Lucas ARRIESSE
commited on
Commit
·
f461b05
1
Parent(s):
2f60010
Enable Table Extraction
Browse files- api/docs.py +1 -1
- requirements.txt +1 -1
api/docs.py
CHANGED
|
@@ -36,7 +36,7 @@ NSMAP = {
|
|
| 36 |
# ================================== Converting of files to .txt ====================================
|
| 37 |
|
| 38 |
KREUZBERG_CONFIG: ExtractionConfig = ExtractionConfig(
|
| 39 |
-
force_ocr=False, ocr_backend=None)
|
| 40 |
|
| 41 |
# Unfortunately needs to be kept to 1, as libreoffice isn't built to support parallel instances
|
| 42 |
LO_CONVERSION_MUTEX = asyncio.Semaphore(1)
|
|
|
|
| 36 |
# ================================== Converting of files to .txt ====================================
|
| 37 |
|
| 38 |
KREUZBERG_CONFIG: ExtractionConfig = ExtractionConfig(
|
| 39 |
+
force_ocr=False, ocr_backend=None, extract_tables=True)
|
| 40 |
|
| 41 |
# Unfortunately needs to be kept to 1, as libreoffice isn't built to support parallel instances
|
| 42 |
LO_CONVERSION_MUTEX = asyncio.Semaphore(1)
|
requirements.txt
CHANGED
|
@@ -11,4 +11,4 @@ beautifulsoup4
|
|
| 11 |
aiolimiter
|
| 12 |
httpx
|
| 13 |
Jinja2
|
| 14 |
-
kreuzberg
|
|
|
|
| 11 |
aiolimiter
|
| 12 |
httpx
|
| 13 |
Jinja2
|
| 14 |
+
kreuzberg[gmft]
|