[PARQUET-485] Decouple data page delimiting from column reader / scanner classes, create test fixtures

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: cpp-1.0.0
• Component/s: parquet-cpp
• Assignee: Wes McKinney
• Reporter: Wes McKinney
• Created: 30/Jan/16 15:49
• Updated: 2 days ago 03:28
• Resolved: 02/Feb/16 18:50

Description

It is difficult to test the column reader classes with mock data because the data page resolution is tightly coupled to the actual file format layout in ColumnReader::ReadNewPage.

I plan to separate these concerns, so that the column readers can be tested with a sequence of data pages encoded in memory, but never actually assembled into a file stream layout with thrift-serialized page headers. Patch forthcoming.