relaton-render 0.7.2 → 0.7.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 6b7a094cb951866a90ce67736e150d7723366ad8f4693ac62de82fbd90fccbdc
-  data.tar.gz: 5041e1e397d38d30c6e48b06fa0b78e09afb987c519393c7ba771a892cfb653f
+  metadata.gz: b64bd51a5487927e3c700f4658188e342593f85bd56abd0efb92b51d2329fe59
+  data.tar.gz: 21fb237d1d039c9c136bc5e3f64396437f007f690fdc1e54d557e40c722c98c5
 SHA512:
-  metadata.gz: 758a2f69ce87e0b258a0e7f5920f29f347a9fd311a11bb31199baaaaa8f7cd4d6fc46fcfdad5b8567072372b908a13fd1b1816056da0967cefb90bf6b79f0e84
-  data.tar.gz: a0fcbecba3f10f74b5a4dbd8ca056554c19527e5fa0277b39b007102cec98121bdfda1bc6211aa12cc8d69afbdc5e602a862015153e30b1658b4bb2a37e6f244
+  metadata.gz: ab48727b2054c7d8f665d409545ff07912de7170b9fdaf539bc8788b4185f9a1c7a02a90c068f00980ab98edba5c05ea1e4c60024b13c3a4e0fc855bc34be6a5
+  data.tar.gz: 769716704f5d77011d281ae8eac4abc6c4332763ce5fa29171537c1c80781e2ad5e6424ef3d25721743c1c8ed2653c3cda8dacc978194814cdf6e9f2658eb845

data/lib/relaton/render/fields/date.rb

@@ -4,6 +4,8 @@ module Relaton
       def initialize(date, options)
         @date = date
         @r = options[:renderer]
+        @type = options[:type]
+        @bibitem = options[:bibitem]
       end
 
       def render

data/lib/relaton/render/fields/fields.rb

@@ -55,7 +55,7 @@ module Relaton
       def date_fields_format(hash)
         [%i(date date), %i(date_updated date_updated),
          %i(date_accessed date_accessed)].each do |k|
-          hash[k[0]] = dateformat(hash[k[1]], hash)
+          hash[k[0]] = dateformat(hash[k[1]], hash, k)
           k[0] == :date && hash[:type] != "standard" and
             hash[k[0]] ||= @r.i18n.get["no_date"]
         end
@@ -191,10 +191,12 @@ module Relaton
         range(hash)
       end
 
-      def dateformat(date, _hash)
+      def dateformat(date, hash, type)
         date.nil? and return nil
+        date.is_a?(String) and return date
         %i(from to on).each do |k|
-          date[k] = ::Relaton::Render::Date.new(date[k], renderer: @r).render
+          date[k] = @r.dateklass
+            .new(date[k], renderer: @r, bibitem: hash, type: type).render
         end
         date_range(date)
       end

data/lib/relaton/render/general/citations.rb

@@ -10,17 +10,58 @@ module Relaton
       # takes array of { id, type, author, date, ord, data_liquid }
       def render(ret)
         cites = citations(ret)
+        cites.each_value do |v|
+          v[:renderer] = @renderer.renderer(v[:type] || "misc")
+        end
+        enhance_data(cites)
         cites.each_key do |k|
           cites[k] = render1(cites[k])
         end
         cites
       end
 
