git-fit 0.8.9 → 0.9.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 9aedbd56b7a0006cad60386fa55b030857cef43497e0d798bb334f6ddaf968fb
-  data.tar.gz: 526692de645f4cee05b8b61a3bc3018efbaad202c2d342b7376e7a90e1b1f237
+  metadata.gz: 0f2c48812a658488f860a9e483756ecd635b078b8f94deaf0cc20e31ab5caef9
+  data.tar.gz: 3a1b15b292e344df1a5a503fd3f2160e0e66ca5f3293c7449144164ca84ce53a
 SHA512:
-  metadata.gz: 10c0908e780b6c8f01cc32af8ed20bcc344f197edcf2daeb2f799f99cad78433a7159d15e0b7dffe51bc654fd76c75ebf6e29152a9e8500aa760fb4a85d10b84
-  data.tar.gz: e36c0f4b019e587640e220704cc82a05eb4533db5314fabfc9ea32771d0493c198df90fdf8b490ee1293924a39630dc8a8721abde3e4ebbe8233f8e722325f2e
+  metadata.gz: 0a6e1dec451ae23ddfc9333258f25c528db2d467dcd2b4dfd8f04ada2decaaa9d81c9d3094010c264c82a51c9e001bac7d33a761c75865dafadcc4207ddbe003
+  data.tar.gz: 4157de7684b83443b6b825af613e2ed0e74b8e4c2bac9378fe0edf0f3329a465ca0e1e80c540fed437d1f88825232b96a480e2fd8bbd6c210e6bad5d0319e7c4

data/lib/git_fit/cli/geo_cli.rb

@@ -11,6 +11,7 @@ module GitFit
     option :time,     type: :numeric, aliases: '-t', desc: 'Time budget in seconds'
     option :"dry-run", type: :boolean, desc: "Don't write to DB"
     option :checkpoint, type: :boolean, desc: 'Run WAL checkpoint after detect'
+    option :"cache-dir", type: :string, desc: "Geo cache directory (empty = no cache)", default: "data/cache"
     def detect
       config = GitFit::Config.new
         conn = GitFit::DB::Connection.new(config.db_path)
@@ -25,6 +26,7 @@ module GitFit
         strategy: options[:strategy],
         time_budget: options[:time],
         dry_run: options[:"dry-run"],
+        cache_dir: options[:"cache-dir"],
       )
       detector.run
       checkpoint_db(db) if options[:checkpoint]

data/lib/git_fit/geo/division_detector.rb

@@ -10,10 +10,9 @@ module GitFit
       KEEP_PLACEHOLDER = 'gqqrFurkeU??'
       SAMPLE_INTERVAL_DEG = 0.01
       CACHE_DIR = 'data/cache'
-      POINT_CACHE_FILE = File.join(CACHE_DIR, 'geo-point.jsonl')
 
       def initialize(db:, amap_key: nil, batch: 20, limit: nil, strategy: 'proportional', time_budget: nil,
-                     dry_run: false)
+                     dry_run: false, cache_dir: CACHE_DIR)
         @db = db
         @geocoder = ReverseGeocode.new(amap_key: amap_key)
         @amap_key = amap_key
@@ -23,6 +22,7 @@ module GitFit
         @time_budget = time_budget
         @start_time = Time.now
         @dry_run = dry_run
+        @cache_dir = cache_dir
         @processed = 0
         @skipped_placeholder    = 0
         @skipped_empty_polyline = 0
@@ -32,7 +32,7 @@ module GitFit
         @errors = 0
         @attempted = 0
         @total = 0
-        FileUtils.mkdir_p(CACHE_DIR)
+        FileUtils.mkdir_p(@cache_dir) if @cache_dir && !@cache_dir.empty?
         load_caches
       end
 
@@ -159,10 +159,11 @@ module GitFit
         key_counts = Hash.new(0)
         samples.each do |lat, lng|
           coord_key = "#{lat},#{lng}"
-          raw = lookup_point_cache(coord_key)
+          provider = provider_for(lat, lng)
+          raw = lookup_point_cache(coord_key, provider)
           unless raw
             raw = @geocoder.lookup(lat, lng)
-            save_point_cache(coord_key, raw) if raw[:province]
+            save_point_cache(coord_key, raw, provider) if raw[:province]
           end
           result = normalize_entry(raw)
           next unless result
@@ -218,9 +219,19 @@ module GitFit
         { divisions: division_list, total_sampled: total_points, source: 'amap', processed_at: Time.now.utc.iso8601 }
       end
 
-      def lookup_point_cache(coord_key)
-        return nil unless File.exist?(POINT_CACHE_FILE)
-        File.open(POINT_CACHE_FILE, 'r') do |f|
+      def cache_path_for(provider)
+        return nil unless @cache_dir && !@cache_dir.empty?
+        File.join(@cache_dir, "#{provider}-points.jsonl")
+      end
+
+      def provider_for(lat, lng)
+        @geocoder.provider_name(lat, lng)
+      end
+
+      def lookup_point_cache(coord_key, provider)
+        path = cache_path_for(provider)
+        return nil unless path && File.exist?(path)
+        File.open(path, 'r') do |f|
           f.each_line do |line|
             entry = JSON.parse(line) rescue next
             return entry if entry['ck'] == coord_key
@@ -229,12 +240,14 @@ module GitFit
         nil
       end
 
-      def save_point_cache(coord_key, result)
+      def save_point_cache(coord_key, result, provider)
+        path = cache_path_for(provider)
+        return unless path
         entry = { ck: coord_key, p: result[:province], source: result[:source] }
         entry[:city] = result[:city] if result[:city]
         entry[:dist] = result[:district] if result[:district]
         entry[:town] = result[:township] if result[:township]
-        File.open(POINT_CACHE_FILE, 'a') { |f| f.puts(JSON.generate(entry)) }
+        File.open(path, 'a') { |f| f.puts(JSON.generate(entry)) }
       end
 
       def update_db(run_id, summary)

data/lib/git_fit/geo/reverse_geocode.rb

@@ -25,6 +25,10 @@ module GitFit
         in_china?(lat, lng) ? lookup_china(lat, lng) : lookup_intl(lat, lng)
       end
 
+      def provider_name(lat, lng)
+        in_china?(lat, lng) ? :amap : :nominatim
+      end
+
       private
 
       def in_china?(lat, lng)

data/lib/git_fit/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module GitFit
-  VERSION = '0.8.9'
+  VERSION = '0.9.0'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: git-fit
 version: !ruby/object:Gem::Version
-  version: 0.8.9
+  version: 0.9.0
 platform: ruby
 authors:
 - Lax