File size: 58,696 Bytes
1d9bd9b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 | """Moonley API service. Loads the legal tool registry once and streams grounded research over SSE.
The React interface is hosted only on Vercel; this process is the private HF backend. Run:
MOONLEY_DATA=.../thor_artifacts MOONLEY_STATUTE=".../statute corpus" \
.venv/bin/uvicorn --app-dir phase1/scripts serve_agent:app --host 127.0.0.1 --port 8001
"""
import os, sys, re, json, time, hashlib
from urllib.parse import unquote
def _promote_moonley_environment() -> None:
"""Let unchanged corpus internals consume canonical Moonley configuration."""
for name, value in list(os.environ.items()):
if name.startswith("MOONLEY_"):
os.environ.setdefault("THEMIS_" + name[len("MOONLEY_"):], value)
def _load_env(path: str) -> None:
if os.path.exists(path):
for line in open(path):
line = line.strip()
if line and not line.startswith("#") and "=" in line:
key, value = line.split("=", 1)
os.environ.setdefault(key.strip(), value.strip().strip('"').strip("'"))
HERE = os.path.dirname(os.path.abspath(__file__))
_load_env(os.path.join(HERE, ".env"))
_promote_moonley_environment()
sys.path.insert(0, HERE)
from tools import Corpus as LegacyCorpus
from corpus_v5 import CorpusV5
import agent as A
import requests
from pdf_sources import PdfSourceResolver
from bharat_courts_source import BharatCourtsPdfError, resolve_and_fetch_pdf
from graph_view import graph_node_card
from project_store import ProjectStore, ProjectStoreError, QuotaExceeded
from drafting_service import (
DRAFT_PROFILES,
DraftingError,
TemplateRegistry,
apply_drafting_intake,
draft_docx,
draft_pdf,
draft_profile,
drafting_intake_messages,
drafting_messages,
finalization_messages,
infer_draft_profile,
missing_draft_fields,
public_draft_profile,
revision_messages,
extract_uploaded_template,
)
from knowledge_service import KnowledgeService, KnowledgeServiceError
from research_release import build_research_release
from statute_crosswalk import ACT_NAMES, normalise_act, normalise_section
from fastapi import BackgroundTasks, FastAPI, Request
from fastapi.responses import FileResponse, StreamingResponse, JSONResponse, RedirectResponse, Response
from clerk_auth import ( # noqa: E402 - the local .env must be loaded first
PUBLIC_PATHS,
authenticate_clerk_request,
clerk_settings,
cors_origins,
frontend_auth_config,
)
HDR = {"Authorization": f"Bearer {os.environ.get('DEEPSEEK_API_KEY','')}", "Content-Type": "application/json"}
def llm_fn(msgs):
for _ in range(2):
try:
r = requests.post("https://api.deepseek.com/chat/completions", headers=HDR, timeout=60,
json={"model": "deepseek-v4-flash", "temperature": 0, "max_tokens": 700,
"thinking": {"type": "disabled"}, "messages": msgs})
if r.status_code == 200: return r.json()["choices"][0]["message"]["content"]
except Exception: time.sleep(1)
return "{}"
def fast_llm_fn(msgs):
"""Fast user-facing turn: fail closed instead of holding the interface through retries."""
try:
r = requests.post("https://api.deepseek.com/chat/completions", headers=HDR, timeout=30,
json={"model": "deepseek-v4-flash", "temperature": 0, "max_tokens": 600,
"thinking": {"type": "disabled"}, "messages": msgs})
if r.status_code == 200:
return r.json()["choices"][0]["message"]["content"]
except Exception:
pass
return "{}"
def ds_call(messages, tools):
"""DeepSeek function-calling turn -> the assistant message (with tool_calls or content)."""
r = requests.post("https://api.deepseek.com/chat/completions", headers=HDR, timeout=90,
json={"model": "deepseek-v4-flash", "temperature": 0, "max_tokens": 800,
"thinking": {"type": "disabled"},
"messages": messages, "tools": tools, "tool_choice": "auto"})
return r.json()["choices"][0]["message"]
DATA_DIR = os.environ.get("THEMIS_DATA", ".")
STATUTE_DIR = os.environ.get("THEMIS_STATUTE", ".")
if os.path.exists(os.path.join(DATA_DIR, "release_manifest.json")):
C = CorpusV5(DATA_DIR, STATUTE_DIR, device=os.environ.get("THEMIS_DEVICE", "cpu"))
RUNTIME_KIND = "schema-v5-qwen"
else:
C = LegacyCorpus(DATA_DIR, STATUTE_DIR, device=os.environ.get("THEMIS_DEVICE", "cpu"))
RUNTIME_KIND = "legacy-bge"
RESEARCH_RELEASE = build_research_release(C, A)
# citation resolution for the judgment view (neutral + equivalent -> doc)
def norm_cite(c): return re.sub(r"\s+", " ", (c or "").replace(".", "")).strip().upper()
cite_resolver = {}; nc2doc = {}
for _d, _m in C.meta.items():
if _m.get("neutral_citation"): nc2doc[_m["neutral_citation"]] = _d
for _k in [_m.get("neutral_citation")] + (_m.get("equivalent_citations") or []):
if _k: cite_resolver.setdefault(norm_cite(_k), _d)
CITE_RE = re.compile(r"\[\d{4}\]\s*\d+\s*S\.?C\.?R\.?\s*\d+|\(\d{4}\)\s*\d+\s*SCC\s*\d+|\d{4}\s+INSC\s+\d+|AIR\s+\d{4}\s+SC\s+\d+")
def doc_links(text, self_id):
out = {}
for c in CITE_RE.findall(text or ""):
rid = cite_resolver.get(norm_cite(c))
if rid and rid != self_id and C.is_retrieval_eligible(rid) and c not in out: out[c] = rid
return [{"cite": k, "id": v} for k, v in out.items()]
def resolve_cited(cases_cited, self_id):
out = []
for c in (cases_cited or []):
rid = None
for cstr in (c.get("citations") or []):
for part in re.split(r"\s*[:;]\s*", cstr):
rid = cite_resolver.get(norm_cite(part))
if rid and rid != self_id: break
rid = None
if rid: break
if not rid and c.get("name"):
hits = C.name_lookup(c["name"], 1)
if hits and hits[0]["doc_id"] != self_id: rid = hits[0]["doc_id"]
if rid:
card = graph_node_card(
rid,
C.meta.get(rid, {}),
treatment=c.get("treatment"),
cited_by=C.cite_indeg.get(rid, 0),
good_law_status=C.goodlaw.get(rid, {}).get(
"good_law_status", "unknown"
),
)
if not card["hover"]["case_name"] and c.get("name"):
card["display_name"] = c["name"]
card["name"] = c["name"]
card["hover"]["case_name"] = c["name"]
out.append(card)
else:
out.append(
{
"name": c.get("name"),
"display_name": c.get("name") or "Unresolved cited case",
"citation": ((c.get("citations") or [""])[0]),
"treatment": c.get("treatment"),
"id": None,
"node_id": None,
"judgment_id": None,
"label": None,
}
)
