X-Git-Url: http://jaekl.net/gitweb/?p=quanlib.git;a=blobdiff_plain;f=store.rb;h=92d1ce57d1917ea7309d86527fa96f5aa826581e;hp=0ba3447e8807051c923b258cfcbf9f4270d398b3;hb=HEAD;hpb=fffa1ed35ce07fdec65e1aa14a8f637fbdde9b10

diff --git a/store.rb b/store.rb
index 0ba3447..1a33ca3 100644
--- a/store.rb
+++ b/store.rb
@@ -1,26 +1,39 @@
 
 require 'csv'
 require 'fileutils'
+require 'inifile'
 require 'pg'
 
-require 'series'
+require_relative 'series'
+require_relative 'tconn'
 
 class Store
-  def initialize
-    @basepath = '/arc/quanlib'	# TODO: FIXME: configure this in a sane way
+  def unclassified_csv
+    @basePath + '/csv/unclassified.csv'
+  end
+
+  def initialize(config_file)
     @conn = nil
 
-    #@dburl = 'dbi:Pg:quanlib:localhost'
-    @dbhost = "localhost"
+    config = IniFile.load(config_file)
+    if nil == config
+      puts 'FATAL:  Failed to load config file "' + config_file + '".  Aborting initialization.'
+      return
+    end
+
+    section = config['database']
+    @dbhost = section['host']
     @dbport = 5432
-    @dbname = 'quanlib'
-    @dbuser = 'quanlib'
-    @dbpass = 'quanlib'
+    @dbname = section['name']
+    @dbuser = section['user']
+    @dbpass = section['pass']
+
+    section = config['filesystem']
+    @basePath = section['basePath']
   end
 
   def connect
-    # @conn = PGconn.connect('localhost', 5432, '', '', 'quanlib', 'quanlib', 'quanlib')
-    @conn = PG.connect('localhost', 5432, '', '', 'quanlib', 'quanlib', 'quanlib')
+    @conn = TimedConn.new(PG.connect(@dbhost, @dbport, '', '', @dbname, @dbuser, @dbpass))
     return @conn
   end
 
@@ -35,40 +48,105 @@ class Store
     return path, name
   end
 
-  def create_schema
-    create_authors = 
+  def cross_reference_lists
+puts "@@@@@@@@@@@ CROSS-REF START @@@@@@@@@@@"
+    exec_update("TRUNCATE TABLE Lists CASCADE;", [])
+
+    populate_lists_table
+puts "@@@@@@@@@@@ CROSS-REF DONE @@@@@@@@@@@"
+  end
+
+  def create_schema(skip_class)
+    create_authors =
 <<EOS
       CREATE TABLE Authors (
         id          INTEGER PRIMARY KEY,
         grouping    VARCHAR(64),
-        reading     VARCHAR(128),
-        sort        VARCHAR(128)
+        reading     VARCHAR(256),
+        sort        VARCHAR(256)
       );
 EOS
 
-    create_books = 
+    create_books =
 <<EOS
       CREATE TABLE Books (
-        id          INTEGER PRIMARY KEY,
-        author      INTEGER REFERENCES Authors(id),
-        cover       INTEGER,
-        description TEXT,
-        path        VARCHAR(256),
-        series      INTEGER REFERENCES Series(id),
-        title       VARCHAR(196),
-        volume      VARCHAR(16)
+        id             INTEGER PRIMARY KEY,
+        arrived        TIMESTAMP,
+        author         INTEGER REFERENCES Authors(id),
+        classification INTEGER REFERENCES Classifications(id),
+        cover          INTEGER,
+        language       VARCHAR(64),
+        description    TEXT,
+        path           VARCHAR(256),
+        series         INTEGER REFERENCES Series(id),
+        title          VARCHAR(256),
+        volume         VARCHAR(16)
       );
 EOS
 
-    create_efs = 
+    create_classification =
+<<EOS
+      CREATE TABLE Classifications (
+        id              INTEGER PRIMARY KEY,
+        ddc             VARCHAR(32),
+        lcc             VARCHAR(32),
+        author_grouping VARCHAR(64),
+        author_sort     VARCHAR(128),
+        title_grouping  VARCHAR(256),
+        title           VARCHAR(256)
+      );
+EOS
+
+    create_efs =
 <<EOS
       CREATE TABLE EFS (
-        id          INTEGER,
+        id          INTEGER PRIMARY KEY,
         mimetype    VARCHAR(64)
       );
 EOS
 
