jmdict_db_schema.txt

Entry
  - ent_seq
  * k_ele
    - keb
    * ke_inf
    * ke_pri
  + r_ele
    - reb
    ? re_nokanji
    * re_restr
    * re_inf
    * re_pri
  ? info
    * links
      - link_tag
      - link_desc
      - link_uri
    * bibl
      ? bib_tag
      ? bib_txt
    * etym (UNUSED)
    * audit
      - upd_date
      - upd_detl
  + sense
    * stagk
    * stagr
    * pos
    * xref*  # not sure why it's (#PCDATA)*...
    * ant*   # not sure why it's (#PCDATA)*...
    * field
    * misc
    * s_inf
    * lsource
    * dial
    * gloss
      * pri (UNUSED)
    * example


entry:
  auto_inc (?)
  ent_seq (also unique int)... good enough?

entry < k_ele - keb
              < ke_inf
              < ke_pri

# SIMPLE translation from XML to SQL:
# Unique key for each table,
# plus indexed foreign key (except top-level element)

entity: id, ent_seq
  k_ele: id, fk, blob (<keb>)
    k_ele_inf: id, fk, entity
    k_ele_pri: id, fk, value
  r_ele: id, fk, blob, nokanji(flag)
    r_ele_restr: id, fk, value
    r_ele_inf: id, fk, entity
    r_ele_pri: id, fk, value
...etc, for all tables.

sense: id, fk

Other tables: pretty much as we planned, but add the fk in, and make
the primary key an auto-increment.  Will leave us with fewer tables
and may reduce size slightly.


---------------
Old notes below
---------------

entry  # Keep around just as a full listing...
  id  # ent_seq good enough...?

entry_k_ele
k_ele_inf
k_ele_pri
entry_r_ele
r_ele_restr
r_ele_inf
r_ele_pri
entry_links #unused
entry_bibl #unused
entry_etym #unused
entry_audit
entry_sense
sense_stagk
sense_stagr
sense_pos
sense_xref
sense_ant
sense_field
sense_misc
sense_s_inf
sense_dial
sense_example
sense_lsource
sense_gloss
gloss_pri  #unused
  (id_parent
   seq)
  id_child

k_ele  # k_ele  (id, keb => id, value)
ke_inf
ke_pri
re_restr
re_inf
re_pri
stagk
stagr
pos
xref
ant
field
misc
s_inf
dial
example
pri: UNUSED
  id
  value

r_ele
  id
  value    # reb
  nokanji  # re_nokanji (boolean)

audit
  id
  update_date
  update_details

lsource
  id
  lang (default eng)
  ls_type ("full" or "part"... full implied, part if spec'd... meaning???)
  ls_wasei (t/f flag, "y" if true, not spec'd otherwise)

gloss
  id
  lang (eng def)
  g_gend (can be null) (unused)
  value
  pri?  (t/f: if true, use pri table)

links  # UNUSED
  id
  tag
  desc
  uri

bibl  # UNUSED
  id
  tag
  txt

etym  # UNUSED


SUPPORT TABLES:

entities:
  id  # surrogate in case entity is ever changed
  entity
  value