df

models.py

@@ -33,10 +33,10 @@ from rich.progress import (
 @dataclass(slots=True)
 class PipeObject:
     """Unified pipeline object for tracking files, metadata, tag values, and relationships through the pipeline.
-    
+
     This is the single source of truth for all result data in the pipeline. Uses the hash+store
     canonical pattern for file identification.
-    
+
     Attributes:
         hash: SHA-256 hash of the file (canonical identifier)
         store: Storage backend name (e.g., 'default', 'hydrus', 'test', 'home')
@@ -53,6 +53,7 @@ class PipeObject:
         parent_hash: Hash of the parent file in the pipeline chain (for tracking provenance/lineage)
         extra: Additional fields not covered above
     """
+
     hash: str
     store: str
     provider: Optional[str] = None
@@ -72,21 +73,21 @@ class PipeObject:
 
     def add_relationship(self, rel_type: str, rel_hash: str) -> None:
         """Add a relationship hash.
-        
+
         Args:
             rel_type: Relationship type ('king', 'alt', 'related')
             rel_hash: Hash to add to the relationship
         """
         if rel_type not in self.relationships:
             self.relationships[rel_type] = []
-        
+
         if isinstance(self.relationships[rel_type], list):
             if rel_hash not in self.relationships[rel_type]:
                 self.relationships[rel_type].append(rel_hash)
         else:
             # Single value (e.g., king), convert to that value
             self.relationships[rel_type] = rel_hash
-    
+
     def get_relationships(self) -> Dict[str, Any]:
         """Get all relationships for this object."""
         return self.relationships.copy() if self.relationships else {}
@@ -114,7 +115,10 @@ class PipeObject:
         cmdlet_name = "PipeObject"
         try:
             import pipeline as ctx
-            current = ctx.get_current_cmdlet_name("") if hasattr(ctx, "get_current_cmdlet_name") else ""
+
+            current = (
+                ctx.get_current_cmdlet_name("") if hasattr(ctx, "get_current_cmdlet_name") else ""
+            )
             if current:
                 cmdlet_name = current
             else:
@@ -145,7 +149,7 @@ class PipeObject:
 
         if self.provider:
             data["provider"] = self.provider
-        
+
         if self.tag:
             data["tag"] = self.tag
         if self.title:
@@ -170,7 +174,7 @@ class PipeObject:
             data["action"] = self.action
         if self.parent_hash:
             data["parent_hash"] = self.parent_hash
-        
+
         # Add extra fields
         data.update({k: v for k, v in self.extra.items() if v is not None})
         return data
@@ -178,22 +182,22 @@ class PipeObject:
 
 class FileRelationshipTracker:
     """Track relationships between files for sidecar creation.
-    
+
     Allows tagging files with their relationships to other files:
     - king: The primary/master version of a file
     - alt: Alternate versions of the same content
     - related: Related files (e.g., screenshots of a book)
     """
-    
+
     def __init__(self) -> None:
         self.relationships: Dict[str, Dict[str, Any]] = {}
-    
+
     def register_king(self, file_path: str, file_hash: str) -> None:
         """Register a file as the king (primary) version."""
         if file_path not in self.relationships:
             self.relationships[file_path] = {}
         self.relationships[file_path]["king"] = file_hash
-    
+
     def add_alt(self, file_path: str, alt_hash: str) -> None:
         """Add an alternate version of a file."""
         if file_path not in self.relationships:
@@ -202,7 +206,7 @@ class FileRelationshipTracker:
             self.relationships[file_path]["alt"] = []
         if alt_hash not in self.relationships[file_path]["alt"]:
             self.relationships[file_path]["alt"].append(alt_hash)
-    
+
     def add_related(self, file_path: str, related_hash: str) -> None:
         """Add a related file."""
         if file_path not in self.relationships:
@@ -211,14 +215,14 @@ class FileRelationshipTracker:
             self.relationships[file_path]["related"] = []
         if related_hash not in self.relationships[file_path]["related"]:
             self.relationships[file_path]["related"].append(related_hash)
-    
+
     def get_relationships(self, file_path: str) -> Optional[Dict[str, Any]]:
         """Get relationships for a file."""
         return self.relationships.get(file_path)
-    
+
     def link_files(self, primary_path: str, king_hash: str, *alt_paths: str) -> None:
         """Link files together with primary as king and others as alternates.
-        
+
         Args:
             primary_path: Path to the primary file (will be marked as 'king')
             king_hash: Hash of the primary file
@@ -231,6 +235,7 @@ class FileRelationshipTracker:
                 self.add_alt(primary_path, alt_hash)
             except Exception as e:
                 import sys
+
                 print(f"Error hashing {alt_path}: {e}", file=sys.stderr)
 
