dfdkflj
This commit is contained in:
nose
@@ -1,139 +1,80 @@
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any, Dict, Sequence
|
||||
import json
|
||||
import sys
|
||||
from pathlib import Path
|
||||
|
||||
from . import register
|
||||
import models
|
||||
import pipeline as ctx
|
||||
from helper import hydrus as hydrus_wrapper
|
||||
from ._shared import Cmdlet, CmdletArg, normalize_hash
|
||||
from ._shared import Cmdlet, CmdletArg, SharedArgs, parse_cmdlet_args, get_field, normalize_hash
|
||||
from helper.logger import log
|
||||
from config import get_local_storage_path
|
||||
from helper.local_library import LocalLibraryDB
|
||||
|
||||
CMDLET = Cmdlet(
|
||||
name="get-url",
|
||||
summary="List URLs associated with a file (Hydrus or Local).",
|
||||
usage="get-url [-hash <sha256>]",
|
||||
args=[
|
||||
CmdletArg("-hash", description="Override the Hydrus file hash (SHA256) to target instead of the selected result."),
|
||||
],
|
||||
details=[
|
||||
"- Prints the known URLs for the selected file.",
|
||||
],
|
||||
)
|
||||
from helper.store import FileStorage
|
||||
|
||||
|
||||
def _parse_hash_and_rest(args: Sequence[str]) -> tuple[str | None, list[str]]:
|
||||
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
|
||||
return override_hash, rest
|
||||
|
||||
|
||||
@register(["get-url", "get-urls", "get_url"]) # aliases
|
||||
def get_urls(result: Any, args: Sequence[str], config: Dict[str, Any]) -> int:
|
||||
# Helper to get field from both dict and object
|
||||
def get_field(obj: Any, field: str, default: Any = None) -> Any:
|
||||
if isinstance(obj, dict):
|
||||
return obj.get(field, default)
|
||||
else:
|
||||
return getattr(obj, field, default)
|
||||
class Get_Url(Cmdlet):
|
||||
"""Get url associated with files via hash+store."""
|
||||
|
||||
# Help
|
||||
try:
|
||||
if any(str(a).lower() in {"-?", "/?", "--help", "-h", "help", "--cmdlet"} for a in args):
|
||||
log(json.dumps(CMDLET, ensure_ascii=False, indent=2))
|
||||
return 0
|
||||
except Exception:
|
||||
pass
|
||||
NAME = "get-url"
|
||||
SUMMARY = "List url associated with a file"
|
||||
USAGE = "@1 | get-url"
|
||||
ARGS = [
|
||||
SharedArgs.HASH,
|
||||
SharedArgs.STORE,
|
||||
]
|
||||
DETAIL = [
|
||||
"- Lists all url associated with file identified by hash+store",
|
||||
]
|
||||
|
||||
override_hash, _ = _parse_hash_and_rest(args)
|
||||
|
||||
# Handle @N selection which creates a list - extract the first item
|
||||
if isinstance(result, list) and len(result) > 0:
|
||||
result = result[0]
|
||||
|
||||
found_urls = []
|
||||
|
||||
# 1. Try Local Library
|
||||
file_path = get_field(result, "file_path") or get_field(result, "path")
|
||||
if file_path and not override_hash:
|
||||
try:
|
||||
path_obj = Path(file_path)
|
||||
if path_obj.exists():
|
||||
storage_path = get_local_storage_path(config)
|
||||
if storage_path:
|
||||
with LocalLibraryDB(storage_path) as db:
|
||||
metadata = db.get_metadata(path_obj)
|
||||
if metadata and metadata.get("known_urls"):
|
||||
found_urls.extend(metadata["known_urls"])
|
||||
except Exception as e:
|
||||
log(f"Error checking local library: {e}", file=sys.stderr)
|
||||
|
||||
# 2. Try Hydrus
|
||||
hash_hex = normalize_hash(override_hash) if override_hash else normalize_hash(get_field(result, "hash_hex", None))
