]> sigrok.org Git - libsigrok.git/blobdiff - src/output/wavedrom.c
kingst-la2016: coding style, separate declaration from instructions
[libsigrok.git] / src / output / wavedrom.c
index 47671729271e4f783281237682cbcacade83cc0d..65324bdc45f0a9c60d8c06456253ccdd756a9853 100644 (file)
 
 #define LOG_PREFIX "output/wavedrom"
 
-struct context
-{
+struct context {
        uint32_t channel_count;
-       // TODO: remove this, channels are available with each call
        struct sr_channel **channels;
-
-       // output strings
-       GString **channel_outputs;
+       GString **channel_outputs; /* output strings */
 };
 
-// takes all data collected in context and
-// renders it all to a JSON string.
+/* Converts accumulated output data to a JSON string. */
 static GString *wavedrom_render(const struct context *ctx)
 {
-       // open
-       GString *output = g_string_new("{ \"signal\": [");
-       uint32_t ch, i;
-       char lastChar, currentChar;
+       GString *output;
+       size_t ch, i;
+       char last_char, curr_char;
 
+       output = g_string_new("{ \"signal\": [");
        for (ch = 0; ch < ctx->channel_count; ch++) {
                if (!ctx->channel_outputs[ch])
                        continue;
 
-               // channel strip
+               /* Channel strip. */
                g_string_append_printf(output,
                        "{ \"name\": \"%s\", \"wave\": \"", ctx->channels[ch]->name);
 
-               lastChar = 0;
+               last_char = 0;
                for (i = 0; i < ctx->channel_outputs[ch]->len; i++) {
-                       currentChar = ctx->channel_outputs[ch]->str[i];
-                       // data point
-                       if (currentChar == lastChar) {
+                       curr_char = ctx->channel_outputs[ch]->str[i];
+                       /* Data point. */
+                       if (curr_char == last_char) {
                                g_string_append_c(output, '.');
                        } else {
-                               g_string_append_c(output, currentChar);
-                               lastChar = currentChar;
+                               g_string_append_c(output, curr_char);
+                               last_char = curr_char;
                        }
                }
                if (ch < ctx->channel_count - 1) {
                        g_string_append(output, "\" },");
                } else {
-                       // last channel - no comma
+                       /* Last channel, no comma. */
                        g_string_append(output, "\" }");
                }
        }
-
-       // close
        g_string_append(output, "], \"config\": { \"skin\": \"narrow\" }}");
+
        return output;
 }
 
+static void process_logic(const struct context *ctx,
+       const struct sr_datafeed_logic *logic)
+{
+       size_t sample_count, ch, i;
+       uint8_t *sample, bit;
+       GString *accu;
+
+       if (!ctx->channel_count)
+               return;
+
+       /*
+        * Extract the logic bits for each channel and store them
+        * as wavedrom letters (1/0) in each channel's text string.
+        * This transforms the input which consists of sample sets
+        * that span multiple channels into output stripes per logic
+        * channel which consist of bits for that individual channel.
+        *
+        * TODO Reduce memory consumption during accumulation of
+        * output data.
+        *
+        * Ideally we'd accumulate binary chunks, and defer conversion
+        * to the text format. Analog data already won't get here, only
+        * logic data does. When the per-channel transformation also
+        * gets deferred until later, then the only overhead would be
+        * for disabled logic channels. Which may be acceptable or even
+        * negligable.
+        *
+        * An optional addition to the above mentioned accumulation of
+        * binary data is RLE compression. Mark both the position in the
+        * accumulated data as well as a repetition counter, instead of
+        * repeatedly storing the same sample set. The efficiency of
+        * this approach of course depends on the change rate of input
+        * data. But the approach perfectly matches the WaveDrom syntax
+        * for repeated bit patterns, and thus is easily handled in the
+        * text rendering stage of the output module.
+        */
+       sample_count = logic->length / logic->unitsize;
+       for (i = 0; i < sample_count; i++) {
+               sample = logic->data + i * logic->unitsize;
+               for (ch = 0; ch < ctx->channel_count; ch++) {
+                       accu = ctx->channel_outputs[ch];
+                       if (!accu)
+                               continue;
+                       bit = sample[ch / 8] & (1 << (ch % 8));
+                       g_string_append_c(accu, bit ? '1' : '0');
+               }
+       }
+}
+
+static int receive(const struct sr_output *o,
+       const struct sr_datafeed_packet *packet, GString **out)
+{
+       struct context *ctx;
+
+       *out = NULL;
+
+       if (!o || !o->sdi || !o->priv)
+               return SR_ERR_ARG;
+
+       ctx = o->priv;
+
+       switch (packet->type) {
+       case SR_DF_LOGIC:
+               process_logic(ctx, packet->payload);
+               break;
+       case SR_DF_END:
+               *out = wavedrom_render(ctx);
+               break;
+       }
+
+       return SR_OK;
+}
+
 static int init(struct sr_output *o, GHashTable *options)
 {
        struct context *ctx;
        struct sr_channel *channel;
        GSList *l;
-       uint32_t i;
+       size_t i;
 
