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

1 files changed, 18 insertions, 14 deletions

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