Add calculator categories

migrate.py

@@ -6,59 +6,30 @@ BASE_DIR = Path(__file__).resolve().parent
 CALCLIST = BASE_DIR / 'calculators_list.md'
 OUTPUT_FILE = BASE_DIR / 'hdyc-svelte/src/lib/data/calculators.ts'
 
-FLUID_KEYWORDS = [
-    'flow',
-    'mass flux',
-    'volumetric',
-    'permeability',
-    'viscosity',
-    'kinematic',
-    'surface tension',
-    'molar',
-    'concentration',
-    'flux density',
-    'flow rate',
-    'gallon per',
-    'gallons per',
-    'liter per',
-    'liters per',
-    'cubic per',
-    'cubic meter per',
-    'cubic meters per',
-    'cubic foot per',
-    'cubic feet per',
-    'cubic inch per',
-    'cubic inches per',
-    'kg per',
-    'kilogram per',
-    'kilograms per',
-    'gram per',
-    'grams per',
-    'g per',
-    'lb per',
-    'lbs per',
-    'pound per',
-    'pounds per',
-    'mole per',
-    'moles per',
-    'mol per',
-    'mmol per',
-    'percent by mass',
-    'ppm',
-    'heat transfer coefficient',
-    'per square meter',
-    'per square metre',
-    'per square foot',
-    'per square inch',
-    'per square centimeter',
-    'per square centimetre',
-    'per cubic meter',
-    'per cubic metre',
-    'per cubic foot',
-    'per cubic inch'
+CATEGORY_KEYS = [
+    'length',
+    'weight',
+    'temperature',
+    'volume',
+    'fluids',
+    'area',
+    'speed',
+    'pressure',
+    'energy',
+    'magnetism',
+    'power',
+    'data',
+    'time',
+    'angle',
+    'number-systems',
+    'radiation',
+    'electrical',
+    'force',
+    'light',
+    'other',
 ]
 
-CURRENCY_KEYWORDS = ['currency', 'exchange rate', 'forex']
+CATEGORY_SET = set(CATEGORY_KEYS)
 
 def load_external_descriptions():
     # Placeholder for future enrichment sources.
@@ -85,12 +56,14 @@ def parse_calculators_list():
             parts = [p.strip() for p in line.strip().strip('|').split('|')]
             name_idx = header_map.get('Calculator Name')
             slug_idx = header_map.get('Slug')
+            category_idx = header_map.get('Category')
             factor_idx = header_map.get('Conversion Factor')
-            if None not in (name_idx, slug_idx, factor_idx) and len(parts) > max(name_idx, slug_idx, factor_idx):
+            if None not in (name_idx, slug_idx, category_idx, factor_idx) and len(parts) > max(name_idx, slug_idx, category_idx, factor_idx):
                 name = parts[name_idx]
                 slug = parts[slug_idx]
+                category = parts[category_idx]
                 factor_raw = parts[factor_idx]
-                active_calcs.append((name, slug, factor_raw))
+                active_calcs.append((name, slug, category, factor_raw))
                 
     return active_calcs
 
@@ -107,38 +80,10 @@ def split_conversion_name(name):
         return parts[0].strip(), parts[1].strip()
     return None
 
