This commit is contained in:
2026-06-17 15:05:10 +02:00
parent de959d849d
commit 4884117227
85 changed files with 34611 additions and 0 deletions
+534
View File
@@ -0,0 +1,534 @@
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
sipiq_import_v1.0.py
====================
Verze: 1.0
Datum: 2026-06-17
Autor: Claude Code (pro MUDr. Vladimíra Buzalku)
Popis
-----
Import SIPIQ odpovědí (Qualtrics CSV export, studie 77242113UCO3002 / ICONIC DAWN)
do MongoDB db `feasibility`. Cílem je:
(a) umožnit křížovou analýzu „otázka × otázka" (ploché odpovědi keyed by Qcode),
(b) umožnit zrekonstruovat KOMPLETNÍ SIPIQ tak, jak ho zkoušející vidí v PDF,
jen s vyplněnými odpověďmi (slovník otázek se sekcí/pořadím/popisky).
Dvě kolekce v db `feasibility`:
* sipiq_questions slovník dotazníku (1 dok = 1 logická otázka; section, order,
text, items[], type, options). Idempotentní (upsert dle _id).
* sipiq_responses 1 dok = 1 odpověď (_id = Qualtrics ResponseId). Identita centra/PI
nahoře, ploché answers{}, meta{}, soft-link investigator_oid,
delta bookkeeping (content_sha256, history[], timestamps).
DELTA import (přepíše JEN změněná data):
- nová odpověď -> insert
- existuje, beze změn -> aktualizuje pouze last_seen_at (+ source_file)
- existuje, něco se změnilo -> $set jen změněných polí + push do history[] {key,old,new}
Soft-link na feasibility.investigators:
- primárně pi_email == email / email2 (lowercase)
- fallback příjmení (bez diakritiky, lower) + země (CZ/SK)
- nedestruktivní: kolekci investigators NEMĚNÍ, jen ukládá investigator_oid do response.
Rozsah: default CZ + SK (--scope czsk). --scope all = všech 276.
Použití:
python sipiq_import_v1.0.py --csv "<cesta.csv>" --dry-run
python sipiq_import_v1.0.py --csv "<cesta.csv>" --apply
Závislosti: pymongo (.venv). Mongo 192.168.1.76:27017, bez auth.
"""
import argparse
import csv
import hashlib
import json
import re
import sys
import unicodedata
from datetime import datetime, timezone
try:
from pymongo import MongoClient
except ImportError:
print("CHYBA: pymongo není nainstalován v aktuálním pythonu.", file=sys.stderr)
raise
MONGO_URI = "mongodb://192.168.1.76:27017"
DB_NAME = "feasibility"
COL_Q = "sipiq_questions"
COL_R = "sipiq_responses"
# Qualtrics systémová meta pole (NEjdou do answers)
META_COLS = {
"StartDate", "EndDate", "Status", "IPAddress", "Progress", "Duration (in seconds)",
"Finished", "RecordedDate", "ResponseId", "RecipientLastName", "RecipientFirstName",
"RecipientEmail", "ExternalReference", "LocationLatitude", "LocationLongitude",
"DistributionChannel", "UserLanguage",
}
# Embedded SDL pole povýšená nahoru do dokumentu (queryable identita)
PROMOTE = [
"site_name", "site_address", "site_city", "site_state", "site_postcode", "site_country",
"pi_first_name", "pi_last_name", "pi_phone", "pi_email",
"sdl_site_id", "fire_site_id", "fire_investigator_id", "mailinglist_id",
"survey_generated_by", "Date", "Time",
]
# Sekce dle ověřeného katalogu (mapování báze Q-čísla -> sekce v PDF)
SECTION_BY_QNUM = {}
def _sec(rng, name):
for n in rng:
SECTION_BY_QNUM[n] = name
_sec([2], "J&J Internal Assessment")
_sec([6, 7, 8, 9, 10, 11, 12, 13], "Contact Information")
_sec(range(14, 22), "Confidentiality Statement")
_sec([25, 26, 27], "Interest")
_sec([29, 30, 31, 32, 33, 34], "Protocol Requirements")
_sec([36, 37, 38], "Enrollment")
_sec([40, 41, 42, 43], "Patient Demographics Overview")
_sec([45, 46, 47, 48, 49], "Site Overview")
_sec([51], "Operational Considerations")
_sec([53, 54], "Comments")
_sec([57, 58, 59, 60, 61], "Patient Population")
_sec([63, 64, 65, 66, 67], "Site Experience and Staffing")
_sec([69], "Equipment and Facility Requirements")
_sec([71, 72, 73, 74, 75], "Institutional Review Board, Ethics Committee, and Contracts")
