utility.c

Go to the documentation of this file.

/*
    This file is part of darktable,
    Copyright (C) 2010-2017 Tobias Ellinghaus.
    Copyright (C) 2011 Henrik Andersson.
    Copyright (C) 2011 johannes hanika.
    Copyright (C) 2011 Kanstantsin Shautsou.
    Copyright (C) 2011 Moritz Lipp.
    Copyright (C) 2012 Jérémy Rosen.
    Copyright (C) 2012 Richard Wonka.
    Copyright (C) 2013 Antony Dovgal.
    Copyright (C) 2013 Jean-Sébastien Pédron.
    Copyright (C) 2013, 2020-2021 Pascal Obry.
    Copyright (C) 2013 Stuart Henderson.
    Copyright (C) 2014-2016 Roman Lebedev.
    Copyright (C) 2015 Steven Fosdick.
    Copyright (C) 2016-2018 Peter Budai.
    Copyright (C) 2017, 2022 luzpaz.
    Copyright (C) 2019, 2021 Philippe Weyland.
    Copyright (C) 2020 Alexis Mousset.
    Copyright (C) 2020 hatsunearu.
    Copyright (C) 2020 Heiko Bauke.
    Copyright (C) 2020 Hubert Kowalski.
    Copyright (C) 2020-2021 Ralf Brown.
    Copyright (C) 2021 Benjamin Grimm-Lebsanft.
    Copyright (C) 2021 Christian Birzer.
    Copyright (C) 2021 Hanno Schwalm.
    Copyright (C) 2021 Marco Carrarini.
    Copyright (C) 2021 parafin.
    Copyright (C) 2021 RSL.
    Copyright (C) 2021 wpferguson.
    Copyright (C) 2022 Martin Bařinka.
    Copyright (C) 2022 Nicolas Auffray.
    Copyright (C) 2023 Luca Zulberti.
    Copyright (C) 2024 Aurélien PIERRE.
    Copyright (C) 2024-2025 Guillaume Stutin.
    
    darktable is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation, either version 3 of the License, or
    (at your option) any later version.
    
    darktable 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 General Public License for more details.
    
    You should have received a copy of the GNU General Public License
    along with darktable.  If not, see <http://www.gnu.org/licenses/>.
*/
 
#include <locale.h>
#include "common/paths.h"   // DT_PATH_MAX
 
#include <glib/gstdio.h>
#include "system/macros.h"
#include "gui/screen_metrics.h"
#include "system/mem_alloc.h"
#include "common/file_location.h"
#include "common/grealpath.h"
#include "common/utility.h"
 
/* getpwnam_r availability check */
#if defined __APPLE__ || defined _POSIX_C_SOURCE >= 1 || defined _XOPEN_SOURCE || defined _BSD_SOURCE        \
    || defined _SVID_SOURCE || defined _POSIX_SOURCE || defined __DragonFly__ || defined __FreeBSD__         \
    || defined __NetBSD__ || defined __OpenBSD__
  #include <unistd.h>
#endif
 
#ifdef _WIN32
  /* lowercase: MinGW ships windows.h/winbase.h/fileapi.h, and a Linux filesystem is
   * case-sensitive. The capitalised spellings only ever worked because Windows itself
   * is not, so they silently blocked cross-compiling from Linux. */
  #include <windows.h>
  #include <winbase.h>
  #include <fileapi.h>
#endif
 
#include <math.h>
#include <glib/gi18n.h>
 
#include <sys/stat.h>
#include <ctype.h>
 
#ifdef HAVE_CONFIG_H
  #include <config.h>
#endif
 
#include <librsvg/rsvg.h>
// ugh, ugly hack. why do people break stuff all the time?
#ifndef RSVG_CAIRO_H
#include <librsvg/rsvg-cairo.h>
#endif
 
size_t safe_strlen(const char *str)
{
  return str ? strlen(str) : 0;
}
 
gchar *dt_util_dstrcat(gchar *str, const gchar *format, ...)
{
  va_list args;
  gchar *ns;
  va_start(args, format);
  const size_t clen = str ? strlen(str) : 0;
  const int alen = g_vsnprintf(NULL, 0, format, args);
  const int nsize = alen + clen + 1;
 
  /* realloc for new string */
  ns = g_realloc(str, nsize);
  if(IS_NULL_PTR(str)) ns[0] = '\0';
  va_end(args);
 
