Sangeetha-Grantha

Metadata	Value
Status	Active
Version	1.3.0
Last Updated	2026-09-10
Author	Sangeetha Grantha Team
Document Type	Evidence record

Trinity Krithi Bulk Import

---

[!NOTE] Historical evidence: results, counts, commands, and observations below belong to the original work described here. The editorial update date is not a new test or corpus verification. For present behavior, use current feature map.

Purpose

Import 1,245 krithis for the three Carnatic Trinity composers (Muthuswami Dikshitar: 482, Syama Sastri: 71, Tyagaraja: 692) via the bulk import UI, with supporting raga seed migrations and scraping pipeline fixes.

Implementation Details

CSV Cleanup

• Removed artifact rows where Raga column contained “Part 1-5” or repeated header rows
• Primarily affected the Thyagaraja CSV

Raga Gap Seeding (Migrations 39–40)

• 962 ragas already in DB from TRACK-091/092
• 164 CSV raga names didn’t exact-match after normalization; 27 LOW-confidence matches needed seeding
• Migration 39: 39__seed_malavasri_raga.sql — Mālāvashree raga
• Migration 40: 40__seed_missing_trinity_ragas.sql — 9 base ragas (Kalyāni, Todi, Jujāvanti, Gaula, Nāta, Bauli, Pūrvi, Gauri, Brindāvana Sāranga)

Scraping Pipeline Fix (v1.1.0)

Kotlin-side HTML scraping produced corrupted lyrics (Devanagari pronunciation guides, Word Division duplication, multiple language versions). Fixed by delegating entirely to Python extraction worker:

• ScrapeWorker now passes only CSV metadata, leaving rawLyrics/rawPayload null
• Triggers ImportService.shouldEnqueueHtmlExtraction() → extraction_queue → Python worker
• StructureParser enhanced with Devanagari pronunciation guide boilerplate detection

Raga Transliteration Handling (v1.2.0)

• Normalization pipeline: diacritic removal, vowel length collapse, aspirate collapse, space removal
• RapidFuzz WRatio fuzzy matching with confidence levels (HIGH 90+, MEDIUM 70-89, LOW 60-69)

Code Changes

File	Change
database/for_import/*.csv	Cleaned CSV source files
database/migrations/39__seed_malavasri_raga.sql	Mālāvashree raga seed
database/migrations/40__seed_missing_trinity_ragas.sql	9 base raga seeds
database/migrations/41__unique_import_source_name.sql	Fixed concurrent import source creation bug causing duplicate krithis
modules/backend/api/.../bulkimport/workers/ScrapeWorker.kt	Delegate scraping to Python worker
tools/krithi-extract-enrich-worker/src/structure_parser.py	Boilerplate detection

Results

• Three-batch import strategy: Syama Sastri → Dikshitar → Thyagaraja
• All transliteration variants resolve correctly via fuzzy matching
• Clean lyrics produced through Python extraction pipeline

Section-count mismatches left by the Trinity import are closed in TRACK-133 (108 → 0 rows; durable pallavi-echo parser 2026-09-05).

Ref: application_documentation/10-implementations/track-093-trinity-krithi-bulk-import.md

---

Section index · Documentation home · Feature status