Use the PAGE XML parser from teklia-toolbox

.gitlab-ci.yml

@@ -36,7 +36,7 @@ stages:
 
     # Add Arkindex deps
     - pip install tox
-    - rm -rf ../api-client ../transkribus
+    - rm -rf ../api-client
 
     # Download OpenAPI schema from last backend build
     # Use internal S3 bucket URL to avoid cache issues
@@ -55,7 +55,6 @@ test-local:
   extends: .test
   script:
     - "git clone https://gitlab-ci-token:${CI_JOB_TOKEN}@gitlab.teklia.com/arkindex/api-client ../api-client"
-    - "git clone https://gitlab-ci-token:${CI_JOB_TOKEN}@gitlab.teklia.com/arkindex/transkribus ../transkribus"
     - tox -e local
 
 lint:

.isort.cfg

@@ -8,4 +8,4 @@ line_length = 88
 
 default_section=FIRSTPARTY
 known_first_party = arkindex
-known_third_party = Levenshtein,PIL,apistar,docker,ijson,lxml,natsort,pdf2image,pdfminer,requests,requests_mock,setuptools,shapely,teklia_toolbox,tenacity,transkribus,yaml,zstandard
+known_third_party = Levenshtein,PIL,apistar,docker,ijson,lxml,natsort,pdf2image,pdfminer,requests,requests_mock,setuptools,shapely,teklia_toolbox,tenacity,yaml,zstandard

Dockerfile

@@ -16,10 +16,5 @@ RUN git clone --branch=$CLIENT_BRANCH https://gitlab.teklia.com/arkindex/api-cli
   python setup.py install && \
   rm -rf /tmp/client
 
-# Install Transkribus client from the repo
-RUN git clone --branch=$TRANSKRIBUS_BRANCH https://gitlab.teklia.com/arkindex/transkribus /tmp/transkribus && \
-  cd /tmp/transkribus && python setup.py install && \
-  rm -rf /tmp/transkribus
-
 COPY --from=build /build/dist/arkindex-tasks-*.tar.gz /tmp/tasks.tar.gz
 RUN pip install /tmp/tasks.tar.gz && rm /tmp/tasks.tar.gz

README.md

@@ -47,12 +47,9 @@ This project relies on several internal dependencies. If you are editing the Ark
 
 ```console
 git clone git@gitlab.com:teklia/arkindex/api-client.git ../api-client
-git clone git@gitlab.com:teklia/arkindex/transkribus.git ../transkribus
 workon tasks
 cd ../api-client
 pip install -e .
-cd ../transkribus
-pip install -e .
 ```
 
 To run the unit tests with `tox`, you will have to use an argument to switch its environment to the one named `local`, which will use your local repositories as dependencies:

arkindex_tasks/import_files/pagexml.py

@@ -2,8 +2,8 @@
 import logging
 
 from shapely.geometry import LinearRing
+from teklia_toolbox.pagexml import PageXmlPage
 from teklia_toolbox.requests import retried_request
-from transkribus.pagexml import PageXmlPage
 
 from arkindex_tasks.base import WORKER_RUN_ID
 

base/requirements.txt

 cryptography==39.0.0
-lxml==4.9.2
+lxml==5.1.0
 pdf2image==1.16.0
 Pillow==10.1.0
 zstandard==0.15.2

requirements.txt

@@ -4,6 +4,5 @@ natsort==8.3.1
 pdfminer.six==20221105
 python-magic==0.4.27
 Shapely==1.8.0
-teklia-toolbox==0.1.4rc3
+teklia-toolbox==0.1.4rc4
 tenacity==8.2.3
-transkribus-client==0.3.4rc1

tox.ini

@@ -17,5 +17,4 @@ deps =
 deps =
   git+https://gitlab.teklia.com/arkindex/apistar.git#egg=apistar
   ../api-client
-  ../transkribus
   {[testenv]deps}