Initial commit: CI Agent competitive-intelligence monitoring app

FastAPI + Celery + Next.js + Postgres/Redis app with company monitoring, source collection, LLM-based change analysis, enrichment, and account security (Turnstile, escalating lockout, email verification).
2026-08-05 10:48:20 -04:00
commit 1a4c80958f
365 changed files with 43541 additions and 0 deletions
@@ -0,0 +1,62 @@
+"""Deterministic mock search provider - the default (`SEARCH_PROVIDER=mock`)
+and what every automated test runs against. Never calls a network.
+
+Unlike `MockLLMProvider` (which fabricates plausible-looking structured
+output from real evidence), this provider has no real evidence to work
+from - a company name alone isn't enough to honestly guess an industry,
+headquarters, or competitor list. So except for the one query type where a
+name-derived guess is genuinely meaningful (official website - many
+companies really do live at a domain close to their name), every other
+query honestly reports "no information available" as its snippet text.
+That honesty flows through `discovery_service.py` and the company-profile
+LLM task: fields with no real evidence come back `None`/`[]`, and the
+wizard's Review step shows them as "not found - fill in yourself" rather
+than presenting fabricated data as fact.
+"""
+
+from __future__ import annotations
+
+import re
+
+from app.search.base import SearchResult
+
+_DOMAIN_STRIP_RE = re.compile(r"[^a-z0-9]+")
+_WEBSITE_SUFFIX = " official website"
+
+
+def _guess_domain(name: str) -> str:
+    slug = _DOMAIN_STRIP_RE.sub("", name.lower())
+    return f"{slug}.com" if slug else "example.com"
+
+
+class MockSearchProvider:
+    provider_name = "mock"
+
+    async def search(self, query: str, *, count: int = 5) -> list[SearchResult]:
+        if query.lower().endswith(_WEBSITE_SUFFIX):
+            name = query[: -len(_WEBSITE_SUFFIX)]
+            domain = _guess_domain(name)
+            results = [
+                SearchResult(
+                    title=f"{name} - Official Site",
+                    url=f"https://{domain}",
+                    snippet=(
+                        f"A likely official website for {name}, guessed from its name. "
+                        "Mock search - no real web search performed; set SEARCH_PROVIDER=brave "
+                        "for real results."
+                    ),
+                )
+            ]
+        else:
+            results = [
+                SearchResult(
+                    title=f"Mock search: {query}",
+                    url="https://example.com/mock-search",
+                    snippet=(
+                        "No information available - SEARCH_PROVIDER=mock is the dev/test "
+                        "default and performs no real web search. Set SEARCH_PROVIDER=brave "
+                        "for real discovery."
+                    ),
+                )
+            ]
+        return results[:count]