Initial implementatin of busqueda in crystal

c-busqueda/handler.cr

@@ -0,0 +1,216 @@
+require "http/request"
+require "http/headers"
+require "ishi/html"
+require "json"
+
+class Handler
+  def format_buffer(buffer, canvas_name)
+    # Process the gnuplot output so it works in the page
+    #
+    # buffer is the Ishi output
+    # name is a string to replace for gnuplot_canvas so
+    # we can have multiple charts in a page
+
+    html = buffer.to_s.split("\n")
+    html = html[html.index("<script type=\"text/javascript\">")..html.index("</script>")]
+    html = html.join("\n") + %(
+    <div class="gnuplot">
+    <canvas id="Tile" width="32" height="32" hidden></canvas>
+    <table class="plot">
+    <tr><td>
+    <canvas id="gnuplot_canvas" width="800" height="300" tabindex="0">
+    Sorry, your browser seems not to support the HTML 5 canvas element
+    </canvas>
+    </td></tr>
+    </table>
+    <script type="text/javascript" defer>
+    gnuplot.init(); gnuplot_canvas();
+    </script>
+    </div>
+    )
+    # This ID needs to be unique in case 
+    # we have 2 charts in the same page
+    html.gsub("gnuplot_canvas", canvas_name)
+  end
+
+  def query(sql)
+    # Runs a SQL query against the Rqlite database.
+    #
+    # Returns an array of values (which need to be casted)
+    # Or nil if there are no results
+
+    params = URI::Params.encode({"q": sql})
+    response = HTTP::Client.get URI.new(
+      "http",
+      "10.61.0.1",
+      4001,
+      "/db/query",
+      params)
+
+    # This API only has a values key when there are actual results
+    results = JSON.parse(response.body)["results"][0].as_h
+    if results.has_key?("values")
+      return results["values"].as_a
+    end
+    # No result, return nil
+    nil
+  end
+
+  def normalize_name(s)
+    # Remove diacritics, turn lowercase
+    normalized = s.unicode_normalize(:nfkd).chars
+    normalized.reject! { |c|
+      !c.ascii_letter?
+    }.join("").downcase
+  end
+
+  def feminidad(nombre)
+    nombre = nombre.to_s.upcase
+    sql1 = %(
+      SELECT COALESCE(frecuencia,0)
+      FROM mujeres WHERE nombre='#{nombre}'
+
+    )
+    sql2 = %(
+    SELECT COALESCE(frecuencia,0)
+    FROM hombres WHERE nombre='#{nombre}'
+    )
+
+    # Yes this database is upper case
+    mujeres = query(sql1)
+    mujeres = mujeres.nil? ? 0 : mujeres[0][0].as_i
+    hombres = query(sql2)
+    hombres = hombres.nil? ? 0 : hombres[0][0].as_i
+    if hombres == mujeres == 0
+      return 0.5
+    end
+    return mujeres / (hombres + mujeres)
+  end
+
+  def split_por_genero(nombres)
+    femeninos = Array(Array(String | Int32)).new
+    masculinos = Array(Array(String | Int32)).new
+    nombres.map { |n|
+      fem = feminidad(n[1])
+      # El overlap en 0.5 es intencional!
+      if fem >= 0.5
+        femeninos << n
+      end
+      if fem <= 0.5
+        masculinos << n
+      end
+    }
+    {
+      "f": femeninos,
+      "m": masculinos,
+    }
+  end
+
+  def run(request : HTTP::Request)
+    unless (body = request.body).nil?
+      query = Hash(String, String).from_json(body)
+    else
+      query = {"p": "", "g": "", a: ""}
+    end
+
+    p! query
+
+    # Sanitize input.
+    # Each one either a valid string or nil
+    prefijo = query.fetch("p", "")
+    genero = query.fetch("g", "")
+    año = query.fetch("a", "")
+
+    if !prefijo.empty?
+      prefijo = normalize_name(prefijo)
+    else
+      prefijo = nil
+    end
+
+    if !["f", "m"].includes?(genero)
+      genero = nil
+    end
+
+    if año.empty?
+      año = nil
+      # TODO: check for valid integer
+    end
+
+    if prefijo.nil? && año.nil?
+      # Global totals
+      sql = %(
+      SELECT total, nombre
+      FROM totales
+      ORDER BY total DESC
+      LIMIT 50
+      )
+    elsif prefijo.nil? && !año.nil?
+      # Per-year totals
+      sql = %(
+        SELECT contador, nombre
+        FROM nombres
+        WHERE
+            anio = '#{año}'
+        ORDER BY contador DESC
+        LIMIT 50
+      )
+    elsif !prefijo.nil? && año.nil?
+      # Filter only by prefix
+      sql = %(
+        SELECT total, nombre
+        FROM totales
+        WHERE
+            nombre LIKE '#{prefijo}%'
+        ORDER BY total DESC
+        LIMIT 50
+      )
+    else
+      # We have all the filters
+      sql = %(
+        SELECT contador, nombre
+        FROM nombres
+        WHERE
+            anio = '#{año}' AND
+            nombre LIKE '#{prefijo}%'
+        ORDER BY contador DESC
+        LIMIT 50
+      )
+    end
+    results = query(sql)
+
+    if results.nil?
+      # This is bad 😀
+      return {
+        body:        "Que raro, no tengo *idea*!",
+        status_code: 200,
+        headers:     HTTP::Headers{"Content-Type" => "text/html"},
+      }
+    end
+    datos = results.map { |r|
+      [r[0].as_i, r[1].as_s]
+    }
+
+    if genero
+      datos = split_por_genero(datos)[genero]
+    end
+
+    datos = datos[..10]
+
+    buffer = IO::Memory.new
+    Ishi.new(buffer) do
+      x = (1..10).to_a
+      y = (1..10).to_a
+      canvas_size(800, 300)
+      plot(x,y, style: :boxes, fs: 0.25)
+      .boxwidth(0.5)
+      .ylabel("Popularidad")
+      .xlabel("Nombre")    
+    end
+
+    {
+      body:        format_buffer(buffer, "busqueda"),
+      status_code: 200,
+      headers:     HTTP::Headers{"Content-Type" => "text/html"},
+    }
+  end
+end

c-busqueda/shard.yml

@@ -0,0 +1,6 @@
+name: busqueda
+version: 0.1.0
+
+dependencies:
+    ishi:
+      github: toddsundsted/ishi

functions.yml

@@ -21,3 +21,10 @@ functions:
     image: ralsina/c-historico:latest
     build_args:
       ADDITIONAL_PACKAGE: gnuplot
+  c-busqueda:
+    lang: crystal-http
+    handler: ./c-busqueda
+    image: ralsina/c-busqueda:latest
+    build_args:
+      ADDITIONAL_PACKAGE: gnuplot
+