path: root/pango/pango-utils.c

/* Pango
 * pango-utils.c: Utilities for internal functions and modules
 *
 * Copyright (C) 2000 Red Hat Software
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Library General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.	 See the GNU
 * Library General Public License for more details.
 *
 * You should have received a copy of the GNU Library General Public
 * License along with this library; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 02111-1307, USA.
 */

#include <config.h>
#include <errno.h>
#include <string.h>
#include <stdlib.h>

#include "pango-font.h"
#include "pango-impl-utils.h"

#include <glib/gstdio.h>

#include "mini-fribidi/fribidi.h"

#ifndef HAVE_FLOCKFILE
#  define flockfile(f) (void)1
#  define funlockfile(f) (void)1
#  define getc_unlocked(f) getc(f)
#endif /* !HAVE_FLOCKFILE */

#ifdef G_OS_WIN32

#include <sys/types.h>

#define STRICT
#include <windows.h>

#endif

struct PangoAlias
{
  char *alias;
  int n_families;
  char **families;
  gboolean visible; /* Do we want/need this? */
};

static GHashTable *pango_aliases_ht = NULL;

PangoWarningHistory _pango_warning_history = { FALSE, FALSE, FALSE };

/**
 * pango_trim_string:
 * @str: a string
 * 
 * Trims leading and trailing whitespace from a string.
 * 
 * Return value: A newly-allocated string that must be freed with g_free()
 **/
char *
pango_trim_string (const char *str)
{
  int len;

  g_return_val_if_fail (str != NULL, NULL);
  
  while (*str && g_ascii_isspace (*str))
    str++;

  len = strlen (str);
  while (len > 0 && g_ascii_isspace (str[len-1]))
    len--;

  return g_strndup (str, len);
}

/**
 * pango_split_file_list:
 * @str: a G_SEARCHPATH_SEPARATOR separated list of filenames
 * 
 * Splits a G_SEARCHPATH_SEPARATOR-separated list of files, stripping
 * white space and subsituting ~/ with $HOME/.
 * 
 * Return value: a list of strings to be freed with g_strfreev()
 **/
char **
pango_split_file_list (const char *str)
{
  int i = 0;
  int j;
  char **files;

  files = g_strsplit (str, G_SEARCHPATH_SEPARATOR_S, -1);

  while (files[i])
    {
      char *file = pango_trim_string (files[i]);

      /* If the resulting file is empty, skip it */
      if (file[0] == '\0')
	{
	  g_free(file);
	  g_free (files[i]);
	  
	  for (j = i + 1; files[j]; j++)
	    files[j - 1] = files[j];
	  
	  files[j - 1] = NULL;

	  continue;
	}
#ifndef G_OS_WIN32
      /* '~' is a quite normal and common character in file names on
       * Windows, especially in the 8.3 versions of long file names, which
       * still occur now and then. Also, few Windows user are aware of the
       * Unix shell convention that '~' stands for the home directory,
       * even if they happen to have a home directory.
       */
      if (file[0] == '~' && file[1] == G_DIR_SEPARATOR)
	{
	  char *tmp = g_strconcat (g_get_home_dir(), file + 1, NULL);
	  g_free (file);
	  file = tmp;
	}
      else if (file[0] == '~' && file[1] == '\0')
	{
	  g_free (file);
	  file = g_strdup (g_get_home_dir());
	}
#endif
      g_free (files[i]);
      files[i] = file;
	
      i++;
    }

  return files;
}

/**
 * pango_read_line:
 * @stream: a stdio stream
 * @str: #GString buffer into which to write the result
 * 
 * Reads an entire line from a file into a buffer. Lines may
 * be delimited with '\n', '\r', '\n\r', or '\r\n'. The delimiter
 * is not written into the buffer. Text after a '#' character is treated as
 * a comment and skipped. '\' can be used to escape a # character.
 * '\' proceeding a line delimiter combines adjacent lines. A '\' proceeding
 * any other character is ignored and written into the output buffer
 * unmodified.
 * 
 * Return value: 0 if the stream was already at an %EOF character, otherwise
 *               the number of lines read (this is useful for maintaining
 *               a line number counter which doesn't combine lines with '\') 
 **/
gint
pango_read_line (FILE *stream, GString *str)
{
  gboolean quoted = FALSE;
  gboolean comment = FALSE;
  int n_read = 0;
  int lines = 1;
  
  flockfile (stream);

  g_string_truncate (str, 0);
  
  while (1)
    {
      int c;
      
      c = getc_unlocked (stream);

      if (c == EOF)
	{
	  if (quoted)
	    g_string_append_c (str, '\\');
	  
	  goto done;
	}
      else
	n_read++;

      if (quoted)
	{
	  quoted = FALSE;
	  
	  switch (c)
	    {
	    case '#':
	      g_string_append_c (str, '#');
	      break;
	    case '\r':
	    case '\n':
	      {
		int next_c = getc_unlocked (stream);

		if (!(next_c == EOF ||
		      (c == '\r' && next_c == '\n') ||
		      (c == '\n' && next_c == '\r')))
		  ungetc (next_c, stream);

		lines++;
		
		break;
	      }
	    default:
	      g_string_append_c (str, '\\');	      
	      g_string_append_c (str, c);
	    }
	}
      else
	{
	  switch (c)
	    {
	    case '#':
	      comment = TRUE;
	      break;
	    case '\\':
	      if (!comment)
		quoted = TRUE;
	      break;
	    case '\n':
	      {
		int next_c = getc_unlocked (stream);

		if (!(c == EOF ||
		      (c == '\r' && next_c == '\n') ||
		      (c == '\n' && next_c == '\r')))
		  ungetc (next_c, stream);

		goto done;
	      }
	    default:
	      if (!comment)
		g_string_append_c (str, c);
	    }
	}
    }

 done:

  funlockfile (stream);

  return (n_read > 0) ? lines : 0;
}

/**
 * pango_skip_space:
 * @pos: in/out string position
 * 
 * Skips 0 or more characters of white space.
 * 
 * Return value: %FALSE if skipping the white space leaves
 * the position at a '\0' character.
 **/
gboolean
pango_skip_space (const char **pos)
{
  const char *p = *pos;
  
  while (g_ascii_isspace (*p))
    p++;

