bookshark 1.0.0.pre.2 → 1.0.1

data/lib/bookshark/extractors/bibliographical_book_extractor.rb

@@ -5,31 +5,31 @@ require_relative 'base'
 
 
 module Biblionet
-  module Extractors   
-    
-    class BibliographicalBookExtractor < Base 
+  module Extractors
+
+    class BibliographicalBookExtractor < Base
       attr_reader :bibliographical_book
 
       def initialize(uri=nil)
-        super(uri)        
-        extract_bibliographical_book unless uri.nil? or @page.nil?        
+        super(uri)
+        extract_bibliographical_book unless uri.nil? or @page.nil?
       end
 
       def load_and_extract_book(uri=nil)
         load_page(uri)
         extract_bibliographical_book unless uri.nil? or @page.nil?
-      end  
+      end
 
-      def extract_bibliographical_book(biblionet_id=@biblionet_id, book_page=@page)                
+      def extract_bibliographical_book(biblionet_id=@biblionet_id, book_page=@page)
         # log = Logger.new(File.new(File.dirname(__dir__).to_s + "/logs/book_parsing.log",'a+'))
         log = Logger.new(STDOUT)
-                       
+
         page = BibliographicalBookDataExtractor.new(book_page)
 
         # End extraction if BookDataExtractor couldnt create a nodeset
         return nil if page.nodeset.nil?
 
-        bibliographical_book_hash = Hash.new   
+        bibliographical_book_hash = Hash.new
 
         extracted_details = page.details
 
@@ -47,7 +47,7 @@ module Biblionet
 
         bibliographical_book_hash[:publisher]         = extracted_details[:publisher]
         bibliographical_book_hash[:publication]       = extracted_details[:publication]
-        
+
         return @bibliographical_book = bibliographical_book_hash
       end
 
@@ -63,14 +63,14 @@ module Biblionet
           puts document
         end
         content = content_re.match(document)[0] unless (content_re.match(document)).nil?
-        
+
         # If content is nil, there is something wrong with the html, so return nil
         if content.nil?
           @nodeset = nil
         else
-          @nodeset = Nokogiri::HTML(content) 
-        end        
-      end    
+          @nodeset = Nokogiri::HTML(content)
+        end
+      end
 
       def size
         size_regex = /\d+x\d+/
@@ -101,19 +101,19 @@ module Biblionet
             original_title = detail.gsub(/Τίτλος πρωτοτύπου:/, "").strip
             details_hash[:original_title] = original_title
           end
-          
-          details_hash[:isbn]         = detail[isbn_regex] if detail =~ isbn_regex                 
+
+          details_hash[:isbn]         = detail[isbn_regex] if detail =~ isbn_regex
 
           details_hash[:isbn_13]      = detail[isbn_13_regex] if detail =~ isbn_13_regex
 
-          details_hash[:last_update]  = detail[last_update_regex] if detail =~ last_update_regex 
+          details_hash[:last_update]  = detail[last_update_regex] if detail =~ last_update_regex
 
-          details_hash[:cover_type]   = detail[cover_type_regex] if detail =~ cover_type_regex  
+          details_hash[:cover_type]   = detail[cover_type_regex] if detail =~ cover_type_regex
 
           details_hash[:availability] = detail[availability_regex] if detail =~ availability_regex
 
-          details_hash[:price]        = detail[price_regex] if detail =~ price_regex 
-          
+          details_hash[:price]        = detail[price_regex] if detail =~ price_regex
+
         end
 
         pre_details_text = @nodeset.xpath("//span[@class='small'][1]/preceding::text()").text
@@ -123,13 +123,13 @@ module Biblionet
         series_regex_no_vol = /(?<=\()\p{Word}+( \S)?( \p{Word}+( \S)?)*(?=\))/
         series_name_regex   = /\p{Word}+( \S)?( \p{Word}+( \S)?)*(?= ·)/
         series_volume_regex = /(?<=· )\d+/
-        physical_size_regex = /\d+x\d+/        
+        physical_size_regex = /\d+x\d+/
 
         series_hash = {}
         if pre_details_text =~ series_regex
           series = pre_details_text[series_regex]
           series_hash[:name]    = series[series_name_regex] if series =~ series_name_regex
-          series_hash[:volume]  = series[series_volume_regex] if series =~ series_volume_regex          
+          series_hash[:volume]  = series[series_volume_regex] if series =~ series_volume_regex
         elsif pre_details_text =~ series_regex_no_vol
           series = pre_details_text[series_regex_no_vol]
           series_hash[:name]    = series
@@ -137,7 +137,7 @@ module Biblionet
         end
 
