19 #include "config_auto.h" 34 const char* extension)
35 : file_extension_(extension),
36 title_(
""), imagenum_(-1),
40 if (strcmp(outputbase,
"-") && strcmp(outputbase,
"stdout")) {
42 fout_ = fopen(outfile.
string(),
"wb");
50 if (fout_ !=
nullptr) {
60 if (next == NULL)
return;
65 while (next->next_ != NULL) {
68 next->next_ = remainder;
73 if (!happy_)
return false;
84 if (!happy_)
return false;
94 if (!happy_)
return false;
107 int n = fwrite(s, 1, len, fout_);
108 if (n != len) happy_ =
false;
128 const std::unique_ptr<const char[]> utf8(api->
GetUTF8Text());
135 bool pageBreak =
false;
155 font_info_ = font_info;
160 "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" 161 "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML 1.0 Transitional//EN\"\n" 162 " \"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd\">\n" 163 "<html xmlns=\"http://www.w3.org/1999/xhtml\" xml:lang=\"en\" " 164 "lang=\"en\">\n <head>\n <title>");
168 "<meta http-equiv=\"Content-Type\" content=\"text/html;" 169 "charset=utf-8\" />\n" 172 " <meta name='ocr-capabilities' content='ocr_page ocr_carea ocr_par" 173 " ocr_line ocrx_word");
176 " ocrp_lang ocrp_dir ocrp_font ocrp_fsize ocrp_wconf");
179 "</head>\n<body>\n");
192 if (hocr == NULL)
return false;
209 font_info_ = font_info;
215 "level\tpage_num\tblock_num\tpar_num\tline_num\tword_" 216 "num\tleft\ttop\twidth\theight\tconf\ttext\n");
224 if (tsv == NULL)
return false;
239 const std::unique_ptr<const char[]> unlv(api->
GetUNLVText());
240 if (unlv == NULL)
return false;
256 if (text == NULL)
return false;
271 if (osd == NULL)
return false;
virtual bool EndDocumentHandler()
virtual bool BeginDocumentHandler()
virtual bool AddImageHandler(TessBaseAPI *api)
TessBoxTextRenderer(const char *outputbase)
virtual bool BeginDocumentHandler()
virtual bool AddImageHandler(TessBaseAPI *api)
virtual bool AddImageHandler(TessBaseAPI *api)
char * GetOsdText(int page_number)
char * GetHOCRText(ETEXT_DESC *monitor, int page_number)
char * GetTSVText(int page_number)
virtual bool AddImageHandler(TessBaseAPI *api)
const char * string() const
virtual bool EndDocumentHandler()
TessTsvRenderer(const char *outputbase, bool font_info)
void insert(TessResultRenderer *next)
bool GetBoolVariable(const char *name, bool *value) const
const char * GetStringVariable(const char *name) const
TessTextRenderer(const char *outputbase)
TessResultRenderer(const char *outputbase, const char *extension)
virtual bool BeginDocumentHandler()
void AppendData(const char *s, int len)
virtual bool AddImageHandler(TessBaseAPI *api)
virtual bool AddImageHandler(TessBaseAPI *api)
char * GetBoxText(int page_number)
bool AddImage(TessBaseAPI *api)
TessUnlvRenderer(const char *outputbase)
TessResultRenderer * next()
virtual bool AddImageHandler(TessBaseAPI *api)=0
bool BeginDocument(const char *title)
#define TESSERACT_VERSION_STR
TessOsdRenderer(const char *outputbase)
virtual ~TessResultRenderer()
TessHOcrRenderer(const char *outputbase, bool font_info)
void AppendString(const char *s)
virtual bool EndDocumentHandler()
const char * title() const