        (void)options;
 
        if (!o || !o->sdi)
                return SR_ERR_ARG;
 
-       o->priv = ctx = g_malloc(sizeof(struct context));
+       o->priv = ctx = g_malloc0(sizeof(*ctx));
 
        ctx->channel_count = g_slist_length(o->sdi->channels);
-       ctx->channels = g_malloc(
-               sizeof(struct sr_channel) * ctx->channel_count);
-       ctx->channel_outputs = g_malloc(
-               sizeof(GString *) * ctx->channel_count);
+       ctx->channels = g_malloc0(
+               sizeof(ctx->channels[0]) * ctx->channel_count);
+       ctx->channel_outputs = g_malloc0(
+               sizeof(ctx->channel_outputs[0]) * ctx->channel_count);
 
        for (i = 0, l = o->sdi->channels; l; l = l->next, i++) {
                channel = l->data;
-               if (channel->enabled &&
-                       channel->type == SR_CHANNEL_LOGIC) {
+               if (channel->enabled && channel->type == SR_CHANNEL_LOGIC) {
                        ctx->channels[i] = channel;
                        ctx->channel_outputs[i] = g_string_new(NULL);
-               } else {
-                       ctx->channels[i] = NULL;
-                       ctx->channel_outputs[i] = NULL;
                }
        }
 
@@ -113,7 +176,7 @@ static int init(struct sr_output *o, GHashTable *options)
 static int cleanup(struct sr_output *o)
 {
        struct context *ctx;
-       uint32_t i;
+       GString *s;
 
        if (!o)
                return SR_ERR_ARG;
@@ -122,9 +185,10 @@ static int cleanup(struct sr_output *o)
        o->priv = NULL;
 
        if (ctx) {
-               for (i = 0; i < ctx->channel_count; i++) {
-                       if (ctx->channel_outputs[i])
-                               g_string_free(ctx->channel_outputs[i], TRUE);
+               while (--ctx->channel_count) {
+                       s = ctx->channel_outputs[ctx->channel_count];
+                       if (s)
+                               g_string_free(s, TRUE);
                }
                g_free(ctx->channel_outputs);
                g_free(ctx->channels);
@@ -134,57 +198,9 @@ static int cleanup(struct sr_output *o)
        return SR_OK;
 }
 
-static void process_logic(const struct context *ctx,
-                         const struct sr_datafeed_logic *logic)
-{
-       unsigned int sample_count, ch, i;
-       uint8_t *sample;
-
-       sample_count = logic->length / logic->unitsize;
-
-       // extract the logic bits for each channel and
-       // store them as wavedrom letters (1/0) in each channel's string
-       for (ch = 0; ch < ctx->channel_count; ch++) {
-               if (ctx->channels[ch]) {
-                       for (i = 0; i < sample_count; i++) {
-                               sample = logic->data + i * logic->unitsize;
-
-                               if (ctx->channel_outputs[ch]) {
-                                       g_string_append_c(ctx->channel_outputs[ch],
-                                                                         sample[ch / 8] & (1 << (ch % 8)) ? '1' : '0');
-                               }
-                       }
-               }
-       }
-}
-
-static int receive(const struct sr_output *o,
-                  const struct sr_datafeed_packet *packet, GString **out)
-{
-       struct context *ctx;
-
-       *out = NULL;
-
-       if (!o || !o->sdi || !o->priv)
-               return SR_ERR_ARG;
-
-       ctx = o->priv;
-
-       switch (packet->type) {
-       case SR_DF_LOGIC:
-               process_logic(ctx, packet->payload);
-               break;
-       case SR_DF_END:
-               *out = wavedrom_render(ctx);
-               break;
-       }
-
-       return SR_OK;
-}
-
 SR_PRIV struct sr_output_module output_wavedrom = {
        .id = "wavedrom",
-       .name = "WAVEDROM",
+       .name = "WaveDrom",
        .desc = "WaveDrom.com file format",
        .exts = (const char *[]){"wavedrom", "json", NULL},
        .flags = 0,