|j_ SrSSKrSSKrSSKrSSKrSSKrSSKJr SSKrSSK J r J r J r J r SSKJr SSKJrJrJrJrJr SSS .rS rS rS rS r\R6R9\R6R;\55r\R6RA\SS5r!\RDr"SOSjr#Sr$Sr%Sr&Sr'Sr(Sr)Sr*Sr+Sr,Sr-Sr.SPSjr/SQSjr0Sr1Sr2SRS jr3S!r4SSS"jr5S#r6\"4S$jr7S%r8S&r9S'r:S(S)S*S+S,S-S.S/S0S1S2S3S4. r;STS5jrSRS8jr?S9r@SRS:jrASRS;jrBS<rCS=rDS>rE\FS?:XaS@rGSA\R;a\ R"SB\5 gSC\R;aSDrJ\ R"SE\J5 gSF\R;a\ R"SG\C5 gSH\R;a\ R"SISJ5 gSK\R;a\ R"SLSM5 g\ R"SN\95 gg)Ua BidBrain daily purchases capture. Motorway and Carwow both close their daily sale at 3:30pm and the day's buys populate in the purchases screens about 15 minutes later. This run, scheduled at 3:50pm, reads what was actually bought on Motorway, Carwow and Auction4Cars (added 2026-08-27, its own auctions run over several days rather than closing at a fixed daily time, so this simply catches whatever has completed since the last read), stores it, and emails the NEW buys to dealer_config. ACCOUNTS_EMAIL with the achievable retail price from that day's BidBrain run, so accounts can add the cars to the other systems. The email is sent through the Mail app on this Mac (signed in as whatever account Mail is set up with), no passwords are stored. No buys, no email. The very first run baselines the purchase history without emailing it. python3 purchases_run.py read, store, email new buys python3 purchases_run.py --dry-run read and store, print the email instead python3 purchases_run.py --baseline seed history without emailing it python3 purchases_run.py --baseline --motorway seed just Motorway (--carwow/--auction4cars for the others) python3 purchases_run.py --check-purchase REG re read one purchase from its own platform python3 purchases_run.py --push-dealerkit REG push one purchase to DealerKit [--items ...] python3 purchases_run.py --check-dealerkit REG check one purchase's real DealerKit record Reads screens only. Never bids, never acts on the platforms. N)sync_playwright)dbrunsbrowserdealerkit_write)pounds)motorwaycarwowcazana auction4cars dealerauctionr r ) Auction4CarsDealer Auctioncr/nUHnURS5(d(URS5(dURS5cMCURS5bMW[[US5S- XS9nURSURS55nUbXe:XaMURXFU45 M U$) uiWhich live cars get a fresh Retail est. from this dealership's own rule, as (row, old, new) triples (Steven 2026-09-07: "why is dealer os still adding £500 on to retail estimate pricing and rounding up always with a 5 on the end"; the estimate was only ever computed on the day a car was first read, so every car bought before the rule changed kept the old formula's number). Reworked from the stored cazana_115 (the plain Cazana retail times 1.15). Left alone: a cancelled sale, a car with no Cazana figure, a car with a manual override (the person's own figure always wins), and a car whose figure already matches. cancelled_atshell cazana_115retail_estimate_overrideffffff?marginendingretail_estimate_rawretail_estimate)getdealer_retail_pricefloatappend)rowsrroutrnewolds ./Users/stevendouglas/BidBrain/purchases_run.py reprice_planr%4s C  55 AEE'NNaeeL6I6Q  55+ , 8 !%,"84"?^ee)1551B+CD ;#*  AC=! Jc P[R"5 [5up[[R"5X5nU(d&[ R "SSSSS5 [S5 gUHUup4n[R"USU5 [R"USSS S XES S 9 [S USSUSSUS35 MW SUSSU3n[ R "SS[U5SUS3[U5[U55 [[U5SUS35 g)aRe price every live car's Retail est. with this dealership's own margin and ending (the Run modal's "Re price retail estimates"). The change is written on each car's timeline; a car on DealerKit and still Due in has the new figure sent to DealerKit by the next stage check (daily_run.retail_sync_plan, within five minutes), a finished car's DealerKit price is never overwritten by a formula figure.retail_repriceTz/Every retail estimate already follows the rule.zNothing to re price.Nidfieldz4Retail est. re priced with the dealership's own rulerBidBrain)r+wasnow_valwho regu: £,.0fu -> £u margin £z , ending z retail estimate(s) re priced (z).z re priced () rinit_db_retail_formula_settingsr%list_purchasesrfinishprintset_purchase_retaillog_purchase_eventlen)rrplanr!r#r"words r$reprice_retailr=Ls!JJL-/NF ))+V This dealership's own two knobs for dealer_retail_price: the flat margin and the price ending, read once per sync (Steven 2026-09-01, see dealer_retail_price's own docstring for why the formula itself stays pure). Falls back to the shared factory defaults on any read failure, never blocks a sync over a settings read.retail_estimate_marginretail_estimate_ending_)r get_settings Exceptionr)saveds r$r4r4dsS ! II. 4 II. 3 55 s; A  A cU(a [U5O1Skn[R"5nUR S5=(d 0n[UR S5=(d /5n[5nUHKn[ R U5nU(aUR U5SLdXt;aM:UR U5 MM U$![a Us$f=f)aThe real set of platforms this sync will actually read: requested (or all four when nothing was asked for, the normal full sync, same "smaller set narrows, never widens" shape as daily_run._select_platforms) narrowed to whatever is switched ON in Settings. A platform hidden as a connection, or turned off on its own Settings tab, must never be read for purchases either, only for the live auction list did that rule apply before this fix (found live, 2026-09-01: Auction4Cars and Dealer Auction both hidden and off, yet still read and failed every sync, painting the nav light red app-wide for something the dealership had deliberately switched off).