[HIVE-7730] Extend ReadEntity to add accessed columns from query

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.14.0
• Assignee: Xiaomeng Huang Assign to me
• Reporter: Xiaomeng Huang
• Created: 14/Aug/14 09:03
• Updated: 13/Nov/14 19:39
• Resolved: 02/Sep/14 00:43

Description

Now what we get from HiveSemanticAnalyzerHookContextImpl is limited. If we have hook of HiveSemanticAnalyzerHook, we may want to get more things from hookContext. (e.g. the needed colums from query).
So we should get instance of HiveSemanticAnalyzerHookContext from configuration, extends HiveSemanticAnalyzerHookContext with a new implementation, overide the HiveSemanticAnalyzerHookContext.update() and put what you want to the class.
Hive should store accessed columns to ReadEntity when we set HIVE_STATS_COLLECT_SCANCOLS(or we can add a confVar) is true.
Then external authorization model can get accessed columns when do authorization in compile before execute. Maybe we will remove columnAccessInfo from BaseSemanticAnalyzer, old authorization and AuthorizationModeV2 can get accessed columns from ReadEntity too.
Here is the quick implement in SemanticAnalyzer.analyzeInternal() below:

   boolean isColumnInfoNeedForAuth = SessionState.get().isAuthorizationModeV2()
        && HiveConf.getBoolVar(conf, HiveConf.ConfVars.HIVE_AUTHORIZATION_ENABLED);
    if (isColumnInfoNeedForAuth
        || HiveConf.getBoolVar(this.conf, HiveConf.ConfVars.HIVE_STATS_COLLECT_SCANCOLS) == true) {
      ColumnAccessAnalyzer columnAccessAnalyzer = new ColumnAccessAnalyzer(pCtx);
      setColumnAccessInfo(columnAccessAnalyzer.analyzeColumnAccess()); 
    }
    compiler.compile(pCtx, rootTasks, inputs, outputs);
    // TODO: 
    // after compile, we can put accessed column list to ReadEntity getting from columnAccessInfo if HIVE_AUTHORIZATION_ENABLED is set true

Issue Links

• blocks: SENTRY-392 Authorization for column level security
• blocks: SENTRY-74 Add column-level privileges for Hive/Impala
• is related to: HIVE-7932 It may cause NP exception when add accessed columns to ReadEntity
• links to: review board for HIVE-7730