- each cache can now allocate as much memory as is available
- no more fixed limits
- replaced old performance memory monitor by new one
- added supervision methods as static functions into the classes that provide cache functionality
- steering of ram allocation is done with two simple limits that are ram availability-relative
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3434 6c8d7289-2bf4-0310-a012-ef5d649a1542
- robots.txt is a servlet now
- no need to rewrite the whole file each time a section is added or removed
- user-defined disallows, added manually, won't be overwritten anymore
- new config-setting: httpd.robots.txt, holding names of the disallowed sections
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3423 6c8d7289-2bf4-0310-a012-ef5d649a1542
- permanent cache flush is switched off. The optimized cache flush
works better if it is a large number of collections that is flushed
together
- the flush size can be configured instead the flush divisor. There is
only one size for all flushes
- collection records that shall be removed during collection transition
(jump from one collection file to another) are now not really removed
but only marked in RAM. add-operations to the collection use these
marked collection spaces
- index bulk write operations are now separated for each file of a kelondroFlex
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3414 6c8d7289-2bf4-0310-a012-ef5d649a1542
- new domains now get a chance to get crawled early
- less IO operations
- new balancing method
- better dump order at shutdown time
- bugfixes regarding not found url hashes (no more superfluous cache kill)
- domain access time is now shared over all balancer stacks
- viewing the stack does no more disturbish the balancing algorithm that much
- intelligent selection of best next domain using domain access times
- extra double-check (to double-check the double-check)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3384 6c8d7289-2bf4-0310-a012-ef5d649a1542
- not all constants may be located correctly in the switchboard. Please relocate if you know the appropriate place for them
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3303 6c8d7289-2bf4-0310-a012-ef5d649a1542
now plasmaSwitchboard.searchFromLocal calculates a searchResults structure,
which is parsed in the yacysearch/detailedSearch Servlets.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3244 6c8d7289-2bf4-0310-a012-ef5d649a1542
- redesign of data storage in plasmaSearchRankingProfile
- profiles are extended by new ranking parameters
- new RWI ranking parameters are considered during ranking
- appearance attributes (i.e. emphasised text) is now considered
- faster ranking
- some attributes that had been checked during post-ranking can now be
checked during pre-ranking phase
- removed old ranking parameter on index.html page (will be replaced by profiles in the future)
- ranking can now consider appearances of media content
- snippet-loading for media types now work correctly (fetches only from the wanted media)
- ranking-profiles can be handed over the remote peers and apply there also
- re-search of same query with different domain now also re-triggers remote search
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3105 6c8d7289-2bf4-0310-a012-ef5d649a1542
- for each crawl start, there is now a flag for text and media
- the localCrawl flag is superfluous
- added new crawl profiles
- if an image search is done, only media links are crawled for the snippets
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3100 6c8d7289-2bf4-0310-a012-ef5d649a1542
- added self-healing to database in case that wrong free-pointers exist
- added presentation of media links in snippets (does not yet work correctly)
- code cleanup
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3055 6c8d7289-2bf4-0310-a012-ef5d649a1542
- removed 'deleteComplete' flag; this was used especially for WORDS indexes
- shifted methods from plasmaSwitchboard to plasmaWordIndex
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3051 6c8d7289-2bf4-0310-a012-ef5d649a1542
- added an assortment importer. the old database structures can
be imported with
java -classpath classes yacy -migrateassortments
- modified wordmigration. The indexes from WORDS are now imported
to the collection database. The call is
java -classpath classes yacy -migratewords
(as it was)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3044 6c8d7289-2bf4-0310-a012-ef5d649a1542
to search procedure. This function can currently not used through the
search interface, but only through remote search.
added accumulation of search attributes to enable the audio, video,
image and application selection.
fixed a problem with external URL representation generation
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3036 6c8d7289-2bf4-0310-a012-ef5d649a1542
Such constraints may formulate specific restrictions to web searches
This is implemented by scraping information for constraints from a web
page during parsing, and storing flags to the pages within the web index.
