- Package:
- python3-fitz
- Source:
- python3-fitz
- Description:
- Python binding for MuPDF
- Submitter:
- Alex Pyrgiotis
- Date:
- 2026-06-15 11:35:01 UTC
- Severity:
- normal
- Tags:
Dear Maintainer, PyMuPDF has built-in OCR support, based on Tesseract. See: https://pymupdf.readthedocs.io/en/latest/recipes-ocr.html The python3-fitz package in Debian lacks this support though. Is there any particular reason why this is the case? If not, is it possible to enable it? Thanks, Alex
Hi Alex, and the mupdf maintainers,
The fact that pymupdf is not able to use OCR seem to be related to the
fact that mupdf is compiled without support for it, and so I'm
reassigning this issue to src:mupdf.
I'm attaching a patch to enable OCR using Tesseract and Leptonica.
This would enable an use via pymudf, amongst other things.
It would mean all mupdf users would fetch tesseract and leptonica, but
the data files aren't actually required.
Would you be keep to add such dependencies, to add support for doing
OCR? It would enable a set of use cases, especially automatically doing
OCR and add a layer of invisible text on top of PDFs built by mupdf.
All the best,
Alexis
PS: That's my first time with the BTS, and so I hope I did things right :-)
------ BEGIN PATCH ------
diff --git a/debian/control b/debian/control
index dff28c5..6c30d6c 100644
--- a/debian/control
+++ b/debian/control
@@ -19,9 +19,11 @@ Build-Depends: afdko-bin:native,
libjbig2dec0-dev (>= 0.20),
libjpeg-dev,
liblcms2-dev,
+ libleptonica-dev,
libmujs-dev (>= 1.3.8),
libopenjp2-7-dev,
libssl-dev,
+ libtesseract-dev,
libx11-dev,
libxext-dev,
libxrandr-dev,
diff --git a/debian/rules b/debian/rules
index a496f69..dfc9c0d 100755
--- a/debian/rules
+++ b/debian/rules
@@ -43,6 +43,10 @@ BUILD_FLAGS += USE_SYSTEM_MUJS=yes
BUILD_FLAGS += USE_SYSTEM_LIBS=yes
# Force using system lcms2 (was not included in default system library
set)
BUILD_FLAGS += USE_SYSTEM_LCMS2=yes
+# Enable Tesseract OCR support, linking against the system tesseract and
+# leptonica libraries. This appends "-tesseract" to the build suffix,
+# so keep LIB_DIR below in sync.
+BUILD_FLAGS += tesseract=yes USE_SYSTEM_TESSERACT=yes
USE_SYSTEM_LEPTONICA=yes
BUILD_FLAGS += LD=$(DEB_HOST_GNU_TYPE)-ld
ifeq (,$(filter terse,$(DEB_BUILD_OPTIONS)))
@@ -73,7 +77,7 @@ override_dh_auto_install:
install -m744 -T $(CURDIR)/debian/mupdf.sh
$(CURDIR)/debian/tmp/usr/bin/mupdf
sh debian/install_icons.sh
ifeq (,$(filter nocheck,$(DEB_BUILD_OPTIONS)))
- $(MAKE) -C $(CURDIR)/debian/tests test
LIB_DIR=$(CURDIR)/build/shared-debug-Py_LIMITED_API_0x030d0000
+ $(MAKE) -C $(CURDIR)/debian/tests test
LIB_DIR=$(CURDIR)/build/shared-debug-Py_LIMITED_API_0x030d0000-tesseract
endif
override_dh_gencontrol:
------ END PATCH ------
Hi Alex, and the mupdf maintainers, The fact that pymupdf is not able to use OCR seem to be related to the fact that mupdf is compiled without support for it, and so I'm reassigning this issue to src:mupdf. I'm attaching a patch to enable OCR using Tesseract and Leptonica. This would enable an use via pymudf, amongst other things. It would mean all mupdf users would fetch tesseract and leptonica, but the data files aren't actually required. Would you be keep to add such dependencies, to add support for doing OCR? It would enable a set of use cases, especially automatically doing OCR and add a layer of invisible text on top of PDFs built by mupdf. All the best, Alexis PS: That's my first time with the BTS, and so I hope I did things right :-)------ BEGIN PATCH ------ diff --git a/debian/control b/debian/control index dff28c5..6c30d6c 100644 --- a/debian/control +++ b/debian/control @@ -19,9 +19,11 @@ Build-Depends: afdko-bin:native, libjbig2dec0-dev (>= 0.20), libjpeg-dev, liblcms2-dev, + libleptonica-dev, libmujs-dev (>= 1.3.8), libopenjp2-7-dev, libssl-dev, + libtesseract-dev, libx11-dev, libxext-dev, libxrandr-dev, diff --git a/debian/rules b/debian/rules index a496f69..dfc9c0d 100755 --- a/debian/rules +++ b/debian/rules @@ -43,6 +43,10 @@ BUILD_FLAGS += USE_SYSTEM_MUJS=yes BUILD_FLAGS += USE_SYSTEM_LIBS=yes # Force using system lcms2 (was not included in default system library set) BUILD_FLAGS += USE_SYSTEM_LCMS2=yes +# Enable Tesseract OCR support, linking against the system tesseract and +# leptonica libraries. This appends "-tesseract" to the build suffix, +# so keep LIB_DIR below in sync. +BUILD_FLAGS += tesseract=yes USE_SYSTEM_TESSERACT=yes USE_SYSTEM_LEPTONICA=yes BUILD_FLAGS += LD=$(DEB_HOST_GNU_TYPE)-ld ifeq (,$(filter terse,$(DEB_BUILD_OPTIONS))) @@ -73,7 +77,7 @@ override_dh_auto_install: install -m744 -T $(CURDIR)/debian/mupdf.sh $(CURDIR)/debian/tmp/usr/bin/mupdf sh debian/install_icons.sh ifeq (,$(filter nocheck,$(DEB_BUILD_OPTIONS))) - $(MAKE) -C $(CURDIR)/debian/tests test LIB_DIR=$(CURDIR)/build/shared-debug-Py_LIMITED_API_0x030d0000 + $(MAKE) -C $(CURDIR)/debian/tests test LIB_DIR=$(CURDIR)/build/shared-debug-Py_LIMITED_API_0x030d0000-tesseract endif override_dh_gencontrol:------ END PATCH ------