return out
# --- verified source PDFs -----------------------------------------------------
# Map membership is not availability: the upstream bucket contains a small number
# of application/pdf objects whose payload is actually an HTML error page. Probe a
# bounded byte range, expose the PDF only after its payload is verified, and let the
# browser load the public source directly so large PDFs retain byte-range support.
PDF_SOURCES = PdfSourceResolver(os.path.join(DATA_DIR, "escr_pdfmap.jsonl"))
print(f"[serve_agent] pdfmap: {PDF_SOURCES.mapped_count} unique judgments have mapped source candidates", flush=True)
app = FastAPI(title="Moonley API", description="Private grounded Indian legal research API", version="2")
RUNTIME_WARM = RUNTIME_KIND != "schema-v5-qwen"
PROJECTS = ProjectStore.from_env()
DRAFTING = TemplateRegistry(os.path.join(HERE, "..", "drafting"))
KNOWLEDGE = KnowledgeService(PROJECTS, C)
@app.on_event("startup")
def _warm_runtime():
global RUNTIME_WARM
if RUNTIME_KIND == "schema-v5-qwen" and os.environ.get("THEMIS_WARM_QUERY_MODEL", "1") == "1":
C.warmup()
RUNTIME_WARM = True
print(f"[serve_agent] READY runtime={RUNTIME_KIND} accepted={len(C.eligible_doc_ids)}", flush=True)
# --- Clerk access gate (public hosting) ---
@app.middleware("http")
async def _clerk_gate(request: Request, call_next):
# CORS preflights and the boot/config endpoints must be reachable before sign-in.
if request.method != "OPTIONS" and request.url.path not in PUBLIC_PATHS:
rejection = authenticate_clerk_request(request)
if rejection is not None:
return rejection
return await call_next(request)
# CORS and Clerk's authorized-parties check share one explicit origin allow-list.
from fastapi.middleware.cors import CORSMiddleware
app.add_middleware(CORSMiddleware, allow_origins=cors_origins(), allow_methods=["*"],
allow_headers=["*"], expose_headers=["*"])
def sse(o): return "data: " + json.dumps(o, ensure_ascii=False) + "\n\n"
# --- SESSION CAPTURE (the pooled-verification machine for daily lawyer sessions) ---
# Every search + every 👍/👎 lands in append-only JSONL; each graded result is a future qrel row.
LOG_DIR = os.environ.get("THEMIS_LOG_DIR") or os.path.join(HERE, "..", "logs")
os.makedirs(LOG_DIR, exist_ok=True)
def _log(name, obj):
try:
obj = {"ts": time.strftime("%Y-%m-%dT%H:%M:%S"), **obj}
with open(os.path.join(LOG_DIR, f"{name}.jsonl"), "a", encoding="utf-8") as fh:
fh.write(json.dumps(obj, ensure_ascii=False) + "\n")
except Exception:
pass
LOG_RAW_QUERIES = os.environ.get("THEMIS_LOG_RAW_QUERIES", "0") == "1"
def _query_log_fields(query: str) -> dict:
normalized = re.sub(r"\s+", " ", query or "").strip()
fields = {
"query_sha256": hashlib.sha256(normalized.encode("utf-8")).hexdigest(),
"query_chars": len(normalized),
}
if LOG_RAW_QUERIES:
fields["q"] = normalized[:2000]
return fields
from pydantic import BaseModel, Field
class CaseChatTurn(BaseModel):
role: str
content: str
class QueryBriefRequest(BaseModel):
query: str
refinements: list[str] = Field(default_factory=list)
history: list[CaseChatTurn] = Field(default_factory=list)
active_case_id: str | None = None
recent_case_ids: list[str] = Field(default_factory=list)
class CaseChatRequest(BaseModel):
doc_id: str
question: str
history: list[CaseChatTurn] = Field(default_factory=list)
class SearchRequest(BaseModel):
q: str
original_q: str | None = None
approved: bool = True
search_frame: dict | None = None
brief_revision: int | None = None
route: str = "legal_research"
retrieval_scope: str = "global"
active_case_id: str | None = None
recent_case_ids: list[str] = Field(default_factory=list)
history: list[CaseChatTurn] = Field(default_factory=list)
case_question: str | None = None
class ProjectRequest(BaseModel):
name: str
class DraftChatSource(BaseModel):
id: str = Field(default="", max_length=200)
title: str = Field(default="Saved chat", max_length=120)
content: str = Field(max_length=16_000)
class DraftMatterDetails(BaseModel):
matter_title: str = Field(default="", max_length=500)
parties: str = Field(default="", max_length=4_000)
lower_court: str = Field(default="", max_length=500)
case_number: str = Field(default="", max_length=300)
impugned_order_date: str = Field(default="", max_length=100)
synopsis: str = Field(default="", max_length=8_000)
list_of_dates: str = Field(default="", max_length=8_000)
questions_of_law: str = Field(default="", max_length=8_000)
grounds: str = Field(default="", max_length=8_000)
relief: str = Field(default="", max_length=8_000)
advocate: str = Field(default="", max_length=500)
class DraftRequest(BaseModel):
template_id: str = Field(default="", max_length=100)
template_text: str = Field(default="", max_length=60_000)
document_type: str = Field(default="", max_length=100)
project_id: str | None = None
document_ids: list[str] = Field(default_factory=list)
chat_sources: list[DraftChatSource] = Field(default_factory=list)
matter_details: DraftMatterDetails = Field(default_factory=DraftMatterDetails)
intake_details: dict[str, str] = Field(default_factory=dict)
instructions: str = Field(default="", max_length=6_000)
class DraftExportRequest(BaseModel):
title: str = Field(default="Moonley working draft", max_length=180)
draft: str = Field(max_length=80_000)
class DraftIntakeTurn(BaseModel):
role: str = Field(max_length=20)
content: str = Field(max_length=2_000)
class DraftIntakeRequest(BaseModel):
message: str = Field(max_length=4_000)
document_type: str = Field(default="", max_length=100)
details: dict[str, str] = Field(default_factory=dict)
history: list[DraftIntakeTurn] = Field(default_factory=list)
class DraftFinalizeRequest(DraftExportRequest):
document_type: str = Field(default="", max_length=100)
class DraftRevisionRequest(DraftFinalizeRequest):
instruction: str = Field(max_length=4_000)
def _project_owner(request: Request) -> str:
return str(getattr(request.state, "clerk_user_id", ""))
def _project_error(exc: ProjectStoreError) -> JSONResponse:
return JSONResponse(
