Spaces:
Sleeping
Sleeping
File size: 22,513 Bytes
6ca8e48 | 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 | """
/api/context routes β the core DocForge pipeline endpoint.
POST /api/context β queues a job, returns job_id immediately
GET /api/context/{job_id} β returns job status and result when done
"""
import logging
import time
import uuid
from datetime import date
from urllib.parse import urlparse
from fastapi import APIRouter, BackgroundTasks, HTTPException, Request
from ..models.schemas import (
ContextJobResult,
ContextRequest,
InputType,
JobCreatedResponse,
JobStatus,
OutputFormat,
)
from ..services import cache
from ..services.formatter import format_context_md
from ..services.ingestion.npm_ingester import (
resolve_npm_package,
resolve_pypi_package,
)
from ..services.ingestion.paste_ingester import ingest_paste
from ..services.ingestion.registries_ingester import (
resolve_crates,
resolve_cran,
resolve_hex,
resolve_maven,
resolve_nuget,
resolve_pubdev,
resolve_rubygems,
)
from ..services.ingestion.search_ingester import find_docs_url
from ..services.ingestion.url_ingester import crawl_url
from ..services.structurer import structure_with_gemini
logger = logging.getLogger(__name__)
router = APIRouter(prefix="/api/context", tags=["context"])
@router.post("", response_model=JobCreatedResponse)
async def create_context_job(
request: ContextRequest,
background_tasks: BackgroundTasks,
http_request: Request,
) -> JobCreatedResponse:
"""
Queue a documentation context generation job.
Returns immediately with a job_id for polling.
If a valid Bearer token is provided the user's stored API keys are used for
the AI step; otherwise falls back to env-var keys (dev / open access).
"""
# ββ Auth: prefer stored keys from Bearer token, fall back to request body β
gemini_key: str | None = None
groq_key: str | None = None
auth_header = http_request.headers.get("authorization", "")
if auth_header.startswith("Bearer "):
from ..services.auth_service import decode_token, get_api_keys
payload = decode_token(auth_header.split(" ", 1)[1])
if payload:
keys = await get_api_keys(payload["sub"])
gemini_key = keys.get("gemini_key")
groq_key = keys.get("groq_key")
# CLI local-key mode: keys sent directly in the request body
if not gemini_key and not groq_key:
gemini_key = request.gemini_key or None
groq_key = request.groq_key or None
# Enforce user-supplied key β DocForge does not maintain a server-level key.
# DEV_MODE bypasses this so local dev works without a key configured.
import os as _os
if not gemini_key and not _os.getenv("DEV_MODE", "false").lower() == "true":
raise HTTPException(
status_code=400,
detail=(
"Gemini API key required. "
"Get a free key at https://aistudio.google.com/app/apikey "
"then add it in DocForge Settings."
),
)
job_id = str(uuid.uuid4())
# Store initial job state
await cache.set_job(job_id, {
"status": JobStatus.queued,
"job_id": job_id,
"input": request.input,
"input_type": request.input_type,
})
# Run the heavy pipeline work in the background so the HTTP response
# returns instantly (crawling takes 30β90s)
background_tasks.add_task(
_run_pipeline,
job_id=job_id,
request=request,
gemini_key=gemini_key,
groq_key=groq_key,
)
logger.info("Queued job %s for input=%s type=%s", job_id, request.input, request.input_type)
return JobCreatedResponse(job_id=job_id, status=JobStatus.queued)
@router.get("/{job_id}", response_model=ContextJobResult)
async def get_context_job(job_id: str) -> ContextJobResult:
"""Poll job status. Returns output when status == 'complete'."""