  /* append string */
  va_start(args, format);
  g_vsnprintf(ns + clen, alen + 1, format, args);
  va_end(args);
 
  ns[nsize - 1] = '\0';
 
  return ns;
}
 
guint dt_util_str_occurence(const gchar *haystack, const gchar *needle)
{
  guint o = 0;
  if(haystack && needle)
  {
    const gchar *p = haystack;
    if((p = g_strstr_len(p, strlen(p), needle)) != NULL)
    {
      do
      {
        o++;
      } while((p = g_strstr_len((p + 1), strlen(p + 1), needle)) != NULL);
    }
  }
  return o;
}
 
gchar *dt_util_str_replace(const gchar *string, const gchar *pattern, const gchar *substitute)
{
  const gint occurrences = dt_util_str_occurence(string, pattern);
  gchar *nstring = NULL;
 
  if(occurrences)
  {
    nstring = g_malloc_n(strlen(string) + (occurrences * strlen(substitute)) + 1, sizeof(gchar));
    const gchar *pend = string + strlen(string);
    const gchar *s = string, *p = string;
    gchar *np = nstring;
    if((s = g_strstr_len(s, strlen(s), pattern)) != NULL)
    {
      do
      {
        memcpy(np, p, s - p);
        np += (s - p);
        memcpy(np, substitute, strlen(substitute));
        np += strlen(substitute);
        p = s + strlen(pattern);
      } while((s = g_strstr_len((s + 1), strlen(s + 1), pattern)) != NULL);
    }
    memcpy(np, p, pend - p);
    np[pend - p] = '\0';
  }
  else
    nstring = g_strdup(string); // otherwise it's a hell to decide whether to free this string later.
  return nstring;
}
 
gchar *dt_util_glist_to_str(const gchar *separator, GList *items)
{
  if(IS_NULL_PTR(items)) return NULL;
 
  const unsigned int count = g_list_length(items);
  gchar *result = NULL;
 
  // add the entries to an char* array
  gchar **strings = g_malloc0_n(count + 1, sizeof(gchar *));
  if(!IS_NULL_PTR(items))
  {
    int i = 0;
    for(; items; items = g_list_next(items))
    {
      strings[i++] = items->data;
    }
  }
 
  // join them into a single string
  result = g_strjoinv(separator, strings);
 
  // free the array
  dt_free(strings);
 
  return result;
}
 
GList *dt_util_glist_uniq(GList *items)
{
  if(IS_NULL_PTR(items)) return NULL;
 
  gchar *last = NULL;
  GList *last_item = NULL;
 
  items = g_list_sort(items, (GCompareFunc)g_strcmp0);
  GList *iter = items;
  while(iter)
  {
    gchar *value = (gchar *)iter->data;
    if(!g_strcmp0(last, value))
    {
      dt_free(value);
      items = g_list_delete_link(items, iter);
      iter = last_item;
    }
    else
    {
      last = value;
      last_item = iter;
    }
    iter = g_list_next(iter);
  }
  return items;
}
 
 
gchar *dt_util_fix_path(const gchar *path)
{
  if(IS_NULL_PTR(path) || *path == '\0')
  {
    return NULL;
  }
 
  gchar *rpath = NULL;
 
  /* check if path has a prepended tilde */
  if(path[0] == '~')
  {
    const size_t len = strlen(path);
    char *user = NULL;
    int off = 1;
 
    /* if the character after the tilde is not a slash we parse
     * the path until the next slash to extend this part with the
     * home directory of the specified user
     *
     * e.g.: ~foo will be evaluated as the home directory of the
     * user foo */
 
    if(len > 1 && path[1] != '/')
    {
      while(path[off] != '\0' && path[off] != '/')
      {
        ++off;
      }
 
      user = g_strndup(path + 1, off - 1);
    }
 
    gchar *home_path = dt_loc_get_home_dir(user);
    dt_free(user);
 
    if(IS_NULL_PTR(home_path))
    {
      return g_strdup(path);
    }
 
    rpath = g_build_filename(home_path, path + off, NULL);
    dt_free(home_path);
  }
  else
  {
    rpath = g_strdup(path);
  }
 
  return rpath;
}
 
size_t dt_utf8_strlcpy(char *dest, const char *src, size_t n)
{
  register const gchar *s = src;
  while(s - src < n && *s)
  {
    s = g_utf8_next_char(s);
  }
 
  if(s - src >= n)
  {
    /* We need to truncate; back up one. */
    s = g_utf8_prev_char(s);
    strncpy(dest, src, s - src);
    dest[s - src] = '\0';
    /* Find the full length for return value. */
    while(*s)
    {
      s = g_utf8_next_char(s);
    }
  }
  else
  {
    /* Plenty of room, just copy */
    strncpy(dest, src, s - src);
    dest[s - src] = '\0';
  }
  return s - src;
}
 
gboolean dt_util_test_image_file(const char *filename)
{
  if(g_access(filename, R_OK)) return FALSE;
#ifdef _WIN32
  struct _stati64 stats;
 