+      def enhance_data(cites)
+        ret = extract_uris_for_lookup(cites)
+        ret.empty? and return
+        @renderer.urls_exist_concurrent(ret.keys).each do |k, v|
+          ret[k].each { |u| add_date_accessed(cites[u], k, v) }
+        end
+      end
+
+      def extract_uris_for_lookup(cites)
+        cites.each_with_object({}) do |(k, v), m|
+          u = extract_uri_for_lookup(v) or next
+          m[u] ||= []
+          m[u] << k
+        end
+      end
+
+      def extract_uri_for_lookup(cite)
+        t = cite[:renderer].template_raw
+        c = cite[:data_liquid]
+        t.is_a?(String) or return
+        (/\{\{\s*date_accessed\s*\}\}/.match?(t) &&
+        /\{\{\s*uri\s*\}\}/.match?(t) &&
+        c[:uri_raw] && !c[:date_accessed]) or return
+        c[:uri_raw]
+      end
+
+      def add_date_accessed(data, uri, status)
+        if status
+          data[:data_liquid][:date_accessed] = { on: ::Date.today.to_s }
+          data[:data_liquid] = @renderer.fieldsklass.new(renderer: @renderer)
+            .compound_fields_format(data[:data_liquid])
+        else
+          @renderer.url_warn(uri)
+        end
+      end
+
       def render1(cit)
-        r = @renderer.renderer(cit[:type] || "misc")
         cit[:formattedref] =
-          @renderer.valid_parse(@i18n.l10n(r.render(cit[:data_liquid])))
-        %i(type data_liquid).each { |x| cit.delete(x) }
+          @renderer.valid_parse(
+            @i18n.l10n(cit[:renderer].render(cit[:data_liquid])),
+          )
+        %i(type data_liquid renderer).each { |x| cit.delete(x) }
         cit
       end
 
@@ -55,7 +96,7 @@ module Relaton
       def sort_ord(ret)
         ret.each do |author, v|
           v.each_key do |date|
-            ret[author][date].sort! { |a, b| a[:ord] <=> b[:ord] }
+            ret[author][date].sort_by! { |a| a[:ord] }
           end
         end
       end
@@ -82,7 +123,7 @@ module Relaton
 
       def to_hash(ret)
         ret.each_with_object({}) do |(_k, v), m|
-          v.each do |_k1, v1|
+          v.each_value do |v1|
             v1.each do |b|
               m[b[:id]] = { author: b[:author], date: b[:date],
                             citation: "#{b[:author]} #{b[:date]}",

data/lib/relaton/render/general/render.rb

@@ -1,10 +1,10 @@
 require_relative "render_classes"
 require_relative "citations"
+require_relative "uri"
 require "yaml"
 require "liquid"
 require "date"
 require "relaton_bib"
-require "net/http"
 require_relative "../template/template"
 require_relative "../../../isodoc/i18n"
 
@@ -14,7 +14,7 @@ module Relaton
       attr_reader :template, :journaltemplate, :seriestemplate, :nametemplate,
                   :authorcitetemplate, :extenttemplate, :sizetemplate,
                   :lang, :script, :i18n,
-                  :edition, :edition_ordinal, :date
+                  :edition, :edition_ordinal, :date, :fieldsklass, :dateklass
 
       def initialize(opt = {})
         options = read_config.merge(Utils::string_keys(opt))
@@ -23,7 +23,9 @@ module Relaton
         root_initalize(options)
         render_initialize(options)
         @parse ||= options["parse"]
+        @semaphore = Mutex.new
         @urlcache = {}
+        @url_warned = {}
       end
 
       def read_config
@@ -38,6 +40,7 @@ module Relaton
         @sizetemplateklass = Relaton::Render::Template::Size
         @generaltemplateklass = Relaton::Render::Template::General
         @fieldsklass = Relaton::Render::Fields
+        @dateklass = Relaton::Render::Date
         @parseklass = Relaton::Render::Parse
       end
 
@@ -133,7 +136,6 @@ module Relaton
           raise "No renderer defined for #{type}"
         @type == "general" || @type == type or
           raise "No renderer defined for #{type}"
-
         ret
       end
 
@@ -169,6 +171,7 @@ module Relaton
       end
 
       # expect array of Relaton objects, in sorted order
+      # enhance_data is skipped here, and is done in batch inside Citations
       def render_all(bib, type: "author-date")
         bib = sanitise_citations_input(bib) or return
         Citations.new(type: type, renderer: self, i18n: @i18n)
@@ -183,7 +186,6 @@ module Relaton
       def sanitise_citations_input_string(bib)
         p = Nokogiri::XML(bib) or return
         (p.errors.empty? && p.root.at("./bibitem")) or return nil
-
         p.root.xpath("./bibitem").each_with_object([]) do |b, m|
           m << RelatonBib::XMLParser.from_xml(b.to_xml)
         end
@@ -210,11 +212,16 @@ module Relaton
           data[:uri_raw] && !data[:date_accessed]) or return
         if url_exist?(data[:uri_raw])
           data[:date_accessed] = { on: ::Date.today.to_s }
-        else
-          warn "BIBLIOGRAPHY WARNING: cannot access #{data[:uri_raw]}"
+        else url_warn(data[:uri_raw])
         end
       end
 
