example: add a paper metadata extraction example

README.md

@@ -139,6 +139,7 @@ It defines an index flow like this:
 | [FastAPI Server with Docker](examples/fastapi_server_docker) | Run the semantic search server in a Dockerized FastAPI setup |
 | [Product Recommendation](examples/product_recommendation) | Build real-time product recommendations with LLM and graph database|
 | [Image Search with Vision API](examples/image_search) | Generates detailed captions for images using a vision model, embeds them, enables live-updating semantic search via FastAPI and served on a React frontend|
+| [Paper Metadata](examples/paper_metadata) | Index papers in PDF files, and build metadata tables for each paper |
 
 More coming and stay tuned 👀!
 

examples/paper_metadata/.env.example

@@ -0,0 +1,4 @@
+# Postgres database address for cocoindex
+COCOINDEX_DATABASE_URL=postgres://cocoindex:cocoindex@localhost/cocoindex
+
+OPENAI_API_KEY=

examples/paper_metadata/.gitignore

@@ -0,0 +1 @@
+.env

examples/paper_metadata/README.md

@@ -0,0 +1,60 @@
+# Build embedding index from PDF files and query with natural language
+[![GitHub](https://img.shields.io/github/stars/cocoindex-io/cocoindex?color=5B5BD6)](https://github.com/cocoindex-io/cocoindex)
+
+
+In this example, we will build a bunch of tables for papers in PDF files, including:
+
+- Metadata (title, authors, abstract) for each paper.
+- Author-to-paper mapping, for author-based query.
+- Embeddings for titles and abstract chunks, for semantics search. 
+
+We appreciate a star ⭐ at [CocoIndex Github](https://github.com/cocoindex-io/cocoindex) if this is helpful.
+
+## Steps
+### Indexing Flow
+
+1. We will ingest a list of papers in PDF.
+2. For each file, we:
+   - Extract the first page of the paper.
+   - Convert the first page to Markdown.
+   - Extract metadata (title, authors, abstract) from the first page.
+   - Split the abstract into chunks, and compute embeddings for each chunk.
+3. We will export to the following tables in Postgres with PGVector:
+   - Metadata (title, authors, abstract) for each paper.
+   - Author-to-paper mapping, for author-based query.
+   - Embeddings for titles and abstract chunks, for semantics search. 
+
+
+## Prerequisite
+
+1.  [Install Postgres](https://cocoindex.io/docs/getting_started/installation#-install-postgres) if you don't have one.
+
+2.  dependencies:
+
+    ```bash
+    pip install -e .
+    ```
+3.  Create a `.env` file from `.env.example`, and fill `OPENAI_API_KEY`.
+
+## Run
+
+Update index, which will also setup the tables at the first time:
+
+```bash
+cocoindex update --setup main.py
+```
+
+You can also run the command with `-L`, which will watch for file changes and update the index automatically.
+
+```bash
+cocoindex update --setup -L main.py
+```
+
+## CocoInsight
+I used CocoInsight (Free beta now) to troubleshoot the index generation and understand the data lineage of the pipeline. It just connects to your local CocoIndex server, with zero pipeline data retention. Run following command to start CocoInsight:
+
+```
+cocoindex server -ci main.py
+```
+
+Then open the CocoInsight UI at [https://cocoindex.io/cocoinsight](https://cocoindex.io/cocoinsight).