  // the code this replaced used utf8 paths with no problem
  // utf8 paths will not work in this context for no reason
  // that I can figure out, but converting utf8 to utf16 works
  // fine.
 
  wchar_t *wfilename = g_utf8_to_utf16(filename, -1, NULL, NULL, NULL);
  const int result = _wstati64(wfilename, &stats);
  dt_free(wfilename);
  if(result) return FALSE; // there was an error
 #else
  struct stat stats;
  if(stat(filename, &stats)) return FALSE;
#endif
 
  const gboolean regular = (S_ISREG(stats.st_mode)) != 0;
  const gboolean size_ok = stats.st_size > 0;
  //fprintf(stderr, "ERR: regular %i, size_ok %i.\n\tfor file: %s\n", regular, size_ok, filename);
  return regular && size_ok;
}
 
gboolean dt_util_test_writable_dir(const char *path)
{
  if(IS_NULL_PTR(path)) return FALSE;
#ifdef _WIN32
  struct _stati64 stats;
 
  wchar_t *wpath = g_utf8_to_utf16(path, -1, NULL, NULL, NULL);
  const int result = _wstati64(wpath, &stats);
  dt_free(wpath);
 
  if(result)
  { // error while testing path:
    return FALSE;
  }
#else
  struct stat stats;
  if(stat(path, &stats)) return FALSE;
#endif
  if(S_ISDIR(stats.st_mode) == 0) return FALSE;
  if(g_access(path, W_OK | X_OK) != 0) return FALSE;
  return TRUE;
}
 
gboolean dt_util_dir_exist(const char *dir)
{
  if(IS_NULL_PTR(dir))
    return 1;
 
  return g_file_test(dir, G_FILE_TEST_IS_DIR);
}
 
gboolean dt_util_is_dir_empty(const char *dirname)
{
  int n = 0;
  GDir *dir = g_dir_open(dirname, 0, NULL);
  if(IS_NULL_PTR(dir)) // Not a directory or doesn't exist
    return TRUE;
  while(g_dir_read_name(dir) != NULL)
  {
    if(++n > 1) break;
  }
  g_dir_close(dir);
  if(n == 0) // Directory Empty
    return TRUE;
  else
    return FALSE;
}
 
gchar *dt_util_foo_to_utf8(const char *string)
{
  gchar *tag = NULL;
 
  if(g_utf8_validate(string, -1, NULL)) // first check if it's utf8 already
    tag = g_strdup(string);
  else
    tag = g_convert(string, -1, "UTF-8", "LATIN1", NULL, NULL, NULL); // let's try latin1
 
  if(IS_NULL_PTR(tag)) // hmm, neither utf8 nor latin1, let's fall back to ascii and just remove everything that isn't
  {
    tag = g_strdup(string);
    char *c = tag;
    while(*c)
    {
      if((*c < 0x20) || (*c >= 0x7f)) *c = '?';
      c++;
    }
  }
  return tag;
}
 
// get easter sunday (in the western world)
static void easter(int Y, int* month, int *day)
{
  const int a  = Y % 19;
  const int b  = Y / 100;
  const int c  = Y % 100;
  const int d  = b / 4;
  const int e  = b % 4;
  const int f  = (b + 8) / 25;
  const int g  = (b - f + 1) / 3;
  const int h  = (19*a + b - d - g + 15) % 30;
  const int i  = c / 4;
  const int k  = c % 4;
  const int L  = (32 + 2*e + 2*i - h - k) % 7;
  const int m  = (a + 11*h + 22*L) / 451;
  *month = (h + L - 7*m + 114) / 31;
  *day   = ((h + L - 7*m + 114) % 31) + 1;
}
 
// days are in [1..31], months are in [0..11], see "man localtime"
dt_logo_season_t dt_util_get_logo_season(void)
{
  time_t now;
  time(&now);
  struct tm lt;
  localtime_r(&now, &lt);
 
  // Halloween is active on 31.10. and 01.11.
  if((lt.tm_mon == 9 && lt.tm_mday == 31) || (lt.tm_mon == 10 && lt.tm_mday == 1))
    return DT_LOGO_SEASON_HALLOWEEN;
 
