Installing Python packages (Offline mode)

Use the following steps to install Python in an offline mode environment.

You must download the packages by using an internet-enabled computer, and then transfer the files to the offline computer. The internet-enabled computer must have Python 3.5.2 and the Python package manager Pip installed.

Procedure

  1. On the online computer, verify if Python and Pip are installed.

    Use the which python command to identify the installed versions of Python. The command returns the location of the installed instances. If Python 3.5 is not displayed in the results, go to the Python location and verify the version. For example:

    python -V
    /usr/bin/python -V
    /usr/local/bin/python3.5 -V

    If Python 3.5.2 is not displayed, then you must install it.

    Also, the Python packages must be loaded in a specific order to avoid problems with conflicting dependencies. The preliminary packages are downloaded to pre_pythoninstall, and all the rest are downloaded to pythoninstall.

    If Python 3.5 and Pip are installed, you can skip steps 2 - 3.

  2. Install Python on the online computer.
    1. Log in as the root user, and run the following commands: 
      mkdir -p /home/sifsuser/pythoninstall
      yum -y install dos2unix
      yum -y install yum-utils
      yum -y install openssl openssl-devel
      yum -y install gcc sqlite-devel bzip2 bzip2-devel gcc-c++
    2. Download and extract Python 3.5.2. 
      cd /home/sifsuser/pythoninstall
      wget https://www.python.org/ftp/python/3.5.2/Python-3.5.2.tgz
      tar -xvzf /home/sifsuser/pythoninstall/Python-3.5.2.tgz
    3. Install Python. 
      cd /home/sifsuser/pythoninstall/Python-3.5.2
      ./configure
      make
      make altinstall
      /usr/local/bin/python3.5 -V

      The results should show Python 3.5.2.

  3. Install the Python package installer on the online computer.
    1. Download and install the Python package installer as the root user. 
      cd /home/sifsuser/pythoninstall
      wget https://files.pythonhosted.org/packages/b6/ac/7015eb97dc749283ffdec1c3a88ddb8ae03b8fad0f0e611408f196358da3/pip-9.0.1-py2.py3-none-any.whl -O /home/sifsuser/pythoninstall/pip-9.0.1-py2.py3-none-any.whl
      /usr/local/bin/python3.5 /home/sifsuser/pythoninstall/pip-9.0.1-py2.py3-none-any.whl/pip install --no-index --find-links=/home/sifsuser/pythoninstall/ pip-9.0.1-py2.py3-none-any.whl

      The results should show Successfully installed pip-9.0.1.

    2. Enter the following command to verify the version: 
      pip3.5 --version

      The result should display a Pip version that is equal to or greater than 9.0.1. You can ignore any update messages.

  4. Download the prerequisite packages on the online computer.

    Log in as the root user, and run the following commands:

    cd /home/sifsuser/pre_pythoninstall
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pre_pythoninstall/ numpy==1.12.1
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pre_pythoninstall/ h5py==2.7.0
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pre_pythoninstall/ keras==2.0.2
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pre_pythoninstall/ scipy==1.0.1
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pre_pythoninstall/ pyyaml==3.12
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pre_pythoninstall/ six==1.11.0
    cd /home/sifsuser
    tar -czvf pre_pythoninstall.tar.gz pre_pythoninstall
  5. Download the main packages on the online computer.

    Log in as the root user, and run the following commands:

