possibility to use libzim to parse the zim dictionary

[xiaoyifang]

Actually, they explicitly changed from GPL2 to GPL2+

openzim/libzim@f57bca2

Plus this chart from gnu.org

https://www.gnu.org/licenses/gpl-faq.en.html#AllCompatibility

their previous discussion on this matter

openzim/libzim#30

Originally posted by @shenlebantongying in #27 (comment)

[xiaoyifang]

the cons and pros of using libzim,
pros:

• support super large zim dictionary >40G ,
  it is said gd consume much memory when parsing zim, the memory is come from parsing or from indexing? to be verified
  original discussion Indexing wikipedia zim - once and for all goldendict/goldendict#914
• fulltext support

cons:

• the gd has built-in fulltext support ,which means zim dictionary will use its own fulltext function ,inconsistence?
• kiwix is already out there with better zim support .

ideas and commits are welcome .

[ancientstraits]

Sure, I could contribute to this. Would zim.cc be the only file I would have to change?

[xiaoyifang]

great, AFAIK ,zim.cc is the file.

[xiaoyifang]

add some test data:
goldendict: wikipedia_zh_all_maxi_2022-01.zim 19G ,parse with gd about 20 seconds and search is ok.

the test data is not bad ,before we use libzim ,maybe we should investigate the performance first @ancientstraits