Reference Dates

1library(yamaa)23yamaa_spec("spec.yaml")$run()$output

1import yamaa23yamaa.spec("spec.yaml").run().output

1# Reference solution for the yamaa benchmark sdtm-dm-dates (R track).2#3# Written from the benchmark's full prompt and its inputs alone, as an agent4# would write it. Harbor's oracle agent runs it to check that the prompt can5# be solved and that the grader scores a correct answer 1.67library(dplyr, warn.conflicts = FALSE)8library(readr)910dm_raw <- read_csv(11  "/app/input/dm_raw.csv",12  col_types = cols(.default = col_character())13)14ex <- read_csv(15  "/app/input/ex.csv",16  col_types = cols(.default = col_character())17)18ds <- read_csv(19  "/app/input/ds.csv",20  col_types = cols(.default = col_character())21)22ae <- read_csv(23  "/app/input/ae.csv",24  col_types = cols(.default = col_character())25)2627ex_sum <- ex |>28  mutate(29    EXSTDTC = suppressWarnings(as.Date(EXSTDTC)),30    EXENDTC = suppressWarnings(as.Date(EXENDTC))31  ) |>32  group_by(STUDYID, USUBJID) |>33  summarise(34    RFXSTDTC = suppressWarnings(min(EXSTDTC, na.rm = TRUE)),35    RFXENDTC = suppressWarnings(max(EXENDTC, na.rm = TRUE)),36    .groups = "drop"37  ) |>38  mutate(39    RFXSTDTC = as.Date(ifelse(is.infinite(as.numeric(RFXSTDTC)), NA, RFXSTDTC), origin = "1970-01-01"),40    RFXENDTC = as.Date(ifelse(is.infinite(as.numeric(RFXENDTC)), NA, RFXENDTC), origin = "1970-01-01")41  )4243ds_sum <- ds |>44  filter(DSCAT == "DISPOSITION EVENT") |>45  mutate(DSSTDTC = suppressWarnings(as.Date(DSSTDTC))) |>46  group_by(STUDYID, USUBJID) |>47  summarise(48    DS_MAX = suppressWarnings(max(DSSTDTC, na.rm = TRUE)),49    .groups = "drop"50  ) |>51  mutate(52    DS_MAX = as.Date(ifelse(is.infinite(as.numeric(DS_MAX)), NA, DS_MAX), origin = "1970-01-01")53  )5455ae_sum <- ae |>56  mutate(AEENDTC = suppressWarnings(as.Date(AEENDTC))) |>57  group_by(STUDYID, USUBJID) |>58  summarise(59    AE_MAX = suppressWarnings(max(AEENDTC, na.rm = TRUE)),60    .groups = "drop"61  ) |>62  mutate(63    AE_MAX = as.Date(ifelse(is.infinite(as.numeric(AE_MAX)), NA, AE_MAX), origin = "1970-01-01")64  )6566dm <- dm_raw |>67  left_join(ex_sum, by = c("STUDYID", "USUBJID")) |>68  left_join(ds_sum, by = c("STUDYID", "USUBJID")) |>69  left_join(ae_sum, by = c("STUDYID", "USUBJID")) |>70  mutate(71    DOMAIN = "DM",72    RFSTDTC = RFXSTDTC,73    RFPENDTC = pmax(RFXENDTC, DS_MAX, AE_MAX, na.rm = TRUE),74    RFPENDTC = as.Date(ifelse(is.infinite(as.numeric(RFPENDTC)), NA, RFPENDTC), origin = "1970-01-01"),75    RFENDTC = ifelse(!is.na(RFSTDTC), as.character(RFPENDTC), NA),76    RFENDTC = suppressWarnings(as.Date(RFENDTC, origin = "1970-01-01"))77  ) |>78  select(DOMAIN, STUDYID, USUBJID, RFICDTC, RFXSTDTC, RFXENDTC, RFSTDTC, RFPENDTC, RFENDTC)7980dir.create("/app/output", showWarnings = FALSE)81write_csv(dm, "/app/output/dm.csv", na = "")

