feat(excel): comparables grid computes adjustments by formula from subject characteristics and unit rates; fix bulk e-mail parsing

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Simon-Pierre Boucher committed 18 days ago (Sep 6, 2026) parent f261ace

5 changed files +301 −152

modified backend/app/api/v1/professor.py +2 −2

		`@@ -180,7 +180,7 @@ class CreateStudentsReq(BaseModel):`
180	180	@router.post("/students", status_code=201)
181	181	async def create_students(req: CreateStudentsReq, _: AuthUser = Depends(require_professor),
182	182	settings: Settings = Depends(get_settings)) -> dict:
183		− raw = req.emails if isinstance(req.emails, list) else re.split(r"[\s,;]+", req.emails)
	183	+ raw = req.emails if isinstance(req.emails, list) else re.split(r"[\n;,]+", req.emails)
184	184	names: dict[str, str] = {}
185	185	emails: list[str] = []
186	186	for item in raw:
		`@@ -192,7 +192,7 @@ async def create_students(req: CreateStudentsReq, _: AuthUser = Depends(require_`
192	192	names[m.group(2).strip().lower()] = m.group(1).strip().strip('"')
193	193	emails.append(m.group(2))
194	194	else:
195		− emails.append(item)
	195	+ emails.extend(t for t in re.split(r"\s+", item) if t)
196	196	if req.role not in {"student", "professor"}:
197	197	raise HTTPException(status.HTTP_400_BAD_REQUEST, detail="Rôle invalide.")
198	198	return await users.create_users(emails[:500], settings, req.role, names)

modified backend/app/tools/excel_templates/comparables_ajustes.py +254 −148

		`@@ -1,164 +1,270 @@`
1		−"""Template: adjusted comparables grid (sales comparison) — flexible columns.
2		−
3		−Accepted `params`:
4		− sujet: str
5		− ajustements: [str] # adjustment names (optional; inferred from comparables)
6		− comparables: [ { adresse\|nom, prix, date\|date_vente, temps\|ajustement_temps_pct,
7		− ajustements: {name: montant $ \| "+3 %"}, # or adjustment keys at top level
8		− caracteristiques: {name: texte}, # e.g. garage: "Oui"
9		− ... } ]
10		−Text values in adjustment columns ("Oui", "meilleur") are moved to the characteristics table.
	1	+"""Template: adjusted comparables grid (sales comparison) with live formulas.
	2	+
	3	+Two ways to feed it (both tolerant to strings like "425 000 $", "+3 %", "Oui", "bon") :
	4	+
	5	+A) Characteristics + unit rates (preferred — adjustments are computed by Excel formulas):
	6	+ sujet: {adresse, superficie: 1200, terrain: 6000, garage: "Oui", etat: "bon", ...}
	7	+ taux: {superficie: 120, terrain: 8, garage: 15000, etat: 10000} # $ per unit / per step
	8	+ echelles: {etat: ["mauvais", "moyen", "bon", "très bon"]} # ordinal scales (optional)
	9	+ taux_temps_mensuel: 0.005 # market trend (optional)
	10	+ comparables: [{adresse, prix, date, mois: 3, superficie: 1150, terrain: 5500, garage: "Oui",
	11	+ etat: "bon"}, ...]
	12	+ → ajustement = (sujet − comparable) × taux, in the workbook, cell by cell.
