1 /* GStreamer Split File Source
2 * Copyright (C) 2011 Collabora Ltd. <tim.muller@collabora.co.uk>
4 * This library is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Library General Public
6 * License as published by the Free Software Foundation; either
7 * version 2 of the License, or (at your option) any later version.
9 * This library is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Library General Public License for more details.
14 * You should have received a copy of the GNU Library General Public
15 * License along with this library; if not, write to the
16 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
17 * Boston, MA 02111-1307, USA.
20 * SECTION:element-splitfilesrc
21 * @see_also: #GstFileSrc, #GstMultiFileSrc
23 * Reads data from multiple files, presenting those files as one continuous
24 * file to downstream elements. This is useful for reading a large file that
25 * had to be split into multiple parts due to filesystem file size limitations,
28 * The files to select are chosen via the location property, which supports
29 * (and expects) shell-style wildcards (but only for the filename, not for
30 * directories). The results will be sorted.
33 * <title>Example launch lines</title>
35 * gst-launch splitfilesrc location="/path/to/part-*.mpg" ! decodebin ! ...
36 * ]| Plays the different parts as if they were one single MPEG file.
38 * gst-launch playbin uri="splitfile://path/to/foo.avi.*"
39 * ]| Plays the different parts as if they were one single AVI file.
49 #include "gstsplitfilesrc.h"
50 #include "patternspec.h"
55 #define DEFAULT_PATTERN_MATCH_MODE MATCH_MODE_UTF8
57 #define DEFAULT_PATTERN_MATCH_MODE MATCH_MODE_AUTO
65 #define DEFAULT_LOCATION NULL
67 static void gst_split_file_src_uri_handler_init (gpointer g_iface,
69 static void gst_split_file_src_set_property (GObject * object, guint prop_id,
70 const GValue * value, GParamSpec * pspec);
71 static void gst_split_file_src_get_property (GObject * object, guint prop_id,
72 GValue * value, GParamSpec * pspec);
73 static void gst_split_file_src_finalize (GObject * obj);
75 static gboolean gst_split_file_src_start (GstBaseSrc * basesrc);
76 static gboolean gst_split_file_src_stop (GstBaseSrc * basesrc);
77 static gboolean gst_split_file_src_can_seek (GstBaseSrc * basesrc);
78 static gboolean gst_split_file_src_get_size (GstBaseSrc * basesrc, guint64 * s);
79 static gboolean gst_split_file_src_unlock (GstBaseSrc * basesrc);
80 static GstFlowReturn gst_split_file_src_create (GstBaseSrc * basesrc,
81 guint64 offset, guint size, GstBuffer ** buffer);
83 static GstStaticPadTemplate gst_split_file_src_pad_template =
84 GST_STATIC_PAD_TEMPLATE ("src",
89 GST_DEBUG_CATEGORY_STATIC (splitfilesrc_debug);
90 #define GST_CAT_DEFAULT splitfilesrc_debug
93 G_DEFINE_TYPE_WITH_CODE (GstSplitFileSrc, gst_split_file_src, GST_TYPE_BASE_SRC,
94 G_IMPLEMENT_INTERFACE (GST_TYPE_URI_HANDLER,
95 gst_split_file_src_uri_handler_init));
98 #define WIN32_BLURB " Location string must be in UTF-8 encoding (on Windows)."