 
@@ -245,6 +250,7 @@ def _get_file_hash(filepath: str) -> str:
 
 # ============= Download Module Classes =============
 
+
 class DownloadError(RuntimeError):
     """Raised when the download or Hydrus import fails."""
 
@@ -252,9 +258,10 @@ class DownloadError(RuntimeError):
 @dataclass(slots=True)
 class DownloadOptions:
     """Configuration for downloading media.
-    
+
     Use the add-file cmdlet separately for Hydrus import.
     """
+
     url: str
     mode: str  # "audio" or "video"
     output_dir: Path
@@ -273,13 +280,14 @@ class DownloadOptions:
 
 class SendFunc(Protocol):
     """Protocol for event sender function."""
-    def __call__(self, event: str, **payload: Any) -> None:
-        ...
+
+    def __call__(self, event: str, **payload: Any) -> None: ...
 
 
 @dataclass(slots=True)
 class DownloadMediaResult:
     """Result of a successful media download."""
+
     path: Path
     info: Dict[str, Any]
     tag: List[str]
@@ -291,6 +299,7 @@ class DownloadMediaResult:
 @dataclass(slots=True)
 class DebugLogger:
     """Logs events to a JSON debug file for troubleshooting downloads."""
+
     path: Path
     file: Optional[TextIO] = None
     session_started: bool = False
@@ -383,8 +392,7 @@ def _sanitise_for_json(value: Any, *, max_depth: int = 8, _seen: Optional[set[in
         if isinstance(value, (list, tuple, set)):
             iterable = value if not isinstance(value, set) else list(value)
             return [
-                _sanitise_for_json(item, max_depth=max_depth - 1, _seen=_seen)
-                for item in iterable
+                _sanitise_for_json(item, max_depth=max_depth - 1, _seen=_seen) for item in iterable
             ]
         if is_dataclass(value) and not isinstance(value, type):
             return _sanitise_for_json(asdict(value), max_depth=max_depth - 1, _seen=_seen)
@@ -393,6 +401,7 @@ def _sanitise_for_json(value: Any, *, max_depth: int = 8, _seen: Optional[set[in
 
     return repr(value)
 
+
 class ProgressBar:
     """Rich progress helper for byte-based transfers.
 
@@ -419,7 +428,9 @@ class ProgressBar:
             # Pipeline-backed transfer task is already registered; update its total if needed.
             try:
                 if total is not None and total > 0:
-                    self._pipeline_ui.update_transfer(label=self._pipeline_label, completed=None, total=int(total))
+                    self._pipeline_ui.update_transfer(
+                        label=self._pipeline_label, completed=None, total=int(total)
+                    )
             except Exception:
                 pass
             return
@@ -438,7 +449,10 @@ class ProgressBar:
                 self._pipeline_ui = ui
                 self._pipeline_label = str(label or "download")
                 try:
-                    ui.begin_transfer(label=self._pipeline_label, total=int(total) if isinstance(total, int) and total > 0 else None)
+                    ui.begin_transfer(
+                        label=self._pipeline_label,
+                        total=int(total) if isinstance(total, int) and total > 0 else None,
+                    )
                 except Exception:
                     # If pipeline integration fails, fall back to standalone progress.
                     self._pipeline_ui = None
@@ -503,7 +517,9 @@ class ProgressBar:
         if self._progress is None or self._task_id is None:
             return
         if total is not None and total > 0:
-            self._progress.update(self._task_id, completed=int(downloaded or 0), total=int(total), refresh=True)
+            self._progress.update(
+                self._task_id, completed=int(downloaded or 0), total=int(total), refresh=True
+            )
         else:
             self._progress.update(self._task_id, completed=int(downloaded or 0), refresh=True)
 
