Medios-Macina/cmdlets/delete_tag.py

from __future__ import annotations

from typing import Any, Dict, Sequence
from pathlib import Path
import json
import sys

from . import register
import models
import pipeline as ctx
from helper import hydrus as hydrus_wrapper
from ._shared import Cmdlet, CmdletArg, SharedArgs, normalize_hash, parse_tag_arguments, fetch_hydrus_metadata, should_show_help, get_field
from helper.logger import debug, log


def _refresh_tag_view_if_current(hash_hex: str | None, file_path: str | None, config: Dict[str, Any]) -> None:
    """If the current subject matches the target, refresh tags via get-tag."""
    try:
        from cmdlets import get_tag as get_tag_cmd  # type: ignore
    except Exception:
        return

    try:
        subject = ctx.get_last_result_subject()
        if subject is None:
            return

        def norm(val: Any) -> str:
            return str(val).lower()

        target_hash = norm(hash_hex) if hash_hex else None
        target_path = norm(file_path) if file_path else None

        subj_hashes: list[str] = []
        subj_paths: list[str] = []
        if isinstance(subject, dict):
            subj_hashes = [norm(v) for v in [subject.get("hydrus_hash"), subject.get("hash"), subject.get("hash_hex"), subject.get("file_hash")] if v]
            subj_paths = [norm(v) for v in [subject.get("file_path"), subject.get("path"), subject.get("target")] if v]
        else:
            subj_hashes = [norm(get_field(subject, f)) for f in ("hydrus_hash", "hash", "hash_hex", "file_hash") if get_field(subject, f)]
            subj_paths = [norm(get_field(subject, f)) for f in ("file_path", "path", "target") if get_field(subject, f)]

        is_match = False
        if target_hash and target_hash in subj_hashes:
            is_match = True
        if target_path and target_path in subj_paths:
            is_match = True
        if not is_match:
            return

        refresh_args: list[str] = []
        if hash_hex:
            refresh_args.extend(["-hash", hash_hex])
        get_tag_cmd._run(subject, refresh_args, config)
    except Exception:
        pass


CMDLET = Cmdlet(
    name="delete-tags",
    summary="Remove tags from a Hydrus file.",
    usage="del-tags [-hash <sha256>] <tag>[,<tag>...]",
    alias=["del-tag", "del-tags", "delete-tag"],
    arg=[
        SharedArgs.HASH,
        CmdletArg("<tag>[,<tag>...]", required=True, description="One or more tags to remove. Comma- or space-separated."),
    ],
    detail=[
        "- Requires a Hydrus file (hash present) or explicit -hash override.",
        "- Multiple tags can be comma-separated or space-separated.",
    ],
)

@register(["del-tag", "del-tags", "delete-tag", "delete-tags"])  # Still needed for backward compatibility
def _run(result: Any, args: Sequence[str], config: Dict[str, Any]) -> int:
    # Help
    if should_show_help(args):
        log(json.dumps(CMDLET, ensure_ascii=False, indent=2))
        return 0
    
    # Check if we have a piped TagItem with no args (i.e., from @1 | delete-tag)
    has_piped_tag = (result and hasattr(result, '__class__') and 
                     result.__class__.__name__ == 'TagItem' and 
                     hasattr(result, 'tag_name'))
    
    # Check if we have a piped list of TagItems (from @N selection)
    has_piped_tag_list = (isinstance(result, list) and result and 
                          hasattr(result[0], '__class__') and 
                          result[0].__class__.__name__ == 'TagItem')

    if not args and not has_piped_tag and not has_piped_tag_list:
        log("Requires at least one tag argument")
        return 1
    
    # Parse -hash override and collect tags from remaining args
    override_hash: str | None = None
    rest: list[str] = []
    i = 0
    while i < len(args):
        a = args[i]
        low = str(a).lower()
        if low in {"-hash", "--hash", "hash"} and i + 1 < len(args):
            override_hash = str(args[i + 1]).strip()
            i += 2
            continue
        rest.append(a)
        i += 1
    
    # Check if first argument is @ syntax (result table selection)
    # @5 or @{2,5,8} to delete tags from ResultTable by index
    tags_from_at_syntax = []
    hash_from_at_syntax = None
    file_path_from_at_syntax = None
    
    if rest and str(rest[0]).startswith("@"):
        selector_arg = str(rest[0])
        pipe_selector = selector_arg[1:].strip()
        # Parse @N or @{N,M,K} syntax
        if pipe_selector.startswith("{") and pipe_selector.endswith("}"):
            # @{2,5,8}
            pipe_selector = pipe_selector[1:-1]
        try:
            indices = [int(tok.strip()) for tok in pipe_selector.split(',') if tok.strip()]
        except ValueError:
            log("Invalid selection syntax. Use @2 or @{2,5,8}")
            return 1
        
