this makes YaCy search results VERY fast for all verify=false search cases
and it enhances the search speed also for all other snippet-fetch cases.
With this change my peer performed 100 Queries Per Second (!!!) while doing 10 queries simultanously (!!!)
in an intranet index of 20000 URLs on my 16-core Mac
Check this yourself by doing:
cd bin
./searchtestmulti.sh
after finishing the run, divide 1000 by the given time per query (which is the qps for one thread)
and then multiply again by 10 (because 10 search threads has been started)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7231 6c8d7289-2bf4-0310-a012-ef5d649a1542
*) more beautyful and easier to understand code (IMO)
*) added display= parameter to a lot of links in Wiki.html
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7226 6c8d7289-2bf4-0310-a012-ef5d649a1542
- better crawl star for files paths and smb paths
- added time-out wrapper for dns resolving and reverse resolving to prevent blockings
- fixed intranet scanner result list check boxes
- prevented htcache usage in case of file and smb crawling (not necessary, documents are locally available)
- fixed rss feed loader
- fixes sitemap loader which had not been restricted to single files (crawl-depth must be zero)
- clearing of crawl result lists when a network switch was done
- higher maximum file size for crawler
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7214 6c8d7289-2bf4-0310-a012-ef5d649a1542
- nobody understand the auto-dom filter without a lenghtly introduction about the function of a crawler
- nobody ever used the auto-dom filter other than with a crawl depth of 1
- the auto-dom filter was buggy since the filter did not survive a restart and then a search index contained waste
- the function of the auto-dom filter was in fact to just load a link list from the given start url and then start separate crawls for all these urls restricted by their domain
- the new Site Link-List option shows the target urls in real-time during input of the start url (like the robots check) and gives a transparent feed-back what it does before it can be used
- the new option also fits into the easy site-crawl start menu
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7213 6c8d7289-2bf4-0310-a012-ef5d649a1542
- migrated the 'yacy' user agent to 'yacybot' in many client methods since the 'yacy' user agent is only used for the proxy
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7199 6c8d7289-2bf4-0310-a012-ef5d649a1542
- better control of number of running search threads
- no time-out waiting time when no ranking feeding takes place
- local search queries by a remote peer may be faster up to 300 milliseconds
- a local search may even be faster
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7176 6c8d7289-2bf4-0310-a012-ef5d649a1542
- remove the library
- added two classes from the httpclient-3.1 library as source code to YaCy because these classes were used by the YaCy HTTP Server
- modified the added classes ChunkedInputStream and ContentLengthInputStream in such a way that:
* there are no more dependencies to httpclient-3.1
* these classes had been simplified to serve only the purpose for the YaCy httpd
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7171 6c8d7289-2bf4-0310-a012-ef5d649a1542
- not so much abstraction for a collection, makes use of remove() (no operands) possible
- different way to delete elements in track (destructive, not constructive (less copies of elements in new queue))
- more abstraction for class api since no static class must be used any more
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7169 6c8d7289-2bf4-0310-a012-ef5d649a1542
- 300 ppm is default now (but this is switched off by default; if you switch it on you may want more traffic?)
- better timing for busy queue
- better amount of remote url retrieval
- better time-out values
- better tracking of availability of remote crawl urls
- more logging for result of receipt sending
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7159 6c8d7289-2bf4-0310-a012-ef5d649a1542
- added a remote crawl menu item to the index create menu. This menu also shows a list of peers that provide remote crawl urls
- set remote crawl option by default to off. This option may be important but it also confuses first-time users
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7158 6c8d7289-2bf4-0310-a012-ef5d649a1542
- added use of LookAheadIterator which should prevent mistakes when coding iterators with embedded iterators
- added a fail-safe reaction in case of database corruption using iterators over database elements (no interruption then)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7154 6c8d7289-2bf4-0310-a012-ef5d649a1542
- changed data structure for sorting in search which performs better in that specific case (too many updates)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7150 6c8d7289-2bf4-0310-a012-ef5d649a1542
- fixed bug in WeakPriorityBlockingQueue
- show favicon faster using pre-loading (same technique as used for fast image search)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7130 6c8d7289-2bf4-0310-a012-ef5d649a1542
created a WeakPriorityBlockingQueue as special implementation
of a PriorityBlockingQueue with a weak object binding.
