From 63c5a3cead1fb5cbd9b1bff653f269dce8d8052c Mon Sep 17 00:00:00 2001 From: Ralph Amissah Date: Tue, 20 Apr 2010 19:01:55 -0400 Subject: db name, tables, columns, indexes changes, review (need another version bump 2.2.0) * db (sql) table structure, further review and changes (hence breakage & version bump) * new pgsql db name prefix "sisu_v2b_" * new table column words in doc_objects & endnotes, VARCHAR 3000 to contain list of unique sorted words in object * increase use of VARCHAR * constants takes on related additions * param, extensive db column size checks for metadata --- lib/sisu/v2/constants.rb | 7 +- lib/sisu/v2/db_columns.rb | 158 ++++++++++++++++++++++++++++--------------- lib/sisu/v2/db_create.rb | 27 +++++--- lib/sisu/v2/db_drop.rb | 72 ++++++++++++-------- lib/sisu/v2/db_import.rb | 120 ++++++++++++++++++-------------- lib/sisu/v2/db_indexes.rb | 52 +++++++------- lib/sisu/v2/db_load_tuple.rb | 32 +++++---- lib/sisu/v2/db_remove.rb | 4 +- lib/sisu/v2/db_sqltxt.rb | 34 ++++++---- lib/sisu/v2/param.rb | 57 +++++++++++----- 10 files changed, 349 insertions(+), 214 deletions(-) diff --git a/lib/sisu/v2/constants.rb b/lib/sisu/v2/constants.rb index 3fcb1e3a..9a24736c 100644 --- a/lib/sisu/v2/constants.rb +++ b/lib/sisu/v2/constants.rb @@ -131,8 +131,8 @@ Px[:lv4]= '-' Px[:lv5]= '.' Px[:lv6]= '.' #Px[:lv5_6]= '.' -Db[:name_prefix]="SiSU#{SiSU_version_dir}a_" -Db[:name_prefix_db]="sisu_#{SiSU_version_dir}a_" +Db[:name_prefix]="SiSU#{SiSU_version_dir}b_" +Db[:name_prefix_db]="sisu_#{SiSU_version_dir}b_" Db[:col_title]=800 Db[:col_title_part]=400 Db[:col_title_edition]=10 @@ -148,6 +148,9 @@ Db[:col_classify_identify]=256 Db[:col_classify_library]=30 Db[:col_classify_small]=16 Db[:col_filename]=256 +Db[:col_digest]=64 +Db[:col_filesize]=10 +Db[:col_info_note]=3000 __END__ consider: 〔comment〕 diff --git a/lib/sisu/v2/db_columns.rb b/lib/sisu/v2/db_columns.rb index ee66c59e..0c2eb367 100644 --- a/lib/sisu/v2/db_columns.rb +++ b/lib/sisu/v2/db_columns.rb @@ -208,7 +208,7 @@ module SiSU_DB_columns 'title_note' end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_info_note]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} @@ -1119,7 +1119,7 @@ module SiSU_DB_columns 'rights' end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_info_note]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} @@ -1141,7 +1141,7 @@ module SiSU_DB_columns 'rights_copyright_text' end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_info_note]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} @@ -1163,7 +1163,7 @@ module SiSU_DB_columns 'rights_copyright_translation' end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_info_note]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} @@ -1185,7 +1185,7 @@ module SiSU_DB_columns 'rights_copyright_illustrations' end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_info_note]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} @@ -1207,7 +1207,7 @@ module SiSU_DB_columns 'rights_copyright_photographs' end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_info_note]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} @@ -1229,7 +1229,7 @@ module SiSU_DB_columns 'rights_copyright_preparation' end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_info_note]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} @@ -1251,7 +1251,7 @@ module SiSU_DB_columns 'rights_copyright_digitization' end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_info_note]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} @@ -1273,7 +1273,7 @@ module SiSU_DB_columns 'rights_copyright_audio' end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_info_note]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} @@ -1295,7 +1295,7 @@ module SiSU_DB_columns 'rights_copyright_video' end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_info_note]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} @@ -1317,7 +1317,7 @@ module SiSU_DB_columns 'rights_license' end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_info_note]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} @@ -1355,8 +1355,7 @@ module SiSU_DB_columns 'classify_topic_register' end def create_column - "#{name} VARCHAR(#{Db[:col_classify_txt_long]}) NULL," - #"#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_info_note]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} @@ -1804,21 +1803,18 @@ module SiSU_DB_columns self end =begin -#% misc -@make: - :skin: -@links: +#% src =end - def filename + def src_filename def name - 'filename' + 'src_filename' end def create_column "#{name} VARCHAR(#{Db[:col_filename]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} - IS 'metadata document filename';} + IS 'sisu markup source text filename';} end def tuple t=if defined? @md.fns \ @@ -1831,56 +1827,61 @@ module SiSU_DB_columns end self end - def sisutxt # consider naming sisusrc + def src_fingerprint def name - 'sisutxt' + 'src_fingerprint' #hash/digest, sha256 or md5 end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_digest]}) NULL," + #"#{name} TEXT NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} - IS 'sisu markup text (if shared)';} + IS 'sisu markup source text fingerprint, hash digest sha256 (or md5)';} end def tuple - t=if @md.mod.inspect=~/import|update/ \ - and FileTest.exist?