]> sigrok.org Git - libsigrok.git/blobdiff - src/input/vcd.c
kingst-la2016: move capture download details out of USB xfer handling
[libsigrok.git] / src / input / vcd.c
index 2cb4eeed66518f128e19b277f2a86060c56e31ab..acd30e5d7f05ae91420453b221f7b4176d111a69 100644 (file)
@@ -205,15 +205,6 @@ static void free_channel(void *data)
        g_free(vcd_ch);
 }
 
-/* TODO Drop the local decl when this has become a common helper. */
-void sr_channel_group_free(struct sr_channel_group *cg);
-
-/* Wrapper for GDestroyNotify compatibility. */
-static void cg_free(void *p)
-{
-       sr_channel_group_free(p);
-}
-
 /*
  * Another timestamp delta was observed, update statistics: Update the
  * sorted list of minimum values, and increment the occurance counter.
@@ -821,6 +812,7 @@ static int parse_header_var(struct context *inc, char *contents)
        size_t length;
        char *type, *size_txt, *id, *ref, *idx;
        gboolean is_reg, is_wire, is_real, is_int;
+       gboolean is_str;
        enum sr_channeltype ch_type;
        size_t size, next_size;
        struct vcd_channel *vcd_ch;
@@ -849,11 +841,19 @@ static int parse_header_var(struct context *inc, char *contents)
        is_wire = g_strcmp0(type, "wire") == 0;
        is_real = g_strcmp0(type, "real") == 0;
        is_int = g_strcmp0(type, "integer") == 0;
+       is_str = g_strcmp0(type, "string") == 0;
 
        if (is_reg || is_wire) {
                ch_type = SR_CHANNEL_LOGIC;
        } else if (is_real || is_int) {
                ch_type = SR_CHANNEL_ANALOG;
+       } else if (is_str) {
+               sr_warn("Skipping id %s, name '%s%s', unsupported type '%s'.",
+                       id, ref, idx ? idx : "", type);
+               inc->ignored_signals = g_slist_append(inc->ignored_signals,
+                       g_strdup(id));
+               g_strfreev(parts);
+               return SR_OK;
        } else {
                sr_err("Unsupported signal type: '%s'", type);
                g_strfreev(parts);
@@ -1065,10 +1065,8 @@ static void create_channels(const struct sr_input *in,
                if (vcd_ch->type != ch_type)
                        continue;
                cg = NULL;
-               if (vcd_ch->size != 1) {
-                       cg = g_malloc0(sizeof(*cg));
-                       cg->name = g_strdup(vcd_ch->name);
-               }
+               if (vcd_ch->size != 1)
+                       cg = sr_channel_group_new(sdi, vcd_ch->name, NULL);
                for (size_idx = 0; size_idx < vcd_ch->size; size_idx++) {
                        ch_name = get_channel_name(vcd_ch, size_idx);
                        sr_dbg("sigrok channel idx %zu, name %s, type %s, en %d.",
@@ -1080,8 +1078,6 @@ static void create_channels(const struct sr_input *in,
                        if (cg)
                                cg->channels = g_slist_append(cg->channels, ch);
                }
-               if (cg)
-                       sdi->channel_groups = g_slist_append(sdi->channel_groups, cg);
        }
 }
 
@@ -1126,7 +1122,7 @@ static void keep_header_for_reread(const struct sr_input *in)
 
        inc = in->priv;
 
-       g_slist_free_full(inc->prev.sr_groups, cg_free);
+       g_slist_free_full(inc->prev.sr_groups, sr_channel_group_free_cb);
        inc->prev.sr_groups = in->sdi->channel_groups;
        in->sdi->channel_groups = NULL;
 
@@ -1165,7 +1161,7 @@ static gboolean check_header_in_reread(const struct sr_input *in)
                return FALSE;
        }
 
-       g_slist_free_full(in->sdi->channel_groups, cg_free);
+       g_slist_free_full(in->sdi->channel_groups, sr_channel_group_free_cb);
        in->sdi->channel_groups = inc->prev.sr_groups;
        inc->prev.sr_groups = NULL;
 
@@ -1510,6 +1506,95 @@ static uint8_t vcd_char_to_value(char bit_char, int *warn)
        return ~0;
 }
 