job = await cache.get_job(job_id)
if job is None:
raise HTTPException(status_code=404, detail=f"Job {job_id} not found")
return ContextJobResult(**job)
# ---------------------------------------------------------------------------
# Background pipeline
# ---------------------------------------------------------------------------
async def _run_pipeline(
job_id: str,
request: ContextRequest,
gemini_key: str | None = None,
groq_key: str | None = None,
) -> None:
"""
Full DocForge pipeline:
1. Resolve package / fetch URL / clone repo
2. Crawl documentation
3. Structure with Gemini (using user's stored key if authenticated)
4. Format as .context.md
5. Cache result
"""
t_start = time.time()
async def update_status(status: str, **extra: object) -> None:
job = await cache.get_job(job_id) or {}
job.update({"status": status, "job_id": job_id, **extra})
await cache.set_job(job_id, job)
try:
await update_status(JobStatus.processing)
# ββ Step 1: Resolve metadata ββββββββββββββββββββββββββββββββββββββ
library = request.input
version = "latest"
docs_url: str | None = None
version_tag: str | None = None
copypaste_docs: str | None = None # pre-fetched docs for copy-paste libs
pre_structured: dict | None = None # pre-built structured dict (skips LLM)
if request.input_type == InputType.npm:
# If the user somehow passed a URL as an npm package, treat it as a URL crawl
if request.input.startswith(("http://", "https://")):
logger.warning("npm input looks like a URL β switching to URL crawl: %s", request.input)
raw_docs = await crawl_url(request.input)
structured = await structure_with_gemini(library, version, raw_docs, gemini_key=gemini_key, groq_key=groq_key)
output = format_context_md(structured) if request.output_format == OutputFormat.context_md else __import__("json").dumps(structured, indent=2)
structured["cached_at"] = date.today().isoformat()
await cache.set_cached_context(library, version, structured)
await update_status(JobStatus.complete, library=library, version=version, output=output, components=structured.get("components", []))
return
meta = await resolve_npm_package(request.input)
library = meta["name"]
version = meta["version"]
docs_url = meta.get("homepage")
version_tag = meta.get("_version_tag") # set when version not on npm but GitHub exists
# Copy-paste library: docs already fetched during resolution
if meta.get("_is_copypaste"):
copypaste_docs = meta.get("_raw_docs", "")
pre_structured = meta.get("_structured") or None
logger.info(
"Copy-paste library '%s' detected via %s β skipping crawl (%d chars pre-fetched)",
library, meta.get("_copypaste_method"), len(copypaste_docs),
)
else:
logger.info("Resolved npm: %s@%s docs=%s tag=%s", library, version, docs_url, version_tag)
elif request.input_type == InputType.pypi:
meta = await resolve_pypi_package(request.input)
library = meta["name"]
version = meta["version"]
docs_url = meta.get("homepage")
logger.info("Resolved PyPI: %s@%s docs=%s", library, version, docs_url)
elif request.input_type == InputType.crates:
meta = await resolve_crates(request.input)
library = meta["name"]
version = meta["version"]
docs_url = meta.get("homepage")
logger.info("Resolved crates.io: %s@%s docs=%s", library, version, docs_url)
elif request.input_type == InputType.rubygems:
meta = await resolve_rubygems(request.input)
library = meta["name"]
version = meta["version"]
docs_url = meta.get("homepage")
logger.info("Resolved RubyGems: %s@%s docs=%s", library, version, docs_url)
elif request.input_type == InputType.pubdev:
meta = await resolve_pubdev(request.input)
library = meta["name"]
version = meta["version"]
docs_url = meta.get("homepage")
logger.info("Resolved pub.dev: %s@%s docs=%s", library, version, docs_url)
elif request.input_type == InputType.nuget:
meta = await resolve_nuget(request.input)
library = meta["name"]
version = meta["version"]
docs_url = meta.get("homepage")
logger.info("Resolved NuGet: %s@%s docs=%s", library, version, docs_url)
elif request.input_type == InputType.maven:
meta = await resolve_maven(request.input)
library = meta["name"]
version = meta["version"]
docs_url = meta.get("homepage")
logger.info("Resolved Maven: %s@%s docs=%s", library, version, docs_url)
elif request.input_type == InputType.hex:
meta = await resolve_hex(request.input)
library = meta["name"]
version = meta["version"]
docs_url = meta.get("homepage")
logger.info("Resolved Hex: %s@%s docs=%s", library, version, docs_url)
elif request.input_type == InputType.cran:
meta = await resolve_cran(request.input)
library = meta["name"]
version = meta["version"]
docs_url = meta.get("homepage")
logger.info("Resolved CRAN: %s@%s docs=%s", library, version, docs_url)
elif request.input_type == InputType.url:
# Derive a clean library name from the domain so the cache key is
# human-readable ("reactbits") rather than the raw URL.
# GitHub URLs keep their repo name; plain URLs use the domain stem.
if "github.com" in request.input:
# e.g. https://github.com/DavidHDev/react-bits β react-bits
parts = [p for p in urlparse(request.input).path.split("/") if p]
library = parts[1] if len(parts) >= 2 else request.input
else:
domain = urlparse(request.input).netloc.removeprefix("www.")