         details_hash[:series] = series_hash
-        
+
         details_hash[:physical_size] = (pre_details_text =~ physical_size_regex) ? pre_details_text[physical_size_regex] : nil
 
         format_regex = /(?<=\[).+(?=\])/
@@ -148,20 +148,42 @@ module Biblionet
         details_hash[:format] = format.nil? ? 'Βιβλίο' : format
 
         publisher_node = @nodeset.xpath("//a[@class='booklink' and @href[contains(.,'/com/') ]][1]").first
-        publisher_hash = {}
-        publisher_hash[:text] = publisher_node.text
-        publisher_hash[:b_id] = (publisher_node[:href].split("/"))[2]
+        if !publisher_node.nil?
+          publisher_hash = {}
+          publisher_hash[:text] = publisher_node.text
+          publisher_hash[:b_id] = (publisher_node[:href].split("/"))[2]
 
-        pre_publisher_text = BibliographicalBookExtractor.decode_text(publisher_node.previous_sibling.text)
-        after_publisher_text = BibliographicalBookExtractor.decode_text(publisher_node.next_sibling.text)
+          pre_publisher_text = BibliographicalBookExtractor.decode_text(publisher_node.previous_sibling.text)
+          after_publisher_text = BibliographicalBookExtractor.decode_text(publisher_node.next_sibling.text)
 
-        publication_hash = {}
-        publication_hash[:year] = after_publisher_text[/(?<=, )\d+(?=\.)/]
-        publication_hash[:version] = pre_details_text[/(?<=- )\d+(?=η)/]
-        publication_hash[:place] = pre_details_text[/(?<=- )\p{Word}+( \S)?( \p{Word}+( \S)?)*(?= :)/]
+          publication_hash = {}
+          publication_hash[:year] = after_publisher_text[/(?<=, )\d+(?=\.)/]
+          publication_hash[:version] = pre_details_text[/(?<=- )\d+(?=η)/]
+          publication_hash[:place] = pre_details_text[/(?<=- )\p{Word}+( \S)?( \p{Word}+( \S)?)*(?= :)/]
 
-        details_hash[:publisher] = publisher_hash
-        details_hash[:publication] = publication_hash
+          details_hash[:publisher] = publisher_hash
+          details_hash[:publication] = publication_hash
+        else
+          publisher_node = @nodeset.xpath("//a[@class='subjectlink' and @href[contains(.,'/com/') ]][1]").first
+          if !publisher_node.nil?
+            details_hash[:publisher] = {
+              text: publisher_node.text,
+              b_id: (publisher_node[:href].split("/"))[2]
+            }
+            after_last_author_text = @nodeset
+              .xpath("//a[@class='booklink' and @href[contains(.,'/author/') ]][last()]").last
+              .next_sibling.text.strip
+            puts after_last_author_text
+            details_hash[:publication] = {
+              year: after_last_author_text[/(?<=: )\d+(?=\.)/],
+              version: after_last_author_text[/(?<=- )\d+(?=η)/],
+              place: after_last_author_text[/(?<=- )\p{Word}+( \S)?( \p{Word}+( \S)?)*(?= :)/]
+            }
+          else
+            details_hash[:publisher] = {text: nil, b_id: nil}
+            details_hash[:publication] = {year: nil, version: nil, place: nil}
+          end
+        end
 
         details_hash
       end
@@ -169,4 +191,4 @@ module Biblionet
     end
 
   end
-end
+end

data/lib/bookshark/extractors/book_extractor.rb

@@ -206,7 +206,15 @@ module Biblionet
         book_hash[:category]   = ddcs
         book_hash[:b_id] = biblionet_id
 
-        uri = "http://www.biblionet.gr/main.asp?page=results&Titlesid=#{biblionet_id}"
+        uri = nil
+
+        if @url
+          uri =  "http://www.biblionet.gr/main.asp?page=results&Titlesid=#{biblionet_id}"
+        elsif @filepath
+          uri = File.dirname(@filepath) + "/" + "bg_record_#{biblionet_id}.html"
+        end
+
+        # uri = "http://www.biblionet.gr/main.asp?page=results&Titlesid=#{biblionet_id}"
 
         bibliographical_book_extractor = Biblionet::Extractors::BibliographicalBookExtractor.new
         bibliographical_details = bibliographical_book_extractor.load_and_extract_book(uri)      