+/*
+ * Check the validity of a VCD string value. It's essential to reliably
+ * accept valid data which the community uses in the field, yet robustly
+ * reject invalid data for users' awareness. Since IEEE 1800-2017 would
+ * not discuss the representation of this data type, it's assumed to not
+ * be an official feature of the VCD file format. This implementation is
+ * an educated guess after inspection of other arbitrary implementations,
+ * not backed by any specification or public documentation.
+ *
+ * A quick summary of the implemented assumptions: Must be a sequence of
+ * ASCII printables. Must not contain whitespace. Might contain escape
+ * sequences: A backslash followed by a single character, like '\n' or
+ * '\\'. Or a backslash and the letter x followed by two hex digits,
+ * like '\x20'. Or a backslash followed by three octal digits, like
+ * '\007'. As an exception also accepts a single digit '\0' but only at
+ * the text end. The string value may be empty, but must not be NULL.
+ *
+ * This implementation assumes an ASCII based platform for simplicity
+ * and readability. Should be a given on sigrok supported platforms.
+ */
+static gboolean vcd_string_valid(const char *s)
+{
+       char c;
+
+       if (!s)
+               return FALSE;
+
+       while (*s) {
+               c = *s++;
+               /* Reject non-printable ASCII chars including DEL. */
+               if (c < ' ')
+                       return FALSE;
+               if (c > '~')
+                       return FALSE;
+               /* Deeper inspection of escape sequences. */
+               if (c == '\\') {
+                       c = *s++;
+                       switch (c) {
+                       case 'a': /* BEL, bell aka "alarm" */
+                       case 'b': /* BS, back space */
+                       case 't': /* TAB, tabulator */
+                       case 'n': /* NL, newline */
+                       case 'v': /* VT, vertical tabulator */
+                       case 'f': /* FF, form feed */
+                       case 'r': /* CR, carriage return */
+                       case '"': /* double quotes */
+                       case '\'': /* tick, single quote */
+                       case '?': /* question mark */
+                       case '\\': /* backslash */
+                               continue;
+                       case 'x': /* \xNN two hex digits */
+                               c = *s++;
+                               if (!g_ascii_isxdigit(c))
+                                       return FALSE;
+                               c = *s++;
+                               if (!g_ascii_isxdigit(c))
+                                       return FALSE;
+                               continue;
+                       case '0': /* \NNN three octal digits */
+                       case '1':
+                       case '2':
+                       case '3':
+                       case '4':
+                       case '5':
+                       case '6':
+                       case '7':
+                               /* Special case '\0' at end of text. */
+                               if (c == '0' && !*s)
+                                       return TRUE;
+                               /*
+                                * First digit was covered by the outer
+                                * switch(). Two more digits to check.
+                                */
+                               c = *s++;
+                               if (!g_ascii_isdigit(c) || c > '7')
+                                       return FALSE;
+                               c = *s++;
+                               if (!g_ascii_isdigit(c) || c > '7')
+                                       return FALSE;
+                               continue;
+                       default:
+                               return FALSE;
+                       }
+               }
+       }
+
+       return TRUE;
+}
+
 /* Parse one text line of the data section. */
 static int parse_textline(const struct sr_input *in, char *lines)
 {
@@ -1518,7 +1603,8 @@ static int parse_textline(const struct sr_input *in, char *lines)
        char **words;
        size_t word_count, word_idx;
        char *curr_word, *next_word, curr_first;
-       gboolean is_timestamp, is_section, is_real, is_multibit, is_singlebit;
+       gboolean is_timestamp, is_section;
+       gboolean is_real, is_multibit, is_singlebit, is_string;
        uint64_t timestamp;
        char *identifier, *endptr;
        size_t count;
@@ -1697,6 +1783,7 @@ static int parse_textline(const struct sr_input *in, char *lines)
                 * timestamp.
                 *
                 * Supported input data formats are:
+                * - S<value> <sep> <id> (value not used, VCD type 'string').
                 * - R<value> <sep> <id> (analog channel, VCD type 'real').
                 * - B<value> <sep> <id> (analog channel, VCD type 'integer').
                 * - B<value> <sep> <id> (logic channels, VCD bit vectors).
@@ -1725,6 +1812,7 @@ static int parse_textline(const struct sr_input *in, char *lines)
                is_singlebit |= curr_first == 'l' || curr_first == 'h';
                is_singlebit |= curr_first == 'x' || curr_first == 'z';
                is_singlebit |= curr_first == 'u' || curr_first == '-';
+               is_string = curr_first == 's';
                if (is_real) {
                        char *real_text;
                        float real_val;
@@ -1872,6 +1960,32 @@ static int parse_textline(const struct sr_input *in, char *lines)
                        process_bits(inc, identifier, inc->conv_bits.value, 1);
                        continue;
                }
+               if (is_string) {
+                       const char *str_value;
+
+                       str_value = &curr_word[1];
+                       identifier = next_word;
+                       word_idx++;
+                       if (!vcd_string_valid(str_value)) {
+                               sr_err("Invalid string data: %s", str_value);
+                               ret = SR_ERR_DATA;
+                               break;
+                       }
+                       if (!identifier || !*identifier) {
+                               sr_err("String value without identifier.");
+                               ret = SR_ERR_DATA;
+                               break;
+                       }
+                       sr_spew("Got string data, id '%s', value \"%s\".",
+                               identifier, str_value);
+                       if (!is_ignored(inc, identifier)) {
+                               sr_err("String value for identifier '%s'.",
+                                       identifier);
+                               ret = SR_ERR_DATA;
+                               break;
+                       }
+                       continue;
+               }
 
                /* Design choice: Consider unsupported input fatal. */
                sr_err("Unknown token '%s'.", curr_word);