File size: 28,302 Bytes
c0ff3c7
 
 
06b0521
 
 
 
 
 
c0ff3c7
 
 
 
 
 
 
ba4d952
c0ff3c7
ba4d952
c0ff3c7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
06b0521
c0ff3c7
 
 
 
06b0521
 
c0ff3c7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
06b0521
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c0ff3c7
 
 
06b0521
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c0ff3c7
 
06b0521
c0ff3c7
06b0521
c0ff3c7
 
06b0521
c0ff3c7
06b0521
 
c0ff3c7
 
fdd7b3e
 
 
 
 
 
 
c0ff3c7
06b0521
 
 
c0ff3c7
 
 
 
 
 
 
 
 
 
 
fdd7b3e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c0ff3c7
fdd7b3e
c0ff3c7
fdd7b3e
 
 
c0ff3c7
 
 
6f94d69
 
 
fdd7b3e
6f94d69
c0ff3c7
fdd7b3e
 
 
c0ff3c7
06b0521
 
fdd7b3e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
06b0521
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fdd7b3e
 
 
 
 
 
 
 
06b0521
fdd7b3e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
06b0521
 
fdd7b3e
 
 
 
 
06b0521
 
fdd7b3e
 
 
 
 
 
 
 
 
06b0521
 
 
fdd7b3e
06b0521
 
 
 
fdd7b3e
06b0521
 
 
 
 
 
 
 
 
 
 
 
fdd7b3e
 
06b0521
 
 
 
 
fdd7b3e
 
 
 
 
 
 
 
06b0521
fdd7b3e
c0ff3c7
 
 
 
 
 
 
ba4d952
 
 
06b0521
ba4d952
 
 
 
 
 
 
 
 
c0ff3c7
 
06b0521
c0ff3c7
 
ba4d952
c0ff3c7
ba4d952
 
06b0521
ba4d952
 
06b0521
ba4d952
06b0521
ba4d952
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6f94d69
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c0ff3c7
 
 
 
 
 
 
 
 
3a227e6
c0ff3c7
 
3a227e6
 
 
 
 
 
 
06b0521
 
 
 
c0ff3c7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
06b0521
 
 
c0ff3c7
06b0521
c0ff3c7
 
06b0521
 
 
 
c0ff3c7
 
 
 
 
 
 
 
 
 
 
e29fc76
c0ff3c7
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
#!/usr/bin/env python3
"""CanLex Web (Path B) -- a private web front-end for CanLex legal research.

A thin client that gives a non-Claude user roughly the same experience as Claude
with the CanLex MCP server. For each question it opens one streamable-HTTP
session against the deployed CanLex MCP, declares the four CanLex tools to
Google Gemini, and lets the model agentically iterate -- searching, fetching
sections, and looking up case-law citations -- until it decides it has enough
material to compose a grounded answer.

All configuration comes from environment variables, set as Hugging Face Space
secrets. Run locally with:  python app.py
"""
import asyncio
import json
import os
import queue
import sys
import threading
import urllib.error
import urllib.request
from datetime import timedelta

import gradio as gr
from mcp import ClientSession
from mcp.client.streamable_http import streamablehttp_client


# --- Configuration (Hugging Face Space secrets / environment variables) -------

# The deployed CanLex MCP server. Retrieval logic and the corpus live there; this
# app never carries its own copy. Override only to point at a different server.
MCP_URL = os.environ.get(
    "CANLEX_MCP_URL", "https://beemer0-canlex.hf.space/mcp").strip()

# Google Gemini -- the free-tier key is supplied as the GEMINI_API_KEY secret.
GEMINI_MODEL = os.environ.get("CANLEX_GEMINI_MODEL", "gemini-2.5-pro").strip()
GEMINI_ENDPOINT = ("https://generativelanguage.googleapis.com/v1beta/models/"
                   f"{GEMINI_MODEL}:generateContent")

