public static void Main(String[] args)
{
/* Global option list. */
string globaloptlist = "searchpath={{../data} {../../data}}";
/* Document specific option list. */
string docoptlist = "";
/* Page-specific option list. */
string pageoptlist = "granularity=word";
FileStream outfile;
StreamWriter outfp;
if (args.Length != 2)
{
Console.WriteLine("usage: glyphinfo <infilename> <outfilename>");
return;
}
outfile = File.Create(args.GetValue(1).ToString());
outfp = new StreamWriter(outfile, System.Text.Encoding.UTF8);
TET tet = null;
try
{
tet = new TET();
tet.set_option(globaloptlist);
int doc = tet.open_document(args[0], docoptlist);
if (doc == -1)
{
Console.WriteLine("Error " + tet.get_errnum() + " in "
+ tet.get_apiname() + "(): " + tet.get_errmsg());
return;
}
/* get number of pages in the document */
int n_pages = (int)tet.pcos_get_number(doc, "length:pages");
/* Loop over pages in the document */
for (int pageno = 1; pageno <= n_pages; ++pageno)
{
string text;
int page;
int previouscolor = -1;
page = tet.open_page(doc, pageno, pageoptlist);
if (page == -1)
{
Console.WriteLine("Error " + tet.get_errnum() + " in "
+ tet.get_apiname() + "() on page "
+ pageno + ": " + tet.get_errmsg());
continue; /* try next page */
}
/* Administrative information */
outfp.WriteLine("[ Document: '" +
tet.pcos_get_string(doc, "filename") + "' ]");
outfp.WriteLine("[ Document options: '" + docoptlist + "' ]");
outfp.WriteLine("[ Page options: '" + pageoptlist + "' ]");
outfp.WriteLine("[ ----- Page " + pageno + " ----- ]");
/* Retrieve all text fragments */
while ((text = tet.get_text(page)) != null)
{
/* print the retrieved text */
outfp.WriteLine("[" + text + "]");
/* Loop over all glyphs and print their details */
while (tet.get_char_info(page) != -1)
{
string str;
string fontname;
/* Fetch the font name with pCOS (based on its ID) */
fontname = tet.pcos_get_string(doc,
"fonts[" + tet.fontid + "]/name");
/* Print the character */
str = String.Format("U+{0}", tet.uv.ToString("X4"));
/* ...and its UTF8 representation */
str = str + String.Format(" '" + (char)(tet.uv) + "'");
/* Print font name, size, and position */
str = str + String.Format(" {0} size={1} x={2} y={3}",
fontname, tet.fontsize.ToString("f2"),
tet.x.ToString("f2"), tet.y.ToString("f2"));
/* Print the color id */
str = str + String.Format(" colorid={0}", tet.colorid);
/* check wheather the text color changes */
if (tet.colorid != previouscolor)
{
str = print_color_value(str, tet, doc, tet.colorid);
previouscolor = tet.colorid;
}
/* Examine the "type" member */
if (tet.type == 1)
{
str = str + " ligature_start";
}
else if (tet.type == 10)
{
str = str + " ligature_cont";
}
/* Separators are only inserted for granularity > word*/
else if (tet.type == 12)
{
str = str + " inserted";
}
/* Examine the bit flags in the "attributes" member */
const int ATTR_NONE = 0;
const int ATTR_SUB = 1;
const int ATTR_SUP = 2;
const int ATTR_DROPCAP = 4;
const int ATTR_SHADOW = 8;
const int ATTR_DH_PRE = 16;
const int ATTR_DH_ARTF = 32;
const int ATTR_DH_POST = 64;
if (tet.attributes != ATTR_NONE)
{
if ((tet.attributes & ATTR_SUB) == ATTR_SUB)
{
str = str + "/sub";
}
if ((tet.attributes & ATTR_SUP) == ATTR_SUP)
{
str = str + "/sup";
}
if ((tet.attributes & ATTR_DROPCAP) == ATTR_DROPCAP)
{
str = str + "/dropcap";
}
if ((tet.attributes & ATTR_SHADOW) == ATTR_SHADOW)
