1  #!/bin/sh
 2  # Pre-commit hook helper: warns if new data/prompts/docs files are staged
 3  # but not listed in tests/data/required-files.test.js
 4  #
 5  # Only checks files that match patterns the pipeline loads at runtime.
 6  # Exits 0 always (warning only, never blocks commit).
 7  
 8  MANIFEST="tests/data/required-files.test.js"
 9  
10  if [ ! -f "$MANIFEST" ]; then
11    exit 0
12  fi
13  
14  # Patterns for runtime-critical file types (not code, not html storage)
15  # A = added, C = copied, R = renamed (destination)
16  new_files=$(git diff --cached --name-only --diff-filter=ACR -- \
17    'data/*.json' \
18    'data/*.csv' \
19    'data/*.txt' \
20    'data/compliance/*.json' \
21    'data/franchises/*.txt' \
22    'data/templates/**/*.json' \
23    'prompts/*.md' \
24    'prompts/agents/*.md' \
25    'docs/**/*-best-practices*.md' \
26    'db/schema.sql' \
27    2>/dev/null)
28  
29  if [ -z "$new_files" ]; then
30    exit 0
31  fi
32  
33  missing=""
34  for f in $new_files; do
35    # Skip html storage (generated at runtime, not tracked in manifest)
36    case "$f" in data/html/*) continue ;; esac
37    # Skip search-volume and regions.txt (not loaded at runtime by default)
38    case "$f" in *search-volume*.csv) continue ;; esac
39    case "$f" in data/*/regions.txt) continue ;; esac
40    # Skip pricing-export.json (generated, not a dependency)
41    case "$f" in data/pricing-export.json) continue ;; esac
42    # Skip grammar reports (generated by spintax-grammar-check, not a hard dep)
43    case "$f" in *grammar-report*) continue ;; esac
44  
45    if ! grep -qF "'$f'" "$MANIFEST" && ! grep -qF "\"$f\"" "$MANIFEST"; then
46      missing="$missing\n  $f"
47    fi
48  done
49  
50  if [ -n "$missing" ]; then
51    printf '\n\033[33m⚠ New data/prompt files not in required-files manifest:%s\033[0m\n' "$missing"
52    printf '  Add them to: %s\n\n' "$MANIFEST"
53  fi
54  
55  exit 0