Kruti Dev 010 to Unicode Troubleshooting Guide
This is a troubleshooting resource for confirmed Kruti Dev 010 text. It explains how to diagnose wrong punctuation, dates, fragmented Word runs and mapping/version mistakes, and why simply selecting Mangal in the font menu does not convert legacy text to Unicode.
Changing the display font to Mangal is not conversion
Kruti Dev 010 text stores legacy character codes whose intended Hindi appears only under the matching legacy font. Selecting Mangal changes how those same stored codes are drawn; it does not rewrite them as Unicode Devanagari.
A real conversion changes the underlying text encoding. After conversion, the Unicode result can be displayed in Mangal, Nirmala UI, Noto Sans Devanagari or another Unicode font without changing the text semantics.
Confirm Kruti Dev 010 before diagnosing characters
SahayakTools verifies the Kruti Dev 010 mapping. Kruti Dev 020, 030 and unspecified variants are not assumed compatible, even when common words or keyboard habits look similar.
The repository's independent regression work found that a neighbouring DevLys encoder could not safely stand in for the exact Kruti mapping for every character. A mapping/version mistake can therefore survive simple samples and still corrupt less common text.
Use a keyboard reference to inspect expected Kruti Dev 010 positions, but do not treat a short layout match as proof that an unknown font version is compatible.
Exact punctuation and date regressions
The verified Kruti Dev 010 forward mapping includes Hkkjr → भारत. It also proves that the legacy @ glyph position represents a visible Unicode slash /.
The production regression 25@05@2016 → 25/05/2016 is locked with structured-token protection both on and off. The reverse direction must encode Unicode 25/05/2016 as 25@05@2016 inside a Kruti Dev 010 run and recover the original date on decode.
Other exact punctuation vectors include - → ., ] → , and A → ।. These are why a converter cannot globally preserve every ASCII-looking character before it knows whether that character belongs to a legacy font run.
Word run fragmentation can split one Hindi sequence
Word may split one legacy sequence across adjacent spans because of formatting history rather than language boundaries. A verified regression covers the Kruti Dev run split f + d, which still has to recover कि when both pieces belong to the same confirmed source run.
The generated fragmentation suite also checks examples such as क्ष, भारत, प्रेम, कृपा, साँवरिया and क्षेत्र at every possible encoded split. Converting each span independently can detach a matra or break a conjunct even though the text looked continuous in Word.
Paste from Word with formatting so explicit Kruti Dev 010 runs can be joined semantically. Normal-font English, numbers and already-Unicode Hindi should remain separate.
Diagnose the failure before editing output
Wrong everywhere: verify the exact font/version and make sure the source is legacy Kruti Dev 010 rather than Unicode text displayed in a similar typeface.
Dates show @ instead of /, or punctuation is shifted: confirm that the affected bytes are inside the Kruti Dev 010 source run and were not protected as ordinary Latin text.
Matras or conjuncts break at formatting changes: return to the rich Word source so adjacent legacy spans can be processed together.
English, IDs or numbers change: inspect font-run boundaries. Repository mixed-document regressions preserve values such as QID-2026, A-17, 15/08/2026, English, 10% and ₹500 outside legacy font runs where appropriate.
A few rare characters fail while common words work: suspect an incorrect mapping/version before manually correcting the whole document.
Safe conversion and reverse workflow
1. Keep an untouched copy and verify Kruti Dev 010 in the original document's affected font runs.
2. Paste with formatting into the Kruti Dev 010 to Unicode converter and review exact names, dates, IDs, punctuation and conjuncts.
3. Use the Unicode result with Mangal or another Unicode Devanagari font only after the underlying text has been converted.
4. For Unicode back to Kruti Dev 010, use Copy for Word when mixed Hindi, English, numbers or unsupported symbols require separate font runs. Raw text cannot carry those assignments.
When to stop
Stop rather than guess when the source version is unknown, Word no longer exposes reliable font runs, or important output cannot be checked against a correctly rendered original.
Do not use readable-looking output, a changed display font or a successful short sample as proof that an unknown mapping is safe for an entire official or archival document.