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

h

Browse Source
This commit is contained in:
Nose
2026-01-31 19:00:04 -08:00
parent dcf16e0cc4
commit 6513a3ad04
25 changed files with 617 additions and 397 deletions
Download Patch File Download Diff File Expand all files Collapse all files

87
tool/florencevision.py
View File

@@ -428,7 +428,8 @@ class FlorenceVisionTool:
if self._model is not None and not hasattr(self._model, "_supports_sdpa"):
setattr(self._model, "_supports_sdpa", False)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to set model compatibility flag _supports_sdpa")
try:
self._model.to(device) # type: ignore[union-attr]
@@ -439,7 +440,8 @@ class FlorenceVisionTool:
try:
self._model.eval() # type: ignore[union-attr]
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to set Florence model to eval mode")
try:
md = getattr(self._model, "device", None)
@@ -450,7 +452,8 @@ class FlorenceVisionTool:
dt = None
debug(f"[florencevision] Model loaded: device={md} param_dtype={dt}")
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to inspect Florence model device/dtype")
def tags_for_image(self, media_path: Path) -> List[str]:
"""Return Florence-derived tags for an image.
@@ -472,7 +475,8 @@ class FlorenceVisionTool:
try:
debug(f"[florencevision] Task prompt: {prompt}")
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to emit debug Task prompt for FlorenceVision")
max_tags = max(0, int(self.defaults.max_tags or 0))
@@ -487,7 +491,8 @@ class FlorenceVisionTool:
try:
debug(f"[florencevision] Image loaded: mode={image.mode} size={image.width}x{image.height}")
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to emit debug for image load")
processor = self._processor
model = self._model
@@ -544,19 +549,22 @@ class FlorenceVisionTool:
)
continue
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to debug tensor shape for processor key '%s'", k)
if isinstance(v, (list, tuple)):
has_none = any(x is None for x in v)
debug(f"[florencevision] {k}: {type(v).__name__} len={len(v)} has_none={has_none}")
continue
debug(f"[florencevision] {k}: type={type(v).__name__}")
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed while inspecting processor output keys")
try:
inputs = inputs.to(model.device) # type: ignore[attr-defined]
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to move processor inputs to device %s", getattr(model, 'device', None))
# Align floating-point input tensors with the model's parameter dtype.
try:
@@ -575,7 +583,8 @@ class FlorenceVisionTool:
except Exception:
continue
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to inspect/align model dtype for Florence inputs")
try:
gen_inputs_all = {k: v for k, v in dict(inputs).items() if v is not None}
@@ -602,7 +611,8 @@ class FlorenceVisionTool:
):
gen_inputs["attention_mask"] = attention_mask
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to reconcile attention mask shape with input_ids for Florence processor")
try:
debug(
@@ -612,18 +622,21 @@ class FlorenceVisionTool:
f"pixel_attention_mask={'pixel_attention_mask' in forward_params}"
)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to debug model forward supports")
try:
gen_inputs.setdefault("use_cache", False)
gen_inputs.setdefault("num_beams", 1)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to set default gen_inputs values")
try:
debug(f"[florencevision] generate kwargs: {sorted(list(gen_inputs.keys()))}")
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to debug generate kwargs")
pv = gen_inputs.get("pixel_values")
if pv is None:
@@ -654,7 +667,8 @@ class FlorenceVisionTool:
if not hasattr(model, "_supports_sdpa"):
setattr(model, "_supports_sdpa", False)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to patch model _supports_sdpa flag in retry handler")
generated_ids = _do_generate(gen_inputs)
elif "NoneType" in msg and "shape" in msg:
retry_inputs = dict(gen_inputs)
@@ -676,7 +690,8 @@ class FlorenceVisionTool:
):
retry_inputs["attention_mask"] = am
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed while filling retry_inputs attention_mask in AttributeError handler")
