cjm-transcript-source-select 0.0.1__tar.gz → 0.0.2__tar.gz

{cjm_transcript_source_select-0.0.1/cjm_transcript_source_select.egg-info → cjm_transcript_source_select-0.0.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: cjm-transcript-source-select
-Version: 0.0.1
+Version: 0.0.2
 Summary: FastHTML source selection component for transcript decomposition workflows, with federated database browsing, drag-drop ordering, and keyboard navigation.
 Home-page: https://github.com/cj-mills/cjm-transcript-source-select
 Author: Christian J. Mills
@@ -105,51 +105,51 @@ graph LR
     components_local_files --> html_ids
     components_preview_panel --> html_ids
     components_selection_queue --> html_ids
-    components_source_browser --> services_source_utils
     components_source_browser --> utils
+    components_source_browser --> services_source_utils
     components_source_browser --> html_ids
-    components_step_renderer --> components_helpers
-    components_step_renderer --> components_source_browser
     components_step_renderer --> utils
+    components_step_renderer --> components_helpers
     components_step_renderer --> components_preview_panel
-    components_step_renderer --> html_ids
-    components_step_renderer --> components_local_files
     components_step_renderer --> components_selection_queue
+    components_step_renderer --> components_local_files
+    components_step_renderer --> components_source_browser
     components_step_renderer --> models
-    routes_core --> components_step_renderer
-    routes_core --> models
-    routes_core --> components_selection_queue
+    components_step_renderer --> html_ids
     routes_core --> components_source_browser
+    routes_core --> components_selection_queue
     routes_core --> services_source
-    routes_filtering --> routes_core
-    routes_filtering --> services_source_utils
-    routes_filtering --> models
+    routes_core --> models
+    routes_core --> components_step_renderer
     routes_filtering --> components_source_browser
+    routes_filtering --> services_source_utils
+    routes_filtering --> routes_core
     routes_filtering --> services_source
-    routes_init --> routes_queue
-    routes_init --> routes_local_files
-    routes_init --> models
+    routes_filtering --> models
     routes_init --> routes_filtering
-    routes_init --> routes_tabs
     routes_init --> services_source
+    routes_init --> models
+    routes_init --> routes_queue
+    routes_init --> routes_local_files
     routes_init --> routes_core
+    routes_init --> routes_tabs
     routes_local_files --> components_local_files
     routes_local_files --> routes_core
     routes_local_files --> services_source
     routes_local_files --> models
     routes_queue --> routes_core
-    routes_queue --> services_source_utils
+    routes_queue --> services_source
     routes_queue --> models
+    routes_queue --> services_source_utils
     routes_queue --> components_preview_panel
-    routes_queue --> services_source
-    routes_tabs --> routes_local_files
     routes_tabs --> components_local_files
-    routes_tabs --> services_source_utils
+    routes_tabs --> routes_local_files
     routes_tabs --> routes_core
     routes_tabs --> components_source_browser
-    routes_tabs --> models
     routes_tabs --> components_step_renderer
+    routes_tabs --> models
     routes_tabs --> services_source
+    routes_tabs --> services_source_utils
 ```
 
 *50 cross-module dependencies detected*
@@ -186,6 +186,16 @@ def _get_step_state(
     "Get the selection step state from the workflow state store."
 ```
 
+``` python
+def _check_duplicate_media_path(
+    source_service: SourceService,  # Source service for lookups
+    record_id: str,  # Candidate record ID
+    provider_id: str,  # Candidate provider ID
+    selected_sources: List[Dict[str, str]],  # Current selections
+) -> bool:  # True if adding would duplicate an audio file
+    "Check if adding a source would duplicate an already-selected audio file."
+```
+
 ``` python
 def _get_active_source_tab(
     state_store: WorkflowStateStore,  # The workflow state store
@@ -658,6 +668,7 @@ def _handle_selection_remove(
     request,  # FastHTML request object
     sess,  # FastHTML session object
     record_id: str,  # Job ID to remove
+    provider_id: str,  # Plugin name for the source
     urls: SelectionUrls,  # URL bundle for rendering
 ):  # Queue component with OOB stats, optionally with OOB source list
     "Remove a source from the selection queue."
@@ -698,7 +709,7 @@ def _handle_selection_select_all(
     grouping_mode: str,  # Current grouping mode: "media_path" or "batch_id"
     urls: SelectionUrls,  # URL bundle for rendering
 ):  # Queue component with OOB stats, optionally with OOB source list
-    "Select all transcriptions for a given group."
+    "Select all transcriptions for a given group, skipping duplicate audio sources."
 ```
 
 ``` python
@@ -1088,6 +1099,7 @@ from cjm_transcript_source_select.services.source_utils import (
     group_transcriptions,
     group_transcriptions_by_audio,
     is_source_selected,
+    get_selected_media_paths,
     filter_transcriptions,
     select_all_in_group,
     toggle_source_selection,
@@ -1133,9 +1145,18 @@ def group_transcriptions_by_audio(
 ``` python
 def is_source_selected(
     record_id: str,  # Job ID to check
+    provider_id: str,  # Provider ID to check
     selected_sources: List[Dict[str, str]]  # List of selected sources
 ) -> bool:  # True if source is selected
-    "Check if a source is in the selected list."
+    "Check if a source is in the selected list by (record_id, provider_id) pair."
+```
+
+``` python
+def get_selected_media_paths(
+    selected_sources: List[Dict[str, str]],  # Current selections (record_id, provider_id)
+    all_transcriptions: List[Dict[str, Any]],  # All available transcription records
+) -> Set[str]:  # Media paths already represented in selections
+    "Get the set of media_paths for currently selected sources."
 ```
 
