feat(monitor): 作品导出里,没有作品的博主也占一行
导出原先是从作品摊出来的,于是「这个号这段时间一条作品都没发」直接从表里消失 —— 而在表里,**消失和「一切正常」长得一模一样**。这类博主恰恰是名单上最该被看见的: 还在涨粉、只是没动静,或者号出了问题。 来源换成 博主 ∪ 作品(和作品栏同一份 list_creators),没有作品的那一行只填博主, 作品那几列留空 —— **空,不是 0**。 顺带把账号级指标加进作品导出(粉丝数 / 总获赞 / 主页作品数):没有作品的博主那一行 就只剩它们,不加的话那一行等于空行。「主页作品数」是平台上的总数,和下面几列(我们 监控到的作品)不是一回事,所以名字分得开。 排序改成按「这是谁」,同一个博主的行挨在一起 —— 表格是拿去筛和做透视的。
This commit is contained in:
+61
-1
@@ -439,7 +439,10 @@ async def export_data(
|
|||||||
|
|
||||||
if kind == "notes":
|
if kind == "notes":
|
||||||
single_task = scoped[0] if scoped and len(scoped) == 1 else None
|
single_task = scoped[0] if scoped and len(scoped) == 1 else None
|
||||||
rows = await service.list_notes(session, single_task, False, 5000, platform)
|
rows = _notes_rows(
|
||||||
|
await service.list_creators(session, single_task, platform),
|
||||||
|
await service.list_notes(session, single_task, False, 5000, platform),
|
||||||
|
)
|
||||||
elif kind == "comments":
|
elif kind == "comments":
|
||||||
single_task = scoped[0] if scoped and len(scoped) == 1 else None
|
single_task = scoped[0] if scoped and len(scoped) == 1 else None
|
||||||
rows = await service.list_comments(session, single_task, note_id, 5000, platform)
|
rows = await service.list_comments(session, single_task, note_id, 5000, platform)
|
||||||
@@ -482,6 +485,57 @@ async def export_data(
|
|||||||
)
|
)
|
||||||
|
|
||||||
|
|
||||||
|
def _notes_rows(
|
||||||
|
creators: List[Dict[str, Any]], notes: List[Dict[str, Any]]
|
||||||
|
) -> List[Dict[str, Any]]:
|
||||||
|
"""博主 ∪ 作品,摊成一行行 —— **没有作品的博主也要占一行**,作品那几列留空。
|
||||||
|
|
||||||
|
导出是拿去比对名单的,而「这个号这段时间一条作品都没发」恰恰是名单上最该被看见的
|
||||||
|
一行:还在涨粉、只是没动静,或者号出了问题。按作品导的话他直接消失 —— 而在表里,
|
||||||
|
**消失和「一切正常」长得一模一样**。
|
||||||
|
|
||||||
|
和作品栏用的是同一个来源(`list_creators` 也是 账号快照 ∪ 作品),所以导出来的
|
||||||
|
名单和屏幕上看到的博主是同一批。
|
||||||
|
"""
|
||||||
|
works_by_creator: Dict[str, List[Dict[str, Any]]] = {}
|
||||||
|
for note in notes:
|
||||||
|
works_by_creator.setdefault(note["creator_hash"], []).append(note)
|
||||||
|
|
||||||
|
rows: List[Dict[str, Any]] = []
|
||||||
|
listed: set = set()
|
||||||
|
for creator in creators:
|
||||||
|
creator_hash = creator["creator_hash"]
|
||||||
|
listed.add(creator_hash)
|
||||||
|
works = works_by_creator.get(creator_hash, [])
|
||||||
|
if works:
|
||||||
|
# 作品行自己就带着博主那几列(list_notes 里已经拼好了),直接用。
|
||||||
|
rows.extend(works)
|
||||||
|
continue
|
||||||
|
# 一行只填博主:作品列全是空串(`_cell` 把缺失的键变成空)。
|
||||||
|
rows.append(
|
||||||
|
{
|
||||||
|
"creator_alias": creator["creator_alias"],
|
||||||
|
"creator_name": creator["creator_name"],
|
||||||
|
