zlei6/ragflow
1
0
Fork 0
mirror of https://github.com/infiniflow/ragflow.git synced 2026-07-13 08:28:22 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: Docling parser extracts mathematical formulas (#16645)

Browse Source
This commit is contained in:
Sohaib Ahmed
2026-07-06 08:17:46 +05:00
committed by GitHub
parent 3d2f60c34f
commit 534a3a7faa
2 changed files with 22 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
deepdoc/parser/docling_parser.py
View File

@@ -33,9 +33,14 @@ from PIL import Image
from common.constants import MAXIMUM_PAGE_NUMBER
try:
from docling.document_converter import DocumentConverter
from docling.document_converter import DocumentConverter, PdfFormatOption
from docling.datamodel.base_models import InputFormat
from docling.datamodel.pipeline_options import PdfPipelineOptions
except Exception:
DocumentConverter = None
PdfFormatOption = None
InputFormat = None
PdfPipelineOptions = None
try:
from deepdoc.parser.pdf_parser import RAGFlowPdfParser
@@ -217,20 +222,20 @@ class DoclingParser(RAGFlowPdfParser):
def _iter_doc_items(self, doc) -> Iterable[tuple[str, Any, Optional[_BBox]]]:
for t in getattr(doc, "texts", []):
label = getattr(t, "label", "")
if label in ("formula",):
text = getattr(t, "text", "") or getattr(t, "orig", "")
bbox = _extract_bbox_from_prov(t)
yield (DoclingContentType.EQUATION.value, text, bbox)
continue
parent = getattr(t, "parent", "")
ref = getattr(parent, "cref", "")
label = getattr(t, "label", "")
if (label in ("section_header", "text") and ref in ("#/body",)) or label in ("list_item",):
text = getattr(t, "text", "") or ""
bbox = _extract_bbox_from_prov(t)
yield (DoclingContentType.TEXT.value, text, bbox)
for item in getattr(doc, "texts", []):
if getattr(item, "label", "") in ("FORMULA",):
text = getattr(item, "text", "") or ""
bbox = _extract_bbox_from_prov(item)
yield (DoclingContentType.EQUATION.value, text, bbox)
def _transfer_to_sections(self, doc, parse_method: str) -> list[tuple[str, ...]]:
sections: list[tuple[str, ...]] = []
for typ, payload, bbox in self._iter_doc_items(doc):
@@ -240,6 +245,8 @@ class DoclingParser(RAGFlowPdfParser):
continue
elif typ == DoclingContentType.EQUATION.value:
section = payload.strip()
if not section:
continue
else:
continue
@@ -556,7 +563,11 @@ class DoclingParser(RAGFlowPdfParser):
except Exception as e:
self.logger.warning(f"[Docling] render pages failed: {e}")
conv = DocumentConverter()
do_formula_enrichment = os.environ.get("DOCLING_FORMULA_ENRICHMENT", "0").strip().lower() in ("1", "true", "yes", "on")
self.logger.info(f"[Docling] Local conversion (formula_enrichment={do_formula_enrichment}): {src_path}")
pipeline_options = PdfPipelineOptions()
pipeline_options.do_formula_enrichment = do_formula_enrichment
conv = DocumentConverter(format_options={InputFormat.PDF: PdfFormatOption(pipeline_options=pipeline_options)})
conv_res = conv.convert(str(src_path))
doc = conv_res.document
if callback:

2
docs/guides/dataset/select_pdf_parser.md
View File

@@ -72,6 +72,8 @@ To use an external Docling Serve instance (instead of local in-process Docling),
When `DOCLING_SERVER_URL` is set, RAGFlow sends PDF content to Docling Serve (`/v1/convert/source`, with fallback to `/v1alpha/convert/source`) and ingests the returned markdown/text. If the variable is not set, RAGFlow keeps using local Docling (`USE_DOCLING=true` + installed package) behavior.
- `DOCLING_FORMULA_ENRICHMENT`: (local Docling) Set to `1` to convert formulas to LaTeX (accurate but slow). Default `0` — formulas are still extracted as their original text.
:::note
All MinerU environment variables are optional. When set, these values are used to auto-provision a MinerU OCR model for the tenant on first use. To avoid auto-provisioning, skip the environment variable settings and only configure MinerU from the **Model providers** page in the UI.
:::