source: nutchez-0.1/default.properties

Last change on this file was 66, checked in by waue, 15 years ago

NutchEz - an easy way to nutch

File size: 3.2 KB
Line 
1Name=Nutch
2name=nutch
3version=1.0
4final.name=${name}-${version}
5year=2006
6
7basedir = ./
8src.dir = ./src/java
9lib.dir = ./lib
10conf.dir = ./conf
11plugins.dir = ./src/plugin
12docs.dir = ./docs
13docs.src = ${basedir}/src/web
14xmlcatalog.dir = ${basedir}/src/xmlcatalog
15
16build.dir = ./build
17build.classes = ${build.dir}/classes
18build.webapps = ${build.dir}/webapps
19build.plugins = ${build.dir}/plugins
20build.docs = ${build.dir}/docs
21build.javadoc = ${build.docs}/api
22build.encoding = UTF-8
23
24test.src.dir = ./src/test
25test.build.dir = ${build.dir}/test
26test.build.data =  ${test.build.dir}/data
27test.build.classes = ${test.build.dir}/classes
28test.build.javadoc = ${test.build.dir}/docs/api
29
30javacc.home=/usr/java/javacc
31
32web.src.dir = ./src/web
33src.webapps = ./src/webapps
34
35# Proxy Host and Port to use for building JavaDoc
36javadoc.proxy.host=-J-DproxyHost=
37javadoc.proxy.port=-J-DproxyPort=
38javadoc.link.java=http://java.sun.com/j2se/1.4.2/docs/api/
39javadoc.link.lucene=http://jakarta.apache.org/lucene/docs/api/
40javadoc.link.hadoop=http://lucene.apache.org/hadoop/docs/api/
41javadoc.packages=org.apache.nutch.*
42
43dist.dir=${build.dir}/${final.name}
44
45javac.debug=on
46javac.optimize=on
47javac.deprecation=off
48javac.version= 1.5
49
50#
51# Plugins API
52#
53plugins.api=\
54   org.apache.nutch.protocol.http.api*:\
55   org.apache.nutch.urlfilter.api*:\
56   org.apache.nutch.parse.ms*
57
58#
59# Protocol Plugins
60#
61plugins.protocol=\
62   org.apache.nutch.protocol.file*:\
63   org.apache.nutch.protocol.ftp*:\
64   org.apache.nutch.protocol.http*:\
65   org.apache.nutch.protocol.httpclient*
66
67#
68# URL Filter Plugins
69#
70plugins.urlfilter=\
71   org.apache.nutch.urlfilter.automaton*:\
72   org.apache.nutch.urlfilter.prefix*:\
73   org.apache.nutch.urlfilter.regex*
74
75#
76# Scoring Plugins
77#
78plugins.scoring=\
79   org.apache.nutch.scoring.opic*
80
81#
82# Parse Plugins
83#
84plugins.parse=\
85   org.apache.nutch.parse.ext*:\
86   org.apache.nutch.parse.html*:\
87   org.apache.nutch.parse.js:\
88   org.apache.nutch.parse.mp3*:\
89   org.apache.nutch.parse.msexcel*:\
90   org.apache.nutch.parse.mspowerpoint*:\
91   org.apache.nutch.parse.msword*:\
92   org.apache.nutch.parse.oo*:\
93   org.apache.nutch.parse.pdf*:\
94   org.apache.nutch.parse.rtf*:\
95   org.apache.nutch.parse.rss*:\
96   org.apache.nutch.parse.swf*:\
97   org.apache.nutch.parse.text:\
98   org.apache.nutch.parse.zip
99
100#
101# Analysis Plugins
102#
103plugins.analysis=\
104#  ${plugin.analysis-de}:\
105#  ${plugin.analysis-fr}
106
107#
108# Indexing Filter Plugins
109#
110plugins.index=\
111   org.apache.nutch.indexer.basic*:\
112   org.apache.nutch.indexer.more*
113
114#
115# Query Filter Plugins
116#
117plugins.query=\
118   org.apache.nutch.searcher.basic*:\
119   org.apache.nutch.searcher.more*:\
120   org.apache.nutch.searcher.site*:\
121   org.apache.nutch.searcher.url*
122
123#
124# Ontology Plugins
125#
126plugins.ontology=\
127   org.apache.nutch.ontology.jena*
128
129#
130# Online Clusterer Plugins
131#
132plugins.clustering=\
133   org.apache.nutch.clustering.carrot2*
134
135#
136# Summary Plugins
137#
138plugins.summary=\
139   org.apache.nutch.summary.basic*:\
140   org.apache.nutch.summary.lucene*
141
142#
143# Misc. Plugins
144#
145# (gathers plugins that cannot be dispatched
146# in any category, mainly because they contains
147# many extension points)
148#
149plugins.misc=\
150   org.apache.nutch.analysis.lang*:\
151   org.apache.nutch.microformats.reltag*:\
152   org.creativecommons.nutch*
Note: See TracBrowser for help on using the repository browser.