+      def url_warn(uri)
+        @url_warned[uri] and return
+        warn "BIBLIOGRAPHY WARNING: cannot access #{uri}"
+        @url_warned[uri] = true
+      end
+
       private
 
       def template_hash_fill(templates)
@@ -224,35 +231,6 @@ module Relaton
           m[type] = template
         end
       end
-
-      def url_exist?(url_string)
-        url = URI.parse(url_string)
-        url.host or return true # allow file URLs
-        res = access_url(url) or return false
-        res.is_a?(Net::HTTPRedirection) and return url_exist?(res["location"])
-        res.code[0] != "4"
-      rescue Errno::ENOENT, SocketError
-        false # false if can't find the server
-      end
-
-      def access_url(url)
-        path = url.path or return false
-        path.empty? and path = "/"
-        @urlcache.key?(url.to_s) and return @urlcache[url.to_s]
-        @urlcache[url.to_s] = url_head(url, path)
-        @urlcache[url.to_s]
-      rescue => e
-        warn e.backtrace
-        false
-      end
-
-      def url_head(url, path)
-        Net::HTTP.start(url.host, url.port,
-                        read_timeout: 2, open_timeout: 2,
-                        use_ssl: url.scheme == "https") do |http|
-          http.request_head(path)
-        end
-      end
     end
   end
 end

data/lib/relaton/render/general/uri.rb

@@ -0,0 +1,56 @@
+require "concurrent"
+require "net/http"
+require "uri"
+
+module Relaton
+  module Render
+    class General
+      def url_exist?(url_string)
+        url = URI.parse(url_string)
+        url.host or return true # allow file URLs
+        res = access_url(url) or return false
+        res.is_a?(Net::HTTPRedirection) and return url_exist?(res["location"])
+        res.code[0] != "4"
+      rescue Errno::ENOENT, SocketError
+        false # false if can't find the server
+      end
+
+      def access_url(url)
+        path = url.path or return false
+        path.empty? and path = "/"
+        url_head(url, path)
+      rescue StandardError => e
+        warn e
+        false
+      end
+
+      def url_head(url, path)
+        ret = nil
+        @semaphore.synchronize { ret = @urlcache[url.to_s] }
+        ret and return ret
+        ret = Net::HTTP.start(url.host, url.port,
+                              read_timeout: 2, open_timeout: 2,
+                              use_ssl: url.scheme == "https") do |http|
+          http.request_head(path)
+        end
+        @semaphore.synchronize { @urlcache[url.to_s] = ret }
+        ret
+      end
+
+      def urls_exist_concurrent(urls)
+        responses = Concurrent::Array.new
+        thread_pool = Concurrent::FixedThreadPool.new(5)
+        urls.each do |u|
+          thread_pool.post { responses << url_exist_async?(u) }
+        end
+        thread_pool.shutdown
+        thread_pool.wait_for_termination
+        responses.each_with_object({}) { |n, m| m[n[:url]] = n[:status] }
+      end
+
+      def url_exist_async?(url_string)
+        { url: url_string, status: url_exist?(url_string) }
+      end
+    end
+  end
+end

data/lib/relaton/render/version.rb

@@ -1,5 +1,5 @@
 module Relaton
   module Render
-    VERSION = "0.7.2".freeze
+    VERSION = "0.7.3".freeze
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-render
 version: !ruby/object:Gem::Version
-  version: 0.7.2
+  version: 0.7.3
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2024-03-05 00:00:00.000000000 Z
+date: 2024-03-06 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -213,6 +213,7 @@ files:
 - lib/relaton/render/general/config.yml
 - lib/relaton/render/general/render.rb
 - lib/relaton/render/general/render_classes.rb
+- lib/relaton/render/general/uri.rb
 - lib/relaton/render/parse/parse.rb
 - lib/relaton/render/parse/parse_contributors.rb
 - lib/relaton/render/parse/parse_extract.rb