"creator_fans": creator["creator_fans"],
|
||||||
|
"creator_total_favorited": creator["creator_total_favorited"],
|
||||||
|
"creator_works": creator["creator_works"],
|
||||||
|
}
|
||||||
|
)
|
||||||
|
|
||||||
|
# 兜底:作品里的博主没出现在博主名单里就不该丢。正常情况下不会发生(名单是
|
||||||
|
# 账号快照 ∪ 作品),但「悄悄少几行」是这份导出最不该犯的错。
|
||||||
|
rows.extend(note for note in notes if note["creator_hash"] not in listed)
|
||||||
|
|
||||||
|
# 同一个博主的行挨在一起,按「这是谁」排 —— 表格是拿去筛和做透视的。
|
||||||
|
rows.sort(
|
||||||
|
key=lambda row: (
|
||||||
|
row.get("creator_alias") or row.get("creator_name") or "",
|
||||||
|
-(row.get("published_at") or 0),
|
||||||
|
)
|
||||||
|
)
|
||||||
|
return rows
|
||||||
|
|
||||||
|
|
||||||
def _export_columns(kind: str) -> List[tuple[str, str]]:
|
def _export_columns(kind: str) -> List[tuple[str, str]]:
|
||||||
"""(key, header) pairs per export kind."""
|
"""(key, header) pairs per export kind."""
|
||||||
if kind == "notes":
|
if kind == "notes":
|
||||||
@@ -490,6 +544,12 @@ def _export_columns(kind: str) -> List[tuple[str, str]]:
|
|||||||
# 备注优先 —— 昵称常常认不出是谁(见 notes 表那一层的说明)。
|
# 备注优先 —— 昵称常常认不出是谁(见 notes 表那一层的说明)。
|
||||||
("creator_alias", "博主备注"),
|
("creator_alias", "博主备注"),
|
||||||
("creator_name", "博主昵称"),
|
("creator_name", "博主昵称"),
|
||||||
|
# 账号级指标 —— 没有作品的博主那一行**只剩它们**,没有的话那一行等于空行。
|
||||||
|
# 「主页作品数」是他在平台上的总数,和下面那几列(我们监控到的作品)不是
|
||||||
|
# 一回事,所以名字要分得开。
|
||||||
|
("creator_fans", "粉丝数"),
|
||||||
|
("creator_total_favorited", "总获赞"),
|
||||||
|
("creator_works", "主页作品数"),
|
||||||
("note_alias", "作品备注"),
|
("note_alias", "作品备注"),
|
||||||
("title", "标题"),
|
("title", "标题"),
|
||||||
("note_id", "作品ID"),
|
("note_id", "作品ID"),
|
||||||
|
|||||||
@@ -370,3 +370,82 @@ class TestNotesExportColumns:
|
|||||||
assert re.fullmatch(r"\d{4}-\d{2}-\d{2} \d{2}:\d{2}", by_id["note-a"]["首次发现"])
|
assert re.fullmatch(r"\d{4}-\d{2}-\d{2} \d{2}:\d{2}", by_id["note-a"]["首次发现"])
|
||||||
# 而且不能是原始毫秒。
|
# 而且不能是原始毫秒。
|
||||||
assert by_id["note-a"]["发布时间"] != str(PUBLISHED_A)
|
assert by_id["note-a"]["发布时间"] != str(PUBLISHED_A)
|
||||||
|
|
||||||
|
|
||||||
|
class TestNotesExportWithoutWorks:
|
||||||
|
"""**一条作品都没有的博主,也要在导出里占一行**,作品那几列留空。
|
||||||
|
|
||||||
|
导出是拿去比对名单的,「这个号这段时间一条作品都没发」恰恰是名单上最该被看见的
|
||||||
|
一行 —— 还在涨粉、只是没动静,或者号出了问题。按作品导的话他直接消失,而在表里,
|
||||||
|
**消失和「一切正常」长得一模一样**。
|
||||||
|
"""
|
||||||
|
|
||||||
|
@staticmethod
|
||||||
|
async def _add_creator_without_works(**overrides) -> None:
|
||||||
|
from sqlalchemy import select
|
||||||
|
|
||||||
|
from api.monitor.models import MonitorCreatorStat
|
||||||
|
|
||||||
|
fields = dict(
|
||||||
|
run_id=1, creator_hash="hash-c", nickname="博主丙", fans=8800,