MAX_OUTPUT_TOKENS = 8192   # generous -- covers Gemini 2.5 thinking plus the answer
MAX_TOOL_ITERATIONS = 8    # loop guard for the agent
REQUEST_TIMEOUT = 180      # seconds, applied to the MCP and Gemini calls


def _load_auth() -> list[tuple[str, str]]:
    """Parse CANLEX_WEB_AUTH (one 'username:password' per line) for Gradio auth."""
    creds: list[tuple[str, str]] = []
    for line in os.environ.get("CANLEX_WEB_AUTH", "").splitlines():
        line = line.strip()
        if not line or ":" not in line:
            continue
        user, password = (p.strip() for p in line.split(":", 1))
        if user and password:
            creds.append((user, password))
    if not creds:
        print("WARNING: CANLEX_WEB_AUTH is not set; using the insecure default "
              "login 'canlex' / 'canlex'. Set CANLEX_WEB_AUTH as a Space secret "
              "(one 'username:password' per line) before sharing this app.",
              file=sys.stderr)
        creds = [("canlex", "canlex")]
    return creds


AUTH = _load_auth()


# --- Tool declarations (Gemini function-calling schema) -----------------------

# The four CanLex MCP tools, declared so Gemini can call them. Three of the four
# wrap their arguments inside a single 'params' object on the server side; the
# Gemini schema is kept flat for the model's convenience and re-wrapped at the
# MCP edge in _run_tool.
TOOL_DECLARATIONS = [
    {
        "name": "canlex_search_legislation",
        "description": (
            "Search Canadian federal law, CBSA D-Memoranda, collective "
            "agreements, NJC directives, leading court decisions and IRPA "
            "delegation instruments for material relevant to a question. "
            "Use this first for any topical question. Returns ranked source "
            "passages with citations. Call it multiple times for different "
            "facets of a question, or with the optional 'act' or 'doc_type' "
            "filters to narrow the search."
        ),
        "parameters": {
            "type": "object",
            "properties": {
                "query": {
                    "type": "string",
                    "description": (
                        "Natural-language legal question or keywords, e.g. "
                        "'detention review timelines' or 'innocent "
                        "misrepresentation defence under IRPA s. 40'."
                    ),
                },
                "top_k": {
                    "type": "integer",
                    "description": "Number of sections to return (1-20). Default 6.",
                },
                "act": {
                    "type": "string",
                    "description": (
                        "Optional. Restrict to a single Act, by short name or "
                        "code (e.g. 'IRPA' or 'I-2.5')."
                    ),
                },
                "doc_type": {
                    "type": "string",
                    "description": (
                        "Optional. Restrict to one source type: 'legislation', "
                        "'memorandum' (CBSA D-Memoranda), 'agreement' "
                        "(collective agreements), 'directive' (NJC), "
                        "'caselaw' (court and tribunal decisions), or "
                        "'delegation' (IRPA/IRPR delegation and designation)."
                    ),
                },
            },
            "required": ["query"],
        },
    },
    {
        "name": "canlex_get_section",
        "description": (
            "Fetch the full text of one specific provision when its Act and "
            "section number are known. Use this to follow a cross-reference "
            "the search results mention but did not include."
        ),
        "parameters": {
            "type": "object",
            "properties": {
                "act": {
                    "type": "string",
                    "description": "Act short name or code, e.g. 'IRPA' or 'I-2.5'.",
                },
                "section": {
                    "type": "string",
                    "description": "Section number exactly as cited, e.g. '34', '20.1'.",
                },
            },
            "required": ["act", "section"],
        },
    },
    {
        "name": "canlex_list_acts",
        "description": (
            "List every Act, regulation, agreement, directive, case-law "
            "decision and delegation instrument loaded into the CanLex "
            "corpus. Useful when the user asks 'what does CanLex have on X?' "
            "or when you need to confirm a source is in scope."
        ),
        "parameters": {"type": "object", "properties": {}},
    },
    {
        "name": "canlex_case",
        "description": (
            "Look up a Canadian case on CanLII to check its citation graph -- "
            "cases that cite it, cases it cites, legislation it cites. Use "
            "this to confirm a decision found in search results is still good "
            "law and to find related authorities. Pass a neutral citation "
            "(e.g. '2019 SCC 65', '2016 FCA 93', '2005 FC 1059') or a full "
            "canlii.org URL."
        ),
        "parameters": {
            "type": "object",
            "properties": {
                "case_url": {
                    "type": "string",
                    "description": (
                        "Neutral citation (preferred for SCC/FCA/FC) or full "
                        "canlii.org URL."
                    ),
                },
            },
            "required": ["case_url"],
        },
    },
]