# Plné znění otázek, které Qualtrics v hlavičce CSV ořezává "..." (maticové otázky).
# Zdroj: prázdný SIPIQ PDF (ICONIC ... _SipIQ_V1_13MAY2026.pdf).
STEM_OVERRIDE = {
"Q31": "At your site, at what line(s) of treatment do you most commonly prescribe "
"vedolizumab for patients with moderately to severely active ulcerative colitis?",
"Q63": "Do you or your site staff have experience in performing the following types of "
"study assessments/procedures?",
"Q64": "The following personnel are required to run the study. "
"Will your site have the following available?",
"Q69": "The following equipment and facilities are required to run the studies. "
"Are these available at your site?",
}
def now_iso():
return datetime.now(timezone.utc).astimezone().isoformat(timespec="seconds")
def strip_accents(s):
if not s:
return ""
nfkd = unicodedata.normalize("NFKD", s)
return "".join(c for c in nfkd if not unicodedata.combining(c))
def norm_name(s):
return re.sub(r"\s+", " ", strip_accents(s or "").lower()).strip()
def sanitize_key(qcode):
"""Qcode -> klíč do answers{} (MongoDB-safe): '#' a '.' -> '_'."""
return qcode.replace("#", "_").replace(".", "_")
def qnum(qcode):
"""Číslo otázky z Qcode (Q63#1_2 -> 63, Q40_6_TEXT -> 40)."""
m = re.match(r"Q(\d+)", qcode)
return int(m.group(1)) if m else None
def qbase(qcode):
"""Logická báze otázky (Q63#1_2 -> Q63, Q40_6 -> Q40, Q25 -> Q25)."""
m = re.match(r"(Q\d+)", qcode)
return m.group(1) if m else qcode
def import_id(h3_cell):
try:
return json.loads(h3_cell).get("ImportId", "")
except Exception:
return h3_cell
def split_text(text):
"""Vrátí (stem, item_label). Stem = text otázky, item_label = popisek podčásti."""
parts = [p.strip() for p in re.split(r"\s+-\s+", text)]
stem = parts[0]
if len(parts) == 1:
return stem, None
# poslední část = popisek řádku/části; vyčisti Qualtrics artefakty
label_parts = parts[1:]
# zahodit "Selected Choice" (artefakt single-choice s Other)
label_parts = [p for p in label_parts if p.lower() != "selected choice"]
# zahodit interní statement kód typu "Q63#1"
label_parts = [p for p in label_parts if not re.fullmatch(r"Q\d+#\d+", p)]
label = " - ".join(label_parts) if label_parts else None
return stem, label
def detect_type(qcode, observed):
"""Heuristika typu otázky z Qcode a pozorovaných hodnot."""
has_hash = "#" in qcode
vals = [v for v in observed if v]
yesno = vals and all(v in ("Yes", "No") for v in vals)
numeric = vals and all(re.fullmatch(r"-?\d+(\.\d+)?", v) for v in vals)
if has_hash and yesno:
return "matrix_yesno"
if has_hash and numeric:
return "matrix_percent"
if has_hash:
return "matrix"
if numeric:
return "numeric"
if yesno:
return "yesno"
return "single_or_text"
# ---------------------------------------------------------------------------
def load_csv(path):
with open(path, encoding="utf-8-sig", newline="") as fh:
rows = list(csv.reader(fh))
h1, h2, h3 = rows[0], rows[1], rows[2]
data = rows[3:]
cols = []
for i, (code, text, imp) in enumerate(zip(h1, h2, h3)):
cols.append({"i": i, "code": code, "text": text, "qid": import_id(imp)})
return cols, data
def col_getter(cols, data):
idx = {c["code"]: c["i"] for c in cols}
def get(row, code):
i = idx.get(code)
return (row[i].strip() if i is not None and i < len(row) else "")
return get, idx
def is_question_col(code):
return bool(re.match(r"Q\d", code))
# ---------------------------------------------------------------------------
def build_questions(cols, data):
"""Slovník otázek -> list dokumentů (1 = 1 logická otázka)."""
