Merge pull request #52 from b97pla/dataops-509_fastq_readme_rb

Dataops 509 fastq readme

.github/workflows/tests.yml

@@ -20,4 +20,4 @@ jobs:
         pip install -r requirements/dev .
     - name: Launch tests
       run: |
-        nosetests ./tests
+        python -m unittest discover -s ./tests -t .

config/app.config

@@ -8,6 +8,7 @@ runfolder_directory: tests/resources/runfolders
 general_project_directory: tests/resources/projects
 staging_directory: /tmp/
 project_links_directory: /tmp/
+readme_directory: tests/resources/readme
 dds_conf:
   log_path: dds.log
 port: 9999

delivery/__init__.py

@@ -1 +1 @@
-__version__ = "3.0.1"
+__version__ = "3.1.0"

delivery/app.py

@@ -25,7 +25,8 @@
     FileSystemBasedUnorganisedRunfolderRepository
 from delivery.repositories.staging_repository import DatabaseBasedStagingRepository
 from delivery.repositories.deliveries_repository import DatabaseBasedDeliveriesRepository
-from delivery.repositories.project_repository import GeneralProjectRepository, UnorganisedRunfolderProjectRepository
+from delivery.repositories.project_repository import GeneralProjectRepository, \
+    UnorganisedRunfolderProjectRepository
 from delivery.repositories.delivery_sources_repository import DatabaseBasedDeliverySourcesRepository
 from delivery.repositories.sample_repository import RunfolderProjectBasedSampleRepository
 
@@ -120,13 +121,19 @@ def _assert_is_dir(directory):
     project_links_directory = config["project_links_directory"]
     _assert_is_dir(project_links_directory)
 
-    runfolder_repo = FileSystemBasedRunfolderRepository(runfolder_dir)
-    project_repository = UnorganisedRunfolderProjectRepository(
-        sample_repository=RunfolderProjectBasedSampleRepository()
+    readme_directory = config["readme_directory"]
+    _assert_is_dir(readme_directory)
+
+    sample_repo = RunfolderProjectBasedSampleRepository()
+    runfolder_repo = FileSystemBasedRunfolderRepository(
+        runfolder_dir
     )
     unorganised_runfolder_repo = FileSystemBasedUnorganisedRunfolderRepository(
         runfolder_dir,
-        project_repository=project_repository
+        project_repository=UnorganisedRunfolderProjectRepository(
+            sample_repository=sample_repo,
+            readme_directory=readme_directory
+        )
     )
 
     general_project_dir = config['general_project_directory']

delivery/models/project.py

@@ -35,18 +35,29 @@ def __hash__(self):
 
 class RunfolderProject(BaseProject):
     """
-    Model a project directory in a runfolder on disk. Note that this means that this project model only extends
-    to the idea of projects as subdirectories in a demultiplexed Illumina runfolder.
+    Model a project directory in a runfolder on disk. Note that this means that this project model
+    only extends to the idea of projects as subdirectories in a demultiplexed Illumina runfolder.
     """
 
-    def __init__(self, name, path, runfolder_path, runfolder_name, samples=None, project_files=None):
+    def __init__(
+            self,
+            name,
+            path,
+            runfolder_path,
+            runfolder_name,
+            samples=None,
+            project_files=None
+    ):
         """
         Instantiate a new `RunfolderProject` object
+
         :param name: of the project
         :param path: path to the project
         :param runfolder_path: path the runfolder in which this project is stored.
         :param runfolder_name: name of the runfolder in which this project is stored
         :param samples: list of instances of Sample, representing samples in the project
+        :param project_files: list of instances of RunfolderFile, representing files belonging to
+        the project. These do not include files already accounted for in the `samples` list.
         """
         self.name = name
         self.path = os.path.abspath(path)

delivery/models/runfolder.py

@@ -37,7 +37,63 @@ def __hash__(self):
 
 class RunfolderFile(object):
 
-    def __init__(self, file_path, file_checksum=None):
+    def __init__(
+            self,
+            file_path,
+            base_path=None,
+            file_checksum=None
+    ):
+        """
+        A `RunfolderFile` object representing a file in the runfolder
+
+        If specified, the `base_path` parameter should specify the path that the file will be
+        considered relative to. For example, if `file_path` is `/path/to/example/file_name` and
+        `base_path` is `/path/to`, the file object, if moved or symlinked, will be placed under the
+        intermediate directory, i.e. `example/file_name`.
+
+        :param file_path: the path to the file
+        :param base_path: a path relative to which the file will be considered
+        :param file_checksum: a computed checksum for the file
+        """
         self.file_path = os.path.abspath(file_path)
         self.file_name = os.path.basename(file_path)
+        self.base_path = base_path or os.path.dirname(self.file_path)
         self.checksum = file_checksum
+
+    @classmethod
+    def create_object_from_path(
+            cls,
+            file_path,
+            runfolder_path,
+            filesystem_service,
+            metadata_service,
+            base_path=None,
+            checksums=None
+    ):
+        """
+        Factory method that creates and returns a class instance.
+
+        :param file_path: the path to the file
+        :param runfolder_path: the path to the runfolder containing the file
+        :param filesystem_service: a service which can access the file system
+        :param metadata_service: a MetadataService for reading and writing metadata files
+        :param base_path: a path relative to which the file will be considered
+        :param checksums: a list of pre-computed checksums that may or may not contain an entry for
+        the `file_path` relative to the `runfolder_path`
+        :return: a class object representing a file in the runfolder
+        """
+        checksums = checksums or {}
+        relative_file_path = filesystem_service.relpath(
+            file_path,
+            filesystem_service.dirname(
+                runfolder_path
+            )
+        )
+        checksum = checksums[relative_file_path] \
+            if relative_file_path in checksums \
+            else metadata_service.hash_file(file_path)
+        return cls(
+            file_path,
+            base_path=base_path,
+            file_checksum=checksum
+        )

delivery/models/sample.py

@@ -46,19 +46,31 @@ def __init__(
             lane_no=None,
             read_no=None,
             is_index=None,
-            checksum=None):
+            base_path=None,
+            checksum=None
+    ):
         """
-        Instantiate a new `SampleFile` object
+        A `SampleFile` object
+
+        If specified, the `base_path` parameter should specify the path that the file will be
+        considered relative to. For example, if `file_path` is `/path/to/example/file_name` and
+        `base_path` is `/path/to`, the file object, if moved or symlinked, will be placed under the
+        intermediate directory, i.e. `example/file_name`
 
         :param sample_path: the path to the file
         :param sample_name: the name of the sample
         :param sample_index: the sample index designator
         :param lane_no: the lane number the sequences in the file were derived from
         :param read_no: the read number
         :param is_index: if True, the sequence file contains index sequences
+        :param base_path: a path relative to which the file will be considered
         :param checksum: the MD5 checksum for this SampleFile
         """
-        super(SampleFile, self).__init__(sample_path, file_checksum=checksum)
+        super(SampleFile, self).__init__(
+            sample_path,
+            base_path=base_path,
+            file_checksum=checksum
+        )
         self.sample_name = sample_name
         self.sample_index = sample_index
         self.lane_no = lane_no
@@ -77,4 +89,5 @@ def __hash__(self):
             self.lane_no,
             self.read_no,
             self.is_index,
-            self.checksum))
+            self.checksum
+        ))