  *pos = p;

  return !(*p == '\0');
}

/**
 * pango_scan_word:
 * @pos: in/out string position
 * @out: a #GString into which to write the result
 * 
 * Scans a word into a #GString buffer. A word consists
 * of [A-Za-z_] followed by zero or more [A-Za-z_0-9]
 * Leading white space is skipped.
 * 
 * Return value: %FALSE if a parse error occured. 
 **/
gboolean
pango_scan_word (const char **pos, GString *out)
{
  const char *p = *pos;

  while (g_ascii_isspace (*p))
    p++;
  
  if (!((*p >= 'A' && *p <= 'Z') ||
	(*p >= 'a' && *p <= 'z') ||
	*p == '_'))
    return FALSE;

  g_string_truncate (out, 0);
  g_string_append_c (out, *p);
  p++;

  while ((*p >= 'A' && *p <= 'Z') ||
	 (*p >= 'a' && *p <= 'z') ||
	 (*p >= '0' && *p <= '9') ||
	 *p == '_')
    {
      g_string_append_c (out, *p);
      p++;
    }

  *pos = p;

  return TRUE;
}

/**
 * pango_scan_string:
 * @pos: in/out string position
 * @out: a #GString into which to write the result
 * 
 * Scans a string into a #GString buffer. The string may either
 * be a sequence of non-white-space characters, or a quoted
 * string with '"'. Instead a quoted string, '\"' represents
 * a literal quote. Leading white space outside of quotes is skipped.
 * 
 * Return value: %FALSE if a parse error occured.
 **/
gboolean
pango_scan_string (const char **pos, GString *out)
{
  const char *p = *pos;
  
  while (g_ascii_isspace (*p))
    p++;

  if (!*p)
    return FALSE;
  else if (*p == '"')
    {
      gboolean quoted = FALSE;
      g_string_truncate (out, 0);

      p++;

      while (TRUE)
	{
	  if (quoted)
	    {
	      int c = *p;
	      
	      switch (c)
		{
		case '\0':
		  return FALSE;
		case 'n':
		  c = '\n';
		  break;
		case 't':
		  c = '\t';
		  break;
		}
	      
	      quoted = FALSE;
	      g_string_append_c (out, c);
	    }
	  else
	    {
	      switch (*p)
		{
		case '\0':
		  return FALSE;
		case '\\':
		  quoted = TRUE;
		  break;
		case '"':
		  p++;
		  goto done;
		default:
		  g_string_append_c (out, *p);
		  break;
		}
	    }
	  p++;
	}
    done:
      ;
    }
  else
    {
      g_string_truncate (out, 0);

      while (*p && !g_ascii_isspace (*p))
	{
	  g_string_append_c (out, *p);
	  p++;
	}
    }

  *pos = p;

  return TRUE;
}

/**
 * pango_scan_int:
 * @pos: in/out string position
 * @out: an int into which to write the result
 * 
 * Scans an integer. An integer consists
 * of up to 31 decimal digits. 
 * Leading white space is skipped.
 * 
 * Return value: %FALSE if a parse error occured.
 **/
gboolean
pango_scan_int (const char **pos, int *out)
{
  unsigned int i = 0;
  char buf[32];
  const char *p = *pos;

  while (g_ascii_isspace (*p))
    p++;
  
  if (*p < '0' || *p > '9')
    return FALSE;

  while ((*p >= '0') && (*p <= '9') && i < sizeof(buf))
    {
      buf[i] = *p;
      i++;
      p++;
    }

  if (i == sizeof(buf))
    return FALSE;
  else
    buf[i] = '\0';

  *out = atoi (buf);
  *pos = p;

  return TRUE;
}

static GHashTable *config_hash = NULL;

static void
read_config_file (const char *filename, gboolean enoent_error)
{
  GKeyFile *key_file = g_key_file_new();
  GError *key_file_error = NULL;
  gchar **groups;
  gsize groups_count = 0;
  guint group_index;
  
  if (!g_key_file_load_from_file(key_file,filename, 0, &key_file_error))
    {
      if (key_file_error)
	{
	  if (key_file_error->domain != G_FILE_ERROR || key_file_error->code != G_FILE_ERROR_NOENT || enoent_error)
	    {
	      g_warning ("error opening config file '%s': %s\n",
			  filename, key_file_error->message);
	    }
	  g_error_free(key_file_error);
	}
      g_key_file_free(key_file);
      return;
    }
  
  groups = g_key_file_get_groups (key_file, &groups_count);
  for (group_index = 0; group_index < groups_count; group_index++)
    {
      gsize keys_count = 0;
      const gchar *group = groups[group_index];
      GError *keys_error = NULL;
      gchar **keys;
      
      keys = g_key_file_get_keys(key_file, group, &keys_count, &keys_error);
      
      if (keys)
	{
	  guint key_index;

	  for (key_index = 0; key_index < keys_count; key_index++)
	    {
	      const gchar *key = keys[key_index];
	      GError *key_error = NULL;
	      gchar *value =  g_key_file_get_value(key_file, group, key, &key_error);
	      if (value != NULL)
		{
		  g_hash_table_insert (config_hash,
				       g_strdup_printf ("%s/%s", group, key),
				       value);
		}
	      if (key_error)
		{
		  g_warning ("error getting key '%s/%s' in config file '%s'\n",
			     group, key, filename);
		  g_error_free(key_error);
		}
	    }
	  g_strfreev(keys);
	}