  // Xmas is active from 24.12. until the end of the year
  if(lt.tm_mon == 11 && lt.tm_mday >= 24) return DT_LOGO_SEASON_XMAS;
 
  // Easter is active from 2 days before Easter Sunday until 1 day after
  {
    struct tm easter_sunday = lt;
    easter(lt.tm_year+1900, &easter_sunday.tm_mon, &easter_sunday.tm_mday);
    easter_sunday.tm_mon--;
    easter_sunday.tm_hour = easter_sunday.tm_min = easter_sunday.tm_sec = 0;
    easter_sunday.tm_isdst = -1;
    time_t easter_sunday_sec = mktime(&easter_sunday);
    // we start at midnight, so it's basically +- 2 days
    if(llabs(easter_sunday_sec - now) <= 2 * 24 * 60 * 60) return DT_LOGO_SEASON_EASTER;
  }
 
  return DT_LOGO_SEASON_NONE;
}
 
static cairo_surface_t *_util_get_svg_img(gchar *logo, const float size)
{
  GError *error = NULL;
  cairo_surface_t *surface = NULL;
  char datadir[DT_PATH_MAX] = { 0 };
 
  dt_loc_get_datadir(datadir, sizeof(datadir));
  char *dtlogo = g_build_filename(datadir, "pixmaps", logo, NULL);
  RsvgHandle *svg = rsvg_handle_new_from_file(dtlogo, &error);
  if(svg)
  {
    RsvgDimensionData dimension;
    dimension = dt_get_svg_dimension(svg);
 
    const float ppd = dt_screen_ppd();
 
    const float svg_size = MAX(dimension.width, dimension.height);
    const float factor = size > 0.0 ? size / svg_size : -1.0 * size;
    const float final_width = dimension.width * factor * ppd,
                final_height = dimension.height * factor * ppd;
    const int stride = cairo_format_stride_for_width(CAIRO_FORMAT_ARGB32, final_width);
 
    guint8 *image_buffer = (guint8 *)calloc(stride * final_height, sizeof(guint8));
    // ppd is 1.0 until a display is probed, and a device scale of 1.0 is what the plain
    // cairo call already does -- so there is nothing left for the old startup branch to
    // guard against.
    surface = dt_cairo_image_surface_create_for_data(image_buffer, CAIRO_FORMAT_ARGB32, final_width,
                                                     final_height, stride);
    if(cairo_surface_status(surface) != CAIRO_STATUS_SUCCESS)
    {
      fprintf(stderr, "warning: can't load darktable logo from SVG file `%s'\n", dtlogo);
      cairo_surface_destroy(surface);
      dt_free(image_buffer);
      surface = NULL;
    }
    else
    {
      cairo_t *cr = cairo_create(surface);
      cairo_scale(cr, factor, factor);
      dt_render_svg(svg, cr, dimension.width, dimension.height, 0, 0);
      cairo_destroy(cr);
      cairo_surface_flush(surface);
    }
    g_object_unref(svg);
  }
  else
  {
    fprintf(stderr, "warning: can't load darktable logo from SVG file `%s'\n%s\n", dtlogo, error->message);
    g_error_free(error);
  }
 
  dt_free(logo);
  dt_free(dtlogo);
 
  return surface;
}
 
cairo_surface_t *dt_util_get_logo(const float size)
{
  char *logo;
  const dt_logo_season_t season = dt_util_get_logo_season();
  if(season != DT_LOGO_SEASON_NONE)
    logo = g_strdup_printf("idbutton-%d.svg", (int)season);
  else
    logo = g_strdup("idbutton.svg");
 
  return _util_get_svg_img(logo, size);
}
 
cairo_surface_t *dt_util_get_logo_text(const float size)
{
  return _util_get_svg_img(g_strdup("dt_text.svg"), size);
}
 
// the following two functions (dt_util_latitude_str and dt_util_longitude_str) were taken from libosmgpsmap
// Copyright (C) 2013 John Stowers <john.stowers@gmail.com>
/* these can be overwritten with versions that support
 *   localization */
#define OSD_COORDINATES_CHR_N  "N"
#define OSD_COORDINATES_CHR_S  "S"
#define OSD_COORDINATES_CHR_E  "E"
#define OSD_COORDINATES_CHR_W  "W"
 
static const char *OSD_ELEVATION_ASL = N_("above sea level");
static const char *OSD_ELEVATION_BSL = N_("below sea level");
 
