================================================================================
KHMER PDF UNICODE & TOKENIZATION ANALYSIS REPORT
================================================================================

PDF: 4a798110-ca16-42ac-98fb-b4416e420889.pdf
Total characters: 42,150
Khmer ratio: 82.4%

================================================================================
1. UNICODE NORMALIZATION
================================================================================
Is NFC normalized: True
✅ Text is properly NFC normalized

================================================================================
2. GRAPHEME CLUSTER VALIDATION
================================================================================
⚠️  ISSUE: Found 537 orphaned COENG characters

Sample occurrences:
   Position 63: U+17D2
   Context: វស្រាវគរុកោសលយជ្ំនាន់ថ្មី 
New 
   Position 415: U+17D2
   Context: ាយ ឬបញ្ចូលកែុងទ្្មង់ណាមួយ ឬតាមរ
   Position 518: U+17D2
   Context: សមកា
និកសនេងៗសទ្ៀត សោយមិនមានការ
   Position 617: U+17D2
   Context: ាន់ៗ ្មាប់េតថបទ្
វិចារណកថា ឬការ
   Position 822: U+17D2
   Context: s Kowalski 
បកម្្បជាភាសាម្មារ៖ 

================================================================================
3. INVISIBLE CHARACTERS
================================================================================
✅ No invisible characters detected

================================================================================
4. TOKENIZATION ANALYSIS
================================================================================
✅ Tokenizer loaded successfully
   Vocabulary size: 8,000
   Token count: 27,952
   Unique tokens: 736
   Avg token length: 1.64 chars
   TPC (tokens/char): 0.663

⚠️  Found 14147 broken tokens
   Sample: ['-', '-', '-', 'A', 'G', 'E', '1', '-', '-', '-']

================================================================================
5. CHARACTER DISTRIBUTION
================================================================================
Total: 42,150
Khmer: 34,744 (82.4%)
Latin: 416

⚠️  Found 14 suspicious characters:
   U+002D (HYPHEN-MINUS): 289 times
   U+00A9 (COPYRIGHT SIGN): 1 times
   U+0022 (QUOTATION MARK): 24 times
   U+005B (LEFT SQUARE BRACKET): 1 times
   U+005D (RIGHT SQUARE BRACKET): 1 times
   U+0028 (LEFT PARENTHESIS): 35 times
   U+0029 (RIGHT PARENTHESIS): 40 times
   U+00AB (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK): 1 times
   U+00BB (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK): 1 times
   U+2756 (BLACK DIAMOND MINUS WHITE X): 4 times

================================================================================
6. SUMMARY & RECOMMENDATIONS
================================================================================
⚠️  ISSUES FOUND:
   - 537 broken clusters

RECOMMENDED FIXES:
   3. Fix orphaned COENG characters
      Review manually or use Khmer text validator

================================================================================