-    create_series = 
+    create_fast =
+<<EOS
+      CREATE TABLE FAST (
+        id          VARCHAR(32) PRIMARY KEY,
+        descr       VARCHAR(128)
+      );
+EOS
+
+    # Associative entity, linking FAST and Classifications tables
+    # in a 0..n to 0..m relationship
+    create_fast_classifications =
+<<EOS
+      CREATE TABLE FAST_Classifications (
+        fast           VARCHAR(32) REFERENCES FAST(id),
+        classification INTEGER REFERENCES Classifications(id)
+      );
+EOS
+
+    create_lists =
+<<EOS
+      CREATE TABLE Lists (
+        id             INTEGER PRIMARY KEY,
+        age            VARCHAR(32),
+        category       VARCHAR(32),
+        code           VARCHAR(2),
+        year           INTEGER,
+        author         INTEGER REFERENCES Authors(id),
+        title          VARCHAR(256)
+      );
+EOS
+
+    # Associative entity, linking Lists and Books tables
+    # in a 0..n to 0..m relationship
+    create_lists_books =
+<<EOS
+      CREATE TABLE Lists_Books (
+        list           INTEGER REFERENCES Lists(id),
+        book           INTEGER REFERENCES Books(id)
+      );
+EOS
+
+    create_series =
 <<EOS
       CREATE TABLE Series (
         id          INTEGER PRIMARY KEY,
@@ -82,12 +160,19 @@ EOS
 
     stmts = [
       create_authors,
+      create_classification,
       create_efs,
+      create_fast,
       create_series,
       create_books,
+      create_fast_classifications,
+      create_lists,
+      create_lists_books,
       'CREATE SEQUENCE author_id;',
       'CREATE SEQUENCE book_id;',
+      'CREATE SEQUENCE classification_id;',
       'CREATE SEQUENCE efs_id;',
+      'CREATE SEQUENCE list_id;',
       'CREATE SEQUENCE series_id;'
     ]
 
@@ -95,38 +180,71 @@ EOS
       @conn.exec(stmt)
     end
 
-    populate_series_table()
+    if skip_class == false
+      populate_fast_table
+      populate_classifications_table
+    end
+
+    populate_series_table
   end
 
   def dropSchema
     stmts = [
+      'DROP TABLE Lists_Books;',
+      'DROP TABLE Lists;',
       'DROP TABLE Books;',
+      'DROP TABLE FAST_Classifications;',
       'DROP TABLE Authors;',
+      'DROP TABLE Classifications;',
       'DROP TABLE EFS;',
+      'DROP TABLE FAST;',
       'DROP TABLE Series;',
       'DROP SEQUENCE author_id;',
       'DROP SEQUENCE book_id;',
+      'DROP SEQUENCE classification_id;',
       'DROP SEQUENCE efs_id;',
+      'DROP SEQUENCE list_id;',
       'DROP SEQUENCE series_id;'
     ]
 
     for stmt in stmts do
-      @conn.exec(stmt)
+      begin
+        @conn.exec(stmt)
+      rescue Exception => exc
+        puts 'WARNING:  "' + stmt + '" failed:  ' + exc.to_s
+      end
     end
   end
 
+  def find_all_authors(author_name)
+    result = []
+
+    sqlSelect = "SELECT id FROM Authors WHERE grouping=$1;"
+    args = [author_name]
+
+    @conn.exec_params(sqlSelect, args) do |rs|
+      rs.each do |row|
+        result << row['id']
+      end
+    end
+
+    result
+  end
+
   def find_author(author)
     sqlSelect = "SELECT id FROM Authors WHERE grouping=$1 AND reading=$2 AND sort=$3;"
     args = [author.grouping, author.reading_order, author.sort_order]
+
     @conn.exec_params(sqlSelect, args) do |rs|
       if rs.ntuples > 0
         return rs[0]['id']
       end
     end
+
     return nil
   end
 
