geoseeq 0.6.14.dev7__py3-none-any.whl → 0.7.0__py3-none-any.whl

geoseeq/cli/download.py

@@ -414,7 +414,7 @@ def _make_read_configs(download_results, config_dir="."):
         "reads_1": ["small.fq.gz"],
         "reads_2": [],
         "fastq_checksum": "",
-        "data_type": "short-read",
+        "data_type": "single",
         "bdx_result_dir": "results",
         "geoseeq_uuid": "05bf22e9-9d25-42db-af25-31bc538a7006"
     }
@@ -428,7 +428,7 @@ def _make_read_configs(download_results, config_dir="."):
                 "reads_1": [],
                 "reads_2": [],
                 "fastq_checksum": "",
-                "data_type": "short-read",
+                "data_type": read_type,
                 "bdx_result_dir": "results",
                 "geoseeq_uuid": sample.uuid,
             }

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('0.6.14dev7')  # remember to update pyproject.toml
+    click.echo('0.7.0')  # remember to update pyproject.toml
 
 
 @main.group('advanced')

geoseeq/cli/upload/__init__.py

@@ -4,9 +4,11 @@ from .upload import (
     cli_upload_file,
     cli_upload_folder,
     cli_metadata,
+    cli_upload_smart_table,
+    cli_upload_smart_tree,
 )
 from .upload_reads import cli_upload_reads_wizard
-from .upload_advanced import cli_find_urls_for_reads
+from .upload_advanced import cli_find_urls_for_reads, cli_upload_from_config
 
 @click.group('upload')
 def cli_upload():
@@ -17,10 +19,13 @@ cli_upload.add_command(cli_upload_reads_wizard)
 cli_upload.add_command(cli_upload_file)
 cli_upload.add_command(cli_upload_folder)
 cli_upload.add_command(cli_metadata)
+cli_upload.add_command(cli_upload_smart_table)
+cli_upload.add_command(cli_upload_smart_tree)
 
-@click.group('upload')
+@cli_upload.group('advanced')
 def cli_upload_advanced():
     """Advanced tools to upload files to GeoSeeq."""
     pass
 
-cli_upload_advanced.add_command(cli_find_urls_for_reads)
+cli_upload_advanced.add_command(cli_find_urls_for_reads)
+cli_upload_advanced.add_command(cli_upload_from_config)

geoseeq/cli/upload/upload.py

@@ -263,3 +263,117 @@ def cli_metadata(state, overwrite, yes, private, create, index_col, encoding, pr
             sample.metadata = new_meta
             sample.idem()
     click.echo(f'Wrote metadata for {len(samples)} samples')
+
+
+@click.command('smart-table')
+@use_common_state
+@overwrite_option
+@yes_option
+@private_option
+@click.option('-n', '--geoseeq-file-name', default=None,
+              help='Specify a different name for the file on GeoSeeq than the local file name.',
+              show_default=True)
+@folder_id_arg
+@click.argument('file_path', type=click.Path(exists=True), nargs=1)
+def cli_upload_smart_table(state, overwrite, yes, private, folder_id, geoseeq_file_name, file_path):
+    """Upload a smart table to GeoSeeq.
+
+    This command uploads a smart table to a project or sample on GeoSeeq. It can be used to upload
+    a single file to a folder at once.
+    
+    ---
+
+    Example Usage:
+
+    \b
+    # Upload a smart table from a file
+    $ geoseeq upload smart-table "My Org/My Project/My Sample/My Folder" /path/to/my_table.csv
+
+    \b
+    # Upload a smart table from a file but name it "My Smart Table" on GeoSeeq
+    $ geoseeq upload smart-table "My Org/My Project/My Sample/My Folder" /path/to/my_table.csv -n "My Smart Table"
+
+    ---
+
+    Command Arguments:
+
+    [FOLDER_ID] Can be a folder UUID, GeoSeeq Resource Number (GRN), or an
+    names for an org, project, sample, folder separated by a slash. Can exclude
+    the sample name if the folder is for a project.
+
+    [FILE_PATH] A path to a file on your local machine.
+
+    ---
+    """
+    knex = state.get_knex()
+    result_folder = handle_folder_id(knex, folder_id, yes=yes, private=private)
+    
+    if not geoseeq_file_name:
+        geoseeq_file_name = basename(file_path)
+    
+    if not overwrite and result_folder.result_file(geoseeq_file_name).exists():
+        raise click.UsageError(f'{geoseeq_file_name} already exists in {result_folder}. Use --overwrite to overwrite it.')
+
+    result_file = result_folder.result_file(geoseeq_file_name)
+    smart_table = result_file.as_smart_table()
+    smart_table.import_csv(file_path)
+
+
+@click.command('smart-tree')
+@use_common_state
+@click.option('-m/-nm', '--make-name-map/--no-name-map', default=True, help="Create a sample name map with all samples currently in the project.")
+@overwrite_option
+@yes_option
+@private_option
+@click.option('-n', '--geoseeq-file-name', default=None,
+              help='Specify a different name for the file on GeoSeeq than the local file name.',
+              show_default=True)
+@folder_id_arg
+@click.argument('newick_file_path', type=click.Path(exists=True), nargs=1)
+def cli_upload_smart_tree(state, make_name_map, overwrite, yes, private, folder_id, geoseeq_file_name, newick_file_path):
+    """Upload a smart tree to GeoSeeq.
+
+    This command uploads a smart tree to a project or sample on GeoSeeq. It can be used to upload
+    a single file to a folder at once.
+    
+    ---
+
+    Example Usage:
+
+    \b
+    # Upload a smart tree from a file
+    $ geoseeq upload smart-tree "My Org/My Project/My Sample/My Folder" /path/to/my_tree.nwk
+
+    \b
+    # Upload a smart tree from a file but name it "My Smart Tree" on GeoSeeq
+    $ geoseeq upload smart-tree "My Org/My Project/My Sample/My Folder" /path/to/my_tree.nwk -n "My Smart Tree"
+
+    ---
+
+    Command Arguments:
+
+    [FOLDER_ID] Can be a folder UUID, GeoSeeq Resource Number (GRN), or an
+    names for an org, project, sample, folder separated by a slash. Can exclude
+    the sample name if the folder is for a project.
+
+    [NEWICK_FILE_PATH] A path to a newick file on your local machine.
+
+    ---
+    """
+    knex = state.get_knex()
+    result_folder = handle_folder_id(knex, folder_id, yes=yes, private=private)
+    
+    if not geoseeq_file_name:
+        geoseeq_file_name = basename(newick_file_path)
+    
+    if not overwrite and result_folder.result_file(geoseeq_file_name).exists():
+        raise click.UsageError(f'{geoseeq_file_name} already exists in {result_folder}. Use --overwrite to overwrite it.')
+
+    result_file = result_folder.result_file(geoseeq_file_name)
+    smart_tree = result_file.as_smart_tree()
+    with open(newick_file_path) as f:
+        newick_str = f.read()
+    smart_tree.create_from_newick(newick_str)
+    if make_name_map:
+        smart_tree.add_all_samples_to_map(result_folder.project)
+    smart_tree.idem()