{
str = str + "/shadow";
}
if ((tet.attributes & ATTR_DH_PRE) == ATTR_DH_PRE)
{
str = str + "/dehyphenation_pre";
}
if ((tet.attributes & ATTR_DH_ARTF) == ATTR_DH_ARTF)
{
str = str + "/dehyphenation_artifact";
}
if ((tet.attributes & ATTR_DH_POST) == ATTR_DH_POST)
{
str = str + "/dehyphenation_post";
}
}
outfp.WriteLine(str);
}
outfp.WriteLine("");
}
if (tet.get_errnum() != 0)
{
Console.WriteLine("Error " + tet.get_errnum() + " in "
+ tet.get_apiname() + "() on page "
+ pageno + ": " + tet.get_errmsg());
}
tet.close_page(page);
}
tet.close_document(doc);
}
catch (TETException e)
{
Console.WriteLine("Error " + e.get_errnum() + " in "
+ e.get_apiname() + "(): " + e.get_errmsg());
}
catch (Exception e)
{
Console.WriteLine("General Exception: " + e.ToString());
}
finally
{
if (tet != null)
{
tet.Dispose();
}
}
}
public static void Main(String[] args)
{
/* Global option list. */
string globaloptlist = "searchpath={{../data} {../../data}}";
/* Document specific option list. */
string docoptlist = "";
/* Page-specific option list. */
string pageoptlist = "granularity=line";
/* Search text with at least this size (use 0 to catch all sizes). */
double fontsizetrigger = 10;
/* Catch text where the font name contains this string (use empty string
* to catch all font names).
*/
String fontnametrigger = "Bold";
TET tet = null;
int pageno = 0;
if (args.Length != 1)
{
Console.WriteLine("usage: fontfilter <infilename>");
return;
}
try
{
tet = new TET();
tet.set_option(globaloptlist);
int doc = tet.open_document(args[0], docoptlist);
if (doc == -1)
{
Console.WriteLine("Error " + tet.get_errnum() + " in "
+ tet.get_apiname() + "(): " + tet.get_errmsg());
return;
}
/* Loop over pages in the document */
int n_pages = (int)tet.pcos_get_number(doc, "length:pages");
for (pageno = 1; pageno <= n_pages; ++pageno)
{
int page = tet.open_page(doc, pageno, pageoptlist);
if (page == -1)
{
Console.WriteLine("Error " + tet.get_errnum() + " in "
+ tet.get_apiname() + "(): " + tet.get_errmsg());
return; /* try next page */
}
/* Retrieve all text fragments for the page */
String text;
while ((text = tet.get_text(page)) != null)
{
/* Loop over all characters */
int ci;
while ((ci = tet.get_char_info(page)) != -1)
{
/* We need only the font name and size; the text
* position could be fetched from tet.x and tet.y.
*/
String fontname = tet.pcos_get_string(doc,
"fonts[" + tet.fontid + "]/name");
/* Check whether we found a match */
if (tet.fontsize >= fontsizetrigger &&
fontname.IndexOf(fontnametrigger) != -1)
{
/* print the retrieved font name, size, and text */
Console.WriteLine("[{0} {1:0.00}] {2}", fontname,
tet.fontsize, text);
}
/* In this sample we check only the first character of
* each fragment.
*/
break;
}
}
if (tet.get_errnum() != 0)
{
Console.WriteLine("Error " + tet.get_errnum() + " in "
+ tet.get_apiname() + "(): " + tet.get_errmsg());
}
tet.close_page(page);
}
tet.close_document(doc);
}
catch (TETException e)
{
if (pageno == 0)
{
Console.WriteLine("Error " + e.get_errnum() + " in "
+ e.get_apiname() + "(): " + e.get_errmsg() + "\n");
}
else
{
Console.WriteLine("Error " + e.get_errnum() + " in "
+ e.get_apiname() + "() on page " + pageno + ": "
+ e.get_errmsg() + "\n");
}
}
catch (Exception e)
{
Console.WriteLine("General Exception: " + e.ToString());
}
finally
{
tet.Dispose();
}
}