# The three tools whose MCP signature wraps arguments under a single 'params'
# object. canlex_list_acts takes none and is handled separately in _run_tool.
_PARAMS_WRAPPED = {"canlex_search_legislation", "canlex_get_section", "canlex_case"}


# --- System prompt ------------------------------------------------------------

SYSTEM_INSTRUCTION = """\
You are CanLex Web, a Canadian legal-research assistant. A member of the public \
has asked the legal question shown below through a web form. Answer it by \
agentically using the four CanLex tools to retrieve primary sources, then \
compose a clear, well-organised answer grounded entirely in what those tools \
return.

Tool-use guidance:
- Start with canlex_search_legislation on the user's question. Read the \
results, including the "ANSWERING INSTRUCTIONS" block CanLex returns.
- If a result mentions a cross-referenced provision, regulation or D-Memo that \
bears on the question but is not reproduced, call canlex_get_section or \
canlex_search_legislation again to fetch it. Do not guess its contents.
- For a question that turns on case law, consider calling canlex_case on the \
leading decision's neutral citation to confirm it has not been overtaken.
- You may call tools multiple times; iterate until you have enough material to \
answer well. Aim for thoroughness but stop once further calls would not change \
the answer.

Answering style:
- Write for a reader who cannot see the raw passages. Quote the key operative \
words ("inadmissible for misrepresentation", etc.) and give every citation in \
full, including section numbers and the deciding court.
- Distinguish source kinds: enacted law is binding; CBSA D-Memoranda are \
administrative guidance, persuasive only; collective agreements and NJC \
directives are binding employment-terms instruments for a bargaining unit; \
court decisions are binding precedent depending on the court and jurisdiction.
- State the date the source is current to, and note that the answer reflects \
the law only as of that date.
- Use plain Markdown -- short paragraphs, headings or lists where they aid \
clarity.
- If the retrieved material does not actually answer the question, say so \
plainly rather than stretching it to fit.
- Close with a one-line reminder that this is legal information, not legal \
advice."""


# --- Agentic loop: Gemini <-> MCP --------------------------------------------

class _AgentError(RuntimeError):
    """Surfaced to the UI; the message text is shown verbatim."""


def _gemini_request_body(contents: list[dict]) -> dict:
    """The JSON body sent to Gemini -- identical between the streaming and the
    non-streaming endpoints. Tool declarations turn on function calling; the
    safety filters are relaxed because legal research routinely discusses
    crime, weapons and the like, and the high-threshold defaults spuriously
    block legitimate legal text."""
    return {
        "systemInstruction": {"parts": [{"text": SYSTEM_INSTRUCTION}]},
        "contents": contents,
        "tools": [{"functionDeclarations": TOOL_DECLARATIONS}],
        "toolConfig": {"functionCallingConfig": {"mode": "AUTO"}},
        "generationConfig": {
            "temperature": 0.2,
            "maxOutputTokens": MAX_OUTPUT_TOKENS,
        },
        "safetySettings": [
            {"category": c, "threshold": "BLOCK_ONLY_HIGH"}
            for c in ("HARM_CATEGORY_HARASSMENT", "HARM_CATEGORY_HATE_SPEECH",
                      "HARM_CATEGORY_SEXUALLY_EXPLICIT",
                      "HARM_CATEGORY_DANGEROUS_CONTENT")
        ],
    }