/* this is the classic geocaching notation */
gchar *dt_util_latitude_str(float latitude)
{
  gchar *c = OSD_COORDINATES_CHR_N;
  float integral, fractional;
 
  if(isnan(latitude)) return NULL;
 
  if(latitude < 0)
  {
    latitude = fabsf(latitude);
    c = OSD_COORDINATES_CHR_S;
  }
 
  fractional = modff(latitude, &integral);
 
  return g_strdup_printf("%s %02d\302\260 %06.3f'", c, (int)integral, fractional*60.0);
}
 
gchar *dt_util_longitude_str(float longitude)
{
  gchar *c = OSD_COORDINATES_CHR_E;
  float integral, fractional;
 
  if(isnan(longitude)) return NULL;
 
  if(longitude < 0)
  {
    longitude = fabsf(longitude);
    c = OSD_COORDINATES_CHR_W;
  }
 
  fractional = modff(longitude, &integral);
 
  return g_strdup_printf("%s %03d\302\260 %06.3f'", c, (int)integral, fractional*60.0);
}
 
gchar *dt_util_elevation_str(float elevation)
{
  const gchar *c = OSD_ELEVATION_ASL;
 
  if(isnan(elevation)) return NULL;
 
  if(elevation < 0)
  {
    elevation = fabsf(elevation);
    c = OSD_ELEVATION_BSL;
  }
 
  return g_strdup_printf("%.2f %s %s", elevation, _("m"), _(c));
}
 
/* a few helper functions inspired by
 *  https://projects.kde.org/projects/kde/kdegraphics/libs/libkexiv2/repository/revisions/master/entry/libkexiv2/kexiv2gps.cpp
 */
 
double dt_util_gps_string_to_number(const gchar *input)
{
  double res = NAN;
  gchar dir = toupper(input[strlen(input) - 1]);
  gchar **list = g_strsplit(input, ",", 0);
  if(list)
  {
    if(list[2] == NULL) // format DDD,MM.mm{N|S}
      res = g_ascii_strtoll(list[0], NULL, 10) + (g_ascii_strtod(list[1], NULL) / 60.0);
    else if(list[3] == NULL) // format DDD,MM,SS{N|S}
      res = g_ascii_strtoll(list[0], NULL, 10) + (g_ascii_strtoll(list[1], NULL, 10) / 60.0)
            + (g_ascii_strtoll(list[2], NULL, 10) / 3600.0);
    if(dir == 'S' || dir == 'W') res *= -1.0;
  }
  g_strfreev(list);
  return res;
}
 
gboolean dt_util_gps_rationale_to_number(const double r0_1, const double r0_2, const double r1_1,
                                         const double r1_2, const double r2_1, const double r2_2, char sign,
                                         double *result)
{
  if(IS_NULL_PTR(result)) return FALSE;
  double res = 0.0;
  // Latitude decoding from Exif.
  double num, den, min, sec;
  num = r0_1;
  den = r0_2;
  if(den == 0) return FALSE;
  res = num / den;
 
  num = r1_1;
  den = r1_2;
  if(den == 0) return FALSE;
  min = num / den;
  if(min != -1.0) res += min / 60.0;
 
  num = r2_1;
  den = r2_2;
  if(den == 0)
  {
    // be relaxed and accept 0/0 seconds. See #246077.
    if(num == 0)
      den = 1;
    else
      return FALSE;
  }
  sec = num / den;
  if(sec != -1.0) res += sec / 3600.0;
 
  if(sign == 'S' || sign == 'W') res *= -1.0;
 
  *result = res;
  return TRUE;
}
 
gboolean dt_util_gps_elevation_to_number(const double r_1, const double r_2, char sign, double *result)
{
  if(IS_NULL_PTR(result)) return FALSE;
  double res = 0.0;
  // Altitude decoding from Exif.
  const double num = r_1;
  const double den = r_2;
  if(den == 0) return FALSE;
  res = num / den;
 
  if(sign != '0') res *= -1.0;
 
  *result = res;
  return TRUE;
}
 
 
// make paths absolute and try to normalize on Windows. also deal with character encoding on Windows.
gchar *dt_util_normalize_path(const gchar *_input)
{
#ifdef _WIN32
  gchar *input;
  if(g_utf8_validate(_input, -1, NULL))
    input = g_strdup(_input);
  else
  {
    input = g_locale_to_utf8(_input, -1, NULL, NULL, NULL);
    if(IS_NULL_PTR(input)) return NULL;
  }
#else
  const gchar *input = _input;
#endif
 
  gchar *filename = g_filename_from_uri(input, NULL, NULL);
 
  if(!filename)
  {
    if(g_str_has_prefix(input, "file://")) // in this case we should take care of %XX encodings in the string
                                           // (for example %20 = ' ')
    {
      input += strlen("file://");
      filename = g_uri_unescape_string(input, NULL);
    }
    else
      filename = g_strdup(input);
  }
 
