Apply DB cleanup and per-file logging; bucket decisions for the caller.
(
decisions: list[Decision],
s: store.Store,
*,
filter_db: tuple[str, ...] = (),
)
| 203 | |
| 204 | |
| 205 | def apply_decisions( |
| 206 | decisions: list[Decision], |
| 207 | s: store.Store, |
| 208 | *, |
| 209 | filter_db: tuple[str, ...] = (), |
| 210 | ) -> Classified: |
| 211 | """Apply DB cleanup and per-file logging; bucket decisions for the caller.""" |
| 212 | out = Classified() |
| 213 | for d in decisions: |
| 214 | if isinstance(d, Work): |
| 215 | out.work_files.append(d.gz_path) |
| 216 | elif isinstance(d, SizeSkip): |
| 217 | cmp = ">" if d.direction == "max" else "<=" |
| 218 | logger.debug( |
| 219 | "size-filter skip %s (%d %s %d)", |
| 220 | d.short_path, |
| 221 | d.size, |
| 222 | cmp, |
| 223 | d.threshold, |
| 224 | ) |
| 225 | out.size_filtered += 1 |
| 226 | out.prefilter_skipped += 1 |
| 227 | elif isinstance(d, AlreadyStored): |
| 228 | logger.debug("skipping %s (already stored)", d.short_path) |
| 229 | out.already_stored += 1 |
| 230 | out.prefilter_skipped += 1 |
| 231 | elif isinstance(d, FilterSkip): |
| 232 | logger.debug( |
| 233 | "filter-skip %s (stored=%s/%s, filter=%s)", |
| 234 | d.short_path, |
| 235 | d.stored_extractor, |
| 236 | d.stored_model, |
| 237 | ", ".join(filter_db), |
| 238 | ) |
| 239 | out.prefilter_skipped += 1 |
| 240 | elif isinstance(d, Symlink): |
| 241 | if d.stale_in_db: |
| 242 | # CASCADE removes orphan mappings. |
| 243 | s.delete_manpage(d.short_path) |
| 244 | logger.info( |
| 245 | "removed stale manpage %s (now a symlink to %s)", |
| 246 | d.short_path, |
| 247 | d.canonical_source, |
| 248 | ) |
| 249 | if d.canonical_in_inputs: |
| 250 | logger.info( |
| 251 | "skipping symlink %s -> %s (canonical file is in the input set)", |
| 252 | d.short_path, |
| 253 | d.canonical_source, |
| 254 | ) |
| 255 | else: |
| 256 | logger.info( |
| 257 | "skipping symlink %s -> %s (pass the canonical file to extract it)", |
| 258 | d.short_path, |
| 259 | d.canonical_source, |
| 260 | ) |
| 261 | out.symlinks.append((d.gz_path, d.short_path, d.canonical_source)) |
| 262 | elif isinstance(d, ContentDup): |