 ``` python
@@ -1152,6 +1173,7 @@ def select_all_in_group(
     group_key: str,  # Group key to match against
     grouping_mode: str,  # Grouping mode: "media_path" or "batch_id"
     selected_sources: List[Dict[str, str]],  # Current selections
+    excluded_media_paths: Optional[Set[str]] = None,  # Media paths to skip (already selected)
 ) -> List[Dict[str, str]]:  # Updated selections with new items appended
     "Add all transcriptions matching a group key to the selection list, skipping duplicates."
 ```
@@ -1162,7 +1184,7 @@ def toggle_source_selection(
     provider_id: str,  # Plugin name for the source
     selected_sources: List[Dict[str, str]],  # Current selections
 ) -> List[Dict[str, str]]:  # Updated selections
-    "Toggle a source in or out of the selection list."
+    "Toggle a source in or out of the selection list by (record_id, provider_id) pair."
 ```
 
 ``` python

{cjm_transcript_source_select-0.0.1 → cjm_transcript_source_select-0.0.2}/README.md

@@ -62,51 +62,51 @@ graph LR
     components_local_files --> html_ids
     components_preview_panel --> html_ids
     components_selection_queue --> html_ids
-    components_source_browser --> services_source_utils
     components_source_browser --> utils
+    components_source_browser --> services_source_utils
     components_source_browser --> html_ids
-    components_step_renderer --> components_helpers
-    components_step_renderer --> components_source_browser
     components_step_renderer --> utils
+    components_step_renderer --> components_helpers
     components_step_renderer --> components_preview_panel
-    components_step_renderer --> html_ids
-    components_step_renderer --> components_local_files
     components_step_renderer --> components_selection_queue
+    components_step_renderer --> components_local_files
+    components_step_renderer --> components_source_browser
     components_step_renderer --> models
-    routes_core --> components_step_renderer
-    routes_core --> models
-    routes_core --> components_selection_queue
+    components_step_renderer --> html_ids
     routes_core --> components_source_browser
+    routes_core --> components_selection_queue
     routes_core --> services_source
-    routes_filtering --> routes_core
-    routes_filtering --> services_source_utils
-    routes_filtering --> models
+    routes_core --> models
+    routes_core --> components_step_renderer
     routes_filtering --> components_source_browser
+    routes_filtering --> services_source_utils
+    routes_filtering --> routes_core
     routes_filtering --> services_source
-    routes_init --> routes_queue
-    routes_init --> routes_local_files
-    routes_init --> models
+    routes_filtering --> models
     routes_init --> routes_filtering
-    routes_init --> routes_tabs
     routes_init --> services_source
+    routes_init --> models
+    routes_init --> routes_queue
+    routes_init --> routes_local_files
     routes_init --> routes_core
+    routes_init --> routes_tabs
     routes_local_files --> components_local_files
     routes_local_files --> routes_core
     routes_local_files --> services_source
     routes_local_files --> models
     routes_queue --> routes_core
-    routes_queue --> services_source_utils
+    routes_queue --> services_source
     routes_queue --> models
+    routes_queue --> services_source_utils
     routes_queue --> components_preview_panel
-    routes_queue --> services_source
-    routes_tabs --> routes_local_files
     routes_tabs --> components_local_files
-    routes_tabs --> services_source_utils
+    routes_tabs --> routes_local_files
     routes_tabs --> routes_core
     routes_tabs --> components_source_browser
-    routes_tabs --> models
     routes_tabs --> components_step_renderer
+    routes_tabs --> models
     routes_tabs --> services_source
+    routes_tabs --> services_source_utils
 ```
 
