alexandria-book-collection-manager 0.7.5 → 0.7.6

data/lib/alexandria/book_providers/barnes_and_noble.rb

@@ -35,7 +35,7 @@ require "alexandria/book_providers/web"
 module Alexandria
   class BookProviders
     class BarnesAndNobleProvider < WebsiteBasedProvider
-      include Alexandria::Logging
+      include Logging
 
       SITE = "http://www.barnesandnoble.com"
 
@@ -56,7 +56,7 @@ module Alexandria
       end
 
       def fetch_redirectly(uri_str, limit = 5)
-        raise NoResultsError, "HTTP redirect too deep" if limit.zero?
+        raise NoResultsError, _("HTTP redirect too deep") if limit.zero?
 
         if limit < 10
           sleep 0.1
@@ -67,8 +67,8 @@ module Alexandria
         response = agent.get(uri_str)
         log.debug { response.inspect }
         case response
-        when Net::HTTPSuccess     then response
-        when Net::HTTPRedirection then
+        when Net::HTTPSuccess then response
+        when Net::HTTPRedirection
           redirect = URI.parse response["Location"]
           redirect = URI.parse(uri_str) + redirect if redirect.relative?
           fetch_redirectly(redirect.to_s, (limit - 1))
@@ -79,7 +79,7 @@ module Alexandria
 
       def search(criterion, type)
         req = create_search_uri(type, criterion)
-        puts "Requesting #{req}" if $DEBUG
+        log.debug { "Requesting #{req}" }
         html_data = fetch_redirectly(req)
 
         if type == SEARCH_BY_ISBN
@@ -204,6 +204,6 @@ module Alexandria
           raise NoResultsError
         end
       end
-    end # class BarnesAndNobleProvider
-  end # class BookProviders
-end # module Alexandria
+    end
+  end
+end

data/lib/alexandria/book_providers/douban.rb

@@ -16,7 +16,7 @@ require "yaml"
 module Alexandria
   class BookProviders
     class DoubanProvider < GenericProvider
-      include Alexandria::Logging
+      include Logging
 
       SITE = "http://www.douban.com"
       BASE_URL = "http://api.douban.com/book/subjects?q=%s&max-results=5&alt=json"
@@ -58,7 +58,7 @@ module Alexandria
       def json2yaml(json)
         # insert spaces after : and , except within strings
         # i.e. when followed by numeral, quote, { or [
-        yaml = json.gsub(/(\:|\,)([0-9'"{\[])/) do |_match|
+        yaml = json.gsub(/(:|,)([0-9'"{\[])/) do |_match|
           "#{Regexp.last_match[1]} #{Regexp.last_match[2]}"
         end
         yaml.gsub!(%r{\\/}, "/") # unescape forward slashes

data/lib/alexandria/book_providers/proxis.rb

@@ -14,7 +14,7 @@ module Alexandria
   class BookProviders
     class ProxisProvider < WebsiteBasedProvider
       # include GetText
-      include Alexandria::Logging
+      include Logging
       # GetText.bindtextdomain(Alexandria::TEXTDOMAIN, :charset => "UTF-8")
 
       # Proxis essentially has three book databases, NL, FR and EN.
@@ -37,7 +37,7 @@ module Alexandria
 
       def search(criterion, type)
         req = create_search_uri(type, criterion)
-        puts req if $DEBUG
+        log.debug { req }
         html_data = transport.get_response(URI.parse(req))
 
         results = parse_search_result_data(html_data.body)
@@ -96,7 +96,7 @@ module Alexandria
           if title_link
             result[:title] = text_of(title_link)
             result[:lookup_url] = title_link["href"]
-            unless /^http/.match?(result[:lookup_url])
+            unless result[:lookup_url].start_with?("http")
               result[:lookup_url] = "#{SITE}#{result[:lookup_url]}"
             end
           end
@@ -108,8 +108,8 @@ module Alexandria
         book_search_results
       end
 
-      def data_for_header(th)
-        tr = th.parent
+      def data_for_header(header)
+        tr = header.parent
         td = tr.at("td")
         text_of(td) if td
       end
@@ -141,15 +141,16 @@ module Alexandria
         unless info_headers.empty?
           info_headers.each do |th|
             header_text = th.inner_text
-            if /Type/.match?(header_text)
+            case header_text
+            when /Type/
               book_data[:binding] = data_for_header(th)
-            elsif /Verschijningsdatum/.match?(header_text)
+            when /Verschijningsdatum/
               date = data_for_header(th)
-              date =~ %r{/([\d]{4})}
+              date =~ %r{/(\d{4})}
               book_data[:publish_year] = Regexp.last_match[1].to_i
-            elsif /Auteur/.match?(header_text)
+            when /Auteur/
               book_data[:authors] << data_for_header(th)
-            elsif /Uitgever/.match?(header_text)
+            when /Uitgever/
               book_data[:publisher] = data_for_header(th)
             end
           end
@@ -157,7 +158,7 @@ module Alexandria
 
