Setting a body:summary without a body erases the body and resets the format when updating a node

Problem/Motivation

Given a content type with a field of type text_with_summary (e.g.: the body field, which Drupal adds by default) If I map a source field to the summary field (e.g.: body:summary), but I do not also map a source field to the main part of the field (e.g.: body), then the main part of the field is erased upon import.

Steps to reproduce

1. Download the latest version of Drupal 7 core, the latest version of the Chaos tools module (ctools-7.x-1.11 at time of writing), and the latest version of the Job Scheduler module (job_scheduler-7.x-2.0-alpha3 at time of writing). Clone the Feeds module, branch 7.x-2.x.
2. Install Drupal using the minimal install profile. Enable Field UI, Feeds, Feeds UI, and their dependencies.
3. Go to /admin/structure/types and create a content type. I named mine "Test content type" (test_content_type) and accepted all the default options.
4. Note that Drupal automatically adds a body field to the new content type.
5. Go to /node/add and create a new node of the new content type. Enter something distinct into the Title, Summary, and Body fields. I used Lorem for the Title, Ipsum for the Summary, and Dolor for the Body. Take note of the node ID (mine had a node ID of 1).
6. Go to /admin/structure/feeds and create a Feeds importer. I named mine "Test importer" (test_importer), and used the following options:
   1. Basic settings:
      • Attach to content type: Use standalone form
      • Periodic import: Off
      • Import on submission: Checked
      • Process in background: Unchecked
   2. Fetcher: File upload
      • Allowed file extensions: txt csv tsv xml opml
      • Immediately delete uploaded file after import: Unchecked
      • Supply path to file or directory directly: Unchecked
      • Upload directory: private://feeds
   3. Parser: CSV parser
      • Default delimiter: ,
      • No headers: Unchecked
      • File encoding: UTF-8
   4. Processor: Node
      Settings:
      • Bundle: Test content type
      • Insert new nodes: Insert new nodes
      • Update existing nodes: Update existing nodes
      • Skip hash check: Checked
      • Text format: Plain text
      • Action to take when previously imported nodes are missing in the feed: Skip non-existent nodes
      • Author: anonymous
      • Authorize: Unchecked
      • Expire nodes: Never

      Mapping:

      Source	Target	Target configuration
      nid	Node ID (nid)	Used as unique.
      Title	Title (title)	Not used as unique.
      Summary	Body: Summary (body:summary)
      Body	Body (body)	Text format: Plain text

7. Create a CSV file on your local machine. The header row must match the Sources in the Mapping you set up earlier. The nid field must match the node ID you took note of earlier when you created the node. The rest of the data should be different from the data you entered when you created the node. For example, my CSV file was:
   

   nid,Title,Summary,Body
   1,Foo,Bar,Baz
   

8. Go to /import. Use the default importer options, but upload the CSV file you created in the previous step, and click the Import button.
9. When the import is complete, edit the node you created earlier. Note that the Title matches the value you put in the CSV (e.g.: mine was "Foo"). Note that the Summary matches the value you put in the CSV (e.g.: mine was "Bar"). Note that the body matches the value you put in the CSV (e.g.: mine was "Baz"). Note this likely matches your expectations.
10. Edit the CSV file on your local machine. Remove the Body mapping, and change the Title and Summary to different values. For example, after editing, my CSV file was:
    

    nid,Title,Summary
    1,Fizz,Buzz
    

11. Go to /admin/structure/feeds and edit the importer you created earlier. Go back to Processor -> Mappings. Remove the "Body" Mapping by checking the "Remove" checkbox in the last column of the table, and clicking the Save button at the bottom. Verify that the Body mapping no longer exists.
12. Go to /import. Use the default importer options, but (re-)upload the CSV file you modified, and click the Import button.
13. When the import is complete, edit the node you created earlier. Note that the Title matches the value you put in the CSV (e.g.: mine was "Fizz"). Note that the Summary matches the value you put in the CSV (e.g.: mine was "Buzz"). Note that the body is empty! Note that this likely does not match your expectations: I would expect it to be the value of the Body field before the import (i.e.: "Baz").

Analysis

The text_with_summary field has three (sub-)parts: a summary, a value and a format.

When text_feeds_set_target() builds a new value for the summary (sub-)part of the field, it modifies the entity loaded on line 280 of \FeedsProcessor::process() (that line calls \FeedsProcessor::entityLoad()).

In \FeedsProcessor::map(), there is code to reset the value of a field before data is mapped to it. However, because the definition for the summary field target has a 'real_target' property, \FeedsProcessor::map() resets the whole field, not just the summary.

Proposed resolution

Remove the 'real_target' property so that \FeedsProcessor::map() doesn't reset everything in the field.

Remaining tasks

1. Write and upload a patch
2. Review and feedback
3. RTBC
4. Commit

User interface changes

None.

API changes

None.

Data model changes

None.