- better abstraction of ordering technique
- fixed some bugs according to result numbering (distinguish different counters in Queue)
- fixed a ordering bug in post-ranking (ordering was decreased instead of increased)
- reversed ordering numbering using a reversed ordering. The higher the ranking number the better (now).
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7128 6c8d7289-2bf4-0310-a012-ef5d649a1542
- removed old parser
- removed old importer framework (was only used by removed old parser)
- added a new sitemap parser in parser framework
- linked new parser with parser access in old sitemap processing routines
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7126 6c8d7289-2bf4-0310-a012-ef5d649a1542
- added some clear statements that shall clear static cache size within the pdfbox library
- the pdfbox library contains a memory leak; it is unsafe to run a peer with pdf parser permanently on.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7120 6c8d7289-2bf4-0310-a012-ef5d649a1542
with this fix search results are (even) faster.
Also enhanced: image search. This is now speeded up using a image search result look-ahead
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7105 6c8d7289-2bf4-0310-a012-ef5d649a1542
- storing the startup properties when yacy is started
- using the properties in the restart-script again. this transports also the DATA directory location as parameter of the -gui option that is used when the Mac version of YaCy is started
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7102 6c8d7289-2bf4-0310-a012-ef5d649a1542
- the search request cache can now get as large as 1000 entries
- if more entries arrive, unused are deleted
- the elements may stay in the cache up to 10 minutes and longer if they are used
- the elements are deleted earlier that 10 minutes if the memory gets low
This commit was mainly done for metager-feeding peers that have a query load of 50000 queries each day. Also added:
- a monitor for cache hit/cache miss in PerformanceMemory_p.html (see at bottom of page)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7093 6c8d7289-2bf4-0310-a012-ef5d649a1542
the home path can now be distinguished between
- data home; the path where the DATA directory is created
- application home; everything else
This will make it possible to store application data on Mac releases within the
~/Library/YaCy
directory; a place where Mac applications write their data.
Similar techniques will be possible for debian and windows.
To use the new data path, YaCy can be started with
-start <data path>
or
-gui <data path>
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7092 6c8d7289-2bf4-0310-a012-ef5d649a1542
- permanent storage of auto-dom statistics in profile
- storage of profiles in WorkTable data structure
not finished yet. No functional change yet.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7088 6c8d7289-2bf4-0310-a012-ef5d649a1542
- added a 'remove all' button for new and scheduled rss feed list
- made adding of new rss feeds concurrent so interface is more responsible
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7078 6c8d7289-2bf4-0310-a012-ef5d649a1542
- next-execution-time in scheduler
- deletion of scheduled rss feed loading (now deletes also the scheduling entry)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7075 6c8d7289-2bf4-0310-a012-ef5d649a1542
- fresh recorded rss feeds (not yet loaded or in scheduler)
- rss feeds in scheduler
The first list has a button that can be used to place rss feeds into the scheduler
The second list has a button to delete rss feeds from the scheduler
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7074 6c8d7289-2bf4-0310-a012-ef5d649a1542
the BEncodedHeap now implements Map<byte[], Map<String, byte[]>>
This will make it possible that also different database storage types may be added that implement also the same Map<byte[], Map<String, byte[]>> interface.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7070 6c8d7289-2bf4-0310-a012-ef5d649a1542
- move the header framework classes to cora
- move the ARC caching classes to cora
- refactoring of code to call these classes from cora
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7068 6c8d7289-2bf4-0310-a012-ef5d649a1542
moved http date methods from DateFormatter to HeaderFramework
changed logging to log4j
- added ftp load access to MultiProtocolURI
- ensured termination of RSS feed iteration
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7067 6c8d7289-2bf4-0310-a012-ef5d649a1542
- the scheduler extends the option for re-crawl timing. Many people misunderstood the re-crawl timing feature because that was just a criteria for the url double-check and not a scheduler. Now the scheduler setting is combined with the re-crawl setting and people will have the choice between no re-crawl, re-crawl as was possible so far and a scheduled re-crawl. The 'classic' re-crawl time is set automatically when the scheduling function is selected
- removed the bookmark-based scheduler. This scheduler was not able to transport all attributes of a crawl start and did therefore not support special crawling starts i.e. for forums and wikis
- since the old scheduler was not aber to crawl special forums and wikis, the must-not-match filter was statically fixed to all bad pages for these special use cases. Since the new scheduler can handle these filters, it is possible to remove the default settings for the filters
- removed the busy thread that was used to trigger the bookmark-based scheduler
- removed the crontab for the bookmark-based scheduler
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7051 6c8d7289-2bf4-0310-a012-ef5d649a1542
- every process that is monitored with the API Steering interface can now be scheduled!