>CarwowMotorwayrrplatform_enabledhidden_connectionsF)setrrCrDr_PLATFORM_SETTINGS_KEYadd) requestedbaserEenabledhiddenr namekeys r$_settings_enabled_platformsrTrs'3y>,dD!ii*+1rG /06B 7F %C$((. GKK$-    J  sC CCdataz last_run.jsonc^UcgUS-U-mUS:Xa"[TS- S- 5n[SU-S-5$TS-S-nSVVs/sHnSH nXE-U-U-PM M nnn[[UU4SjS 95$s snnf) urThe price a dealership asks once a car is its own stock to resell: Cazana retail times 1.15 (BidBrain's own uplift, see pricing.py) plus a flat margin, then rounded to this dealership's own price ending. The two endings are genuinely DIFFERENT shapes, not the same rule with a different digit, because they came from two different real sources: ending=95 (Mark's own, the default) rounds to the nearest of just FOUR specific bands, X295/X495/X795/X995, cycling every £1000. Confirmed against his own real pricing across 16 already bought cars (2026-08-21): landed on his exact figure for 12 of 16, clearly the best fit against a flat 20% uplift alone (6 of 16) or the plain uplift with no added margin (0 of 16). His own words, "roughly £500 above the 115% Cazana retail". These four bands are HIS real, empirical pricing, not a generic rule, so they are never touched by the other ending. ending=99 (Steven's own) rounds to the nearest hundred, ending 99, no gaps: X099/X199/X299/.../X999. A first version of this mechanically reused Mark's own four band SPACING with a 99 suffix (X299/X499/X799/ X999), which put a real car's own £10,166 target 133 away from its nearest allowed price when the genuinely nearest £10,199 was only 33 away and simply never offered as an option (Steven caught this live, 2026-09-01: "why does it compute to 10299 instead of 10199?"). Asked directly rather than guessed a second time; "every £100" was his own confirmed choice. Deliberately PURE, margin and ending never read from settings here: two dealerships share this repo and genuinely disagree on both, so every real caller reads pricing.DEFAULT_SETTINGS' own retail_estimate_margin/retail_estimate_ending and passes them in, the same "settings threaded through, function stays testable alone" shape as daily_run._select_platforms. None in, None out, never guessed when there is no Cazana figure to work from.Nrcd)irrY)iic >[UT- 5$N)abs)ctargets r$%dealer_retail_price..s s1v:r&)rS)roundmin) cazana_retailrrk thousandsdeltahundreds candidatesr`s @r$rrsD T !F *F | 6B;#% &S1Wr\""44'I/8/e"6h#h.7"68/8 Z%>? @@8sA;cSSKn0nUR[RR [ SS5SS9nUR S5HQup4SR [U=(d S5R5R55nU(dMMXAU'MS UR5 0n[R"[[S S 95nURS /5URS /5-HnURS 05URS5pU RS5n U RS5U RS54Hrn SR [U =(d S5R5R55nU(dMKU (aXU'U (dM_U RS5X'Mt M X4$!UR5 f=f![ a X4$f=f)aEMap normalised reg to the most recent governing retail value BidBrain computed in the last few days, from the stored assessments (every run saves them), plus the latest run cache which also knows each car's selling plate. Also returns a second map, reg to photo_url, borrowed from that same run cache (Mark 2026-08-21, "add an image to the line" on the purchases report) since the daily run already fetched it, no separate read needed. A car bought a while back, whose cache has long since rotated away, simply has no photo here, same as it has no retail estimate from this source.rNrU bidbrain.dbtimeoutzSELECT reg, governing_value FROM assessments WHERE governing_value IS NOT NULL AND run_at >= datetime('now', '-4 days') ORDER BY idutf-8encoding shortlistheldcarpricing photo_urlr1 selling_vrmgoverning_value)sqlite3connectospathjoinHEREexecutestruppersplitclosejsonloadopenCACHErOSError) r{r connr1gvrfphotosdar_pphotorSs r$ _retail_maprs C ??277<<fmDa? PD ||"#GC CI2,,.4467AqA# F IId573 4{B'!%%*;;A55#QUU9%5qEE+&Ee aeeM&:;GGC rN00288:; %1I1UU#45CF<< ;#    ; s+AGG4C:G%2G%G"% G43G4c b^SSKn0mU4Sjn[R"[[SS95nUR S/5UR S/5-UR S/5-HnUR S 05UR S 5=(d 0pTUR S 5UR S 54H@nU"XdR S 5UR S5=(d UR S5S9 MB M UR[RR[SS5SS9nURS5H upn U"XU S9 M UR5 [ R""5HHn U"U R S 5U R S 5U R S5U R S5S9 MJ U4Sjn [R"[[SS95nUR S/5UR S/5-UR S/5-HUnUR S 05nUR S 5UR S 54HnU "XdR S55 M MW [ R$"S5R'5H upU "X5 M T$![ [4a GNf=f!UR5 f=f![a GNf=f![a GNSR[U=(d S5R5R55nU(dgTR U05nUR 5H(upEUS;dM UR U5S;dM$XSU'M* g)Nrp)Nrpr)rrrr setdefaultitemsr)rSfactsrfrowfvknowns r$put_known_facts..putss GGC rN((*002 3 q"%KKMDA %#''!* *EA"r&rqrrrtrurejectedrvrwr1rymileagere)rr rUrlrmrnzSELECT reg, mileage, value FROM valuation_cache WHERE source='cazana' AND value IS NOT NULL AND valued_at >= datetime('now', '-30 days') ORDER BY valued_at ASCrr)rrrc>SR[U=(d S5R5R55nU(a9Ub5TR U05R S5c[ U5TUS'gggg)Nrpfinance)rrrrrrbool)rSrrfrs r$ put_finance!_known_facts..put_financesk GGC rN((*002 3 5#3#3Ar#:#>#>y#I#Q"&q'E!HY $R1r&r)r{rrrrrr ValueErrorr|r}r~rrrrrDrr5recent_detail_factsr)r{rrrr_rrSrr1rvaluer!rrrs @r$ _known_factsrs E IId573 4{B'!%%*;;aeeJPR>SSA55#QUU9%5%;qe aeeM&:;Cy!1155;Q;kUVUZUZ[jUkm<T  rww||D&-HRST '+||1(2#e C7 (2 JJL ""$A e aeeI&6155CVhihmhmniA B%* IId573 4{B'!