Peach: Local-first video metadata scraper and media manager.
refactor(follow): 新作接口拆出取候选、筛选判据、筛选项与资料补读
longmeidaoa4a1c72434a1
diff --git a/src/peach/web_jav_discovery.py b/src/peach/web_jav_discovery.pyindex e602d0840..cab14e345 100644--- a/src/peach/web_jav_discovery.py+++ b/src/peach/web_jav_discovery.py@@ -261,70 +261,80 @@ def q_jav_releases(contract, args):page = max(1, int(args.get("page") or 1))except (TypeError, ValueError):size, page = RELEASE_PAGE_SIZE, 1- chosen = {facet: str(args.get(facet) or "").strip() for facet in RELEASE_FACETS}- query = str(args.get("q") or "").strip().casefold()sort = args.get("sort") if args.get("sort") in RELEASE_SORTS else "release_date"descending = args.get("direction") != "asc"+ matches = _release_matcher({facet: str(args.get(facet) or "").strip() for facet in RELEASE_FACETS},+ str(args.get("q") or "").strip().casefold())with contract.read_connection() as connection:- unique = {}- for row in _external(connection, contract):- unique.setdefault(web_catalog.normalise_code_key(row["code"]) or row["id"], row)- rows = list(unique.values())- if sort == "taste":- strengths, names, evidence = preference_profile(connection, contract.taste_history_store)- for row in rows:- row["taste_score"], row["reasons"] = preference_score(row, strengths, names, evidence)- refs = web_feeds._discovery_performers(connection, contract, rows)- studios = {}- for row in rows:- row.pop("_source_actors", None)- row["performer_refs"] = refs[int(row["id"])]- row["performers"] = "、".join(ref["name"] for ref in row["performer_refs"])- if row["studio"] and row["studio"] not in studios:- studios[row["studio"]] = web_feeds._studio_name(connection, row["studio"])- row["studio"] = studios.get(row["studio"], row["studio"])- row.pop("entities", None)- rows = _release_order(rows, sort, descending, str(args.get("seed") or ""))-- def matches(row, skip=None):- if query and query not in " ".join([row["code"] or "", row["title"] or "", row["performers"], row["studio"] or "",- *row["tags"]]).casefold():- return False- return all(not value or facet == skip or value in _release_values(row, facet)- for facet, value in chosen.items())-- facets = {}- for facet in RELEASE_FACETS:- counts, first = {}, {}- for row in rows:- if matches(row, skip=facet):- for value in dict.fromkeys(_release_values(row, facet)):- counts[value] = counts.get(value, 0) + 1- first.setdefault(value, row)- facets[facet] = sorted(counts.items(), key=lambda pair: (-pair[1], pair[0]))- if facet == "performer":- facets[facet] = [{**next(ref for ref in first[name]["performer_refs"] if ref["name"] == name), "count": count}- for name, count in facets[facet]]- elif facet == "studio":- facets[facet] = [{"name": name, "count": count, "has_logo": contract.has_logo(name),- "logo_version": contract.logo_version(name)} for name, count in facets[facet]]- else:- facets[facet] = [{"name": name, "count": count} for name, count in facets[facet]]+ rows = _release_order(_release_rows(contract, connection, taste=sort == "taste"), sort, descending,+ str(args.get("seed") or ""))+ facets = {facet: _release_facet(contract, rows, facet, matches) for facet in RELEASE_FACETS}shown = [row for row in rows if matches(row)]picked = shown[(page - 1) * size:page * size]wanted = wants.wanted_keys(connection, (web_catalog.normalise_code_key(row["code"]) for row in picked))for row in picked:row["wanted"] = web_catalog.normalise_code_key(row["code"]) in wanted- row["has_cover"] = contract.has_cover(row["code"])- row["cover_frame"] = contract.cover_frame(row["code"])- row["cover_version"] = contract.cover_version(row["code"])- row["poster_box"] = contract.poster_box(row["code"])- response = {"items": picked, "total": len(shown), "candidate_total": len(rows), "page": page, "size": size,- "sort": sort, "direction": "desc" if descending else "asc",- "facets": {"performers": facets["performer"], "studios": facets["studio"], "tags": facets["tag"]}}+ row.update(has_cover=contract.has_cover(row["code"]), cover_frame=contract.cover_frame(row["code"]),+ cover_version=contract.cover_version(row["code"]), poster_box=contract.poster_box(row["code"]))+ return {"items": picked, "total": len(shown), "candidate_total": len(rows), "page": page, "size": size,+ "sort": sort, "direction": "desc" if descending else "asc",+ "facets": {"performers": facets["performer"], "studios": facets["studio"], "tags": facets["tag"]},+ **_release_metadata(contract, shown, args.get("metadata_job_id"))}+++def _release_rows(contract, connection, *, taste):+ """新作候选:「发现」那批去重,出演换成解析后的规范名、厂牌换成规范名;按口味排时现算口味分。"""+ unique = {}+ for row in _external(connection, contract):+ unique.setdefault(web_catalog.normalise_code_key(row["code"]) or row["id"], row)+ rows = list(unique.values())+ if taste:+ strengths, names, evidence = preference_profile(connection, contract.taste_history_store)+ for row in rows:+ row["taste_score"], row["reasons"] = preference_score(row, strengths, names, evidence)+ refs = web_feeds._discovery_performers(connection, contract, rows)+ studios = {name: web_feeds._studio_name(connection, name) for name in {row["studio"] for row in rows if row["studio"]}}+ for row in rows:+ row.pop("_source_actors", None)+ row.pop("entities", None)+ row["performer_refs"] = refs[int(row["id"])]+ row["performers"] = "、".join(ref["name"] for ref in row["performer_refs"])+ row["studio"] = studios.get(row["studio"], row["studio"])+ return rows+++def _release_matcher(chosen, query):+ """筛选判据:文字匹配番号、标题、出演、厂牌与标签,三个维度各按所选值;`skip` 的那一维不看。"""+ def matches(row, skip=None):+ if query and query not in " ".join([row["code"] or "", row["title"] or "", row["performers"], row["studio"] or "",+ *row["tags"]]).casefold():+ return False+ return all(not value or facet == skip or value in _release_values(row, facet) for facet, value in chosen.items())+ return matches+++def _release_facet(contract, rows, facet, matches):+ """一个维度的筛选项:本维度不计、其余条件生效的行集里出现过的值与部数,部数多的在前。"""+ counts, first = {}, {}+ for row in rows:+ if matches(row, skip=facet):+ for value in dict.fromkeys(_release_values(row, facet)):+ counts[value] = counts.get(value, 0) + 1+ first.setdefault(value, row)+ ordered = sorted(counts.items(), key=lambda pair: (-pair[1], pair[0]))+ if facet == "performer":+ return [{**next(ref for ref in first[name]["performer_refs"] if ref["name"] == name), "count": count}+ for name, count in ordered]+ if facet == "studio":+ return [{"name": name, "count": count, "has_logo": contract.has_logo(name), "logo_version": contract.logo_version(name)}+ for name, count in ordered]+ return [{"name": name, "count": count} for name, count in ordered]+++def _release_metadata(contract, shown, job_id):+ """筛选结果里从没读过来源资料的那几部起一轮读取;`job_id` 是轮询。没有要读的也没在轮询就不回任务。"""unread = [row["code"] for row in shown if row.get("metadata_state") == "pending"]- if unread or args.get("metadata_job_id"):- job = discovery_metadata(contract).ensure(unread, refresh=bool(unread), job_id=args.get("metadata_job_id"))- if job:- response.update(_job_response(job))- return response+ if not unread and not job_id:+ return {}+ job = discovery_metadata(contract).ensure(unread, refresh=bool(unread), job_id=job_id)+ return _job_response(job) if job else {}