plastron-cli 4.6.2__tar.gz → 4.7.0__tar.gz

{plastron_cli-4.6.2 → plastron_cli-4.7.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: plastron-cli
-Version: 4.6.2
+Version: 4.7.0
 Summary: Plastron command line client
 Author-email: University of Maryland Libraries <lib-ssdr@umd.edu>, Josh Westgard <westgard@umd.edu>, Peter Eichman <peichman@umd.edu>, Mohamed Abdul Rasheed <mohideen@umd.edu>, Ben Wallberg <wallberg@umd.edu>, David Steelman <dsteelma@umd.edu>, Marc Andreu Grillo Aguilar <aguilarm@umd.edu>
 Requires-Python: >=3.10
@@ -221,6 +221,24 @@ optional arguments:
                         resource in the result list
 ```
 
+### Fix Page Order (fixpageorder)
+
+```text
+$ plastron fixpageorder --help
+usage: plastron fixpageorder [-h] [-f URIS_FILE] [-n] [uris ...]
+
+Fix the order of pages in an object, using the page titles as a guide
+
+positional arguments:
+  uris                  URIs of objects to fix
+
+options:
+  -h, --help            show this help message and exit
+  -f URIS_FILE, --uris-file URIS_FILE
+                        file containing URIs of objects to fix
+  -n, --dry-run         dry run; do not actually modify the pages
+```
+
 ### Image Size (imgsize)
 
 ```text
@@ -371,6 +389,26 @@ COMMANDS:
         CamelFcrepoSolrIndexingDestinations: direct:solr.Index,direct:solr.LegacyIndex
 ```
 
+### Replace (replace)
+
+```text
+$ plastron replace --help
+usage: plastron replace [-h] [--binary-file BINARY_FILE] [--mime-type MIME_TYPE] URI|PATH
+
+Replace a binary resource in the repository
+
+positional arguments:
+  URI|PATH              URI or repository path of the resource to replace
+
+options:
+  -h, --help            show this help message and exit
+  --binary-file BINARY_FILE
+                        local path to the binary file
+  --mime-type MIME_TYPE
+                        use this MIME type instead of auto-detecting based on the BINARY_FILE
+
+```
+
 ### Stub (stub)
 
 ```text

{plastron_cli-4.6.2 → plastron_cli-4.7.0}/README.md

@@ -199,6 +199,24 @@ optional arguments:
                         resource in the result list
 ```
 
+### Fix Page Order (fixpageorder)
+
+```text
+$ plastron fixpageorder --help
+usage: plastron fixpageorder [-h] [-f URIS_FILE] [-n] [uris ...]
+
+Fix the order of pages in an object, using the page titles as a guide
+
+positional arguments:
+  uris                  URIs of objects to fix
+
+options:
+  -h, --help            show this help message and exit
+  -f URIS_FILE, --uris-file URIS_FILE
+                        file containing URIs of objects to fix
+  -n, --dry-run         dry run; do not actually modify the pages
+```
+
 ### Image Size (imgsize)
 
 ```text
@@ -349,6 +367,26 @@ COMMANDS:
         CamelFcrepoSolrIndexingDestinations: direct:solr.Index,direct:solr.LegacyIndex
 ```
 
+### Replace (replace)
+
+```text
+$ plastron replace --help
+usage: plastron replace [-h] [--binary-file BINARY_FILE] [--mime-type MIME_TYPE] URI|PATH
+
+Replace a binary resource in the repository
+
+positional arguments:
+  URI|PATH              URI or repository path of the resource to replace
+
+options:
+  -h, --help            show this help message and exit
+  --binary-file BINARY_FILE
+                        local path to the binary file
+  --mime-type MIME_TYPE
+                        use this MIME type instead of auto-detecting based on the BINARY_FILE
+
+```
+
 ### Stub (stub)
 