 *50 cross-module dependencies detected*
@@ -143,6 +143,16 @@ def _get_step_state(
     "Get the selection step state from the workflow state store."
 ```
 
+``` python
+def _check_duplicate_media_path(
+    source_service: SourceService,  # Source service for lookups
+    record_id: str,  # Candidate record ID
+    provider_id: str,  # Candidate provider ID
+    selected_sources: List[Dict[str, str]],  # Current selections
+) -> bool:  # True if adding would duplicate an audio file
+    "Check if adding a source would duplicate an already-selected audio file."
+```
+
 ``` python
 def _get_active_source_tab(
     state_store: WorkflowStateStore,  # The workflow state store
@@ -615,6 +625,7 @@ def _handle_selection_remove(
     request,  # FastHTML request object
     sess,  # FastHTML session object
     record_id: str,  # Job ID to remove
+    provider_id: str,  # Plugin name for the source
     urls: SelectionUrls,  # URL bundle for rendering
 ):  # Queue component with OOB stats, optionally with OOB source list
     "Remove a source from the selection queue."
@@ -655,7 +666,7 @@ def _handle_selection_select_all(
     grouping_mode: str,  # Current grouping mode: "media_path" or "batch_id"
     urls: SelectionUrls,  # URL bundle for rendering
 ):  # Queue component with OOB stats, optionally with OOB source list
-    "Select all transcriptions for a given group."
+    "Select all transcriptions for a given group, skipping duplicate audio sources."
 ```
 
 ``` python
@@ -1045,6 +1056,7 @@ from cjm_transcript_source_select.services.source_utils import (
     group_transcriptions,
     group_transcriptions_by_audio,
     is_source_selected,
+    get_selected_media_paths,
     filter_transcriptions,
     select_all_in_group,
     toggle_source_selection,
@@ -1090,9 +1102,18 @@ def group_transcriptions_by_audio(
 ``` python
 def is_source_selected(
     record_id: str,  # Job ID to check
+    provider_id: str,  # Provider ID to check
     selected_sources: List[Dict[str, str]]  # List of selected sources
 ) -> bool:  # True if source is selected
-    "Check if a source is in the selected list."
+    "Check if a source is in the selected list by (record_id, provider_id) pair."
+```
+
+``` python
+def get_selected_media_paths(
+    selected_sources: List[Dict[str, str]],  # Current selections (record_id, provider_id)
+    all_transcriptions: List[Dict[str, Any]],  # All available transcription records
+) -> Set[str]:  # Media paths already represented in selections
+    "Get the set of media_paths for currently selected sources."
 ```
 
 ``` python
@@ -1109,6 +1130,7 @@ def select_all_in_group(
     group_key: str,  # Group key to match against
     grouping_mode: str,  # Grouping mode: "media_path" or "batch_id"
     selected_sources: List[Dict[str, str]],  # Current selections
+    excluded_media_paths: Optional[Set[str]] = None,  # Media paths to skip (already selected)
 ) -> List[Dict[str, str]]:  # Updated selections with new items appended
     "Add all transcriptions matching a group key to the selection list, skipping duplicates."
 ```
@@ -1119,7 +1141,7 @@ def toggle_source_selection(
     provider_id: str,  # Plugin name for the source
     selected_sources: List[Dict[str, str]],  # Current selections
 ) -> List[Dict[str, str]]:  # Updated selections
-    "Toggle a source in or out of the selection list."
+    "Toggle a source in or out of the selection list by (record_id, provider_id) pair."
 ```
 
 ``` python

cjm_transcript_source_select-0.0.2/cjm_transcript_source_select/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.0.2"

{cjm_transcript_source_select-0.0.1 → cjm_transcript_source_select-0.0.2}/cjm_transcript_source_select/_modidx.py

@@ -79,6 +79,8 @@ d = { 'settings': { 'branch': 'main',
                                                                                                             'cjm_transcript_source_select/models.py')},
             'cjm_transcript_source_select.routes.core': { 'cjm_transcript_source_select.routes.core._build_queue_response': ( 'routes/core.html#_build_queue_response',
                                                                                                                               'cjm_transcript_source_select/routes/core.py'),
+                                                          'cjm_transcript_source_select.routes.core._check_duplicate_media_path': ( 'routes/core.html#_check_duplicate_media_path',
+                                                                                                                                    'cjm_transcript_source_select/routes/core.py'),
                                                           'cjm_transcript_source_select.routes.core._get_active_source_tab': ( 'routes/core.html#_get_active_source_tab',
                                                                                                                                'cjm_transcript_source_select/routes/core.py'),
                                                           'cjm_transcript_source_select.routes.core._get_step_state': ( 'routes/core.html#_get_step_state',
@@ -197,6 +199,8 @@ d = { 'settings': { 'branch': 'main',
                                                                                                                                                'cjm_transcript_source_select/services/source_utils.py'),
                                                                     'cjm_transcript_source_select.services.source_utils.filter_transcriptions': ( 'services/source_utils.html#filter_transcriptions',
                                                                                                                                                   'cjm_transcript_source_select/services/source_utils.py'),
+                                                                    'cjm_transcript_source_select.services.source_utils.get_selected_media_paths': ( 'services/source_utils.html#get_selected_media_paths',
+                                                                                                                                                     'cjm_transcript_source_select/services/source_utils.py'),
                                                                     'cjm_transcript_source_select.services.source_utils.group_transcriptions': ( 'services/source_utils.html#group_transcriptions',
                                                                                                                                                  'cjm_transcript_source_select/services/source_utils.py'),
                                                                     'cjm_transcript_source_select.services.source_utils.group_transcriptions_by_audio': ( 'services/source_utils.html#group_transcriptions_by_audio',