geoseeq/cli/upload/upload_advanced.py

@@ -24,6 +24,16 @@ from geoseeq.cli.shared_params import (
 
 from geoseeq.constants import FASTQ_MODULE_NAMES
 from geoseeq.cli.progress_bar import PBarManager
+import pandas as pd
+from typing import Dict, Optional
+from geoseeq.id_constructors.from_ids import (
+    org_from_id,
+    project_from_id,
+    sample_from_id,
+    result_folder_from_id,
+    result_file_from_id,
+)
+from geoseeq.upload_download_manager import GeoSeeqUploadManager
 
 logger = logging.getLogger('geoseeq_api')
 
@@ -90,3 +100,223 @@ def cli_find_urls_for_reads(state, cores, overwrite, yes, regex, private, module
     groups = _group_files(knex, filepaths, module_name, regex, yes)
     for file_name, target_url in _find_target_urls(groups, module_name, proj, filepaths, overwrite, cores, state):
         print(f'{file_name}\t{target_url}', file=state.outfile)
+
+
+def _get_result_file_from_record_with_ids(knex, record: Dict) -> Dict:
+    """Get all relevant objects from a record, handling GRNs, UUIDs, and absolute names without requiring parent objects.
+    
+    Returns a dict with 'org', 'project', 'sample', 'folder', and 'result_file' keys.
+    Objects may be None if not needed/specified.
+    Guaranteed that at least org is not None.
+    """
+    objects = {
+        'org': None,
+        'project': None,
+        'sample': None,
+        'folder': None,
+        'result_file': None
+    }
+
+    # Try to get file directly - if it's a GRN/UUID we don't need parent objects
+    try:
+        objects['result_file'] = result_file_from_id(knex, record['filename'])
+        objects['folder'] = objects['result_file'].folder
+        if hasattr(objects['folder'], 'sample'):
+            objects['sample'] = objects['folder'].sample
+            objects['project'] = objects['sample'].project
+        else:
+            objects['project'] = objects['folder'].project
+        objects['org'] = objects['project'].org
+        return objects
+    except ValueError:
+        pass  # Not a GRN, UUID or abs name. Continue with normal flow
+
+    # Try to get folder directly - if it's a GRN/UUID we don't need parent objects
+    try:
+        objects['folder'] = result_folder_from_id(knex, record['folder'])
+        # Get parent objects from folder
+        if hasattr(objects['folder'], 'sample'):
+            objects['sample'] = objects['folder'].sample
+            objects['project'] = objects['sample'].project
+        else:
+            objects['project'] = objects['folder'].project
+        objects['org'] = objects['project'].org
+        return objects
+    except ValueError:
+        pass  # Not a GRN, UUID or abs name. Continue with normal flow
+
+    # Try to get sample directly if specified
+    if pd.notna(record['sample']):
+        try:
+            objects['sample'] = sample_from_id(knex, record['sample'])
+            objects['project'] = objects['sample'].project
+            objects['org'] = objects['project'].org
+            return objects
+        except ValueError:
+            pass  # Not a GRN, UUID or abs name. Continue with normal flow
+
+    # Try to get project directly
+    try:
+        objects['project'] = project_from_id(knex, record['project'])
+        objects['org'] = objects['project'].org
+        return objects
+    except ValueError:
+        pass  # Not a GRN/UUID, continue
+
+    
+    if objects['org'] is None: # Get org directly if we don't have one yet
+        objects['org'] = org_from_id(knex, record['organization'])
+    
+    return objects
+
+
+def _get_result_file_from_record(knex, record: Dict) -> Dict:
+    """Get all relevant objects from a record, handling GRNs/UUIDs without requiring parent objects.
+    
+    Returns a dict with 'org', 'project', 'sample', 'folder', and 'result_file' keys.
+    Objects may be None if not needed/specified.
+    """
+    objects = _get_result_file_from_record_with_ids(knex, record)
+
+    if objects['project'] is None:
+        objects['project'] = objects['org'].project(record['project'])
+
+    if objects['sample'] is None:
+        if pd.notna(record['sample']):
+            objects['sample'] = objects['project'].sample(record['sample'])
+            parent = objects['sample']
+    else:
+        parent = objects['project']
+
+    if objects['folder'] is None:
+        objects['folder'] = parent.result_folder(record['folder'])
+
+    if objects['result_file'] is None:
+        objects['result_file'] = objects['folder'].result_file(record['filename'])
+
+    objects['result_file'].idem()
+    print(objects)
+    return objects
+
+
+def _add_record_to_upload_manager_local_file(record: Dict, result_file, upload_manager: GeoSeeqUploadManager) -> None:
+    """Add a local file upload to the upload manager."""
+    upload_manager.add_result_file(result_file, record['path'], link_type='upload')
+
+
+def _add_record_to_upload_manager_s3_file(record: Dict, result_file, upload_manager: GeoSeeqUploadManager) -> None:
+    """Add an S3 file link to the upload manager.
+    
+    Handles two types of S3 URLs:
+    1. https://endpoint/bucket/key - Full URL with endpoint included
+    2. s3://bucket/key - S3 protocol URL that needs endpoint added
+    """
+    path = record['path']
+    
+    if path.startswith('s3://'):
+        # Convert s3:// URL to https:// URL
+        if not record['endpoint_url']:
+            raise ValueError("endpoint_url is required for s3:// URLs")
+        
+        # Remove s3:// prefix and combine with endpoint
+        bucket_and_key = path[5:]  # len('s3://') == 5
+        path = f"{record['endpoint_url'].rstrip('/')}/{bucket_and_key}"
+    elif not path.startswith('https://'):
+        raise ValueError("S3 URLs must start with either 's3://' or 'https://'")
+
+    upload_manager.add_result_file(result_file, path, link_type='s3')
+
+
+def _upload_one_record(knex, record: Dict, overwrite: bool, upload_manager: GeoSeeqUploadManager) -> Dict:
+    """Process a single record from the config file and add it to the upload manager."""
+    objects = _get_result_file_from_record(knex, record)
+    if not objects['result_file']:
+        raise ValueError(f"Could not find or create result_file from record: {record}")
+
+    # Add to upload manager based on type
+    if record['type'].lower() == 'local':
+        _add_record_to_upload_manager_local_file(record, objects["result_file"], upload_manager)
+    elif record['type'].lower() == 's3':
+        _add_record_to_upload_manager_s3_file(record, objects["result_file"], upload_manager)
+    else:
+        raise ValueError(f"Unknown file type: {record['type']}")
+
+    return objects
+
+
+REQUIRED_COLUMNS = [
+    'organization', 'project', 'sample', 'folder', 
+    'filename', 'path', 'type', 'endpoint_url'
+]
+
+
+@click.command('from-config')
+@use_common_state
+@click.option('--cores', default=1, help='Number of uploads to run in parallel')
+@click.option('--sep', default=',', help='Separator character for the CSV file')
+@overwrite_option
+@yes_option
+@click.argument('config_file', type=click.Path(exists=True))
+def cli_upload_from_config(state, cores, sep, overwrite, yes, config_file):
+    """Upload files to GeoSeeq based on a configuration CSV file.
+
+    \b
+    The CSV file must have the following columns:
+    - organization: Organization name, GRN, or UUID (optional if project/sample/folder specified by GRN/UUID)
+    - project: Project name, GRN, or UUID (optional if sample/folder specified by GRN/UUID)
+    - sample: Sample name, GRN, or UUID (optional, also optional if folder specified by GRN/UUID)
+    - folder: Folder name, GRN, or UUID
+    - filename: Name to give the file on GeoSeeq
+    - path: Path to local file or S3 URL
+    - type: Either "local" or "s3"
+    - endpoint_url: S3 endpoint URL (required for S3 files)
+
+    \b
+    When using GRNs or UUIDs, you can omit the parent object IDs. For example:
+    - If folder is a GRN/UUID, organization/project/sample can be blank
+    - If sample is a GRN/UUID, organization/project can be blank
+    - If project is a GRN/UUID, organization can be blank
+
+    \b
+    Example config.csv:
+    organization,project,sample,folder,filename,path,type,endpoint_url
+    MyOrg,MyProject,Sample1,reads,file1.fastq,/path/to/file1.fastq,local,
+    ,grn:project:uuid,Sample2,reads,file2.fastq,/path/to/file2.fastq,local,
+    ,,grn:sample:uuid,reads,file3.fastq,/path/to/file3.fastq,local,
+    ,,,grn:folder:uuid,file4.fastq,s3://bucket/file4.fastq,s3,https://s3.amazonaws.com
+
+    \b
+    Example with tab separator:
+    $ geoseeq upload advanced from-config --sep $'\t' config.tsv
+    """
+    knex = state.get_knex()
+    
+    # Read and validate config file
+    df = pd.read_csv(config_file, sep=sep)
+    missing_cols = set(REQUIRED_COLUMNS) - set(df.columns)
+    if missing_cols:
+        raise click.UsageError(f"Config file missing required columns: {missing_cols}")
+    
+    # Create upload manager
+    upload_manager = GeoSeeqUploadManager(
+        n_parallel_uploads=cores,
+        progress_tracker_factory=PBarManager().get_new_bar,
+        log_level=state.log_level,
+        overwrite=overwrite,
+        use_cache=state.use_cache,
+    )
+    
+    # Process records and add to upload manager
+    objects_by_record = {}  # Store objects for human readable paths
+    for _, record in df.iterrows():
+        objects = _upload_one_record(knex, record, overwrite, upload_manager)
+        objects_by_record[record['path']] = objects
+    
+    # Show preview with both technical and human readable paths
+    click.echo(upload_manager.get_preview_string(), err=True)
+    
+    if not yes:
+        click.confirm('Do you want to proceed with these uploads?', abort=True)
+    
+    # Perform uploads
+    upload_manager.upload_files()

geoseeq/cli/upload/upload_reads.py

@@ -2,7 +2,7 @@ import logging
 import click
 import requests
 from os.path import basename
-
+import pandas as pd
 from multiprocessing import Pool, current_process
 
 from geoseeq.cli.constants import *
@@ -67,8 +67,12 @@ def _get_regex(knex, filepaths, module_name, lib, regex):
     return regex
 