# observed hodnoty per Qcode (pro typ + options)
qcols = [c for c in cols if is_question_col(c["code"])]
observed = {c["code"]: set() for c in qcols}
for row in data:
for c in qcols:
v = (row[c["i"]].strip() if c["i"] < len(row) else "")
if v:
observed[c["code"]].add(v)
groups = {} # base -> dict
order_seen = []
for c in qcols:
base = qbase(c["code"])
if base not in groups:
groups[base] = {
"_id": base,
"order": c["i"],
"qnum": qnum(c["code"]),
"section": SECTION_BY_QNUM.get(qnum(c["code"]), "Other"),
"qids": [],
"text": split_text(c["text"])[0],
"items": [],
"_obs": set(),
"_types": [],
}
order_seen.append(base)
g = groups[base]
base_qid = re.match(r"(QID\d+)", c["qid"] or "")
if base_qid and base_qid.group(1) not in g["qids"]:
g["qids"].append(base_qid.group(1))
stem, label = split_text(c["text"])
key = sanitize_key(c["code"])
item = {"key": key, "qcode": c["code"], "qid": c["qid"]}
if label:
item["label"] = label
g["items"].append(item)
g["_obs"] |= observed[c["code"]]
g["_types"].append(detect_type(c["code"], observed[c["code"]]))
out = []
for n, base in enumerate(order_seen):
g = groups[base]
obs = sorted(g.pop("_obs"))
types = g.pop("_types")
# typ skupiny: nejčastější netriviální
gtype = max(set(types), key=types.count) if types else "single_or_text"
g["type"] = gtype
# options jen u kategorických (yesno/single)
if gtype in ("yesno", "matrix_yesno"):
g["options"] = ["Yes", "No"]
elif gtype == "single_or_text" and obs and len(obs) <= 12:
g["options"] = obs
else:
g["options"] = []
if base in STEM_OVERRIDE:
g["text"] = STEM_OVERRIDE[base]
g["order"] = n # přečíslovat 0..N dle pořadí v CSV
# pokud má jen 1 item bez labelu, items vynech (je to prostá otázka)
if len(g["items"]) == 1 and "label" not in g["items"][0]:
g["items"] = []
out.append(g)
return out
# ---------------------------------------------------------------------------
def build_response(cols, get, row, source_file):
rid = get(row, "ResponseId")
answers = {}
for c in cols:
if is_question_col(c["code"]):
v = (row[c["i"]].strip() if c["i"] < len(row) else "")
if v:
answers[sanitize_key(c["code"])] = v
def g(*names):
for nm in names:
v = get(row, nm)
if v:
return v
return None
meta = {
"start_date": get(row, "StartDate") or None,
"end_date": get(row, "EndDate") or None,
"recorded_date": get(row, "RecordedDate") or None,
"status": get(row, "Status") or None,
"progress": int(get(row, "Progress")) if get(row, "Progress").isdigit() else get(row, "Progress") or None,
"finished": get(row, "Finished") in ("True", "1", "TRUE"),
"duration_sec": int(get(row, "Duration (in seconds)")) if get(row, "Duration (in seconds)").isdigit() else None,
"user_language": get(row, "UserLanguage") or None,
"distribution_channel": get(row, "DistributionChannel") or None,
"ip_address": get(row, "IPAddress") or None,
"location_lat": get(row, "LocationLatitude") or None,
"location_lng": get(row, "LocationLongitude") or None,
"survey_date": get(row, "Date") or None,
"survey_time": get(row, "Time") or None,
}
doc = {
"_id": rid,
"study": "77242113UCO3002",
"site_country": get(row, "site_country") or None,
"site_name": get(row, "site_name") or None,
"site_city": get(row, "site_city") or None,
"site_state": get(row, "site_state") or None,
"site_postcode": get(row, "site_postcode") or None,
"site_address": get(row, "site_address") or None,
"pi_first_name": get(row, "pi_first_name") or None,
