Medios-Macina/Store/HydrusNetwork.py

from __future__ import annotations

import re
import sys
from pathlib import Path
from typing import Any, Dict, List, Optional, Tuple

from SYS.logger import debug, log
from SYS.utils_constant import mime_maps

from Store._base import Store


class HydrusNetwork(Store):
    """File storage backend for Hydrus client.
    
    Each instance represents a specific Hydrus client connection.
    Maintains its own HydrusClient with session key.
    """
    
    def __init__(self, instance_name: str, api_key: str, url: str) -> None:
        """Initialize Hydrus storage backend.
        
        Args:
            instance_name: Name of this Hydrus instance (e.g., 'home', 'work')
            api_key: Hydrus Client API access key
            url: Hydrus client URL (e.g., 'http://192.168.1.230:45869')
        """
        from API.HydrusNetwork import HydrusNetwork as HydrusClient
        
        self._instance_name = instance_name
        self._api_key = api_key
        self._url = url
        # Create persistent client with session key for this instance
        self._client = HydrusClient(url=url, access_key=api_key)

    def name(self) -> str:
        return self._instance_name

    def get_name(self) -> str:
        return self._instance_name

    def add_file(self, file_path: Path, **kwargs: Any) -> str:
        """Upload file to Hydrus with full metadata support.
        
        Args:
            file_path: Path to the file to upload
            tag: Optional list of tag values to add
            url: Optional list of url to associate with the file
            title: Optional title (will be added as 'title:value' tag)
            
        Returns:
            File hash from Hydrus
            
        Raises:
            Exception: If upload fails
        """
        from SYS.utils import sha256_file

        tag_list = kwargs.get("tag", [])
        url = kwargs.get("url", [])
        title = kwargs.get("title")
        
        # Add title to tags if provided and not already present
        if title:
            title_tag = f"title:{title}"
            if not any(str(candidate).lower().startswith("title:") for candidate in tag_list):
                tag_list = [title_tag] + list(tag_list)

        try:
            # Compute file hash
            file_hash = sha256_file(file_path)
            debug(f"File hash: {file_hash}")

            # Use persistent client with session key
            client = self._client
            if client is None:
                raise Exception("Hydrus client unavailable")

            # Check if file already exists in Hydrus
            file_exists = False
            try:
                metadata = client.fetch_file_metadata(hashes=[file_hash])
                if metadata and isinstance(metadata, dict):
                    files = metadata.get("file_metadata", [])
                    if files:
                        file_exists = True
                        log(
                            f"ℹ️  Duplicate detected - file already in Hydrus with hash: {file_hash}",
                            file=sys.stderr,
                        )
            except Exception:
                pass

            # Upload file if not already present
            if not file_exists:
                log(f"Uploading to Hydrus: {file_path.name}", file=sys.stderr)
                response = client.add_file(file_path)

                # Extract hash from response
                hydrus_hash: Optional[str] = None
                if isinstance(response, dict):
                    hydrus_hash = response.get("hash") or response.get("file_hash")
                    if not hydrus_hash:
                        hashes = response.get("hashes")
                        if isinstance(hashes, list) and hashes:
                            hydrus_hash = hashes[0]

                if not hydrus_hash:
                    raise Exception(f"Hydrus response missing file hash: {response}")

                file_hash = hydrus_hash
                log(f"Hydrus: {file_hash}", file=sys.stderr)

            # Add tags if provided (both for new and existing files)
            if tag_list:
                try:
                    # Use default tag service
                    service_name = "my tags"
                except Exception:
                    service_name = "my tags"

                try:
                    debug(f"Adding {len(tag_list)} tag(s) to Hydrus: {tag_list}")
                    client.add_tag(file_hash, tag_list, service_name)
                    log(f"Tags added via '{service_name}'", file=sys.stderr)
                except Exception as exc:
                    log(f"⚠️  Failed to add tags: {exc}", file=sys.stderr)

            # Associate url if provided (both for new and existing files)
            if url:
                log(f"Associating {len(url)} URL(s) with file", file=sys.stderr)
                for url in url:
                    if url:
                        try:
                            client.associate_url(file_hash, str(url))
                            debug(f"Associated URL: {url}")
                        except Exception as exc:
                            log(f"⚠️  Failed to associate URL {url}: {exc}", file=sys.stderr)

            return file_hash

        except Exception as exc:
            log(f"❌ Hydrus upload failed: {exc}", file=sys.stderr)
            raise

    def search(self, query: str, **kwargs: Any) -> list[Dict[str, Any]]:
        """Search Hydrus database for files matching query.
        