 
-def _group_files(knex, filepaths, module_name, regex, yes):
+def _group_files(knex, filepaths, module_name, regex, yes, name_map):
     """Group the files into samples, confirm, and return the groups."""
+    if name_map is not None:
+        name_map_filename, cur_col, new_col = name_map
+        name_map = pd.read_csv(name_map_filename)[[cur_col, new_col]]
+        name_map = name_map.set_index(cur_col).to_dict()
     seq_length, seq_type = module_name.split('::')[:2]
     groups = knex.post('bulk_upload/group_files', json={
         'filenames': list(filepaths.keys()),
@@ -76,7 +80,11 @@ def _group_files(knex, filepaths, module_name, regex, yes):
         'regex': regex
     })
     for group in groups:
-        click.echo(f'sample_name: {group["sample_name"]}', err=True)
+        sample_name = group["sample_name"]
+        if name_map:
+            sample_name = name_map.get(sample_name, sample_name)
+            group["sample_name"] = sample_name
+        click.echo(f'sample_name: {sample_name}', err=True)
         click.echo(f'  module_name: {module_name}', err=True)
         for field_name, filename in group['fields'].items():
             path = filepaths[filename]
@@ -173,10 +181,11 @@ def flatten_list_of_bams(filepaths):
 @private_option
 @link_option
 @no_new_versions_option
+@click.option('--name-map', default=None, nargs=3, help="A file to use for converting names. Takes three arguments: a file name, a column name for current names, and a column name for new names.")
 @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, no_new_versions, module_name, project_id, fastq_files):
+def cli_upload_reads_wizard(state, cores, overwrite, yes, regex, private, link_type, no_new_versions, name_map, module_name, project_id, fastq_files):
     """Upload fastq read files to GeoSeeq.
 
     This command automatically groups files by their sample name, lane number

geoseeq/constants.py

@@ -1,5 +1,6 @@
 from os import environ
 from os.path import join
+from typing import Literal
 
 FIVE_MB = 5 * (1024 ** 2)
 FASTQ_MODULE_NAMES = [
@@ -13,4 +14,6 @@ DEFAULT_ENDPOINT = "https://backend.geoseeq.com"
 
 CONFIG_FOLDER = environ.get("XDG_CONFIG_HOME", join(environ["HOME"], ".config"))
 CONFIG_DIR = environ.get("GEOSEEQ_CONFIG_DIR", join(CONFIG_FOLDER, "geoseeq"))
-PROFILES_PATH = join(CONFIG_DIR, "profiles.json")
+PROFILES_PATH = join(CONFIG_DIR, "profiles.json")
+
+OBJECT_TYPE_STR = Literal['org', 'project', 'sample', 'sample_result_folder', 'project_result_folder', 'sample_result_file', 'project_result_file']

geoseeq/id_constructors/from_ids.py

@@ -22,51 +22,51 @@ from .from_names import (
     project_result_file_from_name,
     result_file_from_name,
 )
-from .utils import is_grn_or_uuid, is_name
+from .utils import is_grn_or_uuid, is_name, is_abs_name
 from geoseeq.knex import with_knex
-
+from geoseeq.constants import OBJECT_TYPE_STR
 logger = logging.getLogger("geoseeq_api")  # Same name as calling module
 
 
-def _generic_from_id(knex, id, from_uuid_func, from_name_func):
+def _generic_from_id(knex, id, from_uuid_func, from_name_func, object_type_str: OBJECT_TYPE_STR):
     """Return the object which the id points to."""
     logger.debug(f'Getting object from id: {id}, knex: {knex}, from_uuid_func: {from_uuid_func}, from_name_func: {from_name_func}')
     if is_grn_or_uuid(id):
         id = id.split(':')[-1]  # if this is a GRN, get the UUID. Won't hurt if it's already a UUID.
         return from_uuid_func(knex, id)
-    if is_name(id):
+    if is_abs_name(id, object_type_str):
         return from_name_func(knex, id)
-    raise ValueError(f'"{id}" is not a GRN, UUID, or name')
+    raise ValueError(f'"{id}" is not a GRN, UUID, or absolute name for {object_type_str}')
 
 
 @with_knex
 def org_from_id(knex, id):
     """Return the organization object which the id points to."""
-    return _generic_from_id(knex, id, org_from_uuid, org_from_name)
+    return _generic_from_id(knex, id, org_from_uuid, org_from_name, 'org')
 
 
 @with_knex
 def project_from_id(knex, id):
     """Return the project object which the id points to."""
-    return _generic_from_id(knex, id, project_from_uuid, project_from_name)
+    return _generic_from_id(knex, id, project_from_uuid, project_from_name, 'project')
 
 
 @with_knex
 def sample_from_id(knex, id):
     """Return the sample object which the id points to."""
-    return _generic_from_id(knex, id, sample_from_uuid, sample_from_name)
+    return _generic_from_id(knex, id, sample_from_uuid, sample_from_name, 'sample')
 
 
 @with_knex
 def sample_result_folder_from_id(knex, id):
     """Return the sample result folder object which the id points to."""
-    return _generic_from_id(knex, id, sample_result_folder_from_uuid, sample_result_folder_from_name)
+    return _generic_from_id(knex, id, sample_result_folder_from_uuid, sample_result_folder_from_name, 'sample_result_folder')
 
 
 @with_knex
 def project_result_folder_from_id(knex, id):
     """Return the project result folder object which the id points to."""
-    return _generic_from_id(knex, id, project_result_folder_from_uuid, project_result_folder_from_name)
+    return _generic_from_id(knex, id, project_result_folder_from_uuid, project_result_folder_from_name, 'project_result_folder')
 
 
 @with_knex
@@ -75,19 +75,22 @@ def result_folder_from_id(knex, id):
     