"pi_last_name": get(row, "pi_last_name") or None,
"pi_email": (get(row, "pi_email") or "").lower() or None,
"pi_phone": get(row, "pi_phone") or None,
"sdl_site_id": get(row, "sdl_site_id") or None,
"fire_site_id": get(row, "fire_site_id") or None,
"fire_investigator_id": get(row, "fire_investigator_id") or None,
"mailinglist_id": get(row, "mailinglist_id") or None,
"survey_generated_by": get(row, "survey_generated_by") or None,
"recipient_email": (get(row, "RecipientEmail") or "").lower() or None,
"recipient_last_name": get(row, "RecipientLastName") or None,
"recipient_first_name": get(row, "RecipientFirstName") or None,
"meta": meta,
"is_full_sipiq": any(k.startswith(("Q57", "Q58", "Q59", "Q63", "Q66", "Q71")) for k in answers),
"interested": answers.get("Q25"),
"answers": answers,
"investigator_oid": None,
"investigator_match": None,
"source_file": source_file,
}
return doc
def content_hash(doc):
payload = {k: doc[k] for k in doc if k not in
("content_sha256", "first_imported_at", "last_seen_at", "last_updated_at", "history",
"investigator_oid", "investigator_match", "source_file")}
blob = json.dumps(payload, sort_keys=True, ensure_ascii=False, default=str)
return hashlib.sha256(blob.encode("utf-8")).hexdigest()
# ---------------------------------------------------------------------------
def load_investigators(db):
inv = list(db.investigators.find(
{"zeme": {"$in": ["Czech Republic", "Slovakia"]}},
{"prijmeni": 1, "jmeno": 1, "email": 1, "email2": 1, "zeme": 1, "KROK": 1, "pracoviste": 1},
))
by_email = {}
by_name = {}
for d in inv:
for ef in ("email", "email2"):
e = (d.get(ef) or "").lower().strip()
if e:
by_email.setdefault(e, d)
nm = norm_name(d.get("prijmeni"))
if nm:
by_name.setdefault((nm, d.get("zeme")), []).append(d)
return inv, by_email, by_name
def soft_link(doc, by_email, by_name):
e = (doc.get("pi_email") or "").lower().strip()
if e and e in by_email:
d = by_email[e]
return d["_id"], f"email:{e}", d
e2 = (doc.get("recipient_email") or "").lower().strip()
if e2 and e2 in by_email:
d = by_email[e2]
return d["_id"], f"recipient_email:{e2}", d
nm = norm_name(doc.get("pi_last_name"))
cand = by_name.get((nm, doc.get("site_country")), [])
if len(cand) == 1:
return cand[0]["_id"], f"prijmeni:{nm}", cand[0]
if len(cand) > 1:
return None, f"prijmeni_ambiguous:{nm}({len(cand)})", None
return None, "NENALEZENO", None
# ---------------------------------------------------------------------------
def main():
ap = argparse.ArgumentParser()
ap.add_argument("--csv", required=True)
ap.add_argument("--scope", choices=["czsk", "all"], default="czsk")
ap.add_argument("--apply", action="store_true", help="ostrý zápis (jinak dry-run)")
ap.add_argument("--dry-run", action="store_true")
args = ap.parse_args()
dry = not args.apply
source_file = args.csv.replace("\\", "/").split("/")[-1]
cols, data = load_csv(args.csv)
get, idx = col_getter(cols, data)
# filtr rozsahu
if args.scope == "czsk":
data = [r for r in data if get(r, "site_country") in ("Czech Republic", "Slovakia")]
print(f"Zdroj: {source_file} | rozsah={args.scope} | odpovědí k importu: {len(data)}")
# --- slovník otázek (staví se z PLNÉHO CSV, ne jen scope) ---
cols_all, data_all = load_csv(args.csv)
questions = build_questions(cols_all, data_all)
print(f"Slovník otázek: {len(questions)} logických otázek "
f"(z toho {sum(1 for q in questions if q['items'])} vícedílných).")
