Begin work on biergarten data generator pipeline

pipeline/src/main.cpp

@@ -0,0 +1,154 @@
+/**
+ * @file main.cpp
+ * @brief Entry point for the brewery data pipeline
+ *
+ * Pipeline Overview:
+ * This is the main data processing pipeline that:
+ * 1. Initializes an in-memory SQLite database
+ * 2. Loads world city data from a JSON file (50k+ cities)
+ * 3. Initializes the brewery generation system (currently mocked)
+ * 4. Demonstrates brewery generation for sample cities
+ *
+ * Architecture:
+ * ┌─────────────┐
+ * │  JSON File  │ (world_city_data.json - 50k+ cities)
+ * └──────┬──────┘
+ *        │
+ *        ▼
+ * ┌─────────────────────┐
+ * │  JsonLoader::Load   │ Parse and validate JSON
+ * └──────┬──────────────┘
+ *        │
+ *        ▼
+ * ┌─────────────────────┐
+ * │  SQLite Database    │ Store cities in-memory
+ * └──────┬──────────────┘
+ *        │
+ *        ▼
+ * ┌─────────────────────┐
+ * │  BreweryGenerator   │ Mock generation (hash-based)
+ * │  .GenerateBrewery() │ Future: LLM-based generation
+ * └─────────────────────┘
+ *
+ * Command Line Arguments:
+ * - argv[1]: Path to GGUF model file (default: ./model.gguf)
+ * - argv[2]: Path to cache directory for JSON downloads (default: /tmp)
+ * - argv[3]: Git commit hash for reproducible data version (default: c5eb7772)
+ *
+ * The pipeline automatically downloads the geographic data from GitHub on first
+ * run and caches it locally to avoid repeated network calls.
+ *
+ * Example Usage - Auto-download (stable 2026-03-28 build):
+ * @code
+ *   ./brewery-pipeline ./llama-7b.gguf
+ * @endcode
+ *
+ * Example Usage - Custom commit:
+ * @code
+ *   ./brewery-pipeline ./llama-7b.gguf /tmp main
+ * @endcode
+ *
+ * Exit Codes:
+ * - 0: Pipeline completed successfully
+ * - 1: Pipeline failed (exception caught)
+ */
+
+#include "data_downloader.h"
+#include "database.h"
+#include "generator.h"
+#include "json_loader.h"
+#include <curl/curl.h>
+#include <iostream>
+
+int main(int argc, char *argv[]) {
+  try {
+    // Initialize libcurl globally (thread-safe mode)
+    curl_global_init(CURL_GLOBAL_DEFAULT);
+
+    // Parse command-line arguments
+    std::string modelPath = argc > 1 ? argv[1] : "./model.gguf";
+    std::string cacheDir = argc > 2 ? argv[2] : "/tmp";
+    std::string commit =
+        argc > 3 ? argv[3] : "c5eb7772"; // Default: stable 2026-03-28
+
+    // Construct cache path for downloaded JSON
+    std::string jsonPath = cacheDir + "/countries+states+cities.json";
+
+    // Step 0: Download geographic data from GitHub (cached locally)
+    // On first run, downloads 45MB JSON. On subsequent runs, uses cached file.
+    // Commit hash allows pinning to specific data versions for reproducibility.
+    std::cout << "\n[Pipeline] Downloading geographic data from GitHub...\n";
+    DataDownloader downloader;
+    downloader.DownloadCountriesDatabase(jsonPath, commit);
+
+    SqliteDatabase db;
+
+    // Step 1: Initialize empty in-memory database
+    std::cout << "Initializing in-memory SQLite database...\n";
+    db.Initialize();
+
+    // Step 2: Load world city data from JSON file
+    // This populates the database with ~50k city records
+    // Each record includes: city name, country, latitude, longitude, population
+    JsonLoader::LoadWorldCities(jsonPath, db);
+
+    // Step 3: Initialize brewery generator
+    // Current: Mock implementation using deterministic hashing
+    // Future: LLM-based generation with llama.cpp
+    std::cout << "Initializing brewery generator...\n";
+    LlamaBreweryGenerator generator;
+    generator.LoadModel(modelPath);
+
+    // Step 4: Query geographic data from database
+    std::cout << "\n=== GEOGRAPHIC DATA OVERVIEW ===\n";
+
+    auto countries = db.QueryCountries(50);
+    auto states = db.QueryStates(50);
+    auto cities = db.QueryCities();
+
+    std::cout << "\nTotal records loaded:";
+    std::cout << "\n  Countries: " << db.QueryCountries(0).size();
+    std::cout << "\n  States: " << db.QueryStates(0).size();
+    std::cout << "\n  Cities: " << cities.size() << "\n";
+
+    // Display 50 countries
+    std::cout << "\n--- 50 COUNTRIES ---\n";
+    for (size_t i = 0; i < countries.size(); i++) {
+      std::cout << (i + 1) << ". " << countries[i].iso2 << " ("
+                << countries[i].iso3 << ") " << countries[i].name << "\n";
+    }
+
+    // Display 50 states
+    std::cout << "\n--- 50 STATES ---\n";
+    for (size_t i = 0; i < states.size(); i++) {
+      std::cout << (i + 1) << ". " << states[i].iso2 << ": " << states[i].name
+                << "\n";
+    }
+
+    // Display 50 cities
+    std::cout << "\n--- 50 CITIES ---\n";
+    for (size_t i = 0; i < std::min(size_t(50), cities.size()); i++) {
+      std::cout << (i + 1) << ". " << cities[i].second << "\n";
+    }
+
+    // Step 5: Demonstrate brewery generation on sample cities
+    std::cout << "\n=== SAMPLE BREWERY GENERATION ===\n\n";
+    for (size_t i = 0; i < std::min(size_t(5), cities.size()); i++) {
+      const auto &[cityId, cityName] = cities[i];
+      auto brewery = generator.GenerateBrewery(cityName, i);
+      std::cout << "  " << cityName << ": " << brewery.name << "\n";
+      std::cout << "    → " << brewery.description << "\n";
+    }
+
+    std::cout << "\n✓ Pipeline completed successfully\n";
+
+    // Cleanup
+    curl_global_cleanup();
+    return 0;
+
+  } catch (const std::exception &e) {
+    std::cerr << "✗ Pipeline failed: " << e.what() << "\n";
+    curl_global_cleanup();
+    return 1;
+  }
+}