genpostings.patch

Index: src/sphinx.cpp
===================================================================
--- src/sphinx.cpp	(revision 3296)
+++ src/sphinx.cpp	(working copy)
@@ -9079,6 +9079,9 @@
 }
 
 
+FILE * g_fpPost = NULL;
+CSphVector<int> g_Docids;
+
 void CSphHitBuilder::cidxHit ( CSphAggregateHit * pHit, const CSphRowitem * pAttrs )
 {
 	assert (
@@ -9093,6 +9096,19 @@
 		( m_pDict->GetSettings().m_bWordDict && strcmp ( (char*)m_tLastHit.m_sKeyword, (char*)pHit->m_sKeyword ) ) ); // OPTIMIZE?
 	bool bNextDoc = bNextWord || ( m_tLastHit.m_iDocID!=pHit->m_iDocID );
 
+	if ( pHit->m_iWordID )
+	{
+		fwrite ( &bNextWord, 1, 1, g_fpPost );
+		if ( bNextWord )
+		{
+			BYTE uLen = strlen((char*)pHit->m_sKeyword);
+			fwrite ( &uLen, 1, 1, g_fpPost );
+			fwrite ( pHit->m_sKeyword, 1, uLen, g_fpPost );
+		}
+		fwrite ( &pHit->m_iDocID, 1, 4, g_fpPost );
+		fwrite ( &pHit->m_iWordPos, 1, 4, g_fpPost );
+	}
+
 	if ( bNextDoc )
 	{
 		// finish hitlist, if any
@@ -10631,6 +10647,10 @@
 
 int CSphIndex_VLN::Build ( const CSphVector<CSphSource*> & dSources, int iMemoryLimit, int iWriteBuffer )
 {
+	g_fpPost = fopen("postings.bin", "wb+");
+	if ( !g_fpPost )
+		sphDie("failed to write postings");
+
 	PROFILER_INIT ();
 
 	assert ( dSources.GetLength() );
@@ -10976,6 +10996,8 @@
 			if ( !pSource->m_tDocInfo.m_iDocID )
 				break;
 
+			g_Docids.Add ( pSource->m_tDocInfo.m_iDocID );
+
 			if ( bGotJoined )
 				dAllIds.Add ( pSource->m_tDocInfo.m_iDocID );
 
@@ -11992,6 +12014,15 @@
 	ARRAY_FOREACH ( i, dSources )
 		dSources[i]->PostIndex ();
 
+	fclose(g_fpPost);
+
+	ARRAY_FOREACH(i, g_Docids)
+		g_Docids[i] -= m_iMinDocid; // adjust to match what cidxHit() logged
+
+	FILE * fp = fopen("docids.bin", "wb+");
+	fwrite(&g_Docids[0], 1, sizeof(int)*g_Docids.GetLength(), fp);
+	fclose(fp);
+
 	PROFILER_DONE ();
 	PROFILE_SHOW ();
 	dFileWatchdog.AllIsDone();