diff options
Diffstat (limited to 'lib/sisu/v3dv/db_sqltxt.rb')
| -rw-r--r-- | lib/sisu/v3dv/db_sqltxt.rb | 60 |
1 files changed, 29 insertions, 31 deletions
diff --git a/lib/sisu/v3dv/db_sqltxt.rb b/lib/sisu/v3dv/db_sqltxt.rb index 49dcf10a..e65a8521 100644 --- a/lib/sisu/v3dv/db_sqltxt.rb +++ b/lib/sisu/v3dv/db_sqltxt.rb @@ -59,46 +59,45 @@ module SiSU_DbText class Prepare def special_character_escape(str) - str.gsub!(/'/,"''") #string.gsub!(/'/,"\047") #string.gsub!(/'/,"\\'") - str.gsub!(/(\\)/m,'\1\1') #ok but with warnings, double backslash on sqlite #str.gsub!(/[\\]/m,'\\x5C') #ok but with warnings, but not for sqlite #str.gsub!(/(\\)/m,'\1') #ok for sqlite not for pgsql - str.gsub!(/#{Mx[:br_line]}|#{Mx[:br_nl]}/,"<br />\n") - str.gsub!(/#{Mx[:tag_o]}\S+?#{Mx[:tag_c]}/,'') #check - str.gsub!(/#{Mx[:lnk_o]}\s*(\S+?\.(?:png|jpg))(?:\s+\d+x\d+)?(.+?)#{Mx[:lnk_c]}\S+/,'[image: \1] \2') - str.gsub!(/#{Mx[:lnk_o]}\s*(.+?)\s*#{Mx[:lnk_c]}(?:file|ftp):\/\/\S+?([.,!?]?(?:\s|$))/,'\1\2') - str.gsub!(/#{Mx[:lnk_o]}\s*(.+?)\s*#{Mx[:lnk_c]}#{Mx[:url_o]}\S+?#{Mx[:url_c]}/,'\1') - str + str=str.gsub(/'/,"''"). #string.gsub!(/'/,"\047") #string.gsub!(/'/,"\\'") + gsub(/(\\)/m,'\1\1'). #ok but with warnings, double backslash on sqlite #str.gsub!(/[\\]/m,'\\x5C') #ok but with warnings, but not for sqlite #str.gsub!(/(\\)/m,'\1') #ok for sqlite not for pgsql + gsub(/#{Mx[:br_line]}|#{Mx[:br_nl]}/,"<br />\n"). + gsub(/#{Mx[:tag_o]}\S+?#{Mx[:tag_c]}/,''). #check + gsub(/#{Mx[:lnk_o]}\s*(\S+?\.(?:png|jpg))(?:\s+\d+x\d+)?(.+?)#{Mx[:lnk_c]}\S+/,'[image: \1] \2'). + gsub(/#{Mx[:lnk_o]}\s*(.+?)\s*#{Mx[:lnk_c]}(?:file|ftp):\/\/\S+?([.,!?]?(?:\s|$))/,'\1\2'). + gsub(/#{Mx[:lnk_o]}\s*(.+?)\s*#{Mx[:lnk_c]}#{Mx[:url_o]}\S+?#{Mx[:url_c]}/,'\1') 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/m,'\2') - s.gsub!(/^(?:block|group|poem|code)\{/m,''); s.gsub!(/^\}(?:block|group|poem|code)/m,'') - s.gsub!(/\A(?:@\S+:\s+.+)\Z/m,'') + s=s.gsub(/([*\/_-])\{(.+?)\}\1/m,'\2'). + gsub(/^(?:block|group|poem|code)\{/m,'').gsub(/^\}(?:block|group|poem|code)/m,''). + gsub(/\A(?:@\S+:\s+.+)\Z/m,'') if s =~/^:A~/ if defined? @md.creator \ and defined? @md.creator.author \ and not @md.creator.author.empty? - s.gsub!