|
||||
|
||||
# If we haven't found URLs yet, or if we want to merge them (maybe?), let's check Hydrus if we have a hash
|
||||
# But usually if it's local, we might not want to check Hydrus unless requested.
|
||||
# However, the user said "they can just work together".
|
||||
|
||||
if hash_hex:
|
||||
try:
|
||||
client = hydrus_wrapper.get_client(config)
|
||||
if client:
|
||||
payload = client.fetch_file_metadata(hashes=[hash_hex], include_file_urls=True)
|
||||
items = payload.get("metadata") if isinstance(payload, dict) else None
|
||||
meta = items[0] if (isinstance(items, list) and items and isinstance(items[0], dict)) else None
|
||||
hydrus_urls = (meta.get("known_urls") if isinstance(meta, dict) else None) or []
|
||||
for u in hydrus_urls:
|
||||
if u not in found_urls:
|
||||
found_urls.append(u)
|
||||
except Exception as exc:
|
||||
# Only log error if we didn't find local URLs either, or if it's a specific error
|
||||
if not found_urls:
|
||||
log(f"Hydrus lookup failed: {exc}", file=sys.stderr)
|
||||
|
||||
if found_urls:
|
||||
for u in found_urls:
|
||||
text = str(u).strip()
|
||||
if text:
|
||||
# Emit a rich object that looks like a string but carries context
|
||||
# We use a dict with 'title' which ResultTable uses for display
|
||||
# and 'url' which is the actual data
|
||||
# We also include the source file info so downstream cmdlets can use it
|
||||
|
||||
# Create a result object that mimics the structure expected by delete-url
|
||||
# delete-url expects a file object usually, but here we are emitting URLs.
|
||||
# If we emit a dict with 'url' and 'source_file', delete-url can use it.
|
||||
|
||||
rich_result = {
|
||||
"title": text, # Display as just the URL
|
||||
"url": text,
|
||||
"source_file": result, # Pass the original file context
|
||||
"file_path": get_field(result, "file_path") or get_field(result, "path"),
|
||||
"hash_hex": hash_hex
|
||||
}
|
||||
ctx.emit(rich_result)
|
||||
return 0
|
||||
|
||||
if not hash_hex and not file_path:
|
||||
log("Selected result does not include a file path or Hydrus hash", file=sys.stderr)
|
||||
return 1
|
||||
def run(self, result: Any, args: Sequence[str], config: Dict[str, Any]) -> int:
|
||||
"""Get url for file via hash+store backend."""
|
||||
parsed = parse_cmdlet_args(args, self)
|
||||
|
||||
ctx.emit("No URLs found.")
|
||||
return 0
|
||||
# Extract hash and store from result or args
|
||||
file_hash = parsed.get("hash") or get_field(result, "hash")
|
||||
store_name = parsed.get("store") or get_field(result, "store")
|
||||
|
||||
if not file_hash:
|
||||
log("Error: No file hash provided")
|
||||
return 1
|
||||
|
||||
if not store_name:
|
||||
log("Error: No store name provided")
|
||||
return 1
|
||||
|
||||
# Normalize hash
|
||||
file_hash = normalize_hash(file_hash)
|
||||
if not file_hash:
|
||||
log("Error: Invalid hash format")
|
||||
return 1
|
||||
|
||||
# Get backend and retrieve url
|
||||
try:
|
||||
storage = FileStorage(config)
|
||||
backend = storage[store_name]
|
||||
|
||||
url = backend.get_url(file_hash)
|
||||
|
||||
if url:
|
||||
for url in url:
|
||||
# Emit rich object for pipeline compatibility
|
||||
ctx.emit({
|
||||
"url": url,
|
||||
"hash": file_hash,
|
||||
"store": store_name,
|
||||
})
|
||||
return 0
|
||||
else:
|
||||
ctx.emit("No url found")
|
||||
return 0
|
||||
|
||||
except KeyError:
|
||||
log(f"Error: Storage backend '{store_name}' not configured")
|
||||
return 1
|
||||
except Exception as exc:
|
||||
log(f"Error retrieving url: {exc}", file=sys.stderr)
|
||||
return 1
|
||||
|
||||
|
||||
# Register cmdlet
|
||||
register(["get-url", "get_url"])(Get_Url)
|
||||
|
||||
|
||||
|
||||
Reference in New Issue
Block a user