%%*;;aeeJPR>SSAeR Ae aeeM&:;Cy!12<T  ,,Y7==?FC  @ LQ Z     JJL        Z      L stC2J<3K(6"KK()A!K9B3L 5L <KKK%%K(( K65K69 LL LL L.-L.c/n[R"U5n[R"US5nUHnURS5(aMURS5=(d SnURS5=(d SS:a$U(aXc:waUR USU45 MwURS5=(d SS:dMU(dMU(aMUR USU45 M U$) aWhich purchases get their On finance? answer from their own purchase page (2026-09-16): a car whose page shows negative equity is marked NEGATIVE EQUITY unless it already is; a car with a settlement figure and no answer yet is marked as on finance. A NO answer a person typed is never overwritten; only the platform's own figures move it up. (purchase id, answer) rows. Pure, tested.Tr on_financerpnegative_equityrr*finance_settlement)rnegative_equity_answerfinance_answerrr)on_fileoptionsr negyesr!curs r$_finance_page_updatesr1s C # #G ,C  GT *C  55 eeL!'R EE# $ )Q .33: JJ$~ &ee().Q! 3CC JJ$~ & Jr&c $/nUGHnURS5(dURS5(aM2URSR[URS5=(d S5R5R 555=(d 0nURS5cM[ R "U[US55nU(dMURUSXdRS5=(d S45 GM U$) zWhich purchases get their On finance? answer from the platform: a car with no answer yet whose listing said whether finance is outstanding. (purchase id, answer, platform) rows. A person's own answer is never overwritten. Pure, tested.rrrpr1rr*platformz the platform) rrrrrrrrr)rrrr r!ranswers r$_finance_updatesrFs C  55  !%%"7"7  IIbggc!%%,"4"5;;=CCEF G M2 55  # ""7D9,>? 6 JJ$z):)LnM N Jr&c S=pEUGHZnURSR[URS5=(d S5R5R 555=(d 0nURS5(d#URS5(a USUS'US- nURS5bMURS5b<USUS'URS5cURS5bUSUS'US- nGMURS5(dGM[ [ US5S -S 5US'[[ US5X#S 9US'US- nGM] XE4$) zFill each purchase's mileage, Cazana figure and Retail est. from what is already known for its plate, so the live Cazana lookup below only ever runs for a car nobody has valued. Returns (mileages filled, valuations filled). Pure over its inputs, tested.rrpr1rr)rrr rr)rrrrrrcrr)rrrrmilesvalsr!rs r$ _reuse_knownrXsG E  IIbggc!%%,"4"5;;=CCEF G M2uuYAEE)$4$4YO8P8\+,->+?A'( x"'ak(:T(A1"E,':58;MV\'l#$  ;r&c^URS5=(d SR5R5m[U4SjS55$)zA purchase whose sale was cancelled or voided on the platform. Carwow marks these voided; Motorway shows a cancel or withdrawn status.statusrpc3,># UH oT;v M g7fr]).0tss r$ _is_cancelled..ts?>!Av>s)voidcancel withdrawn)rstriplowerany)r!rs @r$ _is_cancelledrps; x B%%'--/A ?>? ??r&))payment) winning_bid)r) photo_urls)zservice history)service_history_photosdocument_paths)V5) v5_photosv5_local_paths)zcollection date)collection_datecJU(dg[U5R5S;$)zsA stored value that really carries something: not empty, and not an empty JSON list or object saved as text ("[]").F)z[]z{}null)rr)rs r$ _has_factrs   q6<<>!5 55r&c&URS5=(d /n[U[5(a[R"U5n[U[5(aU(dg[SU55(+$![ [ 4a /nNHf=f)zzWhether a car's stored gallery is our own saved files rather than the platform's own web links (2026-09-10). Pure, tested.rFc3r# UH-n[U5R5RS5v M/ g7f))zhttp://zhttps://N)rr startswith)rus r$r#_photos_are_ours..s)Tt!3q6<<>,,-DEEts57) r isinstancerrloads TypeErrorrlistr)existingurlss r$_photos_are_oursrsz << % +D$ ::d#D dD ! !TtTT TT :& D sA;;BBc^[VVs/sH#up[U4SjU55(aM!UPM% nnnTRS5=(d SnSU;a'US:Xa![T5(dUR S5 US;a9TRS5bTRS 5(aUR S 5 U$s snnf) aWhich of PAGE_FACTS the row on file still lacks, plus, for a Carwow car, photos still hosted by Carwow rather than saved as files of ours (2026-09-10), plus, on Motorway and Carwow, the real buyer's fee while the one on file is still an estimate or not there at all (2026-09-14, Steven: seven of nine Carwow cars showed no fee; Carwow only puts the breakdown on the page once the car is collected). Nothing here is chased for ever any more: _page_read_needed caps every chase.c3X># UHn[TRU55v M! g7fr])rr)rrrs r$r!_missing_facts..s%9e^dYZ)HLLQRO:T:T^ds'*rrprrGzphotos of oursrGrH motorway_feepayment_estimatedfee) PAGE_FACTSrrrr)rwhatfieldsr plats` r$_missing_factsrs%/ fJLDc9e^d9e6e4JC f << # )rDstx/8H8R8R #$ %%8<<+G+OS[S_S_`sStSt 5 J gs CC<cU(dgURS5(ag[URS5=(d S5SSnU(a-XA[R"[S9- R 5:ag[ U5nS U;d S U;dS U;agU(dgXA[R"US9- R 5:ag[U5=(a' XA[R"US9- R 5:$) a3Whether a purchase's own page is worth reading today. A car not on file yet: yes. A car bought within the last `days` days: yes, whatever it has, its documents and collection date are still arriving. A car bought within `chase_days` days that still lacks anything the page can give (payment breakdown, photos, service history, V5, collection date): yes. Older than that, only the payment breakdown and photos are worth chasing, a seller who never uploaded a V5 is not going to now. Anything else is complete and left alone. Before this (Steven 2026-09-08) every purchase's page was read every day, 30 odd pages for a sync that usually had one or two new cars. Steven the same day: "should this not also be looking for v5 and service history or any other thing that we try and capture that isnt there yet?" Pure, tested.TrF bought_daterpN daysrrr)rrdatetime timedeltaCHASE_CAP_DAYS isoformatrr)rtodayr chase_daysboughtmissings r$_page_read_neededrs ||N## m,2 3CR 8F &H$6$6N$KKVVXXX&GGx72ew6F  (,,$77BBDD = `V0B0B 0S(S'^'^'```r&cURS5=(d SnUS;dURS5b0$U=(d [RnURS5bUSnOvUS:XaURS5nO^UbUO#[R"5RS5n[R"URS5XPRS 55nUc0$U"X45nUc0$XF[ US -S 5S S .