Download as
Browse Code »

Commit c7dd64c8599aeee9ddbbc69a97ab1073cd3baa2e

Authored by Guilherme Andrade Del Cantoni
1 parent 5583bbf9
Exists in master and in 1 other branch Pre_versao_2.0

Implementação inicial do dockerfile para projeto OCR-SERVER

Showing 3 changed files with 62 additions and 2 deletions   Show diff stats
Dockerfile 0 → 100644
  Diff comments   View file @c7dd64c
... ... @@ -0,0 +1,46 @@
  1 +FROM ubuntu
  2 +
  3 +RUN apt-get -y update
  4 +RUN apt-get -y install libfile-find-rule-perl libfile-find-rule-perl-perl imagemagick gettext tesseract-ocr-por tesseract-ocr-eng pdftk poppler-utils unpaper
  5 +
  6 +RUN apt-get -y install git
  7 +RUN git clone https://github.com/coherentgraphics/cpdf-binaries.git
  8 +RUN cp cpdf-binaries/Linux-Intel-64bit/cpdf /usr/bin
  9 +
  10 +RUN useradd -m ocr
  11 +
  12 +COPY usr/local/bin/ocr /usr/local/bin/ocr
  13 +RUN chmod +x /usr/local/bin/ocr
  14 +COPY etc/init.d/ocr-ubuntu /etc/init.d/ocr
  15 +
  16 +RUN chmod +x /etc/init.d/ocr; update-rc.d ocr defaults
  17 +
  18 +RUN apt-get -y install build-essential
  19 +RUN perl -MCPAN -e 'install File::Touch'
  20 +
  21 +RUN mkdir /var/ocr-server/
  22 +RUN mkdir -p /var/ocr-server/Entrada
  23 +RUN mkdir -p /var/ocr-server/Saida
  24 +RUN mkdir -p /var/ocr-server/Originais_Processados
  25 +RUN mkdir -p /var/ocr-server/Erro
  26 +COPY entrypoint.sh /entrypoint.sh
  27 +RUN chmod +x entrypoint.sh
  28 +
  29 +RUN perl -MCPAN -e 'install File::Find::Rule;'
  30 +RUN perl -MCPAN -e 'install File::Touch;'
  31 +RUN perl -MCPAN -e 'install Sys::Syslog;'
  32 +RUN perl -MCPAN -e 'install IPC::Open3;'
  33 +RUN perl -MCPAN -e 'install IO::Select;'
  34 +
  35 +RUN apt-get -y install tesseract-ocr
  36 +RUN ln -s /usr/bin/pdftk /usr/local/bin/pdftk
  37 +RUN ln -s /usr/bin/pdfimages /usr/local/bin/pdfimages
  38 +RUN ln -s /usr/bin/tesseract /usr/local/bin/tesseract
  39 +RUN ln -s /usr/bin/pdfinfo /usr/local/bin/pdfinfo
  40 +RUN ln -s /usr/bin/pdffonts /usr/local/bin/pdffonts
  41 +RUN ln -s /usr/bin/pdftoppm /usr/local/bin/pdftoppm
  42 +RUN ln -s /usr/bin/cpdf /usr/local/bin/cpdf
  43 +
  44 +VOLUME /var/ocr-server/
  45 +CMD ["bash", "entrypoint.sh"]
  46 +
... ...
entrypoint.sh 0 → 100644
  Diff comments   View file @c7dd64c
... ... @@ -0,0 +1,12 @@
  1 +#!/usr/bin/env bash
  2 +
  3 +mkdir /var/ocr-server/
  4 +mkdir /var/ocr-server/Entrada
  5 +mkdir /var/ocr-server/Saida
  6 +mkdir /var/ocr-server/Originais_Processados
  7 +mkdir /var/ocr-server/Erro
  8 +chmod -R 777 /var/ocr-server
  9 +
  10 +service ocr start
  11 +
  12 +tail -f /var/log/dmesg
... ...
usr/local/bin/ocr
  Diff comments   View file @c7dd64c
... ... @@ -84,8 +84,10 @@ my $CONVERT = '/usr/bin/convert';
84 84 # If it is needed further filtering
85 85 #my $FILTER = '/usr/local/bin/textcleaner -g -e stretch -f 25 -o 10 -u -s 1 -T -p 10 ';
86 86  
87   -my @BASE_DIRS = ( '/mnt/protocolo_sede/DIGITALIZAÇÃO/ARQUIVOS PROTOCOLO/OCR/',
88   - '/mnt/protocolo_sede/DIGITALIZAÇÃO/ARQUIVOS_PROCESSOS/OCR/' );
  87 +#my @BASE_DIRS = ( '/mnt/protocolo_sede/DIGITALIZAÇÃO/ARQUIVOS PROTOCOLO/OCR/',
  88 +# '/mnt/protocolo_sede/DIGITALIZAÇÃO/ARQUIVOS_PROCESSOS/OCR/' );
  89 +
  90 +my @BASE_DIRS = ('/var/ocr-server/');
89 91  
90 92 my %SUB_DIRS = ( 'IN'=>'Entrada', 'OUT'=>'Saida', 'PROC'=>'Originais_Processados', 'TEMP'=>'/tmp/ocr_tmp', 'ERROR' => 'Erro' );
91 93  
... ...