{"error": exc.code, "message": exc.message},
status_code=exc.status_code,
headers={"Cache-Control": "no-store"},
)
@app.get("/api/v2/auth/config")
def auth_config():
return frontend_auth_config()
@app.get("/api/v2/projects")
def list_projects(request: Request):
try:
projects = PROJECTS.list_projects(_project_owner(request))
return JSONResponse(
{"projects": projects, "storage": PROJECTS.status()},
headers={"Cache-Control": "no-store"},
)
except ProjectStoreError as exc:
return _project_error(exc)
@app.post("/api/v2/projects")
def create_project(request: Request, body: ProjectRequest):
try:
project = PROJECTS.create_project(_project_owner(request), body.name)
return JSONResponse(
{"project": project, "storage": PROJECTS.status()},
status_code=201,
headers={"Cache-Control": "no-store"},
)
except ProjectStoreError as exc:
return _project_error(exc)
@app.get("/api/v2/projects/{project_id}")
def get_project(project_id: str, request: Request):
try:
return JSONResponse(
{"project": PROJECTS.get_project(_project_owner(request), project_id)},
headers={"Cache-Control": "no-store"},
)
except ProjectStoreError as exc:
return _project_error(exc)
@app.patch("/api/v2/projects/{project_id}")
def rename_project(project_id: str, request: Request, body: ProjectRequest):
try:
return JSONResponse(
{"project": PROJECTS.rename_project(_project_owner(request), project_id, body.name)},
headers={"Cache-Control": "no-store"},
)
except ProjectStoreError as exc:
return _project_error(exc)
@app.delete("/api/v2/projects/{project_id}")
def delete_project(project_id: str, request: Request):
try:
KNOWLEDGE.delete_project(_project_owner(request), project_id)
PROJECTS.delete_project(_project_owner(request), project_id)
return Response(status_code=204, headers={"Cache-Control": "no-store"})
except ProjectStoreError as exc:
return _project_error(exc)
except KnowledgeServiceError as exc:
return JSONResponse(
{"error": "knowledge_delete_failed", "message": str(exc)},
status_code=502,
headers={"Cache-Control": "no-store"},
)
@app.post("/api/v2/projects/{project_id}/documents")
async def upload_project_document(project_id: str, request: Request, background_tasks: BackgroundTasks):
try:
content_length = request.headers.get("content-length", "").strip()
if content_length and int(content_length) > PROJECTS.limits.max_file_bytes:
raise QuotaExceeded(
f"Each document must be {PROJECTS.limits.max_file_bytes // (1024 * 1024)} MiB or smaller."
)
filename = unquote(request.headers.get("x-document-name", ""))
content = bytearray()
async for chunk in request.stream():
content.extend(chunk)
if len(content) > PROJECTS.limits.max_file_bytes:
raise QuotaExceeded(
f"Each document must be {PROJECTS.limits.max_file_bytes // (1024 * 1024)} MiB or smaller."
)
document = PROJECTS.add_document(_project_owner(request), project_id, filename, bytes(content))
background_tasks.add_task(
KNOWLEDGE.ingest, _project_owner(request), project_id, document["id"]
)
return JSONResponse(
{"document": document, "project": PROJECTS.get_project(_project_owner(request), project_id)},
status_code=201,
headers={"Cache-Control": "no-store"},
)
except (ValueError, ProjectStoreError) as exc:
if isinstance(exc, ProjectStoreError):
return _project_error(exc)
return JSONResponse({"error": "invalid_content_length"}, status_code=400)
@app.delete("/api/v2/projects/{project_id}/documents/{document_id}")
def delete_project_document(project_id: str, document_id: str, request: Request):
try:
KNOWLEDGE.delete(_project_owner(request), project_id, document_id)
PROJECTS.delete_document(_project_owner(request), project_id, document_id)
return Response(status_code=204, headers={"Cache-Control": "no-store"})
except ProjectStoreError as exc:
return _project_error(exc)
except KnowledgeServiceError as exc:
return JSONResponse(
{"error": "knowledge_delete_failed", "message": str(exc)},
status_code=502,
headers={"Cache-Control": "no-store"},
)
@app.get("/api/v2/projects/{project_id}/documents/{document_id}")
def download_project_document(project_id: str, document_id: str, request: Request):
try:
document = PROJECTS.document_record(_project_owner(request), project_id, document_id)
path = PROJECTS.document_path(_project_owner(request), project_id, document_id)
return FileResponse(
path,
media_type=document.get("media_type") or "application/octet-stream",
filename=document.get("name") or "document",
headers={"Cache-Control": "private, no-store"},
)
except ProjectStoreError as exc:
return _project_error(exc)
@app.post("/api/v2/projects/{project_id}/documents/{document_id}/ingest", status_code=202)
def ingest_project_document(
project_id: str, document_id: str, request: Request, background_tasks: BackgroundTasks
):
try:
PROJECTS.document_record(_project_owner(request), project_id, document_id)
background_tasks.add_task(KNOWLEDGE.ingest, _project_owner(request), project_id, document_id)
return JSONResponse(
{"status": "queued", "document_id": document_id},
status_code=202,
headers={"Cache-Control": "no-store"},
)
except ProjectStoreError as exc:
return _project_error(exc)
@app.get("/api/v2/statute-crosswalk")
def statute_crosswalk(act: str, section: str):
code, number = normalise_act(act), normalise_section(section)
if not code or not number:
return JSONResponse(
{
"error": "invalid_provision",
"message": "Choose IPC, BNS, CrPC, BNSS, IEA, or BSA and enter a section number.",
},
status_code=400,
)
result = C.statute_crosswalk(code, number)
result["provision"] = C.statute_provision(code, number)
for item in result.get("corresponding") or []:
item["provision"] = C.statute_provision(item["act"], item["section"])
result["supported_acts"] = [
{"act": value, "name": ACT_NAMES[value]} for value in ("IPC", "BNS", "CRPC", "BNSS", "IEA", "BSA")
]
return JSONResponse(result, headers={"Cache-Control": "private, max-age=3600"})
@app.get("/api/v2/statute-lookup")
def statute_lookup(act: str, section: str):
"""Exact statutory-text lookup; never infer or substitute another section."""
