EMu XML-to-JSON converter for cultural data

A script to convert EMu XML exports to JSON for cultural data. Note - For record-sets > 1000, this can be slow. It is based on a combination of standards: Darwin core, Latimer core, Audubon Core and the very draft H2I extension to ABCD. We are currently testing it as a sustainable way to share data with partners from Digital Benin and Mapping Philippine Material Culture

How to setup & use

In EMu:

1. Export a set of records as XML.

   • Catalog export of record data called: H2I_JSON_Export_DarwinCore
   • Catalog export of associated Multimedia called: H2I_JSON_Export_AudubonCore

2. When possible, check that the XML output is well-formed.

   • 'Scholarly XML' VSCode add-on
   • xmllint on Mac/Unix/Linux or xsltproc on Windows -- Try this in terminal/shell:
     • xmllint --noout file.xml && echo $?
   • Online XML validator Warning - avoid online-validators for sensitive data.

3. Set up these two CSV's following instructions/examples from the 'Input' section:

   • emu_fields.csv - to map EMu-column-names to corresponding standard-term names.
   • emu_conditions.csv - for EMu-fields and values that need conditional mapping or redaction.

Not in EMu:

1. Clone this repo

2. Set local environment variables by adding a text file named .env in the root directory of this repo. Open it and follow the .env.example file. More info here if needed.

   • IN_PATH = the path to your input XML file
   • MAP_PATH = the path to your emu_conditions.csv
   • OUT_PATH & LOG_PATH = the path where you want the output JSON and log files to go
   • FROM_ADD & TO_ADD1 = the sender and recipient email addresses for notifications
   • other examples-variables are included for using 'mutt to send notifications from a server

3. Install Python 3.9 or later. To send email notifications from a server, also install mutt. (e.g. Ubuntu wiki)

4. Install the python packages listed in required.txt with pip or pip3:

   pip3 install charset-normalizer json pandas python-decouple xml xmltodict

5. Run the script: python3 emu_xml_to_json.py

   • The newest file in your .env file's 'IN_PATH' is the default XML-input.

   • Alternatively, you can manually specify a different XML-input like so:

     python3 emu_xml_to_json.py data_in/2021-08-08/sample.xml

6. Output JSON, XML and log are zipped and emailed. See JSON output in emu_to_json.json, or check for errors in xml_log_YYYYMMDD.txt

Input

EMu XML Export

An XML file containing records exported from EMu as XML, with some or all EMu-fields listed in emu_fields.csv

• example here of well-formed input
• example here of input with a badly-encoded character

emu_fields.csv (example)

A 5-column CSV that maps EMu-column-names to corresponding standard-term names, using the following columns:

• emu = EMu column-names
  • Exclude any "Ref" and "_tab" suffixes here. (This should match the lowest-level tag name in the EMu XML.)
• json_field = corresponding h2i standard term names
• repeatable = blank or 'yes' to indicate if multiple values can be assigned to json_field
• emu_group = in the EMu export's 'Group' name, or the table or Reference column name
  • Include the "Ref" and/or "_tab" suffix (if any) for the corresponding emu field.
• json_container = the group name for a set of json_fields that should be nested together in the output JSON

emu_conditions.csv (example)

A 7-column CSV that defines logic for conditionally redacting or mapping rows in multi-value-tables to standard terms.

• if_field1 = the input EMu-field whose value defines a condition
• if_logic1 = the logical comparison for the condition (e.g. if the field "IS" or "IS NOT" equal to if_value1)
• if_value1 = the input value.
  • Use "NOT NULL" for "any non-blank input value"
• then_field = the input EMu-field (if any) whose value should be transformed or redacted.
• json_field = the output json_field that should be set (conditionally) to the value in static_value.
  • Use "NULL" to redact an output value from the then_field
• static_value = the output value used if an input field matches conditions in the if_field1 & if_value1
• json_container = the output field's group, if any

Output

• emu_to_json.json = Records as JSON objects, with EMu-fields/data as key/value pairs.
• emu_prepped.xml (optional) = XML with EMu column-names as xml-tags instead of xml-attributes
• xml_log_YYYY-MM-DD.txt = Log of successful or failed output.
• email notifications (currently requires mutt to send email, or gmail-only recipients)
  • Log-messages comprise the email-body
  • Output files are zipped and attached