      if (keys_error)
	{
	  g_warning ("error getting keys in group '%s' of config file '%s'\n",
		     filename, group);
	  g_error_free(keys_error);
	}
    }
  g_strfreev(groups);
  g_key_file_free(key_file);
}

static void
read_config (void)
{
  if (!config_hash)
    {
      char *filename;
      const char *home;
      const char *envvar;
      
      config_hash = g_hash_table_new_full (g_str_hash, g_str_equal,
					   (GDestroyNotify)g_free,
					   (GDestroyNotify)g_free);
      filename = g_build_filename (pango_get_sysconf_subdirectory (),
				   "pangorc",
				   NULL);
      read_config_file (filename, FALSE);
      g_free (filename);

      home = g_get_home_dir ();
      if (home && *home)
	{
	  filename = g_build_filename (home, ".pangorc", NULL);
	  read_config_file (filename, FALSE);
	  g_free (filename);
	}

      envvar = g_getenv ("PANGO_RC_FILE");
      if (envvar)
	read_config_file (envvar, TRUE);
    }
}

/**
 * pango_config_key_get:
 * @key: Key to look up, in the form "SECTION/KEY".
 * 
 * Looks up a key in the Pango config database
 * (pseudo-win.ini style, read from $sysconfdir/pango/pangorc,
 *  ~/.pangorc, and getenv (PANGO_RC_FILE).)
 * 
 * Return value: the value, if found, otherwise %NULL. The value is a
 * newly-allocated string and must be freed with g_free().
 **/
char *
pango_config_key_get (const char *key)
{
  g_return_val_if_fail (key != NULL, NULL);
  
  read_config ();

  return g_strdup (g_hash_table_lookup (config_hash, key));
}

#ifdef G_OS_WIN32

/* DllMain function needed to tuck away the DLL name */

G_WIN32_DLLMAIN_FOR_DLL_NAME(static, dll_name)
#endif

/**
 * pango_get_sysconf_subdirectory:
 *
 * On Unix, returns the name of the "pango" subdirectory of SYSCONFDIR
 * (which is set at compile time). On Win32, returns a subdirectory of
 * the Pango installation directory (which is deduced at run time from
 * the DLL's location, or stored in the Registry).
 *
 * Return value: the Pango sysconf directory. The returned string should
 * not be freed. 
 */
G_CONST_RETURN char *
pango_get_sysconf_subdirectory (void)
{
#ifdef G_OS_WIN32
  static gchar *result = NULL;

  if (result == NULL)
    result = g_win32_get_package_installation_subdirectory
      (PACKAGE " " VERSION, dll_name, "etc\\pango");

  return result;
#else
  return SYSCONFDIR "/pango";
#endif
}

/**
 * pango_get_lib_subdirectory:
 *
 * On Unix, returns the name of the "pango" subdirectory of LIBDIR
 * (which is set at compile time). On Win32, returns the Pango
 * installation directory (which is deduced at run time from the DLL's
 * location, or stored in the Registry). The returned string should
 * not be freed.
 *
 * Return value: the Pango lib directory. The returned string should
 * not be freed. 
 */
G_CONST_RETURN char *
pango_get_lib_subdirectory (void)
{
#ifdef G_OS_WIN32
  static gchar *result = NULL;

  if (result == NULL)
    result = g_win32_get_package_installation_subdirectory
      (PACKAGE " " VERSION, dll_name, "lib\\pango");

  return result;
#else
  return LIBDIR "/pango";
#endif
}

/**
 * pango_parse_style:
 * @str: a string to parse.
 * @style: a #PangoStyle to store the result in.
 * @warn: if %TRUE, issue a g_warning() on bad input.
 * 
 * Parses a font style. The allowed values are "normal",
 * "italic" and "oblique", case variations being
 * ignored.
 * 
 * Return value: %TRUE if @str was successfully parsed.
 **/ 
gboolean
pango_parse_style (const char *str,
		   PangoStyle *style,
		   gboolean    warn)
{
  if (*str == '\0')
    return FALSE;

  switch (str[0])
    {
    case 'n':
    case 'N':
      if (g_ascii_strcasecmp (str, "normal") == 0)
	{
	  *style = PANGO_STYLE_NORMAL;
	  return TRUE;
	}
      break;
    case 'i':
    case 'I':
      if (g_ascii_strcasecmp (str, "italic") == 0)
	{
	  *style = PANGO_STYLE_ITALIC;
	  return TRUE;
	}
      break;
    case 'o':
    case 'O':
      if (g_ascii_strcasecmp (str, "oblique") == 0)
	{
	  *style = PANGO_STYLE_OBLIQUE;
	  return TRUE;
	}
      break;
    }
  if (warn)
    g_warning ("style must be normal, italic, or oblique");
  
  return FALSE;
}

/**
 * pango_parse_variant:
 * @str: a string to parse.
 * @variant: a #PangoVariant to store the result in.
 * @warn: if %TRUE, issue a g_warning() on bad input.
 * 
 * Parses a font variant. The allowed values are "normal"
 * and "smallcaps" or "small_caps", case variations being
 * ignored.
 * 
 * Return value: %TRUE if @str was successfully parsed.
 **/ 
gboolean
pango_parse_variant (const char   *str,
		     PangoVariant *variant,
		     gboolean	   warn)
{
  if (*str == '\0')
    return FALSE;

  switch (str[0])
    {
    case 'n':
    case 'N':
      if (g_ascii_strcasecmp (str, "normal") == 0)
	{
	  *variant = PANGO_VARIANT_NORMAL;
	  return TRUE;
	}
      break;
    case 's':
    case 'S':
      if (g_ascii_strcasecmp (str, "small_caps") == 0 ||
	  g_ascii_strcasecmp (str, "smallcaps") == 0)
	{
	  *variant = PANGO_VARIANT_SMALL_CAPS;
	  return TRUE;
	}
      break;
    }
  
  if (warn)
    g_warning ("variant must be normal or small_caps");
  return FALSE;
}

