Struct TextDictReader 

pub struct TextDictReader { /* private fields */ }

Sorted dictionary for text ordinal columns.

Zero-copy: the dictionary data is a shared slice of the .fast file. Lazy: the offset table is built on first access (not at load time), avoiding mmap page faults during segment loading.

Implementations

impl TextDictReader

pub fn new_lazy(data: OwnedBytes, count: u32) -> Self

Create a lazy text dictionary from pre-sliced data. No scanning is performed — offsets are built on first get()/ordinal() call.

pub fn open( file_data: &OwnedBytes, dict_start: usize, count: u32, ) -> Result<Self>

Open a zero-copy text dictionary from file_data starting at dict_start. Scans to find the dict end position for slicing, but defers offset building.

pub fn open_from_raw(raw_dict: &OwnedBytes, count: u32) -> Result<Self>

Open from raw dict bytes (already length-prefixed entries).

pub fn get(&self, ordinal: u32) -> Option<&str>

Get string by ordinal — zero-copy borrow from the underlying file data.

pub fn ordinal(&self, text: &str) -> Option<u64>

Binary search for a string → ordinal.

pub fn len(&self) -> u32

Number of entries in the dictionary.

pub fn is_empty(&self) -> bool

Whether the dictionary is empty.

pub fn iter(&self) -> impl Iterator<Item = &str>

Iterate all entries.