Skip to content

Commit

Permalink
Display full names when catching parsing errors
Browse files Browse the repository at this point in the history 
Related to #690.
  • Loading branch information
@dadoonet
dadoonet committed Mar 2, 2019
1 parent 2f69b8c commit 9821622
Show file tree
Hide file tree
Showing 4 changed files with 34 additions and 33 deletions.
4 changes: 2 additions & 2 deletions core/src/main/java/fr/pilato/elasticsearch/crawler/fs/FsParserAbstract.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -36,6 +36,7 @@
import fr.pilato.elasticsearch.crawler.fs.framework.ByteSizeValue;
import fr.pilato.elasticsearch.crawler.fs.framework.SignTool;
import fr.pilato.elasticsearch.crawler.fs.settings.FsSettings;
import fr.pilato.elasticsearch.crawler.fs.tika.TikaDocParser;
import fr.pilato.elasticsearch.crawler.fs.tika.XmlDocParser;
import org.apache.logging.log4j.LogManager;
import org.apache.logging.log4j.Logger;
Expand All @@ -61,7 +62,6 @@
import static fr.pilato.elasticsearch.crawler.fs.framework.FsCrawlerUtil.isFileSizeUnderLimit;
import static fr.pilato.elasticsearch.crawler.fs.framework.FsCrawlerUtil.isIndexable;
import static fr.pilato.elasticsearch.crawler.fs.framework.FsCrawlerUtil.localDateTimeToDate;
import static fr.pilato.elasticsearch.crawler.fs.tika.TikaDocParser.generate;

public abstract class FsParserAbstract extends FsParser {
private static final Logger logger = LogManager.getLogger(FsParserAbstract.class);
Expand Down Expand Up @@ -471,7 +471,7 @@ private void indexFile(FileAbstractModel fileAbstractModel, ScanStatistic stats,
doc.setObject(XmlDocParser.generateMap(inputStream));
} else {
// Extracting content with Tika
generate(fsSettings, inputStream, filename, doc, messageDigest, filesize);
TikaDocParser.generate(fsSettings, inputStream, filename, fullFilename, doc, messageDigest, filesize);
}

// We index the data structure
Expand Down
2 changes: 1 addition & 1 deletion rest/src/main/java/fr/pilato/elasticsearch/crawler/fs/rest/UploadApi.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -105,7 +105,7 @@ public UploadResponse post(
// Path

// Read the file content
TikaDocParser.generate(settings, filecontent, filename, doc, messageDigest, filesize);
TikaDocParser.generate(settings, filecontent, filename, filename, doc, messageDigest, filesize);

String url = null;
if (Boolean.parseBoolean(simulate)) {
Expand Down
60 changes: 30 additions & 30 deletions tika/src/main/java/fr/pilato/elasticsearch/crawler/fs/tika/TikaDocParser.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -52,9 +52,9 @@ public class TikaDocParser {

private final static Logger logger = LogManager.getLogger(TikaDocParser.class);

public static void generate(FsSettings fsSettings, InputStream inputStream, String filename, Doc doc, MessageDigest messageDigest,
long filesize) throws IOException {
logger.trace("Generating document [{}]", filename);
public static void generate(FsSettings fsSettings, InputStream inputStream, String filename, String fullFilename, Doc doc,
MessageDigest messageDigest, long filesize) throws IOException {
logger.trace("Generating document [{}]", fullFilename);
// Extracting content with Tika
// See #38: https://github.com/dadoonet/fscrawler/issues/38
int indexedChars = 100000;
Expand Down Expand Up @@ -102,8 +102,8 @@ public static void generate(FsSettings fsSettings, InputStream inputStream, Stri
current = current.getCause();
}

logger.warn("Failed to extract [" + indexedChars + "] characters of text for [" + filename + "] {}", sb.toString());
logger.debug("Failed to extract [" + indexedChars + "] characters of text for [" + filename + "]", e);
logger.warn("Failed to extract [{}] characters of text for [{}] {}", indexedChars, fullFilename, sb.toString());
logger.debug("Failed to extract [" + indexedChars + "] characters of text for [" + fullFilename + "]", e);
}

// Adding what we found to the document we want to index
Expand Down Expand Up @@ -134,18 +134,18 @@ public static void generate(FsSettings fsSettings, InputStream inputStream, Stri
// File

