Mark/pile
1
0
Fork 0
Code Actions Releases Activity

Add ListExtractor

Browse Source
This commit is contained in:
rm-dr
2026-03-10 20:24:56 -07:00
parent 280bbcb83e
commit 48ac93c78e
22 changed files with 386 additions and 93 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
crates/pile-dataset/src/extract/epub/epub_meta.rs
View File

@@ -3,7 +3,7 @@ use pile_config::Label;
use std::{collections::HashMap, sync::OnceLock};
use tracing::trace;
use crate::{Item, PileValue, SyncReadBridge, extract::Extractor};
use crate::{Item, PileValue, SyncReadBridge, extract::ObjectExtractor};
pub struct EpubMetaExtractor<'a> {
item: &'a Item,
@@ -78,7 +78,7 @@ impl<'a> EpubMetaExtractor<'a> {
}
#[async_trait::async_trait]
impl Extractor for EpubMetaExtractor<'_> {
impl ObjectExtractor for EpubMetaExtractor<'_> {
async fn field<'a>(
&'a self,
name: &Label,

4
crates/pile-dataset/src/extract/epub/epub_text.rs
View File

@@ -3,7 +3,7 @@ use pile_config::Label;
use std::{collections::HashMap, sync::OnceLock};
use tracing::debug;
use crate::{Item, PileValue, SyncReadBridge, extract::Extractor};
use crate::{Item, PileValue, SyncReadBridge, extract::ObjectExtractor};
pub struct EpubTextExtractor<'a> {
item: &'a Item,
@@ -88,7 +88,7 @@ fn strip_html(html: &str) -> String {
}
#[async_trait::async_trait]
impl Extractor for EpubTextExtractor<'_> {
impl ObjectExtractor for EpubTextExtractor<'_> {
async fn field<'a>(
&'a self,
name: &Label,

10
crates/pile-dataset/src/extract/epub/mod.rs
View File

@@ -9,7 +9,7 @@ pub use epub_text::*;
use crate::{
Item, PileValue,
extract::{Extractor, MapExtractor},
extract::{MapExtractor, ObjectExtractor},
};
pub struct EpubExtractor<'a> {
@@ -23,11 +23,11 @@ impl<'a> EpubExtractor<'a> {
inner: HashMap::from([
(
Label::new("text").unwrap(),
PileValue::Extractor(Arc::new(EpubTextExtractor::new(item))),
PileValue::ObjectExtractor(Arc::new(EpubTextExtractor::new(item))),
),
(
Label::new("meta").unwrap(),
PileValue::Extractor(Arc::new(EpubMetaExtractor::new(item))),
PileValue::ObjectExtractor(Arc::new(EpubMetaExtractor::new(item))),
),
]),
};
@@ -37,7 +37,7 @@ impl<'a> EpubExtractor<'a> {
}
#[async_trait::async_trait]
impl Extractor for EpubExtractor<'_> {
impl ObjectExtractor for EpubExtractor<'_> {
async fn field<'a>(
&'a self,
name: &pile_config::Label,
@@ -45,7 +45,7 @@ impl Extractor for EpubExtractor<'_> {
#[expect(clippy::unwrap_used)]
if name.as_str() == "text" {
match self.inner.inner.get(name).unwrap() {
PileValue::Extractor(x) => return x.field(name).await,
PileValue::ObjectExtractor(x) => return x.field(name).await,
_ => unreachable!(),
};
}