/**
 * pango_parse_weight:
 * @str: a string to parse.
 * @weight: a #PangoWeight to store the result in.
 * @warn: if %TRUE, issue a g_warning() on bad input.
 * 
 * Parses a font weight. The allowed values are "heavy",
 * "ultrabold", "bold", "normal", "light", "ultraleight" 
 * and integers. Case variations are ignored.
 * 
 * Return value: %TRUE if @str was successfully parsed.
 **/ 
gboolean
pango_parse_weight (const char  *str,
		    PangoWeight *weight,
		    gboolean     warn)
{
  if (*str == '\0')
    return FALSE;

  switch (str[0])
    {
    case 'b':
    case 'B':
      if (g_ascii_strcasecmp (str, "bold") == 0)
	{
	  *weight = PANGO_WEIGHT_BOLD;
	  return TRUE;
	}
      break;
    case 'h':
    case 'H':
      if (g_ascii_strcasecmp (str, "heavy") == 0)
	{
	  *weight = PANGO_WEIGHT_HEAVY;
	  return TRUE;
	}
      break;
    case 'l':
    case 'L':
      if (g_ascii_strcasecmp (str, "light") == 0)
	{
	  *weight = PANGO_WEIGHT_LIGHT;
	  return TRUE;
	}
      break;
    case 'n':
    case 'N':
      if (g_ascii_strcasecmp (str, "normal") == 0)
	{
	  *weight = PANGO_WEIGHT_NORMAL;
	  return TRUE;
	}
      break;
    case 'u':
    case 'U':
      if (g_ascii_strcasecmp (str, "ultralight") == 0)
	{
	  *weight = PANGO_WEIGHT_ULTRALIGHT;
	  return TRUE;
	}
      else if (g_ascii_strcasecmp (str, "ultrabold") == 0)
	{
	  *weight = PANGO_WEIGHT_ULTRABOLD;
	  return TRUE;
	}
      break;
    case '0':
    case '1':
    case '2':
    case '3':
    case '4':
    case '5':
    case '6':
    case '7':
    case '8':
    case '9':
      {
	char *end;

	*weight = strtol (str, &end, 10);
	if (*end != '\0')
	  {
	    if (warn)
	      g_warning ("failed parsing numerical weight '%s'", str);
	    return FALSE;
	  }
	return TRUE;
      }
    }
  
  if (warn)
    g_warning ("weight must be ultralight, light, normal, bold, ultrabold, heavy, or an integer");
  return FALSE;
}

/**
 * pango_parse_stretch:
 * @str: a string to parse.
 * @stretch: a #PangoStretch to store the result in.
 * @warn: if %TRUE, issue a g_warning() on bad input.
 * 
 * Parses a font stretch. The allowed values are 
 * "ultra_condensed", "extra_condensed", "condensed", 
 * "semi_condensed", "normal", "semi_expanded", "expanded", 
 * "extra_expanded" and "ultra_expanded". Case variations are 
 * ignored and the '_' characters may be omitted.
 * 
 * Return value: %TRUE if @str was successfully parsed.
 **/ 
gboolean
pango_parse_stretch (const char   *str,
		     PangoStretch *stretch,
		     gboolean	   warn)
{
  if (*str == '\0')
    return FALSE;

  switch (str[0])
    { 
    case 'c':
    case 'C':
      if (g_ascii_strcasecmp (str, "condensed") == 0)
	{
	  *stretch = PANGO_STRETCH_CONDENSED;
	  return TRUE;
	}
      break;
    case 'e':
    case 'E':
      if (g_ascii_strcasecmp (str, "extra_condensed") == 0 ||
	  g_ascii_strcasecmp (str, "extracondensed") == 0)
	{
	  *stretch = PANGO_STRETCH_EXTRA_CONDENSED;
	  return TRUE;
	}
     if (g_ascii_strcasecmp (str, "extra_expanded") == 0 ||
	 g_ascii_strcasecmp (str, "extraexpanded") == 0)
	{
	  *stretch = PANGO_STRETCH_EXTRA_EXPANDED;
	  return TRUE;
	}
      if (g_ascii_strcasecmp (str, "expanded") == 0)
	{
	  *stretch = PANGO_STRETCH_EXPANDED;
	  return TRUE;
	}
      break;
    case 'n':
    case 'N':
      if (g_ascii_strcasecmp (str, "normal") == 0)
	{
	  *stretch = PANGO_STRETCH_NORMAL;
	  return TRUE;
	}
      break;
    case 's':
    case 'S':
      if (g_ascii_strcasecmp (str, "semi_condensed") == 0 ||
	  g_ascii_strcasecmp (str, "semicondensed") == 0)
	{
	  *stretch = PANGO_STRETCH_SEMI_CONDENSED;
	  return TRUE;
	}
      if (g_ascii_strcasecmp (str, "semi_expanded") == 0 ||
	  g_ascii_strcasecmp (str, "semiexpanded") == 0)
	{
	  *stretch = PANGO_STRETCH_SEMI_EXPANDED;
	  return TRUE;
	}
      break;
    case 'u':
    case 'U':
      if (g_ascii_strcasecmp (str, "ultra_condensed") == 0 ||
	  g_ascii_strcasecmp (str, "ultracondensed") == 0)
	{
	  *stretch = PANGO_STRETCH_ULTRA_CONDENSED;
	  return TRUE;
	}
      if (g_ascii_strcasecmp (str, "ultra_expanded") == 0 ||
	  g_ascii_strcasecmp (str, "ultraexpanded") == 0)
	{
	  *stretch = PANGO_STRETCH_ULTRA_EXPANDED;
	  return TRUE;
	}
      break;
    }