     Guess the result folder is a sample result folder. If not, try a project result folder.
     """
-    return _generic_from_id(knex, id, result_folder_from_uuid, result_folder_from_name)
+    try:
+        return _generic_from_id(knex, id, result_folder_from_uuid, result_folder_from_name, 'sample_result_folder')
+    except ValueError:
+        return _generic_from_id(knex, id, result_folder_from_uuid, result_folder_from_name, 'project_result_folder')
 
 
 @with_knex
 def sample_result_file_from_id(knex, id):
     """Return the sample result file object which the id points to."""
-    return _generic_from_id(knex, id, sample_result_file_from_uuid, sample_result_file_from_name)
+    return _generic_from_id(knex, id, sample_result_file_from_uuid, sample_result_file_from_name, 'sample_result_file')
 
 
 @with_knex
 def project_result_file_from_id(knex, id):
     """Return the project result file object which the id points to."""
-    return _generic_from_id(knex, id, project_result_file_from_uuid, project_result_file_from_name)
+    return _generic_from_id(knex, id, project_result_file_from_uuid, project_result_file_from_name, 'project_result_file')
 
 
 @with_knex
@@ -96,7 +99,10 @@ def result_file_from_id(knex, id):
     
     Guess the result file is a sample result file. If not, try a project result file.
     """
-    return _generic_from_id(knex, id, result_file_from_uuid, result_file_from_name)
+    try:
+        return _generic_from_id(knex, id, result_file_from_uuid, result_file_from_name, 'sample_result_file')
+    except ValueError:
+        return _generic_from_id(knex, id, result_file_from_uuid, result_file_from_name, 'project_result_file')
 
 
 

geoseeq/id_constructors/utils.py

@@ -1,4 +1,5 @@
 import uuid
+from geoseeq.constants import OBJECT_TYPE_STR
 
 
 def is_grn(el):
@@ -29,6 +30,22 @@ def is_name(el):
     return False
 
 
+def is_abs_name(el, object_type_str: OBJECT_TYPE_STR) -> bool:
+    """Return True if `el` is an absolute name for the given object type."""
+    if is_grn_or_uuid(el):
+        return False
+    n_required_slashes = {
+        'org': 0,
+        'project': 1,
+        'sample': 2,
+        'sample_result_folder': 3,
+        'sample_result_file': 4,
+        'project_result_folder': 2,
+        'project_result_file': 3,
+    }[object_type_str]
+    return el.count('/') == n_required_slashes
+
+
 def is_grn_or_uuid(el):
     """Return True if `el` is a GRN or a UUID"""
     return is_grn(el) or is_uuid(el)

geoseeq/result/result_file.py

@@ -18,9 +18,10 @@ from geoseeq.knex import GeoseeqOtherError
 from .utils import *
 from .file_upload import ResultFileUpload
 from .file_download import ResultFileDownload
+from .smart_objects import ResultFileSmartObjects
 
 
-class ResultFile(RemoteObject, ResultFileUpload, ResultFileDownload):
+class ResultFile(RemoteObject, ResultFileUpload, ResultFileDownload, ResultFileSmartObjects):
     remote_fields = [
         "uuid",
         "created_at",

geoseeq/result/result_folder.py

@@ -225,6 +225,10 @@ class SampleResultFolder(ResultFolder, SampleBioInfoFolder):
 
     def get_fields(self, *args, **kwargs):
         return self.get_result_files(*args, **kwargs)
+    
+    @property
+    def project(self):
+        return self.sample.project
 
     def __str__(self):
         return f"<Geoseeq::SampleResultFolder {self.module_name} {self.replicate} {self.uuid} />"
@@ -288,6 +292,10 @@ class ProjectResultFolder(ResultFolder):
 
     def get_fields(self, *args, **kwargs):
         return self.get_result_files(*args, **kwargs)
+    
+    @property
+    def project(self):
+        return self.grp
 
     def __str__(self):
         return f"<Geoseeq::ProjectResultFolder {self.module_name} {self.replicate} {self.uuid} />"

geoseeq/result/smart_objects.py

@@ -0,0 +1,40 @@
+
+
+class ResultFileSmartObjects:
+    """Interface for ResultFiles to easily create and use smart objects like tables"""
+
+    def is_smart_table(self):
+        """Return True iff this ResultFile is a smart table."""
+        return self.stored_data.get("__type__") == "smartTable"
+
+    def as_smart_table(self, description=""):
+        """Return a SmartTable object for this ResultFile. Will create a new SmartTable if it doesn't exist yet."""
+        if self.exists() and not self.is_smart_table():
+            raise ValueError("ResultFile exists but is not a smart table")
+        from geoseeq.smart_table import SmartTable  # import here to avoid circular import
+        if self.exists():  # already a smart table
+            smart_table = SmartTable(self.knex, self.name, connected_file_id=self.uuid).get()
+        else:  # create a new smart table
+            smart_table = SmartTable(self.knex, self.name, connected_file_id=self.uuid, description=description)
+            smart_table.create(self.parent)
+        return smart_table
+
+    def is_smart_tree(self):
+        """Return True iff this ResultFile is a smart tree."""
+        return self.stored_data.get("__type__") == "phylodynamics"
+    
+    def as_smart_tree(self, newick_str=None, sample_name_id_map=None):
+        """Return a SmartTree object for this ResultFile. Will create a new SmartTree if it doesn't exist yet."""
+        if self.exists() and not self.is_smart_tree():
+            raise ValueError("ResultFile exists but is not a smart tree")
+        from geoseeq.smart_tree import SmartTree  # import here to avoid circular import
+        if self.exists(): # already a smart tree
+            smart_tree = SmartTree.from_blob(self, self.stored_data)
+        else:  # create a new smart tree
+            smart_tree = SmartTree(self)
+            if newick_str is not None:
+                smart_tree.create_from_newick(newick_str, sample_name_id_map)
+                smart_tree.create()
+        return smart_tree
+
+        

