Reference

Returned by: Extract.source, JinkoClient.create_reference_from_doi, JinkoClient.create_reference_from_pdf, JinkoClient.get_reference, JinkoClient.iter_references, JinkoClient.list_references

Also has every member of ProjectItem.

Member	Kind	Description
`bibliography`	property	Return bibliography metadata extracted from this reference.
`doi`	property	Return this reference's DOI, if available.
`file`	property	Return the File object associated with this reference.
`file_content`	property	Return the content of the file associated with this reference. Equivalent to `Reference.file.content()`.
`list_extracts`	method	List extracts belonging to this reference.
`iter_extracts`	method	Iterate over all extracts belonging to this reference.
`create_extract`	method	Create an extract from this reference.
`create_extract_from_pdf_quote`	method	Locate a quote in this reference PDF and create an Extract.

bibliography

Type: dict[str, Any]

Return bibliography metadata extracted from this reference.

doi

Type: str | None

Return this reference's DOI, if available.

file

Type: File

Return the File object associated with this reference.

file_content

Type: bytes

Return the content of the file associated with this reference. Equivalent to Reference.file.content().

list_extracts

list_extracts(
    *,
    limit: int | None = None,
    after: str | None = None,
    kind: Literal["Textual", "Rectangular", "Latex", "DataTable"] | None = None,
    classification_type: (
        Literal["Statement", "Hypothesis", "Data"] | None
    ) = None,
    classification_level: (
        Literal["Weak", "Medium", "Strong", "Excellent"] | None
    ) = None
) -> Page[Extract]

List extracts belonging to this reference.

Use kind or classification filters to narrow the results when needed.

iter_extracts

iter_extracts(
    *,
    kind: Literal["Textual", "Rectangular", "Latex", "DataTable"] | None = None,
    classification_type: (
        Literal["Statement", "Hypothesis", "Data"] | None
    ) = None,
    classification_level: (
        Literal["Weak", "Medium", "Strong", "Excellent"] | None
    ) = None
) -> Iterator[Extract]

Iterate over all extracts belonging to this reference.

This follows pagination automatically and yields every matching extract.

create_extract

create_extract(
    *,
    anchors: Sequence[Mapping[str, Any]],
    text: str | None = None,
    latex: str | None = None,
    mathjs: str | None = None,
    data_table: DataTable | str | None = None,
    color: (
        Literal["Cyan", "Green", "Grey", "Orange", "Pink", "Purple", "Red"]
        | None
    ) = None,
    classification_type: (
        Literal["Statement", "Hypothesis", "Data"] | None
    ) = None,
    classification_level: (
        Literal["Weak", "Medium", "Strong", "Excellent"] | None
    ) = None,
    name: str | None = None,
    folder: Folder | str | None = None,
    description: str | None = None,
    version: str | dict[str, str] | None = None
) -> Extract

Create an extract from this reference.

anchors must be a non-empty sequence of mappings. Each anchor must provide page, x, y, width, and height.

x and y start from the top-left corner of the page. width and height extend rightward and downward from that point.

Multiple anchors are typically used for multi-line text selections. For non-text rectangular extracts, only the first anchor is meaningful.

Provide at most one of text, latex, and data_table. Text creates a textual extract; latex creates a formula extract, with an optional MathJS expression; and data_table creates an extract linked to that data table. When all three are omitted, a rectangular extract is created.

When color is None, the extract color defaults to Green.

When classification_type is provided without a level, the extract is created as unevaluated. Hypothesis classifications do not support the Excellent level.

create_extract_from_pdf_quote

create_extract_from_pdf_quote(
    quote: str,
    *,
    page_hints: int | Sequence[int] | None = None,
    color: (
        Literal["Cyan", "Green", "Grey", "Orange", "Pink", "Purple", "Red"]
        | None
    ) = None,
    classification_type: (
        Literal["Statement", "Hypothesis", "Data"] | None
    ) = None,
    classification_level: (
        Literal["Weak", "Medium", "Strong", "Excellent"] | None
    ) = None,
    kind: Literal["Textual", "Rectangular", "Latex", "DataTable"] | None = None,
    text: str | None = None,
    latex: str | None = None,
    mathjs: str | None = None,
    data_table: DataTable | str | None = None,
    name: str | None = None,
    folder: Folder | str | None = None,
    description: str | None = None,
    version: str | dict[str, str] | None = None
) -> Extract

Locate a quote in this reference PDF and create an Extract.

Matching tolerates common text-encoding and whitespace differences, attempts to match mathematical expressions, and can locate quotes interrupted once by page breaks, tables, or other intervening content. If multiple locations are equally good matches, a ValueError is raised. You can restrict the search to one or more 1-based page numbers with page_hints.

By default, the Extract is textual and uses quote as its text. Provide text to store different text, latex and optional mathjs for a formula Extract, or data_table for a DataTable Extract. Pass kind="Rectangular" to create an Extract without content.

The attached PDF is downloaded, unless the SDK has already cached it.

This feature requires jinko-sdk[pdf] and a searchable native text layer; it does not perform OCR. Other options behave as they do for create_extract.

Reference​

bibliography​

doi​

file​

file_content​

list_extracts​

iter_extracts​

create_extract​

create_extract_from_pdf_quote​