library = domain.split(".")[0] # "reactbits" from "reactbits.dev"
logger.info("URL input '%s' β library name '%s' for cache/structuring", request.input, library)
# ββ Step 2: Check cache (skip expensive crawl if already done) ββββ
cached = await cache.get_cached_context(library, version)
if cached:
output = format_context_md(cached) if request.output_format == OutputFormat.context_md else str(cached)
await update_status(
JobStatus.complete,
library=library,
version=version,
output=output,
components=cached.get("components", []),
)
logger.info("Cache hit for %s@%s, job %s done in %.1fs", library, version, job_id, time.time() - t_start)
return
# ββ Step 3: Crawl documentation βββββββββββββββββββββββββββββββββββ
raw_docs = ""
# Copy-paste library β docs were fetched during resolution, skip crawl
if copypaste_docs is not None:
raw_docs = copypaste_docs
elif request.input_type == InputType.paste:
raw_docs = await ingest_paste(request.content or request.input)
elif request.input_type == InputType.github:
from ..services.ingestion.github_ingester import ingest_github_repo
raw_docs = await ingest_github_repo(request.input)
elif request.input_type == InputType.url:
if "github.com" in request.input:
from ..services.ingestion.github_ingester import ingest_github_repo
raw_docs = await ingest_github_repo(request.input)
else:
import asyncio as _asyncio
from ..services.ingestion.url_ingester import extract_github_url
from ..services.ingestion.copypaste_ingester import (
check_jsrepo, check_shadcn_registry, detect_and_ingest,
)
# library is already set to the clean name_hint in Step 1
# ββ Parallel: all three lightweight checks at once βββββββββββββ
# check_shadcn_registry hits /r/registry.json etc. β it does NOT
# need a GitHub URL, so it runs alongside the other two.
github_url, jsrepo_name_docs, shadcn_result = await _asyncio.gather(
extract_github_url(request.input),
check_jsrepo(library, None),
check_shadcn_registry(request.input, library),
)
# shadcn registry returns a pre-built structured dict β skip LLM entirely
if shadcn_result is not None:
pre_structured = shadcn_result
logger.info(
"URL '%s' β shadcn registry match for '%s' (%d components, skipping LLM)",
request.input, library, len(pre_structured.get("components", [])),
)
else:
# Prefer jsrepo registry hits for the raw-docs path
quick_docs = jsrepo_name_docs
# If we also have a GitHub URL, do a targeted jsrepo check
# (raw GitHub paths + jsrepo/{owner}/{repo}).
jsrepo_targeted: str | None = None
if github_url and not quick_docs:
jsrepo_targeted = await check_jsrepo(library, github_url)
cp_docs = jsrepo_targeted or quick_docs
if cp_docs:
raw_docs = cp_docs
logger.info("URL '%s' β registry match for '%s' (%d chars)", request.input, library, len(raw_docs))
elif github_url:
logger.info("URL '%s' β GitHub repo found: %s", request.input, github_url)
# copy-paste detection with full github_url context
cp_result = await detect_and_ingest(
name=library,
github_url=github_url,
homepage=request.input,
)
if cp_result:
raw_docs = cp_result.get("_raw_docs", "")
logger.info(
"URL '%s' β copy-paste lib (method=%s) %d chars",
request.input, cp_result.get("_copypaste_method"), len(raw_docs),
)
else:
# Not copy-paste β GitHub ingester still beats Playwright
from ..services.ingestion.github_ingester import ingest_github_repo
raw_docs = await ingest_github_repo(github_url)
logger.info("URL '%s' β GitHub ingestion (%d chars)", request.input, len(raw_docs))
else:
# No GitHub found β fall back to Playwright
raw_docs = await crawl_url(request.input)
logger.info("URL '%s' β Playwright crawl (%d chars)", request.input, len(raw_docs))
# Sparse content β try GitHub search by domain name.
# Only accept a fallback repo whose homepage points back to the
# same domain we started from (prevents picking up unrelated
# packages like MCP wrappers that happen to mention the name).
if len(raw_docs) < 20_000:
import httpx
input_domain = urlparse(request.input).netloc.removeprefix("www.")
