Added the latex serializer

Signed-off-by: Peter Staar <[email protected]>

Author: PeterStaar-IBM

docling_core/transforms/serializer/latex.py

@@ -4607,6 +4607,46 @@ def save_as_html(
         with open(filename, "w", encoding="utf-8") as fw:
             fw.write(html_out)
 
+    def save_as_latex(
+        self,
+        filename: Union[str, Path],
+        artifacts_dir: Optional[Path] = None,
+        from_element: int = 0,
+        to_element: int = sys.maxsize,
+        labels: Optional[set[DocItemLabel]] = None,
+        image_mode: ImageRefMode = ImageRefMode.PLACEHOLDER,
+        page_no: Optional[int] = None,
+        included_content_layers: Optional[set[ContentLayer]] = None,
+        include_annotations: bool = True,
+        page_break_command: Optional[str] = None,
+    ):
+        """Save to LaTeX."""
+        if isinstance(filename, str):
+            filename = Path(filename)
+
+        artifacts_dir, reference_path = self._get_output_paths(filename, artifacts_dir)
+
+        if image_mode == ImageRefMode.REFERENCED:
+            os.makedirs(artifacts_dir, exist_ok=True)
+
+        new_doc = self._make_copy_with_refmode(
+            artifacts_dir, image_mode, page_no, reference_path=reference_path
+        )
+
+        latex_out = new_doc.export_to_latex(
+            from_element=from_element,
+            to_element=to_element,
+            labels=labels,
+            image_mode=image_mode,
+            page_no=page_no,
+            included_content_layers=included_content_layers,
+            include_annotations=include_annotations,
+            page_break_command=page_break_command,
+        )
+
+        with open(filename, "w", encoding="utf-8") as fw:
+            fw.write(latex_out)
+
     def _get_output_paths(
         self, filename: Union[str, Path], artifacts_dir: Optional[Path] = None
     ) -> Tuple[Path, Optional[Path]]:
@@ -4703,6 +4743,57 @@ def export_to_html(  # noqa: C901
 
         return ser_res.text
 
+    def export_to_latex(
+        self,
+        from_element: int = 0,
+        to_element: int = sys.maxsize,
+        labels: Optional[set[DocItemLabel]] = None,
+        enable_chart_tables: bool = True,
+        image_mode: ImageRefMode = ImageRefMode.PLACEHOLDER,
+        page_no: Optional[int] = None,
+        included_content_layers: Optional[set[ContentLayer]] = None,
+        include_annotations: bool = True,
+        page_break_command: Optional[str] = None,
+        indent: int = 2,
+        image_placeholder: str = "% image",
+        escape_latex: bool = True,
+    ) -> str:
+        """Serialize to LaTeX."""
+        from docling_core.transforms.serializer.latex import (
+            LaTeXDocSerializer,
+            LaTeXParams,
+        )
+
+        my_labels = labels if labels is not None else DOCUMENT_TOKENS_EXPORT_LABELS
+        my_layers = (
+            included_content_layers
+            if included_content_layers is not None
+            else DEFAULT_CONTENT_LAYERS
+        )
+
+        params = LaTeXParams(
+            labels=my_labels,
+            layers=my_layers,
+            pages={page_no} if page_no is not None else None,
+            start_idx=from_element,
+            stop_idx=to_element,
+            image_mode=image_mode,
+            enable_chart_tables=enable_chart_tables,
+            include_annotations=include_annotations,
+            page_break_command=page_break_command,
+            indent=indent,
+            image_placeholder=image_placeholder,
+            escape_latex=escape_latex,
+        )
+
+        serializer = LaTeXDocSerializer(
+            doc=self,
+            params=params,
+        )
+        ser_res = serializer.serialize()
+
+        return ser_res.text
+
     @staticmethod
     def load_from_doctags(  # noqa: C901
         doctag_document: DocTagsDocument, document_name: str = "Document"

docling_core/types/doc/document.py

@@ -0,0 +1,43 @@
+\subsection{Summer activities}
+
+\subsection{Swimming in the lake}
+
+Duck
+
+\begin{figure}[h]
+% image
+\caption{Figure 1: This is a cute duckling}
+\end{figure}
+
+\subsection{Let's swim!}
+
+To get started with swimming, first lay down in a water and try not to drown:
+
+\begin{itemize}
+\item ∞ You can relax and look around
+\item ∞ Paddle about
+\item ∞ Enjoy summer warmth
+\end{itemize}
+
+Also, don't forget:
+
+\begin{itemize}
+\item 1. Wear sunglasses
+\item 2. Don't forget to drink water
+\item 3. Use sun cream
+\end{itemize}
+
+Hmm, what else…
+
+\begin{itemize}
+\item -Another activity item
+\item -Yet another one
+\item -Stopping it here
+\end{itemize}
+
+Some text.
+
+\begin{itemize}
+\item -Starting the next page with a list item.
+\item -Second item.
+\end{itemize}

