main time-consuming process. This shall be done in concurrency.
- added concurrent processes to call the PixelGrabber and framework to
do that (queues)
It is now possible to create 4k-Images (3840x2160) i.e. with the Network
Graphics servlet
compressed deflater output, then assign an exact-sized byte[] which
makes resizing afterwards superfluous
- after all enhancements all class objects were removed; result is just
one short static method
- made objects final where possible
quite a bit to remove all code that handles transparency. With this
highly specialized png writer it is possible to write png images much
faster that with the JRE built-in png writer.
In a second step it can be possible to add concurrency to increase
computation speed further.
- multiple hosts can be listed (comma-separated) as host argument
- new 'bf'-attribut (branch factor): the maximum number of edges per
node
- the bf-value is computed automatically
- ordering of nodes when the graphic is drawed: mostly the drawing ends
with an limitation eg. number of nodes. When this happens, it should be
ensured that more 'interesting' nodes are painted in advance. This is
now done by sorting all nodes by the number of links they have in de
distant sub-graph.
superfluous. The target is to make a solr document as the core of YaCy
documents which would cause that many conversions can be removed. On the
way to this target the Equivalence of URIMetadataRow and URIMetadataNode
had to be removed to expose the usage of the old URIMetadataRow data
structure.
This refactoring already removes unneccessary conversions and should
make memory usage during indexing lower.
which had been then mixed with remote RWIs. Now these Solr documents are
feeded into the result set as they appear during local and remote
search. That makes the search much faster.
URIMetadataNode which creates the opportunity to access Solr objects
directly and use their information richness
- lazy initialization of the URIMetadataNode object - should cause less
computation and memory usage during search.
- removed dead code
- showing also outbound links to other domains if there are any
- the outbound links browser shows also the link structure image
- showing even inbound links if the web structure graph has information
about that
- removed the left menu and made the HostBrowser a part of the top menu
for search
- moved the file search also to the top menu
- added hover information in the HostBrowser to explain what the click
means
- because the HostBrowser also links to the Metadata viewer ViewFile,
there should be a button to switch back to the HostBrowser: added that
also.
MultiProtocolURI during normalform computation because that should
always be done and also be done during initialization of the
MultiProtocolURI Object. The new normalform method takes only one
argument which should be 'true' unless you know exactly what you are
doing.
there is no need for the RWI index if the index is not transfered to
another peer. Therefore the creation of RWI index data is now suppressed
if DHT is disabled. This applies for all intranet and portal mode
configurations, but not for public robinson modes. A robinson may switch
back to public mode and then transmit its data. That means if someone
wants to switch never to DHT mode, it would be more appropriate to
choose the portal mode.
on the jaudiotagger library. The parser is disabled by default as it
needs to store temporary files for non file:// protocols, which might be
disliked. For your local MP3-collection it loads nicely Artist,
Title, Album etc. from the audio files meta data.
This can be used to add another stemming to solr using stemming files
that are expressed as synonyms for grammatical alternatives. The
synonym/stemming files must have the following form:
- each line is a comma-separated list of synonyms
- the list of synonyms may be enclosed with {} (like the GSA synonyms
file)
- the file may contain comments which are lines starting with a '#'
The synonym file(s) must be placed in DATA/DICTIONARIES/synonyms/ and
are activated by default whenever a synonym file is in place.
Then, for each word that is found in a document all synonyms are added
to a long text field which is stored into synonyms_t. Processes using
the synonyms must query with that field as optional matcher.
this new search interface is something completely new for search, but
completely common on desktops: browser a web space like one would browse
a file system in a file browser. The file listing is created using the
search index and a faceted restriction to specific domains.
solr and the gsa search interface. That means that all JavaScript in
browsers now can Cross-Origin access all YaCy search interfaces, which
opens the option of 'YaCy Client in Browser' and 'End-Point Fail-over'
concepts.
Therefore a property like solrConnected() must be true all the time.
Removal of this method causes removal of all write operations to the old
metadata index.
create an document entry. This makes remote search much slower.
