add-category

Add Search Category

1. Gather Requirements

Ask the user for:

• OSM type(s) — e.g. shop-leather, amenity-cafe. Multiple types separated by |
• Group(s) — which @group to attach (e.g. @shop, @category_eat). Grep ^@ in data/categories.txt for existing groups
• Emoji (optional) — U+XXXX Unicode format, placed as synonyms on the en: line

2. Ensure Translations Exist in types_strings.txt

Convert OSM type dashes to dots: shop-leather → [type.shop.leather].

Check if the entry exists in data/strings/types_strings.txt.

If absent — create translations first:

1. Ask user for the English name and translation context (what this place/thing is)
2. Run translation tool:

   python3 tools/python/translate.py --context "context here" "English Name"
   

   This outputs translations in both categories.txt and strings.txt formats via DeepL+Google
3. Proof-read the generated translations in all languages using the provided context — fix obvious errors
4. Add the [type.osm.type] entry to data/strings/types_strings.txt:

   [type.shop.leather]
   comment = context used for translation
   en = Leather Shop
   ar = متجر السلع الجلدية
   ...sorted alphabetically by language code...
   zh-Hant = 皮具店
   

If present — read existing translations to use as the base.

3. Build the categories.txt Entry

First translation = types_strings.txt match

The first synonym for each language MUST exactly match the translation in types_strings.txt. This is a hard requirement.

Additional synonyms

Append popular search synonyms after | — only terms users would actually type in the search box. Keep it minimal.

Prefix digits

A digit 1-9 before a synonym controls the minimum characters needed for the suggestion to appear:

• 3Cafe → user must type 3+ chars
• No prefix → full text required

Slavic language rule

For ru, uk, be: short nouns (under 6 letters) need both nominative and genitive forms:

ru:Вино|вина

For sr: threshold is 8 letters. Longer nouns rely on error correction.

4. Entry Format

osm-type1|osm-type2|@group1|@group2
en:4Type Name|synonym|U+1F6B0
ar:Arabic Name
be:Belarusian Name
...languages sorted alphabetically by code...
zh-Hant:Traditional Chinese

• en: line always first after the header
• Other languages sorted alphabetically by language code
• Each entry ends with a blank line
• Insert near related categories (same group or alphabetically by OSM type)
• Supported language codes: ar, be, bg, ca, cs, da, de, el, en, es, et, eu, fa, fi, fr, he, hi, hu, id, it, ja, ko, lt, lv, mr, nb, nl, pl, pt, pt-BR, ro, ru, sk, sl, sr, sv, sw, th, tr, uk, vi, zh-Hans, zh-Hant

5. Displayed Category (@category_*)

Only if adding a category shown in the app's search screen. Requires three coordinated file changes:

a) data/categories.txt

# First keyword should match [category_name] definition in strings.txt!
@category_name
en:Display Name|synonym
...translations...

b) data/strings/strings.txt

[category_name]
comment = Search category for ...; any changes should be duplicated in categories.txt @category_name!
tags = android-app,android-libs-car,apple-maps
en = Display Name
...translations for all supported locales...

The en = value must match the first keyword on the en: line (without prefix digit).

c) libs/search/displayed_categories.cpp

Add "category_name" to the m_keys initializer list.

6. After Editing

1. Sort languages:

   python3 tools/python/sort_categories.py --in-place
   

2. Build and run tests:

   cmake --build build-agent --target search_tests && ctest -j --test-dir build-agent --stop-on-failure --output-on-failure -R search_tests
   

3. If a displayed category was added, also run:

   cmake --build build-agent --target indexer_tests editor_tests && ctest -j --test-dir build-agent --stop-on-failure --output-on-failure -R "indexer_tests|editor_tests"