	13	+
	14	+B) Direct dollar adjustments (no `taux`): comparables: [{adresse, prix, temps: "+2 %",
	15	+ ajustements: {Superficie: -5000, Garage: -12000}, caracteristiques: {Garage: "Oui"}}]
11	16	"""
12	17
13	18	from __future__ import annotations
14	19
	20	+import unicodedata
15	21	from typing import Any
16	22
17		−from openpyxl.utils import get_column_letter
18		−
19		−from app.tools.coerce import cell, num, pct
20		−
21		−META_KEYS = {"adresse", "nom", "comparable", "prix", "prix_vente", "date", "date_vente",
22		− "temps", "ajustement_temps_pct", "date_pct", "ajust_temps", "ajustements",
23		− "caracteristiques", "notes", "note", "source"}
24		−
25		−DEFAULT_COMPS = [
26		− {"adresse": "Comparable 1", "prix": 415000, "temps": "2 %",
27		− "ajustements": {"Superficie": -5000, "Garage": -12000},
28		− "caracteristiques": {"Garage": "Oui", "État": "Bon"}},
29		− {"adresse": "Comparable 2", "prix": 439000, "temps": "1 %",
30		− "ajustements": {"Superficie": 8000, "Terrain": -6000, "État": -10000},
31		− "caracteristiques": {"Garage": "Non", "État": "Très bon"}},
32		− {"adresse": "Comparable 3", "prix": 402000, "temps": "3 %",
33		− "ajustements": {"Terrain": 4000, "État": 15000},
34		− "caracteristiques": {"Garage": "Non", "État": "Moyen"}},
35		−]
36		−
37		−
38		−def _normalise(comps: list[Any]) -> tuple[list[dict[str, Any]], list[str], list[str]]:
39		− out: list[dict[str, Any]] = []
40		− adj_names: list[str] = []
41		− char_names: list[str] = []
42		−
43		− def add(names: list[str], k: str) -> None:
44		− if k not in names:
45		− names.append(k)
46		−
47		− for i, raw in enumerate(comps):
48		− c = dict(raw) if isinstance(raw, dict) else {"adresse": str(raw)}
49		− item: dict[str, Any] = {
50		− "nom": str(c.get("adresse") or c.get("nom") or c.get("comparable") or f"Comparable {i + 1}"),
51		− "prix": num(c.get("prix", c.get("prix_vente")), 0.0) or 0.0,
52		− "date": str(c.get("date") or c.get("date_vente") or ""),
53		− "temps": pct(c.get("temps", c.get("ajustement_temps_pct", c.get("date_pct",
54		− c.get("ajust_temps")))), 0.0) or 0.0,
55		− "adj": {}, "chars": {}, "note": str(c.get("notes") or c.get("note") or ""),
56		− }
57		− candidates: dict[str, Any] = {}
58		− if isinstance(c.get("ajustements"), dict):
59		− candidates.update(c["ajustements"])
	23	+from openpyxl.utils import get_column_letter as L
	24	+
	25	+from app.tools.coerce import boolish, cell, num, pct
	26	+
	27	+META = {"adresse", "nom", "comparable", "prix", "prix_vente", "date", "date_vente", "temps",
	28	+ "ajustement_temps_pct", "date_pct", "ajust_temps", "mois", "ajustements",
	29	+ "caracteristiques", "notes", "note", "source", "filename"}
	30	+QUALITY_WORDS = ["très mauvais", "mauvais", "passable", "moyen", "bon", "très bon", "excellent", "neuf"]
	31	+
	32	+DEFAULT = {
	33	+ "sujet": {"adresse": "Sujet — bungalow fictif, Gatineau", "superficie": 1200, "terrain": 6000,
	34	+ "garage": "Oui", "etat": "bon"},
	35	+ "taux": {"superficie": 120, "terrain": 8, "garage": 15000, "etat": 10000},
	36	+ "taux_temps_mensuel": 0.005,
	37	+ "comparables": [
	38	+ {"adresse": "45 rue Laurier", "prix": 425000, "date": "2026-03", "mois": 6, "superficie": 1150,
	39	+ "terrain": 5500, "garage": "Oui", "etat": "bon"},
	40	+ {"adresse": "12 rue Front", "prix": 398000, "date": "2026-01", "mois": 8, "superficie": 1250,
	41	+ "terrain": 7000, "garage": "Non", "etat": "moyen"},
	42	+ {"adresse": "88 boul. Saint-Joseph", "prix": 449000, "date": "2026-04", "mois": 5,
	43	+ "superficie": 1300, "terrain": 6200, "garage": "Oui", "etat": "très bon"},
	44	+ ],
	45	+}
	46	+
	47	+
	48	+def _norm(s: str) -> str:
	49	+ s = unicodedata.normalize("NFD", str(s)).encode("ascii", "ignore").decode().lower().strip()
	50	+ return s.replace(" ", "_").replace("-", "_")
	51	+
	52	+
	53	+def _label(key: str) -> str:
	54	+ return key.replace("_", " ").strip().capitalize()
	55	+
	56	+
	57	+def _rank(value: Any, scale: list[str] \| None) -> float \| None:
	58	+ """Ordinal text → rank (0-based). Booleans → 1/0. Numbers pass through."""