 ```text

plastron_cli-4.7.0/VERSION

@@ -0,0 +1 @@
+4.7.0

plastron_cli-4.7.0/src/plastron/cli/commands/fixpageorder.py

@@ -0,0 +1,106 @@
+import logging
+import re
+from argparse import Namespace, FileType
+from collections.abc import Iterable
+
+from plastron.cli import get_uris
+from plastron.cli.commands import BaseCommand
+from plastron.models.ore import Proxy
+from plastron.repo import RepositoryResource
+from plastron.repo.pcdm import PCDMObjectResource
+
+logger = logging.getLogger(__name__)
+
+
+def configure_cli(subparsers):
+    parser = subparsers.add_parser(
+        name='fixpageorder',
+        description='Fix the order of pages in an object, using the page titles as a guide',
+    )
+    parser.add_argument(
+        '-f', '--uris-file',
+        action='store',
+        type=FileType(),
+        help='file containing URIs of objects to fix'
+    )
+    parser.add_argument(
+        '-n', '--dry-run',
+        action='store_true',
+        help='dry run; do not actually modify the pages',
+    )
+    parser.add_argument(
+        'uris',
+        nargs='*',
+        help='URIs of objects to fix',
+    )
+    parser.set_defaults(cmd_name='fixpageorder')
+
+
+class Command(BaseCommand):
+    def __call__(self, args: Namespace):
+        # mimicking a click.Context object to bridge between argparse and click commands
+        ctx = Namespace(obj=self.context)
+        return fix_page_order(ctx, uris=get_uris(args), dry_run=args.dry_run)
+
+
+def fix_page_order(ctx: Namespace, uris: Iterable[str], dry_run: bool = False):
+    page_pattern = re.compile(r'page (\d+)', re.IGNORECASE)
+    for uri in uris:
+        logger.info(f'Retrieving {uri}...')
+        resource = ctx.obj.repo.get_resource(uri, PCDMObjectResource)
+        proxy_by_page: dict[int, Proxy] = {}
+        proxy_by_uri: dict[str, Proxy] = {}
+        resource_by_uri: dict[str, RepositoryResource] = {}
+        logger.info('Checking proxies...')
+        for proxy in resource.get_proxies():
+            resource_by_uri[proxy.url] = proxy
+            proxy_obj = proxy.describe(Proxy)
+            title = str(proxy_obj.title)
+            if m := page_pattern.search(title):
+                proxy_by_page[int(m.group(1))] = proxy_obj
+                proxy_by_uri[proxy.url] = proxy_obj
+            else:
+                raise RuntimeError(f'Could not find page number in "{proxy_obj.title}"')
+
+        logger.info('Current order:')
+        for page_n, proxy_obj in proxy_by_page.items():
+            logger.info(f'{page_n:3d}: {proxy_obj.uri} ({proxy_obj.title})')
+
+        sorted_pages = [proxy_by_page[i] for i in sorted(proxy_by_page.keys())]
+
+        logger.info('Desired order:')
+        for page_n, proxy_obj in enumerate(sorted_pages, 1):
+            logger.info(f'{page_n:3d}: {proxy_obj.uri} ({proxy_obj.title})')
+
+        if list(proxy_by_page.values()) == sorted_pages:
+            logger.info(f'No changes to needed; skipping {uri}')
+            continue
+
+        logger.info(f'Modifications required for {uri}')
+
+        if dry_run:
+            logger.info(f'Dry run; skipping modifications to the proxies of {uri}')
+            continue
+
+        with ctx.obj.repo.transaction():
+            for page_n, proxy_obj in enumerate(sorted_pages, 1):
+                logger.info(f'Checking whether {proxy_obj.uri} ({proxy_obj.title}) needs an update...')
+
+                prev_proxy = proxy_by_page.get(page_n - 1, None)
+                next_proxy = proxy_by_page.get(page_n + 1, None)
+
+                if prev_proxy:
+                    proxy_obj.prev = prev_proxy
+                else:
+                    proxy_obj.prev.clear()
+
+                if next_proxy:
+                    proxy_obj.next = next_proxy
+                else:
+                    proxy_obj.next.clear()
+
+                if proxy_obj.has_changes:
+                    logger.info(f'Updating {proxy_obj.uri} ({proxy_obj.title})')
+                    resource_by_uri[proxy_obj.uri].update()
+                else:
+                    logger.info('No changes to {proxy_obj.uri} ({proxy_obj.title})')

{plastron_cli-4.6.2 → plastron_cli-4.7.0}/src/plastron/cli/commands/importcommand.py

