Commit 2a07ba186d82f6cbd0278905e79f66515c7285c0

Authored by Matthijs Brouwer
1 parent ac3fc9e5

upgrade 6.4.0

docker/Dockerfile
1 1 # Automatically generated Dockerfile
2   -# - Build 2017-01-28 11:32
3   -# - Lucene/Solr version 6.3.0
  2 +# - Build 2017-01-31 10:25
  3 +# - Lucene/Solr version 6.4.0
4 4 # - Mtas release 20170110
5 5 #
6 6  
... ... @@ -15,12 +15,12 @@ WORKDIR "/root"
15 15  
16 16 RUN mkdir lib
17 17  
18   -ADD https://github.com/meertensinstituut/mtas/releases/download/20170110/mtas-6.3.0.jar /root/lib/
  18 +ADD https://github.com/meertensinstituut/mtas/releases/download/20170110/mtas-6.4.0.jar /root/lib/
19 19  
20 20 RUN apt-get update && apt-get install -y lsof software-properties-common python-software-properties apache2 curl subversion \
21 21 && locale-gen en_US.UTF-8 en_US && update-locale LANG=en_US.UTF-8 LANGUAGE=en_US:en \
22   -&& solrurl=$(curl -s 'http://www.apache.org/dyn/closer.lua/lucene/solr/6.3.0/solr-6.3.0.tgz' | grep -o '<strong>[^<]*</strong>' | sed 's/<[^>]*>//g' | head -1) \
23   -&& curl -o /root/solr-6.3.0.tgz -O $solrurl \
  22 +&& solrurl=$(curl -s 'http://www.apache.org/dyn/closer.lua/lucene/solr/6.4.0/solr-6.4.0.tgz' | grep -o '<strong>[^<]*</strong>' | sed 's/<[^>]*>//g' | head -1) \
  23 +&& curl -o /root/solr-6.4.0.tgz -O $solrurl \
24 24 && mathurl=$(curl -s 'http://www.apache.org/dyn/closer.lua/commons/math/binaries/commons-math3-3.6.1-bin.tar.gz' | grep -o '<strong>[^<]*</strong>' | sed 's/<[^>]*>//g' | head -1) \
25 25 && curl -o /root/lib/commons-math3-3.6.1-bin.tar.gz -O $mathurl \
26 26 && tar xzf lib/commons-math3-3.6.1-bin.tar.gz -C lib commons-math3-3.6.1/commons-math3-3.6.1.jar --strip-components=1 \
... ... @@ -31,8 +31,8 @@ RUN apt-get update &amp;&amp; apt-get install -y lsof software-properties-common python-
31 31 && echo oracle-java8-installer shared/accepted-oracle-license-v1-1 select true | /usr/bin/debconf-set-selections \
32 32 && apt-get install -y oracle-java8-installer \
33 33 && rm -rf /var/lib/apt/lists/* \
34   -&& tar xzf solr-6.3.0.tgz solr-6.3.0/bin/install_solr_service.sh --strip-components=2 \
35   -&& bash ./install_solr_service.sh solr-6.3.0.tgz && rm install_solr_service.sh && rm -rf solr-6.3.0.tgz \
  34 +&& tar xzf solr-6.4.0.tgz solr-6.4.0/bin/install_solr_service.sh --strip-components=2 \
  35 +&& bash ./install_solr_service.sh solr-6.4.0.tgz && rm install_solr_service.sh && rm -rf solr-6.4.0.tgz \
36 36 && service apache2 stop \
37 37 && echo "ServerName localhost" | tee /etc/apache2/conf-available/fqdn.conf \
38 38 && a2enmod proxy \
... ... @@ -55,8 +55,8 @@ RUN apt-get update &amp;&amp; apt-get install -y lsof software-properties-common python-
55 55 && chmod -R 755 /var/www/html \
56 56 && printf "echo\n" >> /start.sh \
57 57 && printf "echo \"================ Mtas -- Multi Tier Annotation Search =================\"\n" >> /start.sh \
58   -&& printf "echo \" Timestamp 2017-01-28 11:32\"\n" >> /start.sh \
59   -&& printf "echo \" Lucene/Solr version 6.3.0\"\n" >> /start.sh \
  58 +&& printf "echo \" Timestamp 2017-01-31 10:25\"\n" >> /start.sh \
  59 +&& printf "echo \" Lucene/Solr version 6.4.0\"\n" >> /start.sh \
60 60 && printf "echo \" Mtas release 20170110\"\n" >> /start.sh \
61 61 && printf "echo \" See https://meertensinstituut.github.io/mtas/ for more information\"\n" >> /start.sh \
62 62 && printf "echo \"=======================================================================\"\n" >> /start.sh \
... ... @@ -66,7 +66,7 @@ RUN apt-get update &amp;&amp; apt-get install -y lsof software-properties-common python-
66 66 && mkdir demo1 && mkdir demo1/lib && mkdir demo1/conf \
67 67 && echo "name=demo1" > demo1/core.properties \
68 68 && cp lib/commons-math3-3.6.1.jar demo1/lib/ \
69   -&& cp lib/mtas-6.3.0.jar demo1/lib/ \
