gstreamer/libs/gst/bytestream/bytestream.c
Wim Taymans 086de421dc Totally rewritten registry handling.
Original commit message from CVS:
Totally rewritten registry handling.
- move the registry save/load code into a gstregistry subclass, this
will make it possible to use other registries (flat file, web based,
RDBMS type, etc..)
- a simple GMarkup xml registry is implemented
- use standard statically linked plugins for core elements.
- GstPlugin has a very well defined set of functions now
A little bytestream hack..
Added more info to -inspect.
Some more debugging info for clocking.
Small cleanups

I use ./gst-register --gst-plugin-path=/opt/src/sourceforge/gst-plugins/gst-libs:/opt/src/sourceforge/gst-plugins/
to register core and gst-plugins now.
2002-05-08 20:40:48 +00:00

499 lines
14 KiB
C

/* GStreamer
* Copyright (C) 2001 Erik Walthinsen <omega@temple-baptist.com>
*
* gstbytestream.c: adds a convenient bytestream based API to a pad.
*
* 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 <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <gst/gstinfo.h>
#include "bytestream.h"
/* #define BS_DEBUG */
#ifdef BS_DEBUG
# define bs_print(format,args...) GST_DEBUG (GST_CAT_BUFFER, format, ## args)
# define bs_status(bs) gst_bytestream_print_status(bs)
#else
# define bs_print(format,args...)
# define bs_status(bs)
#endif
guint8 *gst_bytestream_assemble (GstByteStream * bs, guint32 len);
/**
* gst_bytestream_new:
* @pad: the pad to attach the bytestream to
*
* creates a bytestream from the given pad
*
* Returns: a new #GstByteStream object
*/
GstByteStream *
gst_bytestream_new (GstPad * pad)
{
GstByteStream *bs = g_new (GstByteStream, 1);
bs->pad = pad;
bs->event = NULL;
bs->buflist = NULL;
bs->headbufavail = 0;
bs->listavail = 0;
bs->assembled = NULL;
return bs;
}
void
gst_bytestream_destroy (GstByteStream * bs)
{
GSList *walk;
if (bs->event)
gst_event_free (bs->event);
walk = bs->buflist;
while (walk) {
gst_buffer_unref (GST_BUFFER (walk->data));
walk = g_slist_next (walk);
}
g_slist_free (bs->buflist);
if (bs->assembled)
g_free (bs->assembled);
g_free (bs);
}
/* HOW THIS WORKS:
*
* The fundamental structure is a singly-linked list of buffers. The
* buffer on the front is the oldest, and thus the first to read data
* from. The number of bytes left to be read in this buffer is stored
* in bs->headbufavail. The number of bytes available in the entire
* list (including the head buffer) is in bs->listavail.
*
* When a request is made for data (peek), _fill_bytes is called with
* the number of bytes needed, but only if the listavail indicates
* that there aren't already enough. This calls _get_next_buf until
* the listavail is sufficient to satisfy the demand.
*
* _get_next_buf pulls a buffer from the pad the bytestream is attached
* to, and shoves it in the list. There are actually two things it can
* do. If there's already a buffer in the list, and the _is_span_fast()
* test returns true, it will merge it with that last buffer. Otherwise
* it will simply tack it onto the end of the list.
*
* The _peek itself first checks the simple case of the request fitting
* within the head buffer, and if so creates a subbuffer and returns.
* Otherwise, it creates a new buffer and allocates space for the request
* and calls _assemble to fill it. We know we have to copy because this
* case only happens when the _merge wasn't feasible during _get_next_buf.
*
* The _flush method repeatedly inspects the head buffer and flushes as
* much data from it as it needs to, up to the size of the buffer. If
* the flush decimates the buffer, it's stripped, unref'd, and removed.