1# Reference solution for the yamaa benchmark sdtm-dm-dates (Python track).2#3# Written from the benchmark's full prompt and its inputs alone, as an agent4# would write it. Harbor's oracle agent runs it to check that the prompt can5# be solved and that the grader scores a correct answer 1.67"""Build the DM dataset (one record per enrolled subject) from DM_RAW,8EX, DS, and AE source datasets, following CDISC SDTM conventions.910Reads /app/input/dm_raw.csv, ex.csv, ds.csv, ae.csv and writes11/app/output/dm.csv with columns:12DOMAIN, STUDYID, USUBJID, RFICDTC, RFXSTDTC, RFXENDTC, RFSTDTC,13RFPENDTC, RFENDTC.14"""1516import csv17import os18from datetime import date1920INPUT_DIR = "/app/input"21OUTPUT_DIR = "/app/output"222324def read_csv(name):25    path = os.path.join(INPUT_DIR, name)26    with open(path, newline="", encoding="utf-8") as f:27        return list(csv.DictReader(f))282930def clean(value):31    """Return the stripped string, or None when the value is missing."""32    if value is None:33        return None34    value = value.strip()35    return value if value else None363738def parse_date(value):39    """Parse an ISO date string to a date for comparison; None if absent."""40    value = clean(value)41    if value is None:42        return None43    try:44        return date.fromisoformat(value)45    except ValueError:46        return None474849def main():50    dm_raw = read_csv("dm_raw.csv")51    ex = read_csv("ex.csv")52    ds = read_csv("ds.csv")53    ae = read_csv("ae.csv")5455    def key(row):56        return (clean(row.get("STUDYID")), clean(row.get("USUBJID")))5758    # Per-subject exposure dates.59    ex_starts = {}60    ex_ends = {}61    for row in ex:62        k = key(row)63        start = parse_date(row.get("EXSTDTC"))64        if start is not None:65            ex_starts.setdefault(k, []).append((start, clean(row.get("EXSTDTC"))))66        end = parse_date(row.get("EXENDTC"))67        if end is not None:68            ex_ends.setdefault(k, []).append((end, clean(row.get("EXENDTC"))))6970    # Per-subject latest disposition event date (category DISPOSITION71    # EVENT; the date is the record's start date).72    ds_event_dates = {}73    for row in ds:74        if clean(row.get("DSCAT")) != "DISPOSITION EVENT":75            continue76        d = parse_date(row.get("DSSTDTC"))77        if d is not None:78            ds_event_dates.setdefault(key(row), []).append(79                (d, clean(row.get("DSSTDTC")))80            )8182    # Per-subject latest adverse event end date.83    ae_end_dates = {}84    for row in ae:85        d = parse_date(row.get("AEENDTC"))86        if d is not None:87            ae_end_dates.setdefault(key(row), []).append(88                (d, clean(row.get("AEENDTC")))89            )9091    def earliest(pairs):92        return min(pairs)[1] if pairs else ""9394    def latest(pairs):95        return max(pairs)[1] if pairs else ""9697    out_rows = []98    for row in dm_raw:99        k = key(row)100        rfxstdtc = earliest(ex_starts.get(k, []))101        rfxendtc = latest(ex_ends.get(k, []))102        rfstdtc = rfxstdtc103104        candidates = []105        if ex_ends.get(k):106            candidates.append(max(ex_ends[k]))107        if ds_event_dates.get(k):108            candidates.append(max(ds_event_dates[k]))109        if ae_end_dates.get(k):110            candidates.append(max(ae_end_dates[k]))111        rfpendtc = max(candidates)[1] if candidates else ""112113        # The reference end date exists only for a subject who entered114        # the reference period (RFSTDTC present).115        rfendtc = rfpendtc if rfstdtc else ""116117        out_rows.append(118            {119                "DOMAIN": "DM",120                "STUDYID": clean(row.get("STUDYID")) or "",121                "USUBJID": clean(row.get("USUBJID")) or "",122                "RFICDTC": clean(row.get("RFICDTC")) or "",123                "RFXSTDTC": rfxstdtc,124                "RFXENDTC": rfxendtc,125                "RFSTDTC": rfstdtc,126                "RFPENDTC": rfpendtc,127                "RFENDTC": rfendtc,128            }129        )130131    fieldnames = [132        "DOMAIN",133        "STUDYID",134        "USUBJID",135        "RFICDTC",136        "RFXSTDTC",137        "RFXENDTC",138        "RFSTDTC",139        "RFPENDTC",140        "RFENDTC",141    ]142    os.makedirs(OUTPUT_DIR, exist_ok=True)143    out_path = os.path.join(OUTPUT_DIR, "dm.csv")144    with open(out_path, "w", newline="", encoding="utf-8") as f:145        writer = csv.DictWriter(f, fieldnames=fieldnames)146        writer.writeheader()147        writer.writerows(out_rows)148149150if __name__ == "__main__":151    main()

