From a8e60b3aecbf6b208d35cb4fe9e285b117408ed2 Mon Sep 17 00:00:00 2001
From: Dhirendra Singh <dhi.singh@gmail.com>
Date: Tue, 16 Mar 2021 11:04:11 +1100
Subject: [PATCH] Write out map of PlanId > AgentId > GroupId; #4

---
 R/plan.R                                   |  34 ++
 tests/expected/4.plan/plan2agent2group.csv | 428 +++++++++++++++++++++
 tests/testthat/testPlan.R                  |  12 +-
 3 files changed, 473 insertions(+), 1 deletion(-)
 create mode 100644 tests/expected/4.plan/plan2agent2group.csv

diff --git a/R/plan.R b/R/plan.R
index 733f67c..f620f05 100644
--- a/R/plan.R
+++ b/R/plan.R
@@ -514,4 +514,38 @@ combinePlans <- function(groupIds,
     nextId <- nextId + length(unique(all$PlanId))
   }
   write.table(allplans, out_csv, row.names=FALSE, col.names=TRUE, quote=TRUE, sep=",", append=FALSE)
+}
+
+
+writePlan2Agent2GroupMap <- function(groupIds,
+                               matched_persons_csv_prefix, 
+                               plans_csv,
+                               out_csv) {
+  # example inputs
+  # groupIds <- getGroupIds('../data/vistaCohorts.csv.gz')
+  # matched_persons_csv_prefix <- '../output/3.match/match_'
+  # plans_csv <- '../output/4.plan/plan.csv'
+  # out_csv <- '../output/4.plan/plan2agent2group.csv'
+  
+  echo(paste0('Creating map of PlanID to AgentId in ', out_csv, '\n'))
+  plans<-read.csv(plans_csv, header=T, stringsAsFactors=F, strip.white=T)
+  plan2agent <- data.frame()
+  for (gid in groupIds) {
+    cohort<-read.csv(paste0(matched_persons_csv_prefix, gid, ".csv"), header=T, stringsAsFactors=F, strip.white=T)
+    cohortAgentIds <- unique(cohort$AgentId)
+    cohortPlans <- plans %>% filter(GroupId == gid)
+    cohortPlanIds <- unique(cohortPlans$PlanId)
+    if (length(cohortAgentIds) != length(cohortPlanIds)) {
+      echo(paste0("For Group ", gid, 
+                  ", the number of generated vista-like plans (", length(cohortPlanIds), 
+                  ") is not the same as the number of sampled census agents (", length(cohortAgentIds), 
+                  "). Skipping this group." ))
+      next
+    }
+    df <- data.frame(PlanId=cohortPlanIds, AgentId=cohortAgentIds)
+    df$GroupId <- gid
+    plan2agent <- rbind(plan2agent, df)
+  }
+  write.table(plan2agent, out_csv, row.names=FALSE, col.names=TRUE, quote=TRUE, sep=",", append=FALSE)
+  
 }
\ No newline at end of file
diff --git a/tests/expected/4.plan/plan2agent2group.csv b/tests/expected/4.plan/plan2agent2group.csv
new file mode 100644
index 0000000..2480674
--- /dev/null
+++ b/tests/expected/4.plan/plan2agent2group.csv
@@ -0,0 +1,428 @@
+"PlanId","AgentId","GroupId"
+1,"213021342P2856225",1
+2,"213021342P2856544",1
+3,"213021342P2856532",1
+4,"213021342P2856080",1
+5,"213021342P2856288",1
+6,"213021342P2856540",1
+7,"213021342P2855370",1
+8,"213021342P2855433",1
+9,"213021342P2856174",1
+10,"213021342P2857306",1
+11,"213021342P2856518",1
+12,"213021342P2856082",1
+13,"213021342P2855368",1
+14,"213021342P2862001",1
+15,"213021342P2855660",1
