dariah / mtas

#Apache Solr

Mtas can be used as plugin for Apache Solr

**Prerequisites**

- Installed [Apache Solr](https://lucene.apache.org/solr/)
- Currently supported and advised version is $context.get("latestReleaseLuceneVersion")

Start with a new Solr core.

**Libraries**

Add the `mtas-${context.get("latestReleaseFullVersion")}.jar` to the `lib` directory of the new Solr core. 
A prebuilt `mtas-${context.get("latestReleaseFullVersion")}.jar` is available from the [download](download.html) page.

Furthermore, add the [Apache Commons Mathematics Library](http://commons.apache.org/proper/commons-math/) to the `lib` directory of the new Solr core.

**Solrconfig.xml**

Some changes have to be made within the `solrconfig.xml` file, elements have to be added to the `<config/>` or existing elements have te be adjusted:

Define a new **mtas searchComponent**: 

```console
<searchComponent name="mtas" class="mtas.solr.handler.component.MtasSolrSearchComponent"/>
```

Add this component to the select requestHandler by inserting the following within the 
`<requestHandler/>` with name `"/select"`:

``` console 
<arr name="last-components">
  <str>mtas</str>
</arr>
```   

Define a new **mtas_cql queryParser** and **mtas_join queryParser**:

```console
<queryParser name="mtas_cql" class="mtas.solr.search.MtasSolrCQLQParserPlugin"/>
<queryParser name="mtas_join" class="mtas.solr.search.MtasSolrJoinQParserPlugin"/>
``` 

Define a new **mtas requestHandler**:

```console
<requestHandler name="/mtas" class="mtas.solr.handler.MtasRequestHandler" />
```

Define a new updateRequestProcessorChain:

```console
<updateRequestProcessorChain name="mtasUpdateProcessor">
  <processor class="mtas.solr.update.processor.MtasUpdateRequestProcessorFactory" />
  <processor class="solr.LogUpdateProcessorFactory" />
  <processor class="solr.RunUpdateProcessorFactory" />
</updateRequestProcessorChain>
```

Define or adjust the update requestHandler with this updateRequestProcessorChain:

```console
<requestHandler name="/update" class="solr.UpdateRequestHandler">
  <lst name="defaults">
    <str name="update.chain">mtasUpdateProcessor</str>
  </lst>    
</requestHandler>
```
   
Finally, in this instruction we will use a classic schema instead of the 
managed-schema. So the configuration must contain:

```console
<schemaFactory class="ClassicIndexSchemaFactory"/>
```    
  
**Schema.xml**

We extend a (classic) schema with one (or multiple) fields that may contain 
annotated text, e.g.

```console
<field name="text" type="mtas" required="false" multiValued="false" indexed="true" stored="true" />
```   

We define the referred Mtas fieldType by

```console
<fieldType name="mtas" class="solr.TextField" postingsFormat="MtasCodec">
  <analyzer type="index">
    <charFilter class="mtas.analysis.util.MtasCharFilterFactory" type="url" prefix="http://localhost/demo/" postfix="" />
    <tokenizer class="mtas.analysis.util.MtasTokenizerFactory" configFile="folia.xml" />
  </analyzer>
  <analyzer type="query">
    <tokenizer class="solr.WhitespaceTokenizerFactory" />
    <filter class="mtas.analysis.util.MtasPrefixTokenFilterFactory" prefix="t" />
  </analyzer>
</fieldType>
```

The charFilter with class *mtas.analysis.util.MtasCharFilterFactory* in the 
index analyzer contains an obligatory attribute `type` and two optional attributes 
`prefix` and `postfix`. The *type* can be *url* or *file*, referring to either an external
url or a file on the filesystem. On indexing, the optional *prefix* and *postfix* attributes 
will be added to the provided value, resulting in a full url or location of a file. 

The tokenizer with class *mtas.analysis.util.MtasTokenizerFactory* in the index analyzer
has an attribute `configFile` containing the name of the required tokenizer configuration.

The filter in the query analyzer contains an obligatory attribute `prefix` defining the 
assumed prefix when this field will be queried directly within Solr.

See [configuration](indexing_configuration.html) for more information about
the definition of a tokenizer configuration.

**Multiple tokenize configurations**

If multiple tokenizer configurations are required, the Mtas fieldType has to be 
defined slightly different: 

```console
<fieldType name="mtas_config" class="solr.TextField" postingsFormat="MtasCodec">
  <analyzer type="index">
    <charFilter class="mtas.analysis.util.MtasCharFilterFactory" config="mtas.xml" />
    <tokenizer class="mtas.analysis.util.MtasTokenizerFactory" config="mtas.xml" />
  </analyzer>
</fieldType>
<fieldType name="mtas" class="mtas.solr.schema.MtasPreAnalyzedField"
  followIndexAnalyzer="mtas_config" defaultConfiguration="default"
  configurationFromField="type" setNumberOfTokens="numberOfTokens"
  setNumberOfPositions="numberOfPositions" setSize="size"
  setError="error" setPrefix="prefixes" postingsFormat="MtasCodec">
  <analyzer type="query">
    <tokenizer class="solr.WhitespaceTokenizerFactory" />
    <filter class="mtas.analysis.util.MtasPrefixTokenFilterFactory" prefix="t" />
  </analyzer>
</fieldType>
```

An additional fieldType (here named mtas_config) is defined, containing 
only an index analyzer. Both the charFilter and tokenizer within this analyzer have
an attribute `config` referring to a Mtas configuration file. Depending on the required 
tokenizer configuration, for the charFilter this file will define *type*, *prefix* and *postfix* and for the tokenizer this file 
will define the *configFile*. An example of a Mtas configuration file is added below.

The Mtas fieldType is defined with class *mtas.solr.schema.MtasPreAnalyzedField*, 
an obligatory attribute `followIndexAnalyzer` referring to the additional fieldType
we defined before. The optional attribute `defaultConfiguration` contains the name 
of the default configuration to be used, and the obligatory attribute 
`configurationFromField` contains the name of the field defining the required 
configuration. The optional attributes `setNumberOfTokens`, 
`setNumberOfPositions`, `setSize`, `setPrefix` and `setError` define fields that may be filled
with respectively number of tokens, number of positions, filesize, prefixes and possible errors.

**Example of a Mtas configuration file**

```console
<?xml version="1.0" encoding="UTF-8" ?>
<mtas>
  <configurations type="mtas.analysis.util.MtasTokenizerFactory">
    <configuration name="folia" file="folia.xml" />
    <configuration name="tei" file="tei.xml" />
  </configurations>
  <configurations type="mtas.analysis.util.MtasCharFilterFactory">
    <configuration name="folia" type="url" prefix="http://www.mycompany.com/archive/" postfix=".xml" />
    <configuration name="tei" type="file" prefix="/storage/tei/" postfix="" />
  </configurations>
</mtas>
```