1/* opensas reference solution for the yamaa benchmark sdtm-dm-dates.2   Reads the benchmark inputs and derives the requested output.3   Execute with the existing opensas runtime: opensas /app/output/result.sas. */45data ae;6  length STUDYID USUBJID AETERM AESTDTC AEENDTC $1024;7  infile "/app/input/ae.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;8  input STUDYID : $char1024. USUBJID : $char1024. AESEQ AETERM : $char1024. AESTDTC : $char1024. AEENDTC : $char1024.;9run;1011data dm_raw;12  length STUDYID USUBJID RFICDTC $1024;13  infile "/app/input/dm_raw.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;14  input STUDYID : $char1024. USUBJID : $char1024. RFICDTC : $char1024.;15run;1617data ds;18  length STUDYID USUBJID DSCAT DSDECOD DSSTDTC $1024;19  infile "/app/input/ds.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;20  input STUDYID : $char1024. USUBJID : $char1024. DSSEQ DSCAT : $char1024. DSDECOD : $char1024. DSSTDTC : $char1024.;21run;2223data ex;24  length STUDYID USUBJID EXTRT EXSTDTC EXENDTC $1024;25  infile "/app/input/ex.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;26  input STUDYID : $char1024. USUBJID : $char1024. EXSEQ EXTRT : $char1024. EXDOSE EXSTDTC : $char1024. EXENDTC : $char1024.;27run;2829proc sql;30create table starts as select STUDYID,USUBJID,min(EXSTDTC) as RFXSTDTC from ex where not missing(EXSTDTC) group by STUDYID,USUBJID;31create table ends as select STUDYID,USUBJID,max(EXENDTC) as RFXENDTC from ex group by STUDYID,USUBJID;32create table alldates as select STUDYID,USUBJID,EXENDTC as DTC from ex union all select STUDYID,USUBJID,DSSTDTC as DTC from ds where DSCAT='DISPOSITION EVENT' union all select STUDYID,USUBJID,AEENDTC as DTC from ae;33create table lastdate as select STUDYID,USUBJID,max(DTC) as RFPENDTC from alldates group by STUDYID,USUBJID;34create table joined as select a.*,b.RFXSTDTC,c.RFXENDTC,d.RFPENDTC from dm_raw a left join starts b on a.STUDYID=b.STUDYID and a.USUBJID=b.USUBJID left join ends c on a.STUDYID=c.STUDYID and a.USUBJID=c.USUBJID left join lastdate d on a.STUDYID=d.STUDYID and a.USUBJID=d.USUBJID; quit;35data result; set joined; length DOMAIN $2 RFSTDTC RFENDTC $1024; DOMAIN='DM'; RFSTDTC=RFXSTDTC; if not missing(RFSTDTC) then RFENDTC=RFPENDTC; run;36proc sql;37  create table final as select DOMAIN, STUDYID, USUBJID, RFICDTC, RFXSTDTC, RFXENDTC, RFSTDTC, RFPENDTC, RFENDTC from result;38quit;39proc export data=final outfile="/app/output/dm.csv" dbms=csv replace; run;