         image_url = nil
         if (cover_img = doc.at("img[@id$='imgProduct']"))
-          image_url = if /^http/.match?(cover_img["src"])
+          image_url = if cover_img["src"].start_with?("http")
                         cover_img["src"]
                       else
                         "#{SITE}/#{cover_img['src']}" # TODO: use html <base>

data/lib/alexandria/book_providers/pseudomarc.rb

@@ -41,9 +41,9 @@ module Alexandria
       notes: ["520", "a"]
     }.freeze
 
-    def self.get_fields(data, type, stripping, m = USMARC_MAPPINGS)
+    def self.get_fields(data, type, stripping, mappings = USMARC_MAPPINGS)
       field = ""
-      m[type][1..m[type].length - 1].each do |part|
+      mappings[type][1..mappings[type].length - 1].each do |part|
         if data.first[part]
           part_data = data.first[part].strip
           if part_data =~ stripping
@@ -58,69 +58,68 @@ module Alexandria
       field
     end
 
-    def self.marc_text_to_book(marc, m = USMARC_MAPPINGS)
+    def self.marc_text_to_book(marc, mappings = USMARC_MAPPINGS)
       details = marc_text_to_details(marc)
-      unless details.empty?
-        title = nil
-        title_data = details[m[:title][0]]
-        if title_data
-          title_data_all = get_fields(title_data, :title, %r{(.*)[/:]$}, m)
-          title = title_data_all if title_data_all
-        end
+      return if details.empty?
 
-        authors = []
-        author_data = details[m[:authors][0]]
-        author_data&.each do |ad|
-          author = ad[m[:authors][1]]
-          if author
-            author = author.strip
-            author = Regexp.last_match[1] if author =~ /(.*),$/
-            authors << author
-          end
-        end
+      title = nil
+      title_data = details[mappings[:title][0]]
+      if title_data
+        title_data_all = get_fields(title_data, :title, %r{(.*)[/:]$}, mappings)
+        title = title_data_all if title_data_all
+      end
 
-        isbn = nil
-        binding = nil
-        isbn_data = details[m[:isbn][0]]
-        if isbn_data
-          isbn = Regexp.last_match[1] if isbn_data.first[m[:isbn][1]] =~ /([-0-9xX]+)/
+      authors = []
+      author_data = details[mappings[:authors][0]]
+      author_data&.each do |ad|
+        author = ad[mappings[:authors][1]]
+        if author
+          author = author.strip
+          author = Regexp.last_match[1] if author =~ /(.*),$/
+          authors << author
         end
+      end
 
-        binding_data = details[m[:binding][0]]
-        if binding_data
-          if binding_data.first[m[:binding][1]] =~ /([a-zA-Z][a-z\s]+[a-z])/
-            binding = Regexp.last_match[1]
-          end
-        end
+      isbn = nil
+      binding = nil
+      isbn_data = details[mappings[:isbn][0]]
+      if isbn_data && isbn_data.first[mappings[:isbn][1]] =~ /([-0-9xX]+)/
+        isbn = Regexp.last_match[1]
+      end
 
-        publisher = nil
-        publisher_data = details[m[:publisher][0]]
-        publisher = publisher_data.first[m[:publisher][1]] if publisher_data
+      binding_data = details[mappings[:binding][0]]
+      if binding_data &&
+          binding_data.first[mappings[:binding][1]] =~ /([a-zA-Z][a-z\s]+[a-z])/
+        binding = Regexp.last_match[1]
+      end
 
-        year = nil
-        publication_data = details[m[:year][0]]
-        if publication_data
-          year = publication_data.first[m[:year][1]]
-          year = Regexp.last_match[1].to_i if year =~ /(\d+)/
-        end
+      publisher = nil
+      publisher_data = details[mappings[:publisher][0]]
+      publisher = publisher_data.first[mappings[:publisher][1]] if publisher_data
 
-        notes = ""
-        notes_data = details[m[:notes][0]]
-        notes_data&.each do |note|
-          txt = note[m[:notes][1]]
-          notes += txt if txt
-        end
+      year = nil
+      publication_data = details[mappings[:year][0]]
+      if publication_data
+        year = publication_data.first[mappings[:year][1]]
+        year = Regexp.last_match[1].to_i if year =~ /(\d+)/
+      end
 
-        if title.nil? && isbn.nil?
-          # probably didn't undertand the MARC dialect
-          return nil
-        end
+      notes = ""
+      notes_data = details[mappings[:notes][0]]
+      notes_data&.each do |note|
+        txt = note[mappings[:notes][1]]
+        notes += txt if txt
+      end
 