- added input methods in Steering interface to set a scheduling time
- added a view on the steering api that shows only crawl jobs inside the Crawl Profile servlet
- added a scheduling call process in the cleanup process handler that triggers the scheduled processes
This causes that the cleanup now also looks for scheduled processes. Such processes are therefore not executed at
the same time as given in the target execution time but they will be executed within the cleanup process time window.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7050 6c8d7289-2bf4-0310-a012-ef5d649a1542
- added recording date, last execution date and next execution date for a scheduler (scheduler to be implemented next)
- extended database access methods for more data formats, especially for date insert/retrieval
- extended 'Steering' interface to show new database fields
- migrated Steering to new http client
- extended cora http client to transmit authentication and also added some convenience methods (http response code)
- simplified database back-end (not so much specialized methods for multiple properties)
- extended date formatter to produce a special format to show dates in html ( in spaces of date format)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7049 6c8d7289-2bf4-0310-a012-ef5d649a1542
- added new network definition 'allip' which can be used in networks where intranet and internet-addresses shall be indexed
- added a auto-switch-off for global search if there are no global peers
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7030 6c8d7289-2bf4-0310-a012-ef5d649a1542
The gui currently does nothing else than providing a search window that sends the search string to the browser
The gui is started when YaCy is started with the option -g or --gui, like
./startYACY.sh -g
The gui will primary be used to provide a 'real' macintosh version that can be started and operated like any other macintosh application. A special mac application wrapper will follow.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7021 6c8d7289-2bf4-0310-a012-ef5d649a1542
- confirm that database objects that are not supposed to grow do not have a index memory management that is designed for growth
- changed index sorting method in such a way that it allocates less objects during quicksort
- database classes classes renaming (shorter, naming addresses that objects hold in RAM)
- added a large number of asserts to check if objects actually take the RAM that they should have
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7019 6c8d7289-2bf4-0310-a012-ef5d649a1542
- added statistics about database index caches in PerformanceMemory_p.html
- adoped many classes to use the new statistics
- added missing close statements
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7018 6c8d7289-2bf4-0310-a012-ef5d649a1542
de.lng: Updated German translation for additional String in ConfigUpdate_p.html
XHTML 1.0 Strict fixes for all the other .html files
yacy/ui/css/yacyui-portalsearch.css: added .hidden class that was removed from ConfigProperties_p.html
Switchboard.java: Added URL for thread Remote Crawl Job and set URL for Remote Crawl URL Loader to null to fix empty href=""
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6996 6c8d7289-2bf4-0310-a012-ef5d649a1542
monitoring: replaced unused 'idletime' by uploading bytes
added some kind of 'upload-throttling' at dht-out :-)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6983 6c8d7289-2bf4-0310-a012-ef5d649a1542
some file types are containers for several files. These containers had been parsed in such a way that the set of resulting parsed content was merged into one single document before parsing. Using this parser infrastructure it is not possible to parse document containers that contain individual files. An example is a rss file where the rss messages can be treated as individual documents with their own url reference. Another example is a surrogate file which was treated with a special operation outside of the parser infrastructure.
This commit introduces a redesigned parser interface and a new abstract parser implementation. The new parser interface has now only one entry point and returns always a set of parsed documents. In case of single documents the parser method returns a set of one documents.
To be compliant with the new interface, the zip and tar parser had been also completely redesigned. All parsers are now much more simple and cleaner in its structure. The switchboard operations had been extended to operate with sets of parsed files, not single parsed files.
additionally, parsing of jar manifest files had been added.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6955 6c8d7289-2bf4-0310-a012-ef5d649a1542
see http://www.gnu.org/licenses/license-list.html for explanation
Since (as far as I know) nobody else has ever contributed to these files I may be allowed to just apply an older license.