100 #define WIN32_BLURB /* nothing */
104 gst_split_file_src_class_init (GstSplitFileSrcClass * klass)
106 GstBaseSrcClass *gstbasesrc_class = GST_BASE_SRC_CLASS (klass);
107 GstElementClass *gstelement_class = GST_ELEMENT_CLASS (klass);
108 GObjectClass *gobject_class = G_OBJECT_CLASS (klass);
110 gobject_class->set_property = gst_split_file_src_set_property;
111 gobject_class->get_property = gst_split_file_src_get_property;
112 gobject_class->finalize = gst_split_file_src_finalize;
114 g_object_class_install_property (gobject_class, PROP_LOCATION,
115 g_param_spec_string ("location", "File Location",
116 "Wildcard pattern to match file names of the input files. If "
117 "the location is an absolute path or contains directory components, "
118 "only the base file name part will be considered for pattern "
119 "matching. The results will be sorted." WIN32_BLURB,
120 DEFAULT_LOCATION, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
122 gstbasesrc_class->start = GST_DEBUG_FUNCPTR (gst_split_file_src_start);
123 gstbasesrc_class->stop = GST_DEBUG_FUNCPTR (gst_split_file_src_stop);
124 gstbasesrc_class->create = GST_DEBUG_FUNCPTR (gst_split_file_src_create);
125 gstbasesrc_class->get_size = GST_DEBUG_FUNCPTR (gst_split_file_src_get_size);
126 gstbasesrc_class->unlock = GST_DEBUG_FUNCPTR (gst_split_file_src_unlock);
127 gstbasesrc_class->is_seekable =
128 GST_DEBUG_FUNCPTR (gst_split_file_src_can_seek);
130 GST_DEBUG_CATEGORY_INIT (splitfilesrc_debug, "splitfilesrc", 0,
131 "splitfilesrc element");
133 gst_element_class_add_pad_template (gstelement_class,
134 gst_static_pad_template_get (&gst_split_file_src_pad_template));
136 gst_element_class_set_static_metadata (gstelement_class, "Split-File Source",
138 "Read a sequentially named set of files as if it was one large file",
139 "Tim-Philipp Müller <tim.muller@collabora.co.uk>");
143 gst_split_file_src_init (GstSplitFileSrc * splitfilesrc)
148 gst_split_file_src_finalize (GObject * obj)
150 GstSplitFileSrc *src = GST_SPLIT_FILE_SRC (obj);
152 g_free (src->location);
153 src->location = NULL;
155 G_OBJECT_CLASS (gst_split_file_src_parent_class)->finalize (obj);
159 gst_split_file_src_can_seek (GstBaseSrc * basesrc)
165 gst_split_file_src_unlock (GstBaseSrc * basesrc)
167 /* This is not actually that useful, since all normal file
168 * operations are fully blocking anyway */
170 GstSplitFileSrc *src = GST_SPLIT_FILE_SRC (basesrc);
172 GST_DEBUG_OBJECT (src, "cancelling pending I/O operation if there is one");
173 /* g_cancellable_cancel (src->cancellable); */
174 GST_DEBUG_OBJECT (src, "done");
181 gst_split_file_src_get_size (GstBaseSrc * basesrc, guint64 * size)
183 GstSplitFileSrc *src = GST_SPLIT_FILE_SRC (basesrc);
185 *size = src->parts[src->num_parts - 1].stop + 1;
190 gst_split_file_src_set_location (GstSplitFileSrc * src, const char *location)
192 GST_OBJECT_LOCK (src);
193 g_free (src->location);
195 if (location != NULL && g_str_has_prefix (location, "splitfile://"))
196 src->location = gst_uri_get_location (location);
198 src->location = g_strdup (location);
200 if (!g_utf8_validate (src->location, -1, NULL)) {
201 g_warning ("splitfilesrc 'location' property must be in UTF-8 "
202 "encoding on Windows");
205 GST_OBJECT_UNLOCK (src);
209 gst_split_file_src_set_property (GObject * object, guint prop_id,
210 const GValue * value, GParamSpec * pspec)
212 GstSplitFileSrc *src = GST_SPLIT_FILE_SRC (object);
216 gst_split_file_src_set_location (src, g_value_get_string (value));
219 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
225 gst_split_file_src_get_property (GObject * object, guint prop_id,
226 GValue * value, GParamSpec * pspec)
228 GstSplitFileSrc *src = GST_SPLIT_FILE_SRC (object);
232 GST_OBJECT_LOCK (src);
233 g_value_set_string (value, src->location);
234 GST_OBJECT_UNLOCK (src);
237 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
243 gst_split_file_src_array_sortfunc (gchar ** a, gchar ** b)
245 return strcmp (*a, *b);
249 gst_split_file_src_find_files (GstSplitFileSrc * src, const gchar * dirname,
250 const gchar * basename, GError ** err)
257 if (dirname == NULL || basename == NULL)
