Note: 0x96 is not legal utf8, and the file is marked as utf8 - which is causing the problem.

Sometimes it would have been substituted with a substitution character, but I found at least one case where an invalid utf8 byte would lead to a blank string.  I will release a fix - it will make it clearer when there is an issue - but the files still need cleaning/encoding properly.

You probably need to replace 0x96 with 0x2D (hyphen minus) as part of the cleaning.

The first two blank lines (line 187 and 200) both correspond to lines in the file that contain a character hex 0x96. 

That corresponds to an EN dash in the windows-1252 code pages, but is not defined in the iso-8859-1 code page that we use for string representations.  I suspect the conversion from string to utf8 fails for some reason because of that character.  I'll debug locally.

@Gavin Halliday Not sure the following information would help for debugging the issue: I desprayed the logical file to a landing zone. I sprayed the file back and got another logical file with the same size as the original logical file. But, I can see the missing data rows in the second logical file.

@Gordon Smith @Kunal Aswani @Jeremy Clements