	59	+ n = num(value, None)
	60	+ if n is not None and not isinstance(value, bool):
	61	+ return n
	62	+ b = boolish(value)
	63	+ if b is not None and not (isinstance(value, str) and scale and _norm(value) in [_norm(x) for x in scale]):
	64	+ return 1.0 if b else 0.0
	65	+ if isinstance(value, str):
	66	+ words = [_norm(x) for x in (scale or QUALITY_WORDS)]
	67	+ v = _norm(value)
	68	+ if v in words:
	69	+ return float(words.index(v))
	70	+ for i, w in enumerate(words): # "bon état" → "bon"
	71	+ if w and w in v:
	72	+ return float(i)
	73	+ return None
	74	+
	75	+
	76	+def _time_pct(c: dict[str, Any]) -> float \| None:
	77	+ for k in ("temps", "ajustement_temps_pct", "date_pct", "ajust_temps"):
	78	+ if k in c:
	79	+ return pct(c[k], None)
	80	+ return None
	81	+
	82	+
	83	+def build(p: dict[str, Any]) -> dict[str, Any]:
	84	+ if not p.get("comparables"):
	85	+ p = {DEFAULT, {k: v for k, v in p.items() if k != "comparables"}} if p.get("taux") is None else {DEFAULT, p}
	86	+ comps_raw = [c if isinstance(c, dict) else {"adresse": str(c)} for c in (p.get("comparables") or [])]
	87	+ taux_raw = {_norm(k): v for k, v in (p.get("taux") or p.get("unit_rates") or p.get("taux_unitaires") or {}).items()}
	88	+ scales = {_norm(k): list(v) for k, v in (p.get("echelles") or p.get("scales") or {}).items() if isinstance(v, list)}
	89	+ sujet_raw = p.get("sujet") or {}
	90	+ sujet: dict[str, Any] = {_norm(k): v for k, v in sujet_raw.items()} if isinstance(sujet_raw, dict) else {}
	91	+ sujet_label = (sujet_raw.get("adresse") or sujet_raw.get("nom") if isinstance(sujet_raw, dict)
	92	+ else str(sujet_raw or "Sujet — immeuble fictif, Gatineau")) or "Sujet"
	93	+ monthly = pct(p.get("taux_temps_mensuel", p.get("taux_temps_pct_mensuel")), None)
	94	+ mode_rates = bool(taux_raw)
	95	+
	96	+ # ---- characteristics (mode A) or adjustments (mode B)
	97	+ keys: list[str] = []
	98	+ for c in comps_raw:
60	99	for k, v in c.items():
61		− if k not in META_KEYS and not isinstance(v, (dict, list)):
62		− candidates[k] = v
63		− for k, v in candidates.items():
64		− name = str(k).replace("_", " ").strip().capitalize()
65		− n = num(v, None)
66		− if n is None or (isinstance(v, str) and v.strip().lower() in {"oui", "non"}):
67		− if v not in (None, ""):
68		− item["chars"][name] = str(v)
69		− add(char_names, name)
	100	+ nk = _norm(k)
	101	+ if nk in META or isinstance(v, (dict, list)):
70	102	continue
71		− if isinstance(v, str) and v.strip().endswith("%"):
72		− n = n * item["prix"] # percent of price → dollars
73		− item["adj"][name] = n
74		− add(adj_names, name)
75		− if isinstance(c.get("caracteristiques"), dict):
76		− for k, v in c["caracteristiques"].items():
77		− name = str(k).replace("_", " ").strip().capitalize()
78		− item["chars"][name] = str(v)
79		− add(char_names, name)
80		− out.append(item)
81		− return out, adj_names, char_names
	103	+ if nk not in keys:
	104	+ keys.append(nk)
	105	+ for k in (c.get("ajustements") or {}) if isinstance(c.get("ajustements"), dict) else {}:
	106	+ if _norm(k) not in keys:
	107	+ keys.append(_norm(k))
	108	+ for k in taux_raw:
	109	+ if k not in keys and k in sujet:
	110	+ keys.append(k)
	111	+ if mode_rates:
	112	+ keys = [k for k in keys if k in taux_raw] # only characteristics with a rate are adjusted
	113	+ text_keys: list[str] = [] # qualitative columns kept as text (mode B or no rank)
82	114
	115	+ n = len(comps_raw)
	116	+ sheet: dict[str, Any] = {"name": "Comparables", "title": f"Grille de comparables ajustés — {sujet_label}",
	117	+ "inputs": [], "tables": [], "charts": [], "notes": []}
