trustgraph-flow 0.21.14__tar.gz → 0.21.16__tar.gz

{trustgraph-flow-0.21.14 → trustgraph-flow-0.21.16}/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: trustgraph-flow
-Version: 0.21.14
+Version: 0.21.16
 Summary: TrustGraph provides a means to run a pipeline of flexible AI processing components in a flexible means to achieve a processing pipeline.
 Home-page: https://github.com/trustgraph-ai/trustgraph
-Download-URL: https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v0.21.14.tar.gz
+Download-URL: https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v0.21.16.tar.gz
 Author: trustgraph.ai
 Author-email: security@trustgraph.ai
 Classifier: Programming Language :: Python :: 3

trustgraph-flow-0.21.16/scripts/pdf-ocr-mistral

@@ -0,0 +1,6 @@
+#!/usr/bin/env python3
+
+from trustgraph.decoding.mistral_ocr import run
+
+run()
+

{trustgraph-flow-0.21.14 → trustgraph-flow-0.21.16}/setup.py

@@ -59,6 +59,7 @@ setuptools.setup(
         "pulsar-client",
         "pymilvus",
         "pypdf",
+        "mistralai",
         "pyyaml",
         "qdrant-client",
         "rdflib",
@@ -98,6 +99,7 @@ setuptools.setup(
         "scripts/object-extract-row",
         "scripts/oe-write-milvus",
         "scripts/pdf-decoder",
+        "scripts/pdf-ocr-mistral",
         "scripts/prompt-generic",
         "scripts/prompt-template",
         "scripts/rows-write-cassandra",

trustgraph-flow-0.21.16/trustgraph/decoding/mistral_ocr/processor.py

@@ -0,0 +1,190 @@
+
+"""
+Simple decoder, accepts PDF documents on input, outputs pages from the
+PDF document as text as separate output objects.
+"""
+
+from pypdf import PdfWriter, PdfReader
+from io import BytesIO
+import base64
+import uuid
+import os
+
+from mistralai import Mistral
+from mistralai import DocumentURLChunk, ImageURLChunk, TextChunk
+from mistralai.models import OCRResponse
+
+from ... schema import Document, TextDocument, Metadata
+from ... schema import document_ingest_queue, text_ingest_queue
+from ... log_level import LogLevel
+from ... base import ConsumerProducer
+
+module = ".".join(__name__.split(".")[1:-1])
+
+default_input_queue = document_ingest_queue
+default_output_queue = text_ingest_queue
+default_subscriber = module
+default_api_key = os.getenv("MISTRAL_TOKEN")
+
+pages_per_chunk = 5
+
+def chunks(lst, n):
+    "Yield successive n-sized chunks from lst."
+    for i in range(0, len(lst), n):
+        yield lst[i:i + n]
+
+def replace_images_in_markdown(markdown_str: str, images_dict: dict) -> str:
+    """
+    Replace image placeholders in markdown with base64-encoded images.
+
+    Args:
+        markdown_str: Markdown text containing image placeholders
+        images_dict: Dictionary mapping image IDs to base64 strings
+
+    Returns:
+        Markdown text with images replaced by base64 data
+    """
+    for img_name, base64_str in images_dict.items():
+        markdown_str = markdown_str.replace(
+            f"![{img_name}]({img_name})", f"![{img_name}]({base64_str})"
+        )
+    return markdown_str
+
+def get_combined_markdown(ocr_response: OCRResponse) -> str:
+    """
+    Combine OCR text and images into a single markdown document.
+
+    Args:
+        ocr_response: Response from OCR processing containing text and images
+
+    Returns:
+        Combined markdown string with embedded images
+    """
+    markdowns: list[str] = []
+    # Extract images from page
+    for page in ocr_response.pages:
+        image_data = {}
+        for img in page.images:
+            image_data[img.id] = img.image_base64
+        # Replace image placeholders with actual images
+        markdowns.append(replace_images_in_markdown(page.markdown, image_data))
+
+    return "\n\n".join(markdowns)
+
+class Processor(ConsumerProducer):
+
+    def __init__(self, **params):
+
+        input_queue = params.get("input_queue", default_input_queue)
+        output_queue = params.get("output_queue", default_output_queue)
+        subscriber = params.get("subscriber", default_subscriber)
+        api_key = params.get("api_key", default_api_key)
+
+        super(Processor, self).__init__(
+            **params | {
+                "input_queue": input_queue,
+                "output_queue": output_queue,
+                "subscriber": subscriber,
+                "input_schema": Document,
+                "output_schema": TextDocument,
+            }
+        )
+
+        if api_key is None:
+            raise RuntimeError("Mistral API key not specified")
+
+        self.mistral = Mistral(api_key=api_key)
+
+        # Used with Mistral doc upload
+        self.unique_id = str(uuid.uuid4())
+
+        print("PDF inited")
+
+    def ocr(self, blob):
+
+        print("Parse PDF...", flush=True)
+
+        pdfbuf = BytesIO(blob)
+        pdf = PdfReader(pdfbuf)
+
+        for chunk in chunks(pdf.pages, pages_per_chunk):
+            
+            print("Get next pages...", flush=True)
+
+            part = PdfWriter()
+            for page in chunk:
+                part.add_page(page)
+
+            buf = BytesIO()
+            part.write_stream(buf)
+
+            print("Upload chunk...", flush=True)
+
+            uploaded_file = self.mistral.files.upload(
+                file={
+                    "file_name": self.unique_id,
+                    "content": buf.getvalue(),
+                },
+                purpose="ocr",
+            )
+
+            signed_url = self.mistral.files.get_signed_url(
+                file_id=uploaded_file.id, expiry=1
+            )
+
+            print("OCR...", flush=True)
+
+            processed = self.mistral.ocr.process(
+                model="mistral-ocr-latest",
+                include_image_base64=True,
+                document={
+                    "type": "document_url",
+                    "document_url": signed_url.url,
+                }
+            )
+
+            print("Extract markdown...", flush=True)
+
+            markdown = get_combined_markdown(processed)
+
+            print("OCR complete.", flush=True)
+
+            return markdown
+
+    async def handle(self, msg):
+
+        print("PDF message received")
+
+        v = msg.value()
+
+        print(f"Decoding {v.metadata.id}...", flush=True)
+
+        markdown = self.ocr(base64.b64decode(v.data))
+
+        r = TextDocument(
+            metadata=v.metadata,
+            text=markdown.encode("utf-8"),
+        )
+
+        await self.send(r)
+
+        print("Done.", flush=True)
+
+    @staticmethod
+    def add_args(parser):
+
+        ConsumerProducer.add_args(
+            parser, default_input_queue, default_subscriber,
+            default_output_queue,
+        )
+
+        parser.add_argument(
+            '-k', '--api-key',
+            default=default_api_key,
+            help=f'Mistral API Key'
+        )
+
+def run():
+
+    Processor.launch(module, __doc__)
+

