xml_extractor.h

Go to the documentation of this file.

// Copyright (c) 2009-2025 Avanquest Software. All rights reserved.
 
#ifndef PDFSDK_XML_EXTRACTOR_H_INCLUDED_
#define PDFSDK_XML_EXTRACTOR_H_INCLUDED_

 
#include <pdfsdk/core.h>
#include <pdfsdk/text_edit.h>
 
#ifdef PDF_XMLEXTRACTOR_EXPORTS
#define PDF_XMLEXTRACTOR_API PDFSDK_EXPORT_
#else
#define PDF_XMLEXTRACTOR_API PDFSDK_IMPORT_
#endif
 
#ifdef __cplusplus
extern "C" {
#endif
 
/* Receives the serialized UTF-8 XML in a single call. `xml` points to `numBytes` bytes (not
   null-terminated); `clientData` is the opaque pointer passed to the serialize function. Returning a
   non-success code aborts serialization and is propagated to the caller. */
typedef PDErrCode(PDFSDK_CALLCONV* PDTXmlWriteProc)(const char* xml, size_t numBytes, void* clientData);
 
/* Serializes the text block content (paragraphs, spans, font and paragraph attributes) to UTF-8 XML
   and passes it to `proc` in a single call (no buffer pre-sizing, the XML is built once). */
PDF_XMLEXTRACTOR_API PDErrCode PDFSDK_CALLCONV PDTTextBlockSerializeToXML(PDTLayout layout,
                                                                         PDTTextBlock block,
                                                                         PDTXmlWriteProc proc,
                                                                         void* clientData);
 
/* Replaces the text block content from UTF-8 XML produced by PDTTextBlockSerializeToXML, applying the
   line-fit algorithm (shrinking font size and line spacing) so that the new text fits the original
   line count recorded per paragraph in the XML. */
PDF_XMLEXTRACTOR_API PDErrCode PDFSDK_CALLCONV PDTTextBlockDeserializeFromXML(PDDoc doc,
                                                                             PDTLayout layout,
                                                                             PDTTextBlock block,
                                                                             const char* pXml,
                                                                             size_t numBytes);
 
/* Serializes every text block on the layout (page) to a single UTF-8 XML document and passes it to
   `proc` in a single call: a <page> root carrying the XFA namespace/APIVersion once, with one <block>
   child per text block. The XML is built once (no buffer pre-sizing). */
PDF_XMLEXTRACTOR_API PDErrCode PDFSDK_CALLCONV PDTLayoutSerializeToXML(PDTLayout layout,
                                                                      PDTXmlWriteProc proc,
                                                                      void* clientData);
 
/* Applies a page UTF-8 XML produced by PDTLayoutSerializeToXML back to the layout's text blocks: the
   i-th <block> updates the i-th text block, processing min(blocks-in-xml, layout text block count). */
PDF_XMLEXTRACTOR_API PDErrCode PDFSDK_CALLCONV PDTLayoutDeserializeFromXML(PDDoc doc,
                                                                          PDTLayout layout,
                                                                          const char* pXml,
                                                                          size_t numBytes);
 
#ifdef __cplusplus
}
#endif
 
#endif  // PDFSDK_XML_EXTRACTOR_H_INCLUDED_