Examples with Document org.icij.datashare.text.Document used on opensource projects

Example 16 with Document

use of org.icij.datashare.text.Document in project datashare by ICIJ.

the class ElasticsearchIndexerTest method test_delete_by_query.

@Test
public void test_delete_by_query() throws Exception {
    Document doc = new org.icij.datashare.text.Document("docId", project("prj"), Paths.get("doc.txt"), "content", Language.FRENCH, Charset.defaultCharset(), "application/pdf", new HashMap<>(), INDEXED, new HashSet<>(), 4324L);
    indexer.add(TEST_INDEX, doc);
    indexer.add(TEST_INDEX, create(PERSON, "Joe Foo", asList(2L), "docId", "root", CORENLP, Language.FRENCH));
    indexer.add(TEST_INDEX, create(PERSON, "John Doe", asList(12L), "docId", "root", CORENLP, Language.FRENCH));
    assertThat(indexer.deleteAll(TEST_INDEX)).isTrue();
    Object[] documents = indexer.search(TEST_INDEX, Document.class).execute().toArray();
    assertThat(documents.length).isEqualTo(0);
}

Example 17 with Document

use of org.icij.datashare.text.Document in project datashare by ICIJ.

the class ElasticsearchIndexerTest method test_search_with_status.

@Test
public void test_search_with_status() throws IOException {
    Document doc = new org.icij.datashare.text.Document("id", project("prj"), Paths.get("doc.txt"), "content", Language.FRENCH, Charset.defaultCharset(), "application/pdf", new HashMap<>(), INDEXED, new HashSet<>(), 123L);
    indexer.add(TEST_INDEX, doc);
    List<? extends Entity> lst = indexer.search(TEST_INDEX, Document.class).ofStatus(INDEXED).execute().collect(toList());
    assertThat(lst.size()).isEqualTo(1);
    assertThat((int) indexer.search(TEST_INDEX, Document.class).ofStatus(DONE).execute().count()).isEqualTo(0);
}

Example 18 with Document

use of org.icij.datashare.text.Document in project datashare by ICIJ.

the class ElasticsearchIndexerTest method test_group_tag_untag_documents.

@Test
public void test_group_tag_untag_documents() throws IOException {
    Document doc1 = new org.icij.datashare.text.Document("id1", project("prj"), Paths.get("doc1.txt"), "content1", Language.FRENCH, Charset.defaultCharset(), "application/pdf", new HashMap<>(), INDEXED, new HashSet<>(), 123L);
    Document doc2 = new org.icij.datashare.text.Document("id2", project("prj"), Paths.get("doc2.txt"), "content2", Language.FRENCH, Charset.defaultCharset(), "application/pdf", new HashMap<>(), INDEXED, new HashSet<>(), 123L);
    indexer.add(TEST_INDEX, doc1);
    indexer.add(TEST_INDEX, doc2);
    assertThat(indexer.tag(project(TEST_INDEX), asList("id1", "id2"), tag("foo"), tag("bar"))).isTrue();
    assertThat(((Document) indexer.get(TEST_INDEX, "id1")).getTags()).containsOnly(tag("foo"), tag("bar"));
    assertThat(((Document) indexer.get(TEST_INDEX, "id2")).getTags()).containsOnly(tag("foo"), tag("bar"));
    assertThat(indexer.untag(project(TEST_INDEX), asList("id1", "id2"), tag("foo"), tag("bar"))).isTrue();
    assertThat(((Document) indexer.get(TEST_INDEX, "id1")).getTags()).isEmpty();
    assertThat(((Document) indexer.get(TEST_INDEX, "id2")).getTags()).isEmpty();
}

Example 19 with Document

use of org.icij.datashare.text.Document in project datashare by ICIJ.

the class ElasticsearchIndexerTest method test_update_named_entity.

@Test
public void test_update_named_entity() throws IOException {
    Document parent = new org.icij.datashare.text.Document("id", project("prj"), Paths.get("doc.txt"), "content Madeline", Language.FRENCH, Charset.defaultCharset(), "text/plain", new HashMap<>(), DONE, new HashSet<>(), 123L);
    NamedEntity ne = create(PERSON, "Madeline", asList(8L), parent.getId(), "root", CORENLP, Language.ENGLISH);
    indexer.add(TEST_INDEX, parent);
    indexer.add(TEST_INDEX, ne);
    ne.hide();
    indexer.update(TEST_INDEX, ne);
    NamedEntity neFromES = indexer.get(TEST_INDEX, ne.getId(), parent.getId());
    assertThat(neFromES.isHidden()).isTrue();
}

Example 20 with Document

use of org.icij.datashare.text.Document in project datashare by ICIJ.

the class ElasticsearchIndexerTest method test_search_with_json_query.

@Test
public void test_search_with_json_query() throws IOException {
    Document doc = new org.icij.datashare.text.Document("id", project("prj"), Paths.get("doc.txt"), "content", Language.FRENCH, Charset.defaultCharset(), "application/pdf", new HashMap<>(), INDEXED, new HashSet<>(), 123L);
    indexer.add(TEST_INDEX, doc);
    String query = "{\"bool\":{\"must\":[{\"match_all\":{}},{\"bool\":{\"should\":[{\"query_string\":{\"query\":\"*\"}}]}},{\"match\":{\"type\":\"Document\"}}]}}";
    List<? extends Entity> lst = indexer.search(TEST_INDEX, Document.class).set(JsonObjectMapper.MAPPER.readTree(query)).execute().collect(toList());
    assertThat(lst.size()).isEqualTo(1);
}