赵国庆 starts spending, so the report now covers six buyers. The refreshed
export also backfills his 10/14 rows, which the previous file lacked.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
32 Adpingids added, 1 removed, no reassignments. Introduces a sixth
buyer, 赵国庆 (9 ids), who has no spend anywhere in the data yet and so
does not appear in the report.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Pulled 武汉自投投放下单_AM&Media Buy.xlsx and the 2610 consumption
workbook out of DingTalk via the browser pane. The roster adds 48
Adpingids and now officially carries every id that roster_extra.json was
covering, so nothing falls back to that file any more and no spend is
left unattributed.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2609 (8/31-10/7) fully supersedes 2026Week39平台消耗.xlsx, which is
removed; 2610 adds 10/8 and wins the overlapping October days.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The consumption export keeps changing filename (每日数据 -> Week39...),
which broke the glob each time. build_data.py now treats any workbook
containing a 平台消耗统计 sheet as a data source and orders them by the
latest date they contain, so the fresher file wins on overlapping days.
2026Week39平台消耗.xlsx covers 8/31-9/27 and fully supersedes the 2609
file, which is removed; 2608 stays for the earlier August dates.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Despite the filename, that workbook is the 投手 roster (no Date or 花费
columns), and is a strict superset of the previous one: 30 Adpingids
added, none removed or reassigned. Spend is unchanged because no new
consumption data arrived; only the 负责 counts move.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Every roster the user sends arrives under a different filename, which
forced a code change each time. The roster now lives at a fixed
roster.xlsx; the three superseded copies are removed (git history keeps
them, and each was a strict subset of this one).
This roster claims all 14 previously unattributed Adpingids, so the
report has no unattributed spend. It also moves 5617321 from 周梓君 to
詹涵粤; that id has never spent, so only the 负责 counts shift.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
投手List.xlsx is a strict superset of the previous roster (19 added,
none removed or reassigned) and claims every previously unmatched
Adpingid, so the report has no unattributed spend.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
build_data.py now also computes each buyer's total for the preceding
3-day window under identical rules (exclusions, attribution), exposed as
prev_total plus meta.prev_range. index.html renders it as the first
panel, with green/red deltas and a total row; the panel stays hidden if
no prior period exists.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
New roster is a strict superset of 投手.xlsx (14 added, none removed, no
reassignments). build_data.py now picks the roster by an explicit
precedence list, so the old 投手.xlsx stays as a fallback.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
New roster_extra.json fills in Adpingids missing from 投手.xlsx. It only
applies when the roster has no entry for that id, so a future roster
always wins.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Monthly data files no longer carry a 投手 sheet, so build_data.py now
prefers a standalone 投手.xlsx and falls back to the newest embedded
sheet. This roster is content-identical to the 2608 embedded one, so
data.json is unchanged.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The 2609 file carries only 8/31 and has no 投手 sheet, so build_data.py
now globs every *每日数据-武汉萌奇熊.xlsx, merges 平台消耗统计 per date
(newer filename wins for a repeated date), and takes the roster from the
newest file that still has a 投手 sheet. Future month rollovers need no
code change.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Adping-level only; com.vuclip.viu stays out of the package exclusion list
per request, so its other Adpingids remain in the report.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
exclude.json gains an "adpings" map (18 ids, value = package for
reference) alongside the existing package list. build_data.py drops those
Adpingids from both the 投手 ownership counts and the spend rows.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Package now renders as pkg_GEO, where GEO is the region with the highest
total spend for that package. The new Adpingid column shows the
top-spending Adping within that region, so label and id stay consistent.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
build_data.py now emits active_ids (Adpingids with spend in the window)
and owned_ids (Adpingids assigned to the buyer, exclusions removed).
active_ids sums to the 覆盖 Adpingid 数 card.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
- 各投手本周期内投放花费: draw value labels at each bar end via a small
inline Chart.js plugin (no extra CDN dependency); right padding added
so labels stay inside the panel.
- Prefix all spend figures with $: summary card, both chart axes and
tooltips, buyer subtotals, and package detail tables.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Add exclude.json (12 package names, permanent) and build_data.py so the
report is reproducible. Exclusion is exact-match on package name, applied
both to 平台消耗统计 rows and to Adpingids whose 投手 pkg is excluded.
Exact match avoids false positives (e.g. in.swiggy.android vs "igg").
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Refresh xlsx source and regenerate data.json from
平台消耗统计 (Date/Adpingid/花费) x 投手 (Adpingid/Media Buyer).
@某某 merged with 某某; Adpingid without a buyer excluded.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