In this first step, only information for index pages ("index of", directory listings)
are scraped and stored in flags
- added new flag class kelondroBitfield
- added scraper method in condenser
- added bitfield structure for all scrape types (see also condenser)
- added bitfield structure for appearance locations (see RWIEntry)
- added handover protocol for remote search and index distribution
- extended kelondroColumn class to hold bitfield types
- added another search attribute on search page (index.html)
- extended search-filter to enable filtering of non-matching constraints
- set all new database types to be default
- refactoring: moved word hash generation to condenser class
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2999 6c8d7289-2bf4-0310-a012-ef5d649a1542
- the test phase of the new collection data structure is finished
- test data that had been generated is void. There will be no migration
- the new collection files are located in DATA/INDEX/PUBLIC/TEXT/RICOLLECTION
- the index dump is void. There will be no migration
- the new index dump is in DATA/INDEX/PUBLIC/TEXT/RICACHE
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2983 6c8d7289-2bf4-0310-a012-ef5d649a1542
- moved all url and index(RWI) entries to index package
- better naming to distinguish RWI entries and URL entries
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2937 6c8d7289-2bf4-0310-a012-ef5d649a1542
- bugfix for performaceMemory
- refactoring of index ram cache: renamed indexRAMCacheRI to indexRAMRI, to make space for a cached indexRI, which should be named indexRAMCacheRI
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2925 6c8d7289-2bf4-0310-a012-ef5d649a1542
- enhanced local domain detection
- bugfixing for memory assignment in kelondroFlexSplit
- automatic memory assignment to caches according to available RAM
- bugfixes for details during search process
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2924 6c8d7289-2bf4-0310-a012-ef5d649a1542
- generalized object caching and added new object caching class
- added object caching wherever kelondroTree was used
- added object caching also to usage of kelondroFlex
- added object buffering (a write cache) to NURLs
- added many assert statements; fixed bugs here and there
- added missing close methods to latest added classes
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2858 6c8d7289-2bf4-0310-a012-ef5d649a1542
- added test migration method to migrate the old LURL to a new LURL
the new LURL will be splitted into different tables for each month
this solves several problems:
- the biggest table in YaCy is splitted in different parts and can
also be managed in filesystems that are limited to 2GB
- the oldest entries can easily be identified, used for re-crawl und
deleted
- The complete database can be limited to a specific size (as wanted many times)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2755 6c8d7289-2bf4-0310-a012-ef5d649a1542
- more synchronization
- bugfix for remove in collections
- bugfix in kelondroFlex (wrong exception condition!)
- options to use RAM, FLEX and TREE tables for Crawl URL stacker
- default for Crawl URL stacker is now FLEX (!)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2746 6c8d7289-2bf4-0310-a012-ef5d649a1542
- snippets will generate an entry in responseHeader.db
- there is now another default profile for snippet loading
- pages from snippet-loading will be indexed, indexing depth = 0
- better organization of default profiles
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2733 6c8d7289-2bf4-0310-a012-ef5d649a1542
The bugfix includes a migration function which starts automatically
after startup of yacy.
This applies only to you, if you are using the new collection index.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2711 6c8d7289-2bf4-0310-a012-ef5d649a1542
- method loadResourceContent defined as deprecated.
Please do not use this function to avoid OutOfMemory Exceptions
when loading large files
- new function getResourceContentStream to get an inputstream of a cache file
- new function getResourceContentLength to get the size of a cached file
*) httpc.java:
- Bugfix: resource content was loaded into memory even if this was not requested
*) Crawler:
- new option to hold loaded resource content in memory
- adding option to use the worker class without the worker pool
(needed by the snippet fetcher)
*) plasmaSnippetCache
- snippet loader does not use a crawl-worker from pool but uses
a newly created instance to avoid blocking by normal crawling
activity.
- now operates on streams instead of byte arrays to avoid OutOfMemory
Exceptions when operating on large files
- snippet loader now forces the crawl-worker to keep the loaded
resource in memory to avoid IO
*) plasmaCondenser: adding new function getWords that can directly operate on input streams
*) Parsers
- keep resource in memory whenever possible (to avoid IO)
- when parsing from stream the content length must be passed to the parser function now.
this length value is needed by the parsers to decide if the parsed resource content is to large
to hold it in memory and must be stored to file
- AbstractParser.java: new function to pass the contentLength of a resource to the parsers
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2701 6c8d7289-2bf4-0310-a012-ef5d649a1542
snippet loading. Some access to header-db had been twice and even
more times in some cases. Snippet resource loading fixed.