{cjm_transcript_source_select-0.0.1 → cjm_transcript_source_select-0.0.2}/cjm_transcript_source_select/components/selection_queue.py

@@ -73,7 +73,7 @@ def _render_queue_item(
             lucide_icon("x", size=4, cls=str(text_dui.base_content.opacity(60))),
             cls=combine_classes(btn, btn_styles.ghost, btn_sizes.xs, m.l(1)),
             hx_post=remove_url,
-            hx_vals=json.dumps({"record_id": record_id}),
+            hx_vals=json.dumps({"record_id": record_id, "provider_id": provider_id}),
             hx_target=SelectionHtmlIds.as_selector(SelectionHtmlIds.QUEUE_CONTAINER),
             hx_swap="outerHTML",
             data_action="remove",

{cjm_transcript_source_select-0.0.1 → cjm_transcript_source_select-0.0.2}/cjm_transcript_source_select/components/source_browser.py

@@ -154,7 +154,7 @@ def _render_source_row(
                 hx_vals=json.dumps({"record_id": record_id, "provider_id": provider_id}),
                 hx_target=SelectionHtmlIds.as_selector(SelectionHtmlIds.QUEUE_CONTAINER),
                 hx_swap="outerHTML",
-                name=f"source_{record_id}"
+                name=f"source_{provider_id}_{record_id}"
             ),
             cls=str(w(12))
         ),
@@ -260,7 +260,9 @@ def _render_source_list(
         
         # Add record rows
         for record in records:
-            is_selected = is_source_selected(record.get("record_id", ""), selected_sources)
+            is_selected = is_source_selected(
+                record.get("record_id", ""), record.get("provider_id", ""), selected_sources
+            )
             table_rows.append(_render_source_row(
                 record, is_selected, add_url, remove_url, preview_url,
                 is_first=is_first_record

{cjm_transcript_source_select-0.0.1 → cjm_transcript_source_select-0.0.2}/cjm_transcript_source_select/routes/core.py

@@ -39,6 +39,22 @@ def _get_step_state(
     step_states = workflow_state.get("step_states", {})
     return step_states.get("selection", {})
 
+def _check_duplicate_media_path(
+    source_service: SourceService,  # Source service for lookups
+    record_id: str,  # Candidate record ID
+    provider_id: str,  # Candidate provider ID
+    selected_sources: List[Dict[str, str]],  # Current selections
+) -> bool:  # True if adding would duplicate an audio file
+    """Check if adding a source would duplicate an already-selected audio file."""
+    candidate = source_service.get_transcription_by_id(record_id, provider_id)
+    if not candidate or not candidate.media_path:
+        return False
+    for s in selected_sources:
+        existing = source_service.get_transcription_by_id(s["record_id"], s["provider_id"])
+        if existing and existing.media_path == candidate.media_path:
+            return True
+    return False
+
 # %% ../../nbs/routes/core.ipynb #3zll5oy1hsc
 def _get_active_source_tab(
     state_store: WorkflowStateStore,  # The workflow state store

{cjm_transcript_source_select-0.0.1 → cjm_transcript_source_select-0.0.2}/cjm_transcript_source_select/routes/filtering.py

@@ -14,14 +14,15 @@ from cjm_fasthtml_interactions.core.state_store import get_session_id
 
 from ..models import SelectionUrls
 from cjm_transcript_source_select.routes.core import (
-    WorkflowStateStore, _get_step_state, _update_step_state, _build_queue_response
+    WorkflowStateStore, _get_step_state, _update_step_state, _build_queue_response,
+    _check_duplicate_media_path
 )
 from cjm_transcript_source_select.components.source_browser import (
     _render_source_list
 )
 from ..services.source import SourceService
 from cjm_transcript_source_select.services.source_utils import (
-    filter_transcriptions, toggle_source_selection, reorder_item
+    filter_transcriptions, toggle_source_selection, reorder_item, is_source_selected
 )
 
 # %% ../../nbs/routes/filtering.ipynb #c4457084
@@ -100,6 +101,13 @@ def _handle_selection_toggle_focused(
     step_state = _get_step_state(state_store, workflow_id, session_id)
     selected_sources = step_state.get("selected_sources", [])
     
+    # Only check for duplicate media_path when adding (not removing)
+    if not is_source_selected(record_id, provider_id, selected_sources):
+        if _check_duplicate_media_path(source_service, record_id, provider_id, selected_sources):
+            return _build_queue_response(
+                state_store, workflow_id, source_service, session_id, selected_sources, urls
+            )
+    
     selected_sources = toggle_source_selection(record_id, provider_id, selected_sources)
     _update_step_state(state_store, workflow_id, session_id, selected_sources)
     