code, number = normalise_act(act), normalise_section(section)
if not code or not number:
return JSONResponse(
{"found": False, "error": "invalid_provision", "message": "Use IPC, BNS, CrPC, BNSS, IEA, or BSA with an exact section number."},
status_code=400,
)
provision = C.statute_provision(code, number)
if not provision:
return JSONResponse(
{"found": False, "act": code, "section": number, "message": "The exact provision is not available in the private statute source; do not guess it."},
headers={"Cache-Control": "private, max-age=300"},
)
return JSONResponse(
{"found": True, "act": code, "act_name": ACT_NAMES.get(code), "section": number, "provision": provision},
headers={"Cache-Control": "private, max-age=3600"},
)
@app.get("/api/v2/drafting/templates")
def drafting_templates():
return JSONResponse(
{
"version": DRAFTING.version,
"templates": DRAFTING.list(),
"document_types": [
public_draft_profile(profile)
for profile in DRAFT_PROFILES.values()
],
"knowledge": KNOWLEDGE.status(),
},
headers={"Cache-Control": "private, max-age=300"},
)
@app.get("/api/v2/drafting/templates/{template_id}/pdf")
def drafting_template_pdf(template_id: str):
try:
template = DRAFTING.get(template_id)
return FileResponse(
template["path"],
media_type="application/pdf",
filename=template["filename"],
headers={"Cache-Control": "private, max-age=3600"},
)
except DraftingError as exc:
return JSONResponse({"error": "template_not_found", "message": str(exc)}, status_code=404)
@app.get("/api/v2/drafting/templates/{template_id}/text")
def drafting_template_text(template_id: str):
try:
template = DRAFTING.get(template_id)
return JSONResponse(
{
"template": {
key: value
for key, value in template.items()
if key not in {"path", "filename"}
},
"text": DRAFTING.text(template_id),
"editable": True,
},
headers={"Cache-Control": "private, no-store"},
)
except DraftingError as exc:
return JSONResponse({"error": "template_not_found", "message": str(exc)}, status_code=404)
@app.post("/api/v2/drafting/templates/extract")
async def extract_private_drafting_template(request: Request):
"""Extract one authenticated user's template without retaining the uploaded file."""
max_bytes = 10 * 1024 * 1024
content_length = request.headers.get("content-length", "").strip()
if content_length and int(content_length) > max_bytes:
return JSONResponse(
{"error": "template_too_large", "message": "Template must be 10 MiB or smaller."},
status_code=413,
headers={"Cache-Control": "no-store"},
)
filename = unquote(request.headers.get("x-template-name", ""))
content = bytearray()
async for chunk in request.stream():
content.extend(chunk)
if len(content) > max_bytes:
return JSONResponse(
{"error": "template_too_large", "message": "Template must be 10 MiB or smaller."},
status_code=413,
headers={"Cache-Control": "no-store"},
)
try:
extracted = extract_uploaded_template(filename, bytes(content), request.headers.get("content-type", ""))
return JSONResponse(
extracted,
headers={"Cache-Control": "no-store"},
)
except DraftingError as exc:
return JSONResponse(
{"error": "template_extraction_failed", "message": str(exc)},
status_code=400,
headers={"Cache-Control": "no-store"},
)
def draft_llm_fn(messages, *, max_tokens: int = 5000, timeout: int = 150):
try:
response = requests.post(
"https://api.deepseek.com/chat/completions",
headers=HDR,
timeout=timeout,
json={
"model": "deepseek-v4-flash",
"temperature": 0,
"max_tokens": max_tokens,
"thinking": {"type": "disabled"},
"messages": messages,
},
)
if response.status_code == 200:
return str(response.json()["choices"][0]["message"]["content"] or "").strip()
except Exception:
pass
return ""
@app.post("/api/v2/drafting/intake")
def drafting_intake(request: Request, body: DraftIntakeRequest):
message = re.sub(r"\x00", "", body.message or "").strip()[:4_000]
if not message:
return JSONResponse(
{"error": "empty_message", "message": "Tell Moonley what you want drafted."},
status_code=400,
)
current_profile = draft_profile(body.document_type)
prompt_profile = current_profile or draft_profile(infer_draft_profile(message))
messages = drafting_intake_messages(
message,
prompt_profile,
body.details,
[turn.dict() for turn in body.history[-8:]],
)
output = draft_llm_fn(messages, max_tokens=800, timeout=60)
model_returned = bool(output)
if not output and current_profile:
missing = missing_draft_fields(current_profile, body.details)
if missing:
output = json.dumps(
{
"document_type": current_profile["id"],
"updates": {missing[0]["key"]: message},
"acknowledgement": "Noted.",
}
)
state = apply_drafting_intake(
message,
body.document_type,
body.details,
output,
)
_log(
"drafting_intake",
{
"owner_sha256": hashlib.sha256(_project_owner(request).encode("utf-8")).hexdigest(),
"document_type": state.get("document_type"),
"detail_count": len(state.get("details") or {}),
"ready": bool(state.get("ready")),
"model_returned": model_returned,
},
)
return JSONResponse(
{
**state,
"model_call": {
"provider": "deepseek",
"attempted": True,
"succeeded": model_returned,
},
},
headers={"Cache-Control": "no-store"},
)
@app.post("/api/v2/drafting/generate")
def generate_draft(request: Request, body: DraftRequest):
owner = _project_owner(request)
try:
profile = draft_profile(body.document_type)
if profile:
missing = missing_draft_fields(profile, body.intake_details)
if missing:
return JSONResponse(
{
"error": "draft_intake_incomplete",
"message": f"Complete the drafting chat first: {missing[0]['label']} is still required.",
"missing_fields": [field["key"] for field in missing],
},
status_code=409,
)
template_id = body.template_id or str(profile.get("template_id") or "")
else:
template_id = body.template_id
if template_id:
template = DRAFTING.get(template_id)
edited_template = re.sub(r"\x00", "", body.template_text or "").strip()[:60_000]
template_text = edited_template or DRAFTING.text(template_id)
elif profile:
template = {
"id": profile["id"],
"title": profile["title"],
"description": profile["description"],
"category": "Chat-led",
}
edited_template = re.sub(r"\x00", "", body.template_text or "").strip()[:60_000]
template_text = edited_template or str(profile.get("structure") or "")
else:
raise DraftingError("Tell Moonley what document to draft first.")
sources = []
document_ids = list(dict.fromkeys(body.document_ids))[:8]
if document_ids and not body.project_id:
raise DraftingError("Choose the project that owns the selected documents.")