*/
/* get the next buffer
* if the buffer can be merged with the head buffer, do so
* else add it onto the head of the
*/
static gboolean
gst_bytestream_get_next_buf (GstByteStream * bs)
{
GstBuffer *nextbuf, *lastbuf;
GSList *end;
g_assert (!bs->event);
bs_print ("get_next_buf: pulling buffer\n");
nextbuf = gst_pad_pull (bs->pad);
if (GST_IS_EVENT (nextbuf))
{
bs->event = GST_EVENT (nextbuf);
return FALSE;
}
if (!nextbuf)
return FALSE;
bs_print ("get_next_buf: got buffer of %d bytes\n", GST_BUFFER_SIZE (nextbuf));
/* first see if there are any buffers in the list at all */
if (bs->buflist) {
bs_print ("gst_next_buf: there is at least one buffer in the list\n");
/* now find the end of the list */
end = g_slist_last (bs->buflist);
/* get the buffer that's there */
lastbuf = GST_BUFFER (end->data);
/* see if we can marge cheaply */
if (gst_buffer_is_span_fast (lastbuf, nextbuf)) {
bs_print ("get_next_buf: merging new buffer with last buf on list\n");
/* it is, let's merge them (this is really an append, but...) */
end->data = gst_buffer_merge (lastbuf, nextbuf);
/* add to the length of the list */
bs->listavail += GST_BUFFER_SIZE (nextbuf);
/* have to check to see if we merged with the head buffer */
if (end == bs->buflist) {
bs->headbufavail += GST_BUFFER_SIZE (nextbuf);
}
gst_buffer_unref (lastbuf);
gst_buffer_unref (nextbuf);
/* if we can't, we just append this buffer */
}
else {
bs_print ("get_next_buf: adding new buffer to the end of the list\n");
end = g_slist_append (end, nextbuf);
/* also need to increment length of list and buffer count */
bs->listavail += GST_BUFFER_SIZE (nextbuf);
}
/* if there are no buffers in the list */
}
else {
bs_print ("get_next_buf: buflist is empty, adding new buffer to list\n");
/* put this on the end of the list */
bs->buflist = g_slist_append (bs->buflist, nextbuf);
/* and increment the number of bytes in the list */
bs->listavail = GST_BUFFER_SIZE (nextbuf);
/* set the head buffer avail to the size */
bs->headbufavail = GST_BUFFER_SIZE (nextbuf);
}
return TRUE;
}
static gboolean
gst_bytestream_fill_bytes (GstByteStream * bs, guint32 len)
{
/* as long as we don't have enough, we get more buffers */
while (bs->listavail < len) {
bs_print ("fill_bytes: there are %d bytes in the list, we need %d\n", bs->listavail, len);
if (!gst_bytestream_get_next_buf (bs))
return FALSE;
}
return TRUE;
}
GstBuffer *
gst_bytestream_peek (GstByteStream * bs, guint32 len)
{
GstBuffer *headbuf, *retbuf = NULL;
g_return_val_if_fail (bs != NULL, NULL);
g_return_val_if_fail (len > 0, NULL);
bs_print ("peek: asking for %d bytes\n", len);
/* make sure we have enough */
bs_print ("peek: there are %d bytes in the list\n", bs->listavail);
if (len > bs->listavail) {
if (!gst_bytestream_fill_bytes (bs, len))
return NULL;
bs_print ("peek: there are now %d bytes in the list\n", bs->listavail);
}
bs_status (bs);
/* extract the head buffer */
headbuf = GST_BUFFER (bs->buflist->data);
/* if the requested bytes are in the current buffer */
bs_print ("peek: headbufavail is %d\n", bs->headbufavail);
if (len <= bs->headbufavail) {
bs_print ("peek: there are enough bytes in headbuf (need %d, have %d)\n", len, bs->headbufavail);
/* create a sub-buffer of the headbuf */
retbuf = gst_buffer_create_sub (headbuf, GST_BUFFER_SIZE (headbuf) - bs->headbufavail, len);