You may consider this as a dual-licensing and may use and optionally replicate the older files under GPL 3.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6952 6c8d7289-2bf4-0310-a012-ef5d649a1542
- a site-operation heuristic that loads all direct links from a portal page if the site-operator is used
- a direct crawl for search results from scroogle for the given search terms
The configuration page can be found directly beside the network configuration page
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6951 6c8d7289-2bf4-0310-a012-ef5d649a1542
- found and fixed a possible memory leak in YaCy internal RSS feed system
- some refactoring in RSS feed mechanisms to make this possible
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6950 6c8d7289-2bf4-0310-a012-ef5d649a1542
lets try that. If we run into a memory problem because of too many 2-letter-words, then we must introduce whitelists for 2-letter words.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6947 6c8d7289-2bf4-0310-a012-ef5d649a1542
- cleaned uo the code. The new eclipse helios provided new warnings for dead code. This change cleans up most of these warnings
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6945 6c8d7289-2bf4-0310-a012-ef5d649a1542
Here a new concept called 'search heuristics' is introduced. A heuristic is a kind of 'shortcut' to good results in IT, here for good search results. In this case it will be used to get a very transparent way to compare what YaCy is able to produce as search result and what g**gle produces as search result. Here is what your can do now:
- add the phrase 'heuristic:scroogle' to your search query, like 'oil spill heuristic:scroogle' and then a call to scroogle is made to get anonymous search results from g**gle.
- these results are _not_ taken as meta-search results, but are used to instantly feed a crawling and indexing process. This happens very fast, here 20 results from scroogle are taken and loaded all simultanously, parsed and indexed immediately and from the results of the parsed content the search result is feeded, along to the normal p2p search
- when new results from that heuristic (more to come) get part of the search results, then it is verified if such results are redundant to existing (they had been part of the normal YaCy search result anyway) or if they had been completely new to YaCy.
- in the search results the new search results from heuristics are marked with a 'H ++' and search results from heuristics that had been already found by YaCy are marked with a 'H ='. That means:
- you can now see YaCy and Scroogle search results in one result page but you also see that you would not have 'missed' the g**gle results when you would only have used YaCy.
- to make it short: YaCy now subsumes g**gle results. If you use only YaCy, you miss nothing.
to come: a configuration page that let you configure the usage of heuristics and get this feature by default.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6944 6c8d7289-2bf4-0310-a012-ef5d649a1542
implemented a hint from dulcedo "use site: - operator as crawl start point".
YaCy already was able to search using a site-constraint. This function is now extended with a instant crawling feature.
When you now use the site-operator, then the landing page of the site iand every page that is linked from this page are loaded, indexed and selected for the search result within that search request. When the remote server responds quickly enough, then this process can result in search results during the normal search result preparation .. just in some seconds.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6941 6c8d7289-2bf4-0310-a012-ef5d649a1542
a cache access shall not made directly to the cache any more, all loading attempts shall use the LoaderDispatcher.
To control the usage of the cache, a enum instance from CrawlProfile.CacheStrategy shall be used.
Some direct loading methods without the usage of a cache strategy have been removed. This affects also the verify-option
of the yacysearch servlet. If there is a 'verify=false' now after this commit this does not necessarily mean that no snippets
are generated. Instead, all snippets that can be retrieved using the cache only are presented. This still means that the search hit was not verified because the snippet was generated using the cache. If a cache-based generation of snippets is not possible, then the verify=false causes that the link is not rejected.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6936 6c8d7289-2bf4-0310-a012-ef5d649a1542
- removed concurrency overhead for small number of index normalizations as it happens during remote search
- removed 'load only parseable' constraint for snippet fetch because some resources may not have any url file extension and these had therefore not been parseable and searcheable since they may become parseable after loading when their mime type is known
- this partly fixes some problems with http://forum.yacy-websuche.de/viewtopic.php?p=20300#p20300 but more changes are necessary to get all expected search results
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6926 6c8d7289-2bf4-0310-a012-ef5d649a1542
- a new news db will be created (news1024.db), the old one (news.db) can be deleted
- peers with too large news payload are not ignored any more (they may have been invisible because they had a too large news payload!)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6917 6c8d7289-2bf4-0310-a012-ef5d649a1542
- crawl lists for each domain now uses a HandleSet which should use less memory than LinkedLists
- but: fill more entries into the domain lists (all available entries)
- fixes to selection criteria (best domain selection)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6909 6c8d7289-2bf4-0310-a012-ef5d649a1542