# Gemini's streamGenerateContent endpoint, used when alt=sse is requested,
# sends one Server-Sent Event per partial GenerateContentResponse. Each chunk
# carries an incremental slice of the turn's content -- a text delta or a
# (complete) functionCall part. The accumulated parts list is what gets sent
# back as the assistant turn for the next round.
_STREAM_ENDPOINT = GEMINI_ENDPOINT.replace(
    ":generateContent", ":streamGenerateContent") + "?alt=sse"


async def _gemini_stream(api_key: str, contents: list[dict]):
    """Async generator over Gemini's streaming response.

    Yields dicts of one of three shapes:
      {"type": "text_delta", "text": str}        -- a partial answer fragment
      {"type": "function_call", "call": dict}    -- a complete tool call
      {"type": "finish", "reason": str|None,     -- end of stream; `parts` is
       "parts": list[dict]}                          the whole assistant turn
    """
    body = _gemini_request_body(contents)
    request = urllib.request.Request(
        _STREAM_ENDPOINT,
        data=json.dumps(body).encode("utf-8"),
        headers={"Content-Type": "application/json",
                 "x-goog-api-key": api_key,
                 "Accept": "text/event-stream"},
        method="POST",
    )
    try:
        # `timeout` is a kwarg of urlopen; passing it positionally to
        # asyncio.to_thread would forward it as `data` (POST body) and break
        # the request.
        response = await asyncio.to_thread(
            lambda: urllib.request.urlopen(request, timeout=REQUEST_TIMEOUT))
    except urllib.error.HTTPError as exc:
        detail = await asyncio.to_thread(exc.read)
        text = detail.decode("utf-8", "replace")[:600]
        raise _AgentError(f"Gemini API returned HTTP {exc.code}: {text}") from None
    except urllib.error.URLError as exc:
        raise _AgentError(f"Could not reach the Gemini API: {exc.reason}") from None

    accumulated_parts: list[dict] = []
    finish_reason = None
    try:
        while True:
            raw = await asyncio.to_thread(response.readline)
            if not raw:
                break
            line = raw.decode("utf-8", "replace").rstrip()
            if not line.startswith("data: "):
                continue
            try:
                chunk = json.loads(line[6:])
            except ValueError:
                continue
            candidate = (chunk.get("candidates") or [{}])[0]
            for part in (candidate.get("content") or {}).get("parts") or []:
                accumulated_parts.append(part)
                if part.get("thought"):
                    # Gemini 2.5 thinking summary -- preserved in history for
                    # the model's own context, never streamed to the user.
                    continue
                if "text" in part:
                    yield {"type": "text_delta", "text": part["text"]}
                elif "functionCall" in part:
                    yield {"type": "function_call", "call": part["functionCall"]}
            if candidate.get("finishReason"):
                finish_reason = candidate["finishReason"]
    finally:
        await asyncio.to_thread(response.close)
    yield {"type": "finish", "reason": finish_reason, "parts": accumulated_parts}


async def _run_tool(session: ClientSession, name: str, args: dict) -> str:
    """Execute a Gemini function call against the MCP, returning text output."""
    if name == "canlex_list_acts":
        mcp_args: dict = {}
    elif name in _PARAMS_WRAPPED:
        # The MCP server's tools accept their schema as a single 'params' object.
        mcp_args = {"params": args or {}}
    else:
        return f"(unknown tool '{name}')"
    try:
        result = await session.call_tool(name, mcp_args)
    except Exception as exc:                        # MCP transport errors
        return f"(tool '{name}' failed: {type(exc).__name__}: {exc})"
    text = "\n".join(
        block.text for block in result.content
        if getattr(block, "type", None) == "text" and getattr(block, "text", None)
    ).strip()
    if result.isError:
        return f"(tool '{name}' reported an error: {text or 'no detail'})"
    return text or "(no content returned)"