/* otherwise we need to figure out how to assemble one */
}
else {
bs_print ("peek: current buffer is not big enough for len %d\n", len);
retbuf = gst_buffer_new ();
GST_BUFFER_SIZE (retbuf) = len;
GST_BUFFER_DATA (retbuf) = gst_bytestream_assemble (bs, len);
if (GST_BUFFER_OFFSET (headbuf) != -1)
GST_BUFFER_OFFSET (retbuf) = GST_BUFFER_OFFSET (headbuf) + (GST_BUFFER_SIZE (headbuf) - bs->headbufavail);
}
return retbuf;
}
guint8 *
gst_bytestream_peek_bytes (GstByteStream * bs, guint32 len)
{
GstBuffer *headbuf;
guint8 *data = NULL;
g_return_val_if_fail (bs != NULL, NULL);
g_return_val_if_fail (len > 0, NULL);
bs_print ("peek_bytes: asking for %d bytes\n", len);
if (bs->assembled) {
g_free (bs->assembled);
bs->assembled = NULL;
}
/* make sure we have enough */
bs_print ("peek_bytes: there are %d bytes in the list\n", bs->listavail);
if (len > bs->listavail) {
if (!gst_bytestream_fill_bytes (bs, len))
return NULL;
bs_print ("peek_bytes: there are now %d bytes in the list\n", bs->listavail);
}
bs_status (bs);
/* extract the head buffer */
headbuf = GST_BUFFER (bs->buflist->data);
/* if the requested bytes are in the current buffer */
bs_print ("peek_bytes: headbufavail is %d\n", bs->headbufavail);
if (len <= bs->headbufavail) {
bs_print ("peek_bytes: there are enough bytes in headbuf (need %d, have %d)\n", len, bs->headbufavail);
/* create a sub-buffer of the headbuf */
data = GST_BUFFER_DATA (headbuf) + (GST_BUFFER_SIZE (headbuf) - bs->headbufavail);
/* otherwise we need to figure out how to assemble one */
}
else {
bs_print ("peek_bytes: current buffer is not big enough for len %d\n", len);
data = gst_bytestream_assemble (bs, len);
bs->assembled = data;
bs->assembled_len = len;
}
return data;
}
guint8 *
gst_bytestream_assemble (GstByteStream * bs, guint32 len)
{
guint8 *data = g_malloc (len);
GSList *walk;
guint32 copied = 0;
GstBuffer *buf;
/* copy the data from the curbuf */
buf = GST_BUFFER (bs->buflist->data);
bs_print ("assemble: copying %d bytes from curbuf at %d to *data\n", bs->headbufavail,
GST_BUFFER_SIZE (buf) - bs->headbufavail);
memcpy (data, GST_BUFFER_DATA (buf) + GST_BUFFER_SIZE (buf) - bs->headbufavail, bs->headbufavail);
copied += bs->headbufavail;
/* asumption is made that the buffers all exist in the list */
walk = g_slist_next (bs->buflist);
while (copied < len) {
buf = GST_BUFFER (walk->data);
if (GST_BUFFER_SIZE (buf) < (len - copied)) {
bs_print ("assemble: copying %d bytes from buf to output offset %d\n", GST_BUFFER_SIZE (buf), copied);
memcpy (data + copied, GST_BUFFER_DATA (buf), GST_BUFFER_SIZE (buf));
copied += GST_BUFFER_SIZE (buf);
}
else {
bs_print ("assemble: copying %d bytes from buf to output offset %d\n", len - copied, copied);
memcpy (data + copied, GST_BUFFER_DATA (buf), len - copied);
copied = len;
}
walk = g_slist_next (walk);
}
return data;
}
gboolean
gst_bytestream_flush (GstByteStream * bs, guint32 len)
{
bs_print ("flush: flushing %d bytes\n", len);
/* make sure we have enough */
bs_print ("flush: there are %d bytes in the list\n", bs->listavail);
if (len > bs->listavail) {
if (!gst_bytestream_fill_bytes (bs, len))
return FALSE;
bs_print ("flush: there are now %d bytes in the list\n", bs->listavail);
}
gst_bytestream_flush_fast (bs, len);
return TRUE;
}
void
