DatasourceProvider reads file-content in "bytes" format and causing parsing to fail

[xiangce]

Original Exception raised by the parser

  File "/work/insights/insights-core/insights/core/__init__.py", line 95, in _handle_content
    self.parse_content(context.content)
  File "/work/insights/insights-core/insights/parsers/ld_library_path.py", line 56, in parse_content
    user, _, raw = [s.strip() for s in line.partition(' ')]
AttributeError: 'int' object has no attribute 'partition'

Collected meta_data:

# cat /tmp/insights-vm37-39.gsslab.pek2.redhat.com-20210308120914/meta_data/insights.specs.Specs.ld_library_path_of_user.json | python3 -m json.tool
{
    "exec_time": 0.0009329319000244141,
    "ser_time": 0.0006611347198486328,
    "errors": [],
    "name": "insights.specs.Specs.ld_library_path_of_user",
    "results": {
        "object": {
            "relative_path": "insights_commands/echo_user_LD_LIBRARY_PATH"
        },
        "type": "insights.core.spec_factory.DatasourceProvider"
    }
}

The content:

# insights-inspect insights.specs.Specs.ld_library_path_of_user /tmp/insights-vm37-39.gsslab.pek2.redhat.com-20210308120914

In [1]: type(ld_library_path_of_user)
Out[1]: insights.core.spec_factory.SerializedRawOutputProvider

In [2]: type(ld_library_path_of_user.content)
Out[2]: bytes

In [3]: ld_library_path_of_user.content
Out[3]: b'rh1adm /usr/sap/RH1/SYS/exe/run:/usr/sap/RH1/SYS/exe/uc/linuxx86_64:/sapdb/clients/RH1/lib\nsr1adm /usr/sap/SR1/HDB02/exe/krb5/lib/krb5/plugins/preauth:/usr/sap/SR1/HDB02/exe/krb5/lib:/usr/sap/SR1/HDB02/exe:/usr/sap/SR1/HDB02/exe/Python/lib:/usr/sap/SR1/HDB02/exe/filter:/usr/sap/SR1/HDB02/exe/dat_bin_dir:/usr/sap/SR1/HDB02/exe/plugins/afl:/usr/sap/SR1/HDB02/exe/plugins/lcapps:/usr/sap/SR1/HDB02/exe/plugins/repository:/usr/sap/SR1/HDB02/exe/plugins/epmmds:/usr/sap/SR1/SYS/global/hdb/federation:/usr/sap/SR1/SYS/global/hdb/plugins/3rd_party_libs\nrh2adm /usr/sap/RH2/SYS/exe/run:/usr/sap/RH2/SYS/exe/uc/linuxx86_64:/sapdb/clients/RH2/lib'

I'm unsure if it's OK to modify the load method to read the file to a list directly:

self.loaded = True
with open(self.path, 'r') as f:
        return f.readlines()

@csams , @bfahr - would you please have a look?

[csams]

Hey @xiangce we can change the SerializedRawOutputProvider in the deserializer to a SerializedOutputProvider.

The raw providers exist in case we ever need to collect binary data instead of text. I defaulted datasource providers to raw binary since it was most general, but we really need datasource functions to be able to return different objects for text and raw bytes.

[bfahr]

Tested with mocked cloud_cfg. Results before change:

$ insights cat insights.specs.Specs.cloud_cfg insights-devboxseven-20210323173222.tar.gz 
SerializedRawOutputProvider("'/tmp/insights-9yjamkj8/insights-devboxseven-20210323173222/data/etc/cloud/cloud.cfg'")

After change:

$ insights cat insights.specs.Specs.cloud_cfg insights-devboxseven-20210323173222.tar.gz 
SerializedOutputProvider("'/tmp/insights-g2h3aonk/insights-devboxseven-20210323173222/data/etc/cloud/cloud.cfg'")
{"version": 1, "config": [{"subnets": [{"type": "dhcp"}, {"type": "dhcp6"}], "type": "physical", "name": "eth0"}]}

$ insights inspect insights.specs.Specs.cloud_cfg insights-devboxseven-20210323173222.tar.gz 

IPython Console Usage Info:

Enter 'cloud_cfg.' and tab to get a list of properties 
Example:
In [1]: cloud_cfg.<property_name>
Out[1]: <property value>

To exit ipython enter 'exit' and hit enter or use 'CTL D'

Starting IPython Interpreter Now 

In [1]: type(cloud_cfg)
Out[1]: insights.core.spec_factory.SerializedOutputProvider

In [2]: type(cloud_cfg.content)
Out[2]: list

In [3]: cloud_cfg.content
Out[3]: ['{"version": 1, "config": [{"subnets": [{"type": "dhcp"}, {"type": "dhcp6"}], "type": "physical", "name": "eth0"}]}']