# --- Mongo ---
client = MongoClient(MONGO_URI, serverSelectionTimeoutMS=8000)
db = client[DB_NAME]
client.admin.command("ping")
inv, by_email, by_name = load_investigators(db)
print(f"Investigatorů CZ+SK v DB: {len(inv)}")
# --- response dokumenty + soft-link ---
docs = []
link_rows = []
for r in data:
doc = build_response(cols, get, r, source_file)
oid, how, matched = soft_link(doc, by_email, by_name)
doc["investigator_oid"] = oid
doc["investigator_match"] = how
doc["content_sha256"] = content_hash(doc)
docs.append(doc)
link_rows.append((doc, how, matched))
# --- delta proti DB ---
existing = {d["_id"]: d for d in db[COL_R].find({}, {"content_sha256": 1})}
to_insert = [d for d in docs if d["_id"] not in existing]
to_update, unchanged = [], []
for d in docs:
if d["_id"] in existing:
if existing[d["_id"]].get("content_sha256") != d["content_sha256"]:
to_update.append(d)
else:
unchanged.append(d)
# ===================== REPORT =====================
print("\n=== SOFT-LINK na investigators ===")
matched_k7 = matched_other = unmatched = 0
for doc, how, m in link_rows:
krok = (m or {}).get("KROK", "")
tag = "" if m else ""
if m and str(krok).startswith("7"):
matched_k7 += 1
elif m:
matched_other += 1
else:
unmatched += 1
print(f" {tag} {doc.get('site_country','?')[:2]} {str(doc.get('pi_last_name'))[:18]:18} "
f"{str(doc.get('pi_email'))[:32]:32} -> {how[:40]:40} {('KROK '+str(krok)) if m else ''}")
print(f" Souhrn: napárováno KROK7={matched_k7}, jiný KROK={matched_other}, nenapárováno={unmatched}")
print("\n=== DELTA ===")
print(f" INSERT (nové): {len(to_insert)}")
print(f" UPDATE (změněné): {len(to_update)}")
print(f" beze změny: {len(unchanged)}")
# ukázka 1 dokumentu
if docs:
s = dict(docs[0])
s["answers"] = {k: s["answers"][k] for k in list(s["answers"])[:6]}
s["answers"][""] = f"(+{len(docs[0]['answers'])-6} dalších)"
print("\n=== UKÁZKA response dokumentu (zkráceno) ===")
print(json.dumps(s, ensure_ascii=False, indent=2, default=str)[:1800])
if dry:
print("\n[DRY-RUN] Nic se nezapsalo. Ostrý běh: přidej --apply")
client.close()
return
# ===================== ZÁPIS =====================
# 1) slovník otázek (idempotentní upsert)
nq = 0
for q in questions:
db[COL_Q].replace_one({"_id": q["_id"]}, q, upsert=True)
nq += 1
print(f"\n[APPLY] sipiq_questions: upsertnuto {nq}")
# 2) responses (delta)
ts = now_iso()
ni = nu = ns = 0
for d in docs:
cur = db[COL_R].find_one({"_id": d["_id"]})
if cur is None:
d["first_imported_at"] = ts
d["last_seen_at"] = ts
d["last_updated_at"] = ts
d["history"] = []
db[COL_R].insert_one(d)
ni += 1
elif cur.get("content_sha256") != d["content_sha256"]:
changes = diff_docs(cur, d)
db[COL_R].update_one({"_id": d["_id"]}, {
"$set": {**{k: d[k] for k in d if k not in ("_id",)},
"last_seen_at": ts, "last_updated_at": ts},
"$push": {"history": {"changed_at": ts, "source_file": source_file, "changes": changes}},
})
nu += 1
else:
db[COL_R].update_one({"_id": d["_id"]},
{"$set": {"last_seen_at": ts, "source_file": source_file}})
ns += 1
print(f"[APPLY] sipiq_responses: insert={ni}, update={nu}, beze změny={ns}")
client.close()
def diff_docs(old, new):
"""Field-level diff pro history (jen answers + povýšená pole + meta)."""
changes = []
def walk(prefix, o, n):
keys = set((o or {}).keys()) | set((n or {}).keys())
for k in sorted(keys):
ov, nv = (o or {}).get(k), (n or {}).get(k)
if isinstance(ov, dict) or isinstance(nv, dict):
walk(f"{prefix}{k}.", ov or {}, nv or {})
elif ov != nv:
changes.append({"key": f"{prefix}{k}", "old": ov, "new": nv})
for field in ("answers", "meta"):
walk(f"{field}.", old.get(field, {}), new.get(field, {}))
for k in ("site_name", "pi_email", "pi_last_name", "interested", "is_full_sipiq"):
if old.get(k) != new.get(k):
changes.append({"key": k, "old": old.get(k), "new": new.get(k)})
return changes
if __name__ == "__main__":
main()