gst_bytestream_flush_fast (GstByteStream * bs, guint32 len)
{
GstBuffer *headbuf;
g_assert (len <= bs->listavail);
if (bs->assembled) {
g_free (bs->assembled);
bs->assembled = NULL;
}
/* repeat until we've flushed enough data */
while (len > 0) {
headbuf = GST_BUFFER (bs->buflist->data);
bs_print ("flush: analyzing buffer that's %d bytes long, offset %d\n", GST_BUFFER_SIZE (headbuf),
GST_BUFFER_OFFSET (headbuf));
/* if there's enough to complete the flush */
if (bs->headbufavail > len) {
/* just trim it off */
bs_print ("flush: trimming %d bytes off end of headbuf\n", len);
bs->headbufavail -= len;
bs->listavail -= len;
len = 0;
/* otherwise we have to trim the whole buffer */
}
else {
bs_print ("flush: removing head buffer completely\n");
/* remove it from the list */
bs->buflist = g_slist_delete_link (bs->buflist, bs->buflist);
/* trim it from the avail size */
bs->listavail -= bs->headbufavail;
/* record that we've trimmed this many bytes */
len -= bs->headbufavail;
/* unref it */
gst_buffer_unref (headbuf);
/* record the new headbufavail */
if (bs->buflist) {
bs->headbufavail = GST_BUFFER_SIZE (GST_BUFFER (bs->buflist->data));
bs_print ("flush: next headbuf is %d bytes\n", bs->headbufavail);
}
else {
bs_print ("flush: no more bytes at all\n");
}
}
bs_print ("flush: bottom of while(), len is now %d\n", len);
}
}
gboolean
gst_bytestream_seek (GstByteStream *bs, GstSeekType type, gint64 offset)
{
GstRealPad *peer = GST_RPAD_PEER (bs->pad);
if (gst_pad_send_event (GST_PAD (peer), gst_event_new_seek (type, offset, TRUE))) {
GstBuffer *nextbuf;
gst_bytestream_flush_fast (bs, bs->listavail);
do {
nextbuf = gst_pad_pull (bs->pad);
}
while (!GST_IS_EVENT (nextbuf));
return TRUE;
}
return FALSE;
}
guint64
gst_bytestream_tell (GstByteStream *bs)
{
return 0;
}
GstBuffer *
gst_bytestream_read (GstByteStream * bs, guint32 len)
{
GstBuffer *buf = gst_bytestream_peek (bs, len);
if (!buf)
return NULL;
gst_bytestream_flush_fast (bs, len);
return buf;
}
/**
* gst_bytestream_get_status
* @bs: a bytestream
* @avail_out: total number of bytes buffered
* @event_out: an event
*
* When an event occurs, the bytestream will return NULL. You must
* retrieve the event using this API before reading more bytes from
* the stream.
*
* It is possible for the bytestream to return NULL due to running
* out of buffers, however, this indicates a bug because an EOS
* event should have been sent.
*/
void
gst_bytestream_get_status (GstByteStream *bs,
guint32 *avail_out,
GstEvent **event_out)
{
if (avail_out)
*avail_out = bs->listavail;
if (event_out)
{
*event_out = bs->event;
bs->event = NULL;
}
}
void
gst_bytestream_print_status (GstByteStream * bs)
{
GSList *walk;
GstBuffer *buf;
bs_print ("STATUS: head buffer has %d bytes available\n", bs->headbufavail);
bs_print ("STATUS: list has %d bytes available\n", bs->listavail);
walk = bs->buflist;
while (walk) {
buf = GST_BUFFER (walk->data);
walk = g_slist_next (walk);
bs_print ("STATUS: buffer starts at %d and is %d bytes long\n", GST_BUFFER_OFFSET (buf), GST_BUFFER_SIZE (buf));
}
}
static gboolean
plugin_init (GModule *module, GstPlugin *plugin)
{
gst_plugin_set_longname (plugin, "GstByteStream: a byte-oriented layer on top of buffer-passing");
return TRUE;
}
GstPluginDesc plugin_desc = {
GST_VERSION_MAJOR,
GST_VERSION_MINOR,
"gstbytestream",
plugin_init
};