        # Get the last ResultTable from pipeline context
        try:
            last_table = ctx._LAST_RESULT_TABLE
            if last_table:
                # Extract tags from selected rows
                for idx in indices:
                    if 1 <= idx <= len(last_table.rows):
                        # Look for a TagItem in _LAST_RESULT_ITEMS by index
                        if idx - 1 < len(ctx._LAST_RESULT_ITEMS):
                            item = ctx._LAST_RESULT_ITEMS[idx - 1]
                            if hasattr(item, '__class__') and item.__class__.__name__ == 'TagItem':
                                tag_name = get_field(item, 'tag_name')
                                if tag_name:
                                    log(f"[delete_tag] Extracted tag from @{idx}: {tag_name}")
                                    tags_from_at_syntax.append(tag_name)
                                # Also get hash from first item for consistency
                                if not hash_from_at_syntax:
                                    hash_from_at_syntax = get_field(item, 'hash_hex')
                                if not file_path_from_at_syntax:
                                    file_path_from_at_syntax = get_field(item, 'file_path')
                
                if not tags_from_at_syntax:
                    log(f"No tags found at indices: {indices}")
                    return 1
            else:
                log("No ResultTable in pipeline (use @ after running get-tag)")
                return 1
        except Exception as exc:
            log(f"Error processing @ selection: {exc}", file=__import__('sys').stderr)
            return 1
    
    # Handle @N selection which creates a list - extract the first item
    # If we have a list of TagItems, we want to process ALL of them if no args provided
    # This handles: delete-tag @1 (where @1 expands to a list containing one TagItem)
    # Also handles: delete-tag @1,2 (where we want to delete tags from multiple files)
    
    # Normalize result to a list for processing
    items_to_process = []
    if isinstance(result, list):
        items_to_process = result
    elif result:
        items_to_process = [result]
    
    # If we have TagItems and no args, we are deleting the tags themselves
    # If we have Files (or other objects) and args, we are deleting tags FROM those files
    
    # Check if we are in "delete selected tags" mode (TagItems)
    is_tag_item_mode = (items_to_process and hasattr(items_to_process[0], '__class__') and 
                        items_to_process[0].__class__.__name__ == 'TagItem')
    
    if is_tag_item_mode:
        # Collect all tags to delete from the TagItems
        # Group by hash/file_path to batch operations if needed, or just process one by one
        # For simplicity, we'll process one by one or group by file
        pass
    else:
        # "Delete tags from files" mode
        # We need args (tags to delete)
        if not args and not tags_from_at_syntax:
             log("Requires at least one tag argument when deleting from files")
             return 1
    
    # Process each item
    success_count = 0
    
    # If we have tags from @ syntax (e.g. delete-tag @{1,2}), we ignore the piped result for tag selection
    # but we might need the piped result for the file context if @ selection was from a Tag table
    # Actually, the @ selection logic above already extracted tags.
    
    if tags_from_at_syntax:
        # Special case: @ selection of tags.
        # We already extracted tags and hash/path.
        # Just run the deletion once using the extracted info.
        # This preserves the existing logic for @ selection.
        
        tags = tags_from_at_syntax
        hash_hex = normalize_hash(override_hash) if override_hash else normalize_hash(hash_from_at_syntax)
        file_path = file_path_from_at_syntax
        
        if _process_deletion(tags, hash_hex, file_path, config):
            success_count += 1
            
    else:
        # Process items from pipe (or single result)
        # If args are provided, they are the tags to delete from EACH item
        # If items are TagItems and no args, the tag to delete is the item itself
        
        tags_arg = parse_tag_arguments(rest)
        
        for item in items_to_process:
            tags_to_delete = []
            item_hash = normalize_hash(override_hash) if override_hash else normalize_hash(get_field(item, "hash_hex"))
            item_path = (
                get_field(item, "path")
                or get_field(item, "file_path")
                or get_field(item, "target")
            )
            item_source = get_field(item, "source")
            
            if hasattr(item, '__class__') and item.__class__.__name__ == 'TagItem':
                # It's a TagItem
                if tags_arg:
                    # User provided tags to delete FROM this file (ignoring the tag name in the item?)
                    # Or maybe they want to delete the tag in the item AND the args?
                    # Usually if piping TagItems, we delete THOSE tags.
                    # If args are present, maybe we should warn?
                    # For now, if args are present, assume they override or add to the tag item?
                    # Let's assume if args are present, we use args. If not, we use the tag name.
                    tags_to_delete = tags_arg
                else:
                    tag_name = get_field(item, 'tag_name')
                    if tag_name:
                        tags_to_delete = [tag_name]
            else:
                # It's a File or other object
                if tags_arg:
                    tags_to_delete = tags_arg
                else:
                    # No tags provided for a file object - skip or error?
                    # We already logged an error if no args and not TagItem mode globally,
                    # but inside the loop we might have mixed items? Unlikely.
                    continue
            
            if tags_to_delete and (item_hash or item_path):
                if _process_deletion(tags_to_delete, item_hash, item_path, config, source=item_source):
                    success_count += 1

    if success_count > 0:
        return 0
    return 1

def _process_deletion(tags: list[str], hash_hex: str | None, file_path: str | None, config: Dict[str, Any], source: str | None = None) -> bool:
    """Helper to execute the deletion logic for a single target."""
    