{cjm_transcript_source_select-0.0.1 → cjm_transcript_source_select-0.0.2}/cjm_transcript_source_select/routes/queue.py

@@ -14,14 +14,15 @@ from cjm_fasthtml_interactions.core.state_store import get_session_id
 
 from ..models import SelectionUrls
 from cjm_transcript_source_select.routes.core import (
-    WorkflowStateStore, _get_step_state, _update_step_state, _build_queue_response
+    WorkflowStateStore, _get_step_state, _update_step_state, _build_queue_response,
+    _check_duplicate_media_path
 )
 from cjm_transcript_source_select.components.preview_panel import (
     _render_preview_panel
 )
 from ..services.source import SourceService
 from cjm_transcript_source_select.services.source_utils import (
-    select_all_in_group, reorder_sources
+    select_all_in_group, reorder_sources, get_selected_media_paths
 )
 
 # %% ../../nbs/routes/queue.ipynb #a5934339
@@ -40,10 +41,17 @@ def _handle_selection_add(
     step_state = _get_step_state(state_store, workflow_id, session_id)
     selected_sources = step_state.get("selected_sources", [])
     
-    # Check if already selected
-    if not any(s.get("record_id") == record_id for s in selected_sources):
-        selected_sources.append({"record_id": record_id, "provider_id": provider_id})
-        _update_step_state(state_store, workflow_id, session_id, selected_sources)
+    # Check if already selected by (record_id, provider_id) pair
+    already_selected = any(
+        s.get("record_id") == record_id and s.get("provider_id") == provider_id
+        for s in selected_sources
+    )
+    
+    if not already_selected:
+        # Reject if another source with the same audio file is already queued
+        if not _check_duplicate_media_path(source_service, record_id, provider_id, selected_sources):
+            selected_sources.append({"record_id": record_id, "provider_id": provider_id})
+            _update_step_state(state_store, workflow_id, session_id, selected_sources)
     
     return _build_queue_response(state_store, workflow_id, source_service, session_id, selected_sources, urls)
 
@@ -55,6 +63,7 @@ def _handle_selection_remove(
     request,  # FastHTML request object
     sess,  # FastHTML session object
     record_id: str,  # Job ID to remove
+    provider_id: str,  # Plugin name for the source
     urls: SelectionUrls,  # URL bundle for rendering
 ):  # Queue component with OOB stats, optionally with OOB source list
     """Remove a source from the selection queue."""
@@ -62,8 +71,11 @@ def _handle_selection_remove(
     step_state = _get_step_state(state_store, workflow_id, session_id)
     selected_sources = step_state.get("selected_sources", [])
     
-    # Remove the item
-    selected_sources = [s for s in selected_sources if s.get("record_id") != record_id]
+    # Remove by (record_id, provider_id) pair
+    selected_sources = [
+        s for s in selected_sources
+        if not (s.get("record_id") == record_id and s.get("provider_id") == provider_id)
+    ]
     _update_step_state(state_store, workflow_id, session_id, selected_sources)
     
     return _build_queue_response(state_store, workflow_id, source_service, session_id, selected_sources, urls)
@@ -120,13 +132,17 @@ def _handle_selection_select_all(
     grouping_mode: str,  # Current grouping mode: "media_path" or "batch_id"
     urls: SelectionUrls,  # URL bundle for rendering
 ):  # Queue component with OOB stats, optionally with OOB source list
-    """Select all transcriptions for a given group."""
+    """Select all transcriptions for a given group, skipping duplicate audio sources."""
     session_id = get_session_id(sess)
     step_state = _get_step_state(state_store, workflow_id, session_id)
     selected_sources = step_state.get("selected_sources", [])
     
     all_transcriptions = source_service.query_transcriptions(limit=500)
-    selected_sources = select_all_in_group(all_transcriptions, group_key, grouping_mode, selected_sources)
+    excluded = get_selected_media_paths(selected_sources, all_transcriptions)
+    selected_sources = select_all_in_group(
+        all_transcriptions, group_key, grouping_mode, selected_sources,
+        excluded_media_paths=excluded,
+    )
     
     _update_step_state(state_store, workflow_id, session_id, selected_sources=selected_sources)
     
@@ -180,11 +196,11 @@ def init_queue_router(
         )
 
     @router
-    def remove(request, sess, record_id: str):
+    def remove(request, sess, record_id: str, provider_id: str):
         """Remove a source from the selection queue."""
         return _handle_selection_remove(
             state_store, workflow_id, source_service,
-            request, sess, record_id, urls=urls,
+            request, sess, record_id, provider_id, urls=urls,
         )
 
