+ j8Rt^RIHt^RIt^RIt^RIt^RIHt^RIH t ]!] 4P4PPt ]!] 4]P9d#]PP!^]!] 44^RIHtHtHtRtRZRRllt/R R bR R bR RbRRbRRbRRbRRbRRbRRbRRbRRbRR bR!R"bR#R$bR%R&bR'R(bR)R*bR+R,R-R.R/R0R1R2R3R4R5R6R7R8R9R:R;RR?R@RARBRCRDRERF/CtRGt]P2!RH4t]P2!RI4t]P2!RJ]P84tRKRLlt]P2!RM4t]P2!RN4t ]P2!RO]P84t!]P2!RP]P84t"]P2!RQ]P84t#RRRSlt$!RTRU]4t%!RVRW]%4t&!RXRY]%4t'R#)[u)Email-driven property source ingestion. Reads property-alert emails from Mail.app via osascript. Bypasses scraper bot blocks because emails are pushed officially by the platforms after the user sets up a saved search. Supports (as of 2026-05-27): - Leggett (notifications@hestia.leggett.fr) — daily ~10 properties/email - Properstar (noreply@properstar.com) — saved-search updates Why this matters: leggett.fr is Cloudflare-blocked for our scraper, but the email alerts arrive freely with the same properties + a parseable REF that maps to frenchestateagents.com URLs (verified — REF format identical to what the legacy scraper used before being blocked). Adding a new sender: 1. Set up a saved search on the platform (logged in, "save this search + email me alerts"). Verify alerts arrive. 2. Inspect a sample email body — find regex patterns for url/price/region. 3. Add a parse_() function returning Iterator[PropertyHit]. 4. Add a Source subclass mirroring the pattern below. 5. Register in sources/__init__.py. 6. Add to campaigns.yaml. ) annotationsN)Path)Iterator) PropertyHitSearchCriteriaSourceu on run argv set senderFilter to item 1 of argv set daysBack to (item 2 of argv) as integer set cutoffDate to (current date) - (daysBack * days) tell application "Mail" set targetAccount to first account whose name is "Gmail" set theBox to first mailbox of targetAccount whose name is "INBOX" set msgs to messages of theBox set results to "" set i to 1 set total to count of msgs if total > 500 then set total to 500 repeat while i ≤ total set m to item i of msgs try set theDate to date received of m if theDate < cutoffDate then exit repeat set s to (sender of m as string) if s contains senderFilter then set results to results & "===EMAIL===" & return set results to results & "SENDER: " & s & return set results to results & "DATE: " & (theDate as string) & return set results to results & "SUBJECT: " & (subject of m as string) & return set results to results & "BODY:" & return try set results to results & (content of m as string) & return end try end if end try set i to i + 1 end repeat return results end tell end run c$V^8dQhRRRRRR/#) sender_filterstr days_backintreturnz list[dict])formats"\/Users/jonathan/Documents/Zakelijk/ClaudeOS/03_Lab/farmmatch/scraper/sources/email_ingest.py __annotate__rTs!####Z#c F\P!RR\V\V4.RR^()]+zEUR\s*([\d.,]+)z(\d+)\s*Kamers?z(\d+)\s*Bed\.?z(\d+)\s*Bad\.?c V^8dQhRRRR/#r`r)rs"rrrsA A DA %:A rc#~a"VPRR4p\PV4p\V4Uau.uFGo\;QJdV3RlR4F 'gK RM RM!V3RlR44'gKESNKI ppVEFpVP V4pV\ ^VR, 4V^,pRp\PV4p V 'd<\V P^4PRR4PR R44pRp \PV4p V 'd\V P^44p Rp \PV4p V 'd\V P^44p VPR 4pVUu.uFkpVP!4'gKVP!4R 8wgK2\#VP!44^8gKRR V9gK[VP!4NKm ppV'dV^,R ,MRp\%TRV'd VR,MRVRV V RVRVPR4RVPR4/R7xEK R#uupi \dELzi;i \dELRi;i \dEL*i;iuupi5i)zYield PropertyHits from one Properstar alert email body. Properstar emails contain 1-N listing blocks; each has a properstar.