def _summarize_call(name: str, args: dict) -> str:
    """Render a tool call as a one-line user-facing status string."""
    args = args or {}
    if name == "canlex_search_legislation":
        bits = [f"`{args.get('query', '')}`"]
        if args.get("act"):
            bits.append(f"in {args['act']}")
        if args.get("doc_type"):
            bits.append(f"({args['doc_type']} only)")
        return "Searching " + " ".join(bits)
    if name == "canlex_get_section":
        return f"Fetching {args.get('act', '?')} s. {args.get('section', '?')}"
    if name == "canlex_case":
        return f"Looking up case {args.get('case_url', '?')}"
    if name == "canlex_list_acts":
        return "Listing the CanLex corpus"
    return f"Calling {name}"


def _format_sources(tool_log: list[tuple[str, dict, str]]) -> str:
    """Render every tool call's output as one Markdown document for display."""
    if not tool_log:
        return ""
    blocks = []
    for i, (name, args, output) in enumerate(tool_log, 1):
        blocks.append(
            f"### Call {i}: `{name}`\n\n"
            f"_Arguments:_ `{json.dumps(args, ensure_ascii=False)}`\n\n"
            f"{output}"
        )
    return "\n\n---\n\n".join(blocks)


async def _agentic_answer(question: str):
    """Run the Gemini-driven agentic loop against a single MCP session.

    Yields tuples of (status, answer_md, sources_md). The final yield carries
    the composed answer; earlier yields are progress updates the UI can show.
    """
    api_key = os.environ.get("GEMINI_API_KEY", "").strip()
    if not api_key:
        raise _AgentError(
            "GEMINI_API_KEY is not set. Add it as a Space secret -- create a "
            "free key at Google AI Studio (https://aistudio.google.com/apikey).")

    yield "_Connecting to the CanLex retrieval service..._", "", ""

    async with streamablehttp_client(
        MCP_URL,
        timeout=timedelta(seconds=REQUEST_TIMEOUT),
        sse_read_timeout=timedelta(seconds=REQUEST_TIMEOUT),
    ) as (read, write, _):
        async with ClientSession(read, write) as session:
            await session.initialize()

            contents: list[dict] = [
                {"role": "user", "parts": [{"text": question}]}
            ]
            tool_log: list[tuple[str, dict, str]] = []
            trace: list[str] = []

            answer_buf = ""

            def status_md(thinking: bool = True) -> str:
                lines = [f"- {line}" for line in trace]
                if thinking:
                    lines.append("- _Thinking..._")
                return "\n".join(lines) if lines else ""

            for step in range(MAX_TOOL_ITERATIONS):
                yield status_md(), answer_buf, _format_sources(tool_log)

                # Stream Gemini's next turn. Stream text deltas to the answer
                # panel optimistically; revert to the pre-turn answer if it
                # turns out to be a tool-calling turn (the streamed text was
                # then commentary, kept in the trace instead).
                turn_text = ""
                turn_calls: list[dict] = []
                turn_parts: list[dict] = []
                optimistic = True

                async for chunk in _gemini_stream(api_key, contents):
                    if chunk["type"] == "text_delta" and optimistic:
                        turn_text += chunk["text"]
                        yield (status_md(),
                               answer_buf + turn_text,
                               _format_sources(tool_log))
                    elif chunk["type"] == "function_call":
                        turn_calls.append(chunk["call"])
                        if optimistic and turn_text:
                            # Roll the answer panel back; the commentary moves
                            # into the trace once the tool labels are drawn.
                            optimistic = False
                            yield (status_md(),
                                   answer_buf,
                                   _format_sources(tool_log))
                    elif chunk["type"] == "finish":
                        turn_parts = chunk["parts"] or []
                        # Capture any text-only finish reason so the caller can
                        # surface a useful error for an empty answer.
                        finish_reason = chunk.get("reason")