	118	+ row = 4
	119	+ inputs_map: dict[str, str] = {} # key → cell of subject value / rate
	120	+
	121	+ if mode_rates:
	122	+ sheet["inputs_title"] = "Hypothèses (cellules bleues modifiables)"
	123	+ # subject values in column B, rates in column D (labels in A and C)
	124	+ r = row
	125	+ for k in keys:
	126	+ sv = _rank(sujet.get(k), scales.get(k))
	127	+ if sv is None:
	128	+ sv = 0.0
	129	+ sheet["inputs"].append({"cell": f"B{r}", "label": f"Sujet — {_label(k)}"
	130	+ + (f" ({'/'.join(scales[k])})" if k in scales else ""),
	131	+ "value": sv, "format": "number"})
	132	+ sheet["inputs"].append({"cell": f"D{r}", "label": f"Taux — {_label(k)} ($/unité ou $/cran)",
	133	+ "value": num(taux_raw.get(k), 0.0) or 0.0, "format": "currency"})
	134	+ inputs_map[k] = f"$B${r}"
	135	+ inputs_map[k + "__taux"] = f"$D${r}"
	136	+ r += 1
	137	+ if monthly is not None:
	138	+ sheet["inputs"].append({"cell": f"D{r}", "label": "Tendance du marché (%/mois)", "value": monthly,
	139	+ "format": "percent"})
	140	+ inputs_map["__monthly"] = f"$D${r}"
	141	+ r += 1
	142	+ row = r + 2
	143	+
	144	+ # ---- main grid
	145	+ header_row = row
	146	+ first = header_row + 1
	147	+ last = first + n - 1
	148	+ columns: list[dict[str, Any]] = [{"header": "Comparable", "type": "text"}, {"header": "Prix de vente ($)", "type": "currency"},
	149	+ {"header": "Date de vente", "type": "text"}]
	150	+ col = 4 # D
	151	+ if mode_rates and monthly is not None:
	152	+ columns += [{"header": "Mois écoulés", "type": "number"}, {"header": "Ajust. temps (%)", "type": "percent"}]
	153	+ c_mois, c_temps = col, col + 1
	154	+ col += 2
	155	+ else:
	156	+ columns += [{"header": "Ajust. temps (%)", "type": "percent"}]
	157	+ c_mois, c_temps = None, col
	158	+ col += 1
	159	+ columns.append({"header": "Prix ajusté temps ($)", "type": "currency"})
	160	+ c_pat = col
	161	+ col += 1
	162	+ c_char: dict[str, int] = {}
	163	+ c_adj: dict[str, int] = {}
	164	+ if mode_rates:
	165	+ for k in keys:
	166	+ columns.append({"header": f"{_label(k)} (comp.)", "type": "number"})
	167	+ c_char[k] = col
	168	+ col += 1
	169	+ for k in keys:
	170	+ columns.append({"header": f"Ajust. {_label(k)} ($)", "type": "currency"})
	171	+ c_adj[k] = col
	172	+ col += 1
	173	+ else:
	174	+ for k in keys:
	175	+ columns.append({"header": f"{_label(k)} ($)", "type": "currency"})
	176	+ c_adj[k] = col
	177	+ col += 1
	178	+ c_total, c_final, c_gross, c_gross_pct = col, col + 1, col + 2, col + 3
	179	+ columns += [{"header": "Total ajustements ($)", "type": "currency"}, {"header": "Prix ajusté ($)", "type": "currency"},
	180	+ {"header": "Ajust. bruts ($)", "type": "currency"}, {"header": "Ajust. bruts (%)", "type": "percent"}]
83	181
84		−def build(p: dict[str, Any]) -> dict[str, Any]:
85		− comps_raw = p.get("comparables") or DEFAULT_COMPS
86		− comps, adj_names, char_names = _normalise(list(comps_raw))
87		− for extra in p.get("ajustements") or []:
88		− name = str(extra).strip().capitalize()
89		− if name and name not in adj_names:
90		− adj_names.append(name)
91		− sujet = str(p.get("sujet", "Sujet — immeuble fictif, Gatineau"))
92		− n = len(comps)
93		− first_row = 5
94		− last = first_row + n - 1
95		− # columns: A nom, B prix, C date, D temps %, E prix ajusté temps, F.. adjustments, then totals
96		− adj_start = 6 # F
97		− adj_end = adj_start + len(adj_names) - 1 if adj_names else adj_start - 1
98		− col_total = adj_end + 1
99		− col_final = adj_end + 2
100		− col_gross = adj_end + 3
101		− col_gross_pct = adj_end + 4
102		− L = get_column_letter
103	182	rows: list[list[Any]] = []
104		− for i, c in enumerate(comps):
105		− r = first_row + i
106		− row: list[Any] = [c["nom"], c["prix"], c["date"] or "—", c["temps"], f"=B{r}*(1+D{r})"]
107		− for name in adj_names:
108		− row.append(c["adj"].get(name, 0.0))
109		− adj_range = f"{L(adj_start)}{r}:{L(adj_end)}{r}" if adj_names else None
110		− row.append(f"=SUM({adj_range})" if adj_range else 0)
111		− row.append(f"=E{r}+{L(col_total)}{r}")
112		− row.append(f"=SUMPRODUCT(ABS({adj_range}))" if adj_range else 0)
113		− row.append(f"=IF(E{r}=0,0,{L(col_gross)}{r}/E{r})")
114		− rows.append(row)
115		− columns = ([{"header": "Comparable", "type": "text"}, {"header": "Prix de vente ($)", "type": "currency"},
116		− {"header": "Date de vente", "type": "text"}, {"header": "Ajust. temps (%)", "type": "percent"},
117		− {"header": "Prix ajusté temps ($)", "type": "currency"}]
118		− + [{"header": f"{a} ($)", "type": "currency"} for a in adj_names]
119		− + [{"header": "Total ajustements ($)", "type": "currency"},
120		− {"header": "Prix ajusté ($)", "type": "currency"},
121		− {"header": "Ajust. bruts ($)", "type": "currency"},
122		− {"header": "Ajust. bruts (%)", "type": "percent"}])
123		− grid = {"anchor": "A4", "columns": columns, "rows": rows}
124		− F = L(col_final)
125		− G = L(col_gross_pct)
126		− stats_anchor = last + 3
127		− stats = {
128		− "anchor": f"A{stats_anchor}",
	183	+ chars_text: list[list[Any]] = []
	184	+ for i, c in enumerate(comps_raw):
	185	+ cn = {_norm(k): v for k, v in c.items()}
	186	+ r = first + i
	187	+ name = str(cn.get("adresse") or cn.get("nom") or cn.get("comparable") or f"Comparable {i + 1}")
	188	+ line: list[Any] = [name, num(cn.get("prix", cn.get("prix_vente")), 0.0) or 0.0,
	189	+ str(cn.get("date") or cn.get("date_vente") or "—")]
	190	+ tp = _time_pct(cn)
	191	+ if c_mois is not None:
	192	+ line.append(num(cn.get("mois"), 0.0) or 0.0)
	193	+ line.append(f"={L(c_mois)}{r}*{inputs_map['__monthly']}" if tp is None else tp)
	194	+ else:
	195	+ line.append(tp if tp is not None else 0.0)
	196	+ line.append(f"=B{r}*(1+{L(c_temps)}{r})")
	197	+ adj_vals = {_norm(k): v for k, v in (cn.get("ajustements") or {}).items()} if isinstance(cn.get("ajustements"), dict) else {}
	198	+ text_row: list[Any] = [name]
	199	+ if mode_rates:
	200	+ for k in keys:
	201	+ rank = _rank(cn.get(k), scales.get(k))
	202	+ line.append(rank if rank is not None else 0.0)
	203	+ for k in keys:
	204	+ line.append(f"=({inputs_map[k]}-{L(c_char[k])}{r})*{inputs_map[k + '__taux']}")
	205	+ for k, v in cn.items():
	206	+ if k not in META and k not in keys and not isinstance(v, (dict, list)):
	207	+ if k not in text_keys:
	208	+ text_keys.append(k)
	209	+ else:
	210	+ for k in keys:
	211	+ v = adj_vals.get(k, cn.get(k))
	212	+ nv = num(v, None)
	213	+ if nv is None:
	214	+ line.append(0.0)
	215	+ if v not in (None, ""):
	216	+ if k not in text_keys:
	217	+ text_keys.append(k)
	218	+ elif isinstance(v, str) and v.strip().endswith("%"):
	219	+ line.append(f"={L(c_pat)}{r}*{nv}")
	220	+ else:
	221	+ line.append(nv)
	222	+ chars_text.append(text_row)
	223	+ adj_range = f"{L(min(c_adj.values()))}{r}:{L(max(c_adj.values()))}{r}" if c_adj else None
	224	+ line.append(f"=SUM({adj_range})" if adj_range else 0)
	225	+ line.append(f"={L(c_pat)}{r}+{L(c_total)}{r}")
	226	+ line.append(f"=SUMPRODUCT(ABS({adj_range}))" if adj_range else 0)
	227	+ line.append(f"=IF({L(c_pat)}{r}=0,0,{L(c_gross)}{r}/{L(c_pat)}{r})")
	228	+ rows.append(line)
	229	+ sheet["tables"].append({"anchor": f"A{header_row}", "columns": columns, "rows": rows})
	230	+
	231	+ # ---- statistics
	232	+ F, G = L(c_final), L(c_gross_pct)
	233	+ s0 = last + 3
	234	+ sheet["tables"].append({
	235	+ "anchor": f"A{s0}",
129	236	"columns": [{"header": "Statistique", "type": "text"}, {"header": "Valeur", "type": "currency"}],
130	237	"rows": [
131		− ["Minimum des prix ajustés", f"=MIN({F}{first_row}:{F}{last})"],
132		− ["Maximum des prix ajustés", f"=MAX({F}{first_row}:{F}{last})"],
133		− ["Moyenne simple", f"=AVERAGE({F}{first_row}:{F}{last})"],
134		− ["Médiane", f"=MEDIAN({F}{first_row}:{F}{last})"],
135		− ["Comparable le moins ajusté (rang)", f"=MATCH(MIN({G}{first_row}:{G}{last}),{G}{first_row}:{G}{last},0)"],
136		− ["Prix ajusté du comparable le moins ajusté", f"=INDEX({F}{first_row}:{F}{last},B{stats_anchor + 5})"],
	238	+ ["Minimum des prix ajustés", f"=MIN({F}{first}:{F}{last})"],
	239	+ ["Maximum des prix ajustés", f"=MAX({F}{first}:{F}{last})"],
	240	+ ["Moyenne simple (indicatif)", f"=AVERAGE({F}{first}:{F}{last})"],
	241	+ ["Médiane", f"=MEDIAN({F}{first}:{F}{last})"],
	242	+ ["Comparable le moins ajusté (rang)", f"=MATCH(MIN({G}{first}:{G}{last}),{G}{first}:{G}{last},0)"],
	243	+ ["Prix ajusté du comparable le moins ajusté", f"=INDEX({F}{first}:{F}{last},B{s0 + 5})"],
	244	+ ["Pondération réconciliée (poids inverses des ajust. bruts)",
	245	+ f"=SUMPRODUCT({F}{first}:{F}{last},1/(1+{G}{first}:{G}{last}))/SUMPRODUCT(1/(1+{G}{first}:{G}{last}))"],
137	246	],
138	247	"row_formats": {4: "integer"},
139		− }
140		− tables = [grid, stats]
141		− if char_names:
142		− crow = stats_anchor + 9
143		− tables.append({
144		− "anchor": f"A{crow}",
145		− "columns": [{"header": "Comparable", "type": "text"}] + [{"header": c, "type": "text"} for c in char_names],
146		− "rows": [[c["nom"]] + [cell(c["chars"].get(name, "—")) for name in char_names] for c in comps],
	248	+ })
	249	+ next_row = s0 + 10
	250	+
	251	+ # ---- qualitative characteristics kept as text
	252	+ if text_keys:
	253	+ sheet["tables"].append({
	254	+ "anchor": f"A{next_row}",
	255	+ "columns": [{"header": "Comparable", "type": "text"}] + [{"header": _label(k), "type": "text"} for k in text_keys],
	256	+ "rows": [[str(c.get("adresse") or c.get("nom") or f"Comparable {i + 1}")]
	257	+ + [cell({_norm(k2): v for k2, v in c.items()}.get(k, "—")) for k in text_keys]
	258	+ for i, c in enumerate(comps_raw)],
147	259	})
148		− return {
149		− "filename": p.get("filename", "comparables_ajustes.xlsx"),
150		− "style": "uqo",
151		− "objective": f"Grille de comparables ajustés — {sujet}",
152		− "sheets": [{
153		− "name": "Comparables",
154		− "title": f"Grille de comparables ajustés — {sujet}",
155		− "inputs": [],
156		− "tables": tables,
157		− "notes": [
158		− "On ajuste le comparable vers le sujet : le comparable est meilleur → ajustement négatif.",
159		− "Ordre : conditions de vente, financement, marché (temps), puis caractéristiques physiques.",
160		− "La réconciliation pondère les indications (poids plus fort au comparable le moins ajusté) ; ce n'est pas une moyenne.",
161		− ] + ([f"Caractéristiques qualitatives ({', '.join(char_names)}) présentées dans le tableau du bas ; "
162		− "leur traduction en $ est une hypothèse à justifier."] if char_names else []),
163		− }],
164		− }
	260	+ if mode_rates and scales:
	261	+ sheet["notes"].append("Échelles ordinales : " + " ; ".join(f"{_label(k)} = {' < '.join(v)} (rang 0, 1, 2…)" for k, v in scales.items()))
	262	+ sheet["notes"] += [
	263	+ "Ajustement = (caractéristique du sujet − celle du comparable) × taux : le comparable est meilleur → ajustement négatif."
