HotPDF är ett nativt VCL PDF-bibliotek för Delphi- och C++Builder-program som behöver direkt PDF-skapande och redigering, formulär, annoteringar, kryptering, digitala signaturer, Unicode-teckensnitt, standardmedveten utdata och preflight-rapporter utan extern PDF-runtime.
Den här artikeln är skriven för developers producing multilingual invoices, certificates, labels, or reports from Delphi. Den behandlar Unicode text shaping for complex scripts som produktionsnära dokumentteknik, inte som ett isolerat komponentanrop.
Den praktiska risken är att text can appear plausible in a sample PDF while ligatures, bidirectional order, fallback fonts, or copy-and-search behavior fail for real customer names. Därför behöver flödet ett skrivet kontrakt, observerbar diagnostik och realistiska regressionsfiler.
Arkitekturbeslut
Make the text pipeline locale-aware. font fallback order for Arabic, Hebrew, Indic, CJK, and mixed Latin text / normalization rules for copied text, database values, and template placeholders
- font fallback order for Arabic, Hebrew, Indic, CJK, and mixed Latin text
- normalization rules for copied text, database values, and template placeholders
- right-to-left paragraph handling and mixed-direction number policy
- whether text must remain searchable, selectable, and accessible after output
Implementeringsflöde
Resolve fonts and shaping before pagination. The order below keeps the workflow reviewable for Delphi and C++Builder teams.
- normalize source text and record the locale used for formatting
- select fonts that contain the required glyphs before measuring layout
- shape and position text before page breaks are finalized
- embed or subset fonts according to licensing and PDF standard requirements
- verify visual output and extracted text with multilingual regression samples
Valideringsbevis
Proof that text is readable and extractable. Keep these fields with the output or support record.
- font selected for every script range and fallback reason when it changed
- glyph coverage warnings, embedding mode, and subset identifier
- extracted Unicode text compared with the original application value
- viewer screenshots for representative right-to-left and combining-mark cases
Visual output is not enough
Complex-script support involves character normalization, shaping, glyph positioning, embedding, ToUnicode maps, and reading order. A PDF that only looks right in one viewer can still fail search, selection, accessibility, or downstream extraction.
Regression files worth keeping
Keep more than successful samples. A useful Unicode text shaping for complex scripts regression set contains normal files, boundary files, and intentional failure files so the behavior is stable across releases.
- database collation can alter composed characters before the PDF layer sees them
- font substitution on a developer machine can hide missing embedded fonts
- line breaks in bidirectional text can reorder punctuation and numbers
- search may fail when ToUnicode data is missing even if the page renders correctly
- normalize source text and record the locale used for formatting
- select fonts that contain the required glyphs before measuring layout
Engineering review notes for Unicode text shaping for complex scripts
Use these review notes to make sure the feature has moved beyond a demo and can be defended during release, support, and customer escalation.
- Decision: font fallback order for Arabic, Hebrew, Indic, CJK, and mixed Latin text. Implementation pressure point: select fonts that contain the required glyphs before measuring layout. Acceptance evidence: extracted Unicode text compared with the original application value. Regression trigger: search may fail when ToUnicode data is missing even if the page renders correctly
- Decision: normalization rules for copied text, database values, and template placeholders. Implementation pressure point: shape and position text before page breaks are finalized. Acceptance evidence: viewer screenshots for representative right-to-left and combining-mark cases. Regression trigger: database collation can alter composed characters before the PDF layer sees them
- Decision: right-to-left paragraph handling and mixed-direction number policy. Implementation pressure point: embed or subset fonts according to licensing and PDF standard requirements. Acceptance evidence: font selected for every script range and fallback reason when it changed. Regression trigger: font substitution on a developer machine can hide missing embedded fonts
- Decision: whether text must remain searchable, selectable, and accessible after output. Implementation pressure point: verify visual output and extracted text with multilingual regression samples. Acceptance evidence: glyph coverage warnings, embedding mode, and subset identifier. Regression trigger: line breaks in bidirectional text can reorder punctuation and numbers
Gränsfall
- database collation can alter composed characters before the PDF layer sees them
- font substitution on a developer machine can hide missing embedded fonts
- line breaks in bidirectional text can reorder punctuation and numbers
- search may fail when ToUnicode data is missing even if the page renders correctly
Delphi / C++Builder notes
HotPDF Component should sit behind a small service boundary that receives files, streams, profiles, and credentials, then returns output paths, warnings, metrics, and validation status. Important terms include Unicode, text shaping, font embedding, ToUnicode, bidirectional text, fallback font.
Delphi-kodexempel
Följande Delphi-skiss visar en praktisk servicegräns för detta ämne. Håll policykontroller, loggning och validering utanför det smala produktanropet så att arbetsflödet går att testa.
procedure DrawShapedRun(Pdf: THotPDF; const Text: UnicodeString; const Script: TScriptProfile);
begin
Pdf.CurrentPage.SetFont(Script.FontName, [], Script.Size, 0, Script.Vertical);
if Script.RequiresReorder then
Pdf.CurrentPage.TextOut(Script.X, Script.Y, 0, ShapeUnicodeRun(Text, Script))
else
Pdf.CurrentPage.TextOut(Script.X, Script.Y, 0, Text);
RecordGlyphCoverage(Script.FontName, Text);
end;Produktionschecklista
- Run the workflow on an empty file, a normal customer file, and a worst-case file
- Open the generated PDF with the target viewer, validator, printer, or downstream application
- Log product version, profile version, input hash, output path, elapsed time, and warning count
- Keep passwords, certificates, temporary files, and customer data under explicit retention rules
- Add regression documents when a customer file exposes a new edge case