geoseeq/smart_table.py

@@ -53,11 +53,12 @@ class SmartTable(RemoteObject):
 
         without_default_columns: if False the server creates 3 example columns.
         """
-
+        if description:
+            self.description = description
         data = {
             "name": self.name,
             "folder_id": result_folder.uuid,
-            "description": description,
+            "description": self.description,
         }
         url = f"table?without_default_columns={without_default_columns}"
         blob = self.knex.post(url, json=data)

geoseeq/smart_tree.py

@@ -0,0 +1,57 @@
+class SmartTree:
+
+    def __init__(self, result_file):
+        self.result_file = result_file  # the result file that contains the smart tree
+        self.content = None
+
+    def create_from_newick(self, newick_str, sample_name_id_map=None):
+        self.content = {
+            "__type__": "phylodynamics",
+            "version": "1.0",
+            "tree": {
+                "value": newick_str,
+                "kind": "newick",
+                "sampleNameIdMap": {} if sample_name_id_map is None else sample_name_id_map,
+            }
+        }
+        return self
+    
+    def add_sample_to_map(self, sample):
+        if self.content is None:
+            raise ValueError("Must create tree before adding samples")
+        self.content["tree"]["sampleNameIdMap"][sample.name] = sample.uuid
+        return self
+    
+    def add_all_samples_to_map(self, project):
+        for sample in project.get_samples():
+            self.add_sample_to_map(sample)
+
+    @classmethod
+    def from_blob(cls, result_file, blob):
+        smart_tree = cls(result_file)
+        smart_tree.content = blob
+        return smart_tree
+    
+    def save(self):
+        if self.content is None:
+            raise ValueError("Must create tree before saving")
+        self.result_file.upload_json(self.content)
+        return self
+    
+    def idem(self):
+        if self.content is None:
+            raise ValueError("Must create tree before saving")
+        self.result_file.upload_json(self.content)
+        return self
+    
+    def get(self):
+        self.result_file.get()
+        self.content = self.result_file.stored_data
+        return self
+    
+    def create(self):
+        if self.content is None:
+            raise ValueError("Must create tree before saving")
+        self.result_file.upload_json(self.content)
+        return self
+        

geoseeq/upload_download_manager.py

@@ -77,8 +77,17 @@ class GeoSeeqUploadManager:
         self.chunk_size_mb = chunk_size_mb
         self.use_atomic_upload = use_atomic_upload
 
-    def add_result_file(self, result_file, local_path):
-        self._result_files.append((result_file, local_path))
+    def add_result_file(self, result_file, local_path, link_type=None):
+        """Add a result file to be uploaded or linked.
+        
+        Args:
+            result_file: A ResultFile object to upload to or link
+            local_path: Path to local file or URL to link
+            link_type: One of 'upload', 's3', 'ftp', 'azure', 'sra', 'http'
+        """
+        if link_type is None:
+            link_type = self.link_type
+        self._result_files.append((result_file, local_path, link_type))
 
     def add_local_file_to_result_folder(self, result_folder, local_path, geoseeq_file_name=None):
         if not geoseeq_file_name:
@@ -96,19 +105,20 @@ class GeoSeeqUploadManager:
 
     def get_preview_string(self):
         out = ["Upload Preview:"]
-        for result_file, local_path in self._result_files:
-            out.append(f"{local_path} -> {result_file}")
+        for result_file, local_path, link_type in self._result_files:
+            action = "link" if link_type != 'upload' else "upload"
+            out.append(f"{local_path} -> {result_file} ({action})")
         return "\n".join(out)
 
     def upload_files(self):
         upload_args = [(
                 result_file, local_path,
                 self.session, self.progress_tracker_factory(local_path),
-                self.link_type, self.overwrite, self.log_level,
+                link_type, self.overwrite, self.log_level,
                 self.n_parallel_uploads > 1, self.use_cache, self.no_new_versions,
                 self.threads_per_upload, self.num_retries, self.ignore_errors,
                 self.chunk_size_mb, self.use_atomic_upload
-            ) for result_file, local_path in self._result_files
+            ) for result_file, local_path, link_type in self._result_files
         ]
         out = []
         if self.n_parallel_uploads == 1:

{geoseeq-0.6.14.dev7.dist-info → geoseeq-0.7.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.4
 Name: geoseeq
-Version: 0.6.14.dev7
+Version: 0.7.0
 Summary: GeoSeeq command line tools and python API
 Author: David C. Danko
 Author-email: "David C. Danko" <dcdanko@biotia.io>
@@ -18,6 +18,7 @@ Requires-Dist: pandas
 Requires-Dist: biopython
 Requires-Dist: tqdm
 Dynamic: author
+Dynamic: license-file
 