- removed synchronization of add method if ip_s is activated to prevent
that a user configuration causes bad behavior. The disadvantage of that
is, that a index dump can cause data loss if an indexing is running
during index dump
- catched more exceptions and more NPE
- better abstraction in MirrorSolrConnector
- slight performance enhancement when only the index count is requested
(rows=0 is sufficient to get a total count)
dump commands
- adjusted the apicall.sh script to get the downloaded text as output to
stdout which is necessary to parse the content out of it
- added indexdump.sh script which creates a solr dump and prints out the
storage path for the index dump
- added synchronization to the Fulltext class to prevent that data is
stored to a non-existing solr index while this index is disabled during
the storage of the dump
structures in cora:
- added the Peer object, which is a fresh version of Seed
- added the Peers object, which is a fresh version of Network
- added the Network api access class to retrieve a list of peers based
on the Network.xml servlet in all YaCy peers.
- indexUrlMustMatch and indexUrlMustNotMatch which can be used to select
loaded pages for indexing. Default patterns are in such a way that all
loaded pages are also indexed (as before) but when doing an expert crawl
start, then the user may select only specific urls to be indexed.
- crawlerNoDepthLimitMatch is a new pattern that can be used to remove
the crawl depth limitation. This filter a never-match by default (which
causes that the depth is used) but the user can select paths which will
be loaded completely even if a crawl depth is reached.
- added indexing to Tables columns to support larger bookmark
collections
- added RDF output (HTTP) for public bookmarks at /YMarks.rdf
- YMarkRDF also provides a Jena RDF Model as "internal" API
- various other changes/fixes for YMarks (mainly backend)
- a solr field collection_sxt can be used to store a set of crawl tags
- when this field is activated, a crawl tag can be assigned when crawls
are started
- the content of the collection field can be comma-separated, all of
them are assigned to the documents when they are indexed as result of
such a crawl start
- a search result can be drilled down to a specific collection; this is
currently only available in the solr interface and also in the gsa
interface using the 'site' option
- this adds a mandatory field for gsa queries (the google api demands
that field all the time)
- reduced danger that a non-existing RWI database causes NPEs
- added Solr queries to did-you-mean: this makes it possible that our
did-you-mean algorithm works together with only Solr and without RWIs
title_count_i, title_chars_val, title_words_val
description_count_i, description_chars_val, description_words_val
- added many asserts to ensure data type correctness from YaCy to Solr
and vice versa
- made many fixes according to new findings from these asserts (!)
- fixed type definition found by the verifier
- added multivalue-string fields for solr with extension 'sxt'
- added multivalue-integer fields for solr with extension 'val'
- renamed some solr attributes from txt to sxt
- changed solr query line to an explicit AND/OR structure
- added a country code second level domain list to Domains class; with
parser
- added a host string parser to get domain class name, country-code
second-level domain and subdomain out of it
- removed old coordinate attributes
if adminAccountForLocalhost=true leave automatic password unchanged so access from local host is granted but remote access is preventet from the 1st second.
Previously Yacy could read file listings in UTF-8, but couldn't send commands to the FTP server in UTF-8 (the second byte of every multi-byte character was ignored), which caused a lot of errors on the server side.
Now it handles UTF-8 correctly.
Node to Row objects
- removed peerDeparture in solr remote search in case that peer does not
answer (this may be normal because it is allowed to switch this off)
- when doing a remote search, node peers are selected for solr queries
- the solr query is done concurrently to the standard YaCy rwi search
- the solr search result is feeded into the same data structure that
prepares the rwi search result
- the same remote seach that is done to several outside peers is done to
the local solr index
- the search process works now also without any 'old' RWI data using
solr
solr as fulltext search index.
- added slow migration from old metadata to solr index entries: each
entry from the old metadata is removed from that data structure and
written into solr.
to have a fall-back option in case that memory problems as reported in
http://forum.yacy-websuche.de/viewtopic.php?p=26901#p26901
for full-solr installation are too strong and we have to work with an
'small memory footprint' peer system.
for production as a replacement of the metadata-db.
This intermediate release 1.041 will switch on the previously optional
solr index and the old metadata-db will still work as it did before.
Solr+metadata are accessed in mixed mode, no migration is done yet.