-def guess_category(name):
-    name_l = name.lower()
-    if any(keyword in name_l for keyword in CURRENCY_KEYWORDS):
-        return 'currency'
-    if any(keyword in name_l for keyword in FLUID_KEYWORDS):
-        return 'fluids'
-    if any(x in name_l for x in ['acre-foot', 'acre-feet', 'acrefoot', 'acre feet']):
-        return 'volume'
-    if 'temp scale' in name_l or 'newton (temp' in name_l:
-        return 'temperature'
-    if any(x in name_l for x in ['force', 'torque', 'newton', 'dyne', 'foot-pound']): return 'force'
-    if any(x in name_l for x in ['acre', 'hectare', 'square']): return 'area'
-    if any(x in name_l for x in ['meter', 'inch', 'feet', 'yard', 'mile', 'cable', 'fathom', 'rod', 'chain', 'nautical', 'league']): return 'length'
-    if any(x in name_l for x in ['gram', 'pound', 'ounce', 'carat', 'stone', 'slug', 'ton', 'pennyweight', 'grain', 'momme']): return 'weight'
-    if any(x in name_l for x in ['celsius', 'fahrenheit', 'kelvin', 'rankine', 'delisle', 'reaumur', 'réaumur', 'romer', 'rømer']): return 'temperature'
-    if any(x in name_l for x in ['liter', 'gallon', 'cup', 'pint', 'quart', 'fluid', 'milliliter', 'spoon', 'drop']): return 'volume'
-    if any(x in name_l for x in ['pascal', 'bar', 'psi', 'atmosphere', 'mmhg', 'torr', 'water', 'mercury']): return 'pressure'
-    if any(x in name_l for x in ['magnet', 'magnetic', 'tesla', 'gauss', 'oersted', 'weber', 'maxwell', 'gilbert', 'ampere-turn', 'ampere turns', 'ampere per meter', 'magnetomotive']): return 'magnetism'
-    if any(x in name_l for x in ['joule', 'calorie', 'btu', 'erg', 'therm', 'electron-volt']): return 'energy'
-    if any(x in name_l for x in ['thermal', 'heat', 'conductivity', 'resistance', 'capacity', 'expansion', 'transfer']): return 'temperature'
-    if any(x in name_l for x in ['watt', 'horsepower']): return 'power'
-    if any(x in name_l for x in ['byte', 'bit', 'nibble', 'baud']): return 'data'
-    if 'light' in name_l or any(x in name_l for x in ['lumen', 'lux', 'candela']): return 'light'
-    if any(x in name_l for x in ['degree', 'degrees', 'radian', 'radians', 'arcminute', 'arcminutes', 'arcsecond', 'arcseconds', 'gradian', 'gradians', 'mil', 'mils', 'quadrant', 'quadrants', 'sextant', 'sextants', 'turn', 'turns', 'points (compass', 'points-compass']): return 'angle'
-    if any(x in name_l for x in ['second', 'minute', 'hour', 'day', 'week', 'month', 'year']): return 'time'
-    if any(x in name_l for x in ['binary', 'hex', 'octal', 'decimal', 'ascii', 'fraction']): return 'number-systems'
-    if any(x in name_l for x in ['becquerel', 'curie', 'gray', 'rad', 'sievert', 'rem', 'roentgen', 'rutherford']): return 'radiation'
-    if any(x in name_l for x in ['volt', 'amp', 'ohm', 'siemens', 'farad', 'henry', 'coulomb']): return 'electrical'
-    if ' per ' in name_l or 'knot' in name_l or 'mach' in name_l or 'rpm' in name_l: return 'speed' # RPM might be frequency, close enough
-    if any(x in name_l for x in ['binary', 'hex', 'octal', 'decimal', 'base']):
-        return 'number-systems'
-    return 'other'
+def normalize_category(raw: str) -> str:
+    normalized = raw.strip().lower().replace(' ', '-')
+    normalized = re.sub(r'[^a-z0-9-]', '', normalized)
+    return normalized
 
 def process():
     external_descriptions = load_external_descriptions()
@@ -146,7 +91,8 @@ def process():
     
     calculators_ts_entries = []
     
-    for raw_name, slug, factor_raw in active_rows:
+    seen_slugs = set()
+    for raw_name, slug, category_raw, factor_raw in active_rows:
         if raw_name == 'Calculator Name' or not slug:
             continue
 
@@ -159,7 +105,14 @@ def process():
         else:
             in1, in2 = "From", "To"
             
-        category = guess_category(display_name)
+        category = normalize_category(category_raw)
+        if not category:
+            raise ValueError(f'Category required for {display_name}')
+        if category not in CATEGORY_SET:
+            raise ValueError(f'Unknown category \"{category_raw}\" resolved to \"{category}\" for {slug}')
+        if slug in seen_slugs:
+            continue
+        seen_slugs.add(slug)
         desc_html = external_descriptions.get(slug, "")
         
         c_type = 'standard'
@@ -368,7 +321,6 @@ export const categories: Record<string, { label: string; icon: string }> = {
   speed:           { label: 'Speed / Velocity',    icon: '💨' },
   pressure:        { label: 'Pressure',            icon: '🔽' },
   energy:          { label: 'Energy',              icon: '⚡' },
-  currency:        { label: 'Currency',            icon: '💱' },
   magnetism:       { label: 'Magnetism',           icon: '🧲' },
   power:           { label: 'Power',               icon: '🔌' },
   data:            { label: 'Data Storage',        icon: '💾' },