258 goto invalid_location;
260 GST_INFO_OBJECT (src, "checking in directory '%s' for pattern '%s'",
263 dir = g_dir_open (dirname, 0, err);
267 if (DEFAULT_PATTERN_MATCH_MODE == MATCH_MODE_UTF8 &&
268 !g_utf8_validate (basename, -1, NULL)) {
272 /* mode will be AUTO on linux/unix and UTF8 on win32 */
273 pspec = pattern_spec_new (basename, DEFAULT_PATTERN_MATCH_MODE);
275 files = g_ptr_array_new ();
277 while ((name = g_dir_read_name (dir))) {
278 GST_TRACE_OBJECT (src, "check: %s", name);
279 if (pattern_match_string (pspec, name)) {
280 GST_DEBUG_OBJECT (src, "match: %s", name);
281 g_ptr_array_add (files, g_build_filename (dirname, name, NULL));
288 g_ptr_array_sort (files, (GCompareFunc) gst_split_file_src_array_sortfunc);
289 g_ptr_array_add (files, NULL);
291 pattern_spec_free (pspec);
294 return (gchar **) g_ptr_array_free (files, FALSE);
299 g_set_error_literal (err, G_FILE_ERROR, G_FILE_ERROR_INVAL,
300 "No filename specified.");
306 g_set_error_literal (err, G_FILE_ERROR, G_FILE_ERROR_INVAL,
307 "Filename pattern must be UTF-8 on Windows.");
312 pattern_spec_free (pspec);
314 g_set_error_literal (err, G_FILE_ERROR, G_FILE_ERROR_NOENT,
315 "Found no files matching the pattern.");
321 gst_split_file_src_start (GstBaseSrc * basesrc)
323 GstSplitFileSrc *src = GST_SPLIT_FILE_SRC (basesrc);
324 GCancellable *cancel;
325 gboolean ret = FALSE;
328 gchar *basename = NULL;
329 gchar *dirname = NULL;
333 GST_OBJECT_LOCK (src);
334 if (src->location != NULL && src->location[0] != '\0') {
335 basename = g_path_get_basename (src->location);
336 dirname = g_path_get_dirname (src->location);
338 GST_OBJECT_UNLOCK (src);
340 files = gst_split_file_src_find_files (src, dirname, basename, &err);
342 if (files == NULL || *files == NULL)
345 src->num_parts = g_strv_length (files);
346 src->parts = g_new0 (GstFilePart, src->num_parts);
348 cancel = src->cancellable;
351 for (i = 0; i < src->num_parts; ++i) {
352 GFileInputStream *stream;
357 file = g_file_new_for_path (files[i]);
358 stream = g_file_read (file, cancel, &err);
359 g_object_unref (file);
362 goto open_read_error;
364 info = g_file_input_stream_query_info (stream, "standard::*", NULL, &err);
366 g_object_unref (stream);
367 goto query_info_error;
370 size = g_file_info_get_size (info);
371 g_object_unref (info);
373 src->parts[i].stream = stream;
374 src->parts[i].path = g_strdup (files[i]);
375 src->parts[i].start = offset;
376 src->parts[i].stop = offset + size - 1;
378 GST_DEBUG ("[%010" G_GUINT64_FORMAT "-%010" G_GUINT64_FORMAT "] %s",
379 src->parts[i].start, src->parts[i].stop, src->parts[i].path);
384 GST_INFO ("Successfully opened %u file parts for reading", src->num_parts);
388 src->cancellable = g_cancellable_new ();
403 if (err->code == G_IO_ERROR_CANCELLED)
406 GST_ELEMENT_ERROR (src, RESOURCE, OPEN_READ, ("%s", err->message),
407 ("Failed to find files in '%s' for pattern '%s'",
408 GST_STR_NULL (dirname), GST_STR_NULL (basename)));
413 if (err->code == G_IO_ERROR_CANCELLED)
416 GST_ELEMENT_ERROR (src, RESOURCE, OPEN_READ, ("%s", err->message),
417 ("Failed to open file '%s' for reading", files[i]));
422 if (err->code == G_IO_ERROR_CANCELLED)
425 GST_ELEMENT_ERROR (src, RESOURCE, OPEN_READ, ("%s", err->message),
426 ("Failed to query info for file '%s'", files[i]));
431 GST_DEBUG_OBJECT (src, "I/O operation cancelled from another thread");
437 gst_split_file_src_stop (GstBaseSrc * basesrc)
439 GstSplitFileSrc *src = GST_SPLIT_FILE_SRC (basesrc);
442 for (i = 0; i < src->num_parts; ++i) {
443 if (src->parts[i].stream != NULL)
444 g_object_unref (src->parts[i].stream);
445 g_free (src->parts[i].path);
451 g_object_unref (src->cancellable);
452 src->cancellable = NULL;
458 gst_split_file_src_find_part_for_offset (GstSplitFileSrc * src, guint64 offset,
464 /* TODO: could use gst_util_array_binary_search() here */
466 for (i = 0; i < src->num_parts; ++i) {
467 if (offset >= part->start && offset <= part->stop) {
478 gst_split_file_src_create (GstBaseSrc * basesrc, guint64 offset, guint size,
481 GstSplitFileSrc *src = GST_SPLIT_FILE_SRC (basesrc);
482 GstFilePart cur_part;
483 GInputStream *stream;