     @router

{cjm_transcript_source_select-0.0.1 → cjm_transcript_source_select-0.0.2}/cjm_transcript_source_select/services/source_utils.py

@@ -4,11 +4,12 @@
 
 # %% auto #0
 __all__ = ['extract_batch_id', 'extract_model_name', 'group_transcriptions', 'group_transcriptions_by_audio',
-           'is_source_selected', 'filter_transcriptions', 'select_all_in_group', 'toggle_source_selection',
-           'reorder_item', 'reorder_sources', 'calculate_next_tab', 'check_audio_exists', 'validate_browse_path']
+           'is_source_selected', 'get_selected_media_paths', 'filter_transcriptions', 'select_all_in_group',
+           'toggle_source_selection', 'reorder_item', 'reorder_sources', 'calculate_next_tab', 'check_audio_exists',
+           'validate_browse_path']
 
 # %% ../../nbs/services/source_utils.ipynb #su-imports
-from typing import Any, List, Dict
+from typing import Any, List, Dict, Optional, Set
 from pathlib import Path
 import json
 
@@ -85,10 +86,27 @@ def group_transcriptions_by_audio(
 # %% ../../nbs/services/source_utils.ipynb #su-is-source-selected
 def is_source_selected(
     record_id: str,  # Job ID to check
+    provider_id: str,  # Provider ID to check
     selected_sources: List[Dict[str, str]]  # List of selected sources
 ) -> bool:  # True if source is selected
-    """Check if a source is in the selected list."""
-    return any(s.get("record_id") == record_id for s in selected_sources)
+    """Check if a source is in the selected list by (record_id, provider_id) pair."""
+    return any(
+        s.get("record_id") == record_id and s.get("provider_id") == provider_id
+        for s in selected_sources
+    )
+
+# %% ../../nbs/services/source_utils.ipynb #yt3azuiiy3g
+def get_selected_media_paths(
+    selected_sources: List[Dict[str, str]],  # Current selections (record_id, provider_id)
+    all_transcriptions: List[Dict[str, Any]],  # All available transcription records
+) -> Set[str]:  # Media paths already represented in selections
+    """Get the set of media_paths for currently selected sources."""
+    selected_keys = {(s.get("record_id"), s.get("provider_id")) for s in selected_sources}
+    return {
+        t.get("media_path") for t in all_transcriptions
+        if (t.get("record_id"), t.get("provider_id")) in selected_keys
+        and t.get("media_path")
+    }
 
 # %% ../../nbs/services/source_utils.ipynb #tg25xqgkaa
 def filter_transcriptions(
@@ -113,6 +131,7 @@ def select_all_in_group(
     group_key: str,  # Group key to match against
     grouping_mode: str,  # Grouping mode: "media_path" or "batch_id"
     selected_sources: List[Dict[str, str]],  # Current selections
+    excluded_media_paths: Optional[Set[str]] = None,  # Media paths to skip (already selected)
 ) -> List[Dict[str, str]]:  # Updated selections with new items appended
     """Add all transcriptions matching a group key to the selection list, skipping duplicates."""
     # Filter transcriptions by group key
@@ -121,14 +140,24 @@ def select_all_in_group(
     else:
         matching = [t for t in transcriptions if t.get("media_path") == group_key]
     
-    # Deduplicate against existing selections
-    existing_record_ids = {s.get("record_id") for s in selected_sources}
+    # Deduplicate against existing selections using (record_id, provider_id) pairs
+    existing_keys = {(s.get("record_id"), s.get("provider_id")) for s in selected_sources}
+    used_paths = set(excluded_media_paths) if excluded_media_paths else set()
     result = list(selected_sources)
     for t in matching:
         record_id = t.get("record_id")
-        if record_id and record_id not in existing_record_ids:
-            result.append({"record_id": record_id, "provider_id": t.get("provider_id", "")})
-            existing_record_ids.add(record_id)
+        provider_id = t.get("provider_id", "")
+        media_path = t.get("media_path")
+        key = (record_id, provider_id)
+        if not record_id or key in existing_keys:
+            continue
+        # Skip if media_path already represented
+        if excluded_media_paths is not None and media_path and media_path in used_paths:
+            continue
+        result.append({"record_id": record_id, "provider_id": provider_id})
+        existing_keys.add(key)
+        if media_path:
+            used_paths.add(media_path)
     
     return result
 
@@ -138,9 +167,11 @@ def toggle_source_selection(
     provider_id: str,  # Plugin name for the source
     selected_sources: List[Dict[str, str]],  # Current selections
 ) -> List[Dict[str, str]]:  # Updated selections
-    """Toggle a source in or out of the selection list."""
-    if any(s.get("record_id") == record_id for s in selected_sources):
-        return [s for s in selected_sources if s.get("record_id") != record_id]
+    """Toggle a source in or out of the selection list by (record_id, provider_id) pair."""
+    if any(s.get("record_id") == record_id and s.get("provider_id") == provider_id
+           for s in selected_sources):
+        return [s for s in selected_sources
+                if not (s.get("record_id") == record_id and s.get("provider_id") == provider_id)]
     else:
         return selected_sources + [{"record_id": record_id, "provider_id": provider_id}]
 

