Return docid in case of errors while rendering the document template

This commit is contained in:
Louis Dureuil 2024-12-11 15:26:18 +01:00
parent e974be9518
commit bfca54cc2c
No known key found for this signature in database
3 changed files with 25 additions and 3 deletions

View File

@ -38,6 +38,16 @@ pub struct RenderPromptError {
pub fault: FaultSource,
}
impl RenderPromptError {
pub(crate) fn missing_context_with_external_docid(
external_docid: String,
inner: liquid::Error,
) -> RenderPromptError {
Self {
kind: RenderPromptErrorKind::MissingContextWithExternalDocid(external_docid, inner),
fault: FaultSource::User,
}
}
pub(crate) fn missing_context(inner: liquid::Error) -> RenderPromptError {
Self { kind: RenderPromptErrorKind::MissingContext(inner), fault: FaultSource::User }
}
@ -47,6 +57,8 @@ impl RenderPromptError {
pub enum RenderPromptErrorKind {
#[error("missing field in document: {0}")]
MissingContext(liquid::Error),
#[error("missing field in document `{0}`: {1}")]
MissingContextWithExternalDocid(String, liquid::Error),
}
impl From<RenderPromptError> for crate::Error {

View File

@ -119,6 +119,7 @@ impl Prompt {
'doc: 'a, // lifetime of the allocator, will live for an entire chunk of documents
>(
&self,
external_docid: &str,
document: impl crate::update::new::document::Document<'a> + Debug,
field_id_map: &RefCell<GlobalFieldsIdsMap>,
doc_alloc: &'doc Bump,
@ -130,9 +131,12 @@ impl Prompt {
self.max_bytes.unwrap_or_else(default_max_bytes).get(),
doc_alloc,
);
self.template
.render_to(&mut rendered, &context)
.map_err(RenderPromptError::missing_context)?;
self.template.render_to(&mut rendered, &context).map_err(|liquid_error| {
RenderPromptError::missing_context_with_external_docid(
external_docid.to_owned(),
liquid_error,
)
})?;
Ok(std::str::from_utf8(rendered.into_bump_slice())
.expect("render can only write UTF-8 because all inputs and processing preserve utf-8"))
}

View File

@ -130,6 +130,7 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
);
} else if new_vectors.regenerate {
let new_rendered = prompt.render_document(
update.external_document_id(),
update.current(
&context.rtxn,
context.index,
@ -139,6 +140,7 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
&context.doc_alloc,
)?;
let old_rendered = prompt.render_document(
update.external_document_id(),
update.merged(
&context.rtxn,
context.index,
@ -158,6 +160,7 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
}
} else if old_vectors.regenerate {
let old_rendered = prompt.render_document(
update.external_document_id(),
update.current(
&context.rtxn,
context.index,
@ -167,6 +170,7 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
&context.doc_alloc,
)?;
let new_rendered = prompt.render_document(
update.external_document_id(),
update.merged(
&context.rtxn,
context.index,
@ -216,6 +220,7 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
);
} else if new_vectors.regenerate {
let rendered = prompt.render_document(
insertion.external_document_id(),
insertion.inserted(),
context.new_fields_ids_map,
&context.doc_alloc,
@ -229,6 +234,7 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
}
} else {
let rendered = prompt.render_document(
insertion.external_document_id(),
insertion.inserted(),
context.new_fields_ids_map,
&context.doc_alloc,