use of org.apache.tika.eval.db.TableInfo in project tika by apache.
the class SimpleComparerTest method testAccessException.
@Test
public void testAccessException() throws Exception {
EvalFilePaths fpsA = new EvalFilePaths(Paths.get("file6_accessEx.pdf.json"), getResourceAsFile("/test-dirs/extractsA/file6_accessEx.pdf.json").toPath());
EvalFilePaths fpsB = new EvalFilePaths(Paths.get("file6_accessEx.pdf.json"), getResourceAsFile("/test-dirs/extractsB/file6_accessEx.pdf.json").toPath());
comparer.compareFiles(fpsA, fpsB);
for (TableInfo t : new TableInfo[] { ExtractComparer.EXCEPTION_TABLE_A, ExtractComparer.EXCEPTION_TABLE_B }) {
List<Map<Cols, String>> table = writer.getTable(t);
Map<Cols, String> rowA = table.get(0);
//debugPrintRow(rowA);
assertEquals(Integer.toString(EXCEPTION_TYPE.ACCESS_PERMISSION.ordinal()), rowA.get(Cols.PARSE_EXCEPTION_ID));
assertNull(rowA.get(Cols.ORIG_STACK_TRACE));
assertNull(rowA.get(Cols.SORT_STACK_TRACE));
}
}
use of org.apache.tika.eval.db.TableInfo in project tika by apache.
the class SimpleComparerTest method oneOff.
@Test
@Ignore("useful for testing 2 files not in test set")
public void oneOff() throws Exception {
Path p1 = Paths.get("");
Path p2 = Paths.get("");
EvalFilePaths fpsA = new EvalFilePaths(Paths.get("file1.pdf.json"), p1);
EvalFilePaths fpsB = new EvalFilePaths(Paths.get("file1.pdf.json"), p2);
comparer.compareFiles(fpsA, fpsB);
for (TableInfo t : new TableInfo[] { ExtractComparer.COMPARISON_CONTAINERS, ExtractComparer.EXTRACT_EXCEPTION_TABLE_A, ExtractComparer.EXTRACT_EXCEPTION_TABLE_B, ExtractComparer.EXCEPTION_TABLE_A, ExtractComparer.EXCEPTION_TABLE_B, ExtractComparer.PROFILES_A, ExtractComparer.PROFILES_B, ExtractComparer.CONTENTS_TABLE_A, ExtractComparer.CONTENTS_TABLE_B, ExtractComparer.CONTENT_COMPARISONS }) {
debugPrintTable(t);
}
}
use of org.apache.tika.eval.db.TableInfo in project tika by apache.
the class SimpleComparerTest method testDebug.
@Test
@Ignore
public void testDebug() throws Exception {
Path commonTokens = Paths.get(getResourceAsFile("/common_tokens_short.txt").toURI());
AbstractProfiler.loadCommonTokens(commonTokens, "en");
EvalFilePaths fpsA = new EvalFilePaths(Paths.get("file1.pdf.json"), getResourceAsFile("/test-dirs/extractsA/file1.pdf.json").toPath());
EvalFilePaths fpsB = new EvalFilePaths(Paths.get("file1.pdf.json"), getResourceAsFile("/test-dirs/extractsB/file1.pdf.json").toPath());
comparer.compareFiles(fpsA, fpsB);
for (TableInfo t : new TableInfo[] { ExtractComparer.COMPARISON_CONTAINERS, ExtractComparer.EXTRACT_EXCEPTION_TABLE_A, ExtractComparer.EXTRACT_EXCEPTION_TABLE_B, ExtractComparer.EXCEPTION_TABLE_A, ExtractComparer.EXCEPTION_TABLE_B, ExtractComparer.PROFILES_A, ExtractComparer.PROFILES_B, ExtractComparer.CONTENTS_TABLE_A, ExtractComparer.CONTENTS_TABLE_B, ExtractComparer.CONTENT_COMPARISONS }) {
//debugPrintTable(t);
}
}
use of org.apache.tika.eval.db.TableInfo in project tika by apache.
the class ExtractComparerBuilder method updateTableInfosWithPrefixes.
@Override
protected void updateTableInfosWithPrefixes(Map<String, String> attrs) {
String tablePrefixA = localAttrs.get(TABLE_PREFIX_A_KEY);
String tablePrefixB = localAttrs.get(TABLE_PREFIX_B_KEY);
tablePrefixA = (tablePrefixA == null || tablePrefixA.endsWith("_")) ? tablePrefixA : tablePrefixA + "_";
tablePrefixB = (tablePrefixB == null || tablePrefixB.endsWith("_")) ? tablePrefixB : tablePrefixB + "_";
if (tablePrefixA != null) {
for (TableInfo tableInfo : tableInfosA) {
tableInfo.setNamePrefix(tablePrefixA);
}
}
if (tablePrefixB != null) {
for (TableInfo tableInfo : tableInfosB) {
tableInfo.setNamePrefix(tablePrefixB);
}
}
if (tablePrefixA != null || tablePrefixB != null) {
String aAndB = (tablePrefixA == null) ? "" : tablePrefixA;
aAndB = (tablePrefixB == null) ? aAndB : aAndB + tablePrefixB;
for (TableInfo tableInfo : tableInfosAandB) {
tableInfo.setNamePrefix(aAndB);
}
}
}