-  def init_db
+  def init_db(skip_class)
     sql = "SELECT 1 FROM pg_tables WHERE tableowner='quanlib' AND tablename='books'"
     found = false
     @conn.exec(sql).each do |row|
@@ -134,12 +252,11 @@ EOS
     end
 
     if ! found
-      create_schema()
+      create_schema(skip_class)
     end
   end
 
   def load_author(id)
-    #puts 'DEBUG:  load_author(' + id + ')'
     sqlSelect = "SELECT grouping, reading, sort FROM Authors WHERE id=$1"
     args = [id]
     @conn.exec_params(sqlSelect, args) do |rs|
@@ -148,10 +265,8 @@ EOS
       end
       row = rs[0]
       author = Author.new(row['grouping'], row['reading'], row['sort'])
-      #puts 'DEBUG:  author:  ' + author.inspect()
       return author
     end
-    #puts 'DEBUG:  NOT FOUND'
     return nil
   end
 
@@ -161,7 +276,7 @@ EOS
       id = next_id('author_id')
       sqlInsert = "INSERT INTO Authors(id, grouping, reading, sort) VALUES ($1, $2, $3, $4);"
       args = [id, author.grouping, author.reading_order, author.sort_order]
-      begin 
+      begin
         rs = @conn.exec_params(sqlInsert, args)
       rescue Exception => e
         puts sqlInsert + ":  " + args.inspect()
@@ -175,8 +290,7 @@ EOS
   end
 
   def load_book(id)
-    #puts 'DEBUG:  load_book(' + id + ')'
-    sql = "SELECT author, cover, description, path, series, title, volume FROM Books WHERE id=$1;"
+    sql = "SELECT author, classification, cover, description, language, path, series, title, volume FROM Books WHERE id=$1;"
     book = nil
 
     begin
@@ -189,38 +303,39 @@ EOS
 
         book = Book.new(self)
         book.author = load_author(row['author'])
+        book.classification_id = row['classification']
         book.cover = load_cover(row['cover'])
         book.description = row['description']
+        book.language = row['language']
         book.path = row['path']
         book.series_id = row['series']
         book.title = row['title']
         book.volume = row['volume']
-      end    
+      end
     rescue Exception => e
       puts sql + ": " + id
       puts e.message
       puts $@
     end
 
-    #puts 'DEBUG:  loaded book:   ' + book.inspect()
     return book
   end
 
   def store_book(book)
-    sql = "INSERT INTO Books (id, author, cover, description, path, series, title, volume) VALUES ($1, $2, $3, $4, $5, $6, $7, $8);"
+    sql = "INSERT INTO Books (id, arrived, author, classification, cover, description, language, path, series, title, volume) VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10, $11);"
 
     book_id = next_id('book_id')
 
     author_id = store_author(book.author)
     (efs_id, mime_type) = store_cover(book)
 
-    args = [book_id, author_id, efs_id, book.description(), book.path(), book.series_id(), book.title(), book.volume()]
+    args = [book_id, book.arrived, author_id, book.classification_id, efs_id, book.description, book.language, book.path, book.series_id, book.title, book.volume]
 
     begin
       rs = @conn.exec_params(sql, args)
     rescue Exception => e
       puts sql + ": " + args.inspect()
-      puts e.message 
+      puts e.message
       puts $@
     ensure
       rs.clear if rs
@@ -229,6 +344,38 @@ EOS
     return book_id
   end
 
+  def find_classification(author_grouping, title_grouping)
+    sql = "SELECT id FROM Classifications WHERE author_grouping = $1 AND title_grouping = $2;"
+    @conn.exec_params(sql, [author_grouping, title_grouping]) do |rs|
+      if rs.ntuples > 0
+        return rs[0]['id']
+      end
+    end
+    return nil
+  end
+
+  def load_classification(id)
+    sql  = "SELECT ddc, lcc, author_grouping, author_sort, title_grouping, title "
+    sql += " FROM Classifications WHERE id=$1"
+    @conn.exec_params(sql, [id]) do |rs|
+      if rs.ntuples > 0
+        row = rs[0]
+        ddc = row['ddc']
+        lcc = row['lcc']
+        author_grouping = row['author_grouping']
+        author = row['author_sort']
+        title_grouping = row['title_grouping']
+        title = row['title']
+
+        result = Classification.new(ddc, lcc, author_grouping, author, title_grouping, title)
+        result.id = id
+        return result
+      end
+    end
+
+    return nil
+  end
+
   def load_cover(id)
     if nil == id
       return nil
@@ -246,15 +393,9 @@ EOS
 