  if (warn)
    g_warning ("stretch must be ultra_condensed, extra_condensed, condensed, semi_condensed, normal, semi_expanded, expanded, extra_expanded, or ultra_expanded");
  return FALSE;
}

static const char canon_map[256] = {
   0,   0,   0,   0,   0,   0,   0,   0,    0,   0,   0,   0,   0,   0,   0,   0, 
   0,   0,   0,   0,   0,   0,   0,   0,    0,   0,   0,   0,   0,   0,   0,   0, 
   0,   0,   0,   0,   0,   0,   0,   0,    0,   0,   0,   0,   0,  '-',  0,   0, 
  '0', '1', '2', '3', '4', '5', '6', '7',  '8', '9',  0,   0,   0,   0,   0,   0, 
   0,  'a', 'b', 'c', 'd', 'e', 'f', 'g',  'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o',
  'p', 'q', 'r', 's', 't', 'u', 'v', 'w',  'x', 'y', 'z',  0,   0,   0,   0,  '-',
   0,  'a', 'b', 'c', 'd', 'e', 'f', 'g',  'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o',
  'p', 'q', 'r', 's', 't', 'u', 'v', 'w',  'x', 'y', 'z',  0,   0,   0,   0,   0
};
   
static gboolean
lang_equal (gconstpointer v1,
	    gconstpointer v2)
{
  const guchar *p1 = v1;
  const guchar *p2 = v2;

  while (canon_map[*p1] && canon_map[*p1] == canon_map[*p2])
    {
      p1++, p2++;
    }

  return (canon_map[*p1] == canon_map[*p2]);
}

static guint
lang_hash (gconstpointer key)
{
  const guchar *p = key;
  guint h = 0;
  while (canon_map[*p])
    {
      h = (h << 5) - h + canon_map[*p];
      p++;
    }

  return h;
}

static PangoLanguage *
pango_language_copy (PangoLanguage *language)
{
  return language; /* language tags are const */
}

static void
pango_language_free (PangoLanguage *language)
{
  return; /* nothing */
}

GType
pango_language_get_type (void)
{
  static GType our_type = 0;
  
  if (our_type == 0)
    our_type = g_boxed_type_register_static (I_("PangoLanguage"),
                                             (GBoxedCopyFunc)pango_language_copy,
                                             (GBoxedFreeFunc)pango_language_free);
  return our_type;
}

/**
 * pango_language_from_string:
 * @language: a string representing a language tag
 * 
 * Take a RFC-3066 format language tag as a string and convert it to a
 * #PangoLanguage pointer that can be efficiently copied (copy the
 * pointer) and compared with other language tags (compare the
 * pointer.)
 *
 * This function first canonicalizes the string by converting it to
 * lowercase, mapping '_' to '-', and stripping all characters other
 * than letters and '-'.
 * 
 * Return value: an opaque pointer to a #PangoLanguage structure.
 *               this will be valid forever after.
 **/
PangoLanguage *
pango_language_from_string (const char *language)
{
  static GHashTable *hash = NULL;
  char *result;
  int len;
  char *p;

  if (G_UNLIKELY (!hash))
    hash = g_hash_table_new (lang_hash, lang_equal);
  else
    {
      result = g_hash_table_lookup (hash, language);
      if (result)
	return (PangoLanguage *)result;
    }

  len = strlen (language);
  result = g_malloc (len + 1);

  p = result;
  while ((*(p++) = canon_map[*(guchar *)language++]))
    ;

  g_hash_table_insert (hash, result, result);

  return (PangoLanguage *)result;
}

/**
 * pango_language_matches:
 * @language: a language tag (see pango_language_from_string()),
 *            %NULL is allowed and matches nothing but '*'
 * @range_list: a list of language ranges, separated by ';', ':',
 *   ',', or space characters.
 *   Each element must either be '*', or a RFC 3066 language range
 *   canonicalized as by pango_language_from_string()
 * 
 * Checks if a language tag matches one of the elements in a list of
 * language ranges. A language tag is considered to match a range
 * in the list if the range is '*', the range is exactly the tag,
 * or the range is a prefix of the tag, and the character after it
 * in the tag is '-'.
 *
 * Return value: %TRUE if a match was found.
 **/
gboolean
pango_language_matches (PangoLanguage *language,
			const char    *range_list)
{
  const char *lang_str = pango_language_to_string (language);
  const char *p = range_list;
  gboolean done = FALSE;

  while (!done)
    {
      const char *end = strpbrk (p, ";:, \t");
      if (!end)
	{
	  end = p + strlen (p);
	  done = TRUE;
	}

      if (strncmp (p, "*", 1) == 0 ||
	  (lang_str && strncmp (lang_str, p, end - p) == 0 &&
	   (lang_str[end - p] == '\0' || lang_str[end - p] == '-')))
	return TRUE;

      if (!done)
	p = end + 1;
    }

  return FALSE;
}

typedef struct {
  const char lang[4];
  const char *str;
} LangInfo;

static int
lang_info_compare (const void *key, const void *val)
{
  const LangInfo *lang_info = val;