async with httpx.AsyncClient(timeout=10) as _client:
from ..services.ingestion.npm_ingester import _github_search
fallback = await _github_search(_client, library)
if fallback:
# Quick check: does the GitHub repo's homepage match?
try:
meta_resp = await _client.get(
f"https://api.github.com/repos/{fallback.removeprefix('https://github.com/')}",
headers={"User-Agent": "docforge/1.0", "Accept": "application/vnd.github+json"},
timeout=5,
)
if meta_resp.is_success:
repo_homepage = meta_resp.json().get("homepage") or ""
if input_domain not in repo_homepage:
logger.info(
"GitHub fallback '%s' homepage '%s' doesn't match domain '%s' β skipping",
fallback, repo_homepage, input_domain,
)
fallback = None
except Exception:
fallback = None # if we can't verify, skip it
if fallback:
logger.info(
"Sparse Playwright content (%d chars) for '%s' β trying GitHub: %s",
len(raw_docs), request.input, fallback,
)
from ..services.ingestion.github_ingester import ingest_github_repo
github_docs = await ingest_github_repo(fallback)
if len(github_docs) > len(raw_docs):
raw_docs = github_docs
else:
# npm or pypi: crawl the resolved docs URL
if version_tag and not docs_url:
# Version not on npm but GitHub repo known β use git tag directly
from ..services.ingestion.github_ingester import ingest_github_repo
repo_url = meta.get("repository", "")
import re as _re
repo_url = _re.sub(r"^git\+", "", repo_url)
logger.info("Using GitHub tag %s for %s", version_tag, repo_url)
raw_docs = await ingest_github_repo(repo_url, version_tag=version_tag)
elif docs_url:
if "github.com" in docs_url:
from ..services.ingestion.github_ingester import ingest_github_repo
raw_docs = await ingest_github_repo(docs_url, version_tag=version_tag)
else:
raw_docs = await crawl_url(docs_url)
else:
# Web search fallback: query DuckDuckGo for docs
found_url = await find_docs_url(f"{library} {version}")
if found_url:
logger.info("Web search found docs for %s: %s", library, found_url)
if "github.com" in found_url:
from ..services.ingestion.github_ingester import ingest_github_repo
raw_docs = await ingest_github_repo(found_url)
else:
raw_docs = await crawl_url(found_url)
else:
logger.warning("No docs URL found for %s, sending minimal stub", library)
raw_docs = f"Library: {library}\nVersion: {version}\nNo documentation URL found."
logger.info("Crawl done for %s@%s: %d chars in %.1fs", library, version, len(raw_docs), time.time() - t_start)
# ββ Step 4: Structure with Gemini βββββββββββββββββββββββββββββββββ
# Skip LLM for copy-paste libraries with pre-built structured data
if pre_structured is not None:
structured = pre_structured
structured.setdefault("library", library)
structured.setdefault("version", version)
logger.info("Using pre-built structure for %s@%s (%d components)", library, version, len(structured.get("components", [])))
else:
structured = await structure_with_gemini(library, version, raw_docs, gemini_key=gemini_key, groq_key=groq_key)
# Filter to requested components if specified
if request.components:
requested = {c.lower() for c in request.components}
structured["components"] = [
c for c in structured.get("components", [])
if c.get("name", "").lower() in requested
]
# ββ Step 5: Format output βββββββββββββββββββββββββββββββββββββββββ
if request.output_format == OutputFormat.context_md:
output = format_context_md(structured)
else:
import json
output = json.dumps(structured, indent=2)
# ββ Step 6: Cache and complete ββββββββββββββββββββββββββββββββββββ
structured["cached_at"] = date.today().isoformat()
await cache.set_cached_context(library, version, structured)
total = time.time() - t_start
logger.info("Pipeline complete for %s@%s in %.1fs", library, version, total)
await update_status(
JobStatus.complete,
library=library,
version=version,
output=output,
components=structured.get("components", []),
)
except Exception as exc:
logger.exception("Pipeline failed for job %s: %s", job_id, exc)
await update_status(JobStatus.failed, error=str(exc)) |