trustgraph-flow-0.21.16/trustgraph/embeddings/ollama/__init__.py

@@ -0,0 +1,3 @@
+
+from . processor import *
+

trustgraph-flow-0.21.16/trustgraph/embeddings/ollama/__main__.py

@@ -0,0 +1,7 @@
+#!/usr/bin/env python3
+
+from . processor import run
+
+if __name__ == '__main__':
+    run()
+

trustgraph-flow-0.21.16/trustgraph/flow_version.py

@@ -0,0 +1 @@
+__version__ = "0.21.16"

{trustgraph-flow-0.21.14 → trustgraph-flow-0.21.16}/trustgraph_flow.egg-info/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: trustgraph-flow
-Version: 0.21.14
+Version: 0.21.16
 Summary: TrustGraph provides a means to run a pipeline of flexible AI processing components in a flexible means to achieve a processing pipeline.
 Home-page: https://github.com/trustgraph-ai/trustgraph
-Download-URL: https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v0.21.14.tar.gz
+Download-URL: https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v0.21.16.tar.gz
 Author: trustgraph.ai
 Author-email: security@trustgraph.ai
 Classifier: Programming Language :: Python :: 3

{trustgraph-flow-0.21.14 → trustgraph-flow-0.21.16}/trustgraph_flow.egg-info/SOURCES.txt

@@ -30,6 +30,7 @@ scripts/metering
 scripts/object-extract-row
 scripts/oe-write-milvus
 scripts/pdf-decoder
+scripts/pdf-ocr-mistral
 scripts/prompt-generic
 scripts/prompt-template
 scripts/rows-write-cassandra
@@ -72,6 +73,9 @@ trustgraph/chunking/token/__init__.py
 trustgraph/chunking/token/__main__.py
 trustgraph/chunking/token/chunker.py
 trustgraph/decoding/__init__.py
+trustgraph/decoding/mistral_ocr/__init__.py
+trustgraph/decoding/mistral_ocr/__main__.py
+trustgraph/decoding/mistral_ocr/processor.py
 trustgraph/decoding/pdf/__init__.py
 trustgraph/decoding/pdf/__main__.py
 trustgraph/decoding/pdf/pdf_decoder.py

{trustgraph-flow-0.21.14 → trustgraph-flow-0.21.16}/trustgraph_flow.egg-info/requires.txt

@@ -14,6 +14,7 @@ langchain-core
 langchain-text-splitters
 minio
 mistralai
+mistralai
 neo4j
 ollama
 openai

trustgraph-flow-0.21.14/trustgraph/flow_version.py

@@ -1 +0,0 @@
-__version__ = "0.21.14"