Furthermore the snippet loading during remote search within the
remote peer has been disabled, but can be switched on remotely by
new flag 'includesnippet=true'
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2688 6c8d7289-2bf4-0310-a012-ef5d649a1542
Extracted text of files that are larger than 5MB is stored in a temp file instead of keeping it in memory
*) plasmaParserDocument.java; getText now returnes an inputStream instead of a byte array
*) plasmaParserDocument.java: new function getTextBytes returns the parsed content as byte array
Attention: the caller of this function has to ensure that enough memory is available to do this
to avoid OutOfMemory Exceptions
*) httpd.java: better error handling if the soaphander is not installed
*) pdfParser.java:
- better handling of documents with exotic charsets
- better handling of large documents
- better error logging of encrypted documents
*) rtfParser.java: Bugfix for UTF-8 support
*) tarParser.java: better handling of large documents
*) zipParser.java: better handling of large documents
*) plasmaCrawlEURL.java: new errorcode for encrypted documents
*) plasmaParserDocument.java: the extracted text can now be passed
to this object as byte array or temp file
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2679 6c8d7289-2bf4-0310-a012-ef5d649a1542
- Set MaxWordCount for dhtInCache to indexDistribution.dhtReceiptLimit
so that the inCache gets flushed when the limit is passed
- Modify flushCacheSome to flush enough words to get below MaxWordCount immediately
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2649 6c8d7289-2bf4-0310-a012-ef5d649a1542
*) better logging of parser failures
*) simplified usage of plasmaparser through switchboard
*) restructuring of crawler
- crawler now returns an error message if it is used in sync mode (e.g. by snippet fetcher)
*) snippet-fetcher: more verbose error messages
*) serverByteBuffer.java: adding new function append(String,encoding)
*) serverFileUtils.java: adding functions to copy only a given number of bytes between streams
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2641 6c8d7289-2bf4-0310-a012-ef5d649a1542
- serverFileUtils.java:
-- adding methods to copy from stream to writer and readers to writers
-- moving httpc writeX methods into serverFileUtils class
- serverCharBuffer.java: removing inheritance from Writer class
- replacing htmlFilterOutputStream by htmlFilterWriter class which handles
content as char stream
- htmlFilterContentTransformer.java: deactivating getText mode
(still needs to be migrated to use char streams instead of byte streams)
- changes in several classes to use htmlFilterWriter instead of htmlFilterOutputStream
- changes in Scraper and Transformer classes to operate on chars instead of bytes
- httpdProxyHandler.java: bugfix. clientTimeout setting was missing in config file
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2617 6c8d7289-2bf4-0310-a012-ef5d649a1542
- added new image 'B' in front of search results for bookmark generation
- added news generation when a public bookmark is added
- the '+' in front of search results has new meaning: positive rating for that result
- added news generation when a '+' is hit
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2613 6c8d7289-2bf4-0310-a012-ef5d649a1542
there are now two full-controlled caches for incoming indexes:
- dhtIn
- dhtOut
during indexing, all indexes that shall not be transported to remote peers
because they belong to the own peer are stored to dhtIn. It is furthermore
ensured that received indexes are not again transmitted to other peers
directly. They may, however be transmitted later if the network grows.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2574 6c8d7289-2bf4-0310-a012-ef5d649a1542
*) next step of restructuring for new crawlers
- adding first testversion of ftp crawler class
-- does not create a htCache entry yet
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2483 6c8d7289-2bf4-0310-a012-ef5d649a1542
- adding interface class (plasma/crawler/plasmaCrawlWorker.java) for protocol specific crawl-worker threads
- moving reusable code into abstract crawl-worker class AbstractCrawlWorker.java
- the load method of the worker threads should not be called directly anymore (e.g. by the snippet fetcher)
to crawl a page and wait for the result use function plasmaCrawlLoader.loadSync([...])
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2474 6c8d7289-2bf4-0310-a012-ef5d649a1542
- catchup of OutOfMemoryError in server threads
- automatic adoption of word cache size after a Short Mem Cycle
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2426 6c8d7289-2bf4-0310-a012-ef5d649a1542
for indexing, the plasmaWordIndex.
The new data structure is ready-to-use, but currently disabled.
It can be activated by setting the static
plasmaWordIndex.useCollectionIndex
to true. This shall be done for testing purpose.
The new index is stored to
DATA/INDEX/PUBLIC/TEXT
The directory PLASMA shall be used only for crawler in the future.
Attention: during testing the data structure in INDEX may change,
and created indexes with the new data structure may get useless.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2348 6c8d7289-2bf4-0310-a012-ef5d649a1542
* replaced indexTreeMapContainer by indexRowSetContainer
* deleted indexTreeMapContainer and abstract class
This is another step to the new database structure
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2343 6c8d7289-2bf4-0310-a012-ef5d649a1542
to store the index entry. This is another step to move to the new database structure.
