From 9aeb2343e59d1ca8df6f5e764aa33858b334a783 Mon Sep 17 00:00:00 2001 From: Ralph Amissah Date: Sat, 7 Sep 2013 01:37:52 -0400 Subject: v5: dal, book index, hash key, strip whitespace, a (required) fix --- lib/sisu/v5/dal_doc_str.rb | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'lib') diff --git a/lib/sisu/v5/dal_doc_str.rb b/lib/sisu/v5/dal_doc_str.rb index 5b6c03bf..87e5c908 100644 --- a/lib/sisu/v5/dal_doc_str.rb +++ b/lib/sisu/v5/dal_doc_str.rb @@ -196,7 +196,8 @@ module SiSU_DAL_DocumentStructureExtract ? term_node : (term_node + '+0') term_nodes << term_node - @use,plus=rgx_idx_ocn_seg.match(term_node)[1,2] + use,plus=rgx_idx_ocn_seg.match(term_node)[1,2] + @use=use.strip idx_hash[@use]={ sub: [], plus: plus } unless idx_hash[@use] and defined? idx_hash[@use] when Array subterm_nodes=[] -- cgit v1.2.3 From af29258b1f388ec20a2695db7e3f1122bdbe47c8 Mon Sep 17 00:00:00 2001 From: Ralph Amissah Date: Sat, 7 Sep 2013 01:42:18 -0400 Subject: v5: db, book index, convert hash back to string --- lib/sisu/v5/db_import.rb | 30 +++++++++++++++++++++++++----- 1 file changed, 25 insertions(+), 5 deletions(-) (limited to 'lib') diff --git a/lib/sisu/v5/db_import.rb b/lib/sisu/v5/db_import.rb index 5f89c968..bcb22de8 100644 --- a/lib/sisu/v5/db_import.rb +++ b/lib/sisu/v5/db_import.rb @@ -209,6 +209,26 @@ module SiSU_DbImport end def pf_db_import_transaction_close end + def book_idx_hash_to_str(book_idx) + book_idx=book_idx ? book_idx : '' + book_idx_str,book_subidx_part='','' + if not book_idx.empty? + book_idx_str='' + book_idx.each_pair do |k0,v0| + book_idx_str << %{#{k0}+#{v0[:plus]}} + book_subidx_part='' + if v0[:sub].length > 0 + v0[:sub].each do |subterms| + subterms.each_pair do |k1,v1| + book_subidx_part << %{\n #{k1}+#{v1[:plus]} | } + end + end + book_idx_str=book_idx_str + ':' + book_subidx_part + end + end + end + book_idx_str + end def db_import_metadata #% import documents - populate database print %{ #{@cX.grey}import documents dbi_unit #{@cX.off} } if @opt.cmd =~/vVM/ @tp={} @@ -285,7 +305,7 @@ module SiSU_DbImport plaintext=@col[:body].dup plaintext=strip_markup(plaintext) @col[:plaintext]=clean_searchable_text(plaintext) - book_idx=data.idx ? data.idx : '' + book_idx=book_idx_hash_to_str(data.idx) @col[:book_idx]=clean_searchable_text(book_idx) if @en[0] then @en_a,@en_z=@en[0].first,@en[0].last end @@ -321,7 +341,7 @@ module SiSU_DbImport plaintext=@col[:body].dup plaintext=strip_markup(plaintext) @col[:plaintext]=clean_searchable_text(plaintext) - book_idx=data.idx ? data.idx : '' + book_idx=book_idx_hash_to_str(data.idx) @col[:book_idx]=clean_searchable_text(book_idx) @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] @@ -353,7 +373,7 @@ module SiSU_DbImport plaintext=@col[:body].dup plaintext=strip_markup(plaintext) @col[:plaintext]=clean_searchable_text(plaintext) - book_idx=data.idx ? data.idx : '' + book_idx=book_idx_hash_to_str(data.idx) @col[:book_idx]=clean_searchable_text(book_idx) @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] @@ -383,7 +403,7 @@ module SiSU_DbImport plaintext=@col[:body].dup plaintext=strip_markup(plaintext) @col[:plaintext]=clean_searchable_text(plaintext) - book_idx=data.idx ? data.idx : '' + book_idx=book_idx_hash_to_str(data.idx) @col[:book_idx]=clean_searchable_text(book_idx) @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] @@ -441,7 +461,7 @@ module SiSU_DbImport plaintext=@col[:body].dup plaintext=strip_markup(plaintext) @col[:plaintext]=clean_searchable_text(plaintext) - book_idx=data.idx ? data.idx : '' + book_idx=book_idx_hash_to_str(data.idx) @col[:book_idx]=clean_searchable_text(book_idx) t=SiSU_DbTuple::LoadDocuments.new(@conn,@col,@opt,@file_maint) @tuple_array << t.tuple -- cgit v1.2.3