try:
import torch
@@ -692,14 +707,16 @@ class FlorenceVisionTool:
elif "pixel_attention_mask" in forward_params and "pixel_attention_mask" not in retry_inputs:
retry_inputs["pixel_attention_mask"] = mask
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to build mask or adjust retry_inputs in AttributeError handler")
try:
debug(
f"[florencevision] generate retry kwargs: {sorted(list(retry_inputs.keys()))}"
)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to debug generate retry kwargs")
generated_ids = _do_generate(retry_inputs)
else:
@@ -708,7 +725,8 @@ class FlorenceVisionTool:
try:
debug(f"[florencevision] generated_ids type={type(generated_ids).__name__}")
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to debug generated_ids type")
seq = getattr(generated_ids, "sequences", generated_ids)
generated_text = processor.batch_decode(seq, skip_special_tokens=False)[0]
@@ -719,7 +737,8 @@ class FlorenceVisionTool:
debug(f"[florencevision] prompt run failed: {type(exc).__name__}: {exc}")
debug("[florencevision] traceback:\n" + traceback.format_exc())
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to emit debug for prompt run failure: %s", exc)
raise
parsed = None
@@ -766,12 +785,14 @@ class FlorenceVisionTool:
debug(f"[florencevision] post_process[{k!r}] type={type(parsed.get(k)).__name__}")
debug("[florencevision] post_process[key] repr:\n" + _debug_repr(parsed.get(k)))
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed while debugging parsed post_process output for prompt %s", task_prompt)
else:
debug(f"[florencevision] post_process_generation: type={type(parsed).__name__}")
debug("[florencevision] post_process repr:\n" + _debug_repr(parsed))
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to post-process generated output for prompt %s", task_prompt)
return generated_text, parsed, seq
@@ -800,7 +821,8 @@ class FlorenceVisionTool:
try:
debug(f"[florencevision] candidate label strings ({len(labels)}): {labels!r}")
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to emit candidate label strings debug")
out: List[str] = []
seen: set[str] = set()
@@ -848,7 +870,8 @@ class FlorenceVisionTool:
for raw_lab, cleaned, reason in dropped:
debug(f"[florencevision] drop reason={reason} raw={raw_lab!r} cleaned={cleaned!r}")
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to emit cleaned/dropped tags debug info")
return labels, caption_candidates, out, dropped
@@ -871,7 +894,12 @@ class FlorenceVisionTool:
try:
return max(cleaned, key=lambda s: len(str(s)), default=None)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to choose best caption from cleaned candidates")
try:
return max(raw, key=lambda s: len(str(s)), default=None)
except Exception:
return None
try:
return max(raw, key=lambda s: len(str(s)), default=None)
except Exception:
@@ -936,7 +964,8 @@ class FlorenceVisionTool:
try:
debug(f"[florencevision] grounding prompt: {grounding_prompt}")
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to emit grounding prompt debug")
grd_text, grd_parsed, _grd_seq = _run_prompt(grounding_prompt)
_grd_labels, grd_captions, grd_cleaned, _grd_dropped = _extract_labels_and_captions(grounding_prompt, grd_text, grd_parsed)
@@ -962,6 +991,8 @@ class FlorenceVisionTool:
is_combo = "<|detailed_caption|>" in prompt and "<|grounding|>" in prompt
only_task_tokens = not final_tags or all(t in {"object_detection", "grounding", "tag"} for t in final_tags)
except Exception:
from SYS.logger import logger
logger.exception("Failed to compute is_combo/only_task_tokens for prompt '%s'", prompt)
is_combo = False
only_task_tokens = False
@@ -973,13 +1004,15 @@ class FlorenceVisionTool:
try:
self.defaults.task = "<|detailed_caption|>"
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to set self.defaults.task to '<|detailed_caption|>' during od retry")
final_tags = self.tags_for_image(media_path)
finally:
try:
self.defaults.task = original_task
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to restore self.defaults.task after od retry")
self._od_tag_retrying = False
self._last_caption = caption_text if caption_text else None