+16,"213021342P2857475",1
+17,"213021342P2856204",1
+18,"213021342P2861650",1
+19,"213021342P2861920",1
+20,"213031351P3698884",1
+21,"213031351P3696688",1
+22,"213031351P3698281",1
+23,"213031351P3698969",1
+24,"213031351P3698758",1
+25,"213031351P3698574",1
+26,"213031352P897951",1
+27,"213031352P897879",1
+28,"213031352P898786",1
+29,"213031352P899064",1
+30,"213031352P898218",1
+31,"213031352P897996",1
+32,"213031352P898050",1
+33,"213031352P899113",1
+34,"213031352P900119",1
+35,"213031352P898527",1
+36,"213031352P898369",1
+37,"213031352P899067",1
+38,"213031352P897897",1
+39,"213031352P902780",1
+40,"213021342P2854997",1
+41,"213021342P2857938",1
+42,"213021342P2855019",1
+43,"213021342P2854099",1
+44,"213021342P2854952",1
+45,"213021342P2854639",1
+46,"213021342P2854744",1
+47,"213021342P2854251",1
+48,"213021342P2854515",1
+49,"213021342P2861655",1
+50,"213021342P2853789",1
+51,"213021342P2854134",1
+52,"213021342P2861951",1
+53,"213021342P2861768",1
+54,"213021342P2854140",1
+55,"213021342P2854473",1
+56,"213021342P2853966",1
+57,"213021342P2858160",1
+58,"213021342P2854173",1
+59,"213021342P2854704",1
+60,"213021342P2853848",1
+61,"213021342P2854567",1
+62,"213021342P2861914",1
+63,"213021342P2855021",1
+64,"213031351P3698130",1
+65,"213031351P3697709",1
+66,"213031351P3697994",1
+67,"213031351P3697562",1
+68,"213031351P3697759",1
+69,"213031351P3697847",1
+70,"213031351P3697454",1
+71,"213031351P3702532",1
+72,"213031351P3697573",1
+73,"213031351P3702442",1
+74,"213031352P899424",1
+75,"213031352P897426",1
+76,"213031352P896900",1
+77,"213031352P897787",1
+78,"213031352P897063",1
+79,"213031352P897627",1
+80,"213031352P896660",1
+81,"213031352P899745",1
+82,"213031352P899422",1
+83,"213031352P897095",1
+84,"213031352P896818",1
+85,"213031352P899773",1
+86,"213031352P897480",1
+87,"213031352P896797",1
+88,"213031352P896617",1
+89,"213031352P897474",1
+90,"213031352P899314",1
+91,"213031352P897802",1
+92,"213031352P902876",1
+93,"213031352P897304",1
+94,"213031352P899313",1
+95,"213031352P896755",1
+96,"213021342P2860113",2
+97,"213021342P2857603",2
+98,"213021342P2852537",2
+99,"213021342P2852538",2
+100,"213021342P2848185",2
+101,"213021342P2861804",2
+102,"213021342P2857761",2
+103,"213021342P2857333",2
+104,"213021342P2858919",2
+105,"213021342P2859001",2
+106,"213031351P3701983",2
+107,"213031351P3699831",2
+108,"213031352P902243",2
+109,"213031352P899464",2
+110,"213031352P900187",2
+111,"213021342P2857974",2
+112,"213021342P2856830",2
+113,"213021342P2857095",2
+114,"213021342P2857961",2
+115,"213021342P2857233",2
+116,"213021342P2857109",2
+117,"213031351P3701359",2
+118,"213031351P3701377",2
+119,"213031351P3699155",2
+120,"213031351P3699232",2
+121,"213031351P3699554",2
+122,"213031352P901759",2
+123,"213031352P889088",2
+124,"213021342P2849361",3
+125,"213021342P2848267",3
+126,"213021342P2849392",3
+127,"213021342P2859208",3
+128,"213021342P2849557",3
+129,"213021342P2848428",3
+130,"213021342P2859221",3