{cjm_transcript_source_select-0.0.1 → cjm_transcript_source_select-0.0.2/cjm_transcript_source_select.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: cjm-transcript-source-select
-Version: 0.0.1
+Version: 0.0.2
 Summary: FastHTML source selection component for transcript decomposition workflows, with federated database browsing, drag-drop ordering, and keyboard navigation.
 Home-page: https://github.com/cj-mills/cjm-transcript-source-select
 Author: Christian J. Mills
@@ -105,51 +105,51 @@ graph LR
     components_local_files --> html_ids
     components_preview_panel --> html_ids
     components_selection_queue --> html_ids
-    components_source_browser --> services_source_utils
     components_source_browser --> utils
+    components_source_browser --> services_source_utils
     components_source_browser --> html_ids
-    components_step_renderer --> components_helpers
-    components_step_renderer --> components_source_browser
     components_step_renderer --> utils
+    components_step_renderer --> components_helpers
     components_step_renderer --> components_preview_panel
-    components_step_renderer --> html_ids
-    components_step_renderer --> components_local_files
     components_step_renderer --> components_selection_queue
+    components_step_renderer --> components_local_files
+    components_step_renderer --> components_source_browser
     components_step_renderer --> models
-    routes_core --> components_step_renderer
-    routes_core --> models
-    routes_core --> components_selection_queue
+    components_step_renderer --> html_ids
     routes_core --> components_source_browser
+    routes_core --> components_selection_queue
     routes_core --> services_source
-    routes_filtering --> routes_core
-    routes_filtering --> services_source_utils
-    routes_filtering --> models
+    routes_core --> models
+    routes_core --> components_step_renderer
     routes_filtering --> components_source_browser
+    routes_filtering --> services_source_utils
+    routes_filtering --> routes_core
     routes_filtering --> services_source
-    routes_init --> routes_queue
-    routes_init --> routes_local_files
-    routes_init --> models
+    routes_filtering --> models
     routes_init --> routes_filtering
-    routes_init --> routes_tabs
     routes_init --> services_source
+    routes_init --> models
+    routes_init --> routes_queue
+    routes_init --> routes_local_files
     routes_init --> routes_core
+    routes_init --> routes_tabs
     routes_local_files --> components_local_files
     routes_local_files --> routes_core
     routes_local_files --> services_source
     routes_local_files --> models
     routes_queue --> routes_core
-    routes_queue --> services_source_utils
+    routes_queue --> services_source
     routes_queue --> models
+    routes_queue --> services_source_utils
     routes_queue --> components_preview_panel
-    routes_queue --> services_source
-    routes_tabs --> routes_local_files
     routes_tabs --> components_local_files
-    routes_tabs --> services_source_utils
+    routes_tabs --> routes_local_files
     routes_tabs --> routes_core
     routes_tabs --> components_source_browser
-    routes_tabs --> models
     routes_tabs --> components_step_renderer
+    routes_tabs --> models
     routes_tabs --> services_source
+    routes_tabs --> services_source_utils
 ```
 
 *50 cross-module dependencies detected*
@@ -186,6 +186,16 @@ def _get_step_state(
     "Get the selection step state from the workflow state store."
 ```
 
+``` python
+def _check_duplicate_media_path(
+    source_service: SourceService,  # Source service for lookups
+    record_id: str,  # Candidate record ID
+    provider_id: str,  # Candidate provider ID
+    selected_sources: List[Dict[str, str]],  # Current selections
+) -> bool:  # True if adding would duplicate an audio file
+    "Check if adding a source would duplicate an already-selected audio file."
+```
+
 ``` python
 def _get_active_source_tab(
     state_store: WorkflowStateStore,  # The workflow state store
@@ -658,6 +668,7 @@ def _handle_selection_remove(
     request,  # FastHTML request object
     sess,  # FastHTML session object
     record_id: str,  # Job ID to remove
+    provider_id: str,  # Plugin name for the source
     urls: SelectionUrls,  # URL bundle for rendering
 ):  # Queue component with OOB stats, optionally with OOB source list
     "Remove a source from the selection queue."
@@ -698,7 +709,7 @@ def _handle_selection_select_all(
     grouping_mode: str,  # Current grouping mode: "media_path" or "batch_id"
     urls: SelectionUrls,  # URL bundle for rendering
 ):  # Queue component with OOB stats, optionally with OOB source list
-    "Select all transcriptions for a given group."
+    "Select all transcriptions for a given group, skipping duplicate audio sources."
 ```
 
 ``` python
@@ -1088,6 +1099,7 @@ from cjm_transcript_source_select.services.source_utils import (
     group_transcriptions,
     group_transcriptions_by_audio,
     is_source_selected,
+    get_selected_media_paths,
     filter_transcriptions,
     select_all_in_group,
     toggle_source_selection,
@@ -1133,9 +1145,18 @@ def group_transcriptions_by_audio(
 ``` python
 def is_source_selected(
     record_id: str,  # Job ID to check
+    provider_id: str,  # Provider ID to check
     selected_sources: List[Dict[str, str]]  # List of selected sources
 ) -> bool:  # True if source is selected
-    "Check if a source is in the selected list."
+    "Check if a source is in the selected list by (record_id, provider_id) pair."
+```
+
+``` python
+def get_selected_media_paths(
+    selected_sources: List[Dict[str, str]],  # Current selections (record_id, provider_id)
+    all_transcriptions: List[Dict[str, Any]],  # All available transcription records
+) -> Set[str]:  # Media paths already represented in selections
+    "Get the set of media_paths for currently selected sources."
 ```
 