  return strncmp (key, lang_info->lang, 2);
}

/* The following array is supposed to contain enough text to tickle all necessary fonts for each
 * of the languages in the following. Yes, it's pretty lame. Not all of the languages
 * in the following have sufficient text to excercise all the accents for the language, and
 * there are obviously many more languages to include as well.
 */
static const LangInfo lang_texts[] = {
  { "ar", "Arabic  \330\247\331\204\330\263\331\204\330\247\331\205 \330\271\331\204\331\212\331\203\331\205" },
  { "cs", "Czech (\304\215esky)  Dobr\303\275 den" },
  { "da", "Danish (Dansk)  Hej, Goddag" },
  { "el", "Greek (\316\225\316\273\316\273\316\267\316\275\316\271\316\272\316\254) \316\223\316\265\316\271\316\254 \317\203\316\261\317\202" },
  { "en", "English Hello" },
  { "eo", "Esperanto Saluton" },
  { "es", "Spanish (Espa\303\261ol) \302\241Hola!" },
  { "et", "Estonian  Tere, Tervist" },
  { "fi", "Finnish (Suomi)  Hei, Hyv\303\244\303\244 p\303\244iv\303\244\303\244" },
  { "fr", "French (Fran\303\247ais)" },
  { "de", "German Gr\303\274\303\237 Gott" },
  { "he", "Hebrew   \327\251\327\234\327\225\327\235" },
  { "it", "Italiano  Ciao, Buon giorno" },
  { "ja", "Japanese (\346\227\245\346\234\254\350\252\236) \343\201\223\343\202\223\343\201\253\343\201\241\343\201\257, \357\275\272\357\276\235\357\276\206\357\276\201\357\276\212" },
  { "ko", "Korean (\355\225\234\352\270\200)   \354\225\210\353\205\225\355\225\230\354\204\270\354\232\224, \354\225\210\353\205\225\355\225\230\354\213\255\353\213\210\352\271\214" },
  { "mt", "Maltese   \304\212aw, Sa\304\247\304\247a" },
  { "nl", "Nederlands, Vlaams Hallo, Dag" },
  { "no", "Norwegian (Norsk) Hei, God dag" },
  { "pl", "Polish   Dzie\305\204 dobry, Hej" },
  { "ru", "Russian (\320\240\321\203\321\201\321\201\320\272\320\270\320\271)" },
  { "sk", "Slovak   Dobr\303\275 de\305\210" },
  { "sv", "Swedish (Svenska) Hej p\303\245 dej, Goddag" },
  { "tr", "Turkish (T\303\274rk\303\247e) Merhaba" },
  { "zh", "Chinese (\344\270\255\346\226\207,\346\231\256\351\200\232\350\257\235,\346\261\211\350\257\255)" }
};

/**
 * pango_language_get_sample_string:
 * @language: a #PangoLanguage
 * 
 * Get a string that is representative of the characters needed to
 * render a particular language. This function is a bad hack for
 * internal use by renderers and Pango.
 * 
 * Return value: the sample string. This value is owned by Pango
 *   and must not be freed.
 **/
G_CONST_RETURN char *
pango_language_get_sample_string (PangoLanguage *language)
{
  const char *result;
  
  if (language)
    {
      const char *lang_str = pango_language_to_string (language);
      
      LangInfo *lang_info = bsearch (lang_str, lang_texts,
				     G_N_ELEMENTS (lang_texts), sizeof (LangInfo),
				     lang_info_compare);

      if (lang_info)
	result = lang_info->str;
      else
	result = "French (Fran\303\247ais)";     /* Assume iso-8859-1 */
    }
  else
    {
      /* Complete junk
       */

      result = "\330\247\331\204\330\263\331\204\330\247\331\205 \330\271\331\204\331\212\331\203\331\205 \304\215esky \316\225\316\273\316\273\316\267\316\275\316\271\316\272\316\254 Fran\303\247ais \346\227\245\346\234\254\350\252\236 \355\225\234\352\270\200 \320\240\321\203\321\201\321\201\320\272\320\270\320\271 \344\270\255\346\226\207,\346\231\256\351\200\232\350\257\235,\346\261\211\350\257\255 T\303\274rk\303\247e";
    }

  return result;
}

/**
 * pango_log2vis_get_embedding_levels:
 * @text:      the text to itemize.
 * @length:    the number of bytes (not characters) to process, or -1
 *             if @text is nul-terminated and the legnth should be calculated.
 * @pbase_dir: input base direction, and output resolved direction.
 *
 * This will return the bidirectional embedding levels of the input paragraph
 * as defined by the Unicode Bidirectional Algorithm available at:
 *
 *   http://www.unicode.org/reports/tr9/
 *
 * If the input base direction is a weak direction, the direction of the
 * characters in the text will determine the final resolved direction.
 *
 * Return value: a newly allocated array of embedding levels, one item per
 *               character (not byte), that should be freed using g_free.
 *
 * Since: 1.4
 */
guint8 *
pango_log2vis_get_embedding_levels (const gchar    *text,
				    int             length,
				    PangoDirection *pbase_dir)
{
  FriBidiCharType fribidi_base_dir;
  guint8 *embedding_levels_list;

  switch (*pbase_dir)
    {
    case PANGO_DIRECTION_LTR:
    case PANGO_DIRECTION_TTB_RTL:
      fribidi_base_dir = FRIBIDI_TYPE_L;
      break;
    case PANGO_DIRECTION_RTL:
    case PANGO_DIRECTION_TTB_LTR:
      fribidi_base_dir = FRIBIDI_TYPE_R;
      break;
    case PANGO_DIRECTION_WEAK_RTL:
      fribidi_base_dir = FRIBIDI_TYPE_WR;
      break;
    /*
    case PANGO_DIRECTION_WEAK_LTR:
    case PANGO_DIRECTION_NEUTRAL:
    */
    default:
      fribidi_base_dir = FRIBIDI_TYPE_WL;
      break;
    }

#ifdef FRIBIDI_HAVE_UTF8
  {
    if (length < 0)
      length = strlen (text);
    embedding_levels_list = fribidi_log2vis_get_embedding_levels_new_utf8 (text, length, &fribidi_base_dir);
  }
#else
  {
    gunichar *text_ucs4;
    int n_chars;
    text_ucs4 = g_utf8_to_ucs4_fast (text, length, &n_chars);
    embedding_levels_list = g_new (guint8, n_chars);
    fribidi_log2vis_get_embedding_levels ((FriBidiChar*)text_ucs4, n_chars,
					  &fribidi_base_dir,
					  (FriBidiLevel*)embedding_levels_list);
    g_free (text_ucs4);
  }
#endif