+131,"213021342P2849384",3
+132,"213021342P2848740",3
+133,"213021342P2848806",3
+134,"213021342P2849270",3
+135,"213031351P3702034",3
+136,"213031351P3702087",3
+137,"213031351P3702142",3
+138,"213031351P3694306",3
+139,"213031351P3694639",3
+140,"213031351P3694875",3
+141,"213031351P3694499",3
+142,"213031351P3694329",3
+143,"213031351P3694751",3
+144,"213031351P3694764",3
+145,"213031351P3694584",3
+146,"213031352P902450",3
+147,"213031352P893322",3
+148,"213031352P896152",3
+149,"213031352P892917",3
+150,"213031352P893383",3
+151,"213021342P2859513",3
+152,"213021342P2859592",3
+153,"213021342P2859801",3
+154,"213021342P2859893",3
+155,"213021342P2860005",3
+156,"213021342P2861209",3
+157,"213021342P2845347",3
+158,"213021342P2845019",3
+159,"213021342P2845728",3
+160,"213021342P2845382",3
+161,"213021342P2847513",3
+162,"213021342P2845236",3
+163,"213021342P2845661",3
+164,"213021342P2846276",3
+165,"213021342P2846460",3
+166,"213021342P2845370",3
+167,"213021342P2845752",3
+168,"213021342P2844836",3
+169,"213021342P2847123",3
+170,"213021342P2844941",3
+171,"213021342P2844331",3
+172,"213021342P2844985",3
+173,"213021342P2846917",3
+174,"213021342P2844964",3
+175,"213021342P2858618",3
+176,"213021342P2845415",3
+177,"213021342P2845085",3
+178,"213021342P2852819",3
+179,"213021342P2845741",3
+180,"213021342P2844213",3
+181,"213021342P2845710",3
+182,"213021342P2847221",3
+183,"213021342P2846120",3
+184,"213021342P2846573",3
+185,"213021342P2845410",3
+186,"213021342P2846068",3
+187,"213021342P2852908",3
+188,"213021342P2858780",3
+189,"213021342P2844763",3
+190,"213021342P2858601",3
+191,"213031351P3700124",3
+192,"213031351P3700267",3
+193,"213031351P3700301",3
+194,"213031351P3700317",3
+195,"213031351P3700357",3
+196,"213031351P3700386",3
+197,"213031351P3700430",3
+198,"213031351P3701716",3
+199,"213031351P3701720",3
+200,"213031351P3701735",3
+201,"213031351P3701840",3
+202,"213031351P3692673",3
+203,"213031351P3693257",3
+204,"213031351P3693244",3
+205,"213031351P3693328",3
+206,"213031351P3692694",3
+207,"213031351P3692479",3
+208,"213031351P3693073",3
+209,"213031351P3693004",3
+210,"213031351P3699721",3
+211,"213031351P3696934",3
+212,"213031351P3693283",3
+213,"213031351P3693657",3
+214,"213031351P3699805",3
+215,"213031351P3693790",3
+216,"213031351P3692856",3
+217,"213031351P3692097",3
+218,"213031351P3692838",3
+219,"213031351P3693476",3
+220,"213031351P3699795",3
+221,"213031351P3692436",3
+222,"213031351P3693540",3
+223,"213031351P3692067",3
+224,"213031352P900606",3
+225,"213031352P900634",3
+226,"213031352P901794",3
+227,"213031352P901966",3
+228,"213031352P902025",3
+229,"213031352P902163",3
+230,"213031352P890833",3
+231,"213031352P889666",3
+232,"213031352P891976",3
+233,"213031352P890509",3
+234,"213031352P889583",3
+235,"213031352P896045",3
+236,"213031352P895754",3
+237,"213031352P895785",3
+238,"213031352P890723",3
+239,"213031352P891538",3
+240,"213031352P891619",3
+241,"213031352P891361",3
+242,"213031352P891952",3
+243,"213031352P890654",3