#ifdef _WIN32
  dt_free(input);
#endif
 
  if(g_path_is_absolute(filename) == FALSE)
  {
    char *current_dir = g_get_current_dir();
    char *tmp_filename = g_build_filename(current_dir, filename, NULL);
    dt_free(filename);
    filename = g_realpath(tmp_filename);
    if(IS_NULL_PTR(filename))
    {
      dt_free(current_dir);
      dt_free(tmp_filename);
      dt_free(filename);
      return NULL;
    }
    dt_free(current_dir);
    dt_free(tmp_filename);
  }
 
#ifdef _WIN32
  // on Windows filenames are case insensitive, so we can end up with an arbitrary number of different spellings for the same file.
  // another problem is that path separators can either be / or \ leading to even more problems.
 
  // TODO:
  // this handles filenames in the formats <drive letter>:\path\to\file or \\host-name\share-name\file
  // some other formats like \Device\... are not supported
 
  GFile *gfile = g_file_new_for_path(filename);
  dt_free(filename);
  if(IS_NULL_PTR(gfile))
    return NULL;
  filename = g_file_get_path(gfile);
  g_object_unref(gfile);
  if(!filename)
    return NULL;
 
  const char first = g_ascii_toupper(filename[0]);
  if(first >= 'A' && first <= 'Z' && filename[1] == ':') // path format is <drive letter>:\path\to\file
  {
    filename[0] = first;
    return filename;
  }
  else if(first == '\\' && filename[1] == '\\') // path format is \\host-name\share-name\file
    return filename;
  else
  {
    dt_free(filename);
    return NULL;
  }
#endif
 
  return filename;
}
 
#ifdef WIN32
// returns TRUE if the path is a Windows UNC (\\server\share\...\file)
const gboolean dt_util_path_is_UNC(const gchar *filename)
{
  return filename[0] == G_DIR_SEPARATOR && filename[1] == G_DIR_SEPARATOR;
}
#endif
 
// gets the directory components of a file name, like g_path_get_dirname(), but works also with Windows networks paths (\\hostname\share\file)
gchar *dt_util_path_get_dirname(const gchar *filename)
{
  gchar *dirname = g_path_get_dirname(filename);
 
  /* Remove trailing slash, as g_path_get_dirname() leaves it for Windows UNC and this messes up film roll name */
  if(dirname[0])
  {
    int last = strlen(dirname) - 1;
    if(G_IS_DIR_SEPARATOR(dirname[last]))
      dirname[last] = '\0';
  }
  return dirname;
}
 
 
GDateTime *dt_util_get_file_datetime(const char *const path)
{
  if(IS_NULL_PTR(path)) return NULL;
 
  GFile *file = g_file_new_for_path(path);
  GError *error = NULL;
  GFileInfo *info = g_file_query_info(file, G_FILE_ATTRIBUTE_STANDARD_NAME "," G_FILE_ATTRIBUTE_TIME_MODIFIED,
                                      G_FILE_QUERY_INFO_NONE, NULL, &error);
  if(IS_NULL_PTR(info))
  {
    if(error) g_error_free(error);
    g_object_unref(file);
    return NULL;
  }
 
  const guint64 datetime = g_file_info_get_attribute_uint64(info, G_FILE_ATTRIBUTE_TIME_MODIFIED);
  g_object_unref(file);
  g_object_unref(info);
  return g_date_time_new_from_unix_local(datetime);
}
 
 
guint dt_util_string_count_char(const char *text, const char needle)
{
  guint count = 0;
  while(text[0])
  {
    if(text[0] == needle) count ++;
    text ++;
  }
  return count;
}
 
void dt_util_str_to_loc_numbers_format(char *data)
{
  const struct lconv *currentLocalConv = localeconv();
  const gchar loc_decimal_point = currentLocalConv->decimal_point[0];
  const gchar *en_decimal_point = ".";
  g_strdelimit(data, en_decimal_point, loc_decimal_point);
}
 