  69 +&& cp lib/mtas-6.4.0.jar demo1/lib/ \
70 70 && cp data/solrconfig.xml demo1/conf/ \
71 71 && cp data/schemaBasic.xml demo1/conf/schema.xml \
72 72 && cp -r data/mtas demo1/conf/ && cp data/mtas.xml demo1/conf/ \
... ...
... ... @@ -4,13 +4,13 @@
4 4 <timestamp>${maven.build.timestamp}</timestamp>
5 5 <maven.build.timestamp.format>yyyy-MM-dd HH:mm</maven.build.timestamp.format>
6 6 <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
7   - <currentDevelopmentVersion>6.3.0</currentDevelopmentVersion>
  7 + <currentDevelopmentVersion>6.4.0</currentDevelopmentVersion>
8 8 <currentDevelopmentRelease>20170110</currentDevelopmentRelease>
9 9 </properties>
10 10 <modelVersion>4.0.0</modelVersion>
11 11 <groupId>dev.meertens.mtas</groupId>
12 12 <artifactId>mtas</artifactId>
13   - <version>6.3.0</version>
  13 + <version>6.4.0</version>
14 14 <packaging>jar</packaging>
15 15 <licenses>
16 16 <license>
... ... @@ -216,27 +216,27 @@
216 216 <dependency>
217 217 <groupId>org.apache.lucene</groupId>
218 218 <artifactId>lucene-core</artifactId>
219   - <version>6.3.0</version>
  219 + <version>6.4.0</version>
220 220 </dependency>
221 221 <dependency>
222 222 <groupId>org.apache.lucene</groupId>
223 223 <artifactId>lucene-analyzers-common</artifactId>
224   - <version>6.3.0</version>
  224 + <version>6.4.0</version>
225 225 </dependency>
226 226 <dependency>
227 227 <groupId>org.apache.lucene</groupId>
228 228 <artifactId>lucene-queryparser</artifactId>
229   - <version>6.3.0</version>
  229 + <version>6.4.0</version>
230 230 </dependency>
231 231 <dependency>
232 232 <groupId>org.apache.lucene</groupId>
233 233 <artifactId>lucene-codecs</artifactId>
234   - <version>6.3.0</version>
  234 + <version>6.4.0</version>
235 235 </dependency>
236 236 <dependency>
237 237 <groupId>org.apache.solr</groupId>
238 238 <artifactId>solr-core</artifactId>
239   - <version>6.3.0</version>
  239 + <version>6.4.0</version>
240 240 </dependency>
241 241 <dependency>
242 242 <groupId>org.apache.commons</groupId>
... ...
src/mtas/analysis/parser/MtasCRMParser.java
... ... @@ -516,10 +516,11 @@ public class MtasCRMParser extends MtasBasicParser {
516 516 MtasParserType tmpCurrentType;
517 517 MtasParserObject currentObject;
518 518 if ((tmpCurrentType = crmSentenceTypes.get(name)) != null) {
  519 + String filteredText = text.replaceAll("[^0-9\\-]", "");
519 520 currentObject = new MtasParserObject(tmpCurrentType);
520 521 currentObject.setUnknownAncestorNumber(unknownAncestors);
521   - currentObject.setRealOffsetStart(currentOffset);
522   - currentObject.setText(text);
  522 + currentObject.setRealOffsetStart(currentOffset);
  523 + currentObject.setText(filteredText);
523 524 if (!prevalidateObject(currentObject, currentList)) {
524 525 return null;
525 526 } else {
... ... @@ -567,10 +568,11 @@ public class MtasCRMParser extends MtasBasicParser {
567 568 MtasParserType tmpCurrentType;
568 569 MtasParserObject currentObject;
569 570 if ((tmpCurrentType = crmClauseTypes.get(name)) != null) {
  571 + String filteredText = text.replaceAll("[^0-9\\-]", "");
570 572 currentObject = new MtasParserObject(tmpCurrentType);
571 573 currentObject.setUnknownAncestorNumber(unknownAncestors);
572 574 currentObject.setRealOffsetStart(currentOffset);
573   - currentObject.setText(text);
  575 + currentObject.setText(filteredText);
574 576 if (!prevalidateObject(currentObject, currentList)) {
575 577 return null;
576 578 } else {
... ...
src/mtas/analysis/token/MtasTokenCollection.java
... ... @@ -137,7 +137,7 @@ public class MtasTokenCollection {
137 137 } else {
138 138 row[7] = null;
139 139 row[8] = null;
140   - row[9] = token.getPositions().toString();
  140 + row[9] = Arrays.toString(token.getPositions());
141 141 }
142 142 }
143 143 if (token.getParentId() != null) {
... ...