 ``` python
@@ -1152,6 +1173,7 @@ def select_all_in_group(
     group_key: str,  # Group key to match against
     grouping_mode: str,  # Grouping mode: "media_path" or "batch_id"
     selected_sources: List[Dict[str, str]],  # Current selections
+    excluded_media_paths: Optional[Set[str]] = None,  # Media paths to skip (already selected)
 ) -> List[Dict[str, str]]:  # Updated selections with new items appended
     "Add all transcriptions matching a group key to the selection list, skipping duplicates."
 ```
@@ -1162,7 +1184,7 @@ def toggle_source_selection(
     provider_id: str,  # Plugin name for the source
     selected_sources: List[Dict[str, str]],  # Current selections
 ) -> List[Dict[str, str]]:  # Updated selections
-    "Toggle a source in or out of the selection list."
+    "Toggle a source in or out of the selection list by (record_id, provider_id) pair."
 ```
 
 ``` python

{cjm_transcript_source_select-0.0.1 → cjm_transcript_source_select-0.0.2}/settings.ini

@@ -1,16 +1,10 @@
 [DEFAULT]
-# All sections below are required unless otherwise specified.
-# See https://github.com/AnswerDotAI/nbdev/blob/main/settings.ini for examples.
-
-### Python library ###
 repo = cjm-transcript-source-select
-lib_name = %(repo)s
-version = 0.0.1
+lib_name = cjm-transcript-source-select
+version = 0.0.2
 min_python = 3.12
 license = apache2
 black_formatting = False
-
-### nbdev ###
 doc_path = _docs
 lib_path = cjm_transcript_source_select
 nbs_path = nbs
@@ -18,29 +12,27 @@ recursive = True
 tst_flags = notest
 put_version_in_init = True
 update_pyproject = True
-
-### Docs ###
 branch = main
 custom_sidebar = False
-doc_host = https://%(user)s.github.io
-doc_baseurl = /%(repo)s
-git_url = https://github.com/%(user)s/%(repo)s
-title = %(lib_name)s
-
-### PyPI ###
+doc_host = https://cj-mills.github.io
+doc_baseurl = /cjm-transcript-source-select
+git_url = https://github.com/cj-mills/cjm-transcript-source-select
+title = cjm-transcript-source-select
 audience = Developers
 author = Christian J. Mills
 author_email = 9126128+cj-mills@users.noreply.github.com
-copyright = 2026 onwards, %(author)s
+copyright = 2026 onwards, Christian J. Mills
 description = FastHTML source selection component for transcript decomposition workflows, with federated database browsing, drag-drop ordering, and keyboard navigation.
 keywords = nbdev jupyter notebook python
 language = English
 status = 3
 user = cj-mills
+requirements = cjm-plugin-system cjm-transcription-plugin-system cjm-fasthtml-app-core cjm-fasthtml-daisyui cjm_fasthtml_lucide_icons cjm_fasthtml_file_browser cjm_fasthtml_keyboard_navigation duckdb pandas cjm_workflow_state cjm_source_provider cjm_fasthtml_interactions
+readme_nb = index.ipynb
+allowed_metadata_keys = 
+allowed_cell_metadata_keys = 
+jupyter_hooks = False
+clean_ids = True
+clear_all = False
+skip_procs = 
 
-### Optional ###
-requirements = cjm-plugin-system cjm-transcription-plugin-system cjm-fasthtml-app-core cjm-fasthtml-daisyui cjm_fasthtml_lucide_icons cjm_fasthtml_file_browser cjm_fasthtml_keyboard_navigation duckdb pandas cjm_workflow_state cjm_source_provider cjm_fasthtml_interactions 
-# dev_requirements = 
-# console_scripts =
-# conda_user = 
-# package_data =

cjm_transcript_source_select-0.0.1/cjm_transcript_source_select/__init__.py

@@ -1 +0,0 @@
-__version__ = "0.0.1"