(self.path, "")[1:]
-        return path
-
-    def list_abnormal_objects(self):
-        """Return a list of files that have been modified.
-        
-        This function will return a list of tuples where the first element
-        is the status of the file and the second element is the path to the file.
-        """
-        modified_files = []
-
-        # list remote samples
-        for sample in self.project.get_samples():
-            sample_path = os.path.join(self.path, "sample_results", sample.name)
-            sample_on_fs = SampleOnFilesystem(sample, sample_path)
-            modified_files.extend(sample_on_fs.list_abnormal_objects())
-
-        # list remote project result folders
-        for result_folder in self.project.get_result_folders():
-            result_folder_path = os.path.join(self.path, "project_results", result_folder.name)
-
-            result_folder_on_fs = ResultFolderOnFilesystem(result_folder, result_folder_path, "project")
-            modified_files.extend(result_folder_on_fs.list_abnormal_objects())
-
-        # list local samples
-        for local_sample in os.listdir(os.path.join(self.path, "sample_results")):
-            local_sample_path = os.path.join(self.path, "sample_results", local_sample)
-            if not os.path.isdir(local_sample_path):
-                continue
-            sample_on_fs = SampleOnFilesystem.from_path(local_sample_path)
-            modified_files.extend(sample_on_fs.list_abnormal_objects())
-
-        # list local project result folders
-        for local_result_folder in os.listdir(os.path.join(self.path, "project_results")):
-            local_result_folder_path = os.path.join(self.path, "project_results", local_result_folder)
-            if not os.path.isdir(local_result_folder_path):
-                continue
-            result_folder_on_fs = ResultFolderOnFilesystem.from_path(local_result_folder_path)
-            modified_files.extend(result_folder_on_fs.list_abnormal_objects())
-        return dedupe_modified_files(modified_files)
-                
-
-

geoseeq/file_system/main.py

@@ -1,122 +0,0 @@
-from fuse import FUSE, Operations
-import os
-
-
-class GeoSeeqProjectFileSystem(Operations):
-    """Mount a GeoSeeq project as a filesystem.
-    
-    The project will automatically have this directory structure:
-    - <root>/project_results/<project_result_folder_name>/...
-    - <root>/sample_results/<sample_name>/...
-    - <root>/metadata/sample_metadata.csv
-    - <root>/.config/config.json
-    """
-
-    def __init__(self, root, project):
-        self.root = root
-        self.project = project
-
-    def access(self, path, mode):
-        pass
-
-    def chmod(self, path, mode):
-        pass
-
-    def chown(self, path, uid, gid):
-        pass
-
-    def getattr(self, path, fh=None):
-        pass
-
-    def readdir(self, path, fh):
-        pass
-
-    def readlink(self, path):
-        pass
-
-    def mknod(self, path, mode, dev):
-        pass
-
-    def rmdir(self, path):
-        pass
-
-    def mkdir(self, path, mode):
-        pass
-
-    def statfs(self, path):
-        pass
-
-    def unlink(self, path):
-        pass
-
-    def symlink(self, name, target):
-        pass
-
-    def rename(self, old, new):
-        pass
-
-    def link(self, target, name):
-        pass
-
-    def utimens(self, path, times=None):
-        pass
-
-    def open(self, path, flags):
-        tkns = path.split('/')
-        if tkns[0] == 'project_results':
-            result_folder_name, result_file_name = tkns[2], '/'.join(tkns[3:])
-            result_folder = self.project.get_result_folder(result_folder_name).get()
-            result_file = result_folder.get_file(result_file_name).get()
-            result_file.download(path)
-        elif tkns[0] == 'sample_results':
-            sample_name, result_folder_name, result_file_name = tkns[2], tkns[3], '/'.join(tkns[4:])
-            sample = self.project.get_sample(sample_name).get()
-            result_folder = sample.get_result_folder(result_folder_name).get()
-            result_file = result_folder.get_file(result_file_name).get()
-            result_file.download(path)
-        elif tkns[0] == 'metadata':
-            raise NotImplementedError('TODO')
-        
-        return os.open(self._full_local_path(path), flags)
-
-    def create(self, path, mode, fi=None):
-        tkns = path.split('/')
-        if tkns[0] == 'project_results':
-            result_name, file_name = tkns[2], '/'.join(tkns[3:])
-            result_folder = self.project.get_result_folder(result_name).idem()
-            result_file = result_folder.get_file(file_name).create()
-            result_file.download(path)  # nothing to download at this point
-        elif tkns[0] == 'sample_results':
-            sample_name, result_folder_name, result_file_name = tkns[2], tkns[3], '/'.join(tkns[4:])
-            sample = self.project.get_sample(sample_name).idem()
-            result_folder = sample.get_result_folder(result_folder_name).idem()
-            result_file = result_folder.get_file(result_file_name).create()
-            result_file.download(path)  # nothing to download at this point
-        elif tkns[0] == 'metadata':
-            raise NotImplementedError('TODO')
-
-    def read(self, path, length, offset, fh):
-        os.lseek(fh, offset, os.SEEK_SET)
-        return os.read(fh, length)
-
-    def write(self, path, buf, offset, fh):
-        pass
-
-    def truncate(self, path, length, fh=None):
-        pass
-
-    def flush(self, path, fh):
-        pass
-
-    def release(self, path, fh):
-        pass
-
-    def fsync(self, path, fdatasync, fh):
-        pass
-    
-    def _full_local_path(self, partial):
-        if partial.startswith("/"):
-            partial = partial[1:]
-        return os.path.join(self.root, partial)
-
-