test/data/doc/2408.09869v3_enriched.gt.tex

@@ -0,0 +1,109 @@
+\begin{itemize}
+\item item of leading list
+\end{itemize}
+
+\section*{Title of the Document}
+
+Author 1
+Affiliation 1
+
+Author 2
+Affiliation 2
+
+\subsection{1. Introduction}
+
+This paper introduces the biggest invention ever made. ...
+
+\begin{itemize}
+\item list item 1
+\item list item 2
+\item list item 3
+  \begin{enumerate}
+\item list item 3.a
+\item list item 3.b
+\item list item 3.c
+    \begin{enumerate}
+\item list item 3.c.i
+    \end{enumerate}
+  \end{enumerate}
+\item list item 4
+\end{itemize}
+
+\begin{table}[h]
+\caption{This is the caption of table 1.}
+\begin{tabular}{|l|l|l|}
+\hline
+Product & Years & Years \\ \hline
+Product & 2016 & 2017 \\ \hline
+Apple & 49823 & 695944 \\ \hline
+\end{tabular}
+\end{table}
+
+\begin{figure}[h]
+% image
+\caption{This is the caption of figure 1.}
+\end{figure}
+
+\begin{figure}[h]
+% image
+\caption{This is the caption of figure 2.}
+\end{figure}
+
+\begin{itemize}
+\item item 1 of list
+\end{itemize}
+
+\begin{itemize}
+\item item 1 of list after empty list
+\item item 2 of list after empty list
+\end{itemize}
+
+\begin{itemize}
+\item item 1 of neighboring list
+\item item 2 of neighboring list
+  \begin{itemize}
+\item item 1 of sub list
+\item Here a code snippet: \texttt{print("Hello world")} (to be displayed inline)
+Here a code snippet: \texttt{print("Hello world")} (to be displayed inline)
+\item Here a formula: $E=mc^2$ (to be displayed inline)
+Here a formula: $E=mc^2$ (to be displayed inline)
+  \end{itemize}
+\end{itemize}
+
+Here a code block:
+
+\begin{verbatim}
+print("Hello world")
+\end{verbatim}
+
+Here a formula block:
+
+$$E=mc^2$$
+
+% missing-key-value-item
+
+% missing-form-item
+
+Some formatting chops: \textbf{bold} \textit{italic} \underline{underline} \sout{strikethrough} $_{subscript}$ $^{superscript}$ \href{.}{hyperlink} \& \href{https://github.com/DS4SD/docling}{\sout{\underline{\textit{\textbf{everything at the same time.}}}}}
+
+\begin{enumerate}
+\item Item 1 in A
+\item Item 2 in A
+\item Item 3 in A
+  \begin{enumerate}
+\item Item 1 in B
+\item Item 2 in B
+    \begin{enumerate}
+\item Item 1 in C
+\item Item 2 in C
+    \end{enumerate}
+\item Item 3 in B
+  \end{enumerate}
+\item Item 4 in A
+\end{enumerate}
+
+\begin{itemize}
+\item List item without parent list group
+\end{itemize}
+
+The end.

test/data/doc/activities.gt.tex

@@ -0,0 +1,12 @@
+\section*{Rich tables}
+
+\begin{table}[h]
+\begin{tabular}{|l|l|}
+\hline
+cell 0,0 & cell 0,1 \\ \hline
+cell 1,0 & \textit{text in italic} \\ \hline
+\begin{itemize} \item list item 1 \item list item 2 \end{itemize} & cell 2,1 \\ \hline
+cell 3,0 & \begin{table}[h] \begin{tabular}{|l|l|l|} \hline inner cell 0,0 & inner cell 0,1 & inner cell 0,2 \\ \hline inner cell 1,0 & inner cell 1,1 & inner cell 1,2 \\ \hline \end{tabular} \end{table} \\ \hline
+Some text in a generic group.  More text in the group. & cell 4,1 \\ \hline
+\end{tabular}
+\end{table}

test/data/doc/construct_doc.gt.tex

@@ -0,0 +1,11 @@
+\section*{DocLayNet: A Large Human-Annotated Dataset for Document-Layout Analysis}
+
+\begin{figure}[h]
+% image
+\caption{Figure 1: Four examples of complex page layouts across different document categories}
+% annotation[classification]: bar chart
+% annotation[description]: ...
+% annotation[molecule_data]: CC1=NNC(C2=CN3C=CN=C3C(CC3=CC(F)=CC(F)=C3)=N2)=N1
+\end{figure}
+
+% annotation[description]: A description annotation for this table.

test/data/doc/construct_rich_table_doc.gt.tex

@@ -0,0 +1,32 @@
+\section*{Contribution guideline example}
+
+This is simple.
+
+Foo \textit{emphasis} \textbf{strong emphasis} \textit{\textbf{both}} .
+
+Create your feature branch: \texttt{git checkout -b feature/AmazingFeature} .
+
+\begin{enumerate}
+\item Pull the \href{https://github.com/docling-project/docling}{\textbf{repository}} .
+Pull the \href{https://github.com/docling-project/docling}{\textbf{repository}} .
+\item Create your feature branch ( \texttt{git checkout -b feature/AmazingFeature} )
+Create your feature branch ( \texttt{git checkout -b feature/AmazingFeature} )
+\item Commit your changes ( \texttt{git commit -m 'Add some AmazingFeature'} )
+Commit your changes ( \texttt{git commit -m 'Add some AmazingFeature'} )
+\item Push to the branch ( \texttt{git push origin feature/AmazingFeature} )
+Push to the branch ( \texttt{git push origin feature/AmazingFeature} )
+\item Open a Pull Request
+\item \textbf{Whole list item has same formatting}
+\item List item has \textit{mixed or partial} formatting
+List item has \textit{mixed or partial} formatting
+\end{enumerate}
+
+\section*{\textit{Whole heading is italic}}
+
+Some \texttt{formatted_code}
+
+\subsection{\textit{Partially formatted} heading to\_escape \texttt{not_to_escape} $E=mc^2$ \& ampersand}
+
+\textit{Partially formatted} heading to\_escape \texttt{not_to_escape} $E=mc^2$ \& ampersand
+
+The end.