* URL, a description line, a room/bed/bath/type line, and EUR price. rrdc3,<"TF qS9xK R#5iNr).0markerus& r #parse_properstar..s_)^v{)^sTFi N.rerrfEURrgproperstar_emailrjzProperstar listingrlrorprrqr)rrrsrtrurvrxroomsry)z /listing/z/annoncez /property/z /eigendom/)r+RE_PROPERSTAR_URLfindallsetanyfindmaxRE_PROPERSTAR_PRICErr r}rrRE_PROPERSTAR_BEDSRE_PROPERSTAR_ROOMSr%r&r)r)r2rurlsr listing_urlsrrr5nearbyrurbedsrrrmr3r desc_linesros& ` rparse_properstarrso 99VR D  $ $T *Dt9a 3_)^_333_)^_ _ 9 iinc!S3Y's3 ' ' /  BHHQK//R8@@bIJ  & &v .  288A;' ' ' /  BHHQK(  T"!& !&2xxz  hhjE1 69"((*o6J OT\^ BHHJ  .8jmD)R %&1+c"7K{9!5eii/  M          s6J=I= I= I=I=AJ=:JJ=6JJ=0J& J=J8<J8J83J8<J8A2J= J J=JJ= J#J="J##J=& J51J=4J55J=cF]tRtRtRtRtRtRRltRtR R R llt R t R#) _EmailSourceBasei;uShared logic: read once per orchestrator run, then serve from cache. The orchestrator calls search() once per (region × source × campaign). For email sources the data isn't region-specific (emails come pre- filtered by the user's saved searches), so we parse once on the first call and serve cached PropertyHits on subsequent calls, with downstream criteria filters applied per call. Fc RVnR#r_cacheselfs&r__init___EmailSourceBase.__init__Fs 04 rcV^8dQhRR/#r rzlist[PropertyHit]r)rs"rr_EmailSourceBase.__annotate__Is""("rc \hr)NotImplementedErrorrs&r_load_EmailSourceBase._loadIs!!rc PVPfVP4VnR#R#r)rrrs&r _ensure_cache_EmailSourceBase._ensure_cacheLs ;; **,DK rNc$V^8dQhRRRRRR/#)r criteriar known_urlszset[str] | Nonerrbr)rs"rrrPs#~*6Krc# "VPVP9dR#VP4T;'g \4pVPEFpVP V9dKVP 'd0VP'dVPVP 8wdKXVP'd(VP;'g^VP8dKVP'd0VP'dVPVP8dKVP'd)VP;'g^VP8dEK VxEK R#5ir) rv countriesrrrrrregionrw min_priceru max_price min_bedroomsrx)rrrknownhits&&& rr_EmailSourceBase.searchPs   4>> 1  ##ce;;Cww%3#4#4#49J9Jhoo9]!!!syy~~A9K9K&K!!!ciiiCII@R@R4R$$$#,,*;*;!x?T?T)TIs:6E'?E'9E' .E':E' %E'3E'.E'4E' E'rr) __name__ __module__ __qualname____firstlineno____doc__ requires_authrrrr__static_attributes__rrrrr;s'M5"'rrc8]tRtRtRtR.tRRltRRltRtR #) LeggettEmailSourceidrirlcV^8dQhRR/#r rztuple[bool, str]r)rs"rrLeggettEmailSource.__annotate__h44(4rc \R^R7pT'gR#R\ T4 R23# \d#pR\T4P 3uRp?#Rp?ii;i)leggettr FNT emails in 30d)Fz+no emails in 30d (subscribe at leggett.fr?)r7 Exceptiontyperr)rr/es& rhealthLeggettEmailSource.healthhs^ 0!)r:FG GF }N333  0T!W--./ / 0 ) AA AAcV^8dQhRR/#rr)rs"rrrq(rc j\R^R7p.pVFpVP\V44K V#)rr)r7extendrrr/hitsrs& rrLeggettEmailSource._loadqs3i26"$A KK a( ) rrN rrrrnamerrrrrrrrrds DI4rrc:]tRtRtRt.R OtRRltRRltRtR#) ProperstarEmailSourceiyrcV^8dQhRR/#rr)rs"rr"ProperstarEmailSource.__annotate__}rrc \R^R7pT'gR#R\ T4 R23# \d#pR\T4P 3uRp?#Rp?ii;i)properstar.comrFNTr)Fzrsr0#   (^ # # % , , 3 3 z?#(("HHOOAs:'==#L#V!o! !+!( ! x ! y ! F! h! F! ! F! g! ! !,!  !!" 5#!$ h  F W h h y x F(4*2&A! J` 45::GH**-r}}=: ~JJLMjj!34jj!3R]]CZZ 12==Ajj!2BMMBA L&v&R)*,r