Merge pull request #1 from jvanz/minor-fixes-to-aassumpcao-changes

Minor fixes to aassumpcao changes
jvanz · Dec 5, 2020 · 9357129 · 9357129
2 parents 5178830 + bf65f15
commit 9357129
Show file tree

Hide file tree

Showing 4 changed files with 7 additions and 10 deletions.
diff --git a/.gitignore b/.gitignore
@@ -134,3 +134,5 @@ dmypy.json
 data/*
 *scratch/
 tests/text_extraction_tests_aassumpcao.py
+fake_gazette.json
+fake_gazette.txt
diff --git a/Dockerfile b/Dockerfile
@@ -3,7 +3,11 @@ FROM python:3.8
 RUN adduser --system gazette
 
 RUN apt-get update -y
-RUN apt-get -y install default-jre wait-for-it
+RUN apt-get -y install default-jre wait-for-it libtesseract4 tesseract-ocr-por
+
+# install Apache Tika
+ADD http://archive.apache.org/dist/tika/tika-app-1.24.jar /tika-app.jar
+RUN chmod 755 /tika-app.jar 
 
 RUN mkdir /mnt/code
 COPY . /mnt/code
@@ -12,9 +16,5 @@ ENV PYTHONPATH=/mnt/code
 
 RUN pip install --no-cache-dir -r requirements.txt
 
-# install Apache Tika
-ADD http://archive.apache.org/dist/tika/tika-app-1.24.jar /tika-app.jar
-RUN chmod 755 /tika-app.jar 
-
 USER gazette
 
diff --git a/tests/data/fake_gazette.json b/tests/data/fake_gazette.json
diff --git a/tests/data/fake_gazette.txt b/tests/data/fake_gazette.txt