     (efspath, efsname) = construct_efs_path(id)
 
-    fullpath = @basepath + '/efs/' + efspath + '/' + efsname
+    fullpath = @basePath + '/efs/' + efspath + '/' + efsname
 
     return Cover.new(nil, fullpath, mime_type)
-
-    #File.open(fullpath, 'rb') do |is|
-    #  return Cover.new(is, fullpath, mime_type)
-    #end
-    #
-    #return nil
   end
 
   def store_cover(book)
@@ -275,7 +416,7 @@ EOS
 
     (efspath, efsname) = construct_efs_path(efs_id)
 
-    efspath = @basepath + '/efs/' + efspath
+    efspath = @basePath + '/efs/' + efspath
 
     FileUtils.mkdir_p(efspath)
 
@@ -291,15 +432,37 @@ EOS
     ensure
       rs.clear if rs
     end
-    
+
     return efs_id, mimetype
   end
 
+  def exec_id_query(sql, args)
+    ids = []
+    @conn.exec_params(sql, args) do |rs|
+      rs.each do |row|
+        ids.push(row['id'])
+      end
+    end
+    return ids
+  end
+
+  def exec_update(sql, args)
+    begin
+      rs = @conn.exec_params(sql, args)
+    rescue Exception => e
+      puts sql + ": " + args.inspect()
+      puts e.message
+      puts $@
+    ensure
+      rs.clear if rs
+    end
+  end
+
   def next_id(seq_name)
     id = nil
     @conn.exec("SELECT nextval('" + seq_name + "');") do |rs|
       id = rs[0]['nextval']
-    end 
+    end
     return id
   end
 
@@ -337,73 +500,156 @@ EOS
     return nil
   end
 
+  def populate_classifications_table
+    puts "Populating the Classifications table..."
+    first = true
+    CSV.foreach(@basePath + '/csv/class.csv') do |row|
+      if first
+        # skip the header row
+        first = false
+      else
+
+        # First, add a row to the Classifications table
+
+        id = next_id('classification_id')
+        ddc = row[0]
+        lcc = row[1]
+        author_grouping = row[2]
+        author_sort = row[3]
+        title_grouping = row[4]
+        title = row[5]
+
+        sqlInsert = "INSERT INTO Classifications (id, ddc, lcc, author_grouping, author_sort, title_grouping, title) VALUES ($1, $2, $3, $4, $5, $6, $7);"
+        args = [id, ddc, lcc, author_grouping, author_sort, title_grouping, title]
+        exec_update(sqlInsert, args)
+
+        # Second, link up with the appropriate FAST table entries
+
+        fast = []
+        input = row[6]
+        if input.length > 0
+          fast = input.split(';')
+        end
+
+        fast.each do |fast_id|
+          sqlInsert = "INSERT INTO FAST_Classifications (fast, classification) VALUES ($1, $2);"
+          args = [fast_id, id]
+          exec_update(sqlInsert, args)
+        end
+      end
+    end
+  end
+
+  def populate_fast_table
+    puts "Populating the FAST table..."
+    first = true
+    CSV.foreach(@basePath + '/csv/fast.csv') do |row|
+      if first
+        first = false	# skip the header row
+      else
+        id = row[0]
+        descr = row[1]
+        sqlInsert = "INSERT INTO FAST (id, descr) VALUES ($1, $2);"
+        exec_update(sqlInsert, [id, descr])
+      end
+    end
+  end
+
+  def populate_lists_table
+    puts "Populating the Lists table..."
+
+    CSV.foreach(@basePath + "/csv/lists.csv", headers: true) do |row|
+      author_ids = find_all_authors(row['author'])
+      if author_ids.empty?
+        specification = [row['age'], row['category'], row['code'], row['year'], row['author'], row['title']]
+          .map { |x| x.inspect }
+          .join(', ')
+
+        puts "WARNING: For list entry (#{specification}), no such author was found."
+
+        next
+      end
+
+      sqlInsert = %Q(
+        INSERT INTO Lists (id, age, category, code, year, author, title)
+        VALUES ($1, $2, $3, $4, $5, $6, $7);
+      )
+      author_ids.each do |author_id|
+        list_id = next_id('list_id')
+        args = [list_id, row['age'], row['category'], row['code'], row['year'], author_id, row['title']]
+        exec_update(sqlInsert, args)
+
+        update_lists_books_table(list_id, author_id, row['title'])
+      end
+    end
+  end
+
+  # Scan for books that match this Lists entry, and add any matches to the Lists_Books associative table
+  def update_lists_books_table(list_id, author_id, title)
+    title_pattern = Book.grouping_for_title(title).gsub('_', '%')
+    sqlSelect = "SELECT id FROM Books WHERE author = $1 AND title LIKE $2;"
+    args = [author_id, title_pattern]
+
+    @conn.exec_params(sqlSelect, args) do |rs|
+      rs.each do |row|
+        sqlInsert = "INSERT INTO Lists_Books (list, book) VALUES ($1, $2)"
+        args = [list_id, row['id']]
+        exec_update(sqlInsert, args)
+      end
+    end
+  end
+
   def populate_series_table
     puts "Populating the Series table..."
-    CSV.foreach(@basepath + '/csv/series.csv') do |row|
+    CSV.foreach(@basePath + '/csv/series.csv') do |row|
       id = next_id('series_id')
       sqlInsert = "INSERT INTO Series (id, age, genre, grouping, code, descr) VALUES ($1, $2, $3, $4, $5, $6);"
       args = [id] + row
-      begin
-        # DEBUG: puts 'SQL> ' + sqlInsert + ':  ' + args.inspect()
-        rs = @conn.exec_params(sqlInsert, args)
-      rescue Exception => e
-        puts sqlInsert + ":  " + args.inspect()
-        puts e.message
-        puts $@
-      ensure
-        rs.clear if rs
-      end
+      exec_update(sqlInsert, args)
     end
   end
 