for document_id in document_ids:
document = PROJECTS.document_record(owner, body.project_id or "", document_id)
text, extraction = KNOWLEDGE.source_text(owner, body.project_id or "", document_id)
if text:
sources.append(
{
"label": f"Project document: {document.get('name')}",
"text": text,
"kind": "document",
"document_id": document_id,
"extraction": extraction,
}
)
for chat in body.chat_sources[:5]:
text = re.sub(r"\x00", "", chat.content or "").strip()[:16_000]
if text:
sources.append({"label": f"Selected chat: {chat.title[:120]}", "text": text, "kind": "chat"})
messages = drafting_messages(
template,
template_text,
body.instructions,
sources,
body.matter_details.dict(),
intake_details=body.intake_details,
profile=profile,
)
draft = draft_llm_fn(messages)
if not draft:
return JSONResponse(
{"error": "draft_generation_unavailable", "message": "The drafting model did not return a draft. Try again."},
status_code=503,
)
_log(
"drafting",
{
"owner_sha256": hashlib.sha256(owner.encode("utf-8")).hexdigest(),
"template_id": template_id or profile.get("id"),
"document_type": body.document_type,
"document_count": len(document_ids),
"chat_count": len(body.chat_sources[:5]),
},
)
return JSONResponse(
{
"draft": draft[:80_000],
"template": {key: value for key, value in template.items() if key not in {"path", "filename"}},
"sources": [
{key: value for key, value in source.items() if key not in {"text"}}
for source in sources
],
"notice": "Working draft only. Verify every fact, authority, annexure and filing requirement before use.",
},
headers={"Cache-Control": "no-store"},
)
except ProjectStoreError as exc:
return _project_error(exc)
except DraftingError as exc:
return JSONResponse({"error": "invalid_draft_request", "message": str(exc)}, status_code=400)
@app.post("/api/v2/drafting/finalize")
def finalize_draft(request: Request, body: DraftFinalizeRequest):
draft = re.sub(r"\x00", "", body.draft or "").strip()[:80_000]
if not draft:
return JSONResponse(
{"error": "empty_draft", "message": "Generate or enter a draft before finalizing."},
status_code=400,
)
profile = draft_profile(body.document_type)
final = draft_llm_fn(
finalization_messages(body.title, draft, profile),
max_tokens=6_000,
timeout=150,
)
if not final:
return JSONResponse(
{"error": "finalization_unavailable", "message": "The drafting model did not return a final version. Your editable draft is unchanged."},
status_code=503,
)
_log(
"drafting_finalize",
{
"owner_sha256": hashlib.sha256(_project_owner(request).encode("utf-8")).hexdigest(),
"document_type": body.document_type,
"input_chars": len(draft),
},
)
return JSONResponse(
{
"draft": final[:80_000],
"notice": "Finalized working draft only. Counsel must verify the record, law and filing requirements.",
},
headers={"Cache-Control": "no-store"},
)
@app.post("/api/v2/drafting/revise")
def revise_draft(request: Request, body: DraftRevisionRequest):
draft = re.sub(r"\x00", "", body.draft or "").strip()[:80_000]
instruction = re.sub(r"\x00", "", body.instruction or "").strip()[:4_000]
if not draft:
return JSONResponse(
{"error": "empty_draft", "message": "Generate or enter a draft before asking for changes."},
status_code=400,
)
if not instruction:
return JSONResponse(
{"error": "empty_instruction", "message": "Tell Moonley what to change or what new draft to prepare."},
status_code=400,
)
profile = draft_profile(body.document_type)
revised = draft_llm_fn(
revision_messages(body.title, draft, instruction, profile),
max_tokens=6_000,
timeout=150,
)
if not revised:
return JSONResponse(
{"error": "revision_unavailable", "message": "The drafting model did not return an update. Your editable draft is unchanged."},
status_code=503,
)
_log(
"drafting_revision",
{
"owner_sha256": hashlib.sha256(_project_owner(request).encode("utf-8")).hexdigest(),
"document_type": body.document_type,
"input_chars": len(draft),
"instruction_chars": len(instruction),
},
)
return JSONResponse(
{
"draft": revised[:80_000],
"model_call": {"provider": "deepseek", "attempted": True, "succeeded": True},
"notice": "AI-updated working draft only. Review every change before finalizing.",
},
headers={"Cache-Control": "no-store"},
)
@app.post("/api/v2/drafting/export/docx")
def export_draft_docx(body: DraftExportRequest):
draft = re.sub(r"\x00", "", body.draft or "").strip()[:80_000]
if not draft:
return JSONResponse(
{"error": "empty_draft", "message": "Generate or enter a draft before exporting."},
status_code=400,
)
title = re.sub(r"\s+", " ", body.title or "").strip()[:180] or "Moonley working draft"
filename = re.sub(r"[^A-Za-z0-9._-]+", "-", title).strip("-.")[:80] or "moonley-working-draft"
return Response(
content=draft_docx(title, draft),
media_type="application/vnd.openxmlformats-officedocument.wordprocessingml.document",
headers={
"Cache-Control": "private, no-store",
"Content-Disposition": f'attachment; filename="{filename}.docx"',
},
)
@app.post("/api/v2/drafting/export/pdf")
def export_draft_pdf(body: DraftExportRequest):
draft = re.sub(r"\x00", "", body.draft or "").strip()[:80_000]
if not draft:
return JSONResponse(
{"error": "empty_draft", "message": "Generate or enter a draft before exporting."},
status_code=400,
)
title = re.sub(r"\s+", " ", body.title or "").strip()[:180] or "Moonley working draft"
filename = re.sub(r"[^A-Za-z0-9._-]+", "-", title).strip("-.")[:80] or "moonley-working-draft"
try:
payload = draft_pdf(title, draft)
except DraftingError as exc:
return JSONResponse({"error": "pdf_export_failed", "message": str(exc)}, status_code=400)
return Response(
content=payload,
media_type="application/pdf",
headers={
"Cache-Control": "private, no-store",
"Content-Disposition": f'attachment; filename="{filename}.pdf"',
},
)
@app.post("/api/v2/query_brief")
@app.post("/api/query_brief")
def query_brief(req: QueryBriefRequest):
query = re.sub(r"\s+", " ", req.query or "").strip()
if not query:
return JSONResponse({"error": "query is required"}, status_code=400)
history = [
turn.model_dump() if hasattr(turn, "model_dump") else turn.dict()
for turn in req.history[-8:]
]
active_doc = _eligible_doc(req.active_case_id or "")
active_case = C._card(active_doc) if active_doc else None
brief = A.query_brief(
query,
req.refinements,
fast_llm_fn,
history=history,
active_case=active_case,
)
route = str(brief.get("route") or "legal_research")
scope = str(brief.get("retrieval_scope") or "global")
if route.startswith("case_") or scope == "case_plus_global":
resolution = A.resolve_case_reference(
C,
brief.get("case_reference") or query,
active_case_id=active_doc,
recent_case_ids=req.recent_case_ids,
)
brief["case_resolution"] = resolution
if resolution.get("status") == "resolved":
brief["active_case"] = resolution.get("case")
elif resolution.get("status") == "ambiguous":
brief["case_message"] = (
"I found more than one plausible case-title match in the corpus. "
"Choose the intended judgment; Moonley will not silently substitute one case for another."
