goyimnose/Medios-Macina
1
0
Fork 0
Code Issues Pull Requests Packages Releases 1 Wiki Activity

f

Browse Source
This commit is contained in:
Nose
2026-01-24 01:38:12 -08:00
parent 4e4c374908
commit 3a4d3f029d
5 changed files with 210 additions and 229 deletions
Download Patch File Download Diff File Expand all files Collapse all files

172
cmdlet/get_url.py
View File

@@ -241,95 +241,32 @@ class Get_Url(Cmdlet):
return None
@staticmethod
def _resolve_title_for_hash(backend: Any, file_hash: str, hit: Any = None) -> str:
"""Best-effort title resolution for a found hash.
Strategy:
- Use the hit's existing title/columns when present.
- Prefer backend.get_metadata(hash) when available (direct lookup).
- Fallback to backend.search('hash:<sha>', limit=1) and read title.
"""
try:
if hit is not None:
from_hit = Get_Url._extract_title_from_result(hit)
if from_hit:
return from_hit
except Exception:
pass
try:
if hasattr(backend, "get_metadata"):
meta = backend.get_metadata(file_hash)
if isinstance(meta, dict):
t = meta.get("title")
if isinstance(t, str) and t.strip():
return t.strip()
except Exception:
pass
try:
if hasattr(backend, "search"):
hits = backend.search(f"hash:{file_hash}", limit=1)
if isinstance(hits, list) and hits:
t2 = Get_Url._extract_title_from_result(hits[0])
if t2:
return t2
except Exception:
pass
return ""
def _extract_size_from_hit(hit: Any) -> int | None:
for key in ("size", "file_size", "filesize", "size_bytes"):
try:
val = get_field(hit, key)
except Exception:
val = None
if val is None:
continue
if isinstance(val, (int, float)):
return int(val)
try:
return int(val)
except Exception:
continue
return None
@staticmethod
def _resolve_size_ext_for_hash(backend: Any, file_hash: str, hit: Any = None) -> tuple[int | None, str]:
"""Best-effort (size, ext) resolution for a found hash."""
# First: see if the hit already includes these fields.
try:
size_val = get_field(hit, "size")
if size_val is None:
size_val = get_field(hit, "file_size")
if size_val is None:
size_val = get_field(hit, "filesize")
if size_val is None:
size_val = get_field(hit, "size_bytes")
size_int = int(size_val) if isinstance(size_val, (int, float)) else None
except Exception:
size_int = None
try:
ext_val = get_field(hit, "ext")
if ext_val is None:
ext_val = get_field(hit, "extension")
ext = str(ext_val).strip().lstrip(".") if isinstance(ext_val, str) else ""
except Exception:
ext = ""
if size_int is not None or ext:
return size_int, ext
# Next: backend.get_metadata(hash) when available.
try:
if hasattr(backend, "get_metadata"):
meta = backend.get_metadata(file_hash)
if isinstance(meta, dict):
size_val2 = meta.get("size")
if size_val2 is None:
size_val2 = meta.get("file_size")
if size_val2 is None:
size_val2 = meta.get("filesize")
if size_val2 is None:
size_val2 = meta.get("size_bytes")
if isinstance(size_val2, (int, float)):
size_int = int(size_val2)
ext_val2 = meta.get("ext")
if ext_val2 is None:
ext_val2 = meta.get("extension")
if isinstance(ext_val2, str) and ext_val2.strip():
ext = ext_val2.strip().lstrip(".")
except Exception:
pass
return size_int, ext
def _extract_ext_from_hit(hit: Any) -> str:
for key in ("ext", "extension"):
try:
ext_val = get_field(hit, key)
except Exception:
ext_val = None
if isinstance(ext_val, str) and ext_val.strip():
return ext_val.strip().lstrip(".")
return ""
def _search_urls_across_stores(self,
pattern: str,
@@ -360,9 +297,6 @@ class Get_Url(Cmdlet):
try:
backend = storage[store_name]
title_cache: Dict[str, str] = {}
meta_cache: Dict[str, tuple[int | None, str]] = {}
# Search only URL-bearing records using the backend's URL search capability.
# This avoids the expensive/incorrect "search('*')" scan.
try:
@@ -431,22 +365,12 @@ class Get_Url(Cmdlet):
search_limit,
store_name,
pattern_hint=target_pattern,
minimal=True,
)
if search_results is None:
continue
search_results = search_results or []
if not search_results and target_pattern and not has_wildcards:
fallback_results = self._execute_search_with_timeout(
backend,
"url:*",
search_limit,
store_name,
pattern_hint=target_pattern,
)
if fallback_results is None:
continue
search_results = fallback_results or []
for hit in (search_results or []):
if len(items) >= MAX_RESULTS:
@@ -459,44 +383,9 @@ class Get_Url(Cmdlet):
file_hash = str(file_hash)
title = title_cache.get(file_hash, "")
if not title:
try:
title = (
get_field(hit, "title")
or get_field(hit, "name")
or get_field(hit, "file_title")
or ""
)
except Exception:
title = ""
if not title:
title = self._resolve_title_for_hash(backend, file_hash, hit)
title_cache[file_hash] = title
size, ext = meta_cache.get(file_hash, (None, ""))
if size is None and not ext:
try:
size = get_field(hit, "size")
if size is None:
size = get_field(hit, "size_bytes")
if size is None:
size = get_field(hit, "file_size")
if size is None:
size = get_field(hit, "filesize")
size = int(size) if isinstance(size, (int, float)) else None
except Exception:
size = None
try:
ext = get_field(hit, "ext") or get_field(hit, "extension")
ext = str(ext).strip().lstrip(".") if isinstance(ext, str) else ""
except Exception:
ext = ""
if size is None and not ext:
size, ext = self._resolve_size_ext_for_hash(backend, file_hash, hit)
meta_cache[file_hash] = (size, ext)
title = self._extract_title_from_result(hit) or ""
size = self._extract_size_from_hit(hit)
ext = self._extract_ext_from_hit(hit)
urls = self._extract_urls_from_hit(hit)
if not urls:
@@ -505,6 +394,7 @@ class Get_Url(Cmdlet):
except Exception:
urls = []
hit_added = False
for url in (urls or []):
if len(items) >= MAX_RESULTS:
break
@@ -526,7 +416,9 @@ class Get_Url(Cmdlet):
ext=str(ext or ""),
)
)
found_stores.add(str(store_name))
hit_added = True
if hit_added:
found_stores.add(str(store_name))
if len(items) >= MAX_RESULTS:
break
except Exception as exc: