zipcode-fr 0.1.0 → 1.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 49ba8c551db2a645e35f59f485073ea85ec78c43
-  data.tar.gz: b4079b54250461c4b1d6a2c9b1a6b2724d7ace87
+  metadata.gz: 2b951d8164fd96461f3b1c4549c231867c77d4bc
+  data.tar.gz: 27f411e06e2fefa64d9e84098c0d84cc0e23bd2e
 SHA512:
-  metadata.gz: e39eef729b7634e97b4441af2abe6e70314bc078603fe295ecab2380672dee6c8f06dfb125c7dee72d8305d7ae8c43aa0cef8b61c32570c5f0c48a26d0d0a3f5
-  data.tar.gz: 616c5eb703167666cbae11c086bf3534fbb4ea258ee9352852588a9aab658a0a7eeb2256d200427e336f307ae99dd162cff9ed2284110d0e8426ab8ea5dd8d57
+  metadata.gz: fe56c6094a32e80322b04d4b0aaf6dc1dc91679ed14c49f9c6393d79290185f83dfa4cffdafa088d73843716b3dcf15a282f939e1fbff0c45486105798567532
+  data.tar.gz: 6a6705c55e4846bfadc5ba5f3e269e1e1917fbbcd23bfe771cc393a80f01e9696d42066dfb5f9955364c3c7433d127c117d4c0dd007d8cee74c98eff1481ff5d

data/lib/zipcode-fr.rb

@@ -1,4 +1,7 @@
 module ZipCode
+  # TODO: factor index system out
+  # TODO: factor country-independent code out
+  # rubocop:disable Metrics/ModuleLength
   module FR
     require 'csv'
 
@@ -8,8 +11,8 @@ module ZipCode
 
     def load
       # TODO: non-optimal, but not overly long either
-      index!(:name, reader)
-      index!(:zip, reader)
+      index!(:name, reader, [:word_prefix, :match])
+      index!(:zip, reader, :prefix)
       @loaded = true
     end
 
@@ -57,20 +60,77 @@ module ZipCode
       [:insee, :name, :zip, :alt_name].zip(row).to_h
     end
 
-    def index!(name, data, key = nil)
+    def index!(name, data, modes = nil, key: nil)
       key ||= name
       index = Hash.new { |h, k| h[k] = [] unless h.frozen? }
 
-      data.each do |pos, record|
-        val = record[key]
-        val.length.times { |i| index[val[0..i].hash] << pos }
+      modes = [modes] unless modes.is_a?(Enumerable)
+      modes.each do |mode|
+        data.each(&appender(index, key, mode))
       end
 
+      index.each { |_, v| v.uniq! }
       index.freeze
 
       @indexes[name] = index
     end
 
+    # TODO: create an appender registry
+    # rubocop:disable Metrics/AbcSize
+    # rubocop:disable Metrics/MethodLength
+    private def appender(idx, key, mode)
+      case mode
+      when :prefix
+        -> (pos, record) { append_prefixes(idx, pos, record[key]) }
+      when :infix
+        -> (pos, record) { append_infixes(idx, pos, record[key]) }
+      when :word
+        -> (pos, record) { append_words(idx, pos, record[key]) }
+      when :word_prefix
+        -> (pos, record) { append_word_prefixes(idx, pos, record[key]) }
+      else
+        -> (pos, record) { append_match(idx, pos, record[key]) }
+      end
+    end
+
+    private def append_match(idx, pos, val)
+      idx[val.hash] << pos
+    end
+
+    private def append_words(idx, pos, val)
+      each_word(val) { |w| idx[w.hash] << pos }
+    end
+
+    private def append_word_prefixes(idx, pos, val)
+      each_word(val) do |word|
+        each_prefix(word) { |prefix| idx[prefix.hash] << pos }
+      end
+    end
+
+    private def append_prefixes(idx, pos, val, min_size: 1)
+      each_prefix(val, min_size: min_size) { |prefix| idx[prefix.hash] << pos }
+    end
+
+    private def each_word(val, &block)
+      val.split.each(&block)
+    end
+
+    private def each_prefix(val, min_size: 1)
+      min_size.upto(val.length) { |i| yield val[0...i] }
+    end
+
+    private def each_suffix(val, min_size: 1)
+      min_size.upto(val.length) { |i| yield val[-i..-1] }
+    end
+
+    private def append_infixes(idx, pos, val, min_size: 1)
+      each_prefix(val, min_size: min_size) do |prefix|
+        each_suffix(prefix, min_size: min_size) do |infix|
+          idx[infix.hash] << pos
+        end
+      end
+    end
+
     private def index(name)
       if @indexes.key?(name)
         @indexes[name]
@@ -79,6 +139,12 @@ module ZipCode
       end
     end
 
+    def memsize_of_index(name)
+      require 'objspace'
+      ObjectSpace.memsize_of(@indexes[name]) +
+        @indexes[name].reduce(0) { |a, (_, v)| a + ObjectSpace.memsize_of(v) }
+    end
+
     private def read_at(*positions, count: 1)
       Enumerator.new do |y|
         open do |io|