)
else:
reference = re.sub(r"\s+", " ", str(brief.get("case_reference") or query)).strip()
brief["case_message"] = (
f"I could not find an exact or reliable close match for {reference!r} in the Supreme Court corpus. "
"Add a citation, year, another party name, or subject if you want me to search differently."
)
crosswalks = []
for mention in A.extract_statute_mentions(" ".join([query, *req.refinements])):
result = C.statute_crosswalk(mention["act"], mention["section"])
if result.get("found"):
result["provision"] = C.statute_provision(
mention["act"], mention["section"]
)
for item in result.get("corresponding") or []:
item["provision"] = C.statute_provision(
item["act"], item["section"]
)
crosswalks.append(result)
if crosswalks:
brief["statute_crosswalks"] = crosswalks
if brief.get("mode") == "research":
provisions = list(brief.get("provisions") or [])
for item in crosswalks:
label = f"{item['from']} corresponds directly to {item['to']}"
if label not in provisions:
provisions.append(label)
brief["provisions"] = provisions[:8]
_log("query_briefs", {
**_query_log_fields(query),
"refinement_count": len([x for x in req.refinements if str(x).strip()]),
"history_turn_count": len(history),
"route": brief.get("route"),
"case_resolution": (brief.get("case_resolution") or {}).get("status"),
})
return JSONResponse(brief)
def _eligible_results(rows):
out, seen = [], set()
for card in rows or []:
if not isinstance(card, dict):
continue
d = str(card.get("judgment_id") or card.get("doc_id") or "")
if not d or d in seen or not C.is_retrieval_eligible(d):
continue
copy = dict(card)
copy["doc_id"] = d
copy["judgment_id"] = d
out.append(copy)
seen.add(d)
return out
def _search_response(
q: str,
*,
original_q: str | None = None,
approved_frame: dict | None = None,
route: str = "legal_research",
retrieval_scope: str = "global",
active_case_id: str | None = None,
case_question: str | None = None,
history: list[dict] | None = None,
primary_limit: int = 6,
more_limit: int = 14,
):
q = re.sub(r"\s+", " ", q or "").strip()
original_q = re.sub(r"\s+", " ", original_q or q).strip()
if not q:
return JSONResponse({"error": "query is required"}, status_code=400)
case_doc = _eligible_doc(active_case_id or "")
requested_scope = retrieval_scope if retrieval_scope in {"case", "graph", "case_plus_global", "global"} else "global"
if route in {"case_lookup", "case_question"}:
scope = "case"
elif route == "case_lineage":
scope = "graph"
else:
scope = requested_scope if requested_scope in {"global", "case_plus_global"} else "global"
frame = dict(approved_frame or {}) if approved_frame else None
if case_doc and scope == "case_plus_global":
frame = dict(frame or {})
known = list(frame.get("known_citations") or [])
case_name = str(C.meta.get(case_doc, {}).get("case_name") or "").strip()
if case_name and case_name not in known:
known.insert(0, case_name)
frame["known_citations"] = known[:4]
t0 = time.time()
def gen():
yield sse({"t": "meta", "corpus": C.coverage(), "grounding": "stored-source-only", "research_release": RESEARCH_RELEASE})
if case_doc:
card = C._card(case_doc)
yield sse({
"t": "case_context",
"route": route,
"retrieval_scope": scope,
"source": "verified_doc_id",
"case": card,
})
final, pending_more, more_sent = [], [], False
try:
if case_doc and scope == "case":
events = A.case_context_stream(
C,
re.sub(r"\s+", " ", str(case_question or q)).strip()[:1200],
case_doc,
history or [],
fast_llm_fn,
)
elif case_doc and scope == "graph":
events = A.case_lineage_stream(C, case_question or q, case_doc)
else:
events = A.structured_search_stream(
C, q, llm_fn, approved_frame=frame, identity_query=original_q
)
for ev in events:
if ev.get("t") == "_trace": # stage-level instrumentation -> log only
_log("trace", {**_query_log_fields(q), "stage": ev.get("stage"), "data": ev.get("data")})
continue
if ev.get("t") == "results":
final = _eligible_results(ev.get("results"))
pending_more = final[primary_limit:]
ev = {**ev, "results": final[:primary_limit]}
elif ev.get("t") == "more_results":
combined = _eligible_results(pending_more + list(ev.get("results") or []))
pending_more = []
more_sent = True
ev = {**ev, "results": combined[:more_limit]}
if not ev["results"]:
continue
elif ev.get("t") == "done" and pending_more and not more_sent:
yield sse({"t": "more_results", "results": pending_more[:more_limit]})
pending_more = []
yield sse(ev)
except Exception as e:
yield sse({"t": "error", "message": str(e)[:200]}); yield sse({"t": "done"})
_log("searches", {**_query_log_fields(q), "latency_s": round(time.time() - t0, 1),
"result_ids": [c.get("doc_id") for c in final[:20]]})
query_log = _query_log_fields(q)
print(
f"[agent] query_sha256={query_log['query_sha256']} "
f"query_chars={query_log['query_chars']} {time.time()-t0:.1f}s",
flush=True,
)
return StreamingResponse(gen(), media_type="text/event-stream",
headers={"Cache-Control": "no-cache", "X-Accel-Buffering": "no", "Connection": "keep-alive"})
@app.get("/api/search_stream")
def search_stream(q: str, request: Request):
return _search_response(q)
@app.post("/api/v2/search_stream")
def search_stream_v2(req: SearchRequest, request: Request):
direct_case = req.route in {"case_lookup", "case_question", "case_lineage"}
if not req.approved and not direct_case:
return JSONResponse({"error": "query understanding must be approved"}, status_code=409)
if direct_case and not _eligible_doc(req.active_case_id or ""):
return JSONResponse({"error": "selected judgment is required"}, status_code=409)
return _search_response(
req.q,
original_q=req.original_q,
approved_frame=req.search_frame,
route=req.route,
retrieval_scope=req.retrieval_scope,
active_case_id=req.active_case_id,
case_question=req.case_question,
history=[
turn.model_dump() if hasattr(turn, "model_dump") else turn.dict()
for turn in req.history[-6:]
],
)
def _graph_card(t, src_dst):
tm = C.meta.get(t, {})
return graph_node_card(
t,
tm,
treatment=C.edge_meta.get(src_dst, {}).get("treatment"),
cited_by=C.cite_indeg.get(t, 0),
good_law_status=C.goodlaw.get(t, {}).get("good_law_status", "unknown"),
)
@app.get("/api/deep_search_stream")
def deep_search_stream(q: str, request: Request):
# the agent IS the unified deep pipeline — alias so the frontend's 'deep' toggle never 404s
return _search_response(q)
def _eligible_doc(value: str):
d = value if value in C.meta else nc2doc.get(value)
return d if d and C.is_retrieval_eligible(d) else None
def _pdf_aliases(metadata: dict) -> list[str]:
return [
value
for value in [
metadata.get("neutral_citation"),
*(metadata.get("equivalent_citations") or []),
]
if value
]
@app.get("/api/v2/judgment")
@app.get("/api/judgment")
def judgment(id: str, q: str = ""):
d = _eligible_doc(id)
if not d: return JSONResponse({"error": "not found"}, status_code=404)
m = C.meta.get(d, {}); jv = C.judgment_view(d)
jv["judgment_id"] = str(d)
jv["bench"] = m.get("bench"); jv["author_judge"] = m.get("author_judge"); jv["acts"] = m.get("acts")
jv["case_number"] = m.get("case_number"); jv["year"] = m.get("year")
jv["text_raw"] = jv.get("text", "")
aliases = _pdf_aliases(m)
pdf_status = PDF_SOURCES.probe(d, aliases=aliases)
pdf_public = pdf_status.public_dict()