GList *dt_util_str_to_glist(const gchar *separator, const gchar *text)
{
  if(IS_NULL_PTR(text)) return NULL;
  GList *list = NULL;
  gchar *item = NULL;
  gchar *entry = g_strdup(text);
  gchar *prev = entry;
  int len = strlen(prev);
  while (len)
  {
    gchar *next = g_strstr_len(prev, -1, separator);
    if (next)
    {
      const gchar c = next[0];
      next[0] = '\0';
      item = g_strdup(prev);
      next[0] = c;
      prev = next + strlen(separator);
      len = strlen(prev);
      list = g_list_prepend(list, item);
      if(!len) list = g_list_prepend(list, g_strdup(""));
    }
    else
    {
      item = g_strdup(prev);
      len = 0;
      list = g_list_prepend(list, item);
    }
  }
  list = g_list_reverse(list);
  dt_free(entry);
  return list;
}
 
// format exposure time given in seconds to a string in a unified way
char *dt_util_format_exposure(const float exposuretime)
{
  char *result = NULL;
  if(exposuretime >= 1.0f)
  {
    if(nearbyintf(exposuretime) == exposuretime)
      result = g_strdup_printf("%.0f\"", exposuretime);
    else
      result = g_strdup_printf("%.1f\"", exposuretime);
  }
  /* want to catch everything below 0.3 seconds */
  else if(exposuretime < 0.29f)
    result = g_strdup_printf("1/%.0f", 1.0 / exposuretime);
 
  /* catch 1/2, 1/3 */
  else if(nearbyintf(1.0f / exposuretime) == 1.0f / exposuretime)
    result = g_strdup_printf("1/%.0f", 1.0 / exposuretime);
 
  /* catch 1/1.3, 1/1.6, etc. */
  else if(10 * nearbyintf(10.0f / exposuretime) == nearbyintf(100.0f / exposuretime))
    result = g_strdup_printf("1/%.1f", 1.0 / exposuretime);
 
  else
    result = g_strdup_printf("%.1f\"", exposuretime);
 
  return result;
}
 
char *dt_read_file(const char *const filename, size_t *filesize)
{
  if (filesize) *filesize = 0;
  FILE *fd = g_fopen(filename, "rb");
  if(IS_NULL_PTR(fd)) return NULL;
 
  fseek(fd, 0, SEEK_END);
  const size_t end = ftell(fd);
  rewind(fd);
 
  char *content = (char *)malloc(sizeof(char) * end);
  if(IS_NULL_PTR(content)) return NULL;
 
  const size_t count = fread(content, sizeof(char), end, fd);
  fclose(fd);
  if (count == end)
  {
    if (filesize) *filesize = end;
    return content;
  }
  dt_free(content);
  return NULL;
}
 
void dt_copy_file(const char *const sourcefile, const char *dst)
{
  // Copied in fixed-size chunks, not slurped whole. The previous version sized a single
  // allocation from ftell() and read the entire file into it, which was wrong twice over:
  //
  //   - ftell() returns a SIGNED long and -1 on failure. Assigned to a size_t that becomes
  //     SIZE_MAX, and g_malloc_n() aborts the process on a failed allocation -- it is not
  //     the _try_ variant and does not return NULL for the caller's IS_NULL_PTR check.
  //   - the largest caller is the "copy original file" export format
  //     (imageio/format/copy.c), so "the whole file" is a raw: tens to hundreds of MB held
  //     in RAM to copy bytes that were never needed all at once.
  //
  // A fixed buffer has neither problem and needs no size query at all.
  FILE *fin = g_fopen(sourcefile, "rb");
  FILE *fout = g_fopen(dst, "wb");
 
  if(!IS_NULL_PTR(fin) && !IS_NULL_PTR(fout))
  {
    char buffer[64 * 1024];
    size_t bytes_read = fread(buffer, sizeof(char), sizeof(buffer), fin);
 
    while(bytes_read > 0)
    {
      if(fwrite(buffer, sizeof(char), bytes_read, fout) != bytes_read) break;
      bytes_read = fread(buffer, sizeof(char), sizeof(buffer), fin);
    }
  }
 
  if(!IS_NULL_PTR(fout)) fclose(fout);
  if(!IS_NULL_PTR(fin)) fclose(fin);
}
 
void dt_copy_resource_file(const char *src, const char *dst)
{
  char share[DT_PATH_MAX] = { 0 };
  dt_loc_get_datadir(share, sizeof(share));
  gchar *sourcefile = g_build_filename(share, src, NULL);
  dt_copy_file(sourcefile, dst);
  dt_free(sourcefile);
}
 