   def query_books_by_author(pattern)
-    sql = 
+    sql =
 <<EOS
-      SELECT b.id FROM Authors a 
-      INNER JOIN Books b ON b.author=a.id 
+      SELECT b.id FROM Authors a
+      INNER JOIN Books b ON b.author=a.id
       LEFT OUTER JOIN Series s on s.id=b.series
-      WHERE upper(a.grouping) LIKE $1 
+      WHERE upper(a.grouping) LIKE $1
       ORDER BY a.grouping, b.series, b.volume, b.title
 EOS
-    book_ids = []
-    @conn.exec_params(sql, [pattern]) do |rs|
-      rs.each do |row|
-        book_ids.push(row['id'])
-      end
-    end
-    return book_ids
+    return exec_id_query(sql, [pattern])
+  end
+
+  def query_books_by_ddc
+    sql =
+<<EOS
+      SELECT b.id FROM Classifications c
+      INNER JOIN Books b ON b.classification=c.id
+      ORDER BY c.ddc
+EOS
+    return exec_id_query(sql, [])
   end
 
   def query_books_by_series_id(id)
-    sql = 
+    sql =
 <<EOS
       SELECT b.id FROM Books b
       WHERE b.series = $1
       ORDER BY b.volume,b.title
 EOS
-    book_ids = []
-    @conn.exec_params(sql, [id]) do |rs|
-      rs.each do |row|
-        book_ids.push(row['id'])
-      end
-    end
-    return book_ids
+    return exec_id_query(sql, [id])
   end
 
   def query_series_by_age(pattern)
-    sql = 
+    sql =
 <<EOS
-      SELECT s.id FROM Series s
+      SELECT s.id
+      FROM Series s
       WHERE s.age LIKE $1
-      ORDER BY s.descr
+      ORDER BY s.grouping,s.descr
 EOS
-    series_ids = []
-    @conn.exec_params(sql, [pattern]) do |rs|
-      rs.each do |row|
-        series_ids.push(row['id'])
-      end
-    end
-    return series_ids
+    return exec_id_query(sql, [pattern])
   end
 end