A side effect of this change is, that index storage uses much less RAM space,
which affects the index RAM cache.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2341 6c8d7289-2bf4-0310-a012-ef5d649a1542
indexContainers from RAM must be cloned explicitely to prevent
side-effects on stored indexContainer objects in Cache
* changed behaviour of urlReference deletion from indexContainers:
deletion does not user retrieval of all Elements from the assortments
* added textual configuration of kelondroRow and kelondroColumn definition
* update of kelondroRow usage in yacyNews
* modified kelondroAttrSeq to use modified kelondroColumn parser
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2339 6c8d7289-2bf4-0310-a012-ef5d649a1542
* store() is now called explicitely
* more urls are written to the EURL table
* the EURL stack does not store the complete entry any more, now only the URL hash
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2323 6c8d7289-2bf4-0310-a012-ef5d649a1542
This shall be seen as an experiment to exclude all cases where
there could be a DNS lookup during URL comparisment.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2290 6c8d7289-2bf4-0310-a012-ef5d649a1542
this was done because testing showed that cache-delete operations
slowed down record access most, even more that actual IO operations.
Cache-delete operations appeared when entries were shifted from low-priority
positions to high-priority positions. During a fill of x entries to a database,
x/2 delete situation happen which caused two or more delete operations.
removing the cache control means that these delete operations are not
necessary any more, but it is more difficult to decide which cache elements
shall be removed in case that the cache is full. There is not yet a stable
solution for this case, but the advantage of a faster cache is more important
that the flush problem.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2244 6c8d7289-2bf4-0310-a012-ef5d649a1542
- needed by Ismael to receive the Steering page properly on shutdown
- now the steering page should always be displayed properly in the web browser
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2129 6c8d7289-2bf4-0310-a012-ef5d649a1542
- Improve performance of plasmaURL.exists() by remembering URL-hashes that are not present
- Use a more realistic estimation of memory usage by the existsIndex cache
- Routine cleanup of the existsIndex to limit its memory usage
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2113 6c8d7289-2bf4-0310-a012-ef5d649a1542
- re-crawl by age of page (enter in minutes)
- auto-domain-filter
- maximum number of pages per domain
NOT YET TESTED!
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1949 6c8d7289-2bf4-0310-a012-ef5d649a1542
the old search page is obsolete and will be removed
* ConfigBasic.html is now the default page instead of index.html
as long as no password is set
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1815 6c8d7289-2bf4-0310-a012-ef5d649a1542
* fixed graphics problem
* fixed some other problems with default values
* 1-2-3 config now appears automatically on start-up if no password is set
* added new config menu
* moved profile to new config menu
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1792 6c8d7289-2bf4-0310-a012-ef5d649a1542
* added ranking coefficient transmission to remote peer (without evaluation on server side, will be added later)
* changed ranking coefficients slightly
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1770 6c8d7289-2bf4-0310-a012-ef5d649a1542
find the same words as before. This was caused because indexContaines stored the url references
with a hashtable. A tree was needed to work with the index conjunction-by-numeration
- added permanent ram cache flush (again)
- removed direct flush of ram cache after a large container is added.
this happens especially during DHT transmission and therefore this fix should
speed up DHT transmission on server side.
- removed unused and out-dated methods
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1765 6c8d7289-2bf4-0310-a012-ef5d649a1542
The cache will fill up and flushed explicitely when it is full.
This shall remove double-access of assortments (indexing and flush)
during indexing process. Hopefully this should reduce IO.
The main idea is: the cache shall mainly be flushed by DHT transfer, and
only indexes that shall be hosted by the own peer are flushed to the
assortments. This needs further work.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1617 6c8d7289-2bf4-0310-a012-ef5d649a1542
- restructuring to allow different import tasks to be controlled via one gui
- adding possibility to import a single assortment file
- adding possibility to set the cache size that should be used
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1504 6c8d7289-2bf4-0310-a012-ef5d649a1542
- changed index collection process: indexes are not first flushed to indexEntity,
but now collected directly from ram cache and assortments
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1489 6c8d7289-2bf4-0310-a012-ef5d649a1542
- replaced usage of temporary IndexEntity by EntryContainer
- added more attributes to word index
- added exact-string search (using quotes in query)
- disabled writing into WORDS during search; EntryContainers are used instead
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1485 6c8d7289-2bf4-0310-a012-ef5d649a1542
- implemented hand-over of new word index attributes during remote search
- implemented word-distance computation during search
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1382 6c8d7289-2bf4-0310-a012-ef5d649a1542
the kelondro database needs specific information about the order of
base64-encoded keys. Since no other package depends on base64
(only the httpd uses base64 for encryption, but does not need to encode these strings)
it is good to move base64 encoding to the new ordering classes in kelondro.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1284 6c8d7289-2bf4-0310-a012-ef5d649a1542