RsvgDimensionData dt_get_svg_dimension(RsvgHandle *svg)
{
  RsvgDimensionData dimension;
  // rsvg_handle_get_dimensions has been deprecated in librsvg 2.52
  #if LIBRSVG_CHECK_VERSION(2,52,0)
    double width;
    double height;
    if(rsvg_handle_get_intrinsic_size_in_pixels(svg, &width, &height)) //only works if SVG document has size specified
    {
      dimension.width = lround(width);
      dimension.height = lround(height);
    }
    else
    {
#define VIEWPORT_SIZE 32767 //use maximum cairo surface size to have enough precision when size is converted to int
      const RsvgRectangle viewport = {
        .x = 0,
        .y = 0,
        .width = VIEWPORT_SIZE,
        .height = VIEWPORT_SIZE,
      };
#undef VIEWPORT_SIZE
      RsvgRectangle rectangle;
      rsvg_handle_get_geometry_for_layer(svg, NULL, &viewport, NULL, &rectangle, NULL);
      dimension.width = lround(rectangle.width);
      dimension.height = lround(rectangle.height);
    }
  #else
    rsvg_handle_get_dimensions(svg, &dimension);
  #endif
  return dimension;
}
 
void dt_render_svg(RsvgHandle *svg, cairo_t *cr, double width, double height, double offset_x, double offset_y)
{
  // rsvg_handle_render_cairo has been deprecated in librsvg 2.52
  #if LIBRSVG_CHECK_VERSION(2,52,0)
    RsvgRectangle viewport = {
      .x = offset_x,
      .y = offset_y,
      .width = width,
      .height = height,
    };
    rsvg_handle_render_document(svg, cr, &viewport, NULL);
  #else
    rsvg_handle_render_cairo(svg, cr);
  #endif
}
 
// check if the path + basenames are the same (<=> only differ by the extension)
gboolean dt_has_same_path_basename(const char *filename1, const char *filename2)
{
  // assume both filenames have an extension
  if(!filename1 || !filename2) return FALSE;
  const char *dot1 = strrchr(filename1, '.');
  if(IS_NULL_PTR(dot1)) return FALSE;
  const char *dot2 = strrchr(filename2, '.');
  if(IS_NULL_PTR(dot2)) return FALSE;
  const int length1 = dot1 - filename1;
  const int length2 = dot2 - filename2;
  if(length1 != length2)
    return FALSE;
  for(int i = length1 - 1; i > 0; i--)
    if(filename1[i] != filename2[i])
      return FALSE;
  return TRUE;
}
 
// set the filename2 extension to filename1 - return NULL if fails - result should be freed
char *dt_copy_filename_extension(const char *filename1, const char *filename2)
{
  // assume both filenames have an extension
  if(!filename1 || !filename2) return NULL;
  const char *dot1 = strrchr(filename1, '.');
  if(IS_NULL_PTR(dot1)) return NULL;
  const char *dot2 = strrchr(filename2, '.');
  if(IS_NULL_PTR(dot2)) return NULL;
  const int name_lgth = dot1 - filename1;
  const int ext_lgth = strlen(dot2);
  char *output = g_malloc(name_lgth + ext_lgth + 1);
  if(output)
  {
    memcpy(output, filename1, name_lgth);
    memcpy(&output[name_lgth], &filename2[strlen(filename2) - ext_lgth], ext_lgth + 1);
  }
  return output;
}
 
// replaces all occurences of a substring in a string
gchar *dt_str_replace(const char *string, const char *search, const char *replace)
{
  gchar **split = g_strsplit(string, search, -1);
  gchar *res = g_strjoinv(replace, split);
  g_strfreev(split);
  return res;
}
 
// Checks for the opposite separator in a string and replace it by the needed one by the current OS
gchar *dt_cleanup_separators(gchar *string)
{
#ifdef WIN32
  string = dt_str_replace(string, "/", G_DIR_SEPARATOR_S);
#else
  string = dt_str_replace(string, "\\", G_DIR_SEPARATOR_S);
#endif
return string;
}
 
// remove trail and lead space of each folders and file name. Result should be freed.
gchar *dt_util_remove_whitespace(const gchar *path)
{
  gchar **split = g_strsplit(path, G_DIR_SEPARATOR_S, -1);
  for(int i = 0; i < g_strv_length(split); i++)
    g_strstrip(split[i]);
 
  char* result = g_strjoinv(G_DIR_SEPARATOR_S, split);
  g_strfreev(split);
 
  return result;
}
// clang-format off
// modelines: These editor modelines have been set for all relevant files by tools/update_modelines.py
// vim: shiftwidth=2 expandtab tabstop=2 cindent
// kate: tab-indents: off; indent-width 2; replace-tabs on; indent-mode cstyle; remove-trailing-spaces modified;
// clang-format on