see freshly deprecated https://developers.google.com/webmasters/ajax-crawling/
Implementation improves parsing of the homepage (ajax page) which uses metatag "fragment" in header and parses supplied html snapshot instead of mostly empty ajax/scripted page.
Implementation supports also hash-bang urls (url with anchor starting with ! like ...path#!hashfragment) but our crawler filters it
(use of hash-bang is controversly discussed and proposal is deprecated, makes no sense to adjust the crawler, but as long as it is used by some sites the minor change/improvement in htmlparser is good for some time).
Quick - how does it work
- if metatag fragment with content "!" is found
- htmlparser tries to get content of htmls snapshot (using a different url)
- htmlparser returns 2 documents (original url and snapshot content - but using same original url)
- after parsing result documents are joined (and stored to index containing content also from snapshot page... as the original ajax page contains typically no parseable html content)
and allow injection of recrawl urls before queue is empty
During recrawl the balancer hangs on the very last urls often on hosts with huge delay time,
by allowing injection earlier progress is more balanced. Max number of injected crawl urls by recrawl job is 2 * max loader.
use getAttributes() to get query parameters as clear text (w/o url encoding)
use getSearchpartMap() to get in internal format (url encoded)
fix for http://mantis.tokeek.de/view.php?id=606
This is a workaround for existing index (not fully reindexed) since intro of schema with docvalues
to prevent solr exception causing recrawljob to fail with
org.apache.solr.core.SolrCore java.lang.IllegalStateException: unexpected docvalues type NONE for field 'load_date_dt' (expected=NUMERIC). Use UninvertingReader or index with docvalues.
1-char tokens and also more-than-1-char tokens, then remove the 1-char
tokens to prevent that we are to strict. This will make it possible to
be a bit more fuzzy in the search where it is appropriate.
reger