-        book = Alexandria::Book.new(title, authors, isbn,
-                                    publisher, year, binding)
-        book.notes = notes unless notes.empty?
-        book
+      if title.nil? && isbn.nil?
+        # probably didn't undertand the MARC dialect
+        return nil
       end
+
+      book = Alexandria::Book.new(title, authors, isbn,
+                                  publisher, year, binding)
+      book.notes = notes unless notes.empty?
+      book
     end
 
     def self.marc_text_to_details(marc)
@@ -132,31 +131,22 @@ module Alexandria
 
           this_line_data = {}
 
-          # puts code
-          # puts data
           d_idx = 0
           while d_idx < data.size
             d_str = data[d_idx..-1]
-            # puts d_str
-            if (idx = d_str =~ /\$([a-z]) ([^\$]+)/)
-              # puts idx
-              sub_code = Regexp.last_match[1]
-              sub_data = Regexp.last_match[2]
-              this_line_data[sub_code] = sub_data
-              # puts "  " + $1
-              # puts "    " + $2
-              # puts idx
-              d_idx += idx + 2 # (2 extra to push beyond this '$a' etc.)
-            else
-              break
-            end
+            idx = d_str =~ /\$([a-z]) ([^$]+)/
+            break unless idx
+
+            sub_code = Regexp.last_match[1]
+            sub_data = Regexp.last_match[2]
+            this_line_data[sub_code] = sub_data
+            d_idx += idx + 2 # (2 extra to push beyond this '$a' etc.)
           end
 
           unless this_line_data.empty?
             details[code] = [] unless details.key?(code)
             details[code] << this_line_data
           end
-
         end
       end
       details

data/lib/alexandria/book_providers/siciliano.rb

@@ -75,9 +75,9 @@ module Alexandria
           if (type == SEARCH_BY_ISBN) && (trying_again == false)
             trying_again = true
             retry
-          else
-            raise ex
           end
+
+          raise ex
         end
       end
 
@@ -143,7 +143,7 @@ module Alexandria
           result[:title] = link.inner_text.strip
           link_to_description = link["href"]
           slash = ""
-          slash = "/" unless %r{^/}.match?(link_to_description)
+          slash = "/" unless link_to_description.start_with?("/")
           result[:url] = "#{SITE}#{slash}#{link_to_description}"
 
           book_search_results << result
@@ -194,15 +194,14 @@ module Alexandria
 
           script.children.each do |ch|
             ch_text = ch.to_s
-            if ch_text =~ /ImgSrc\[[\d]\]="(.+)";/
+            if ch_text =~ /ImgSrc\[\d\]="(.+)";/
               img_link = Regexp.last_match[1]
               image_urls << img_link
             end
           end
         end
         book = Book.new(title, authors, isbn, publisher, publish_year, binding)
-        result = [book, image_urls.first]
-        result
+        [book, image_urls.first]
       rescue StandardError => ex
         trace = ex.backtrace.join("\n> ")
         log.error { "Failed parsing Siciliano product page #{ex.message}\n#{trace}" }

data/lib/alexandria/book_providers/thalia.rb

@@ -31,7 +31,7 @@ require "alexandria/book_providers/web"
 module Alexandria
   class BookProviders
     class ThaliaProvider < WebsiteBasedProvider
-      include Alexandria::Logging
+      include Logging
 
       SITE = "http://www.thalia.de"
       BASE_SEARCH_URL = "#{SITE}/shop/bde_bu_hg_startseite/suche/?%s=%s" # type,term
@@ -48,7 +48,7 @@ module Alexandria
 
       def search(criterion, type)
         req = create_search_uri(type, criterion)
-        puts req if $DEBUG
+        log.debug { req }
         html_data = transport.get_response(URI.parse(req))
         if type == SEARCH_BY_ISBN
           parse_result_data(html_data.body, criterion)
@@ -130,11 +130,10 @@ module Alexandria
           # e.g. .../dave_thomas/ISBN0-9745140-5-5/ID6017044.html
           chosen = results.first # fallback!
           results.each do |rslt|
-            if rslt[:lookup_url] =~ %r{/ISBN(\d+[\d-]*)/}
-              if Regexp.last_match[1].delete("-") == isbn10
-                chosen = rslt
-                break
-              end
+            if rslt[:lookup_url] =~ %r{/ISBN(\d+[\d-]*)/} &&
+                (Regexp.last_match[1].delete("-") == isbn10)
+              chosen = rslt
+              break
             end
           end
           html_data = transport.get_response(URI.parse(chosen[:lookup_url]))
@@ -152,7 +151,7 @@ module Alexandria
               authors = []
               author_links = author_h.parent / :a
               author_links.each do |a|
-                if a["href"] =~ %r{BUCH/sa}
+                if a["href"].include? "BUCH/sa"
                   # 'sa' means search author, there may also be 'ssw' (search keyword) links
                   authors << a.inner_text[0..-2].strip
                   # NOTE stripping the little >> character here...
@@ -167,7 +166,7 @@ module Alexandria
 