 # Geoseeq API Client
 

{geoseeq-0.6.14.dev7.dist-info → geoseeq-0.7.0.dist-info}/RECORD

@@ -2,7 +2,7 @@ geoseeq/__init__.py,sha256=YqYqeHbqjgWI5OBIxkPXNvLISOjVWaNwVFy6AGJ7uwc,982
 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=g9YEczT2ssteTrdgFrRimXo5I0C8-MqftoG-hE3KAkg,489
+geoseeq/constants.py,sha256=KmWH3gwHLd3toTWLWSViV_1eAYzPOPZ5bv6INpgwzQI,665
 geoseeq/file_system_cache.py,sha256=HzVZWtwLD2fjWWSo_UfWmGeBltm9He4lP_OqzKwNGWg,4138
 geoseeq/knex.py,sha256=zcjafsmUn9SC3LlRnvvaXpr-pHYZ0IXk7LpzuUoE3MI,8312
 geoseeq/organization.py,sha256=bJkYL8_D-k6IYAaii2ZbxjwYnXy6lvu6iLXscxKlA3w,2542
@@ -11,8 +11,9 @@ geoseeq/project.py,sha256=Ba3yvrSyPVnEM5_VGSQu6YPH3CTwE3fHs4JjAepEK7g,13880
 geoseeq/remote_object.py,sha256=GYN6PKU7Zz3htIdpFjfZiFejzGqqJHbJyKlefM1Eixk,7151
 geoseeq/sample.py,sha256=HAfMiDPHp1UJgIA2lI6oGnNit4YKyj7nx9X07CCN98U,8316
 geoseeq/search.py,sha256=gawad6Cx5FxJBPlYkXWb-UKAO-UC0_yhvyU9Ca1kaNI,3388
-geoseeq/smart_table.py,sha256=X1y9nBr8BkMNcBqdaiXtlqLCTfgc7lvdjSlAGxppvLo,6098
-geoseeq/upload_download_manager.py,sha256=eNR_Lw6JJstZ2X1Si43MM-0JZGv9GT-4fDjykaoH38A,9368
+geoseeq/smart_table.py,sha256=rihMsFUIn-vn4w6ukVZTHI9bjDSEr8xHExBfX8mwCHM,6169
+geoseeq/smart_tree.py,sha256=bSjDlwmOuNXutYJhytA1RovwRCHV6ZxXXJPiIGFhPaA,1825
+geoseeq/upload_download_manager.py,sha256=DxzwhkpuzuXTQt28ORULspDi2rzXqqIHqROEAs99SxQ,9849
 geoseeq/user.py,sha256=tol8i1UGLRrbMw5jeJDnna1ikRgrCDd50Jxz0a1lSgg,690
 geoseeq/utils.py,sha256=ZXpWb2MetUIeLrExiXb7IaOXYrW1pvrdP3o0KWzbwCs,4035
 geoseeq/work_orders.py,sha256=5uLVVfdKE8qh4gGaHkdBpXJGRTujuSg59knWCqEET4A,8071
@@ -20,11 +21,11 @@ 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=cOLIVzAfxpOJnSd4WKWR-HCkOrKjPMbgYaNXDSd9i90,21305
+geoseeq/cli/download.py,sha256=Znjuc9IFOcIa5_Od9mFXHJdYAJtgw9Bc_wPPcOVXn7s,21298
 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=guwmM9RE-vVZcu7_rMyOLiet-Ua1Jw_VP0mk62cT4eQ,3987
+geoseeq/cli/main.py,sha256=pkpjc_5dZ8T0F92777OKRXT_EFoHelfw4DhKjEcXqIE,3982
 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
@@ -40,10 +41,10 @@ geoseeq/cli/shared_params/config.py,sha256=HQ0xQh_jdt3EKI5VXYqQXzo-s8Rm6YlziMyVX
 geoseeq/cli/shared_params/id_handlers.py,sha256=KtzflnplYVkXsyqI5Ej6r-_BwQnuXVHPr7JcYumTKNc,10700
 geoseeq/cli/shared_params/obj_getters.py,sha256=ZSkt6LnDkVFlNVYKgLrjzg60-6BthZMr3eeD3HNqzac,2741
 geoseeq/cli/shared_params/opts_and_args.py,sha256=_DcJ-TqgrbBaeDd-kuHEx2gLZPQN6EHZYWh8Ag-d8Vg,2091
-geoseeq/cli/upload/__init__.py,sha256=3C9_S9t7chmYU-2ot89NV03x-EtmsjibulErKaU9w1k,627
-geoseeq/cli/upload/upload.py,sha256=JZkhe1q3KOp7-tKyzwi860TQhZoNDnZs4yB2PJhOjl0,10081
-geoseeq/cli/upload/upload_advanced.py,sha256=tXP7SWaLbLay5FVtqOoj-oToBM1ha2NWzywMnE0Foeo,3434
-geoseeq/cli/upload/upload_reads.py,sha256=IlO-9b8dHbA-fx8FK_R0BwyoRgtqQUoowgXJcyx_O3k,11018
+geoseeq/cli/upload/__init__.py,sha256=tRrljqG7uBM6zVBKpDw1zevHz6Vy418HuIYZKwByyWg,862
+geoseeq/cli/upload/upload.py,sha256=U9w18sG6dZ0uJoAMcQDLdrEtgmRaSaPS7-8KM5LkVd4,14220
+geoseeq/cli/upload/upload_advanced.py,sha256=1pjtbe8EidIo4eR1_oVD00rPCTxYj8CRqB8D9buA62M,12432
+geoseeq/cli/upload/upload_reads.py,sha256=m9ujcjSYOG88LhxMj1vEt8v7FpuhRciL4ntE6oNlVuM,11632
 geoseeq/contrib/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 geoseeq/contrib/ncbi/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 geoseeq/contrib/ncbi/api.py,sha256=WQeLoGA_-Zha-QeSO8_i7HpvXyD8UkV0qc5okm11KiA,1056
@@ -52,11 +53,11 @@ geoseeq/contrib/ncbi/cli.py,sha256=j9zEcaZPTryK3a4xluRxigcJKDhRpRxbp3KZSx-Bfhk,2
 geoseeq/contrib/ncbi/setup_logging.py,sha256=Tp1bY1U0f-o739aHpvVYriG2qdd1lFvCYBXZeXQgt-w,175