(@md.fns) - ["#{name}, ","'#{@sisutxt}', "] + t=if defined? @md.dgst \ + and @md.dgst.class==Array \ + and @md.dgst[1]=~/\S+/ + txt=@md.dgst[1] + ["#{name}, ","'#{txt}', "] else ['',''] end end self end - def fulltext + def src_filesize def name - 'fulltext' + 'src_filesize' end def create_column - "#{name} TEXT NULL," + "#{name} VARCHAR(#{Db[:col_filesize]}) NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} - IS 'document full text clean, searchable';} + IS 'sisu markup source text file size';} end def tuple - t=if @md.mod.inspect=~/import|update/ \ - and FileTest.exist?(@md.fns) - ["#{name}, ","'#{@fulltext}', "] - else ['',''] - end + t=if defined? @md.filesize \ + and @md.filesize=~/\S+/ + txt=@md.filesize + special_character_escape(txt) + ["#{name}, ","'#{txt}', "] + else ['',''] + end end self end - def word_count + def src_word_count def name - 'word_count' + 'src_word_count' end def create_column "#{name} TEXT NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} - IS 'document word count';} + IS 'sisu markup source text word count';} end def tuple t=if defined? @md.wc_words \ @@ -1893,23 +1894,47 @@ module SiSU_DB_columns end self end - def digest + def src_txt # consider naming sisusrc def name - 'dgst' + 'src_text' end def create_column "#{name} TEXT NULL," end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} - IS 'document hash digest sha256 (or md5)';} + IS 'sisu markup source text (if shared)';} end def tuple - t=if defined? @md.dgst \ - and @md.dgst=~/\S+/ - txt=@md.dgst - special_character_escape(txt) - ["#{name}, ","'#{txt}', "] + t=if @md.mod.inspect=~/import|update/ \ + and FileTest.exist?(@md.fns) + ["#{name}, ","'#{@sisutxt}', "] + else ['',''] + end + end + self + end +=begin +#% misc +@make: + :skin: +@links: +=end + def fulltext + def name + 'fulltext' + end + def create_column + "#{name} TEXT NULL," + end + def column_comment + %{COMMENT ON COLUMN metadata_and_text.#{name} + IS 'document full text clean, searchable';} + end + def tuple + t=if @md.mod.inspect=~/import|update/ \ + and FileTest.exist?(@md.fns) + ["#{name}, ","'#{@fulltext}', "] else ['',''] end end @@ -1924,12 +1949,35 @@ module SiSU_DB_columns end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} - IS 'metadata document skin name';} + IS 'source text skin name';} + end + def tuple + t=if defined? @md.skin_name \ + and @md.skin_name=~/\S+/ + txt=@md.skin_name + special_character_escape(txt) + ["#{name}, ","'#{txt}', "] + else ['',''] + end + end + self + end + def skin_fingerprint #check + def name + 'skin_fingerprint' + end + def create_column + "#{name} VARCHAR(#{Db[:col_digest]}) NULL," + end + def column_comment + %{COMMENT ON COLUMN metadata_and_text.#{name} + IS 'source text skin fingerprint';} end def tuple - t=if defined? @md.notes.skin_name \ - and @md.notes.skin_name=~/\S+/ - txt=@md.notes.skin_name + t=if defined? @md.dgst_skin \ + and @md.dgst_skin.class==Array \ + and @md.dgst_skin[1]=~/\S+/ + txt=@md.dgst_skin[1] special_character_escape(txt) ["#{name}, ","'#{txt}', "] else ['',''] @@ -1946,7 +1994,7 @@ module SiSU_DB_columns end def column_comment %{COMMENT ON COLUMN metadata_and_text.#{name} - IS 'metadata document skin';} + IS 'source text skin';} end def tuple t=if defined? @md.skin \ @@ -1972,9 +2020,9 @@ module SiSU_DB_columns IS 'metadata document links';} end def tuple - t=if defined? @md.notes.links \ - and @md.notes.links=~/\S+/ - txt=@md.notes.links + t=if defined? @md.links \ + and @md.links=~/\S+/ + txt=@md.links special_character_escape(txt) ["#{name}, ","'#{txt}', "] else ['',''] diff --git a/lib/sisu/v2/db_create.rb b/lib/sisu/v2/db_create.rb index 1fc00168..c1fed045 100644 --- a/lib/sisu/v2/db_create.rb +++ b/lib/sisu/v2/db_create.rb @@ -182,13 +182,16 @@ module SiSU_DB_create #{column.notes_prefix_a.create_column} #{column.notes_prefix_b.create_column} #{column.notes_suffix.create_column} + /* src */ + #{column.src_filename.create_column} + #{column.src_fingerprint.create_column} + #{column.src_filesize.create_column} + #{column.src_word_count.create_column} + #{column.src_txt.create_column} /* misc */ - #{column.filename.create_column} - #{column.sisutxt.create_column} #{column.fulltext.create_column} - #{column.word_count.create_column} - #{column.digest.create_column} #{column.skin_name.create_column} + #{column.skin_fingerprint.create_column} #{column.skin.create_column} #{column.links.create_column.gsub(/,$/,'')} /* subj VARCHAR(64) NULL, */ @@ -215,7 +218,8 @@ module SiSU_DB_create ocns VARCHAR(6), clean TEXT NULL, body TEXT NULL, - seg VARCHAR(120) NULL, + words VARCHAR(3000) NULL, + seg VARCHAR(256) NULL, lev_an VARCHAR(1), lev SMALLINT NULL, lev1 SMALLINT, @@ -254,6 +258,7 @@ module SiSU_DB_create nr SMALLINT, clean TEXT NULL, body TEXT NULL, + words VARCHAR(3000) NULL, ocn SMALLINT, ocnd VARCHAR(6), ocns VARCHAR(6), @@ -276,6 +281,7 @@ module SiSU_DB_create nr SMALLINT, clean TEXT NULL, body TEXT NULL, + words VARCHAR(3000) NULL, ocn SMALLINT, ocnd VARCHAR(6), ocns VARCHAR(6), @@ -298,6 +304,7 @@ module SiSU_DB_create nr SMALLINT, clean TEXT NULL, body TEXT NULL, + words VARCHAR(3000) NULL, ocn SMALLINT, ocnd VARCHAR(6), ocns VARCHAR(6), @@ -430,12 +437,14 @@ module SiSU_DB_create %{#{column.notes_prefix_a.column_comment}}, %{#{column.notes_prefix_b.column_comment}}, %{#{column.notes_suffix.column_comment}}, - %{#{column.filename.column_comment}}, - %{#{column.sisutxt.column_comment}}, + %{#{column.src_filename.column_comment}}, + %{#{column.src_fingerprint.column_comment}}, + %{#{column.src_filesize.column_comment}}, + %{#{column.src_word_count.column_comment}}, + %{#{column.src_txt.column_comment}}, %{#{column.fulltext.column_comment}}, - %{#{column.word_count.column_comment}}, - %{#{column.digest.column_comment}}, %{#{column.skin_name.column_comment}}, + %{#{column.skin_fingerprint.column_comment}}, %{#{column.skin.column_comment}}, %{#{column.links.column_comment}}, ] diff --git a/lib/sisu/v2/db_drop.rb b/lib/sisu/v2/db_drop.rb index 7189da56..35d971af 100644 --- a/lib/sisu/v2/db_drop.rb +++ b/lib/sisu/v2/db_drop.rb @@ -133,37 +133,49 @@ module SiSU_DB_drop ensure end end + def conn_execute_array(sql_arr) + @conn.transaction do |conn| + sql_arr.each do |sql| + conn.execute(sql) + end + end + end def indexes #% drop all indexes - #@conn.do(%{ - # DROP INDEX object_nr ON doc_objects(ocn); - # DROP INDEX body ON doc_objects(body); - # DROP INDEX clean ON doc_objects(clean); - # DROP INDEX lev1 ON doc_objects(lev1); - # DROP INDEX lev2 ON doc_objects(lev2); - # DROP INDEX lev3 ON doc_objects(lev3); - # DROP INDEX lev4 ON doc_objects(lev4); - # DROP INDEX lev5 ON doc_objects(lev5); - # DROP INDEX lev6 ON doc_objects(lev6); - # DROP INDEX endnote_nr ON endnotes(nr); - # DROP INDEX endnote ON endnotes(body); - # DROP INDEX title ON metadata_and_text(title); - # DROP INDEX filename ON metadata_and_text(filename) - # /* - # DROP INDEX object_nr ON doc_objects(ocn) CASCADE; - # DROP INDEX body ON doc_objects(body) CASCADE; - # DROP INDEX clean ON doc_objects(clean) CASCADE; - # DROP INDEX lev1 ON doc_objects(lev1) CASCADE; - # DROP INDEX lev2 ON doc_objects(lev2) CASCADE; - # DROP INDEX lev3 ON doc_objects(lev3) CASCADE; - # DROP INDEX lev4 ON doc_objects(lev4) CASCADE; - # DROP INDEX lev5 ON doc_objects(lev5) CASCADE; - # DROP INDEX lev6 ON doc_objects(lev6) CASCADE; - # DROP INDEX endnote_nr ON endnotes(nr) CASCADE; - # DROP INDEX endnote ON endnotes(body) CASCADE; - # DROP INDEX title ON metadata_and_text(title) CASCADE; - # DROP INDEX filename ON metadata_and_text(filename) CASCADE - # */ - #}) + print "\n drop documents common indexes\n" unless @opt.cmd =~/q/ + sql_arr=[ + %{DROP INDEX idx_text_words;}, + %{DROP INDEX idx_title;}, + %{DROP INDEX idx_author;}, + %{DROP INDEX idx_filename;}, + %{DROP INDEX idx_topics;}, + %{DROP INDEX idx_ocn;}, + %{DROP INDEX idx_digest_clean;}, + %{DROP INDEX idx_digest_all;}, + %{DROP INDEX idx_lev1;}, + %{DROP INDEX idx_lev2;}, + %{DROP INDEX idx_lev3;}, + %{DROP INDEX idx_lev4;}, + %{DROP INDEX idx_lev5;}, + %{DROP INDEX idx_lev6;}, + %{DROP INDEX idx_endnote_words;}, + %{DROP INDEX idx_endnote_nr;}, + %{DROP INDEX idx_digest_en;}, + %{DROP INDEX idx_endnote_words_asterisk;}, + %{DROP INDEX idx_endnote_nr_asterisk;}, + %{DROP INDEX idx_endnote_asterisk;}, + %{DROP INDEX idx_digest_en_asterisk;}, + %{DROP INDEX idx_endnote_words_plus;}, + %{DROP INDEX idx_endnote_nr_plus;}, + %{DROP INDEX idx_endnote_plus;}, + %{DROP INDEX idx_digest_en_plus}, + ] + conn_execute_array(sql_arr) + print "\n drop documents TEXT indexes\n" unless @opt.cmd =~/q/ + sql_arr=[ + %{DROP INDEX idx_clean;}, + %{DROP INDEX idx_endnote}, + ] + conn_execute_array(sql_arr) end self end diff --git a/lib/sisu/v2/db_import.rb b/lib/sisu/v2/db_import.rb index 45aca11b..e351f6fc 100644 --- a/lib/sisu/v2/db_import.rb +++ b/lib/sisu/v2/db_import.rb @@ -122,7 +122,7 @@ module SiSU_DB_import tell.puts_blue unless @opt.cmd =~/q/ tell=SiSU_Screen::Ansi.new(@opt.cmd,'Marshal Load',@fnc) tell.print_grey if @opt.cmd =~/v/ - select_first_match=%{ SELECT metadata_and_text.tid FROM metadata_and_text WHERE metadata_and_text.filename = '#{@opt.fns}'; } + select_first_match=%{ SELECT metadata_and_text.tid FROM metadata_and_text WHERE metadata_and_text.src_filename = '#{@opt.fns}'; } file_exist=@sql_type=~/sqlite/ \ ? @conn.get_first_value(select_first_match) \ : @conn.select_one(select_first_match) @@ -265,7 +265,10 @@ module SiSU_DB_import @col[:body]=SiSU_Format_Shared::CSS_Format.new(@md,data).lev4_minus special_character_escape(@col[:body]) @col[:plaintext]=@col[:body].dup - strip_markup(@col[:plaintext]) + @col[:plaintext]=strip_markup(@col[:plaintext]) + @col[:plaintext]=clean_searchable_text(@col[:plaintext]) + @col[:words]=@col[:plaintext].dup + @col[:words]=unique_words(@col[:words]) if @en[0]; @en_a,@en_z=@en[0].first,@en[0].last end if @en_ast[0]; @en_a_asterisk,@en_z_asterisk=@en_ast[0].first,@en_ast[0].last @@ -298,7 +301,10 @@ module SiSU_DB_import @col[:body]=SiSU_Format_Shared::CSS_Format.new(@md,data).lev4_plus special_character_escape(@col[:body]) @col[:plaintext]=@col[:body].dup - strip_markup(@col[:plaintext]) + @col[:plaintext]=strip_markup(@col[:plaintext]) + @col[:plaintext]=clean_searchable_text(@col[:plaintext]) + @col[:words]=@col[:plaintext].dup + @col[:words]=unique_words(@col[:words]) @en_a,@en_z=@en[0].first,@en[0].last if @en[0] @en_a_asterisk,@en_z_asterisk=@en_ast[0].first,@en_ast[0].last if @en_ast[0] @en_a_plus,@en_z_plus=@en_pls[0].first,@en_pls[0].last if @en_pls[0] @@ -324,7 +330,10 @@ module SiSU_DB_import @col[:body]=SiSU_Format_Shared::CSS_Format.new(@md,data).lev4_plus special_character_escape(@col[:body]) @col[:plaintext]=@col[:body].dup - strip_markup(@col[:plaintext]) + @col[:plaintext]=strip_markup(@col[:plaintext]) + @col[:plaintext]=clean_searchable_text(@col[:plaintext]) + @col[:words]=@col[:plaintext].dup + @col[:words]=unique_words(@col[:words]) @en_a,@en_z=@en[0].first,@en[0].last if @en[0] @en_a_asterisk,@en_z_asterisk=@en_ast[0].first,@en_ast[0].last if @en_ast[0] @en_a_plus,@en_z_plus=@en_pls[0].first,@en_pls[0].last if @en_pls[0] @@ -350,7 +359,10 @@ module SiSU_DB_import @col[:body]=SiSU_Format_Shared::CSS_Format.new(@md,data).lev4_plus special_character_escape(@col[:body]) @col[:plaintext]=@col[:body].dup - strip_markup(@col[:plaintext]) + @col[:plaintext]=strip_markup(@col[:plaintext]) + @col[:plaintext]=clean_searchable_text(@col[:plaintext]) + @col[:words]=@col[:plaintext].dup + @col[:words]=unique_words(@col[:words]) @en_a,@en_z=@en[0].first,@en[0].last if @en[0] @en_a_asterisk,@en_z_asterisk=@en_ast[0].first,@en_ast[0].last if @en_ast[0] @en_a_plus,@en_z_plus=@en_pls[0].first,@en_pls[0].last if @en_pls[0] @@ -389,12 +401,15 @@ module SiSU_DB_import end special_character_escape(@col[:body]) @col[:plaintext]=@col[:body].dup - strip_markup(@col[:plaintext]) + @col[:plaintext]=strip_markup(@col[:plaintext]) + @col[:plaintext]=clean_searchable_text(@col[:plaintext]) + @col[:words]=@col[:plaintext].dup + @col[:words]=unique_words(@col[:words]) t=SiSU_DB_tuple::Load_documents.new(@conn,@col,@opt,@file) @tuple_array << t.tuple @en,@en_ast,@en_pls=[],[],[] @col[:en_a]=@col[:en_z]=nil - @col[:lev]=@col[:plaintext]=@col[:body]='' + @col[:lev]=@col[:plaintext]=@col[:body]=@col[:words]='' end if notedata =~/#{Mx[:en_a_o]}.+?#{Mx[:en_a_c]}/ #% import into database endnotes tables endnote_array=notedata.scan(/#{Mx[:en_a_o]}.+?#{Mx[:en_a_c]}/) @@ -406,9 +421,9 @@ module SiSU_DB_import @id_n+=1 special_character_escape(txt) body=SiSU_Format_Shared::CSS_Format.new(@md,data).endnote(nr,txt) - #special_character_escape(body) - #special_character_escape(txt) strip_markup(txt) + words=txt.dup + words=unique_words(words) if txt.size > (SiSU_DB_columns::Column_size.new.endnote_clean - 1) puts "\n\nTOO LARGE (ENDNOTE - see error log)\n\n" open("#{Dir.pwd}/pg_documents_error_log",'a') do |error| @@ -418,16 +433,17 @@ module SiSU_DB_import end if txt en={ :type => 'endnotes', - :id => @id_n, - :lid => @col[:lid], - :nr => nr, - :txt => txt, - :body => body, - :ocn => @col[:ocn], - :ocnd => @col[:ocnd], - :ocns => @col[:ocns], - :id_t => @@id_t, - :hash => digest_clean + :id => @id_n, + :lid => @col[:lid], + :nr => nr, + :txt => txt, + :body => body, + :words => words, + :ocn => @col[:ocn], + :ocnd => @col[:ocnd], + :ocns => @col[:ocns], + :id_t => @@id_t, + :hash => digest_clean } t=SiSU_DB_tuple::Load_endnotes.new(@conn,en,@opt,@file) @tuple_array << t.tuple @@ -447,6 +463,8 @@ module SiSU_DB_import special_character_escape(txt) body=SiSU_Format_Shared::CSS_Format.new(@md,data).endnote(nr,txt) strip_markup(txt) + words=txt.dup + words=unique_words(words) if txt.size > (SiSU_DB_columns::Column_size.new.endnote_clean - 1) puts "\n\nTOO LARGE (ENDNOTE - see error log)\n\n" open("#{Dir.pwd}/pg_documents_error_log",'a') do |error| @@ -456,16 +474,17 @@ module SiSU_DB_import end if txt en={ :type => 'endnotes_asterisk', - :id => @id_n, - :lid => @col[:lid], - :nr => nr, - :txt => txt, - :body => body, - :ocn => @col[:ocn], - :ocnd => @col[:ocnd], - :ocns => @col[:ocns], - :id_t => @@id_t, - :hash => digest_clean + :id => @id_n, + :lid => @col[:lid], + :nr => nr, + :txt => txt, + :body => body, + :words => words, + :ocn => @col[:ocn], + :ocnd => @col[:ocnd], + :ocns => @col[:ocns], + :id_t => @@id_t, + :hash => digest_clean } t=SiSU_DB_tuple::Load_endnotes.new(@conn,en,@opt,@file) @tuple_array << t.tuple @@ -485,6 +504,8 @@ module SiSU_DB_import special_character_escape(txt) body=SiSU_Format_Shared::CSS_Format.new(@md,data).endnote(nr,txt) strip_markup(txt) + words=txt.dup + words=unique_words(words) if txt.size > (SiSU_DB_columns::Column_size.new.endnote_clean - 1) puts "\n\nTOO LARGE (ENDNOTE - see error log)\n\n" open("#{Dir.pwd}/pg_documents_error_log",'a') do |error| @@ -494,16 +515,17 @@ module SiSU_DB_import end if txt en={ :type => 'endnotes_plus', - :id => @id_n, - :lid => @col[:lid], - :nr => nr, - :txt => txt, - :body => body, - :ocn => @col[:ocn], - :ocnd => @col[:ocnd], - :ocns => @col[:ocns], - :id_t => @@id_t, - :hash => digest_clean + :id => @id_n, + :lid => @col[:lid], + :nr => nr, + :txt => txt, + :body => body, + :words => words, + :ocn => @col[:ocn], + :ocnd => @col[:ocnd], + :ocns => @col[:ocns], + :id_t => @@id_t, + :hash => digest_clean } t=SiSU_DB_tuple::Load_endnotes.new(@conn,en,@opt,@file) @tuple_array << t.tuple @@ -526,25 +548,25 @@ module SiSU_DB_import endnotes(@txt).range @en << endnotes(@txt).standard if @txt =~/#{Mx[:en_a_o]}.+?#{Mx[:en_a_c]}/ @en_ast << endnotes(@txt).asterisk if @txt =~/#{Mx[:en_b_o]}\*.+?#{Mx[:en_b_c]}/ - @en_pls << endnotes(@txt).plus if @txt =~/#{Mx[:en_b_o]}\+.+?#{Mx[:en_b_o]}/ + @en_pls << endnotes(@txt).plus if @txt =~/#{Mx[:en_b_o]}\+.+?#{Mx[:en_b_c]}/ @txt=endnotes(@txt).clean_text end @txt end def standard - x=if @txt =~/#{Mx[:en_a_o]}.+?#{Mx[:en_a_c]}/; @txt.scan(/#{Mx[:en_a_o]}(\d+).+?#{Mx[:en_a_c]}/) - else nil - end + x=(@txt =~/#{Mx[:en_a_o]}.+?#{Mx[:en_a_c]}/) \ + ? @txt.scan(/#{Mx[:en_a_o]}(\d+).+?#{Mx[:en_a_c]}/) \ + : nil end def asterisk - x=if @txt =~/#{Mx[:en_b_o]}\*.+?#{Mx[:en_b_c]}/; @txt.scan(/#{Mx[:en_b_o]}[*](\d+).+?#{Mx[:en_b_c]}/) - else nil - end + x=(@txt =~/#{Mx[:en_b_o]}\*.+?#{Mx[:en_b_c]}/) \ + ? @txt.scan(/#{Mx[:en_b_o]}[*](\d+).+?#{Mx[:en_b_c]}/) \ + : nil end def plus - x=if @txt =~/#{Mx[:en_b_o]}\+.+?#{Mx[:en_b_c]}/; @txt.scan(/#{Mx[:en_b_o]}[+](\d+).+?#{Mx[:en_b_c]}/) - else nil - end + x=(@txt =~/#{Mx[:en_b_o]}\+.+?#{Mx[:en_b_c]}/) \ + ? @txt.scan(/#{Mx[:en_b_o]}[+](\d+).+?