jmdict_plan.txt

JMDict structure
================

JMdict
  entry
    ent_seq (int)  # Just use orig code; don't add autonumber
    k_ele*
      keb (text)
      ke_inf* (info field: ENTITY: &ateji; &iK; &ik; &io; &oK;)
      ke_pri* (priority: TEXT (or could make a table and ref by int to save space...))
    r_ele+
      reb (text)
      re_nokanji? (empty tag: boolean t/f)
      re_restr* (text, matches one "keb" string)
      re_inf* (info: ENTITY: &gikun; &ik; &ok; &uK;)
      re_pri* (see ke_pri)
    info?
      links*   # UNUSED
        link_tag (???)   # UNUSED
        link_desc (text) # UNUSED
        link_uri (text)  # UNUSED
      bibl*    # UNUSED
        bib_tag? (???)   # UNUSED
        bib_txt? (text)  # UNUSED
      etym* (???)   # UNUSED
      audit*
        upd_date (text/date, always YYYY/MM/DD format datestamp)
        upd_detl (text, "Entry amended" (x1818) or "Entry created" (x51861),
                  could have others...)
    sense+
      stagk* (text, matches keb) ("sense tag k")
      stagr* (text, matches reb)
      pos* (text or int ref to entity code)  # check ElementTree parsing of entities
      xref* (text w/ special format - for now, just store)   # What does (#PCDATA)* mean???
      ant* (text, matches keb/reb)
      field* (text or int ref to entity code)
      misc* (text or int ref to entity code)
      s_inf* (text)
      lsource* (xml:lang TEXT (def "eng"),
                ls_type (implied "full", "part" when spec'd (UNUSED) -> constant tbl?  t/f?),
                ls_wasei (t/f flag...?) (always "y" if spec'd))
               (text or empty str (why???))
      dial* (entity code)
      gloss* (xml:lang TEXT (def "eng"), g_gend (abs irrel, spec'd gives gender (UNUSED)))
             (TEXT or pri element)
        pri* (???)  # UNUSED
      example* (text)  # UNUSED


NOTES:
- ElementTree does auto-convert entities.
- It's possible to use ElementTree.parse(source_file, parser) to supply parser instance...
  maybe this could used to track entities.