$)aThe buyer's fee to hold for a Carwow or Motorway car whose page has not shown its breakdown yet (2026-09-14, Steven: "a car should never be added from carwow or motorway from now on without the fees"): the platform's own published band for the price, with its VAT, marked estimated so the real breakdown replaces it the moment the page shows one. Carwow's list price is the car alone; Motorway's is the car alone while the purchase is still going through and the all in figure once the checks are complete, so its bid goes through db.provisional_motorway_bid, which knows the difference from the row's own status. {} when the row already has a fee, has no price, or no band fits. Pure: fee_for is db.fee_for, mw_bands Motorway's band table.rrprrrrGpricerHrg?rT)rr vat_total estimated)rrfee_for get_fee_bandsprovisional_motorway_bidrc)rrmw_bandsrrbandsrs r$ estimated_feer s 77:  $"D ))SWW^-D-P #G ww})M"   $0b6F6F6H6L6LZ6X++CGGG,Q6R6RVYV]V]^lVmVm '-2 .s 3F < bFh.@.@n.U&U%`%`%bbbr&udCarwow never showed the fee within {days} days, the £{fee:,.0f} estimate stands. Check the invoice.cU=(d [RR5nSn[R"5Hn[ X 5(dM[ R[URS5=(d SS9n[S[R"US555(aMu[R"USSUSS9 US - nM U$) zZOne timeline line, once, on every Carwow car whose fee estimate has aged out of the chase.rr)rrc3\# UH"nSURS5=(d S;v M$ g7f)zestimate standstextrpNr)res r$r,note_estimates_that_stand..s%aC`a QUU6]%8b9C`s*,r*syncr,r/r)) rdaterrr5r ESTIMATE_STANDS_NOTEformatrrrpurchase_eventsr9)rnrrs r$note_estimates_that_standrs  *X]]((*E A  "s** #**CGGND[D`_`*a a2CUCUVYZ^V_C`a a a  c$i:F Q# Hr&c UVs/sHMn[URUS[URS5=(d S545U5(dMKUPMO nnU[U5[U5- 4$s snf)z6(rows to read, how many complete ones are left alone).r vehicle_refrp)rrrr:)rrrr!keeps r$_rows_worth_readingr si wt!0a mSQRQVQVWdQeQkikMl=m1npuvAtD w TSY& && xs A A4A4c^SSKJn [U5n/n/n[R"5Vs0sH+oUS[ UR S5=(d S54U_M- nn[RR5n[5nS[R4S[R4S[R 4S ["R$44HMupX;aM U"S S U S 35 U "U5n ['U S [)U 5S35 UR+U 5 MO SU;GaU"S S5 [R0"XS9n ['S[)U 5S35 UR+U 5 UV^s/sH,mTSS:XdM[3U4SjS55(aM*TPM. nn[5XU5upU(a['SUS35 U(GaU"S S[)U5S[)U5S:waSOSU(aSUS 3OSS!3S[)U55 [6R8"US"S#S$9nUR:(aUR:SOUR=5nSn[?U5GHunm[R@"UTS5nU(aTRCU5 US- n[DRFRI[JS%S&TS'5n[RL"U5nU(a([RN"UU/U5nU(aUSTS('[RP"U5nU(a8[RN"UUU5nU(a[RRT"U5TS)'[RV"U5nU(a8[RN"UUU5nU(a[RRT"U5TS*'[RX"UTS5nU(a8[RN"UUU5nU(a[RRT"U5TS+'[RZ"UTS5nU(aUTS,'U"S S-US-S.[)U5STR S'S5S/3US-[)U55 GM UR]5 ['S0US.[)U5S135 SU;GaxUVs/sH>oUSS:XdM UR S35S4:wdM$UR S5(dM<UPM@ nn[5UXg5unnU(a['S5US35 U(GaU"S S[)U5S6[)U5S:waSOSU(aSUS 3OSS!3S[)U55 [6R8"US7S#S$9nUR:(aUR:SOUR=5nSn[?U5GH8unm[R@"UTS5nU(aTRCU5 US- nTR S85SLTS9'[DRFRI[JS%S&TS'5n[RX"UTS5nU(a:[RN"UUU5n[RRT"U=(d U5TS+'[DR^"US#S:9 [R`"UTSUS;9nUV s/sH:n [DRFRcU 5ReS<5(dM8U PM< n!n UV s/sH:n [DRFRcU 5ReS=5(dM8U PM< n"n U!(a[RRT"U!5TS>'U"(a[RRT"U"5TS?'[RZ"UTS5nU(aUTS,'U"S S@US-S.[)U5STR S'S5S/3US-[)U55 GM; UR]5 ['SAUS.[)U5S135 S U;GaUVs/sH'oUSS :XdM UR SC5(dM%UPM) n#n[5U#Xg5un#n$U$(a['SDU$S35 U#(Ga\U"S SE[)U#5SF3S[)U#55 [6R8"USGS#S$9nUR:(aUR:SOUR=5nSn[?U#5Hunm["RfRiTSC5n%U%(aL["RX"UU%RkS55nU(a[RRT"U5TS+'US- nU"S SHUS-S.[)U#5STR S'S5S/3US-[)U#55 M UR]5 ['SIUS.[)U#5S135 SSS5 X44$s snf![,a-n ['U SU 35 UR/U 5 Sn A G MeSn A ff=f![,a*n ['SU 35 UR/S5 Sn A GNSn A ff=fs snf![,a GNCf=f![,an ['S2U 35 Sn A GNSn A ff=fs snfs sn fs sn f![,a GN f=f![,an ['SBU 35 Sn A GNSn A ff=fs snf![,a GNf=f![,an ['SJU 35 Sn A GN7Sn A ff=f!,(df  X44$=f)Ka Read the purchases screens, plus Motorway's dedicated Cancelled list. Each read fails independently and loudly, one being down never hides another. which platform(s) are read (a set of "Motorway"/"Carwow"/"Auction4Cars"/ "Dealer Auction"), None or all four means the normal full read, same "smaller set is a partial run" shape as daily_run.run()'s own platforms parameter. Auction4Cars reads its own real "My Purchase History" screen (Mark 2026-08-27: "i now want to add a4c purchase info in to BB"), the account's completed buys over its own 13 month window, found live to be the real, wider source, "Won" only ever covers the last 3 months and reads 0 on this account (nothing currently mid process). No payment breakdown, documents, V5, collection date or DealerKit push support is built for this platform yet, only the base capture (reg, name, price, mileage, bought date), matching how Carwow's own purchases started before those were added in later, separate sessions; a follow up, not this one. Dealer Auction reads its own real Invoice history screen, Purchases tab (Mark, same day: "ok lets do DA now"), found live under Dashboard > Invoice history: a genuine itemised transaction ledger, no detail page visit needed at all, registration, the full real fee breakdown (vehicle price, buyer fee, VAT) and a real advert id are all already on the one list, the richest of the three platforms' own purchase records built so far. Newest first, a real page or two covers the daily top up (dealerauction.read_purchase_invoices' own max_pages default). It gained its own real photo gallery (Mark 2026-08-27, "still no images in the DA cars") straight after, read off the won