                contents.append({"role": "model", "parts": turn_parts})

                if not turn_calls:
                    # Final turn -- the text was already streamed; finalize.
                    if not turn_text:
                        raise _AgentError(
                            f"Gemini produced an empty answer (finishReason: "
                            f"{finish_reason!s}). If this is MAX_TOKENS, "
                            "raise MAX_OUTPUT_TOKENS in app.py.")
                    answer_buf += turn_text
                    yield status_md(thinking=False), answer_buf, \
                        _format_sources(tool_log)
                    return

                # Tool turn. If the model emitted a commentary fragment before
                # its function calls, surface it once in the trace -- it often
                # explains WHY the next tools are being called.
                if turn_text:
                    snippet = turn_text.strip().replace("\n", " ")
                    if len(snippet) > 140:
                        snippet = snippet[:137].rstrip() + "..."
                    trace.append(f"_{snippet}_")

                # Execute every function call in this turn, then send the
                # responses back as a single 'user' message.
                function_responses = []
                for call in turn_calls:
                    name = call.get("name", "")
                    args = call.get("args") or {}
                    label = _summarize_call(name, args)
                    trace.append(label)
                    yield status_md(), answer_buf, _format_sources(tool_log)

                    output = await _run_tool(session, name, args)
                    tool_log.append((name, args, output))
                    function_responses.append({
                        "functionResponse": {
                            "name": name,
                            "response": {"output": output},
                        }
                    })
                contents.append({"role": "user", "parts": function_responses})

            # Loop budget exhausted -- ask Gemini for a final answer without
            # further tool use rather than leave the user with nothing. We
            # stream this terminal turn too, so the user sees it compose.
            contents.append({"role": "user", "parts": [{"text":
                "You have reached the maximum number of tool calls. Compose "
                "the best answer you can from the material gathered so far, "
                "without calling further tools. If the material is "
                "insufficient, say so plainly."}]})
            turn_text = ""
            async for chunk in _gemini_stream(api_key, contents):
                if chunk["type"] == "text_delta":
                    turn_text += chunk["text"]
                    yield (status_md(thinking=False),
                           answer_buf + turn_text,
                           _format_sources(tool_log))
            answer_buf += turn_text or \
                "_(no answer produced after the tool-call budget was exhausted)_"
            yield status_md(thinking=False), answer_buf, _format_sources(tool_log)


# --- Gradio handler -----------------------------------------------------------

ANSWER_PLACEHOLDER = "*Your answer will appear here.*"


_SENTINEL = object()


def answer(question: str):
    """Generator wrapping the async agent for Gradio's progressive UI.

    The async work runs on a dedicated worker thread with its own event loop
    and stays inside a single asyncio task for the whole question. Items are
    handed back to this sync generator through a thread-safe queue. The
    previous loop.run_until_complete-per-anext pattern created a fresh task
    on every yield, which tripped anyio's cancel-scope check inside the MCP
    streamable-HTTP client ('Attempted to exit cancel scope in a different
    task than it was entered in')."""
    question = (question or "").strip()
    if not question:
        yield "Please enter a legal question above.", ANSWER_PLACEHOLDER, ""
        return

    events: queue.Queue = queue.Queue()

    def worker():
        async def run():
            try:
                async for tup in _agentic_answer(question):
                    events.put(("yield", tup))
            except _AgentError as exc:
                events.put(("agent_error", exc))
            except Exception as exc:                # network blip, MCP transport
                events.put(("error", exc))
            finally:
                events.put((_SENTINEL,))
        try:
            asyncio.run(run())
        except Exception as exc:                    # loop setup failures
            events.put(("error", exc))
            events.put((_SENTINEL,))

    threading.Thread(target=worker, daemon=True).start()