    if not tags:
        return False
    
    def _fetch_existing_tags() -> list[str]:
        existing: list[str] = []
        # Prefer local DB when we have a path and not explicitly hydrus
        if file_path and (source == "local" or (source != "hydrus" and not hash_hex)):
            try:
                from helper.folder_store import FolderDB
                from config import get_local_storage_path
                path_obj = Path(file_path)
                local_root = get_local_storage_path(config) or path_obj.parent
                with FolderDB(local_root) as db:
                    file_hash = db.get_file_hash(path_obj)
                    existing = db.get_tags(file_hash) if file_hash else []
            except Exception:
                existing = []
        elif hash_hex:
            meta, _ = fetch_hydrus_metadata(
                config, hash_hex,
                include_service_keys_to_tags=True,
                include_file_url=False,
            )
            if isinstance(meta, dict):
                tags_payload = meta.get("tags")
                if isinstance(tags_payload, dict):
                    seen: set[str] = set()
                    for svc_data in tags_payload.values():
                        if not isinstance(svc_data, dict):
                            continue
                        display = svc_data.get("display_tags")
                        if isinstance(display, list):
                            for t in display:
                                    if isinstance(t, (str, bytes)):
                                        val = str(t).strip()
                                        if val and val not in seen:
                                            seen.add(val)
                                            existing.append(val)
                            storage = svc_data.get("storage_tags")
                            if isinstance(storage, dict):
                                current_list = storage.get("0") or storage.get(0)
                                if isinstance(current_list, list):
                                    for t in current_list:
                                        if isinstance(t, (str, bytes)):
                                            val = str(t).strip()
                                            if val and val not in seen:
                                                seen.add(val)
                                                existing.append(val)
        return existing

    # Safety: only block if this deletion would remove the final title tag
    title_tags = [t for t in tags if isinstance(t, str) and t.lower().startswith("title:")]
    if title_tags:
        existing_tags = _fetch_existing_tags()
        current_titles = [t for t in existing_tags if isinstance(t, str) and t.lower().startswith("title:")]
        del_title_set = {t.lower() for t in title_tags}
        remaining_titles = [t for t in current_titles if t.lower() not in del_title_set]
        if current_titles and not remaining_titles:
            log("Cannot delete the last title: tag. Add a replacement title first (add-tag \"title:new title\").", file=sys.stderr)
            return False
        
    if not hash_hex and not file_path:
        log("Item does not include a hash or file path")
        return False

    # Handle local file tag deletion
    if file_path and (source == "local" or (not hash_hex and source != "hydrus")):
        try:
            from helper.folder_store import FolderDB
            from pathlib import Path
            
            path_obj = Path(file_path)
            if not path_obj.exists():
                log(f"File not found: {file_path}")
                return False
            
            # Try to get local storage path from config
            from config import get_local_storage_path
            local_root = get_local_storage_path(config)
            
            if not local_root:
                # Fallback: assume file is in a library root or use its parent
                local_root = path_obj.parent
            
            with FolderDB(local_root) as db:
                db.remove_tags(path_obj, tags)
            debug(f"Removed {len(tags)} tag(s) from {path_obj.name} (local)")
            _refresh_tag_view_if_current(hash_hex, file_path, config)
            return True
            
        except Exception as exc:
            log(f"Failed to remove local tags: {exc}")
            return False

    # Hydrus deletion logic
    if not hash_hex:
        return False

    try:
        service_name = hydrus_wrapper.get_tag_service_name(config)
        client = hydrus_wrapper.get_client(config)
        
        if client is None:
            log("Hydrus client unavailable")
            return False
        
        debug(f"Sending deletion request: hash={hash_hex}, tags={tags}, service={service_name}")
        client.delete_tags(hash_hex, tags, service_name)
        
        preview = hash_hex[:12] + ('…' if len(hash_hex) > 12 else '')
        debug(f"Removed {len(tags)} tag(s) from {preview} via '{service_name}'.")
        _refresh_tag_view_if_current(hash_hex, None, config)
        return True
        
    except Exception as exc:
        log(f"Hydrus del-tag failed: {exc}")
        return False