        Args:
            query: Search query (tags, filenames, hashes, etc.)
            limit: Maximum number of results to return (default: 100)
            
        Returns:
            List of dicts with 'name', 'hash', 'size', 'tags' fields
            
        Example:
            results = storage["hydrus"].search("artist:john_doe music")
            results = storage["hydrus"].search("Simple Man")
        """
        limit = kwargs.get("limit", 100)

        try:
            client = self._client
            if client is None:
                raise Exception("Hydrus client unavailable")

            debug(f"Searching Hydrus for: {query}")

            # Parse the query into tags
            # Handle both simple tags and complex queries
            # "*" means "match all" - use system:everything tag in Hydrus
            if query.strip() == "*":
                # Use system:everything to match all files in Hydrus
                tags = ["system:everything"]
            else:
                query_lower = query.lower().strip()
                # If query doesn't have a namespace (no ':'), search all files and filter by title/tags
                # If query has explicit namespace, use it as a tag search
                if ':' not in query_lower:
                    # No namespace provided: search all files, then filter by title/tags containing the query
                    tags = ["system:everything"]
                else:
                    # User provided explicit namespace (e.g., "creator:john" or "system:has_audio")
                    # Use it as a tag search
                    tags = [query_lower]
            
            if not tags:
                debug(f"Found 0 result(s)")
                return []
            
            # Search files with the tags
            search_result = client.search_files(
                tags=tags,
                return_hashes=True,
                return_file_ids=True
            )
            
            # Extract file IDs from search result
            file_ids = search_result.get("file_ids", [])
            hashes = search_result.get("hashes", [])
            
            if not file_ids and not hashes:
                debug(f"Found 0 result(s)")
                return []
            
            # Fetch metadata for the found files
            results = []
            query_lower = query.lower().strip()
            # Split by comma or space for AND logic
            search_terms = set(query_lower.replace(',', ' ').split())  # For substring matching
            
            if file_ids:
                metadata = client.fetch_file_metadata(file_ids=file_ids)
                metadata_list = metadata.get("metadata", [])
                
                for meta in metadata_list:
                    if len(results) >= limit:
                        break
                    
                    file_id = meta.get("file_id")
                    hash_hex = meta.get("hash")
                    size = meta.get("size", 0)
                    
                    # Get tags for this file and extract title
                    tags_set = meta.get("tags", {})
                    all_tags = []
                    title = f"Hydrus File {file_id}"  # Default fallback
                    all_tags_str = ""  # For substring matching
                    
                    # debug(f"[HydrusBackend.search] Processing file_id={file_id}, tags type={type(tags_set)}")
                    
                    if isinstance(tags_set, dict):
                        # Collect both storage_tags and display_tags to capture siblings/parents and ensure title: is seen
                        def _collect(tag_list: Any) -> None:
                            nonlocal title, all_tags_str
                            if not isinstance(tag_list, list):
                                return
                            for tag in tag_list:
                                tag_text = str(tag) if tag else ""
                                if not tag_text:
                                    continue
                                all_tags.append(tag_text)
                                all_tags_str += " " + tag_text.lower()
                                if tag_text.lower().startswith("title:") and title == f"Hydrus File {file_id}":
                                    title = tag_text.split(":", 1)[1].strip()

                        for _service_name, service_tags in tags_set.items():
                            if not isinstance(service_tags, dict):
                                continue

                            storage_tags = service_tags.get("storage_tags", {})
                            if isinstance(storage_tags, dict):
                                for tag_list in storage_tags.values():
                                    _collect(tag_list)

                            display_tags = service_tags.get("display_tags", [])
                            _collect(display_tags)

                        # Also consider top-level flattened tags payload if provided (Hydrus API sometimes includes it)
                        top_level_tags = meta.get("tags_flat", []) or meta.get("tags", [])
                        _collect(top_level_tags)
                    
                    # Resolve extension from MIME type
                    mime_type = meta.get("mime")
                    ext = ""
                    if mime_type:
                        for category in mime_maps.values():
                            for _ext_key, info in category.items():
                                if mime_type in info.get("mimes", []):
                                    ext = info.get("ext", "").lstrip('.')
                                    break
                            if ext:
                                break

