IMPORTING DELIMITED ASCII (GENERIC)
If a file to be imported is recognized as containing ASCII but is not
recognized as XML, you are always given an opportunity to select the
Generic CSV (character separated values) import strategy. This
strategy supports the importing of a wide variety of files that employ
a separator character for fields on a single line, and where each line
represents an event. However, if other, more specific, importing
strategies are offered, you should consider whether they are more
appropriate.
In keeping with RFC 4180 (the standard for "comma separated
values"), it is possible to quote both line continuations and the
separator character. The support exceeds that in RFC 4180 in several
ways (for example, the separator does not have to be a comma), but
follows the standard in supporting only ASCII (not full Unicode)
characters.
To use the generic CSV strategy effectively, the following must be
true of the file.
- Either every line must represent the same event with a fixed event
type or there must be a field at fixed column position that represents
an event type name.
- Either there must be a header line, appearing before any data
lines (after some optional skipped lines), that names all the columns,
or you must accept that the event attributes will have uninformative
names like "1", "2", "3", etc.
- Every event must have a unique attribute (after the previous two
steps) that may be interpreted as a integer timestamp (of any
reasonable precision).
If you select this strategy, you are then asked to choose the
following:
- Number of lines to skip at the start of the file.
- The separator character. To specify a non-printable character,
specify its ASCII code as a decimal number of at least two digits:
e.g., use 09 to represent a Tab.
- The quote character (same conventions).
- A fixed event name to use for every event (blank if you use the following option).
- The label of a column that contains the event name (blank if use
the previous option).
- The label of a column that contains the timestamp.
- The ticks-per-second for the timestamp.
- The column labelling strategy. Either "1", "2" etc will be used
or the importer will read a header line from the file (after skipping
the number of lines designated).
After import, every
apparently numeric attribute of every event type (other than the
timestamp attribute) becomes a sample stream that can be manipulated
in TuningFork. The importer can be fooled by attributes that appear
numeric but turn out not to be later in the import. This property may
cause occasional problems with this strategy.