|
||||||
|
total_favorited=4321, works_count=7, following=2,
|
||||||
|
captured_at=1_700_000_000_000,
|
||||||
|
)
|
||||||
|
fields.update(overrides)
|
||||||
|
|
||||||
|
async with monitor_db.get_session() as session:
|
||||||
|
task_id = await session.scalar(select(MonitorTask.id))
|
||||||
|
session.add(MonitorCreatorStat(task_id=task_id, **fields))
|
||||||
|
|
||||||
|
@staticmethod
|
||||||
|
async def _rows(client):
|
||||||
|
response = await client.get(
|
||||||
|
"/api/monitor/export", params={"kind": "notes", "format": "csv"}
|
||||||
|
)
|
||||||
|
assert response.status_code == 200
|
||||||
|
return list(csv.DictReader(io.StringIO(response.content.decode("utf-8-sig"))))
|
||||||
|
|
||||||
|
@pytest.mark.asyncio
|
||||||
|
async def test_a_creator_without_works_gets_a_row(self, client):
|
||||||
|
await self._add_creator_without_works()
|
||||||
|
|
||||||
|
rows = await self._rows(client)
|
||||||
|
only = [row for row in rows if row["博主昵称"] == "博主丙"]
|
||||||
|
|
||||||
|
assert len(only) == 1, "没有作品的博主从导出里消失了"
|
||||||
|
|
||||||
|
@pytest.mark.asyncio
|
||||||
|
async def test_that_row_keeps_the_creator_info_and_leaves_works_empty(self, client):
|
||||||
|
await self._add_creator_without_works()
|
||||||
|
|
||||||
|
row = next(r for r in await self._rows(client) if r["博主昵称"] == "博主丙")
|
||||||
|
|
||||||
|
# 博主那几列要有 —— 否则这一行就是个空行。
|
||||||
|
assert row["粉丝数"] == "8800"
|
||||||
|
assert row["总获赞"] == "4321"
|
||||||
|
assert row["主页作品数"] == "7"
|
||||||
|
# 作品那几列留空。**空,不是 0** —— 0 会读成「这条作品零互动」。
|
||||||
|
for column in ("作品ID", "标题", "链接", "发布时间", "点赞", "评论", "收藏", "分享"):
|
||||||
|
assert row[column] == "", f"{column} 应该是空的"
|
||||||
|
|
||||||
|
@pytest.mark.asyncio
|
||||||
|
async def test_the_works_are_still_all_there(self, client):
|
||||||
|
"""加了这一类行之后,原来那些作品一行都不能少。"""
|
||||||
|
await self._add_creator_without_works()
|
||||||
|
|
||||||
|
rows = await self._rows(client)
|
||||||
|
|
||||||
|
assert {row["作品ID"] for row in rows if row["作品ID"]} == {"note-a", "note-b"}
|
||||||
|
|
||||||
|
@pytest.mark.asyncio
|
||||||
|
async def test_a_creator_whose_metrics_are_unknown_leaves_them_empty(self, client):
|
||||||
|
"""采不到账号指标时留空,不写 0 —— 和界面一个口径。"""
|
||||||
|
await self._add_creator_without_works(
|
||||||
|
fans=None, total_favorited=None, works_count=None
|
||||||
|
)
|
||||||
|
|
||||||
|
row = next(r for r in await self._rows(client) if r["博主昵称"] == "博主丙")
|
||||||
|
|
||||||
|
assert row["粉丝数"] == ""
|
||||||
|
assert row["总获赞"] == ""
|
||||||
|
assert row["主页作品数"] == ""
|
||||||
|
|||||||
Reference in New Issue
Block a user