Nativ: missing InDesign text diagnostic exercise, 5 November 2026

Download check_idml_text.py and literal-story-exercise.zip to the same folder.
Requires Python 3.10+. No dependencies, network requests or archive extraction.

Try:
python3 check_idml_text.py literal-story-exercise.zip --contains "Keep the final sentence."
Expected: 1 matching story (udemo), 1 phrase match, layoutChecked false.
The phrase is split across two character-formatting runs in the XML.

Then:
python3 check_idml_text.py literal-story-exercise.zip --contains "This sentence is missing."
Expected: 0 matching stories. No claim of deleted or failed translation follows.

The ZIP is a manually authored text-parser fixture, NOT an InDesign document.
It has no complete spreads, fonts or geometry. Do not open it in InDesign.
It is not a generated translation, an overset-text detector or a repaired file.
On a real IDML copy, replace the ZIP filename and choose an exact expected
TARGET-language phrase from your approved translation. Phrase search is
case-sensitive and does not normalize whitespace, punctuation or Unicode.

Limits:
- Only literal Content in story files referenced by designmap.xml is inspected.
- No story-to-page placement, text-frame geometry or overset is calculated.
- Hidden conditions, tracked changes, notes and variables are not resolved.
- Text in images, outlined type, linked InCopy and XML backing stories is not
  comprehensively covered; this is not a text extractor or IDML validator.
- Boundaries in cells/notes/anchored frames are conservatively separated.
- Zero matches can mean changed wording, punctuation, encoding or unsupported
  content, not necessarily lost text. Use Story Editor and a human reviewer.
- Reads are bounded (200 MiB ZIP, 8 MiB per XML part, 64 MiB XML total,
  1000 referenced stories). Unsupported input is refused, not repaired.
- The command prints counts and story identifiers, not the full document text.
  Keep reports containing project identifiers within your normal review team.
- Read on a copy; the program does not write to its input or make translations.