[sqlite] Fix bug in how recover virtual table handles empty.

third_party/sqlite/src/src/recover.c

Index: third_party/sqlite/src/src/recover.c
diff --git a/third_party/sqlite/src/src/recover.c b/third_party/sqlite/src/src/recover.c
index 8e3929d453be4db35ebe8468185e54905485f667..58dd89e1f2f6f9011b028137acccd5a27e7525b8 100644
--- a/third_party/sqlite/src/src/recover.c
+++ b/third_party/sqlite/src/src/recover.c
@@ -1138,6 +1138,7 @@ struct RecoverLeafCursor {
  */
 static int leafCursorLoadPage(RecoverLeafCursor *pCursor, DbPage *pPage){
   const unsigned char *pPageHeader;  /* Header of *pPage */
+  unsigned nCells;                   /* Number of cells in the page */
 
   /* Release the current page. */
   if( pCursor->pPage ){
@@ -1167,10 +1168,17 @@ static int leafCursorLoadPage(RecoverLeafCursor *pCursor, DbPage *pPage){
     return SQLITE_OK;
   }
 
+  /* Leaf contains no data, skip it.  Empty tables, for instance. */
+  nCells = decodeUnsigned16(pPageHeader + kiPageCellCountOffset);;
+  if( nCells<1 ){

[Ryan Hamilton, 2016/02/05 17:51:25]

The style... it burns my eyes!

[Scott Hess - ex-Googler, 2016/02/05 18:18:05]

Yeah!  It's essentially opposite Google/Chromium style :-).  And
psuedo-Hungrarian notation is always a joy.

+    sqlite3PagerUnref(pPage);
+    return SQLITE_OK;
+  }
+
   /* Take ownership of the page and start decoding. */
   pCursor->pPage = pPage;
   pCursor->iCell = 0;
-  pCursor->nCells = decodeUnsigned16(pPageHeader + kiPageCellCountOffset);
+  pCursor->nCells = nCells;
   return SQLITE_OK;
 }

[Ryan Hamilton, 2016/02/05 17:51:25]

This diff appears to be basically the same as the diff to the previous file. Is
one an auto-generated version of the other, or some such?

[Scott Hess - ex-Googler, 2016/02/05 18:18:05]

Oh, sorry, should have mentioned that, too.

In third_party/sqlite, the actual source data is in src/src, and there's a
SQLite-provided script which merges everything into amalgamation/sqlite3.c. 
It's a primitive approach to full-program optimization, and the standard source
SQLite ships.

Additionally, we keep patch files, to keep track of our actual changes.  This is
because we can't have individual patches as CLs, because it would break the
waterfall, but if we squash them then everything is all mixed together.  I
didn't regen the patch files in this case, because I plan to make other
recover.c changes.