Fuzzy matching against csv header#62
Conversation
|
To test run ./fuzzy_load.sh |
| catch (IllegalArgumentException e){ | ||
| System.err.println("Could not connect to the cluster, check your hosts"); | ||
| //e.printStackTrace(); | ||
| System.exit(0); |
There was a problem hiding this comment.
System.exit should really be avoided, why not throw an IOException like on line 461?
Same thing below
| if (boolFalse.equalsIgnoreCase(toparse)) | ||
| return new Boolean("FALSE"); | ||
| throw new ParseException("Boolean was not TRUE (" + boolTrue + ") or FALSE (" + boolFalse + ")", 0); | ||
| throw new ParseException("Boolean was not TRUE (" + boolTrue + ") or FALSE (" + boolFalse + ")" + " it was "+ toparse, 0); |
There was a problem hiding this comment.
nit: would read better with some punctuation, ex:
+ ". It was "+ toparse
| List<String> columnNames = cdp.getColumnNames(); | ||
| try { | ||
| String header = reader.readLine(); | ||
| //this didn' work |
| } | ||
| resultColumns[i]=currentColumn; | ||
| i++; | ||
| columnNames.remove(columnNames.contains(currentColumn)); |
There was a problem hiding this comment.
columnNames.contains(...) returns a boolean; passing any boolean to columnNames.remove(...) will have no effect
Perhaps this was meant to remove currentColumn from columnNames if it exists in the list. In this case it is fine to just call columnNames.remove(currentColumn)
| if (delimiter == null){ | ||
| delimiter = ","; | ||
| } | ||
| List<String> headerList = Arrays.asList(header.split(delimiter)); |
There was a problem hiding this comment.
When performing a basic split, it won't account for the various CSV edge cases. Would be more robust to use a CSV parser to do the split (like Univocity, used elsewhere in this project)
2 participants
No description provided.