484 GCancellable *cancel;
493 cur_part = src->parts[src->cur_part];
494 if (offset < cur_part.start || offset > cur_part.stop) {
495 if (!gst_split_file_src_find_part_for_offset (src, offset, &src->cur_part))
497 cur_part = src->parts[src->cur_part];
500 GST_LOG_OBJECT (src, "current part: %u (%" G_GUINT64_FORMAT " - "
501 "%" G_GUINT64_FORMAT ", %s)", src->cur_part, cur_part.start,
502 cur_part.stop, cur_part.path);
504 buf = gst_buffer_new_allocate (NULL, size, NULL);
506 GST_BUFFER_OFFSET (buf) = offset;
508 gst_buffer_map (buf, &map, GST_MAP_WRITE);
511 cancel = src->cancellable;
514 guint64 bytes_to_end_of_part;
517 /* we want the offset into the file part */
518 read_offset = offset - cur_part.start;
520 GST_LOG ("Reading part %03u from offset %" G_GUINT64_FORMAT " (%s)",
521 src->cur_part, read_offset, cur_part.path);
523 /* FIXME: only seek when needed (hopefully gio is smart) */
524 seekable = G_SEEKABLE (cur_part.stream);
525 if (!g_seekable_seek (seekable, read_offset, G_SEEK_SET, cancel, &err))
528 GST_LOG_OBJECT (src, "now: %" G_GUINT64_FORMAT, g_seekable_tell (seekable));
530 bytes_to_end_of_part = (cur_part.stop - cur_part.start) + 1 - read_offset;
531 to_read = MIN (size, bytes_to_end_of_part);
533 GST_LOG_OBJECT (src, "reading %u bytes from part %u (bytes to end of "
534 "part: %u)", to_read, src->cur_part, (guint) bytes_to_end_of_part);
536 stream = G_INPUT_STREAM (cur_part.stream);
538 /* NB: we won't try to read beyond EOF */
539 if (!g_input_stream_read_all (stream, data, to_read, &read, cancel, &err))
542 GST_LOG_OBJECT (src, "read %u bytes", (guint) read);
552 GST_LOG_OBJECT (src, "%u bytes left to read for this chunk", size);
554 /* corner case, this should never really happen (assuming basesrc clips
555 * requests beyond the file size) */
556 if (read < to_read) {
557 if (src->cur_part == src->num_parts - 1) {
558 /* last file part, stop reading and truncate buffer */
559 gst_buffer_set_size (buf, offset - GST_BUFFER_OFFSET (buf));
562 goto file_part_changed;
567 cur_part = src->parts[src->cur_part];
570 GST_BUFFER_OFFSET_END (buf) = offset;
572 gst_buffer_unmap (buf, &map);
575 GST_LOG_OBJECT (src, "read %" G_GSIZE_FORMAT " bytes into buf %p",
576 gst_buffer_get_size (buf), buf);
582 if (err->code == G_IO_ERROR_CANCELLED)
585 GST_ELEMENT_ERROR (src, RESOURCE, SEEK, (NULL),
586 ("Seek to %" G_GUINT64_FORMAT " in %s failed", read_offset,
589 gst_buffer_unref (buf);
590 return GST_FLOW_ERROR;
594 if (err->code == G_IO_ERROR_CANCELLED)
597 GST_ELEMENT_ERROR (src, RESOURCE, READ, ("%s", err->message),
598 ("Read from %" G_GUINT64_FORMAT " in %s failed", read_offset,
601 gst_buffer_unref (buf);
602 return GST_FLOW_ERROR;
606 GST_ELEMENT_ERROR (src, RESOURCE, READ,
607 ("Read error while reading file part %s", cur_part.path),
608 ("Short read in file part, file may have been modified since start"));
609 gst_buffer_unref (buf);
610 return GST_FLOW_ERROR;
614 GST_DEBUG_OBJECT (src, "I/O operation cancelled from another thread");
616 gst_buffer_unref (buf);
617 return GST_FLOW_FLUSHING;
622 gst_split_file_src_uri_get_type (GType type)
627 static const gchar *const *
628 gst_split_file_src_uri_get_protocols (GType type)
630 static const gchar *protocols[] = { "splitfile", NULL };
632 return (const gchar * const *) protocols;
636 gst_split_file_src_uri_get_uri (GstURIHandler * handler)
638 GstSplitFileSrc *src = GST_SPLIT_FILE_SRC (handler);
641 GST_OBJECT_LOCK (src);
642 if (src->location != NULL)
643 ret = g_strdup_printf ("splitfile://%s", src->location);
646 GST_OBJECT_UNLOCK (src);
652 gst_split_file_src_uri_set_uri (GstURIHandler * handler, const gchar * uri,
655 GstSplitFileSrc *src = GST_SPLIT_FILE_SRC (handler);
657 gst_split_file_src_set_location (src, uri);
663 gst_split_file_src_uri_handler_init (gpointer g_iface, gpointer iface_data)
665 GstURIHandlerInterface *iface = (GstURIHandlerInterface *) g_iface;
667 iface->get_type = gst_split_file_src_uri_get_type;
668 iface->get_protocols = gst_split_file_src_uri_get_protocols;
669 iface->get_uri = gst_split_file_src_uri_get_uri;
670 iface->set_uri = gst_split_file_src_uri_set_uri;