advert's own page, the Invoice history list itself carries none. Real progress is written throughout this whole function (Steven 2026-08-27, live: the purchases page's own run bar was showing a stale message left over from an unrelated earlier daily run, since this function, unlike push_dealerkit/check_dealerkit/refresh_purchase in this same file, never wrote to run_progress.json at all, so the bar had nothing real of its own to show for the several minutes this can genuinely take, one page visit per purchase for its own payment breakdown, photos, documents and gallery). Local import, matching the convention already used everywhere else in this file.r _progressrrrprHrGrrreadingzReading purchases...: z purchases on the screenz purchases read FAILED: Nz.Reading Motorway's cancelled purchases list...) max_pageszMotorway cancelled list: z rows readz Motorway cancelled read FAILED: zMotorway cancelledc3z># UH0nUTRS5=(d SR5;v M2 g7f)rrpN)rr)rrr!s r$rread_all..ds4"L+Ja#$h(=2'D'D'F"F+Js8;)rrrz Motorway: z3 purchases already complete, their pages left alonezReading payment details for z Motorway purchaser)r (z already complete)...r TheadlessrU purchase_docsr1rxrrrrz.Reading payment details for Motorway purchase  of )...zMotorway payment breakdown: z readz(Motorway payment breakdown read FAILED: rvoidedzCarwow: z Carwow purchaser rrexist_okout_dirservice_record_ v5c_logbook_rrz,Reading payment details for Carwow purchase zCarwow payment breakdown: z&Carwow payment breakdown read FAILED: listing_urlzDealer Auction: zReading photos for z Dealer Auction purchases...r z+Reading photos for Dealer Auction purchase zDealer Auction gallery: z$Dealer Auction gallery read FAILED: )6 daily_runr!rTrr5rrrrrrr read_purchasesr read_wonr read_purchase_historyr read_purchase_invoicesr7r:extendrDrread_cancelledrrropen_reader_contextpagesnew_page enumerateread_purchase_breakdownupdater}r~rrread_purchase_photodownload_purchase_mediaread_service_history_photosrdumpsread_v5_photosread_purchase_galleryread_collection_datermakedirsread_purchase_documentsbasenamer ADVERT_URL_REsearchgroup)&cancelled_pages platformsr!rfailuresr!rrrrRfngotrcxmw_rowsmw_donectxdpagegot_nibd mw_doc_dirrlocalhistory_photosrgallery collectedcw_rowscw_donedoc_dirdocsr service_docsv5_docsda_rowsda_donems& ` r$read_allrls N$+I6I DHLNL]L]L_`L_q* s155#7#=2>?BL_G` MM   !E  a$h&=&=>6??@[)<+M+MN+]-Q-QRTHD$ i8D6!? @ &ebS *BCD C T  " i!Q R 6,,QJ1#b'*EF B#'M$Q!J-:*Eq!"L+J"LL$GM 37UK G 7)+^_`)'CCL>Qcknovkw{|k|dgCEdFAHr'*<=bQQT&VS\+WJ!55adSC,/IICIIaL3<<>EE )' 21K!!)!A!A%=IY!ZB! ! % $*,dFOUVW\U])^J%-$@$@$GE$(0(H(HQVPWYc(d#(5:1XAkN.6-Q-QRW-XN-(0(H(HP^`j(k#(BF**UBSA&>$? )1(?(?(FI((0(H(HPY[e(f#(59ZZ5FAkN'/&D&DUAmL\&]G&(0(H(HPWYc(d#(6:jj6GAlO)1(E(EeQ}M](^I(7@"3 4")/]%&UG4G ~ReR@P?QQU.W"#a%W7[!3`IIK8tCL>QVWX y "&P$QJ-8*Cq%%/X5:;%% :N$GP27GK GW )\]^)'CCL>QailmtiuyzizbeACbDAHr'*<=bQQT&VS\+?H!55aDQC,/IICIIaL3<<>EE )' 213!!'!?!?qGW!XB! ! % -.UU>-Bd-JAkN')ggll4RSTYRZ&[G&,&B&B5!MJZ&[G&(.(F(FugW^(_26**U=Mg2N,KK$?#)#A#A%=IYcj#kD8<+qt!rww?O?OPQ?R?]?]^o?pAtL+q26&i$Q"'':J:J1:M:X:XYg:hq$G&i+6:jj6N"2 3&6:jj6I"2 3)/(C(CE1]K[(\I(7@"3 4")/[%&UG4G ~ReR@P?QQU.W"#a%W7k!3pIIK6ugT#g,uUV y ("&e$QJ-;K*KqPQPUPUVcPdq$Ge27GK GW( 1def)':3w<.Hd%eS\+F!55aSWXC,/IICIIaL3<<>EE )' 21)77>>q?OP%*7*M*MeUVU\U\]^U_*`#*6:jj6IAlO$)QJE")/Z%&UG4G ~ReR@P?QQU.W"#a%W7!3IIK4UG4G ~USTq v >{a &6qc:;%% & 68<= 455 6Mr )! !!JDQCHIIJPb,r&i )! !!HB1#FGGHf&$-% $%!F@DEEFs  v >s2g? A l3+3hl33=h>0l35 i5i5"i5(A2l3Aj 6Gi:A-j 1 l3< j1 j1#j1;j1A3l35AkDk7j6j6k7j;j;A(kA-k0 l3; k7 k7#k7)Al3:Bl =A k<A,l  h;!h6/l36h;;l3> i2i-'l3-i22l3: jj jj  j.j)#l3)j..l36 k k k kk k4k/)l3/k44l3< l l  l  l  l0l+%l3+l00l33 mc SS/nSGH4nUVs/sHo3SU:XdM UPM nnU(dM)URU5 UHnURS5=(d Sn[URS5=(d URS55nURS5(aS OSnURS 5nU(aS [U5S 3OS n URSUSURSS535 URSUUSU 35 URS5 M URS5 GM7 URS5 SRU5$s snf)zThe email body for accounts: what was bought, what was paid, and the retail each car can go up at. Plain text, house style, no dashes.z.Cars bought today, confirmed on the platforms.rprHrGrrrr1zno plate total_pricerz including fees and VATrzAchievable retail .z)Not valued by BidBrain, price it by hand.r0rRz Paid z. zCSent automatically by BidBrain after the day's purchases were read. )rrrr) r"linesrr!carsr1paid paid_noteretail retail_notes r$composerx?sD>r BEH83aJ-4"738  TA%%,,*C!%% .@!