data/lib/bookshark/storage/json_book_records/0/book_63.json

@@ -0,0 +1,61 @@
+{
+  "book": [
+    {
+      "title": "Μέσα από το σκοτεινό καθρέφτη",
+      "subtitle": null,
+      "image": "http://www.biblionet.gr/images/covers/b63.jpg",
+      "author": [
+        {
+          "name": "Γιοστέιν Γκάαρντερ",
+          "b_id": "507"
+        }
+      ],
+      "contributors": {
+        "μετάφραση": [
+          {
+            "name": "Μαρία Αγγελίδου",
+            "b_id": "796"
+          }
+        ]
+      },
+      "publisher": {
+        "text": "Εκδοτικός Οίκος Α. Α. Λιβάνη",
+        "b_id": "271"
+      },
+      "isbn": "960-236-891-8",
+      "isbn_13": "978-960-236-891-6",
+      "award": [
+
+      ],
+      "description": "\"Bλέπουμε τον κόσμο μέσα από έναν καθρέφτη. Kι είναι ο κόσμος ένα μεγάλο μυστήριο για μας: \"Δι' εσόπτρου εν αινίγματι\". Kαμιά φορά καταφέρνουμε να ρίξουμε μια ματιά απ' την άλλη μεριά,να δούμε τι κρύβεται πίσω απ' τον καθρέφτη. Aν καθαρίζαμε όλο τον καθρέφτη και τον κάναμε γυαλί διάφανο, θα βλέπαμε φυσικά πολύ περισσότερα πράγματα. Mόνο που δε θα μπορούσαμε πια να δούμε τον εαυτό μας. Γιατί το γυαλί δε θα 'ταν πια καθρέφτης...\" <br>",
+      "category": [
+        {
+          "ddc": "839.82",
+          "name": "Νορβηγική πεζογραφία - Μυθιστόρημα",
+          "b_id": "400"
+        }
+      ],
+      "b_id": "63",
+      "publication": {
+        "year": "1998",
+        "version": null,
+        "place": "Αθήνα"
+      },
+      "format": "Βιβλίο",
+      "original_language": "νορβηγικά",
+      "original_title": "I et speil, i en gåte",
+      "price": "14,85",
+      "availability": "Κυκλοφορεί",
+      "last_update": null,
+      "series": {
+        "name": "Ξένη Λογοτεχνία",
+        "volume": null
+      },
+      "physical_description": {
+        "pages": "267",
+        "size": "21x14",
+        "cover_type": "Μαλακό εξώφυλλο"
+      }
+    }
+  ]
+}

data/lib/bookshark/storage/json_book_records/0/book_67.json

@@ -0,0 +1,53 @@
+{
+  "book": [
+    {
+      "title": "Ο χορός των ωρών",
+      "subtitle": "Διηγήματα",
+      "image": "http://www.biblionet.gr/images/covers/b67.jpg",
+      "author": [
+        {
+          "name": "Τσιτσέλη Καίη 1926-2001",
+          "b_id": "74"
+        }
+      ],
+      "contributors": {
+      },
+      "publisher": {
+        "text": "Άγρα",
+        "b_id": "44"
+      },
+      "isbn": "960-325-281-6",
+      "isbn_13": "978-960-325-281-8",
+      "award": [
+
+      ],
+      "description": null,
+      "category": [
+        {
+          "ddc": "889.3",
+          "name": "Νεοελληνική πεζογραφία - Διήγημα",
+          "b_id": "3"
+        }
+      ],
+      "b_id": "67",
+      "publication": {
+        "year": "1998",
+        "version": null,
+        "place": "Αθήνα"
+      },
+      "format": "Βιβλίο",
+      "original_language": null,
+      "original_title": null,
+      "price": "11,15",
+      "availability": "Κυκλοφορεί",
+      "last_update": "5/7/2007",
+      "series": {
+      },
+      "physical_description": {
+        "pages": "153",
+        "size": "17x11",
+        "cover_type": "Μαλακό εξώφυλλο"
+      }
+    }
+  ]
+}

