Fix: docker images

Author: gchehab

Dockerfile

@@ -1,11 +1,6 @@
 
 FROM ubuntu:21.04
 
-# Cópia de arquivos do projeto OCR-SERVER
-COPY usr/local/bin/ocr /usr/local/bin/ocr
-COPY etc/init.d/ocr-ubuntu /etc/init.d/ocr
-COPY entrypoint.sh /entrypoint.sh
-
 WORKDIR /tmp
 
 ENV TZ=Etc/UTC
@@ -14,16 +9,23 @@ RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone
 # Instalação dos pacotes pré-requisitos do ocr-server 2
 RUN apt-get -y update && \
     apt-get install -y tesseract-ocr tesseract-ocr-por tesseract-ocr-eng tesseract-ocr-spa leptonica-progs \
-    poppler-utils pdftk unpaper ocaml ghostscript imagemagick libcamlpdf-ocaml \
+    poppler-utils pdftk unpaper ocaml ghostscript imagemagick libcamlpdf-ocaml rsyslog \
     wget perl libfile-find-rule-perl libfile-touch-perl libunix-syslog-perl
 
 RUN wget \
     https://raw.githubusercontent.com/coherentgraphics/cpdf-binaries/master/Linux-Intel-64bit/cpdf \
-    -o /usr/local/bin/cpdf && \
+    -O /usr/local/bin/cpdf && \
     chmod 755 /usr/local/bin/cpdf
 
+RUN update-rc.d rsyslog defaults
+
 RUN useradd -m ocr
 
+# Cópia de arquivos do projeto OCR-SERVER
+COPY usr/local/bin/ocr /usr/local/bin/ocr
+COPY etc/init.d/ocr-ubuntu /etc/init.d/ocr
+COPY entrypoint.sh /entrypoint.sh
+
 RUN chmod +x /usr/local/bin/ocr && \
     chmod +x /etc/init.d/ocr && \
     update-rc.d ocr defaults

entrypoint.sh

@@ -1,7 +1,7 @@
 #!/usr/bin/env bash
 
 # Inicializa serviço de log
-/etc/init.d/rsyslog start
+service rsyslog start
 
 # Cria estrutura de pastas para monitoramento de arquivos
 mkdir -p /var/ocr-server/
@@ -14,4 +14,7 @@ chmod -R 777 /var/ocr-server
 # Iniciar serviço do OCR-Server
 service ocr start
 
-tail -f /var/log/syslog
+while [ 1 ]; do
+	tail -f /var/log/syslog
+	sleep 1;
+done

usr/local/bin/ocr

@@ -44,9 +44,10 @@
 #		Now using '-oem 1' option to tesseract as there is an issue with default OS trained data
 #		Fix: detection of files with unicode filenames
 #		Fix: now uses current shell to spawn process, plain open3 where not using default shell env, needed for java progs
-#	2.1.1	Fixed docker on Ubuntu 20.04
+#	2.1.1	Fixed docker on Ubuntu 21.04
 #		Color conversion to gray is off by default
 #		Fix: new images format from INPE samples
+#		Fix: now uses default tesseract algorithms
 #
 #	TODO: 	- Changes get_imgs and OCR processing to enable pages with more than one image -- it
 #		would not work on previous versions that assumed #pages = #imgs. Version 1.0.1 counts them
@@ -99,8 +100,8 @@ my $COLOR_THRES = .03; # Min color spread, below this value, will convert image
 # Command dependencies
 
 # depends on tesseract-ocr an tesseract-ocr-por 3.05-dev or higher -- for pdf/a Tesseract 4.0 is recomended
-my $TESSERACT = 'tesseract --oem 1'; 		# if Tesseract => 4.0
-#my $TESSERACT = 'tesseract';			# if Tesseract < 4.0
+#my $TESSERACT = 'tesseract --oem 2'; 		# if Tesseract => 4.0
+my $TESSERACT = 'tesseract';			# if Tesseract < 4.0
 
 # Depends on pdftk 2.02 or higher
 my $PDFTK = 'pdftk';
@@ -142,6 +143,7 @@ my %SUB_DIRS = ( 'IN'=>'Entrada', 'OUT'=>'Saida', 'PROC'=>'Originais_Processados
 # Safeguard im case of cpuinfo has not identified correctly the number of CPUs 
 $MAX_PGS = ($MAX_PGS==0) ? 4 : $MAX_PGS;
 
+$ENV{'SHELL'} = exists $ENV{'SHELL'} ? $ENV{'SHELL'} : '/bin/bash';
 $ENV{'PATH'} = '/usr/local/bin:/usr/bin:/bin';
 $ENV{'IFS'} = '\t\n';