Search Algorithm Strategies

1. Zoom Level Calibration

Objective: Determine the optimal zoom level to maximize the number of unique locations found for a given keyword and location.

Input:
- Keyword (e.g., "carpenters")
- Location (e.g., "Barcelona, Spain")
- Zoom Range (e.g., 12-18)
Process:
- Iterate through the defined range of zoom levels.
- For each zoom level:
  - Perform a Google Maps search using the googleMaps function.
  - Store the results in a JSON file using the --dst option (e.g., ./tmp/search/test-zoomlevel-<zoom>.json).
  - Count the total number of valid results returned.
- Maintain a record of (Zoom Level -> Result Count).
Output:
- The zoom level that yielded the maximum number of results.
Notes:
- This process assumes that for a fixed point, the "best" zoom captures the most relevant density without being too broad (losing small entities) or too narrow (missing context).
- Overlaps/Duplicates should be handled by the underlying search function or post-processing if multi-point scanning is used later.

Objective: Scan a larger, defined area (e.g., "Madrid" or "Spain") using the optimal zoom level to ensure comprehensive coverage.

Input:
- Target Area Boundaries (Polygon/Box).
- Calibrated Zoom Level (from Step 1).
- List of Provinces/Towns (if segmenting by admin regions).
Grid Generation:
- Determine the lat/long delta that corresponds to the calibrated zoom level's viewport size.
- Create a grid of search coordinates covering the Target Area.
Execution & State Management:
- This is a long-running task.
- State Store: Maintain a persistent state (JSON/DB) tracking:
  - Queue of pending coordinates.
  - Completed coordinates.
  - Failed coordinates.
- Process the queue sequentially or in parallel batches.
Aggregation:
- Combine all result files.
- Perform global deduplication (by place_id or title + address).