#{Mx[:en_b_c]}/) \ + : nil end def clean_text(base_url=nil) if base_url diff --git a/lib/sisu/v2/db_indexes.rb b/lib/sisu/v2/db_indexes.rb index 3cbcc20c..fb5ddd76 100644 --- a/lib/sisu/v2/db_indexes.rb +++ b/lib/sisu/v2/db_indexes.rb @@ -73,38 +73,44 @@ module SiSU_DB_index def base print "\n create documents common indexes\n" unless @opt.cmd =~/q/ sql_arr=[ - %{CREATE INDEX object_nr ON doc_objects(ocn);}, - %{CREATE INDEX digest_clean ON doc_objects(digest_clean);}, - %{CREATE INDEX digest_all ON doc_objects(digest_all);}, - %{CREATE INDEX lev1 ON doc_objects(lev1);}, - %{CREATE INDEX lev2 ON doc_objects(lev2);}, - %{CREATE INDEX lev3 ON doc_objects(lev3);}, - %{CREATE INDEX lev4 ON doc_objects(lev4);}, - %{CREATE INDEX lev5 ON doc_objects(lev5);}, - %{CREATE INDEX lev6 ON doc_objects(lev6);}, - %{CREATE INDEX endnote_nr ON endnotes(nr);}, - %{CREATE INDEX digest_en ON endnotes(digest_clean);}, - %{CREATE INDEX endnote_nr_asterisk ON endnotes_asterisk(nr);}, - %{CREATE INDEX endnote_asterisk ON endnotes_asterisk(clean);}, - %{CREATE INDEX digest_en_asterisk ON endnotes_asterisk(digest_clean);}, - %{CREATE INDEX endnote_nr_plus ON endnotes_plus(nr);}, - %{CREATE INDEX endnote_plus ON endnotes_plus(clean);}, - %{CREATE INDEX digest_en_plus ON endnotes_plus(digest_clean);}, - %{CREATE INDEX title ON metadata_and_text(title);}, - %{CREATE INDEX filename ON metadata_and_text(filename)}, + %{CREATE INDEX idx_text_words ON doc_objects(words);}, + %{CREATE INDEX idx_ocn ON doc_objects(ocn);}, + %{CREATE INDEX idx_digest_clean ON doc_objects(digest_clean);}, + %{CREATE INDEX idx_digest_all ON doc_objects(digest_all);}, + %{CREATE INDEX idx_lev1 ON doc_objects(lev1);}, + %{CREATE INDEX idx_lev2 ON doc_objects(lev2);}, + %{CREATE INDEX idx_lev3 ON doc_objects(lev3);}, + %{CREATE INDEX idx_lev4 ON doc_objects(lev4);}, + %{CREATE INDEX idx_lev5 ON doc_objects(lev5);}, + %{CREATE INDEX idx_lev6 ON doc_objects(lev6);}, + %{CREATE INDEX idx_endnote_words ON endnotes(words);}, + %{CREATE INDEX idx_endnote_nr ON endnotes(nr);}, + %{CREATE INDEX idx_digest_en ON endnotes(digest_clean);}, + %{CREATE INDEX idx_endnote_words_asterisk ON endnotes_asterisk(words);}, + %{CREATE INDEX idx_endnote_nr_asterisk ON endnotes_asterisk(nr);}, + %{CREATE INDEX idx_endnote_asterisk ON endnotes_asterisk(clean);}, + %{CREATE INDEX idx_digest_en_asterisk ON endnotes_asterisk(digest_clean);}, + %{CREATE INDEX idx_endnote_words_plus ON endnotes_plus(words);}, + %{CREATE INDEX idx_endnote_nr_plus ON endnotes_plus(nr);}, + %{CREATE INDEX idx_endnote_plus ON endnotes_plus(clean);}, + %{CREATE INDEX idx_digest_en_plus ON endnotes_plus(digest_clean);}, + %{CREATE INDEX idx_title ON metadata_and_text(title);}, + %{CREATE INDEX idx_author ON metadata_and_text(creator_author);}, + %{CREATE INDEX idx_filename ON metadata_and_text(src_filename);}, + %{CREATE INDEX idx_topics ON metadata_and_text(classify_topic_register)}, ] conn_execute_array(sql_arr) end def text - print "\n create documents text indexes\n" unless @opt.cmd =~/q/ + print "\n create documents TEXT indexes\n" unless @opt.cmd =~/q/ sql_arr=[ - %{CREATE INDEX clean ON doc_objects(clean);}, - %{CREATE INDEX endnote ON endnotes(clean);} + %{CREATE INDEX idx_clean ON doc_objects(clean);}, + %{CREATE INDEX idx_endnote ON endnotes(clean);} ] conn_execute_array(sql_arr) end base - @opt.cmd=~/D/ || ((@opt.mod=~/psql/) ? '' : text) + text #@opt.cmd=~/D/ || ((@opt.mod=~/psql/) ? '' : text) end end end diff --git a/lib/sisu/v2/db_load_tuple.rb b/lib/sisu/v2/db_load_tuple.rb index cc00b74a..23b6249d 100644 --- a/lib/sisu/v2/db_load_tuple.rb +++ b/lib/sisu/v2/db_load_tuple.rb @@ -79,11 +79,11 @@ module SiSU_DB_tuple end def tuple #% import line sql_entry=if @col[:en_a] - "INSERT INTO doc_objects (lid, metadata_tid, lev, lev_an, clean, body, ocn, ocnd, ocns, seg, lev1, lev2, lev3, lev4, lev5, lev6, en_a, en_z, t_of, t_is, node, parent, digest_clean, digest_all) " + - "VALUES (#{@col[:lid]}, #{@col[:tid]}, #{@col[:lev]}, '#{@col[:lev_an]}', '#{@col[:plaintext]}', '#{@col[:body]}', '#{@col[:ocn]}', '#{@col[:ocnd]}', '#{@col[:ocns]}', '#{@col[:seg]}', '#{@col[:lv1]}', '#{@col[:lv2]}', '#{@col[:lv3]}', '#{@col[:lv4]}', '#{@col[:lv5]}', '#{@col[:lv6]}', '#{@col[:en_a]}', '#{@col[:en_z]}', '#{@col[:t_of]}', '#{@col[:t_is]}', '#{@col[:node]}', '#{@col[:parent]}', '#{@col[:digest_clean]}', '#{@col[:digest_all]}');" + "INSERT INTO doc_objects (lid, metadata_tid, lev, lev_an, clean, body, words, ocn, ocnd, ocns, seg, lev1, lev2, lev3, lev4, lev5, lev6, en_a, en_z, t_of, t_is, node, parent, digest_clean, digest_all) " + + "VALUES (#{@col[:lid]}, #{@col[:tid]}, #{@col[:lev]}, '#{@col[:lev_an]}', '#{@col[:plaintext]}', '#{@col[:body]}', '#{@col[:words]}', '#{@col[:ocn]}', '#{@col[:ocnd]}', '#{@col[:ocns]}', '#{@col[:seg]}', '#{@col[:lv1]}', '#{@col[:lv2]}', '#{@col[:lv3]}', '#{@col[:lv4]}', '#{@col[:lv5]}', '#{@col[:lv6]}', '#{@col[:en_a]}', '#{@col[:en_z]}', '#{@col[:t_of]}', '#{@col[:t_is]}', '#{@col[:node]}', '#{@col[:parent]}', '#{@col[:digest_clean]}', '#{@col[:digest_all]}');" else - "INSERT INTO doc_objects (lid, metadata_tid, lev, lev_an, clean, body, ocn, ocnd, ocns, seg, lev1, lev2, lev3, lev4, lev5, lev6, t_of, t_is, node, parent, digest_clean, digest_all) " + - "VALUES (#{@col[:lid]}, #{@col[:tid]}, #{@col[:lev]}, '#{@col[:lev_an]}', '#{@col[:plaintext]}', '#{@col[:body]}', '#{@col[:ocn]}', '#{@col[:ocnd]}', '#{@col[:ocns]}', '#{@col[:seg]}', '#{@col[:lv1]}', '#{@col[:lv2]}', '#{@col[:lv3]}', '#{@col[:lv4]}', '#{@col[:lv5]}', '#{@col[:lv6]}', '#{@col[:t_of]}', '#{@col[:t_is]}', '#{@col[:node]}', '#{@col[:parent]}', '#{@col[:digest_clean]}', '#{@col[:digest_all]}');" + "INSERT INTO doc_objects (lid, metadata_tid, lev, lev_an, clean, body, words, ocn, ocnd, ocns, seg, lev1, lev2, lev3, lev4, lev5, lev6, t_of, t_is, node, parent, digest_clean, digest_all) " + + "VALUES (#{@col[:lid]}, #{@col[:tid]}, #{@col[:lev]}, '#{@col[:lev_an]}', '#{@col[:plaintext]}', '#{@col[:body]}', '#{@col[:words]}', '#{@col[:ocn]}', '#{@col[:ocnd]}', '#{@col[:ocns]}', '#{@col[:seg]}', '#{@col[:lv1]}', '#{@col[:lv2]}', '#{@col[:lv3]}', '#{@col[:lv4]}', '#{@col[:lv5]}', '#{@col[:lv6]}', '#{@col[:t_of]}', '#{@col[:t_is]}', '#{@col[:node]}', '#{@col[:parent]}', '#{@col[:digest_clean]}', '#{@col[:digest_all]}');" end if @opt.cmd =~/M/ if @opt.cmd =~/V/ @@ -191,12 +191,14 @@ module SiSU_DB_tuple #{@tp.column.notes_prefix_a.tuple[0]} #{@tp.column.notes_prefix_b.tuple[0]} #{@tp.column.notes_suffix.tuple[0]} -#{@tp.column.filename.tuple[0]} -#{@tp.column.sisutxt.tuple[0]} +#{@tp.column.src_filename.tuple[0]} +#{@tp.column.src_fingerprint.tuple[0]} +#{@tp.column.src_filesize.tuple[0]} +#{@tp.column.src_word_count.tuple[0]} +#{@tp.column.src_txt.tuple[0]} #{@tp.column.fulltext.tuple[0]} -#{@tp.column.word_count.tuple[0]} -#{@tp.column.digest.tuple[0]} #{@tp.column.skin_name.tuple[0]} +#{@tp.column.skin_fingerprint.tuple[0]} #{@tp.column.skin.tuple[0]} #{@tp.column.links.tuple[0]} tid) @@ -272,12 +274,14 @@ tid) #{@tp.column.notes_prefix_a.tuple[1]} #{@tp.column.notes_prefix_b.tuple[1]} #{@tp.column.notes_suffix.tuple[1]} -#{@tp.column.filename.tuple[1]} -#{@tp.column.sisutxt.tuple[1]} +#{@tp.column.src_filename.tuple[1]} +#{@tp.column.src_fingerprint.tuple[1]} +#{@tp.column.src_filesize.tuple[1]} +#{@tp.column.src_word_count.tuple[1]} +#{@tp.column.src_txt.tuple[1]} #{@tp.column.fulltext.tuple[1]} -#{@tp.column.word_count.tuple[1]} -#{@tp.column.digest.tuple[1]} #{@tp.column.skin_name.tuple[1]} +#{@tp.column.skin_fingerprint.tuple[1]} #{@tp.column.skin.tuple[1]} #{@tp.column.links.tuple[1]} #{@id} @@ -311,8 +315,8 @@ tid) @conn,@en,@opt,@file=conn,en,opt,file end def tuple - sql_entry="INSERT INTO #{@en[:type]} (nid, document_lid, nr, clean, body, ocn, ocnd, ocns, metadata_tid, digest_clean) " + - "VALUES ('#{@en[:id]}', '#{@en[:lid]}', '#{@en[:nr]}', '#{@en[:txt]}', '#{@en[:body]}', '#{@en[:ocn]}', '#{@en[:ocnd]}', '#{@en[:ocns]}', '#{@en[:id_t]}', '#{@en[:hash]}');" + sql_entry="INSERT INTO #{@en[:type]} (nid, document_lid, nr, clean, body, words, ocn, ocnd, ocns, metadata_tid, digest_clean) " + + "VALUES ('#{@en[:id]}', '#{@en[:lid]}', '#{@en[:nr]}', '#{@en[:txt]}', '#{@en[:body]}', '#{@en[:words]}', '#{@en[:ocn]}', '#{@en[:ocnd]}', '#{@en[:ocns]}', '#{@en[:id_t]}', '#{@en[:hash]}');" if @opt.cmd =~/M/ @file.puts sql_entry else diff --git a/lib/sisu/v2/db_remove.rb b/lib/sisu/v2/db_remove.rb index e7942a15..5a7f1244 100644 --- a/lib/sisu/v2/db_remove.rb +++ b/lib/sisu/v2/db_remove.rb @@ -72,9 +72,9 @@ module SiSU_DB_remove : false end del_id=if driver_sqlite3 - @conn.get_first_value(%{ SELECT tid FROM metadata_and_text WHERE filename = '#{@opt.fns}'; }).to_i + @conn.get_first_value(%{ SELECT tid FROM metadata_and_text WHERE src_filename = '#{@opt.fns}'; }).to_i else - x=@conn.select_one(%{ SELECT metadata_and_text.tid FROM metadata_and_text WHERE metadata_and_text.filename = '#{@opt.fns}'; }) + x=@conn.select_one(%{ SELECT metadata_and_text.tid FROM metadata_and_text WHERE metadata_and_text.src_filename = '#{@opt.fns}'; }) x ? (x.join.to_i) : nil end if del_id diff --git a/lib/sisu/v2/db_sqltxt.rb b/lib/sisu/v2/db_sqltxt.rb index 68e9ef8a..17a92683 100644 --- a/lib/sisu/v2/db_sqltxt.rb +++ b/lib/sisu/v2/db_sqltxt.rb @@ -72,9 +72,10 @@ module SiSU_DB_text end def clean_searchable_text(arr) #produce clean, searchable, plaintext from document source txt_arr,en=[],[] + arr=arr.class==String ? arr.split(/\n+/m) : arr arr.each do |s| - s.gsub!(/([*\/_-])\{(.+?)\}\1/,'\2') - s.gsub!(/^(?:group|poem|code)\{/,''); s.gsub!(/^\}(?:group|poem|code)/,'') + s.gsub!(/([*\/_-])\{(.+?)\}\1/m,'\2') + s.gsub!(/^(?:group|poem|code)\{/m,''); s.gsub!(/^\}(?:group|poem|code)/m,'') s.gsub!(/\A(?:@\S+:\s+.+)\Z/m,'') if s =~/^:A~/ if defined? @md.creator \ @@ -82,26 +83,26 @@ module SiSU_DB_text and not @md.creator.author.empty? s.gsub!(/@author/,@md.creator.author) else - tell=SiSU_Screen::Ansi.new('v','WARNING Document Author information missing; provide @creator: :author:') - tell.warn + tell=SiSU_Screen::Ansi.new('v','WARNING Document Author information missing; provide @creator: :author:',@md.fnb) + tell.warn unless @md.cmd.inspect =~/q/ end if defined? @md.title \ and defined? @md.title.full \ and not @md.title.full.empty? s.gsub!(/@title/,@md.title.full) else - tell=SiSU_Screen::Ansi.new('v','WARNING Document Title missing; provide @title:') - tell.warn + tell=SiSU_Screen::Ansi.new('v','WARNING Document Title missing; provide @title:',@md.fnb) + tell.warn unless @md.cmd.inspect =~/q/ end end - s.gsub!(/^(?:_[1-9]\*?|_\*)\s+/,'') - s.gsub!(/^(?:[1-9]\~(\S+)?)\s+/,'') - s.gsub!(/^(?::?[A-C]\~(\S+)?)\s+/,'') - s.gsub!(/^%{1,3} .+/,'') #removed even if contained in code block - s.gsub!(/
/,' ') - en << s.scan(/~\{\s*(.+?)\s*\}~/) - s.gsub!(/~\{.+?\}~/,'') - s.gsub!(/ \s+/,' ') + s.gsub!(/^(?:_[1-9]\*?|_\*)\s+/m,'') + s.gsub!(/^(?:[1-9]\~(\S+)?)\s+/m,'') + s.gsub!(/^(?::?[A-C]\~(\S+)?)\s+/m,'') + s.gsub!(/^%{1,3} .+/m,'') #removed even if contained in code block + s.gsub!(/