    while True:
        kind, *payload = events.get()
        if kind is _SENTINEL:
            return
        if kind == "yield":
            yield payload[0]
        elif kind == "agent_error":
            yield (f"**{payload[0]}**", ANSWER_PLACEHOLDER, "")
        elif kind == "error":
            exc = payload[0]
            # Unwrap ExceptionGroup (from anyio TaskGroups in the MCP client)
            # so the user sees the actual root cause, not the wrapper.
            lines = []
            def _walk(e, depth=0):
                indent = "  " * depth
                lines.append(f"{indent}- `{type(e).__name__}: {e}`")
                inner = getattr(e, "exceptions", None)
                if inner:
                    for sub in inner:
                        _walk(sub, depth + 1)
            _walk(exc)
            yield ("**Could not complete the request.**\n\n"
                   + "\n".join(lines) +
                   "\n\nThe MCP service may be waking from sleep -- "
                   "try again in a moment.", ANSWER_PLACEHOLDER, "")


# --- UI -----------------------------------------------------------------------

INTRO = """\
# CanLex -- Canadian Legal Research

Ask a question about Canadian **border, customs, immigration, criminal, drug,
labour or related federal law**. CanLex finds the governing statutory
provisions, D-Memoranda, collective-agreement terms and leading court
decisions, then composes an answer that cites them.

The CanLex corpus contains 31 federal Acts and regulations -- including the
Immigration and Refugee Protection Act, the Customs Act and the Criminal Code
-- alongside the CBSA D-Memoranda, the FB (Border Services) collective
agreement, the National Joint Council directives, leading decisions of the
Supreme Court, the Federal Courts and the federal labour and immigration
tribunals, and the IRPA/IRPR instruments of delegation and designation.

The assistant iterates over the corpus -- searching, fetching sections and
looking up case-law citations -- before composing a grounded answer. A complex
question may take 30 seconds or more.

Legal information, not legal advice -- always verify against the primary sources.
"""

EXAMPLE_QUESTIONS = [
    "What are the detention review timelines for a permanent resident?",
    "When is a foreign national inadmissible for serious criminality?",
    "What overtime provisions apply to FB-group Border Services officers?",
    "Can the CBSA seize goods for an undervalued customs declaration?",
]

with gr.Blocks(title="CanLex", analytics_enabled=False) as demo:
    gr.Markdown(INTRO)

    question = gr.Textbox(
        label="Your legal question",
        placeholder="e.g. What are the detention review timelines for a "
                    "permanent resident?",
        lines=3,
    )
    with gr.Row():
        submit = gr.Button("Ask CanLex", variant="primary")
        clear = gr.Button("Clear")

    gr.Examples(examples=EXAMPLE_QUESTIONS, inputs=question, label="Examples")

    # Three panels: a progress trace (also used to surface errors), the final
    # composed answer, and the raw tool outputs the agent gathered.
    progress_md = gr.Markdown(value="")
    answer_md = gr.Markdown(value=ANSWER_PLACEHOLDER)
    with gr.Accordion("Retrieved source passages (every tool call)", open=False):
        sources_md = gr.Markdown()

    submit.click(answer, [question], [progress_md, answer_md, sources_md])
    question.submit(answer, [question], [progress_md, answer_md, sources_md])
    clear.click(lambda: ("", "", ANSWER_PLACEHOLDER, ""), None,
                [question, progress_md, answer_md, sources_md])


if __name__ == "__main__":
    print(f"CanLex Web starting -- MCP: {MCP_URL}; model: {GEMINI_MODEL}; "
          f"{len(AUTH)} login(s) configured.", file=sys.stderr)
    demo.queue()
    demo.launch(
        server_name="0.0.0.0",
        server_port=int(os.environ.get("PORT", "7860")),
        auth=AUTH,
        auth_message="Sign in to use CanLex.",
        ssr_mode=False,   # no Node in the slim container; render client-side
    )