# A mapped individual object is the fast path. Bharat Courts can still
# resolve the same public archive by year and identity if that map misses.
pdf_public["fallback_available"] = bool(
pdf_public.get("fallback_available")
or (m.get("year") and (m.get("neutral_citation") or m.get("case_name")))
)
pdf_public["fallback_provider"] = "bharat_courts"
pdf_public["route"] = f"/api/v2/pdf?id={d}"
jv["pdf"] = pdf_public
jv["has_pdf"] = pdf_status.verified # compatibility with cached/older frontends
text_provider = m.get("source_provider") or m.get("provider") or "Supreme Court Reports open registry"
jv["grounding"] = {
"text_available": bool((jv.get("text") or "").strip()),
"retrieval_eligible": True,
"text_origin": f"judgment text extracted from {text_provider}",
"pdf_status": pdf_status.status,
"source_name": text_provider,
"source_url": m.get("source_url"),
}
clean_query = re.sub(r"\s+", " ", q or "").strip()[:4000]
if clean_query:
if hasattr(C, "relevant_passages"):
highlights = C.relevant_passages(clean_query, d, k=6)
else:
highlights = C.case_chat_passages(clean_query, d, k=6)
jv["relevance_highlights"] = highlights
jv["highlighting"] = {
"query_specific": True,
"method": "case-local semantic retrieval resolved to stored paragraphs",
"grounding": "stored_paragraph_ids_only",
}
else:
jv["relevance_highlights"] = []
jv["highlighting"] = {"query_specific": False, "grounding": "stored_paragraph_ids_only"}
jv["corpus_notice"] = (
f"Searched {C.coverage()['accepted_judgments']:,} accepted Supreme Court judgments. "
"Unavailable or unmapped judgments were not evaluated."
)
if not pdf_status.verified:
_log("pdf_sources", {"doc_id": d, "status": pdf_status.status, "reason": pdf_status.reason})
# CITATOR from the citation GRAPH (meta.cases_cited is only ~3% populated): note-up + note-down
jv["cited_cases"] = [
_graph_card(t, (d, t))
for t in list(dict.fromkeys(C.out_edges.get(d, [])))
if C.is_retrieval_eligible(t)
][:20]
jv["citing_cases"] = [
_graph_card(s, (s, d))
for s in sorted(set(C.in_edges.get(d, [])), key=lambda x: -C.cite_indeg.get(x, 0))
if C.is_retrieval_eligible(s)
][:20]
if not jv["cited_cases"]: # fallback to the sparse metadata if the graph has nothing
jv["cited_cases"] = resolve_cited(m.get("cases_cited"), d)
jv["links"] = doc_links(jv.get("text"), d)
return JSONResponse(jv)
@app.get("/api/v2/judgment/{judgment_id}/paragraphs")
def judgment_paragraphs(judgment_id: str, offset: int = 0, limit: int = 50):
d = _eligible_doc(judgment_id)
if not d:
return JSONResponse({"error": "judgment not found"}, status_code=404)
limit = max(1, min(int(limit), 100))
offset = max(0, int(offset))
if hasattr(C, "judgment_paragraphs"):
return JSONResponse(C.judgment_paragraphs(d, offset=offset, limit=limit))
cis = C.doc_chunks.get(d, [])
rows = [
{
"paragraph_id": f"{d}:chunk:{ci}",
"label": f"Indexed passage {position + 1}",
"sequence": position + 1,
"text": C.texts[ci],
"html_anchor": f"paragraph-{d}-chunk-{ci}",
"source_kind": "legacy_chunk",
}
for position, ci in enumerate(cis[offset:offset + limit], start=offset)
if str(C.texts[ci]).strip()
]
return JSONResponse({
"judgment_id": str(d),
"paragraphs": rows,
"offset": offset,
"limit": limit,
"total": len(cis),
"next_offset": offset + len(rows) if offset + len(rows) < len(cis) else None,
})
@app.get("/api/v2/graph")
def graph(id: str, direction: str = "both", limit: int = 50):
d = _eligible_doc(id)
if not d:
return JSONResponse({"error": "judgment not found"}, status_code=404)
direction = direction if direction in {"incoming", "outgoing", "both"} else "both"
limit = max(1, min(int(limit), 100))
root = graph_node_card(
d,
C.meta.get(d, {}),
cited_by=C.cite_indeg.get(d, 0),
good_law_status=C.goodlaw.get(d, {}).get("good_law_status", "unknown"),
)
nodes = {d: root}
edges = []
if direction in {"outgoing", "both"}:
for target in list(dict.fromkeys(C.out_edges.get(d, []))):
if len(edges) >= limit or not C.is_retrieval_eligible(target):
continue
card = _graph_card(target, (d, target)); nodes[target] = card
edge = C.edge_meta.get((d, target), {})
edges.append({
"source_id": str(d), "target_id": str(target),
"relation": edge.get("treatment") or "referred_to",
"scope": edge.get("scope") or "unknown",
"confidence": edge.get("confidence"),
"direction": "outgoing", "evidence": edge.get("evidence") or [],
})
if direction in {"incoming", "both"}:
for source in sorted(set(C.in_edges.get(d, [])), key=lambda x: -C.cite_indeg.get(x, 0)):
if len(edges) >= limit or not C.is_retrieval_eligible(source):
continue
card = _graph_card(source, (source, d)); nodes[source] = card
edge = C.edge_meta.get((source, d), {})
edges.append({
"source_id": str(source), "target_id": str(d),
"relation": edge.get("treatment") or "referred_to",
"scope": edge.get("scope") or "unknown",
"confidence": edge.get("confidence"),
"direction": "incoming", "evidence": edge.get("evidence") or [],
})
return JSONResponse({
"judgment_id": str(d), "root": root,
"nodes": list(nodes.values()), "edges": edges,
"unresolved_edges_hidden": True,