@@ -528,21 +544,21 @@ class ProgressBar:
 
     def format_bytes(self, bytes_val: Optional[float]) -> str:
         """Format bytes to human-readable size.
-        
+
         Args:
             bytes_val: Number of bytes or None.
-            
+
         Returns:
             Formatted string (e.g., "123.4 MB", "1.2 GB").
         """
         if bytes_val is None or bytes_val <= 0:
             return "?.? B"
-        
+
         for unit in ("B", "KB", "MB", "GB", "TB"):
             if bytes_val < 1024:
                 return f"{bytes_val:.1f} {unit}"
             bytes_val /= 1024
-        
+
         return f"{bytes_val:.1f} PB"
 
     # NOTE: rich.Progress handles the visual formatting; format_bytes remains as a general utility.
@@ -555,7 +571,14 @@ class ProgressFileReader:
     Progress is written to stderr (so pipelines remain clean).
     """
 
-    def __init__(self, fileobj: Any, *, total_bytes: Optional[int], label: str = "upload", min_interval_s: float = 0.25):
+    def __init__(
+        self,
+        fileobj: Any,
+        *,
+        total_bytes: Optional[int],
+        label: str = "upload",
+        min_interval_s: float = 0.25,
+    ):
         self._f = fileobj
         self._total = int(total_bytes) if total_bytes not in (None, 0, "") else 0
         self._label = str(label or "upload")
@@ -574,7 +597,12 @@ class ProgressFileReader:
         now = time.time()
         if now - self._last < self._min_interval_s:
             return
-        self._bar.update(downloaded=int(self._read), total=int(self._total), label=str(self._label or "upload"), file=sys.stderr)
+        self._bar.update(
+            downloaded=int(self._read),
+            total=int(self._total),
+            label=str(self._label or "upload"),
+            file=sys.stderr,
+        )
         self._last = now
 
     def _finish(self) -> None:
@@ -868,7 +896,13 @@ class PipelineLiveProgress:
             return
         if self._live is not None:
             return
-        if self._console is None or self._pipe_progress is None or self._subtasks is None or self._transfers is None or self._overall is None:
+        if (
+            self._console is None
+            or self._pipe_progress is None
+            or self._subtasks is None
+            or self._transfers is None
+            or self._overall is None
+        ):
             # Not initialized yet; start fresh.
             self.start()
             return
@@ -1081,7 +1115,9 @@ class PipelineLiveProgress:
         except Exception:
             pass
 
-    def update_transfer(self, *, label: str, completed: Optional[int], total: Optional[int] = None) -> None:
+    def update_transfer(
+        self, *, label: str, completed: Optional[int], total: Optional[int] = None
+    ) -> None:
         if not self._enabled:
             return
         if self._transfers is None:
@@ -1123,7 +1159,9 @@ class PipelineLiveProgress:
             return False
         return True
 
-    def begin_pipe(self, pipe_index: int, *, total_items: int, items_preview: Optional[List[Any]] = None) -> None:
+    def begin_pipe(
+        self, pipe_index: int, *, total_items: int, items_preview: Optional[List[Any]] = None
+    ) -> None:
         if not self._ensure_pipe(pipe_index):
             return
         pipe_progress = self._pipe_progress
@@ -1321,7 +1359,11 @@ class PipelineLiveProgress:
         if self._overall_task is not None:
             completed = 0
             try:
-                completed = sum(1 for i in range(len(self._pipe_labels)) if self._pipe_done[i] >= max(1, self._pipe_totals[i]))
+                completed = sum(
+                    1
+                    for i in range(len(self._pipe_labels))
+                    if self._pipe_done[i] >= max(1, self._pipe_totals[i])
+                )
             except Exception:
                 completed = 0
             overall.update(
@@ -1330,6 +1372,7 @@ class PipelineLiveProgress:
                 description=f"Pipeline: {completed}/{len(self._pipe_labels)} pipes completed",
             )
 
+
 class PipelineStageContext:
     """Context information for the current pipeline stage."""
 