                    # Filter results based on query type
                    # If user provided explicit namespace (has ':'), don't do substring filtering
                    # Just include what the tag search returned
                    has_namespace = ':' in query_lower
                    
                    if has_namespace:
                        # Explicit namespace search - already filtered by Hydrus tag search
                        # Include this result as-is
                        file_url = f"{self._url.rstrip('/')}/get_files/file?hash={hash_hex}"
                        results.append({
                            "hash": hash_hex,
                            "url": file_url,
                            "name": title,
                            "title": title,
                            "size": size,
                            "size_bytes": size,
                            "store": self._instance_name,
                                "tag": all_tags,
                            "file_id": file_id,
                            "mime": mime_type,
                            "ext": ext,
                        })
                    else:
                        # Free-form search: check if search terms match the title or tags
                        # Match if ALL search terms are found in title or tags (AND logic)
                        # AND use whole word matching
                        
                        # Combine title and tags for searching
                        searchable_text = (title + " " + all_tags_str).lower()
                        
                        match = True
                        if query_lower != "*":
                            for term in search_terms:
                                # Regex for whole word: \bterm\b
                                # Escape term to handle special chars
                                pattern = r'\b' + re.escape(term) + r'\b'
                                if not re.search(pattern, searchable_text):
                                    match = False
                                    break
                        
                        if match:
                            file_url = f"{self._url.rstrip('/')}/get_files/file?hash={hash_hex}"
                            results.append({
                                "hash": hash_hex,
                                "url": file_url,
                                "name": title,
                                "title": title,
                                "size": size,
                                "size_bytes": size,
                                "store": self._instance_name,
                                "tag": all_tags,
                                "file_id": file_id,
                                "mime": mime_type,
                                "ext": ext,
                            })
            
            debug(f"Found {len(results)} result(s)")
            return results[:limit]

        except Exception as exc:
            log(f"❌ Hydrus search failed: {exc}", file=sys.stderr)
            import traceback
            traceback.print_exc(file=sys.stderr)
            raise

    def get_file(self, file_hash: str, **kwargs: Any) -> Path | str | None:
        """Open file in browser via Hydrus client API URL."""
        import webbrowser
        
        debug(f"[HydrusNetwork.get_file] Starting for hash: {file_hash[:12]}...")
        
        # Build browser URL with access key
        base_url = self._client.url.rstrip('/')
        access_key = self._client.access_key
        browser_url = f"{base_url}/get_files/file?hash={file_hash}&Hydrus-Client-API-Access-Key={access_key}"
        debug(f"[HydrusNetwork.get_file] Opening URL: {browser_url}")
        
        # Open in default browser
        webbrowser.open(browser_url)
        debug(f"[HydrusNetwork.get_file] Browser opened successfully")
        
        # Return the URL string instead of downloading
        debug(f"[HydrusNetwork.get_file] Returning URL: {browser_url}")
        return browser_url

    def get_metadata(self, file_hash: str, **kwargs: Any) -> Optional[Dict[str, Any]]:
        """Get metadata for a file from Hydrus by hash.
        
        Args:
            file_hash: SHA256 hash of the file (64-char hex string)
            
        Returns:
            Dict with metadata fields or None if not found
        """
        try:
            client = self._client
            if not client:
                debug("get_metadata: Hydrus client unavailable")
                return None
            
            # Fetch file metadata
            payload = client.fetch_file_metadata(hashes=[file_hash], include_service_keys_to_tags=True)
            
            if not payload or not payload.get("metadata"):
                return None
            
            meta = payload["metadata"][0]
            
            # Extract title from tags
            title = f"Hydrus_{file_hash[:12]}"
            tags_payload = meta.get("tags", {})
            if isinstance(tags_payload, dict):
                for service_data in tags_payload.values():
                    if isinstance(service_data, dict):
                        display_tags = service_data.get("display_tags", {})
                        if isinstance(display_tags, dict):
                            current_tags = display_tags.get("0", [])
                            if isinstance(current_tags, list):
                                for tag in current_tags:
                                    if str(tag).lower().startswith("title:"):
                                        title = tag.split(":", 1)[1].strip()
                                        break
                    if title != f"Hydrus_{file_hash[:12]}":
                        break
            
            # Determine extension from mime type
            mime_type = meta.get("mime", "")
            ext = ""
            if mime_type:
                from SYS.utils_constant import mime_maps
                for _category, extensions in mime_maps.items():
                    for extension, mime in extensions.items():
                        if mime == mime_type:
                            ext = extension.lstrip(".")
                            break
                    if ext:
                        break
            
            return {
                "hash": file_hash,
                "title": title,
                "ext": ext,
                "size": meta.get("size", 0),
                "mime": mime_type,
            }
            
        except Exception as exc:
            debug(f"Failed to get metadata from Hydrus: {exc}")
            return None

    def get_tag(self, file_identifier: str, **kwargs: Any) -> Tuple[List[str], str]:
        """Get tags for a file from Hydrus by hash.
        