 geoseeq/id_constructors/__init__.py,sha256=w5E0PNQ9UuAxBeZbDI7KBnUoERd85gGz3nScz45bd2o,126
 geoseeq/id_constructors/from_blobs.py,sha256=wZp6P6m7X0lhZAQxorHKJQn1CTBvKyfANzu2VFexp44,6315
-geoseeq/id_constructors/from_ids.py,sha256=vwv_R1uZ5hdeUvpKmNLOTz02PCvwfd6yQ7aAXqi0hZo,3372
+geoseeq/id_constructors/from_ids.py,sha256=SRylajToFdnRON5EhPIvcAi64XuIzseaasZlOooqTpg,3984
 geoseeq/id_constructors/from_names.py,sha256=RqgFjDsAwQcidMkZwX7oB00OvBAKTiilHYetTPogJ40,4174
 geoseeq/id_constructors/from_uuids.py,sha256=loqTeZqDZuvfVVHpS7m-Y_spNaY3h9VRjt0wDiMxHbs,3522
 geoseeq/id_constructors/resolvers.py,sha256=8hp5xJSCoZrAXtMT54Hp4okt63l909XqJU3IQx-VCgc,2676
-geoseeq/id_constructors/utils.py,sha256=CKlZHGMiqi1b6r1KtgD3czSAomH99Gdfx5ziqaahz-0,723
+geoseeq/id_constructors/utils.py,sha256=Up_vXoZDwa2mJGllJ6CmcbaOGUvA-sM0gMmFnXK8wn4,1255
 geoseeq/plotting/__init__.py,sha256=RkGoXxgu7jEfK0B7NmdalPS2AbU7I7dZwDbi4rn9CKM,154
 geoseeq/plotting/constants.py,sha256=CGUlm8WAFG3YRKdicc9Rcy5hFxUdUm2RgK0iXZWLuX8,285
 geoseeq/plotting/highcharts.py,sha256=AGzdW4VSUsL_rfEI-RiVbbtaTLFMARvzLzVUDrKTlnU,4096
@@ -70,10 +71,11 @@ geoseeq/result/bioinfo.py,sha256=QQtbyogrdro9avJSN0713sxLVnVeA24mFw3hWtKDKyw,178
 geoseeq/result/file_chunker.py,sha256=bXq1csuRtqMB5sbH-AfWo6gdPwrivv5DJPuHVj-h08w,1758
 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
+geoseeq/result/result_file.py,sha256=mkFh2DpKO1-kEAARCMYjkc7TmkJh41azyauGIHl_VZo,9173
+geoseeq/result/result_folder.py,sha256=iyO0hwZWokrH6oWhBgHlunWMpCMpejKb8v2sHFhecws,11283
 geoseeq/result/resumable_download_tracker.py,sha256=YEzqHBBnE7L3XokTvlTAhHZ8TcDTIE_pyTQ7YadOfbU,3667
 geoseeq/result/resumable_upload_tracker.py,sha256=2aI09gYz2yw63jEXqs8lmCRKQ79TIc3YuPETvP0Jeek,3811
+geoseeq/result/smart_objects.py,sha256=-krK9h9HcznIglf189uPghOidF_BzOeSxmBWxrtFFo8,1991
 geoseeq/result/utils.py,sha256=C-CxGzB3WddlnRiqFSkrY78I_m0yFgNqsTBRzGU-y8Q,2772
 geoseeq/vc/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 geoseeq/vc/checksum.py,sha256=y8rh1asUZNbE_NLiFO0-9hImLNiTOc2YXQBRKORWK7k,710
@@ -84,12 +86,12 @@ geoseeq/vc/vc_cache.py,sha256=P4LXTbq2zOIv1OhP7Iw5MmypR2vXuy29Pq5K6gRvi-M,730
 geoseeq/vc/vc_dir.py,sha256=A9CLTh2wWCRzZjiLyqXD1vhtsWZGD3OjaMT5KqlfAXI,457
 geoseeq/vc/vc_sample.py,sha256=qZeioWydXvfu4rGMs20nICfNcp46y_XkND-bHdV6P5M,3850
 geoseeq/vc/vc_stub.py,sha256=IQr8dI0zsWKVAeY_5ybDD6n49_3othcgfHS3P0O9tuY,3110
+geoseeq-0.7.0.dist-info/licenses/LICENSE,sha256=IuhIl1XCxXLPLJT_coN1CNqQU4Khlq7x4IdW7ioOJD8,1067
 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.dev7.dist-info/LICENSE,sha256=IuhIl1XCxXLPLJT_coN1CNqQU4Khlq7x4IdW7ioOJD8,1067
-geoseeq-0.6.14.dev7.dist-info/METADATA,sha256=q9ZgGURnAxFyUKqLaF6J5db1e9rrrqEcQbGhJ8MmFUc,4937
-geoseeq-0.6.14.dev7.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-geoseeq-0.6.14.dev7.dist-info/entry_points.txt,sha256=yF-6KDM8zXib4Al0qn49TX-qM7PUkWUIcYtsgt36rjM,45
-geoseeq-0.6.14.dev7.dist-info/top_level.txt,sha256=zZk7mmeaqAYqFJG8nq2DTgSQPbflRjJwkDIhNURPDEU,14
-geoseeq-0.6.14.dev7.dist-info/RECORD,,
+geoseeq-0.7.0.dist-info/METADATA,sha256=lEr_Lja53CYN53DTQTR3CKG3cqPdwgnuB03sbiwtzTs,4953
+geoseeq-0.7.0.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
+geoseeq-0.7.0.dist-info/entry_points.txt,sha256=yF-6KDM8zXib4Al0qn49TX-qM7PUkWUIcYtsgt36rjM,45
+geoseeq-0.7.0.dist-info/top_level.txt,sha256=zZk7mmeaqAYqFJG8nq2DTgSQPbflRjJwkDIhNURPDEU,14
+geoseeq-0.7.0.dist-info/RECORD,,

{geoseeq-0.6.14.dev7.dist-info → geoseeq-0.7.0.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (75.8.0)
+Generator: setuptools (78.1.0)
 Root-Is-Purelib: true
 Tag: py3-none-any