Loaded Document Toolkit

HotPDF can open an existing PDF, apply page and form operations, inspect typed document state, render or search pages, print with device capabilities, and save the result without moving the workflow to another library

Page operations and merging

• THotPDF.DeletePages accepts the standard one-based page-range syntax, removes duplicate selections once, and preserves a valid page tree
• Deleting loaded pages detaches widget annotations and empty AcroForm field branches that belonged only to the deleted pages
• THotPDF.InsertPagesFromDocument and THotPDF.AppendPagesFromDocument accept THPDFMergeOptions
• Form-name conflicts are checked before page insertion when automatic renaming is disabled, so a rejected merge leaves the destination page count unchanged
• THPDFMergeOption values are moIncludeOutlines, moIncludeFormFields, and moRenameConflictingFormFields
• Outline destinations are remapped through the selected source-to-destination page map, while imported form roots are registered in the destination AcroForm
• THotPDF.InsertLoadedBlankPage inserts a caller-sized blank page at an exact zero-based position
• THPDFWatermarkTemplate.MakeCurve adds a cubic Bezier curve to ordered composite loaded-page watermarks
• THPDFPage.AddCurveWatermark provides the same cubic Bezier primitive during new-document authoring
• Loaded document metadata setters mark the Info dictionary dirty so incremental saves retain edited fields
• BeginIncrementalUpdate keeps the source file write-locked and file-backed with a bounded tail probe instead of retaining the complete file in memory
• SaveIncrementalUpdate visits only tracked changes, streams the source prefix to a new target, and uses validated rollback-protected direct append for the same target
• OpenLoadedPageContentStream exposes single and array /Contents values as one seekable logical stream, retains unfiltered segments in source-backed storage, and decodes at most one filtered segment at a time
• Editing a page whose /Contents value is an array appends a fresh indirect segment without coalescing or clearing the original streams, while rendering and compatibility materialization use geometrically growing token and byte buffers
• Loaded annotation setters update contents, title, unique name, modification date, rectangle, flags, color, and border style directly on the existing annotation object
• Annotation rectangle endpoints are normalized, dashed borders keep only positive dash lengths, and visual changes invalidate only the affected page's cached renders
• ActivateLoadedLinks detects bare web and email addresses from positioned glyph runs, creates borderless URI annotations, and skips matching existing links without rescanning the annotation list for every candidate
• ExtractLoadedPagesTextToStream writes UTF-8 page text directly to a caller-owned stream with range, separator, and BOM controls
• THPDFViewerModel exposes page form-field hit testing and deterministic top-to-bottom keyboard navigation without creating a window
• Loaded-signature verification supports RSA PKCS#1 v1.5, parameter-validated RSA-PSS, ECDSA, and RFC 3161 document and signature timestamps; strict byte-range checks report signed-revision length and unsigned trailing bytes while preserving caller stream position
• THPDFSignatureInfo.CertificateTimeValidity classifies certificate bounds and signed-time consistency without changing cryptographic svValid semantics
• VerifyLoadedSignatureWithTrust builds the signer path to caller-supplied trust anchors and validates offline OCSP or CRL evidence, preferring per-signature VRI data before document-level DSS and caller evidence
• AnalyzeLoadedSignatureRevisions walks traditional, stream, hybrid, and compressed-object xref histories, classifies later changes as long-term validation, form filling and signing, annotations, or other modifications, and evaluates DocMDP and FieldMDP independently
• SanitiseLoadedDocument applies JavaScript-only, active-content, or strict untrusted-document policies to the complete object graph, removes unreachable payload objects, and prepares a bounded compact rewrite
• ScrubLoadedRedactionRemnants removes metadata, attachments, Redact annotations, unreachable resources, and retained incremental source bytes before a compact full rewrite
• Bounded cross-reference recovery relocates intact xref sections or rebuilds validated direct-object offsets after damaged startxref and xref data, with caller-controlled scan budgets and structured diagnostics
• Scalable loading keeps file stream bodies source-backed until mutation, lazily resolves compressed objects, and accepts custom THPDFRandomAccessSource implementations
• DecodeLoadedStreamToStream runs bounded filter chains into caller-owned streams while document-wide byte accounting, filter-count limits, indirect-reference cycle checks, and pipeline-depth guards protect hostile inputs
• ExtractLoadedImageToStream copies native image payloads or generic decoded filter output directly into caller-owned streams without materializing a full-size byte array
• ExportLoadedPagesStructured serializes semantic reading order, tables, geometry, styles, and bounded image occurrences as HTML, XHTML, XML, or JSON
• ExtractLoadedTypedTables builds confidence-scored canonical grids with merged spans, repeated headers, cross-page continuation, typed values, and atomic CSV or JSON export
• OptimizeLoadedStreams selects unfiltered, Flate, LZW, ASCII85, ASCIIHex, and RunLength candidates, spills large work buffers to temporary files, can run budgeted Zopfli after a four-strategy zlib baseline, and adopts only replacements that meet a positive savings threshold
• SubsetLoadedFonts removes unreachable glyph programs from supported embedded TrueType and CFF fonts after tracing page, Form XObject, and annotation appearance content, while preserving existing character, CID, glyph, width, and ToUnicode mappings
• SetDocumentAttachmentMetadata renames an embedded file and updates its description before the document is saved
• THPDFLoadedFormField exposes flags, MaxLen, page index, bounds, and common text-field behaviors for property inspectors
• THPDFViewerModel creates caller-owned page snapshots and writes the current page as BMP, JPEG, or PNG to files and streams while reusing the loaded render cache
• THPDFViewerModel rotates the current view and generated snapshots in 90-degree steps without modifying the loaded PDF page
• THPDFViewerModel creates rotation-aware enhanced metafiles while retaining scalable text and vector drawing records
• THPDFViewer.DrawCurrentView and PrintCurrentView fit and center the active vector view on a caller canvas or printer page
• THPDFViewerModel.ProcessKeyInput provides testable page and zoom commands while the Viewer handles arrow-key scrolling
• THPDFViewerModel.ProcessMouseWheel provides testable Ctrl+wheel zoom while the Viewer handles vertical and horizontal scrolling directly
• THPDFViewerModel.GetNextZoomLevel and GetPreviousZoomLevel expose deterministic standard zoom navigation from 5% through 6400%
• THPDFViewerModel.PagePointToView and ViewPointToPage provide reversible coordinates across page rotation, view rotation, page box offsets, and DPI
• THPDFViewerModel.CreatePageRegionSnapshot renders only the requested PDF rectangle for bounded-memory selection snapshots
• THPDFViewer.ResetView restores presentation defaults without unloading the document or changing the current page