data/lib/bookshark/storage/json_book_records/2/book_2110.json

@@ -0,0 +1,59 @@
+{
+  "book": [
+    {
+      "title": "Kreta - Mir ist, Als Wurde Ich Heimkehren",
+      "subtitle": null,
+      "image": null,
+      "author": [
+        {
+          "name": "Karin Guse",
+          "b_id": "879"
+        }
+      ],
+      "contributors": {
+        "φωτογράφιση": [
+          {
+            "name": "Karin Guse",
+            "b_id": "879"
+          }
+        ]
+      },
+      "publisher": {
+        "text": "Ευσταθιάδης Group",
+        "b_id": "101"
+      },
+      "isbn": "960-226-521-3",
+      "isbn_13": "978-960-226-521-5",
+      "award": [
+
+      ],
+      "description": null,
+      "category": [
+        {
+          "ddc": "914.9504",
+          "name": "Κρήτη - Περιγραφή και ταξίδια - Οδηγοί",
+          "b_id": "110"
+        }
+      ],
+      "b_id": "2110",
+      "publication": {
+        "year": "1998",
+        "version": null,
+        "place": "Αθήνα"
+      },
+      "format": "Βιβλίο",
+      "original_language": null,
+      "original_title": null,
+      "price": "4,46",
+      "availability": "Κυκλοφορεί",
+      "last_update": null,
+      "series": {
+      },
+      "physical_description": {
+        "pages": "79",
+        "size": "18x12",
+        "cover_type": "Μαλακό εξώφυλλο"
+      }
+    }
+  ]
+}

data/lib/bookshark/storage/json_book_records/2/book_2111.json

@@ -0,0 +1,65 @@
+{
+  "book": [
+    {
+      "title": "Athens - Attica",
+      "subtitle": "General Guide",
+      "image": null,
+      "author": [
+        {
+          "name": "John Bowman",
+          "b_id": "578"
+        }
+      ],
+      "contributors": {
+        "φωτογράφιση": [
+          {
+            "name": "A. Spyropoulos",
+            "b_id": "1195"
+          },
+          {
+            "name": "G. Sfikas",
+            "b_id": "15982"
+          }
+        ]
+      },
+      "publisher": {
+        "text": "Ευσταθιάδης Group",
+        "b_id": "101"
+      },
+      "isbn": "960-226-378-4",
+      "isbn_13": "978-960-226-378-5",
+      "award": [
+
+      ],
+      "description": null,
+      "category": [
+        {
+          "ddc": "910.49512",
+          "name": "Αθήνα - Οδηγοί",
+          "b_id": "198"
+        }
+      ],
+      "b_id": "2111",
+      "publication": {
+        "year": "1998",
+        "version": null,
+        "place": "Attica"
+      },
+      "format": "Βιβλίο",
+      "original_language": null,
+      "original_title": null,
+      "price": "5,07",
+      "availability": "Κυκλοφορεί",
+      "last_update": null,
+      "series": {
+        "name": "Argonaut Pocket Guide",
+        "volume": null
+      },
+      "physical_description": {
+        "pages": "214",
+        "size": "18x12",
+        "cover_type": "Μαλακό εξώφυλλο"
+      }
+    }
+  ]
+}

data/lib/bookshark/storage/json_book_records/2/book_2112.json

@@ -0,0 +1,69 @@
+{
+  "book": [
+    {
+      "title": "Raubvögel in Griechenland",
+      "subtitle": null,
+      "image": null,
+      "author": [
+        {
+          "name": "G. Handrinos",
+          "b_id": "1185"
+        },
+        {
+          "name": "A. Dimitropoulos",
+          "b_id": "1186"
+        }
+      ],
+      "contributors": {
+        "εικονογράφηση": [
+          {
+            "name": "Ach. Dimitropoulos",
+            "b_id": "1186"
+          },
+          {
+            "name": "G. Handrinos",
+            "b_id": "1185"
+          }
+        ]
+      },
+      "publisher": {
+        "text": "Ευσταθιάδης Group",
+        "b_id": "101"
+      },
+      "isbn": "960-226-576-0",
+      "isbn_13": "978-960-226-576-5",
+      "award": [
+
+      ],
+      "description": null,
+      "category": [
+        {
+          "ddc": "598.495",
+          "name": "Πουλιά - Ελλάς",
+          "b_id": "205"
+        }
+      ],
+      "b_id": "2112",
+      "publication": {
+        "year": "1999",
+        "version": null,
+        "place": "Αθήνα"
+      },
+      "format": "Βιβλίο",
+      "original_language": null,
+      "original_title": null,
+      "price": "9,83",
+      "availability": "Κυκλοφορεί",
+      "last_update": null,
+      "series": {
+        "name": "Griechenlands Natur",
+        "volume": null
+      },
+      "physical_description": {
+        "pages": "199",
+        "size": "18x12",
+        "cover_type": "Μαλακό εξώφυλλο"
+      }
+    }
+  ]
+}