%%.AD56UU=5I5I1rIUU,-FEK/v/?qA K  LL2cU"QUU62%6$78 9 LL74& 2k]C D LL   RI  LLVW 99U !9s E"E"c>Sn[R"SSX0X/SSSS9 g)zSend through the Mail app using the account already signed in on this Mac. The message text is passed as arguments, never built into the script.a on run argv set theSubject to item 1 of argv set theBody to item 2 of argv set theTo to item 3 of argv tell application "Mail" set m to make new outgoing message with properties {subject:theSubject, content:theBody, visible:false} tell m to make new to recipient with properties {address:theTo} send m end tell end run osascriptz-eTr)checkcapture_outputroN) subprocessrun)subjectbodytoscripts r$ send_mailrWs* FNNKvAdB@r&cSSKJn U(a)SRU5S3nUR"SSU5 g USUS3nUR"SS S US 35 g ) zThe sync's terminal write, both halves, through the shared bidbrain.runs.finish. A platform that could not be read is a real failure even though everything else stored fine. Split out of main() so the outcome logic is testable without a browser.rr, z could not be read purchasesFz purchases checked,  newTz Updated. rpN)bidbrainrrr6)rows_nnew_nrSrsummarys r$ _finish_syncrjs] YYx());< K0H0t< K7)1'=>r&cSSKJn SSKJn UR S5 [ R "5 S[R;nS[R;n[5nS[R;aURS5 S [R;aURS 5 S [R;aURS 5 S [R;aURS5 S[R;n[U(aSOSU=(d SS9upgUVs/sHn[U5(dMUPM n nUVs/sHn[U5(aMUPM nnU (Ga[ R"U 5n [[U 5S[U 5S35 [ R "5n [5n U GH!n SR#[%U R'S5=(d S5R)5R+55nU HnUSU ;dUS(d USU S:waM$U R'S5=(a [%US5[%U S5:HnU=(a USU:HnU(d!U R'S5(aMU(dM[ R,"USSSU S3SSS9 U RUS5 GM GM$ U (a[[U 5S35 [/5unnUH}nSR#[%UR'S5=(d S5R)5R+55nUR'S 5(aMiUR'U5US 'M [15unn[3U[55UU5unnU(dU(a[S!US"US#35 UVs/sH;oSS$;dM UR'S%5bM!UR'S&5(dM9UPM= nnU(Ga1[S'[U5S(35 U"S)S'[U5S*3S[U55 [75n[9U5HunnUR'S+5=(d SR+SS5nU(aUSOSn[U5S:aUSOUn[:R<"XSUS&UU4S,9unnUb [?US--S.5US%'[AUUUS/9US0'U"S)S3US-S4[U5S5UR'SS5S63US-[U55 M SSS5 UHnURE[GU55 M [ RH"XcS79n![[U5S8[U!5S935 [K5n"U"(a[S1U"S:[LS;35 [55n#[ RN"5=(d 0R'S=5=(d /n$[Q[ R "5U#U$5H5un%n&n'[ RR"U%S=U&U'S>3S?9 [S@U%SAU&SBU'35 M7 [U[ R "5U$5H/un%n&[ RR"U%S=U&SCS?9 [S@U%SAU&SD35 M1 [W[U5[U!5U5 U(a [SF5 g[ RX"5n(U((d1[SG5 U(a[SHSIR#U5SJ35 gSSK-n)SKU)R\R_5RaSL53n*[cU(5n+U(a3[SM5 [SN[d5 [SOU*5 [U+5 g[gU*U+5 [ Rh"U(Vs/sHoSPM sn5 [SP[U(5SQ[dSR35 U(a[SHSIR#U5SS35 ggs snfs snfs snf![Ban [S1USS2U 35 Sn A GNYSn A ff=f!,(df  GN'=f![Ban [Ss5zlK00 1##%%A''#aeeEl0b1779??ABCT7g%>):a mqQ[}>\55/bC-8H4ISQRS`QaMb4b21U8s?AEE-$8$8XX&&$!G'* 7Z)+KK$(  S\N#00 1 IAv  GGCe *+11399; <*uu[!!#ZZ]AkN/p./NFF%dLNFFKHh8 3H:]8*Tabc"Hd m7o&o1eeL)67eeI6FdKH -c+.>-?@<= >),/B3v^d/e+,)'A!a%SQ\M]L^_eeE2./t&567!eS=MO/& q!"  d 6C SYK*3s8*D 9:2(*  BtfD^DTTYZ [ 9 N --/52::<HNB!1"2C2C2EyRZ![ C  ! !#|VD6AR S Ku$9&dVT U"\11B1B1DhOKC  ! !#|VAT U Ku$9&Ast uP TCHh/ 12$$&G  ,-  K ( 345:: ;X]]002;;JGHIG 7 D  12 e^$ j'" d  gt81w89 HS\N)N+;1 => DIIh/00CDEY6 4~H"!FBqxj(@DEEF 6 2 ,QC0112 9 3A3788949s,a a a(a0 aaa,a?bBa$Ab'b0C+b=/c# b#a=7b=bb b b:!b55b:= c cc c SSKnSSKnSHvn[5n[R"USSS9nUR (aUR SOUR 5nU"U5UR5 sSSS5 s $ g![Ra=nUS:Xa"[R"SS[U55 e[S5 SnAOSnAff=fUR5 O!UR5 f=fSSS5 O!,(df  O=fUR"URS /S S S S 9nUR S:wdGMCUR"UR$-R'5R)5S S=(d S/n [R"SSU S5 [R"SU S35e)aRun work(page) against DealerKit; if the saved login has expired, log back in once with the stored password (auto_login_dealerkit.py, the same thing the daily passes do) and try again, so a job started from Dealer OS never dies on a stale session while nobody is at the Mac (2026-09-03, found live: a check asked for from Dealer OS failed on exactly that). A second failure is the real answer and is raised.rN)r)r dealerkitFr*rzQDealerKit's saved login has expired, logging in again with the stored password...zauto_login_dealerkit.pyT,)r|rrozno reason givenzADealerKit's saved login has expired and logging in again failed: )r}rrrr>r?r@rrDealerkitSessionExpiredrrecord_site_healthrr7r~ executable returncodestdoutstderrr splitlines) workr}rattemptrrYpagerr!msgs r$_dealerkit_with_reloginrGsw  !--auMC#&99399Q<#,,.D Dz   #:: ka<))+uc!fEijj  kk     NNCNN,EFW[bfps t <<1 88ahh&--/::>D  D zPurchase pricez Buyer's feeMileagePhotos Documentsz Retail pricez Due in datezPrice and fee after the chip TransportzCheck inRemovez Price and fee) rrrr documentsrv due_in_datechip transportcheckinremovefiguresc [R"USSS[USUS3US9 URS5c"USS 5 gURS 5(dgUS :Xa[R "US U/S 9 gUS;aG[R "US US:Xa/SQOSS/URS5URS5S.S9 gUS:XaMURS5(a[R "USUSUS9 [R"USS5 gUS :Xa"USS 5 gg)aWrite what one DealerKit follow up (chip, transport, checkin, remove) actually did to the car's timeline and stamp the purchases row (2026-09-05). Shared by the one car job and the automatic run so the two can never disagree. A follow up that found no DealerKit record at all records the car as not on DealerKit (dk_removed_at), so the automatic run stops asking and a later Send can start clean.r*r DealerKit:  outcomer vehicle_idNrdonerr1r)rrr)rrrrr car_price effective_feerrrsentr delivered_on)rr9DK_ITEM_LABELSrstamp_dkmark_dealerkit_pushedset_checked_in_at)ractionresultr/s r$record_followup_outcomerlsB#d)['v(>'?q AR@STZ]_ zz,' CIx( ::f      UF8< & &   UvY_O_3Kfmoteu03 0DSWWUdMe&f h 9  ::n % %  TF>,B L CIy) 8  CIx( r&c hU=(d 0RS5(aURS5(dg[R"XSUS5nUb[R"USU5 gg![ a3n[ SUSS[R"US535 SnAgSnAff=f)a0After a follow up that left a figure alone on DealerKit (a price a person typed there, or one held under another contact), read the record again and store it (2026-09-16), so Dealer OS shows what DealerKit holds, with the Update the price and fee button, and the automatic run stops asking. Never raises.leftrNr1r0z6: DealerKit could not be re read after the follow up, rZ) rrcheck_dealerkit_recordrset_dealerkit_checkrDr7r first_line)rrrr{rs r$refresh_check_if_leftrs Lb  f % %VZZ -E-Ep66tZP\I]^    " "3u:u 5  p 3u:,TUYUdUdefhkUlTmnoops:A44 B1>)B,,B1c @^^SSKJn SRST55n[SUSTSS35 U"S SUSTSS 35 UU4S jn[ U5nTH/n[TXuU5 [S[USXWS35 M1 [R"SSSUSTSS35 U$![ R a&n[R"TS S SUSU3SS9 eSnAf[a&n[R"TS S SUSU3SS9 eSnAff=f)zOne car, one or more of the four follow ups, in one DealerKit session (python3 purchases_run.py --push-dealerkit REG --items chip). Nothing here ever creates a DealerKit record.rr rc34# UHn[Uv M g7fr])r)rrs r$r+push_dealerkit_followups..s:'Q~a('szSending z for r1z to DealerKit...dealerkit-login to DealerKitcp>0nTH,n[R"UTU5X'[UTX5 M. U$r])r push_followupr)rr ractionsrs r$r&push_dealerkit_followups..works=A$224a@CF !