    cd /home/sifsuser/pythoninstall
    yumdownloader dos2unix yum-utils openssl openssl-devel gcc sqlite-devel bzip2 bzip2-devel gcc-c++ --destdir /home/sifsuser/pythoninstall/ --resolve
    wget https://public.dhe.ibm.com/ibmdl/export/pub/software/data/db2/drivers/odbc_cli/linuxx64_odbc_cli.tar.gz -P /home/sifsuser/pythoninstall
    wget https://github.com/pgmpy/pgmpy/archive/dev.zip -O /home/sifsuser/pythoninstall/dev.zip
    wget https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-2.0.0/en_core_web_sm-2.0.0.tar.gz -O /home/sifsuser/pythoninstall/en_core_web_sm-2.0.0.tar.gz
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/  beautifulsoup4==4.5.1
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/  email_reply_parser==0.5.9
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/  flask_cors==3.0.3
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/  flask_restful==0.3.6
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/  flask==0.12.2
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/  gnip_trend_detection==0.5
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ hdfs==2.1.0
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ ibm_db==2.0.8a
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ nltk==3.2.4
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ pandas==0.18.1
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ pypandoc==1.4
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ pyspark==2.3.1
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ requests==2.19.1
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ requests-kerberos==0.12.0
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ scikit-learn==0.18.1
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ segtok==1.5.6
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ spacy==2.0.5
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ statsmodels==0.8.0
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ tensorflow==1.8.0
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ textacy==0.4.1
    /usr/local/bin/pip3.5 download -d /home/sifsuser/pythoninstall/ wrapt==1.10.11
    cd /home/sifsuser/
    tar -czvf pythoninstall.tar.gz pythoninstall
  6. Transfer the package files from the online computer to the offline computer. 
    scp pre_pythoninstall.tar.gz root@<offline_host>:/home/sifsuser
    scp pythoninstall.tar.gz root@<offline_host>:/home/sifsuser
  7. On the offline computer, decompress the transferred files. 
    cd /home/sifsuser/
    tar -xzf /home/sifsuser/pre_pythoninstall.tar.gz
    tar -xzf /home/sifsuser/pythoninstall.tar.gz
  8. Install the required RPMs on the offline computer. 
    cd /home/sifsuser/pythoninstall
    yum localinstall *.rpm
    Note: If you see the message This system is not registered with an entitlement server. You can use subscription-manager to register., as the root user, set enabled=0 in the /etc/yum/pluginconf.d/subscription-manager.conf, and run the command again.
  9. Install Python on the offline computer. 
    cd /home/sifsuser/pythoninstall/Python-3.5.2
    ./configure
    make
    make altinstall
  10. Install the Python package installer on the offline computer. 
    cd /home/sifsuser/pythoninstall 
    /usr/local/bin/python3.5 /home/sifsuser/pythoninstall/pip-9.0.1-py2.py3-none-any.whl/pip install --no-index --find-links=/home/sifsuser/pythoninstall/ pip-9.0.1-py2.py3-none-any.whl
  11. On the offline computer, install the odbc cli database driver. 
    tar -xzf /home/sifsuser/pythoninstall/linuxx64_odbc_cli.tar.gz -C /usr/local/lib/python3.5/site-packages/
  12. On the offline computer, export the database home parameter. 
    export IBM_DB_HOME=/usr/local/lib/python3.5/site-packages/clidriver
  13. On the offline computer, as the root user, create a text file that is named pre_reqs.txtin the /home/sifsuser, and add the following lines to the file: 
    h5py==2.7.0
keras==2.0.2
numpy==1.12.1
pyyaml==3.12
scipy==1.0.1
six==1.11.0
  14. Install the prerequisite Python libraries on the offline computer. 
    cd /home/sifsuser
    /usr/local/bin/pip3.5 install --no-deps --no-index --find-links=/home/sifsuser/pre_pythoninstall -r pre_reqs.txt
  15. On the offline computer, create a file that is named requirements.txt in the /home/sifsuser directory, and include the following content in the file: 
    beautifulsoup4==4.5.1
email_reply_parser==0.5.9
flask_cors==3.0.3
flask_restful==0.3.6
flask==0.12.2
gnip_trend_detection==0.5
hdfs==2.1.0
ibm_db==2.0.8a
nltk==3.2.4
pandas==0.18.1
pypandoc==1.4
pyspark==2.3.1
requests==2.19.1
requests-kerberos==0.12.0
scikit-learn==0.18.1
segtok==1.5.6
spacy==2.0.5
statsmodels==0.8.0
tensorflow==1.8.0
textacy==0.4.1
wrapt==1.10.11
  16. On the offline computer, install the other libraries by running the following commands: 
    /usr/local/bin/pip3.5 install --no-index --find-links=/home/sifsuser/pythoninstall/ pypandoc
    /usr/local/bin/pip3.5 install --no-index --find-links=/home/sifsuser/pythoninstall/ -r requirements.txt
  17. On the offline computer, install the pgympy Bayesian network library: 
    cd /home/sifsuser/pythoninstall/
    unzip /home/sifsuser/pythoninstall/dev.zip
    cd /home/sifsuser/pythoninstall/pgmpy-dev/
    /usr/local/bin/python3.5 setup.py install
  18. On the offline computer, install spaCy. 
    cd /home/sifsuser/pythoninstall
    /usr/local/bin/pip3.5 install --no-index --find-links=/home/sifsuser/pythoninstall/ en_core_web_sm-2.0.0.tar.gz

    For more information about spaCy, see the Models & Language quick start (https://spacy.io/docs/usage/models).

  19. On the offline computer, create a symbolic link for this model to en: 
    /usr/local/bin/python3.5 -m spacy link en_core_web_sm en