	264	+ if mode_rates else "On ajuste le comparable vers le sujet : le comparable est meilleur → ajustement négatif.",
	265	+ "Ordre des ajustements : conditions de vente, financement, marché (temps), puis caractéristiques physiques.",
	266	+ "La réconciliation pondère les indications (plus de poids au comparable le moins ajusté) ; la moyenne simple n'est qu'indicative.",
	267	+ "Repères du cours : ajustements bruts ≤ 25 % et nets ≤ 15 % pour un comparable fiable.",
	268	+ ]
	269	+ return {"filename": p.get("filename", "comparables_ajustes.xlsx"), "style": "uqo",
	270	+ "objective": f"Grille de comparables ajustés — {sujet_label}", "sheets": [sheet]}

modified backend/app/tools/schemas/create_excel.json +1 −1

		`@@ -1,6 +1,6 @@`
1	1	{
2	2	"name": "create_excel",
3		− "description": "Crée un classeur Excel (.xlsx) professionnel aux couleurs UQO avec formules vivantes, cellules d'hypothèses en bleu, feuille Lisez-moi, graphiques. Deux modes : (1) `template` + `params` pour un gabarit prêt : methode_du_cout (superficie_m2, cout_unitaire_m2, couts_indirects_pct, profit_pct, age_effectif, duree_vie_economique, valeur_terrain, ameliorations_site, depreciation_fonctionnelle, depreciation_economique, adresse), comparables_ajustes (sujet, comparables:[{adresse,prix,date_pct,superficie,terrain,garage,etat}]), age_vie (cout_neuf, age_effectif, duree_vie_economique), tableau_amortissement (capital, taux_annuel, amortissement_ans, versements_par_an), six_fonctions (taux, periodes_max), sensibilite (superficie_m2, cout_unitaire_m2, duree_vie_economique, valeur_terrain, ages, variations) ; (2) `spec` libre : {filename, objective, sheets:[{name, title, inputs:[{cell:'B4', label, value, format:'currency\|percent\|number\|integer\|area\|factor', name}], tables:[{anchor:'A8', columns:[{header,type}], rows:[[...]], totals:{label, formula}}], charts:[{type:'bar\|line\|pie', title, categories_range, values_range, anchor}], notes:[...]}]}. Les valeurs commençant par '=' sont des formules Excel (références relatives aux cellules réelles : la 1re ligne de données d'un tableau ancré en A8 est la ligne 9).",
	3	+ "description": "Crée un classeur Excel (.xlsx) professionnel aux couleurs UQO : formules vivantes, cellules d'hypothèses en bleu, feuille Lisez-moi, graphiques. Tolérant : accepte '425 000 $', '+3 %', 'Oui'/'Non', 'bon'. Deux modes.\n(1) `template` + `params` :\n- comparables_ajustes : {sujet:{adresse, superficie, terrain, garage:'Oui', etat:'bon', …}, taux:{superficie:120, terrain:8, garage:15000, etat:10000} ($/unité ou $/cran), echelles:{etat:['moyen','bon','très bon']}, taux_temps_mensuel:0.005, comparables:[{adresse, prix, date, mois, superficie, terrain, garage, etat, …}]} → les ajustements sont CALCULÉS par formules (sujet − comparable) × taux. Sans `taux`, donne directement des ajustements en $ : comparables:[{adresse, prix, temps:'+2 %', ajustements:{Superficie:-5000}, caracteristiques:{Garage:'Oui'}}].\n- methode_du_cout : {adresse, superficie_m2, cout_unitaire_m2, couts_indirects_pct, profit_pct, age_effectif, duree_vie_economique, valeur_terrain, ameliorations_site, depreciation_fonctionnelle, depreciation_economique}.\n- age_vie : {cout_neuf, age_effectif, duree_vie_economique}.\n- tableau_amortissement : {capital, taux_annuel, amortissement_ans, versements_par_an}.\n- six_fonctions : {taux, periodes_max}.\n- sensibilite : {superficie_m2, cout_unitaire_m2, duree_vie_economique, valeur_terrain, ages:[…], variations:[…]}.\n(2) `spec` libre : {filename, objective, sheets:[{name, title, inputs:[{cell:'B4', label, value, format:'currency\|percent\|number\|integer\|area\|text', name}], tables:[{anchor:'A8', columns:[{header,type}] ou ['Col1','Col2'], rows:[[…]] ou [{col:val}], totals:{label, formula}}], charts:[{type:'bar\|line\|pie', title, categories_range, values_range, anchor}], notes:[…]}]}. Les valeurs commençant par '=' sont des formules (références réelles : la 1re ligne de données d'un tableau ancré en A8 est la ligne 9). Reste compact (≤ 60 lignes par appel).",