@@ -1343,7 +1386,7 @@ class PipelineStageContext:
     ):
         self.stage_index = stage_index
         self.total_stages = total_stages
-        self.is_last_stage = (stage_index == total_stages - 1)
+        self.is_last_stage = stage_index == total_stages - 1
         self.pipe_index = int(pipe_index) if pipe_index is not None else None
         self.worker_id = worker_id
         self._on_emit = on_emit
@@ -1377,13 +1420,14 @@ class PipelineStageContext:
 # Consolidated from result_table.py
 # ============================================================================
 
+
 @dataclass
 class InputOption:
     """Represents an interactive input option (cmdlet argument) in a table.
-    
+
     Allows users to select options that translate to cmdlet arguments,
     enabling interactive configuration right from the result table.
-    
+
     Example:
         # Create an option for location selection
         location_opt = InputOption(
@@ -1392,11 +1436,12 @@ class InputOption:
             choices=["local", "hydrus", "0x0"],
             description="Download destination"
         )
-        
+
         # Use in result table
         table.add_input_option(location_opt)
         selected = table.select_option("location")  # Returns user choice
     """
+
     name: str
     """Option name (maps to cmdlet argument)"""
     type: str = "string"
@@ -1409,7 +1454,7 @@ class InputOption:
     """Description of what this option does"""
     validator: Optional[Callable[[str], bool]] = None
     """Optional validator function: takes value, returns True if valid"""
-    
+
     def to_dict(self) -> Dict[str, Any]:
         """Convert to dictionary."""
         return {
@@ -1424,10 +1469,11 @@ class InputOption:
 @dataclass
 class TUIResultCard:
     """Represents a result as a UI card with title, metadata, and actions.
-    
-    Used in hub-ui and TUI contexts to render individual search results 
+
+    Used in hub-ui and TUI contexts to render individual search results
     as grouped components with visual structure.
     """
+
     title: str
     subtitle: Optional[str] = None
     metadata: Optional[Dict[str, str]] = None
@@ -1436,7 +1482,7 @@ class TUIResultCard:
     file_hash: Optional[str] = None
     file_size: Optional[str] = None
     duration: Optional[str] = None
-    
+
     def __post_init__(self):
         """Initialize default values."""
         if self.metadata is None:
@@ -1448,14 +1494,15 @@ class TUIResultCard:
 @dataclass
 class ResultColumn:
     """Represents a single column in a result table."""
+
     name: str
     value: str
     width: Optional[int] = None
-    
+
     def __str__(self) -> str:
         """String representation of the column."""
         return f"{self.name}: {self.value}"
-    
+
     def to_dict(self) -> Dict[str, str]:
         """Convert to dictionary."""
         return {"name": self.name, "value": self.value}
@@ -1464,28 +1511,29 @@ class ResultColumn:
 @dataclass
 class ResultRow:
     """Represents a single row in a result table."""
+
     columns: List[ResultColumn] = field(default_factory=list)
-    
+
     def add_column(self, name: str, value: Any) -> None:
         """Add a column to this row."""
         str_value = str(value) if value is not None else ""
         self.columns.append(ResultColumn(name, str_value))
-    
+
     def get_column(self, name: str) -> Optional[str]:
         """Get column value by name."""
         for col in self.columns:
             if col.name.lower() == name.lower():
                 return col.value
         return None
-    
+
     def to_dict(self) -> List[Dict[str, str]]:
         """Convert to list of column dicts."""
         return [col.to_dict() for col in self.columns]
-    
+
     def to_list(self) -> List[tuple[str, str]]:
         """Convert to list of (name, value) tuples."""
         return [(col.name, col.value) for col in self.columns]
-    
+
     def __str__(self) -> str:
         """String representation of the row."""
-        return " | ".join(str(col) for col in self.columns)
+        return " | ".join(str(col) for col in self.columns)