If this causes not a catastrophe until the end of the weekend, we will
do a YaCy 1.1 main release containing this as default.
attached, the metadata is not written to the metadata-db, even if it is
enabled but instead to solr. This prevents that metadata is written in
two store systems at the same time. It is also the next step to migrate
the current metadata-db to solr.
metadata representation from the solr index. This shall replace metadata
from the built-in database in the future.
- added the Solr-driven metadata into the search index of YaCy which
makes it now possible to run YaCy without the old metadata index. This
is a major stept forward to a full migration to Solr.
a rss/opensearch writer and an enhanced solr xml writer.
The enhanced solr writer has less configuration overhead than the
original writer and should by slightly faster. The rss/opensearch writer
is at this time slightly incomplete compared with the already existing
rss search result form YaCy and also snippets are missing at this time.
To test the new interface, open for example:
http://localhost:8090/solr/select?wt=rss&q=olympia
The wt-code for the new result writers are=
wt=rss for opensearch
wt=exml for the enhanced solr xml writer.
Additionally, the SRU search parameters had been added to the solr
interface which can now also be used for a normal solr/xml search.
writings to the Metadata-DB are now also done to solr. This includes
metadata transfer during search and rwi transfer.
The new/added solr fields are:
## time when resource was loaded
load_date_dt
## date until resource shall be considered as fresh
fresh_date_dt
## id of the host, a 6-byte hash that is part of the document id
host_id_s
## ids of referrer to this document
referrer_id_ss
## the md5 of the raw source
md5_s
## the name of the publisher of the document
publisher_t
## the language used in the document; starts with primary language
language_ss
## an external ranking value
ranking_i
## the size of the raw source
size_i
## number of links to audio resources
audiolinkscount_i
## number of links to video resources
videolinkscount_i
## number of links to application resources
applinkscount_i
or will be superfluous or subject of complete redesign after the
migration to solr. Removing these things now will make the transition to
solr more simple.
- by default, a (empty) solr storage instance is created at
SEGMENTS/solr_36
- the index is written if in /IndexFederated_p.html the flag "embedded
solr search index" is switched on
- a standard solr query interface is available now with a new servlet at
http://127.0.0.1:8090/solr/select
To test this, do the following:
- switch to webportal mode
- switch on the feature as described
- do a crawl. this fills the solr index. The normal YaCy search will NOT
work now!
- do a solr query, like:
http://127.0.0.1:8090/solr/select?q=*:*
http://127.0.0.1:8090/solr/select?q=text_t:Help
play with different search fields as you can see in
/IndexFederated_p.html
You can use the standard solr query attributes as described in
http://wiki.apache.org/solr/SearchHandler
writter faster that the logger is able to print this out to its out
stream. A very large collection of unwritten log outputs had been seen
during strong crawling. The new ArrayBlockingQueue is limited to prevent
this case.
- added cleaning of blacklist cache if cache is modified in interface
- extended cache saving to all cache types
- moved cache location to DATA/LISTS
- fixed static file path which was relative to the application path but
should be relative to data path - which is different in debian and mac
implementations
of the major CPU users during snippet verification. The class was not
efficient for two reasons:
- it used a too complex input stream; generated from sources and UTF8
byte-conversions. The BufferedReader applied a strong overhead.
- to feed data into the SentenceReader, multiple toString/getBytes had
been applied until a buffered Reader from an input stream was possible.
These superfluous conversions had been removed.
- the best source for the Sentence Reader is a String. Therefore the
production of Strings had been forced inside the Document class.
A limit was necessary because some web pages have such huge numbers of
links that it can easily cause a OOM just by the number of links.
The quesion if the number of 1000 links is sufficient or too weak must
be answered with the result of testing this feature.
the segments had been there to create a tenant-infrastructure but were
never be used since that was all much too complex. There will be a
replacement using a solr navigation using a segment field in the search
index.
0-values and no empty strings are written). This may save a lot of
memory (in ram and on disc) if excessive 0-values or empty strings
appear)
- do not allow default boolean values for checkboxes because that does
not make sense: browsers may omit the checkbox attribute name if the box
is not checked. A default value 'true' would not comply with the
semantic of the browsers response.
- add a checkbox in IndexFederated_p for the lazy initialization of solr
fields.
computation. This error was there before the latest IPv6 hack but did
not cause a NPE. The IPv6 hack was not the cause for this bug, but it
discovered the misconfiguration of the 'referer' referrer.
