fix(kb): route plain text uploads to TextParser without markitdown

[xiaoyuyu6420]

Fixes #9598

Problem

select_parser sent .txt / .md / .markdown to MarkitdownParser, which imports the optional markitdown-no-magika dependency at module level. When that dependency is not installed (e.g. manual deployments), uploading a plain text file raises ModuleNotFoundError, which kb_helper swallows into a generic "文档解析失败" message. The user gets no hint that the root cause is a missing optional dependency, and plain text uploads simply fail.

Fix

Plain text formats are decoded directly by the existing TextParser, which needs no extra dependency and auto-detects multiple encodings (utf-8 / gbk / gb2312 / gb18030):

• .txt / .md / .markdown → TextParser
• .rst / .adoc / .xlsx / .docx / .xls → MarkitdownParser (unchanged; these rely on conversion)

Chunking in kb_helper already selects MarkdownChunker by file extension, so markdown heading hierarchy is preserved — the parser swap does not change downstream behavior for .md files.

Testing

• uv run pytest tests/unit/test_parser_selection.py — 3 passed (new)
• uv run pytest tests/test_epub_parser.py tests/unit/test_kb_upload_atomicity.py tests/unit/test_parser_selection.py — 26 passed
• ruff format / ruff check clean

Summary by Sourcery

Bug Fixes:

• Route plain-text knowledge-base uploads through TextParser so .txt, .md, and .markdown files work without the optional Markitdown dependency.

[sourcery-ai]

Hey - I've found 1 issue

Prompt for AI Agents

Please address the comments from this code review:

## Individual Comments

### Comment 1
<location path="tests/unit/test_parser_selection.py" line_range="3" />
<code_context>
+import pytest
+
+from astrbot.core.knowledge_base.parsers.markitdown_parser import MarkitdownParser
+from astrbot.core.knowledge_base.parsers.text_parser import TextParser
+from astrbot.core.knowledge_base.parsers.util import select_parser
</code_context>
<issue_to_address>
**issue (testing):** The test module imports `MarkitdownParser` at collection time, so `ModuleNotFoundError` is raised before any test runs when the optional `markitdown-no-magika` dependency is absent—the exact deployment scenario this change is intended to support.

**Triggers:** When running the test suite in an installation without `markitdown-no-magika`.

**Suggested fix:** Import `MarkitdownParser` lazily inside the rich-format test and skip that test when the optional dependency is unavailable, or mock the parser module during selection testing.
</issue_to_address>

Sourcery assessment

Needs a human reviewer. 1 finding to address first, and if the routing is wrong, .txt or Markdown files could be parsed into incorrect knowledge-base content and embeddings, leaving bad indexed records after a revert. The records are bounded and can be cleared or regenerated, so reverting plus re-ingestion repairs the effect.

Blocking findings: tests/unit/test_parser_selection.py:3

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}