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).

apps/api/app/change_detection/text_diff.py

@@ -0,0 +1,56 @@
+"""Layer 3: bounded text diff.
+
+Runs on noise-stripped text (see noise_filters.py) so navigation/cookie/
+timestamp churn doesn't register as a change. Bounded: only a capped number
+of added/removed lines are kept, so a full page rewrite doesn't produce an
+unbounded diff blob for storage or LLM consumption later.
+"""
+
+from __future__ import annotations
+
+import difflib
+from dataclasses import dataclass, field
+
+from app.change_detection.noise_filters import strip_noise
+
+_MAX_DIFF_LINES = 40
+
+
+@dataclass(frozen=True)
+class TextDiffResult:
+    diff_ratio: float  # 0.0 = identical, 1.0 = completely different
+    added_lines: list[str] = field(default_factory=list)
+    removed_lines: list[str] = field(default_factory=list)
+
+    @property
+    def is_identical(self) -> bool:
+        return self.diff_ratio == 0.0
+
+
+def bounded_text_diff(previous_text: str, current_text: str) -> TextDiffResult:
+    previous_clean = strip_noise(previous_text or "")
+    current_clean = strip_noise(current_text or "")
+
+    if previous_clean == current_clean:
+        return TextDiffResult(diff_ratio=0.0)
+
+    previous_lines = [line for line in previous_clean.splitlines() if line.strip()]
+    current_lines = [line for line in current_clean.splitlines() if line.strip()]
+
+    matcher = difflib.SequenceMatcher(a=previous_lines, b=current_lines, autojunk=False)
+    similarity = matcher.ratio()
+    diff_ratio = round(1.0 - similarity, 4)
+
+    added: list[str] = []
+    removed: list[str] = []
+    for tag, i1, i2, j1, j2 in matcher.get_opcodes():
+        if tag in ("replace", "delete"):
+            removed.extend(previous_lines[i1:i2])
+        if tag in ("replace", "insert"):
+            added.extend(current_lines[j1:j2])
+
+    return TextDiffResult(
+        diff_ratio=diff_ratio,
+        added_lines=added[:_MAX_DIFF_LINES],
+        removed_lines=removed[:_MAX_DIFF_LINES],
+    )