- vocabulary annotation is not done any more into the metadata of urldb
- vocabularies are written into the jena triplestore using a rdf
vocabulary
- vocabularies for rdf tripel must be updated; refactoring done
- with the new navigation tags in the triplestore a faster
pre-urldb-lookup is possible: navigation is processed now within the RWI
during pre-ranking retrieval
- added also a Owl vocabulary stub to add the plain-text url to the
triplestore using the owl:sameas predicate
later
- added abstract add, delete, get methods in the triplestore
- added generation of triples after auto-annotation
- migrated all MultiProtocolURI objects to DigestURI in the parser since
the url hash is needed as subject value in the triples in the triple
store
rewrite engine to customize existing webpages. originally implemented by
Florian Richter.
Conflicts:
source/de/anomic/http/server/HTTPDProxyHandler.java
- new limit to use the table copy (instead of flag): 400MB available. If
less is available, then a copy is never used. If more is available, then
it can be used if there is a remaining space of at least 200MB
- flush caches more often: flush the Digest cache
this causes that the table is not initialized when there is no search is
done. the effect is most strong if YaCy is started headless which causes
no browser pop-up which otherwise would load the search page and
therefore trigger the initialization of the table.
not equal. It might happen that these data is different because one of
that caches is cleaned after a while or when it is too big. The metadata
is then not cleaned, but now wiped after a checkup process at every
application start. This should cause a bit less memory usage.
- added log warnings in case that search processes run into time-out
situations
- better concurrency for Integer formatter (used a non-synchronized
formatter before)
- bugfix for search termination (a poison pill was missing)
- added timeout parameters for search (again) -> target is, that they
are never reached.
- search request are now made using a map boundary
- search results are only computed for the map boundary
- the number of results is adopted to the results in the visible range
- added a double-buffering for the search result markers
- added a search query option for the search results:
/radius/<lat>/<lon>/<radius>
SeedDB.lookupByName searche for lowercase peerNames, while MapColumnIndex.getIndex uses peername as is in the keyset.
Changed the index init to insert lowercase peer names as key
- Changing the format of YaCy's solr.key.list while maintainig backward compatibility
Federated index config screens adjusted accordingly
- modified the Solr update request to use a 3 min Solr autocommit intervall
Error was:
java.lang.UnsupportedOperationException: Client was created outside of
HttpSolrServer
at
org.apache.solr.client.solrj.impl.HttpSolrServer.setDefaultMaxConnectionsPerHost(HttpSolrServer.java:614)
at
net.yacy.cora.services.federated.solr.SolrSingleConnector.<init>(SolrSingleConnector.java:128)
at
net.yacy.cora.services.federated.solr.SolrShardingConnector.<init>(SolrShardingConnector.java:55)
at net.yacy.search.Switchboard.<init>(Switchboard.java:657)
at net.yacy.yacy.startup(yacy.java:222)
at net.yacy.yacy.main(yacy.java:1018)
queue and not from virtual documents generated by the parser.
- The parser now generates nice description texts for NOLOAD entries
which shall make it possible to find media content using the search
index and not using the media prefetch algorithm during search (which
was costly)
- Removed the media-search prefetch process from image search
only links where the content can be parsed. All non-parseable links are
placed into the noload queue. The search process must therefore be able
to filter out non-text search results.
- This fixes the problem that image search results appeared in the text
search.
- The interactive search can retrieve now ALL types of links
- The p2p interface is now extended to retrieve only certain types of
links (text, image, video, apps)
- The search process has an extension to filter the right document type
according to the search query
- a new database for url reference data ('seen links')
- a new database extending the references to the full url metadata
attributes set which shall replace the old metadata database if it is
finished
- migration help classes stub to use old and new metadata databases
simultanously
the links between all pages are now stored. The same index structure as
used for the word index is used to make a reverse link index.
The new file(s) in SEGMENT/default/citation.index.*.blob store the
citation index. This will be used to create much more detailed link
structures for the YaCy apis and to create a better ranking. A ranking
using the citation.index should provide better results especially for
portal indexes and initranets.