$SV 4 r&r*r DealerKit z failed: r,rNr0r$rdealerkit_push_rowTzSent  to DealerKit.)r7r!rr7rrrrr9 RuntimeErrorrrrr6)rrr!labelsrresultsrrs`` r$push_dealerkit_followupsrsI$ YY:': :F HVHE#e*-= >? 8F85U M!RS)$/Q 3 >!$%R 9(='>?@ KK$deF85U N,[\ N  2 2 c$i 6()TUSV6W]gh  c$i 6()TUSV6W]gh s$ B55D !C** D7!DDcUc[R"5nU=(d 0RS5nUbUS:XaSO [ U5n[S[SU55$![a 0nNUf=f![ [ 4a SnN(a[USS?35 O[USS@35 [R8H"nUU ;dM [S2TUSAU U35 M$ [SBUS435 TSCSD;a [SE5 gg![Ra#n[R"TSSSU3SS9 eSnAf[an[R =(d 0nSR#S[UR%5555n [R"TSSS[&R("US53U (aSU S3OS-SS9 eSnAff=f)FaManual, one car at a time trigger: python3 purchases_run.py --push-dealerkit REG [--items price,photos,documents,retail,due_in_date]. Opens a real HEADED DealerKit window (so it can be watched and reviewed, per Mark's own "test first" plan, 2026-08-21) and pushes that purchase in as a Due In record with its photos and service history documents. items is an optional subset of price/photos/documents/retail/due_in_date (2026-08-26, Mark: "choose which data to push" from the row's own kebab menu; retail and due_in_date added the same day, "the Retail Est is the retail price that should be mapped to DK" and "lets now wire up the due in date in BB to DK"), None pushes everything, the original behaviour. A car with no mileage is refused outright by dealerkit_write.add_due_in (Steven 2026-08-27, "NEVER ALLOW Dealerkit to manually guess the mileage"): the old --allow-missing-mileage escape hatch is gone, the fix is always to enter the real mileage on the purchases page first. A DealerkitMissingMileage failure is written to run_progress.json the same way a session expired one is, so it surfaces as a real alert rather than a row that just looks unchanged. Marks db.dealerkit_pushed_at (and the matching per item columns) on success, a real bug fixed the same day this gained the items option: this function never marked it at all before, only the bulk Run modal pass (daily_run.dealerkit_purchases_pass) did, so a row pushed one at a time from the purchases page kept showing "not pushed" forever despite a genuinely successful push.rr c3t># UH-oSR5TR5:XdM)Uv M/ g7fr1Nrrr!r1s r$r!push_dealerkit..)C4aU8>>#3syy{#B4(8 8N is not in the purchases table.rzPushing r1r(rRrrz) to DealerKit...rrc>[R"UTT[5S9n[R"UTSUR S55US'[ R"5n[U4SjU5S5n[TX25nX1S'0US'UH!n[R"XU5USU'M# U$![ a SUS'Nzf=f![ a*nS[R"US 50USU'SnAMkSnAff=f) N)r max_photosr1rr{c3B># UHoSTS:XdMUv M g7f)r*Nr)rr!rs r$r/push_dealerkit..work..:s BAtWD -Aaas r followupserrorr[) r push_purchaserrrrDrr5nextrrrr) rrrows_nowrduerrrrs r$rpush_dealerkit..work)s  ..tSRgRij  #-DDT3u:W]WaWabnWopF7O$$&BBDI"38w {A L)8)F)FtTU)V{#A&  % #"F7O #  L*14??1c3J)K{#A& Ls)-B8C 8C  C  DC<<Dr*rzDealerKit push failed: r,rz; c3Z# UH!up[RX5SU3v M# g7frN)rr)rrfrs r$rrPs+^I]n0067q# UHupTRX5SU3v M g7fr(r)rrfr_LABELs r$rr[s*(hOgtqFJJq,<+=Qqc)BOgs%(zDealerKit: nothing to do items_donerrrrrruDealerKit: the price sent, £r2u, already takes the £ chip_totalz chip offrr r!rz after the send failed: r0z (after the send): failed, rzC after the send: the record was not found yet, the next run does itz' (after the send): record not found yetz (after the send): rr{rTzPushed ralready_existedz& was already in DealerKit's stocklist.z- created as a new Due In record in DealerKit.r$ Vehicle id: rrHrGz9V5 and service history not sourced for this platform yet.)r7r!rr5r#rrKrFOLLOWUP_ACTIONSrsortedr7rrr9rD LAST_OUTCOMESrrrrrrrrrrr6 SEND_ITEMS)r1rr!rfollowrrrrso_farr*noterrreslabel_kr,rs`` @@r$push_dealerkitr<s6$   D C4CT JC C5 ?@AA" O$D$D E EF s5;B'#o.N.N*OO'VF^<< HSZL3v;-r#h-@Q RS 8CJ<}!EF8(.$Fzz*%+Htyy(hvV^VdVdVfOg(hhh)#d)[$JG l   U6,3G03 0DSWWUdMe&f h f\* */DS/I/I  $ $SZx47GGK4HQTQXQXYhQi*j l  ! !#d)[$A#''+BVB[Z[\`Aaawx{xxAMyNySRSTXxYYb#c&0 2 JJw  &3E::k*0b779""1( 777    ! !#d)[JugMefijqfres:t{E F Bug8WG H WW\ " *  ! !