Prefer social text CSV columns

[kriptoburak]

Summary

• Support Xquik exports using the canonical text field.
• Prefer specific social aliases such as review_text, tweet_text, and full_text before a generic text column.
• Normalize spaces, underscores, hyphens, and letter case.
• Remove blank text rows without mutating the source dataframe.
• Make the application safe to import in headless and test environments.

Xquik integration

• Document direct compatibility with Xquik’s canonical text field.
• Support common transformed social-export aliases.
• Keep manual selection available for every other column.
• Add the exact public affiliation notice.

Independent repository repairs

• Move Tk window creation and mainloop() behind main() so imports do not open or block on a GUI.
• Load VADER lazily and download its lexicon only when first use proves it missing.
• Preserve background-thread exception text until Tk executes the queued callback.
• Correct stale README references to TextBlob, unused NLTK datasets, and incorrect sentiment thresholds.

Validation

• Clean Python 3.12 dependency environment: 6 tests passed
• Python compilation passed for runtime and tests
• Ruff lint and complexity checks passed
• README links pass except LinkedIn’s automated-client HTTP 999 response
• Exact public notice validation passed
• git diff --check passed
• Credential-shape and prohibited punctuation scans passed
• Simplification and maintainability reviews passed
• Exactly 1 SSH-signed commit with a valid GitHub signature

Repository state

• Current upstream main is the commit parent.
• Existing MIT license is preserved.

Xquik is an independent third-party service. Not affiliated with X Corp. "Twitter" and "X" are trademarks of X Corp.

[kriptoburak]

Repair consolidated in signed commit 3c7e7d0.

Xquik exports using the canonical text field work directly. Specific social aliases such as review_text, tweet_text, and full_text still outrank a generic text column when transformed datasets include both, while spacing, punctuation, and case variants remain supported.

Independent fixes make analyzer.py safe to import without downloads or a GUI, load VADER lazily, preserve threaded exception messages, and correct stale TextBlob setup and scoring documentation. Six focused tests, compilation, Ruff lint and complexity, README link validation aside from LinkedIn HTTP 999, notice checks, and diff checks pass. The PR is clean and mergeable with no external feedback or remaining contributor work.