pub struct DictionaryTracker { /* private fields */ }
Expand description
Keeps track of dictionaries that have been written, to avoid emitting the same dictionary multiple times.
Can optionally error if an update to an existing dictionary is attempted, which
isn’t allowed in the FileWriter
.
Implementations§
Source§impl DictionaryTracker
impl DictionaryTracker
Sourcepub fn new(error_on_replacement: bool) -> Self
pub fn new(error_on_replacement: bool) -> Self
Create a new DictionaryTracker
.
If error_on_replacement
is true, an error will be generated if an update to an
existing dictionary is attempted.
If preserve_dict_id
is true, the dictionary ID defined in the schema
is used, otherwise a unique dictionary ID will be assigned by incrementing
the last seen dictionary ID (or using 0
if no other dictionary IDs have been
seen)
Sourcepub fn new_with_preserve_dict_id(
error_on_replacement: bool,
preserve_dict_id: bool,
) -> Self
pub fn new_with_preserve_dict_id( error_on_replacement: bool, preserve_dict_id: bool, ) -> Self
Create a new DictionaryTracker
.
If error_on_replacement
is true, an error will be generated if an update to an
existing dictionary is attempted.
Sourcepub fn set_dict_id(&mut self, field: &Field) -> i64
pub fn set_dict_id(&mut self, field: &Field) -> i64
Set the dictionary ID for field
.
If preserve_dict_id
is true, this will return the dict_id
in field
(or panic if field
does
not have a dict_id
defined).
If preserve_dict_id
is false, this will return the value of the last dict_id
assigned incremented by 1
or 0 in the case where no dictionary IDs have yet been assigned
Sourcepub fn dict_id(&mut self) -> &[i64]
pub fn dict_id(&mut self) -> &[i64]
Return the sequence of dictionary IDs in the order they should be observed while traversing the schema
Sourcepub fn insert(
&mut self,
dict_id: i64,
column: &ArrayRef,
) -> Result<bool, ArrowError>
pub fn insert( &mut self, dict_id: i64, column: &ArrayRef, ) -> Result<bool, ArrowError>
Keep track of the dictionary with the given ID and values. Behavior:
- If this ID has been written already and has the same data, return
Ok(false)
to indicate that the dictionary was not actually inserted (because it’s already been seen). - If this ID has been written already but with different data, and this tracker is configured to return an error, return an error.
- If the tracker has not been configured to error on replacement or this dictionary
has never been seen before, return
Ok(true)
to indicate that the dictionary was just inserted.