  *pbase_dir = (fribidi_base_dir == FRIBIDI_TYPE_L) ?  PANGO_DIRECTION_LTR : PANGO_DIRECTION_RTL;

  return embedding_levels_list;
}

/**
 * pango_unichar_direction:
 * @ch: a Unicode character
 *
 * Determines the direction of a character; either
 * %PANGO_DIRECTION_LTR, %PANGO_DIRECTION_RTL, or
 * %PANGO_DIRECTION_NEUTRAL.
 *
 * Return value: the direction of the character, as used in the
 * Unicode bidirectional algorithm.
 */
PangoDirection
pango_unichar_direction (gunichar ch)
{
  FriBidiCharType fribidi_ch_type = fribidi_get_type (ch);

  if (!FRIBIDI_IS_LETTER (fribidi_ch_type))
    return PANGO_DIRECTION_NEUTRAL;
  else if (FRIBIDI_IS_RTL (fribidi_ch_type))
    return PANGO_DIRECTION_RTL;
  else
    return PANGO_DIRECTION_LTR;
}

/**
 * pango_get_mirror_char:
 * @ch: a Unicode character
 * @mirrored_ch: location to store the mirrored character
 *
 * If @ch has the Unicode mirrored property and there is another Unicode
 * character that typically has a glyph that is the mirror image of @ch's
 * glyph, puts that character in the address pointed to by @mirrored_ch.
 *
 * Use g_unichar_get_mirror_char() instead; the docs for that function
 * provide full details.
 * 
 * Return value: %TRUE if @ch has a mirrored character and @mirrored_ch is
 * filled in, %FALSE otherwise
 **/
gboolean 
pango_get_mirror_char (gunichar        ch,
		       gunichar       *mirrored_ch)
{
  return g_unichar_get_mirror_char (ch, mirrored_ch); 
}


static guint
alias_hash (struct PangoAlias *alias)
{
  return g_str_hash (alias->alias);
}

static gboolean
alias_equal (struct PangoAlias *alias1,
	     struct PangoAlias *alias2)
{
  return g_str_equal (alias1->alias,
		      alias2->alias);
}


static void
alias_free (struct PangoAlias *alias)
{
  int i;
  g_free (alias->alias);

  for (i = 0; i < alias->n_families; i++)
    g_free (alias->families[i]);

  g_free (alias->families);
  
  g_slice_free (struct PangoAlias, alias);
}

static void
read_alias_file (const char *filename)
{
  FILE *file;
    
  GString *line_buffer;
  GString *tmp_buffer1;
  GString *tmp_buffer2;
  char *errstring = NULL;
  const char *pos;
  int line = 0;
  struct PangoAlias alias_key;
  struct PangoAlias *alias;
  char **new_families;
  int n_new;
  int i;

  file = g_fopen (filename, "r");
  if (!file)
    return;

  line_buffer = g_string_new (NULL);
  tmp_buffer1 = g_string_new (NULL);
  tmp_buffer2 = g_string_new (NULL);

  while (pango_read_line (file, line_buffer))
    {
      gboolean append = FALSE;
      line++;

      pos = line_buffer->str;
      if (!pango_skip_space (&pos))
	continue;
      
      if (!pango_scan_word (&pos, tmp_buffer1) ||
	  !pango_skip_space (&pos))
	{
	  errstring = g_strdup ("Line is not of the form KEY=VALUE or KEY+=VALUE");
	  goto error;
	}
      
      if (*pos == '+')
	{
	  append = TRUE;
	  pos++;
	}

      if (*(pos++) != '=')
	{
	  errstring = g_strdup ("Line is not of the form KEY=VALUE or KEY+=VALUE");
	  goto error;
	}
      
      if (!pango_scan_string (&pos, tmp_buffer2))
	{
	  errstring = g_strdup ("Error parsing value string");
	  goto error;
	}
      if (pango_skip_space (&pos))
	{
	  errstring = g_strdup ("Junk after value string");
	  goto error;
	}

      alias_key.alias = g_ascii_strdown (tmp_buffer1->str, -1);
      
      /* Remove any existing values */
      alias = g_hash_table_lookup (pango_aliases_ht, &alias_key);
      
      if (!alias)
	{
	  alias = g_slice_new0 (struct PangoAlias);
	  alias->alias = alias_key.alias;
	  
	  g_hash_table_insert (pango_aliases_ht,
			       alias, alias);
	}
      else
	g_free (alias_key.alias);
	  
      new_families = g_strsplit (tmp_buffer2->str, ",", -1);
      
      n_new = 0;
      while (new_families[n_new])
	n_new++;
      
      if (alias->families && append)
	{
	  alias->families = g_realloc (alias->families,
				       sizeof (char *) *(n_new + alias->n_families));
	  for (i = 0; i < n_new; i++)
	    alias->families[alias->n_families + i] = new_families[i];
	  g_free (new_families);
	  alias->n_families += n_new;
	}
      else
	{
	  for (i = 0; i < alias->n_families; i++)
	    g_free (alias->families[i]);
	  g_free (alias->families);
	  
	  alias->families = new_families;
	  alias->n_families = n_new;
	}
    }

  if (ferror (file))
    errstring = g_strdup (g_strerror(errno));
  
 error:

  if (errstring)
    {
      g_warning ("error reading alias file: %s:%d: %s\n", filename, line, errstring);
      g_free (errstring);
    }
      
  g_string_free (line_buffer, TRUE);
  g_string_free (tmp_buffer1, TRUE);
  g_string_free (tmp_buffer2, TRUE);

  fclose (file);
}

static void
pango_load_aliases (void)
{
  char *filename;
  const char *home;

  pango_aliases_ht = g_hash_table_new_full ((GHashFunc)alias_hash,
					    (GEqualFunc)alias_equal,
					    (GDestroyNotify)alias_free,
					    NULL);


  filename = g_strconcat (pango_get_sysconf_subdirectory (),
			  G_DIR_SEPARATOR_S "pango.aliases",
			  NULL);
  read_alias_file (filename);
  g_free (filename);
  
  home = g_get_home_dir ();
  if (home && *home)
    {
      filename = g_strconcat (home,
			      G_DIR_SEPARATOR_S ".pango.aliases",
			      NULL);
      read_alias_file (filename);
      g_free (filename);
    }
}


