fix: snap chunk start to a word boundary in CharacterBasedTextChunker

[inesaranab]

Change Description

CharacterBasedTextChunker snaps end forward to the next boundary char so a chunk never ends mid-word. But start, computed as end - chunk_overlap, gets no equivalent treatment — the subtraction discards the alignment established two lines earlier.

With chunk_size=10, chunk_overlap=3:

text = "This is a test string for chunking purposes"

before: ['This is a test', 'st string for', 'or chunking', 'ing purposes']
after: ['This is a test', ' string for', ' chunking purposes']

Three of the four chunks begin mid-word. An NER recognizer receives 'st', 'or', 'ing' as standalone tokens and can confidently mislabel them as entities, producing a false positive that redacts the wrong span.

Fix: advance start to the next boundary char, mirroring the loop end already has. Three lines.

Present since #1805, which introduced this chunker.

Issue reference

No existing issue. Found while using the chunker via GLiNERRecognizer on long transcripts.

Checklist

• I have reviewed the contribution guidelines
• I agree to follow this project's Code of Conduct
• I confirm that I have the right to submit this contribution and that it does not knowingly contain proprietary or confidential code.
• My code includes unit tests
• All unit tests and lint checks pass locally
• My PR contains documentation updates / additions if required

[Copilot]

Pull request overview

This PR improves CharacterBasedTextChunker to avoid starting overlapped chunks in the middle of words by snapping the computed start forward to a configured boundary character (mirroring the existing logic which snaps end forward), reducing the chance of NER false positives on partial tokens.

Changes:

• Adjust CharacterBasedTextChunker.chunk() to advance start to the next boundary character after applying chunk_overlap.
• Add a unit test asserting that every chunk after the first begins on (or immediately after) a boundary.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
presidio-analyzer/presidio_analyzer/chunkers/character_based_text_chunker.py	Advances start to a boundary char after overlap to prevent mid-word chunk starts.
presidio-analyzer/tests/test_character_based_text_chunker.py	Adds coverage to ensure chunk starts (after the first) land on a word boundary.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[omri374]

Approved! Thanks