+244,"213031352P899975",3
+245,"213031352P895740",3
+246,"213031352P890872",3
+247,"213031352P889385",3
+248,"213031352P891996",3
+249,"213031352P889944",3
+250,"213031352P889192",3
+251,"213031352P889748",3
+252,"213031352P890771",3
+253,"213031352P889987",3
+254,"213031352P889883",3
+255,"213031352P891535",3
+256,"213031352P891557",3
+257,"213031352P889327",3
+258,"213031352P900024",3
+259,"213031352P890946",3
+260,"213031352P902650",3
+261,"213031352P889334",3
+262,"213021342P2860311",4
+263,"213021342P2860346",4
+264,"213021342P2860381",4
+265,"213021342P2860385",4
+266,"213021342P2860416",4
+267,"213021342P2860456",4
+268,"213021342P2860524",4
+269,"213021342P2860659",4
+270,"213021342P2860694",4
+271,"213021342P2861366",4
+272,"213021342P2850427",4
+273,"213021342P2850448",4
+274,"213021342P2851892",4
+275,"213021342P2849441",4
+276,"213021342P2850204",4
+277,"213021342P2848530",4
+278,"213021342P2853079",4
+279,"213021342P2853153",4
+280,"213021342P2848677",4
+281,"213021342P2851028",4
+282,"213021342P2850174",4
+283,"213021342P2853312",4
+284,"213021342P2852670",4
+285,"213021342P2852633",4
+286,"213021342P2848986",4
+287,"213021342P2850084",4
+288,"213021342P2850551",4
+289,"213021342P2848431",4
+290,"213021342P2850758",4
+291,"213021342P2849503",4
+292,"213021342P2852574",4
+293,"213021342P2849146",4
+294,"213021342P2849651",4
+295,"213021342P2853559",4
+296,"213021342P2849587",4
+297,"213021342P2848742",4
+298,"213021342P2861526",4
+299,"213021342P2861984",4
+300,"213031351P3700728",4
+301,"213031351P3700760",4
+302,"213031351P3700979",4
+303,"213031351P3700995",4
+304,"213031351P3702057",4
+305,"213031351P3695622",4
+306,"213031351P3695614",4
+307,"213031351P3695435",4
+308,"213031351P3694951",4
+309,"213031351P3694663",4
+310,"213031351P3694690",4
+311,"213031351P3697213",4
+312,"213031351P3697165",4
+313,"213031351P3695210",4
+314,"213031351P3694446",4
+315,"213031351P3699931",4
+316,"213031351P3695184",4
+317,"213031351P3695470",4
+318,"213031351P3695458",4
+319,"213031351P3695679",4
+320,"213031351P3694932",4
+321,"213031351P3699989",4
+322,"213031351P3695575",4
+323,"213031351P3694651",4
+324,"213031351P3695396",4
+325,"213031352P900986",4
+326,"213031352P901060",4
+327,"213031352P901126",4
+328,"213031352P901218",4
+329,"213031352P901310",4
+330,"213031352P901402",4
+331,"213031352P901442",4
+332,"213031352P902515",4
+333,"213031352P892510",4
+334,"213031352P892743",4
+335,"213031352P894919",4
+336,"213031352P894858",4
+337,"213031352P892696",4
+338,"213031352P892497",4
+339,"213031352P894406",4
+340,"213031352P895087",4
+341,"213031352P893999",4
+342,"213031352P894340",4
+343,"213031352P894206",4
+344,"213031352P896268",4
+345,"213031352P893652",4
+346,"213031352P895873",4
+347,"213031352P894205",4
+348,"213031352P894195",4
+349,"213031352P894589",4
+350,"213031352P896362",4
+351,"213031352P893914",4
+352,"213031352P892945",4
+353,"213031352P894085",4
+354,"213031352P900372",4
+355,"213031352P900342",4
+356,"213031352P893804",4