             year = nil
             date = data_from_label(item_details, "Erschienen:")
-            year = Regexp.last_match[1].to_i if date =~ /([\d]{4})/
+            year = Regexp.last_match[1].to_i if date =~ /(\d{4})/
 
             binding = data_from_label(item_details, "Einband")
 

data/lib/alexandria/book_providers/worldcat.rb

@@ -22,7 +22,7 @@ require "alexandria/book_providers/web"
 module Alexandria
   class BookProviders
     class WorldCatProvider < WebsiteBasedProvider
-      include Alexandria::Logging
+      include Logging
 
       SITE = "https://www.worldcat.org"
       BASE_SEARCH_URL = "#{SITE}/search?q=%s%s&qt=advanced" # type, term
@@ -77,10 +77,9 @@ module Alexandria
         book_search_results = []
         begin
           result_cells = doc / "td.result/div.name/.."
-          # puts result_cells.length
           result_cells.each do |td|
             type_icon = (td % "div.type/img.icn")
-            next unless type_icon && type_icon["src"] =~ /icon-bks/
+            next unless type_icon && type_icon["src"].include?("icon-bks")
 
             name_div = td % "div.name"
             title = name_div.inner_text
@@ -129,22 +128,20 @@ module Alexandria
               html2 = rslt2.body
 
               book, cover_url = parse_result_data(html2, search_isbn, true)
-              first_result = [book, cover_url] if first_result.nil?
 
               log.debug { "got book #{book}" }
 
-              if search_isbn
-                search_isbn_canon = Library.canonicalise_ean(search_isbn)
-                rslt_isbn_canon = Library.canonicalise_ean(book.isbn)
-                if search_isbn_canon == rslt_isbn_canon
-                  log.info { "book #{book} is a match" }
-                  return [book, cover_url]
-                end
-                log.debug { "not a match, checking next" }
-              else
-                # no constraint to match isbn, just return first result
+              return [book, cover_url] unless search_isbn
+
+              first_result = [book, cover_url] if first_result.nil?
+
+              search_isbn_canon = Library.canonicalise_ean(search_isbn)
+              rslt_isbn_canon = Library.canonicalise_ean(book.isbn)
+              if search_isbn_canon == rslt_isbn_canon
+                log.info { "book #{book} is a match" }
                 return [book, cover_url]
               end
+              log.debug { "not a match, checking next" }
             end
 
             # gone through all and no ISBN match, so just return first result
@@ -152,7 +149,6 @@ module Alexandria
               "no more results to check. Returning first result, just an approximation"
             end
             return first_result
-
           end
 
           title_header = doc % "h1.title"
@@ -180,9 +176,9 @@ module Alexandria
             publication_info = (publisher_row / "td").last.inner_text
 
             publication_info =~ if publication_info.index(";")
-                                  /;[\s]*([^\d]+)[\s]*[\d]*/
+                                  /;\s*([^\d]+)\s*\d*/
                                 elsif publication_info.index(":")
-                                  /:[\s]*([^;:,]+)/
+                                  /:\s*([^;:,]+)/
                                 else
                                   /([^;,]+)/
                                 end
@@ -195,20 +191,18 @@ module Alexandria
             year = nil
           end
 
-          isbn = search_isbn
-          unless isbn
-            isbn_row = doc % "tr#details-standardno"
-            if isbn_row
-              isbns = (isbn_row / "td").last.inner_text.split
-              isbn = Library.canonicalise_isbn(isbns.first)
-            else
-              log.warn { "No ISBN found on page" }
-            end
+          isbn_row = doc % "tr#details-standardno"
+          if isbn_row
+            isbns = (isbn_row / "td").last.inner_text.split
+            isbn = Library.canonicalise_isbn(isbns.first)
+          else
+            log.warn { "No ISBN found on page" }
+            isbn = search_isbn
           end
 
-          binding = "" # not given on WorldCat website (as far as I can tell)
+          book_binding = "" # not given on WorldCat website (as far as I can tell)
 
-          book = Book.new(title, authors, isbn, publisher, year, binding)
+          book = Book.new(title, authors, isbn, publisher, year, book_binding)
 
           image_url = nil # hm, it's on the website, but uses JavaScript...
 
@@ -224,6 +218,6 @@ module Alexandria
           raise NoResultsError
         end
       end
-    end # class WorldCatProvider
-  end # class BookProviders
-end # module Alexandria
+    end
+  end
+end