@@ -1,15 +1,16 @@
 import csv
 import logging
-from argparse import FileType, ArgumentTypeError, Namespace
+from argparse import ArgumentTypeError, FileType, Namespace
 from typing import TextIO
 
 from plastron.cli.commands import BaseCommand
-from plastron.jobs.importjob.ndnp import NDNPBatch, write_import_csv
-from plastron.jobs.importjob import ImportConfig, ImportJob
 from plastron.jobs import Jobs
-from plastron.models import get_model_from_name, ModelClassNotFoundError
+from plastron.jobs.importjob import ImportConfig, ImportJob
+from plastron.jobs.importjob.ndnp import NDNPBatch, write_import_csv
 from plastron.utils import datetimestamp, uri_or_curie
 
+from plastron.models import ModelClassNotFoundError, get_model_from_name
+
 logger = logging.getLogger(__name__)
 
 
@@ -212,7 +213,7 @@ class Command(BaseCommand):
             if args.convert_from is not None:
                 if args.convert_from == 'ndnp':
                     params = dict(args.convert_params or [])
-                    batch = NDNPBatch(params['dir'], params['batch_file'])
+                    batch = NDNPBatch(**params)
                     logger.info(f'Converting NDNP batch at {batch.batch_file} to import job {args.job_id}')
                     job = jobs.create_job(job_class=ImportJob, config=ImportConfig(
                         job_id=args.job_id,

{plastron_cli-4.6.2 → plastron_cli-4.7.0}/src/plastron/cli/commands/load.py

@@ -9,6 +9,7 @@ from time import sleep
 
 from plastron.client import Client, ClientError
 from plastron.cli.commands import BaseCommand
+from plastron.client.transactions import transaction
 from plastron.repo import DataReadError
 from plastron.cli import ConfigError
 from plastron.jobs import ItemLog
@@ -247,7 +248,7 @@ def load_item(client: Client, batch_item, args, extra=None):
 
     if args.use_transactions:
         # open transaction
-        with client.transaction(keep_alive=90) as txn_client:  # type: TransactionClient
+        with transaction(client, keep_alive=90) as txn_client:  # type: TransactionClient
             # create item and its components
             try:
                 load_item_internal(txn_client, item, args, extra)

plastron_cli-4.7.0/src/plastron/cli/commands/replace.py

@@ -0,0 +1,65 @@
+import logging
+from argparse import Namespace
+
+from rdflib import Literal
+
+from plastron.cli.commands import BaseCommand
+from plastron.files import LocalFileSource
+from plastron.models.pcdm import PCDMFile
+from plastron.repo import Repository, BinaryResource, RepositoryError
+
+logger = logging.getLogger(__name__)
+
+
+def configure_cli(subparsers):
+    parser = subparsers.add_parser(
+        name='replace',
+        description='Replace a binary resource in the repository'
+    )
+    parser.add_argument(
+        '--binary-file',
+        help='local path to the binary file',
+    )
+    parser.add_argument(
+        '--mime-type',
+        help='use this MIME type instead of auto-detecting based on the BINARY_FILE',
+    )
+    parser.add_argument(
+        'location',
+        help='URI or repository path of the resource to replace',
+        metavar='URI|PATH',
+    )
+    parser.set_defaults(cmd_name='replace')
+
+
+class Command(BaseCommand):
+    def __call__(self, args: Namespace):
+        # mimicking a click.Context object to bridge between argparse and click commands
+        ctx = Namespace(obj=self.context)
+        return replace(ctx, location=args.location, binary_filename=args.binary_file, mime_type=args.mime_type)
+
+
+def replace(ctx, location: str, binary_filename: str, mime_type: str = None):
+    repo: Repository = ctx.obj.repo
+    try:
+        resource = repo.get_resource(location, BinaryResource).read()
+    except RepositoryError as e:
+        raise RuntimeError(str(e)) from e
+
+    if not resource.exists:
+        raise RuntimeError(f'Resource {resource.url} not found')
+
+    source = LocalFileSource(binary_filename)
+
+    try:
+        with repo.transaction():
+            resource.update_binary(source, mime_type=mime_type)
+            logger.info(f'Updating metadata for {resource.url}')
+            file = resource.describe(PCDMFile)
+            file.title = Literal(source.filename)
+            resource.update()
+
+        print(resource.url)
+
+    except RepositoryError as e:
+        raise RuntimeError(str(e)) from e

