From: <bra...@us...> - 2008-06-05 20:35:00
|
Revision: 2280 http://archive-access.svn.sourceforge.net/archive-access/?rev=2280&view=rev Author: bradtofel Date: 2008-06-05 13:34:57 -0700 (Thu, 05 Jun 2008) Log Message: ----------- FEATURE: added method to return iterator from a pathOrUrl (String) Modified Paths: -------------- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/ArcIndexer.java trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/WarcIndexer.java Modified: trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/ArcIndexer.java =================================================================== --- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/ArcIndexer.java 2008-06-04 00:08:01 UTC (rev 2279) +++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/ArcIndexer.java 2008-06-05 20:34:57 UTC (rev 2280) @@ -59,7 +59,7 @@ public ArcIndexer() { canonicalizer = new AggressiveUrlCanonicalizer(); } - + /** * @param arc * @return Iterator of SearchResults for input arc File @@ -67,7 +67,26 @@ */ public CloseableIterator<SearchResult> iterator(File arc) throws IOException { - ARCReader arcReader = ARCReaderFactory.get(arc); + return iterator(ARCReaderFactory.get(arc)); + } + + /** + * @param pathOrUrl + * @return Iterator of SearchResults for input pathOrUrl + * @throws IOException + */ + public CloseableIterator<SearchResult> iterator(String pathOrUrl) + throws IOException { + return iterator(ARCReaderFactory.get(pathOrUrl)); + } + + /** + * @param arcReader + * @return Iterator of SearchResults for input ARCReader + * @throws IOException + */ + public CloseableIterator<SearchResult> iterator(ARCReader arcReader) + throws IOException { arcReader.setParseHttpHeaders(true); Adapter<ArchiveRecord,ARCRecord> adapter1 = Modified: trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/WarcIndexer.java =================================================================== --- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/WarcIndexer.java 2008-06-04 00:08:01 UTC (rev 2279) +++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/WarcIndexer.java 2008-06-05 20:34:57 UTC (rev 2280) @@ -31,21 +31,37 @@ } /** - * @param arc + * @param warc * @return Iterator of SearchResults for input arc File * @throws IOException */ public CloseableIterator<SearchResult> iterator(File warc) throws IOException { + return iterator(WARCReaderFactory.get(warc)); + } + /** + * @param pathOrUrl + * @return Iterator of SearchResults for input pathOrUrl + * @throws IOException + */ + public CloseableIterator<SearchResult> iterator(String pathOrUrl) + throws IOException { + return iterator(WARCReaderFactory.get(pathOrUrl)); + } + /** + * @param arc + * @return Iterator of SearchResults for input arc File + * @throws IOException + */ + public CloseableIterator<SearchResult> iterator(WARCReader reader) + throws IOException { Adapter<ArchiveRecord, WARCRecord> adapter1 = new ArchiveRecordToWARCRecordAdapter(); WARCRecordToSearchResultAdapter adapter2 = new WARCRecordToSearchResultAdapter(); adapter2.setCanonicalizer(canonicalizer); - - WARCReader reader = WARCReaderFactory.get(warc); - + ArchiveReaderCloseableIterator itr1 = new ArchiveReaderCloseableIterator(reader,reader.iterator()); This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site. |