Forms and search

• THotPDF.AddLoadedRadioButton creates a parent radio field with child widgets, shared export values, and valid on/off appearance states
• Loaded form fields can update flags, maximum length, justification, tooltip, per-widget rectangles, rotation, visibility, printing, border width and color, and background color with incremental-save dirty tracking
• GetLoadedFormFieldNameValuePairs and SetLoadedFormFieldNameValuePairs exchange lightweight name=value batches through TStrings
• ExtractLoadedPagesText aggregates a one-based page range with a caller-selected separator
• CMapResourcePath overrides automatic CMap discovery with a caller-selected root containing the standard collection folders
• Assigning CMapResourcePath clears loaded-font decoders even when the path is unchanged; call HPDFClearCMapCaches first when replaced external resources must also be reloaded
• Standard CMap mappings are cached per resolved resource root, support concurrent table access, and use logarithmic lookup when character and range tables are ordered
• Complete UniJIS, UniGB, UniCNS, and UniKS UCS2 and UTF16 mappings are embedded for horizontal and vertical text, including supplementary scalars and standard vertical defaults when /DW2 is absent
• Loaded-font decoders use a document-scoped 32 MiB weighted LRU cache shared by text extraction, SVG export, text replacement, and explicit decoder requests
• GetLoadedFontDecoder returns an independent deep copy while internal read-only consumers reuse the cached parsed core
• LoadedFontDecoderCacheMaxBytes, GetLoadedFontDecoderCacheInfo, and ResetLoadedFontDecoderCacheStatistics control and measure cache reuse
• Low-resolution grayscale and RGB JPEG regions use reduced inverse-DCT scaling and direct scanline resampling without materializing a full-resolution raster
• GetLastJPEGReducedDecodeInfo reports source, scaled, output, scanline, and working-memory geometry for the latest compatible request
• AcquireLoadedImageStream and AcquireLoadedImageRegionAtResolution return immutable shared handles without warm-hit pixel copies, while the cached TBitmap methods preserve caller-owned compatibility
• GetImageDecodeConcurrencyInfo reports active flights, peak parallel owner decodes, duplicate waits, completions, and failures
• THotPDF.SearchLoadedNext is a stateless page-local search primitive that moves forward from a previous THPDFTextMatch
• THPDFLoadedDocumentFacade groups loaded forms, annotations, outlines, fonts, and images into typed collection objects
• THPDFLoadedFormInfo includes the current and default value, field type, description, read-only flag, and choice options

