#!/usr/bin/env bash
# RDLC Fact Regression Tests — TEMPLATE
# Ported from states-project-research/scripts/regression_test.sh (claude-rdlc-wizard v0.1.0)
#
# Customize: replace TODO markers with project-specific assertions about your facts.
# Run: bash scripts/regression_test.sh
# Exit code 0 = all pass, 1 = failures found
#
# Pattern: every defect found becomes a permanent regression check.
# The ratchet only turns forward — assertions get stricter, never looser.

set -euo pipefail

FAIL=0
PASS=0

# TODO: Set the deliverable file paths your project produces.
# Source markdown is the authoritative copy; HTML/PDF are derived but tested too
# because they get forwarded without context.
DOC="${RDLC_DOC:-output/research_document.md}"
HTML="${RDLC_HTML:-output/research_document.html}"

red()   { printf "\033[31mFAIL: %s\033[0m\n" "$1"; }
green() { printf "\033[32mPASS: %s\033[0m\n" "$1"; }
note()  { printf "\033[33mNOTE: %s\033[0m\n" "$1"; }

# --- Helpers ---

check_absent() {
  local label="$1" pattern="$2" file="$3"
  if [ ! -f "$file" ]; then
    note "$label — file not found ($file), skipping"
    return 0
  fi
  if grep -qiE "$pattern" "$file" 2>/dev/null; then
    red "$label — found in $file"
    grep -inE "$pattern" "$file" | head -3
    FAIL=$((FAIL + 1))
  else
    green "$label"
    PASS=$((PASS + 1))
  fi
}

# Like check_absent but excludes the review/changelog section.
# Useful when you intentionally document a corrected error inline (e.g. "Fixed
# from: <wrong claim>") and don't want the assertion to fail on the explanation.
check_absent_excl_review() {
  local label="$1" pattern="$2" file="$3"
  if [ ! -f "$file" ]; then
    note "$label — file not found ($file), skipping"
    return 0
  fi
  local hits
  hits=$(grep -iE "$pattern" "$file" 2>/dev/null \
    | grep -viE "Issue Found|Fix Applied|Fixed to|Softened to|overclaim|reframed from|What Changed|previously claimed|corrected from" \
    | grep -iE "$pattern" || true)
  if [ -n "$hits" ]; then
    red "$label — found in $file (excluding review/changelog)"
    printf '%s\n' "$hits" | head -3
    FAIL=$((FAIL + 1))
  else
    green "$label"
    PASS=$((PASS + 1))
  fi
}

check_present() {
  local label="$1" pattern="$2" file="$3"
  if [ ! -f "$file" ]; then
    note "$label — file not found ($file), skipping"
    return 0
  fi
  if grep -qiE "$pattern" "$file" 2>/dev/null; then
    green "$label"
    PASS=$((PASS + 1))
  else
    red "$label — NOT found in $file"
    FAIL=$((FAIL + 1))
  fi
}

# --- Test sections ---

echo "=== RDLC REGRESSION TESTS ==="
echo ""

# --- 1. Confidence labels exist on every claim ---
# At least one confidence marker must exist in the deliverable
echo "--- Confidence labels ---"
check_present "Confidence vocabulary in use" "VERIFIED|SUPPORTED|INFERRED|UNVERIFIED|GAP|DIRECT" "$DOC"

# --- 2. Source-at-first-mention ---
# At least one source pattern must exist in the deliverable
echo ""
echo "--- Source citations ---"
check_present "At least one source citation" "https?://|PMID:?[[:space:]]*[0-9]|DOI:?[[:space:]]*10\\.|\\[Source:" "$DOC"

# --- 3. AI slop must be absent ---
# Project-level enforcement layered on top of the per-edit slop hook
echo ""
echo "--- AI slop gate ---"
check_absent "No 'deep dive'" "deep dive" "$DOC"
check_absent "No 'game-changer'" "game.changer" "$DOC"
check_absent "No 'cutting-edge'" "cutting.edge" "$DOC"
check_absent "No 'delve'" "delve" "$DOC"
check_absent "No 'tapestry'" "tapestry" "$DOC"
check_absent "No 'paradigm'" "paradigm" "$DOC"
check_absent "No 'smoking gun'" "smoking gun" "$DOC"
check_absent "No 'in today's world'" "in today.s world" "$DOC"

# --- 4. FABRICATED CLAIMS (must NOT exist) ---
# TODO: Add assertions for every fabrication caught during cross-model review.
# Pattern: a claim that was wrong, corrected, and must never reappear.
echo ""
echo "--- Fabricated claims ---"
# Example (replace with project-specific):
# check_absent "No fabricated 'Person X worked on Y' claim" "person_x.*(worked on y|joined y)" "$DOC"

# --- 5. CORRECT FACTS (must exist) ---
# TODO: Add assertions for every load-bearing fact your deliverable depends on.
# Pattern: if this fact is removed or rewritten incorrectly, the deliverable fails its mission.
echo ""
echo "--- Correct facts ---"
# Example (replace with project-specific):
# check_present "Correct title" "Senior Director.*Lawmaker Engagement" "$DOC"
# check_present "Correct salary range" "111,679.*117,262" "$DOC"

# --- 6. AUDIENCE FIREWALL (private content NOT in public deliverable) ---
# TODO: Add per-deliverable forbidden patterns from .rdlc/audience-firewall.conf
echo ""
echo "--- Audience firewall ---"
# Example (replace with project-specific):
# PUBLIC_HTML="output/public_strategic_analysis.html"
# check_absent "No salary in public deliverable" "111,679|117,262|salary range" "$PUBLIC_HTML"
# check_absent "No mock interview content in public deliverable" "mock interview|interviewer strategy" "$PUBLIC_HTML"

# --- 7. HTML matches MD (regenerate-after-edit ratchet) ---
# Catches "I edited the markdown but forgot to regenerate"
if [ -f "$DOC" ] && [ -f "$HTML" ]; then
    if [ "$DOC" -nt "$HTML" ]; then
        red "Stale HTML — $DOC is newer than $HTML"
        echo "    Run: python3 scripts/generate_deliverable.py"
        FAIL=$((FAIL + 1))
    else
        green "HTML freshness — $HTML is current"
        PASS=$((PASS + 1))
    fi
fi

# --- Summary ---
echo ""
echo "=========================="
echo "PASS: $PASS  /  FAIL: $FAIL"
echo "=========================="

if [ "$FAIL" -gt 0 ]; then
    exit 1
fi

exit 0