examples/paper_metadata/main.py

@@ -0,0 +1,183 @@
+import cocoindex
+import io
+import tempfile
+import dataclasses
+import datetime
+
+from marker.config.parser import ConfigParser
+from marker.converters.pdf import PdfConverter
+from marker.models import create_model_dict
+from marker.output import text_from_rendered
+from functools import cache
+from pypdf import PdfReader, PdfWriter
+
+
+@cache
+def get_marker_converter() -> PdfConverter:
+    config_parser = ConfigParser({})
+    return PdfConverter(
+        create_model_dict(), config=config_parser.generate_config_dict()
+    )
+
+
+@dataclasses.dataclass
+class PaperBasicInfo:
+    num_pages: int
+    first_page: bytes
+
+
+@cocoindex.op.function()
+def extract_basic_info(content: bytes) -> PaperBasicInfo:
+    """Extract the first pages of a PDF."""
+    reader = PdfReader(io.BytesIO(content))
+
+    output = io.BytesIO()
+    writer = PdfWriter()
+    writer.add_page(reader.pages[0])
+    writer.write(output)
+
+    return PaperBasicInfo(num_pages=len(reader.pages), first_page=output.getvalue())
+
+
+@dataclasses.dataclass
+class Author:
+    """One author of the paper."""
+
+    name: str
+    email: str | None
+    affiliation: str | None
+
+
+@dataclasses.dataclass
+class PaperMetadata:
+    """
+    Metadata for a paper.
+    """
+
+    title: str
+    authors: list[Author]
+    abstract: str
+
+
+@cocoindex.op.function(gpu=True, cache=True, behavior_version=2)
+def pdf_to_markdown(content: bytes) -> str:
+    """Convert to Markdown."""
+
+    with tempfile.NamedTemporaryFile(delete=True, suffix=".pdf") as temp_file:
+        temp_file.write(content)
+        temp_file.flush()
+        text, _, _ = text_from_rendered(get_marker_converter()(temp_file.name))
+        return text
+
+
+@cocoindex.transform_flow()
+def text_to_embedding(
+    text: cocoindex.DataSlice[str],
+) -> cocoindex.DataSlice[list[float]]:
+    """
+    Embed the text using a SentenceTransformer model.
+    This is a shared logic between indexing and querying, so extract it as a function.
+    """
+    return text.transform(
+        cocoindex.functions.SentenceTransformerEmbed(
+            model="sentence-transformers/all-MiniLM-L6-v2"
+        )
+    )
+
+
+@cocoindex.flow_def(name="PaperMetadata")
+def paper_metadata_flow(
+    flow_builder: cocoindex.FlowBuilder, data_scope: cocoindex.DataScope
+) -> None:
+    """
+    Define an example flow that embeds files into a vector database.
+    """
+    data_scope["documents"] = flow_builder.add_source(
+        cocoindex.sources.LocalFile(path="papers", binary=True),
+        refresh_interval=datetime.timedelta(seconds=10),
+    )
+
+    paper_metadata = data_scope.add_collector()
+    metadata_embeddings = data_scope.add_collector()
+    author_papers = data_scope.add_collector()
+
+    with data_scope["documents"].row() as doc:
+        doc["basic_info"] = doc["content"].transform(extract_basic_info)
+        doc["first_page_md"] = doc["basic_info"]["first_page"].transform(
+            pdf_to_markdown
+        )
+        doc["metadata"] = doc["first_page_md"].transform(
+            cocoindex.functions.ExtractByLlm(
+                llm_spec=cocoindex.LlmSpec(
+                    api_type=cocoindex.LlmApiType.OPENAI, model="gpt-4o"
+                ),
+                output_type=PaperMetadata,
+                instruction="Please extract the metadata from the first page of the paper.",
+            )
+        )
+        doc["title_embedding"] = text_to_embedding(doc["metadata"]["title"])
+        doc["abstract_chunks"] = doc["metadata"]["abstract"].transform(
+            cocoindex.functions.SplitRecursively(
+                custom_languages=[
+                    cocoindex.functions.CustomLanguageSpec(
+                        language_name="abstract",
+                        separators_regex=[r"[.?!]+\s+", r"[:;]\s+", r",\s+", r"\s+"],
+                    )
+                ]
+            ),
+            language="abstract",
+            chunk_size=500,
+            min_chunk_size=200,
+            chunk_overlap=150,
+        )
+
+        paper_metadata.collect(
+            filename=doc["filename"],
+            title=doc["metadata"]["title"],
+            authors=doc["metadata"]["authors"],
+            abstract=doc["metadata"]["abstract"],
+        )
+        metadata_embeddings.collect(
+            id=cocoindex.GeneratedField.UUID,
+            filename=doc["filename"],
+            location="title",
+            text=doc["metadata"]["title"],
+            embedding=doc["title_embedding"],
+        )
+        with doc["metadata"]["authors"].row() as author:
+            author_papers.collect(
+                author_name=author["name"],
+                filename=doc["filename"],
+            )
+
+        with doc["abstract_chunks"].row() as chunk:
+            chunk["embedding"] = text_to_embedding(chunk["text"])
+            metadata_embeddings.collect(
+                id=cocoindex.GeneratedField.UUID,
+                filename=doc["filename"],
+                location="abstract",
+                text=chunk["text"],
+                embedding=chunk["embedding"],
+            )
+
+    paper_metadata.export(
+        "paper_metadata",
+        cocoindex.targets.Postgres(),
+        primary_key_fields=["filename"],
+    )
+    metadata_embeddings.export(
+        "metadata_embeddings",
+        cocoindex.targets.Postgres(),
+        primary_key_fields=["id"],
+        vector_indexes=[
+            cocoindex.VectorIndexDef(
+                field_name="embedding",
+                metric=cocoindex.VectorSimilarityMetric.COSINE_SIMILARITY,
+            )
+        ],
+    )
+    author_papers.export(
+        "author_papers",
+        cocoindex.targets.Postgres(),
+        primary_key_fields=["author_name", "filename"],
+    )

examples/paper_metadata/pyproject.toml

@@ -0,0 +1,13 @@
+[project]
+name = "paper-metadata"
+version = "0.1.0"
+description = "Build index for papers with both metadata and content embeddings"
+requires-python = ">=3.11"
+dependencies = [
+    "cocoindex[embeddings]>=0.1.62",
+    "pypdf>=5.7.0",
+    "marker-pdf>=1.5.2",
+]
+
+[tool.setuptools]
+packages = []