Outlines and resources

• THotPDF.GetLoadedOutlineEntries returns a complete preorder snapshot as THPDFLoadedOutlineEntryArray
• Each THPDFLoadedOutlineEntry has an index, parent index, level, title, and resolved destination page index
• THPDFLoadedOutlineInfo exposes the same hierarchy through THPDFLoadedOutlineCollection
• THPDFLoadedResourceCollection enumerates fonts and loaded images and can extract supported images as caller-owned bitmaps

System certificate signing

• THotPDF.SignPDFWithSystemCertificate signs file or stream input through a Windows certificate-store private key
• THotPDF.GetLoadedSignaturePermissions reads DocMDP, Catalog permission linkage, SigFieldLock, and FieldMDP structures into THPDFSignaturePermissionInfo without treating structural policy as cryptographic validity
• THPDFCertificateStoreSelector selects current-user or local-machine scope, store name, SHA-1 thumbprint, and whether key-provider UI is allowed
• THPDFCMSSignDigestCallback and HPDFCMSSignPDFStreamWithExternalSigner provide the underlying external-digest signer hook
• CNG keys use RSA PKCS#1 SHA-256 signing and legacy CSP keys use the equivalent CryptoAPI path

Printing

• THPDFPrintOptions supports collated copies, paper code, custom paper width and length, orientation, duplex, bin, and quality
• THPDFPrinterCapabilities reports parallel paper codes, names, and physical sizes alongside bins, resolutions, duplex, color, and maximum copies
• HPDFBuildPrintPageSequence exposes the deterministic physical page sequence used for collated and uncollated jobs

VCL viewer components

• THPDFViewerModel owns navigation, zoom state, text-search results, wrapping, and change notifications without creating a window
• THPDFViewer renders the selected page into a scrollable cached bitmap, supports actual size, fit page, fit width, custom zoom, and view rotation, and draws current search hits over the rendered page
• THPDFSearchPanel connects search text, case sensitivity, whole-word matching, and next or previous navigation to a viewer
• THPDFOutlineViewer displays the full recursive bookmark tree, follows document changes from the linked viewer, and navigates the linked viewer
• THPDFViewerAction descendants provide first, prior, next, last, zoom, fit, rotation, reset, find-next, and clear-search actions

Minimal viewer setup

PDF.LoadFromFile('input.pdf');
PDFViewer.PDFDocument := PDF;
OutlineViewer.Viewer := PDFViewer;
SearchPanel.Viewer := PDFViewer;
PDFViewer.ZoomMode := vzmFitWidth;

Ownership and indexing

• Loaded page indexes and destination page indexes are zero-based
• Page-range strings are one-based and use the same parser as extraction and printing
• Facade records are snapshots and collection mutation methods operate on the current loaded document
• Extracted bitmaps are owned by the caller

Related typed inspection and placement records

• THPDFPageElementKind, THPDFPageElement, and THPDFBaselineRole describe text, image, path, and baseline results returned by loaded page-element inspection
• THPDFWatermarkKind, THPDFWatermarkHorizPos, and THPDFWatermarkVertPos select the content and placement of loaded-document watermarks