(/@author/,@md.creator.author) + s=s.gsub(/@author/,@md.creator.author) else SiSU_Screen::Ansi.new('v','WARNING Document Author information missing; provide @creator: :author:',@md.fnb).warn unless @md.opt.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) + s=s.gsub(/@title/,@md.title.full) else SiSU_Screen::Ansi.new('v','WARNING Document Title missing; provide @title:',@md.fnb).warn unless @md.opt.cmd.inspect =~/q/ end end - 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!(/<br>/m,' ') + s=s.gsub(/^(?:_[1-9]\*?|_\*)\s+/m,''). + gsub(/^(?:[1-9]\~(\S+)?)\s+/m,''). + gsub(/^(?::?[A-C]\~(\S+)?)\s+/m,''). + gsub(/^%{1,3} .+/m,''). #removed even if contained in code block + gsub(/<br>/m,' ') en << s.scan(/~\{\s*(.+?)\s*\}~/m) - s.gsub!(/~\{.+?\}~/m,'') - s.gsub!(/ \s+/m,' ') + s=s.gsub(/~\{.+?\}~/m,''). + gsub(/ \s+/m,' ') #special_character_escape(s) s end @@ -109,17 +108,16 @@ module SiSU_DbText txt end def strip_markup(str) #define rules, make same as in dal clean - str.gsub!(/#{Mx[:fa_superscript_o]}(\d+)#{Mx[:fa_superscript_c]}/,'[\1]') - str.gsub!(/(?: \\;|#{Mx[:nbsp]})+/,' ') - str.gsub!(/#{Mx[:tc_o]}#{Mx[:tc_p]}#{Mx[:tc_p]}\d+(.+)#{Mx[:tc_c]}/u,'\1') #tables - str.gsub!(/#{Mx[:tc_p]}#{Mx[:tc_p]}\d+#{Mx[:tc_p]}/u,' ') #tables - str.gsub!(/#{Mx[:tc_p]}/u,' ') #tables tidy later - str.gsub!(/<.+?>/,'') - str.gsub!(/#{Mx[:lnk_o]}.+?\.(?:png|jpg|gif).+?#{Mx[:lnk_c]}(?:file|ftp)\/\/:\S+ /,' [image] ') # else image names found in search - str.gsub!(/#{Mx[:lnk_o]}.+?\.(?:png|jpg|gif).+?#{Mx[:lnk_c]}#{Mx[:url_o]}\S+?#{Mx[:url_c]}/,' [image]') # else image names found in search - str.gsub!(/\s\s+/,' ') - str.strip! - str + str=str.gsub(/#{Mx[:fa_superscript_o]}(\d+)#{Mx[:fa_superscript_c]}/,'[\1]'). + gsub(/(?: \\;|#{Mx[:nbsp]})+/,' '). + gsub(/#{Mx[:tc_o]}#{Mx[:tc_p]}#{Mx[:tc_p]}\d+(.+)#{Mx[:tc_c]}/u,'\1'). #tables + gsub(/#{Mx[:tc_p]}#{Mx[:tc_p]}\d+#{Mx[:tc_p]}/u,' '). #tables + gsub(/#{Mx[:tc_p]}/u,' '). #tables tidy later + gsub(/<.+?>/,''). + gsub(/#{Mx[:lnk_o]}.+?\.(?:png|jpg|gif).+?#{Mx[:lnk_c]}(?:file|ftp)\/\/:\S+ /,' [image] '). # else image names found in search + gsub(/#{Mx[:lnk_o]}.+?\.(?:png|jpg|gif).+?#{Mx[:lnk_c]}#{Mx[:url_o]}\S+?#{Mx[:url_c]}/,' [image]'). # else image names found in search + gsub(/\s\s+/,' '). + strip end def unique_words(str) a=str.scan(/[a-zA-Z0-9\\\/_-]{2,}/) #a=str.scan(/\S+{2,}/) |