// Standard Meta
setMeta(filename, metadata, TikaCoreProperties.CREATOR, doc.getMeta()::setAuthor, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.TITLE, doc.getMeta()::setTitle, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.MODIFIED, doc.getMeta()::setDate, FsCrawlerUtil::localDateTimeToDate);
setMeta(filename, metadata, TikaCoreProperties.KEYWORDS, doc.getMeta()::setKeywords, TikaDocParser::commaDelimitedListToStringArray);
setMeta(filename, metadata, TikaCoreProperties.FORMAT, doc.getMeta()::setFormat, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.IDENTIFIER, doc.getMeta()::setIdentifier, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.CONTRIBUTOR, doc.getMeta()::setContributor, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.COVERAGE, doc.getMeta()::setCoverage, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.MODIFIER, doc.getMeta()::setModifier, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.CREATOR_TOOL, doc.getMeta()::setCreatorTool, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.CREATOR, doc.getMeta()::setAuthor, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.TITLE, doc.getMeta()::setTitle, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.MODIFIED, doc.getMeta()::setDate, FsCrawlerUtil::localDateTimeToDate);
setMeta(fullFilename, metadata, TikaCoreProperties.KEYWORDS, doc.getMeta()::setKeywords, TikaDocParser::commaDelimitedListToStringArray);
setMeta(fullFilename, metadata, TikaCoreProperties.FORMAT, doc.getMeta()::setFormat, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.IDENTIFIER, doc.getMeta()::setIdentifier, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.CONTRIBUTOR, doc.getMeta()::setContributor, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.COVERAGE, doc.getMeta()::setCoverage, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.MODIFIER, doc.getMeta()::setModifier, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.CREATOR_TOOL, doc.getMeta()::setCreatorTool, Function.identity());
String finalParsedContent = parsedContent;
setMeta(filename, metadata, TikaCoreProperties.LANGUAGE, doc.getMeta()::setLanguage, (lang) -> {
setMeta(fullFilename, metadata, TikaCoreProperties.LANGUAGE, doc.getMeta()::setLanguage, (lang) -> {
if (lang != null) {
return lang;
} else if (fsSettings.getFs().isLangDetect() && finalParsedContent != null) {
Expand All @@ -158,20 +158,20 @@ public static void generate(FsSettings fsSettings, InputStream inputStream, Stri
}
return null;
});
setMeta(filename, metadata, TikaCoreProperties.PUBLISHER, doc.getMeta()::setPublisher, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.RELATION, doc.getMeta()::setRelation, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.RIGHTS, doc.getMeta()::setRights, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.SOURCE, doc.getMeta()::setSource, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.TYPE, doc.getMeta()::setType, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.DESCRIPTION, doc.getMeta()::setDescription, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.CREATED, doc.getMeta()::setCreated, FsCrawlerUtil::localDateTimeToDate);
setMeta(filename, metadata, TikaCoreProperties.PRINT_DATE, doc.getMeta()::setPrintDate, FsCrawlerUtil::localDateTimeToDate);
setMeta(filename, metadata, TikaCoreProperties.METADATA_DATE, doc.getMeta()::setMetadataDate, FsCrawlerUtil::localDateTimeToDate);
setMeta(filename, metadata, TikaCoreProperties.LATITUDE, doc.getMeta()::setLatitude, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.LONGITUDE, doc.getMeta()::setLongitude, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.ALTITUDE, doc.getMeta()::setAltitude, Function.identity());
setMeta(filename, metadata, TikaCoreProperties.RATING, doc.getMeta()::setRating, (value) -> value == null ? null : Integer.parseInt(value));
setMeta(filename, metadata, TikaCoreProperties.COMMENTS, doc.getMeta()::setComments, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.PUBLISHER, doc.getMeta()::setPublisher, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.RELATION, doc.getMeta()::setRelation, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.RIGHTS, doc.getMeta()::setRights, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.SOURCE, doc.getMeta()::setSource, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.TYPE, doc.getMeta()::setType, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.DESCRIPTION, doc.getMeta()::setDescription, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.CREATED, doc.getMeta()::setCreated, FsCrawlerUtil::localDateTimeToDate);
setMeta(fullFilename, metadata, TikaCoreProperties.PRINT_DATE, doc.getMeta()::setPrintDate, FsCrawlerUtil::localDateTimeToDate);
setMeta(fullFilename, metadata, TikaCoreProperties.METADATA_DATE, doc.getMeta()::setMetadataDate, FsCrawlerUtil::localDateTimeToDate);
setMeta(fullFilename, metadata, TikaCoreProperties.LATITUDE, doc.getMeta()::setLatitude, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.LONGITUDE, doc.getMeta()::setLongitude, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.ALTITUDE, doc.getMeta()::setAltitude, Function.identity());
setMeta(fullFilename, metadata, TikaCoreProperties.RATING, doc.getMeta()::setRating, (value) -> value == null ? null : Integer.parseInt(value));
setMeta(fullFilename, metadata, TikaCoreProperties.COMMENTS, doc.getMeta()::setComments, Function.identity());

// Add support for more OOTB standard metadata

Expand Down
1 change: 1 addition & 0 deletions tika/src/test/java/fr/pilato/elasticsearch/crawler/fs/tika/TikaDocParserTest.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -745,6 +745,7 @@ private Doc extractFromFile(String filename, FsSettings fsSettings) throws IOExc
fsSettings,
data,
filename,
"/documents/" + filename,
doc,
messageDigest,
0);
Expand Down

0 comments on commit 9821622

Please sign in to comment.