See
http://docs.oracle.com/javase/6/docs/api/java/util/concurrent/ConcurrentLinkedQueue.html
and the following test programm:
public class QueueLengthTimeTest {
public static long countTest(Queue<Integer> q, int c) {
long t = System.currentTimeMillis();
for (int i = 0; i < c; i++) {
q.add(q.size());
}
return System.currentTimeMillis() - t;
}
public static void main(String[] args) {
int c = 1;
for (int i = 0; i < 100; i++) {
Runtime.getRuntime().gc();
long t1 = countTest(new ArrayBlockingQueue<Integer>(c), c);
Runtime.getRuntime().gc();
long t2 = countTest(new LinkedBlockingQueue<Integer>(), c);
Runtime.getRuntime().gc();
long t3 = countTest(new ConcurrentLinkedQueue<Integer>(),
c);
System.out.println("count = " + c + ": ArrayBlockingQueue =
" + t1 + ", LinkedBlockingQueue = " + t2 + ", ConcurrentLinkedQueue = "
+ t3);
c = c * 2;
}
}
}
1. Improved and fixed language detection:
1.1 Identificator.java - recognition fix (improved)
1.2 DCEntry.java - fix (changed detection order due to detection from
tld in many cases is incorrect)
1.3 MultiProtocolURI.java - fixed and enhanced language from tld
detection (all currently used top-level domains; ccTLD added but not
tested).
2. Ukrainian language update.
3. Main Slavic languages langstats (tested and works fine).
ready-prepared crawl list but at the stacks of the domains that are
stored for balanced crawling. This affects also the balancer since that
does not need to prepare the pre-selected crawl list for monitoring. As
a effect:
- it is no more possible to see the correct order of next to-be-crawled
links, since that depends on the actual state of the balancer stack the
next time another url is requested for loading
- the balancer works better since the next url can be selected according
to the current situation and not according to a pre-selected order.
but this did not solve all problems because a bug in the apache http
client prevented that it worked. Thread dump:
Caused by: java.lang.NumberFormatException: For input string:
"1450:400c:c01:0:0:0:69"
at
java.lang.NumberFormatException.forInputString(NumberFormatException.java:48)
at java.lang.Integer.parseInt(Integer.java:458)
at java.lang.Integer.parseInt(Integer.java:499)
at org.apache.http.client.utils.URIUtils.extractHost(URIUtils.java:310)
at
org.apache.http.impl.client.AbstractHttpClient.determineTarget(AbstractHttpClient.java:764)
at
org.apache.http.impl.client.AbstractHttpClient.execute(AbstractHttpClient.java:754)
at net.yacy.cora.protocol.http.HTTPClient.execute(HTTPClient.java:597)
at
net.yacy.cora.protocol.http.HTTPClient.getContentBytes(HTTPClient.java:558)
at net.yacy.cora.protocol.http.HTTPClient.GETbytes(HTTPClient.java:341)
at de.anomic.crawler.retrieval.HTTPLoader.load(HTTPLoader.java:131)
at de.anomic.crawler.retrieval.HTTPLoader.load(HTTPLoader.java:74)
at
net.yacy.repository.LoaderDispatcher.loadInternal(LoaderDispatcher.java:274)
at net.yacy.repository.LoaderDispatcher.load(LoaderDispatcher.java:164)
at net.yacy.repository.LoaderDispatcher.load(LoaderDispatcher.java:150)
at
net.yacy.repository.LoaderDispatcher.loadDocument(LoaderDispatcher.java:355)
at getpageinfo_p.respond(getpageinfo_p.java:97)
during receiving of DHT submissions and when answering remote search
requests. Both events together may have caused IO-deadlocking and this
commit shall fix that.
parser. This makes it possible that every type of document can be a
crawl start point, not only text documents or html documents. Testet
this with a pdf document.
method. This is much faster and produces less blocking when using the
Compressor class which is used by the HTCache. I.e. picture search is
much faster now.
- tags that are automatically generated now enrich the dc:subject
- auto-generated tags have a '$' at the beginning of the tag
- auto-generated tags lead the tag name with a vocabulary name
each tag has the form
$<vocabulary-name>:<tag-printname-space-replaced-by-'_'>
Michael Peter Christen