+357,"213031352P894237",4
+358,"213031352P893157",4
+359,"213031352P893741",4
+360,"213031352P900336",4
+361,"213021342P2859831",4
+362,"213021342P2852897",4
+363,"213021342P2846832",4
+364,"213031351P3693822",4
+365,"213031351P3693639",4
+366,"213031352P891668",4
+367,"213031352P891754",4
+368,"213021342P2860479",5
+369,"213021342P2860543",5
+370,"213021342P2860696",5
+371,"213021342P2860839",5
+372,"213021342P2860866",5
+373,"213021342P2852132",5
+374,"213021342P2851878",5
+375,"213021342P2852049",5
+376,"213021342P2851917",5
+377,"213021342P2851325",5
+378,"213021342P2851086",5
+379,"213021342P2850963",5
+380,"213021342P2851061",5
+381,"213021342P2852041",5
+382,"213021342P2852755",5
+383,"213021342P2852012",5
+384,"213021342P2861637",5
+385,"213021342P2852704",5
+386,"213021342P2852124",5
+387,"213031351P3701113",5
+388,"213031351P3701177",5
+389,"213031351P3702255",5
+390,"213031351P3696145",5
+391,"213031351P3696398",5
+392,"213031351P3697239",5
+393,"213031351P3696373",5
+394,"213031351P3696173",5
+395,"213031352P901322",5
+396,"213031352P901376",5
+397,"213031352P901411",5
+398,"213031352P901468",5
+399,"213031352P901469",5
+400,"213031352P901511",5
+401,"213031352P901664",5
+402,"213031352P896401",5
+403,"213021342P2859853",5
+404,"213021342P2860043",5
+405,"213021342P2860059",5
+406,"213021342P2847827",5
+407,"213021342P2847688",5
+408,"213021342P2848138",5
+409,"213021342P2847278",5
+410,"213021342P2848015",5
+411,"213021342P2852480",5
+412,"213021342P2846836",5
+413,"213021342P2847856",5
+414,"213031351P3693683",5
+415,"213031351P3693940",5
+416,"213031351P3693769",5
+417,"213031351P3694077",5
+418,"213031351P3694112",5
+419,"213031352P900764",5
+420,"213031352P900781",5
+421,"213031352P900861",5
+422,"213031352P900889",5
+423,"213031352P892190",5
+424,"213031352P892223",5
+425,"213031352P896097",5
+426,"213031352P891915",5
+427,"213031352P892352",5
diff --git a/tests/testthat/testPlan.R b/tests/testthat/testPlan.R
index 641baad..cbbf4e9 100644
--- a/tests/testthat/testPlan.R
+++ b/tests/testthat/testPlan.R
@@ -25,7 +25,13 @@ test_that("VISTA-like plans generation works", {
       '../actual/4.plan/plan.csv'
     )
   )
-
+  capture_output(
+    writePlan2AgentMap(getGroupIds('../data/vistaCohorts.csv.gz'),
+                       '../expected/3.match/match_',
+                       '../expected/4.plan/plan.csv',
+                       '../actual/4.plan/plan2agent2group.csv'
+    )
+  )
   files<-c(
     'analysis-start-times-by-activity-qq.pdf',
     'analysis-end-times-by-activity-qq.pdf',
@@ -46,5 +52,9 @@ test_that("VISTA-like plans generation works", {
   
   expect_true(file.exists('../actual/4.plan/plan.csv'))
   expect_true(md5sum('../actual/4.plan/plan.csv') == md5sum('../expected/4.plan/plan.csv'))
+  
+  expect_true(file.exists('../actual/4.plan/plan2agent2group.csv'))
+  expect_true(md5sum('../actual/4.plan/plan2agent2group.csv') == md5sum('../expected/4.plan/plan2agent2group.csv'))
+  
 })
   
\ No newline at end of file