#d)[JugNQ;RXb c BugDE F #Ec 2 Bug0Y0@A B:zz'& s5z6'?;KK$dgc%j\,PQ   EFG LMN(( > Bvbzl"Xb\N3 4) L -. /0  :44 IJ5U  2 2 c$i8OPQs6SYcd   ,,2^PTPZPZP\I]^^ c$i 738O7PQtzWfgmfnnoUpACDJT U  s%> QT',R  T'B T""T'c^^^SSKJn [R"5n[ SU5S5=(d [ SU5S5mT(d[ R "S5e[RR[RR[RR[55SSS 5m[R"[RRT5S S 9 [TS S S9nURS5 SSS5 U4SjmT"S[ R R#5R%SS9STS35 U"SSTS35 ['UU4Sj5 [ R,"SS S5 g!,(df  N=f![(a*nT"S[ R*"US535 SnANYSnAff=f)a~Look at DealerKit's costs screen for one car and write every step down, saving nothing (2026-09-16, Steven: "learn the steps and screens properly to make this work every time"). The car is the first live one on DealerKit whose price never landed there (db.dk_figures_missing), else the first live car on DealerKit. Written to data/logs/dealerkit_survey.log for the diagnostics pull.rr c3`# UH$n[R"U5(dM Uv M& g7fr])rdk_figures_missingrr!s r$r#survey_dealerkit..s<4a2#8#8#;4s. .Nc3# UH<n[R"U5(dM URS5(aM8Uv M> g7f)rN)r on_dealerkitrr@s r$rrAsIMRX\ST`b`o`opq`rQ{||A|ABP|QQQX\sAA AzNo car on DealerKit to survey.rUlogszdealerkit_survey.logTr0wrqrrrpc>[U5 [TSSS9nURUS-5 SSS5 g!,(df  g=f![a gf=f)Nrrqrrrq)r7rwriterD)linerlog_paths r$saysurvey_dealerkit..saysJ d  hg6!t $766   s+ A7A AAA AAzDealerKit costs screen survey seconds)timespecz, car r1rz(Looking at DealerKit's costs screen for cd>[R"UTS[R"T5T5$)Nr1)rsurvey_expense_form_default_supplier)rrrJs r$ra"survey_dealerkit..sC_-P-PQUWZ[`WacrdEdEFIdJLO.Pr&zSurvey stopped: rdealerkit_surveyz0Survey written to data/logs/dealerkit_survey.log)r7r!rr5r#r RunSkippedr}r~rdirnameabspath__file__rKrrGrnowrrrDrr6)r!rrrrIrrJs @@@r$survey_dealerkitrXs$   D <4??ww||BGGOOBGGOOH,EFPVXnoHKK)D9 hg .!   / ():):)>)>)@)J)JT])J)^(__efijofpeqrs #KCPUJ# UH-oSR5TR5:XdM)Uv M/ g7frrrs r$r"check_dealerkit..rrNrz Checking r1r(rRz) on DealerKit...rz on DealerKitc>UR[RSSS3SSS9 [R "UTS5$![a N&f=f)Nrbase_urlz /vehiclesdomcontentloadedi0u) wait_untilror1)gotorSITESrDrr)rrs r$_checkcheck_dealerkit.._checksd  II{3J?@ J!3U  D55dCJGG   s(A AAr*rzDealerKit check failed: r,rzDealerKit checked: vehicle rrimagesz photos, rz documentsz-DealerKit checked: no record for this car yetTdealerkit_check_rowz$ has no DealerKit record at all yet.zChecked z on DealerKit.r0zImages on record: zDocuments on record: purchase_cost_value purchase_costzPurchase price on record: £gznot setzRetail price on record: retail_price)r7r!rr5r#rr7rrrr9rrrrr6_expense_item_amountr)r1r!rrbrr_costrs` @r$check_dealerkitrms6$   D C4CT JC C5 ?@AA Ic%j\CK=0A BC 9SZL !FGH(0#d)[$9L9Q8RRTU[U_U_`hjkUlTmmvw}xBxBCNPQxRwSS]^*Y`jl+t,3u:v.  )4CJ-RS L -. /0 vh/0 12 !&"5!6 78 JJ, -E }44VZZ5PQ &e>Otq ':U^&_ `a $39.3I3UTC~./ /[d e gh3  2 2 c$i8PQRPS6TZde s8 HI$IIc  ^SSKJn [U4Sj[R"55S5nU(d[ TS35eUSnUS;a[ TSUS 35eUUS US UR S 5UR S 5UR S5UR S5UR S5S.n[SUS SUS SUS35 U"SSUS SU35 [5nUS:XaSOSn[R"XVSS9nUR(aURSOUR5nUS:XGa[R"XS 5n U (aURU 5 UR[!U55 ["R$R'[(SSUS 5n [R*"U5n U (a'[R,"X/U 5n U (aU SUS'[R."U5n U (a7[R,"XU 5n U (a[0R2"U 5US '[R4"U5nU(a7[R,"XU 5n U (a[0R2"U 5US!'[R6"XS 5nU(a7[R,"XU 5n U (a[0R2"U 5US"'[R8"XS 5nU(aUUS#'GO[:R"XS 5n U (aURU 5 UR S$5SLUS%'UR[!U55 ["R$R'[(SSUS 5n[:R6"XS 5nU(a9[:R,"XU5n [0R2"U =(d U5US"'["R<"USS&9 [:R>"XS US'9nUVs/sH:n["R$RAU5RCS(5(dM8UPM< nnUVs/sH:n["R$RAU5RCS)5(dM8UPM< nnU(a[0R2"U5US*'U(a[0R2"U5US+'[:R8"XS 5nU(aUUS#'URG5 SSS5 [RJ"U/5 [LRN"S.SS/US SUS035 [US S1US035 gs snfs snf![Da-nURG5 [IS,US SUS-U35UeSnAff=f!,(df  N=f)2aManual, one car at a time re read of a single already known purchase straight off its own platform, Motorway or Carwow, whichever it was bought on (python3 purchases_run.py --check-purchase REG, 2026-08-26, Mark: replacing the per row kebab's push options with "a check motorway or carwow button"). The single car version of read_all()'s own per platform detail pass, same reads, same functions, just for one row rather than every purchase in one sweep, useful when the platform had not finished processing yet at first capture (Motorway "Reviewing documents", a Carwow document bundle not yet available) and the daily 15:45 sync has not come back round to it. db. record_purchases' own fill once, never overwrite rules (see its own docstring) mean this only ever adds what is now newly available, it can never clobber a figure already confirmed. Writes to run_progress.json the same way check_dealerkit does, so a failure surfaces on the page rather than only in the log.rr c3># UHFn[R"URS55[R"T5:XdMBUv MH g7fr)r _norm_regrrs r$r#refresh_purchase..s=J 31aeeEl3r||C7HHQ 3s AA ANrrr1z was bought on z), which has no re read for purchases yet.rr1rRrrror)rrr1rRrrrorz Re reading r(z) on r)r"z on rHr r Tr*rUr,rxrrrrrrr0r2r4r5rrzCould not re read r$purchase_checkzRe read rpz updated from )(r7r!r#rr5rrr7rrr>r?r@r rBrCr r}r~rrrDrErFrrGrHrIrJr rKrLrMrrDrrrrr6)r1r!rrrrsiterYrZr]r^rr_r`rrarbrerfrrgrhrs` r$refresh_purchaserts"$J 1 1 3JKOQH C5 ?@AA #H--C5z9bcdd . V$,,x(g&||M2||M2 C KE |2c&k]%z EF i;s5zl$xjAB  a%3z))!DA #  ! s||~: \:%55e=OPJJrN =-.  WW\\$UT  44U;$<#:d#BK  =-.'',,tV_c%jQ 66u->PQ"::57SE(, 53CG(DC % Gd355e=OY`a+/e4a2773C3CA3F3Q3QRc3d4 e&*]dbgg.>.>q.A.L.L^.\1d],0JJ|,DC(),0JJw,?C()"77=?QR -6C)*  BKK $(3u:,d8*A(NO SZLxj 23! f] \ IIK!3CJC>rz--survey-dealerkitrR--check-dealerkitrec,[[SS55$)Nrz5Usage: python3 purchases_run.py --check-dealerkit REG)rmrzrr&r$raras/ *,c d3fr&--check-purchaserrc,[[SS55$)Nrz4Usage: python3 purchases_run.py --check-purchase REG)rtrzrr&r$raras-= )+a b.dr&r)r@rB))NNr])r)N)r,)K__doc__rrr}r}rplaywright.sync_apir dealer_configrrrrrbidbrain.pricingrbidbrain.readersr r r r r rLr%r=r4rTr~rTrUrVrrrrrrrrrrrrrrrrrr r rrrrlrxrrrrrrrrrrrr<rXrmrt__name__rzrguardrrr&r$rsV8  /77#RR+9O\0.0 54 wwrwwx01  T6?3--.Ab'TBJ*$0@ 6 U*a<l@c~  ' iX 0!/@& ?LF^^<,Mi$;*=8{_iu}, . )F p >"GDWKt#^LHhVi4X z  SXX% #^4 sxx ' 5 '/  ) %'78  ( (+f g sxx ' #&d e ;%Ir&