        Args:
            file_identifier: File hash (SHA256 hex string)
            **kwargs: Optional service_name parameter
            
        Returns:
            Tuple of (tags_list, source_description)
            where source is always "hydrus"
        """
        try:
            from API import HydrusNetwork as hydrus_wrapper

            file_hash = str(file_identifier or "").strip().lower()
            if len(file_hash) != 64 or not all(ch in "0123456789abcdef" for ch in file_hash):
                debug(f"get_tags: invalid file hash '{file_identifier}'")
                return [], "unknown"
            
            # Get Hydrus client and service info
            client = self._client
            if not client:
                debug("get_tags: Hydrus client unavailable")
                return [], "unknown"
            
            # Fetch file metadata
            payload = client.fetch_file_metadata(
                hashes=[file_hash],
                include_service_keys_to_tags=True,
                include_file_url=False
            )
            
            items = payload.get("metadata") if isinstance(payload, dict) else None
            if not isinstance(items, list) or not items:
                debug(f"get_tags: No metadata returned for hash {file_hash}")
                return [], "unknown"
            
            meta = items[0] if isinstance(items[0], dict) else None
            if not isinstance(meta, dict) or meta.get("file_id") is None:
                debug(f"get_tags: Invalid metadata for hash {file_hash}")
                return [], "unknown"
            
            # Extract tags using service name
            service_name = "my tags"
            service_key = hydrus_wrapper.get_tag_service_key(client, service_name)
            
            # Extract tags from metadata
            tags = self._extract_tags_from_hydrus_meta(meta, service_key, service_name)
            
            return tags, "hydrus"
        
        except Exception as exc:
            debug(f"get_tags failed for Hydrus file: {exc}")
            return [], "unknown"

    def add_tag(self, file_identifier: str, tags: List[str], **kwargs: Any) -> bool:
        """Add tags to a Hydrus file.
        """
        try:
            client = self._client
            if client is None:
                debug("add_tag: Hydrus client unavailable")
                return False

            file_hash = str(file_identifier or "").strip().lower()
            if len(file_hash) != 64 or not all(ch in "0123456789abcdef" for ch in file_hash):
                debug(f"add_tag: invalid file hash '{file_identifier}'")
                return False
            service_name = kwargs.get("service_name") or "my tags"
            # Ensure tags is a list
            tag_list = list(tags) if isinstance(tags, (list, tuple)) else [str(tags)]
            if not tag_list:
                return False
            client.add_tag(file_hash, tag_list, service_name)
            return True
        except Exception as exc:
            debug(f"Hydrus add_tag failed: {exc}")
            return False

    def delete_tag(self, file_identifier: str, tags: List[str], **kwargs: Any) -> bool:
        """Delete tags from a Hydrus file.
        """
        try:
            client = self._client
            if client is None:
                debug("delete_tag: Hydrus client unavailable")
                return False

            file_hash = str(file_identifier or "").strip().lower()
            if len(file_hash) != 64 or not all(ch in "0123456789abcdef" for ch in file_hash):
                debug(f"delete_tag: invalid file hash '{file_identifier}'")
                return False
            service_name = kwargs.get("service_name") or "my tags"
            tag_list = list(tags) if isinstance(tags, (list, tuple)) else [str(tags)]
            if not tag_list:
                return False
            client.delete_tag(file_hash, tag_list, service_name)
            return True
        except Exception as exc:
            debug(f"Hydrus delete_tag failed: {exc}")
            return False

    def get_url(self, file_identifier: str, **kwargs: Any) -> List[str]:
        """Get known url for a Hydrus file.
        """
        try:
            client = self._client
            if client is None:
                debug("get_url: Hydrus client unavailable")
                return []

            file_hash = str(file_identifier or "").strip().lower()
            if len(file_hash) != 64 or not all(ch in "0123456789abcdef" for ch in file_hash):
                return []

            payload = client.fetch_file_metadata(hashes=[file_hash], include_file_url=True)
            items = payload.get("metadata") if isinstance(payload, dict) else None
            if not isinstance(items, list) or not items:
                return []
            meta = items[0]
            url = meta.get("url") or []
            return list(url)
        except Exception as exc:
            debug(f"Hydrus get_url failed: {exc}")
            return []