/**
 * pango_lookup_aliases:
 * @fontname: an ascii string
 * @families: will be set to an array of font family names.
 *    this array is owned by pango and should not be freed.
 * @n_families: will be set to the length of the @families array.
 *
 * Look up all user defined aliases for the alias @fontname.
 * The resulting font family names will be stored in @families,
 * and the number of families in @n_families.
 **/
void
pango_lookup_aliases (const char   *fontname,
		      char       ***families,
		      int          *n_families)
{
  struct PangoAlias alias_key;
  struct PangoAlias *alias;
  
  if (pango_aliases_ht == NULL)
    pango_load_aliases ();

  alias_key.alias = g_ascii_strdown (fontname, -1);
  alias = g_hash_table_lookup (pango_aliases_ht, &alias_key);
  g_free (alias_key.alias);

  if (alias)
    {
      *families = alias->families;
      *n_families = alias->n_families;
    }
  else
    {
      *families = NULL;
      *n_families = 0;
    }
}

/**
 * pango_find_base_dir:
 * @text:   the text to process
 * @length: length of @text in bytes (may be -1 if @text is nul-terminated)
 *
 * Searches a string the first character that has a strong 
 * direction, according to the Unicode bidirectional algorithm.
 *
 * Return value: The direction corresponding to the first strong character.
 * If no such character is found, then %PANGO_DIRECTION_NEUTRAL is returned.
 *
 * Since: 1.4
 */
PangoDirection
pango_find_base_dir (const gchar *text,
		     gint         length)
{
  PangoDirection dir = PANGO_DIRECTION_NEUTRAL;
  const gchar *p;

  g_return_val_if_fail (text != NULL || length == 0, PANGO_DIRECTION_NEUTRAL);

  p = text;
  while ((length < 0 || p < text + length) && *p)
    {
      gunichar wc = g_utf8_get_char (p);

      dir = pango_unichar_direction (wc);

      if (dir != PANGO_DIRECTION_NEUTRAL)
	break;

      p = g_utf8_next_char (p);
    }

  return dir;
}

/**
 * pango_is_zero_width:
 * @ch: a Unicode character
 *
 * Checks @ch to see if it is a zero-width character that should not be
 * normally rendered on the screen.
 *
 * Return value: %TRUE if @ch is a zero-width character, %FALSE otherwise
 *
 * Since: 1.10
 */
gboolean
pango_is_zero_width (gunichar ch)
{
/* Zero Width characters:
 *
 *  00AD  SOFT HYPHEN
 *  034F  COMBINING GRAPHEME JOINER
 *
 *  200B  ZERO WIDTH SPACE
 *  200C  ZERO WIDTH NON-JOINER
 *  200D  ZERO WIDTH JOINER
 *  200E  LEFT-TO-RIGHT MARK
 *  200F  RIGHT-TO-LEFT MARK
 *
 *  2028  LINE SEPARATOR
 *
 *  202A  LEFT-TO-RIGHT EMBEDDING
 *  202B  RIGHT-TO-LEFT EMBEDDING
 *  202C  POP DIRECTIONAL FORMATTING
 *  202D  LEFT-TO-RIGHT OVERRIDE
 *  202E  RIGHT-TO-LEFT OVERRIDE
 *
 *  2060  WORD JOINER
 *  2061  FUNCTION APPLICATION
 *  2062  INVISIBLE TIMES
 *  2063  INVISIBLE SEPARATOR
 *
 *  FEFF  ZERO WIDTH NO-BREAK SPACE
 */
  return ((ch & ~(gunichar)0x007F) == 0x2000 && (
		(ch >= 0x200B && ch <= 0x200F) ||
		(ch >= 0x202A && ch <= 0x202E) ||
		(ch >= 0x2060 && ch <= 0x2063) ||
		(ch == 0x2028)
	 )) || G_UNLIKELY (ch == 0x00AD
			|| ch == 0x034F
			|| ch == 0xFEFF);
}

/**
 * pango_quantize_line_geometry:
 * @thickness: pointer to the thickness of a line, in Pango scaled units
 * @position: corresponding position
 *
 * Quantizes the thickness and position of a line, typically an
 * underline or strikethrough, to whole device pixels, that is integer
 * multiples of %PANGO_SCALE. The purpose of this function is to avoid
 * such lines looking blurry.
 *
 * Since: 1.12
 */
void
pango_quantize_line_geometry (int *thickness,
			      int *position)
{
  int thickness_pixels = (*thickness + PANGO_SCALE / 2) / PANGO_SCALE;
  if (thickness_pixels == 0)
    thickness_pixels = 1;
  
  if (thickness_pixels & 1)
    {
      int new_center = ((*position - *thickness / 2) & ~(PANGO_SCALE - 1)) + PANGO_SCALE / 2;
      *position = new_center + (PANGO_SCALE * thickness_pixels) / 2;
    }
  else
    {
      int new_center = ((*position - *thickness / 2 + PANGO_SCALE / 2) & ~(PANGO_SCALE - 1));
      *position = new_center + (PANGO_SCALE * thickness_pixels) / 2;
    }

  *thickness = thickness_pixels * PANGO_SCALE;
}