})
@app.post("/api/v2/judgment_chat")
@app.post("/api/judgment_chat")
def judgment_chat(req: CaseChatRequest):
d = _eligible_doc(req.doc_id)
if not d:
return JSONResponse({"error": "judgment not found"}, status_code=404)
question = re.sub(r"\s+", " ", req.question or "").strip()
if not question:
return JSONResponse({"error": "question is required"}, status_code=400)
jv = C.judgment_view(d)
summary = jv.get("summary") or {}
if not summary.get("available") or not summary.get("text"):
return JSONResponse(
{"error": "case summary unavailable", "code": "summary_unavailable"},
status_code=409,
)
passages = C.case_chat_passages(question, d, k=5)
response = A.case_chat_grounded_response(
summary["text"],
passages,
question,
[turn.dict() for turn in req.history],
jv.get("case_name"),
jv.get("neutral_citation"),
fast_llm_fn,
)
if not response.get("answer"):
return JSONResponse({"error": "case chat unavailable"}, status_code=502)
_log("judgment_chats", {"doc_id": d, **_query_log_fields(question), "summary_source": summary.get("source")})
return JSONResponse({
"doc_id": d,
"judgment_id": str(d),
"answer": response["answer"],
"evidence": response.get("evidence") or [],
"supported": bool(response.get("supported")),
"grounded_in": "case_summary_and_stored_passages",
"summary_source": summary.get("source"),
})
@app.get("/api/v2/pdf/status")
@app.get("/api/pdf/status")
def pdf_status(id: str, refresh: int = 0):
d = _eligible_doc(id)
if not d: return JSONResponse({"error": "not found"}, status_code=404)
m = C.meta.get(d, {})
status = PDF_SOURCES.probe(d, aliases=_pdf_aliases(m), force=bool(refresh))
public = status.public_dict()
public["fallback_available"] = bool(
public.get("fallback_available")
or (m.get("year") and (m.get("neutral_citation") or m.get("case_name")))
)
public["fallback_provider"] = "bharat_courts"
return JSONResponse({"doc_id": d, "judgment_id": str(d), **public})
@app.get("/api/v2/pdf")
@app.get("/api/pdf")
async def pdf(id: str, dl: int = 0):
d = _eligible_doc(id)
if not d: return JSONResponse({"error": "not found"}, status_code=404)
m = C.meta.get(d, {})
aliases = _pdf_aliases(m)
status = PDF_SOURCES.probe(d, aliases=aliases)
if status.verified:
# Redirect instead of downloading the complete document into the Space.
# The public object supports byte ranges used by native PDF viewers.
return RedirectResponse(
status.url,
status_code=307,
headers={"Cache-Control": "private, max-age=3600", "X-PDF-Provider": "aws_open_data"},
)
archive = PDF_SOURCES.archive_candidate(d, aliases)
try:
data, provenance = await resolve_and_fetch_pdf(
year=(archive or {}).get("year") or m.get("year"),
path=(archive or {}).get("path"),
case_name=m.get("case_name") or "",
neutral_citation=m.get("neutral_citation") or "",
equivalent_citations=m.get("equivalent_citations") or [],
decision_date=m.get("date") or "",
)
except BharatCourtsPdfError as exc:
_log("pdf_sources", {"doc_id": d, "status": "bharat_courts_unavailable", "reason": str(exc)[:200]})
return JSONResponse(
{
"error": "pdf unavailable",
"reason": str(exc)[:300],
"pdf_status": "bharat_courts_unavailable",
"official_search_url": status.public_dict()["official_search_url"],
},
status_code=503 if status.status == "temporarily_unavailable" else 422,
)
citation = re.sub(r"[^A-Za-z0-9._-]+", "-", m.get("neutral_citation") or "judgment").strip("-")
disposition = "attachment" if dl else "inline"
_log("pdf_sources", {"doc_id": d, "status": "verified", "provider": provenance["provider"]})
return Response(
content=data,
media_type="application/pdf",
headers={
"Cache-Control": "private, max-age=86400",
"Content-Disposition": f'{disposition}; filename="{citation or "judgment"}.pdf"',
"X-PDF-Provider": "bharat_courts",
},
)
@app.get("/api/v2/health")
def health():
return JSONResponse({
"service": "Moonley API",
"status": "ok",
"auth_configured": clerk_settings().configured,
"api_version": "v2-grounded-preview",
"runtime": RUNTIME_KIND,
"warm": RUNTIME_WARM,
"corpus": C.coverage(),
"research_release": RESEARCH_RELEASE,
"device": C.device,
"pdf_sources": {
"mapped_identities": PDF_SOURCES.mapped_count,
"primary": "aws_open_data",
"fallback": "bharat_courts",
},
"project_storage": PROJECTS.status(),
"knowledge": KNOWLEDGE.status(),
"statute_crosswalk": {
"loaded": True,
"indexed_directions": C.crosswalk.mapping_count,
},
"statute_library": C.statute_library.status(),
"drafting_templates": len(DRAFTING.list()),
}, headers={"Cache-Control": "no-store"})
@app.get("/api/v2/ready")
def ready():
return JSONResponse({
"ready": bool(C.eligible_doc_ids) and RUNTIME_WARM,
"api_version": "v2-grounded-preview",
"runtime": RUNTIME_KIND,
"accepted_judgments": len(C.eligible_doc_ids),
"research_release": RESEARCH_RELEASE,
}, headers={"Cache-Control": "no-store"})
@app.get("/")
def home():
return JSONResponse(
{"service": "Moonley API", "status": "ok", "ui": "https://moonley-pilot.vercel.app"},
headers={"Cache-Control": "no-store"},
)
print(f"[serve_agent] boot configured runtime={RUNTIME_KIND}", flush=True)
|