    def add_url(self, file_identifier: str, url: List[str], **kwargs: Any) -> bool:
        """Associate one or more url with a Hydrus file.
        """
        try:
            client = self._client
            if client is None:
                debug("add_url: Hydrus client unavailable")
                return False
            for u in url:
                client.associate_url(file_identifier, u)
            return True
        except Exception as exc:
            debug(f"Hydrus add_url failed: {exc}")
            return False

    def delete_url(self, file_identifier: str, url: List[str], **kwargs: Any) -> bool:
        """Delete one or more url from a Hydrus file.
        """
        try:
            client = self._client
            if client is None:
                debug("delete_url: Hydrus client unavailable")
                return False
            for u in url:
                client.delete_url(file_identifier, u)
            return True
        except Exception as exc:
            debug(f"Hydrus delete_url failed: {exc}")
            return False

    def get_note(self, file_identifier: str, **kwargs: Any) -> Dict[str, str]:
        """Get notes for a Hydrus file (default note service only)."""
        try:
            client = self._client
            if client is None:
                debug("get_note: Hydrus client unavailable")
                return {}

            file_hash = str(file_identifier or "").strip().lower()
            if len(file_hash) != 64 or not all(ch in "0123456789abcdef" for ch in file_hash):
                return {}

            payload = client.fetch_file_metadata(hashes=[file_hash], include_notes=True)
            items = payload.get("metadata") if isinstance(payload, dict) else None
            if not isinstance(items, list) or not items:
                return {}
            meta = items[0] if isinstance(items[0], dict) else None
            if not isinstance(meta, dict):
                return {}

            notes_payload = meta.get("notes")
            if isinstance(notes_payload, dict):
                return {str(k): str(v or "") for k, v in notes_payload.items() if str(k).strip()}

            return {}
        except Exception as exc:
            debug(f"Hydrus get_note failed: {exc}")
            return {}

    def set_note(self, file_identifier: str, name: str, text: str, **kwargs: Any) -> bool:
        """Set a named note for a Hydrus file (default note service only)."""
        try:
            client = self._client
            if client is None:
                debug("set_note: Hydrus client unavailable")
                return False

            file_hash = str(file_identifier or "").strip().lower()
            if len(file_hash) != 64 or not all(ch in "0123456789abcdef" for ch in file_hash):
                return False

            note_name = str(name or "").strip()
            if not note_name:
                return False
            note_text = str(text or "")

            client.set_notes(file_hash, {note_name: note_text})
            return True
        except Exception as exc:
            debug(f"Hydrus set_note failed: {exc}")
            return False

    def delete_note(self, file_identifier: str, name: str, **kwargs: Any) -> bool:
        """Delete a named note for a Hydrus file (default note service only)."""
        try:
            client = self._client
            if client is None:
                debug("delete_note: Hydrus client unavailable")
                return False

            file_hash = str(file_identifier or "").strip().lower()
            if len(file_hash) != 64 or not all(ch in "0123456789abcdef" for ch in file_hash):
                return False

            note_name = str(name or "").strip()
            if not note_name:
                return False

            client.delete_notes(file_hash, [note_name])
            return True
        except Exception as exc:
            debug(f"Hydrus delete_note failed: {exc}")
            return False

    @staticmethod
    def _extract_tags_from_hydrus_meta(
        meta: Dict[str, Any],
        service_key: Optional[str],
        service_name: str
    ) -> List[str]:
        """Extract current tags from Hydrus metadata dict.
        
        Prefers display_tags (includes siblings/parents, excludes deleted).
        Falls back to storage_tags status '0' (current).
        """
        tags_payload = meta.get("tags")
        if not isinstance(tags_payload, dict):
            return []
        
        svc_data = None
        if service_key:
            svc_data = tags_payload.get(service_key)
        if not isinstance(svc_data, dict):
            return []
        
        # Prefer display_tags (Hydrus computes siblings/parents)
        display = svc_data.get("display_tags")
        if isinstance(display, list) and display:
            return [str(t) for t in display if isinstance(t, (str, bytes)) and str(t).strip()]
        
        # Fallback to storage_tags status '0' (current)
        storage = svc_data.get("storage_tags")
        if isinstance(storage, dict):
            current_list = storage.get("0") or storage.get(0)
            if isinstance(current_list, list):
                return [str(t) for t in current_list if isinstance(t, (str, bytes)) and str(t).strip()]
        
        return []