Commit bcbd6d9adcf8af6f5cbccb49e06c1e775fc54c49

Authored by Guilherme Andrade Del Cantoni
2 parents 28db22f3 4ff2a149

Merge branch 'master' into Pre_versao_2.0

Showing 3 changed files with 61 additions and 2 deletions   Show diff stats
Dockerfile 0 → 100644
... ... @@ -0,0 +1,45 @@
  1 +FROM ubuntu
  2 +
  3 +COPY usr/local/bin/ocr /usr/local/bin/ocr
  4 +COPY etc/init.d/ocr-ubuntu /etc/init.d/ocr
  5 +COPY entrypoint.sh /entrypoint.sh
  6 +
  7 +RUN useradd -m ocr
  8 +
  9 +RUN apt-get -y update && \
  10 + apt-get -y install libfile-find-rule-perl libfile-find-rule-perl-perl imagemagick tesseract-ocr \
  11 + gettext tesseract-ocr-por tesseract-ocr-eng pdftk poppler-utils unpaper git build-essential
  12 +
  13 +RUN git clone https://github.com/coherentgraphics/cpdf-binaries.git && \
  14 + cp cpdf-binaries/Linux-Intel-64bit/cpdf /usr/bin
  15 +
  16 +RUN perl -MCPAN -e 'install File::Touch' && \
  17 + perl -MCPAN -e 'install File::Find::Rule;' && \
  18 + perl -MCPAN -e 'install File::Touch;' && \
  19 + perl -MCPAN -e 'install Sys::Syslog;' && \
  20 + perl -MCPAN -e 'install IPC::Open3;' && \
  21 + perl -MCPAN -e 'install IO::Select;'
  22 +
  23 +RUN chmod +x /usr/local/bin/ocr && \
  24 + chmod +x /etc/init.d/ocr && \
  25 + update-rc.d ocr defaults
  26 +
  27 +RUN mkdir /var/ocr-server/ && \
  28 + mkdir -p /var/ocr-server/Entrada && \
  29 + mkdir -p /var/ocr-server/Saida && \
  30 + mkdir -p /var/ocr-server/Originais_Processados && \
  31 + mkdir -p /var/ocr-server/Erro
  32 +RUN chmod +x entrypoint.sh
  33 +
  34 +RUN ln -s /usr/bin/pdftk /usr/local/bin/pdftk && \
  35 + ln -s /usr/bin/pdfimages /usr/local/bin/pdfimages && \
  36 + ln -s /usr/bin/tesseract /usr/local/bin/tesseract && \
  37 + ln -s /usr/bin/pdfinfo /usr/local/bin/pdfinfo && \
  38 + ln -s /usr/bin/pdffonts /usr/local/bin/pdffonts && \
  39 + ln -s /usr/bin/pdftoppm /usr/local/bin/pdftoppm && \
  40 + ln -s /usr/bin/cpdf /usr/local/bin/cpdf
  41 +
  42 +VOLUME /var/ocr-server/
  43 +
  44 +CMD ["bash", "entrypoint.sh"]
  45 +
... ...
entrypoint.sh 0 → 100644
... ... @@ -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
... ... @@ -112,8 +112,10 @@ my $CONVERT = 'convert';
112 112 # If it is needed further filtering
113 113 #my $FILTER = '/usr/local/bin/textcleaner -g -e stretch -f 25 -o 10 -u -s 1 -T -p 10 ';
114 114  
115   -my @BASE_DIRS = ( '/mnt/protocolo_sede/DIGITALIZAÇÃO/ARQUIVOS PROTOCOLO/OCR/',
116   - '/mnt/protocolo_sede/DIGITALIZAÇÃO/ARQUIVOS_PROCESSOS/OCR/' );
  115 +#my @BASE_DIRS = ( '/mnt/protocolo_sede/DIGITALIZAÇÃO/ARQUIVOS PROTOCOLO/OCR/',
  116 +# '/mnt/protocolo_sede/DIGITALIZAÇÃO/ARQUIVOS_PROCESSOS/OCR/' );
  117 +
  118 +my @BASE_DIRS = ('/var/ocr-server/');
117 119  
118 120 my %SUB_DIRS = ( 'IN'=>'Entrada', 'OUT'=>'Saida', 'PROC'=>'Originais_Processados', 'TEMP'=>'/tmp/ocr_tmp', 'ERROR' => 'Erro' );
119 121  
... ...