{plastron_cli-4.6.2 → plastron_cli-4.7.0}/src/plastron/cli/commands/update.py

@@ -2,10 +2,11 @@ import logging
 from argparse import FileType, Namespace
 
 from plastron.cli.commands import BaseCommand
+from plastron.jobs import ItemLog
 from plastron.jobs.updatejob import UpdateJob
-from plastron.models import get_model_from_name
 from plastron.utils import parse_predicate_list
-from plastron.jobs import ItemLog
+
+from plastron.models import get_model_from_name
 
 logger = logging.getLogger(__name__)
 
@@ -88,9 +89,22 @@ class Command(BaseCommand):
 
         traverse = parse_predicate_list(args.recursive) if args.recursive is not None else []
 
+        uris = set()
+
+        if args.file:
+            try:
+                with open(args.file, 'r') as f:
+                    file_uris = [line.strip() for line in f if line.strip()]
+                uris = uris | set(file_uris)
+            except FileNotFoundError:
+                raise RuntimeError(f"File {args.file} not found")
+
+        if args.uris:
+            uris = uris | set(args.uris)
+
         update_job = UpdateJob(
             repo=self.context.repo,
-            uris=args.uris,
+            uris=uris,
             sparql_update=sparql_update,
             model_class=model_class,
             traverse=traverse,

{plastron_cli-4.6.2 → plastron_cli-4.7.0}/src/plastron_cli.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: plastron-cli
-Version: 4.6.2
+Version: 4.7.0
 Summary: Plastron command line client
 Author-email: University of Maryland Libraries <lib-ssdr@umd.edu>, Josh Westgard <westgard@umd.edu>, Peter Eichman <peichman@umd.edu>, Mohamed Abdul Rasheed <mohideen@umd.edu>, Ben Wallberg <wallberg@umd.edu>, David Steelman <dsteelma@umd.edu>, Marc Andreu Grillo Aguilar <aguilarm@umd.edu>
 Requires-Python: >=3.10
@@ -221,6 +221,24 @@ optional arguments:
                         resource in the result list
 ```
 
+### Fix Page Order (fixpageorder)
+
+```text
+$ plastron fixpageorder --help
+usage: plastron fixpageorder [-h] [-f URIS_FILE] [-n] [uris ...]
+
+Fix the order of pages in an object, using the page titles as a guide
+
+positional arguments:
+  uris                  URIs of objects to fix
+
+options:
+  -h, --help            show this help message and exit
+  -f URIS_FILE, --uris-file URIS_FILE
+                        file containing URIs of objects to fix
+  -n, --dry-run         dry run; do not actually modify the pages
+```
+
 ### Image Size (imgsize)
 
 ```text
@@ -371,6 +389,26 @@ COMMANDS:
         CamelFcrepoSolrIndexingDestinations: direct:solr.Index,direct:solr.LegacyIndex
 ```
 
+### Replace (replace)
+
+```text
+$ plastron replace --help
+usage: plastron replace [-h] [--binary-file BINARY_FILE] [--mime-type MIME_TYPE] URI|PATH
+
+Replace a binary resource in the repository
+
+positional arguments:
+  URI|PATH              URI or repository path of the resource to replace
+
+options:
+  -h, --help            show this help message and exit
+  --binary-file BINARY_FILE
+                        local path to the binary file
+  --mime-type MIME_TYPE
+                        use this MIME type instead of auto-detecting based on the BINARY_FILE
+
+```
+
 ### Stub (stub)
 
 ```text

{plastron_cli-4.6.2 → plastron_cli-4.7.0}/src/plastron_cli.egg-info/SOURCES.txt

@@ -10,6 +10,7 @@ src/plastron/cli/commands/echo.py
 src/plastron/cli/commands/export.py
 src/plastron/cli/commands/extractocr.py
 src/plastron/cli/commands/find.py
+src/plastron/cli/commands/fixpageorder.py
 src/plastron/cli/commands/imgsize.py
 src/plastron/cli/commands/importcommand.py
 src/plastron/cli/commands/list.py
@@ -17,6 +18,7 @@ src/plastron/cli/commands/load.py
 src/plastron/cli/commands/ping.py
 src/plastron/cli/commands/publish.py
 src/plastron/cli/commands/reindex.py
+src/plastron/cli/commands/replace.py
 src/plastron/cli/commands/set.py
 src/plastron/cli/commands/stub.py
 src/plastron/cli/commands/unpublish.py

plastron_cli-4.6.2/VERSION

@@ -1 +0,0 @@
-4.6.2