4	4	"parameters": {
5	5	"type": "object",
6	6	"properties": {

modified backend/tests/tools/test_coerce_and_spec.py +1 −1

		`@@ -43,7 +43,7 @@ def test_comparables_with_text_values_and_dynamic_columns() -> None:`
43	43	assert ws["B5"].value == 415000 and ws["D5"].value == 0.02
44	44	# -3 % of price → dollars
45	45	piscine_col = headers.index("Piscine ($)") + 1
46		− assert abs(ws.cell(row=5, column=piscine_col).value + 0.03 * 415000) < 1e-6
	46	+ assert ws.cell(row=5, column=piscine_col).value == "=E5*-0.03" # live formula: −3 % of time-adjusted price
47	47	# "Oui" landed in the characteristics table, not in a numeric column
48	48	texts = [c.value for row in ws.iter_rows() for c in row if isinstance(c.value, str)]
49	49	assert "Oui" in texts and "Garage" in texts

added backend/tests/tools/test_comparables_rates.py +43 −0

		`@@ -0,0 +1,43 @@`
	1	+import io
	2	+
	3	+from openpyxl import load_workbook
	4	+
	5	+from app.tools.create_excel import build_workbook
	6	+from app.tools.excel_templates import TEMPLATES
	7	+
	8	+
	9	+def _ws(params): # noqa: ANN001, ANN202
	10	+ data, _ = build_workbook(TEMPLATES["comparables_ajustes"](params))
	11	+ return load_workbook(io.BytesIO(data))["Comparables"]
	12	+
	13	+
	14	+def test_rates_mode_builds_live_formulas() -> None:
	15	+ ws = _ws({
	16	+ "sujet": {"adresse": "Bungalow Hull", "superficie": "1 200 pi²", "terrain": 6000, "garage": "Oui", "etat": "bon"},
	17	+ "taux": {"superficie": "120 $", "terrain": 8, "garage": "15 000 $", "etat": 10000},
	18	+ "taux_temps_mensuel": "0,5 %",
	19	+ "comparables": [
	20	+ {"adresse": "45 rue Laurier", "prix": "425 000 $", "date": "mars 2026", "mois": 6, "superficie": 1150,
	21	+ "terrain": 5500, "garage": "Oui", "etat": "bon"},
	22	+ {"adresse": "12 rue Front", "prix": 398000, "mois": 8, "superficie": 1250, "terrain": 7000,
	23	+ "garage": "Non", "etat": "moyen"},
	24	+ ],
	25	+ })
	26	+ values = {c.coordinate: c.value for row in ws.iter_rows() for c in row if c.value is not None}
	27	+ # subject inputs and rates present as blue inputs
	28	+ assert values["B4"] == 1200 and values["D4"] == 120
	29	+ assert any(v == "Oui".lower() or v == 1.0 for k, v in values.items() if k.startswith("B")) # garage → 1
	30	+ # adjustment formulas reference subject/rate cells
	31	+ formulas = [v for v in values.values() if isinstance(v, str) and v.startswith("=(")]
	32	+ assert formulas and all("$B$" in f and "$D$" in f for f in formulas)
	33	+ # time adjustment uses months × monthly rate
	34	+ assert any(isinstance(v, str) and v.startswith("=D") and "*$D$" in v for v in values.values())
	35	+ assert any(h == "Ajust. Garage ($)" for h in values.values())
	36	+ assert any(h == "Médiane" for h in values.values())
	37	+
	38	+
	39	+def test_dollar_mode_still_works_with_text() -> None:
	40	+ ws = _ws({"comparables": [{"adresse": "A", "prix": 415000, "temps": "+2 %", "garage": "Oui",
	41	+ "ajustements": {"Superficie": -5000}}]})
	42	+ values = [c.value for row in ws.iter_rows() for c in row if c.value is not None]
	43	+ assert "Oui" in values and -5000 in values and "Garage" in values
	44