/m,' ') + en << s.scan(/~\{\s*(.+?)\s*\}~/m) + s.gsub!(/~\{.+?\}~/m,'') + s.gsub!(/ \s+/m,' ') #special_character_escape(s) s end @@ -124,6 +125,11 @@ module SiSU_DB_text str.strip! str end + def unique_words(str) + a=str.scan(/[a-zA-Z0-9\\\/_-]{2,}/) #a=str.scan(/\S+{2,}/) + str=a.uniq.sort.join(' ') + str + end end end __END__ diff --git a/lib/sisu/v2/param.rb b/lib/sisu/v2/param.rb index ef91f10d..87dd9aab 100644 --- a/lib/sisu/v2/param.rb +++ b/lib/sisu/v2/param.rb @@ -162,8 +162,8 @@ module SiSU_Param puts "#{n} is #{s.class}: programming error, String expected #{__FILE__}:#{__LINE__}" s else - tell=SiSU_Screen::Ansi.new('v',"#{n} length #{s.length} exceeds set db field length #{l}, metadata dropped") - tell.warn + tell=SiSU_Screen::Ansi.new('v',"#{n} length #{s.length} exceeds set db field length #{l}, metadata dropped",@opt.fns) + tell.warn unless @opt.cmd =~/q/ nil end end @@ -244,7 +244,9 @@ module SiSU_Param validate_length(s,l,n) end def note - @h['note'] #TEXT + s=@h['note'] + l,n=Db[:col_info_note],'title.note' + validate_length(s,l,n) end def short s=(@h['short'] ? @h['short'] : @h['main']) @@ -380,9 +382,9 @@ module SiSU_Param def rights a=@s.split(/[ ]*\n[ ]*/m) @h=build_hash(a) - def copyright # TEXT used db sql + def copyright def text #you may wish to expand to take from all - r=if @h['copyright'] + s=if @h['copyright'] @h['copyright'] elsif @h['text'] @h['text'] @@ -393,26 +395,40 @@ module SiSU_Param tell.warn unless @opt.cmd =~/q/ '' end + l,n=Db[:col_info_note],'rights.copyright.text' + validate_length(s,l,n) end def translation - r=(@h['translation'] ? @h['translation'] : nil) + s=(@h['translation'] ? @h['translation'] : nil) + l,n=Db[:col_info_note],'rights.copyright.translation' + validate_length(s,l,n) end def illustrations - r=(@h['illustrations'] ? @h['illustrations'] : nil) + s=(@h['illustrations'] ? @h['illustrations'] : nil) + l,n=Db[:col_info_note],'rights.copyright.illustrations' + validate_length(s,l,n) end def photographs - r=(@h['photographs'] ? @h['photographs'] : nil) + s=(@h['photographs'] ? @h['photographs'] : nil) + l,n=Db[:col_info_note],'rights.copyright.photographs' + validate_length(s,l,n) end def digitiztion - r=(@h['digitization'] ? @h['digitization'] : nil) + s=(@h['digitization'] ? @h['digitization'] : nil) + l,n=Db[:col_info_note],'rights.copyright.digitization' + validate_length(s,l,n) end def audio - r=(@h['audio'] ? @h['audio'] : nil) + s=(@h['audio'] ? @h['audio'] : nil) + l,n=Db[:col_info_note],'rights.copyright.audio' + validate_length(s,l,n) end self end def license - r=(@h['license'] ? @h['license'] : nil) + s=(@h['license'] ? @h['license'] : nil) + l,n=Db[:col_info_note],'rights.license' + validate_length(s,l,n) end def all s=if @h['all']; @h['all'] @@ -445,6 +461,9 @@ module SiSU_Param if s.empty? tell=SiSU_Screen::Ansi.new(@cmd,'WARNING Document Rights information missing; provide @rights: :copyright:') tell.warn unless @opt.cmd =~/q/ + else + l,n=Db[:col_info_note],'rights.all' + validate_length(s,l,n) end s end @@ -472,7 +491,7 @@ module SiSU_Param end def topic_register s=@h['topic_register'] - l,n=Db[:col_classify_txt_long],'classify.topic_register' + l,n=Db[:col_info_note],'classify.topic_register' validate_length(s,l,n) end def type @@ -743,9 +762,9 @@ module SiSU_Param @doc={ :lv=>[] } @doc[:fns],@doc[:fnb],@doc[:scr_suffix]='','','' @@publisher='SiSU scribe' - attr_accessor :cmd,:make,:mod,:env,:fn,:fns,:fnb,:fnn,:fnt,:fnl,:flv,:fnz,:fnstex,:ocn,:sfx_src,:pdf,:file_type,:dir_out,:dir_tex,:dir_lout,:txt_path,:site_skin,:sisu,:sisu_version,:ruby_version,:title,:subtitle,:full_title,:html_title,:subtitle_tex,:creator,:classify,:author_home,:author,:author_title,:author_nationality,:authors,:authorship,:translator,:illustrator,:prepared_by,:digitized_by,:subject,:description,:publisher,:contributor,:date,:date_created,:date_issued,:date_available,:date_valid,:date_modified,:date_translated,:date_added_to_site,:date_scheme,:date_created_scheme,:date_issued_scheme,:date_available_scheme,:date_valid_scheme,:date_modified_scheme,:type,:format,:identifier,:source,:language,:language_original,:relation,:coverage,:rights,:keywords,:comments,:abstract,:cls_loc,:cls_dewey,:cls_pg,:cls_isbn,:papersize,:papersize_array,:toc,:lv1,:lv2,:lv3,:lv4,:lv5,:lv6,:lvs,:pagenew,:pagebreak,:num_top,:toc_lev_limit,:flag_endnotes,:flag_auto_endnotes,:flag_separate_endnotes,:flag_separate_endnotes_make,:markup,:markup_instruction,:markup_version,:markup_declared,:flag_tables,:vocabulary,:doc_skin,:doc_css,:yaml,:lnk,:prefix_a,:prefix_b,:suffix,:information,:contact,:icon,:image,:ad_url,:ad_png,:ad_alt,:ad_began,:flag_promo,:promo,:ad_home,:stmp,:stmpd,:sc_filename,:sc_number,:sc_date,:sc_time,:sc_info,:yamladdr,:locale,:wc_lines,:wc_words,:wc_bytes,:file_encoding,:file_size,:user,:home,:hostname,:pwd,:firstseg,:programs,:author_copymark,:lang,:en,:dgst,:dgst_skin,:generated,:tags,:tag_array,:concord_make,:seg_names,:seg_autoname_safe,:set_header_title,:set_heading_top,:set_heading_seg,:heading_seg_first,:heading_seg_first_flag,:base_program,:ec,:opt,:sem_tag,:book_idx,:topic_register,:topic_register_array,:original_publication,:original_publication_date,:original_publication_nationality,:original_publication_institution,:writing_focus,:audio,:daisy + attr_accessor :cmd,:make,:mod,:env,:fn,:fns,:fnb,:fnn,:fnt,:fnl,:flv,:fnz,:fnstex,:ocn,:sfx_src,:pdf,:file_type,:dir_out,:dir_tex,:dir_lout,:txt_path,:site_skin,:sisu,:sisu_version,:ruby_version,:title,:subtitle,:full_title,:html_title,:subtitle_tex,:creator,:classify,:author_home,:author,:author_title,:author_nationality,:authors,:authorship,:translator,:illustrator,:prepared_by,:digitized_by,:subject,:description,:publisher,:contributor,:date,:date_created,:date_issued,:date_available,:date_valid,:date_modified,:date_translated,:date_added_to_site,:date_scheme,:date_created_scheme,:date_issued_scheme,:date_available_scheme,:date_valid_scheme,:date_modified_scheme,:type,:format,:identifier,:source,:language,:language_original,:relation,:coverage,:rights,:keywords,:comments,:abstract,:cls_loc,:cls_dewey,:cls_pg,:cls_isbn,:papersize,:papersize_array,:toc,:lv1,:lv2,:lv3,:lv4,:lv5,:lv6,:lvs,:pagenew,:pagebreak,:num_top,:toc_lev_limit,:flag_endnotes,:flag_auto_endnotes,:flag_separate_endnotes,:flag_separate_endnotes_make,:markup,:markup_instruction,:markup_version,:markup_declared,:flag_tables,:vocabulary,:doc_skin,:doc_css,:yaml,:lnk,:prefix_a,:prefix_b,:suffix,:information,:contact,:icon,:image,:ad_url,:ad_png,:ad_alt,:ad_began,:flag_promo,:promo,:ad_home,:stmp,:stmpd,:sc_filename,:sc_number,:sc_date,:sc_time,:sc_info,:yamladdr,:locale,:wc_lines,:wc_words,:wc_bytes,:file_encoding,:filesize,:user,:home,:hostname,:pwd,:firstseg,:programs,:author_copymark,:lang,:en,:dgst,:dgst_skin,:generated,:tags,:tag_array,:concord_make,:seg_names,:seg_autoname_safe,:set_header_title,:set_heading_top,:set_heading_seg,:heading_seg_first,:heading_seg_first_flag,:base_program,:ec,:opt,:sem_tag,:book_idx,:topic_register,:topic_register_array,:original_publication,:original_publication_date,:original_publication_nationality,:original_publication_institution,:writing_focus,:audio,:daisy def initialize(fns_array,opt) - @env=@fn=@fns=@fnb=@fnn=@fnt=@fnl=@flv=@fnz=@fnstex=@ocn=@sfx_src=@pdf=@file_type=@dir_out=@dir_tex=@dir_lout=@txt_path=@make=@flag_endnotes=@flag_auto_endnotes=@flag_separate_endnotes=@flag_separate_endnotes_make=@site_skin=@sisu=@sisu_version=@ruby_version=@title=@subtitle=@full_title=@html_title=@subtitle_tex=@creator=@classify=@author_home=@author=@author_title=@author_nationality=@translator=@illustrator=@prepared_by=@digitized_by=@subject=@description=@publisher=@contributor=@date=@date_created=@date_issued=@date_available=@date_valid=@date_modified=@date_translated=@date_added_to_site=@date_scheme=@date_created_scheme=@date_issued_scheme=@date_available_scheme=@date_valid_scheme=@date_modified_scheme=@type=@format=@identifier=@source=@language=@language_original=@relation=@coverage=@rights=@keywords=@comments=@abstract=@cls_loc=@cls_dewey=@cls_pg=@cls_isbn=@papersize=@toc=@lv1=@lv2=@lv3=@lv4=@lv5=@lv6=@pagenew=@pagebreak=@num_top=@toc_lev_limit=@flag_tables=@vocabulary=@doc_skin=@doc_css=@yaml=@lnk=@prefix_a=@prefix_b=@suffix=@information=@contact=@icon=@ad_url=@ad_png=@ad_alt=@ad_began=@promo=@ad_home=@stmp=@stmpd=@sc_filename=@sc_number=@sc_date=@sc_time=@sc_info=@yamladdr=@locale=@wc_lines=@wc_words=@wc_bytes=@file_encoding=@file_size=@firstseg=@programs=@author_copymark=@lang=@en=@dgst=@dgst_skin=@generated=@heading_seg_first=@base_program=@topic_register=@original_publication_details=@original_publication=@original_publication_date=@original_publication_nationality=@original_publication_institution=@writing_focus=@audio=nil + @env=@fn=@fns=@fnb=@fnn=@fnt=@fnl=@flv=@fnz=@fnstex=@ocn=@sfx_src=@pdf=@file_type=@dir_out=@dir_tex=@dir_lout=@txt_path=@make=@flag_endnotes=@flag_auto_endnotes=@flag_separate_endnotes=@flag_separate_endnotes_make=@site_skin=@sisu=@sisu_version=@ruby_version=@title=@subtitle=@full_title=@html_title=@subtitle_tex=@creator=@classify=@author_home=@author=@author_title=@author_nationality=@translator=@illustrator=@prepared_by=@digitized_by=@subject=@description=@publisher=@contributor=@date=@date_created=@date_issued=@date_available=@date_valid=@date_modified=@date_translated=@date_added_to_site=@date_scheme=@date_created_scheme=@date_issued_scheme=@date_available_scheme=@date_valid_scheme=@date_modified_scheme=@type=@format=@identifier=@source=@language=@language_original=@relation=@coverage=@rights=@keywords=@comments=@abstract=@cls_loc=@cls_dewey=@cls_pg=@cls_isbn=@papersize=@toc=@lv1=@lv2=@lv3=@lv4=@lv5=@lv6=@pagenew=@pagebreak=@num_top=@toc_lev_limit=@flag_tables=@vocabulary=@doc_skin=@doc_css=@yaml=@lnk=@prefix_a=@prefix_b=@suffix=@information=@contact=@icon=@ad_url=@ad_png=@ad_alt=@ad_began=@promo=@ad_home=@stmp=@stmpd=@sc_filename=@sc_number=@sc_date=@sc_time=@sc_info=@yamladdr=@locale=@wc_lines=@wc_words=@wc_bytes=@file_encoding=@filesize=@firstseg=@programs=@author_copymark=@lang=@en=@dgst=@dgst_skin=@generated=@heading_seg_first=@base_program=@topic_register=@original_publication_details=@original_publication=@original_publication_date=@original_publication_nationality=@original_publication_institution=@writing_focus=@audio=nil @data,@fns,@cmd,@mod,@opt=fns_array,opt.fns,opt.cmd,opt.mod,opt #@data used as data @flag_tables,@set_header_title,@set_heading_top,@set_heading_seg,@heading_seg_first_flag,@flag_promo,@book_idx=false,false,false,false,false,false,false @seg_autoname_safe=true @@ -1107,16 +1126,22 @@ module SiSU_Param @papersize=determine_papersize(@mod.inspect) end @papersize_array=@papersize.scan(/(?:a4|letter|legal|book|a5|b5)/i) + fn=@fns=~/\.ssm\.sst$/ ? @fns.gsub(/.sst/,'') : @fns #decide what to do a filesize on .ssm tells very little about actual document size + @filesize=(File.size(fn)).to_s if @sys.openssl !=false skin=@doc_skin \ ? (SiSU_Env::Info_skin.new(@opt,@doc_skin).select) \ : SiSU_Env::Info_skin.new(@opt).select @dgst,@dgst_skin=[],[] if @env.digest.type =~/sha256/ - @dgst=@sys.sha256(@env.source_file_with_path) + dgst=@sys.sha256(@env.source_file_with_path) + @dgst=dgst[1].length==64 ? dgst : nil + puts 'check document (sha256) digest' if not @dgst @dgst_skin=skin ? (@sys.sha256(skin)) : nil else - @dgst=@sys.md5(@env.source_file_with_path) + dgst=@sys.md5(@env.source_file_with_path) + @dgst=dgst[1].length==32 ? dgst : nil + puts 'check document (md5) digest' if not @dgst @dgst_skin=skin ? (@sys.md5(skin)) : nil end end -- cgit v1.2.3