21
tool/playwright.py
View File

@@ -48,7 +48,8 @@ def _resolve_out_dir(arg_outdir: Optional[Union[str, Path]]) -> Path:
try:
p.mkdir(parents=True, exist_ok=True)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to create resolved output dir %s", p)
return p
except Exception:
return Path(tempfile.mkdtemp(prefix="pwdl_"))
@@ -425,17 +426,20 @@ def config_schema() -> List[Dict[str, Any]]:
if context is not None:
context.close()
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to close Playwright context")
try:
if browser is not None:
browser.close()
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to close Playwright browser")
try:
if pw is not None:
pw.stop()
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to stop Playwright engine")
def goto(self, page: Any, url: str) -> None:
"""Navigate with configured timeout."""
@@ -564,7 +568,8 @@ def config_schema() -> List[Dict[str, Any]]:
page.mouse.move(box['x'] + box['width'] / 2, box['y'] + box['height'] / 2)
page.mouse.click(box['x'] + box['width'] / 2, box['y'] + box['height'] / 2)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to perform mouse click for selector '%s'", selector)
resp = page.wait_for_response(
lambda r: r.status == 200 and any(k.lower() == 'content-disposition' for k in r.headers.keys()),
@@ -583,7 +588,8 @@ def config_schema() -> List[Dict[str, Any]]:
try:
debug(f"[playwright] attempt failed (headless={mode}): {traceback.format_exc()}")
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to emit debug info for Playwright attempt failure")
continue
return PlaywrightDownloadResult(ok=False, error=last_error or "no download captured")
@@ -596,7 +602,8 @@ def config_schema() -> List[Dict[str, Any]]:
f"nav_timeout_ms={self.defaults.navigation_timeout_ms}"
)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to debug_dump Playwright defaults")
def _wait_for_block_clear(self, page: Any, timeout_ms: int = 8000) -> bool:
try:

72
tool/ytdlp.py
View File

@@ -131,7 +131,8 @@ def _build_supported_domains() -> set[str]:
domains = extract_domains(regex)
_SUPPORTED_DOMAINS.update(domains)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to build supported domains from yt-dlp extractors")
return _SUPPORTED_DOMAINS
@@ -299,7 +300,8 @@ def _add_browser_cookies_if_available(options: Dict[str, Any], preferred_browser
log(f"Requested browser cookie DB '{preferred_browser}' not found; falling back to autodetect.")
_BROWSER_COOKIE_WARNING_EMITTED = True
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to check browser cookie path for preferred browser '%s'", preferred_browser)
# Auto-detect in common order (chrome/chromium/brave)
for candidate in ("chrome", "chromium", "brave"):
@@ -308,6 +310,8 @@ def _add_browser_cookies_if_available(options: Dict[str, Any], preferred_browser
options["cookiesfrombrowser"] = [candidate]
return
except Exception:
from SYS.logger import logger
logger.exception("Error while checking cookie path for candidate browser '%s'", candidate)
continue
if not _BROWSER_COOKIE_WARNING_EMITTED:
@@ -650,7 +654,8 @@ def format_for_table_selection(
if vcodec != "none" and acodec == "none" and format_id:
selection_format_id = f"{format_id}+ba"
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to compute selection_format_id for format: %s", fmt)
# Format file size
size_str = ""
@@ -661,7 +666,8 @@ def format_for_table_selection(
size_mb = float(size_bytes) / (1024 * 1024)
size_str = f"{size_prefix}{size_mb:.1f}MB"
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to compute size string for format: %s", fmt)
# Build description
desc_parts: List[str] = []
@@ -755,7 +761,8 @@ class YtDlpTool:
if resolved is not None and resolved.is_file():
return resolved
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to initialize cookiefile using resolve_cookies_path")
return None
def resolve_height_selector(self, format_str: Optional[str]) -> Optional[str]:
@@ -908,13 +915,15 @@ class YtDlpTool:
if bundled_ffmpeg_dir.exists():
base_options.setdefault("ffmpeg_location", str(bundled_ffmpeg_dir))
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to inspect bundled ffmpeg directory")
try:
if os.name == "nt":
base_options.setdefault("file_access_retries", 40)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to set Windows-specific yt-dlp options")
if opts.cookies_path and opts.cookies_path.is_file():
base_options["cookiefile"] = str(opts.cookies_path)
@@ -948,7 +957,8 @@ class YtDlpTool:
opts = _dc.replace(opts, mode="audio", ytdl_format=None)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to set opts mode to audio via dataclasses.replace")
elif opts.ytdl_format == "video":
try:
opts = opts._replace(mode="video", ytdl_format=None)
@@ -958,7 +968,8 @@ class YtDlpTool:
opts = _dc.replace(opts, mode="video", ytdl_format=None)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to set opts mode to video via dataclasses.replace")
if opts.no_playlist:
base_options["noplaylist"] = True
@@ -978,7 +989,8 @@ class YtDlpTool:
opts = _dc.replace(opts, mode="audio")
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to set opts mode to audio via dataclasses.replace (configured default)")
ytdl_format = None
else:
# Leave ytdl_format None so that default_format(opts.mode)
@@ -1130,7 +1142,8 @@ class YtDlpTool:
try:
debug("yt-dlp argv: " + " ".join(str(a) for a in argv))
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to debug-print yt-dlp CLI arguments")
def config_schema() -> List[Dict[str, Any]]:
@@ -1150,6 +1163,8 @@ def config_schema() -> List[Dict[str, Any]]:
if _browser_cookie_path_for(b) is not None:
browser_choices.append(b)
except Exception:
from SYS.logger import logger
logger.exception("Error while checking cookie path for browser '%s'", b)
continue
return [
@@ -1410,7 +1425,8 @@ def _download_with_sections_via_cli(
try:
_set_pipe_percent(current)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to set pipeline percent to %d", current)
def start(self) -> None:
if self._thread is not None or self._start_pct >= self._max_pct:
@@ -1426,7 +1442,8 @@ def _download_with_sections_via_cli(
try:
_set_pipe_percent(self._max_pct)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to set pipeline percent to max %d", self._max_pct)
session_id = hashlib.md5((url + str(time.time()) + "".join(random.choices(string.ascii_letters, k=10))).encode()).hexdigest()[:12]
first_section_info = None
@@ -1440,7 +1457,8 @@ def _download_with_sections_via_cli(
try:
_set_pipe_percent(display_pct)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to set pipeline percent to display_pct %d for section %d", display_pct, section_idx)
pipeline.set_status(f"Downloading & clipping clip section {section_idx}/{total_sections}")
@@ -1484,7 +1502,8 @@ def _download_with_sections_via_cli(
try:
cmd.extend(["--ffmpeg-location", str(ytdl_options["ffmpeg_location"])])
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to append ffmpeg_location CLI option")
if ytdl_options.get("format"):
cmd.extend(["-f", ytdl_options["format"]])
if ytdl_options.get("merge_output_format"):
@@ -1547,7 +1566,8 @@ def _download_with_sections_via_cli(
try:
_set_pipe_percent(99)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to set pipeline percent to 99 at end of multi-section job")
return session_id, first_section_info or {}
@@ -1654,7 +1674,8 @@ def _progress_callback(status: Dict[str, Any]) -> None:
if isinstance(value, (int, float)) and value > 0:
return int(value)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to interpret total bytes value: %r", value)
return None
if event == "downloading":
@@ -1669,7 +1690,8 @@ def _progress_callback(status: Dict[str, Any]) -> None:
total=_total_bytes(total),
)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to update pipeline transfer for label '%s'", label)
else:
_YTDLP_PROGRESS_BAR.update(
downloaded=int(downloaded) if downloaded is not None else None,
@@ -1683,7 +1705,8 @@ def _progress_callback(status: Dict[str, Any]) -> None:
if _YTDLP_TRANSFER_STATE.get(label, {}).get("started"):
pipeline.finish_transfer(label=label)
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to finish pipeline transfer for label '%s'", label)
_YTDLP_TRANSFER_STATE.pop(label, None)
else:
_YTDLP_PROGRESS_BAR.finish()
@@ -1848,7 +1871,8 @@ def download_media(opts: DownloadOptions, *, config: Optional[Dict[str, Any]] =
if cand.suffix.lower() in {".json", ".info.json"}:
continue
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to inspect candidate suffix for %s", cand)
media_file = cand
break
if media_file is None and media_candidates:
@@ -1870,10 +1894,13 @@ def download_media(opts: DownloadOptions, *, config: Optional[Dict[str, Any]] =
if name.startswith(prefix):
return name[len(prefix):]
except Exception:
pass
from SYS.logger import logger
logger.exception("Failed to check name prefix for '%s'", name)
try:
return Path(name).suffix
except Exception:
from SYS.logger import logger
logger.exception("Failed to obtain suffix for name '%s'", name)
return ""
try:
@@ -1884,7 +1911,8 @@ def download_media(opts: DownloadOptions, *, config: Optional[Dict[str, Any]] =
try:
media_file.unlink()
except OSError:
pass
from SYS.logger import logger
logger.exception("Failed to unlink duplicate media file %s", media_file)
else:
media_file.rename(new_media_path)
debug(f"Renamed section file: {media_file.name} -> {new_media_name}")