X-Git-Url: http://sigrok.org/gitweb/?a=blobdiff_plain;f=src%2Finput%2Fcsv.c;h=1d230fdcad0e2c3777f329943fc1ddc219a6de0f;hb=de788af4108493211d2c62004bf3d2fd81ad7d67;hp=0a2d7276d996eb14b5f0310999feb1ed2e3c8c06;hpb=6433156c3275df933e4bf6dcfb020c91fca0ae86;p=libsigrok.git
diff --git a/src/input/csv.c b/src/input/csv.c
index 0a2d7276..1d230fdc 100644
--- a/src/input/csv.c
+++ b/src/input/csv.c
@@ -17,6 +17,7 @@
* along with this program. If not, see .
*/
+#include
#include
#include
#include
@@ -66,6 +67,39 @@
* than 0. The default line number to start processing is 1.
*/
+/*
+ * TODO
+ *
+ * - Determine how the text line handling can get improved, regarding
+ * all of robustness and flexibility and correctness.
+ * - The current implementation splits on "any run of CR and LF". Which
+ * translates to: Line numbers are wrong in the presence of empty
+ * lines in the input stream. See below for an (expensive) fix.
+ * - Dropping support for CR style end-of-line markers could improve
+ * the situation a lot. Code could search for and split on LF, and
+ * trim optional trailing CR. This would result in proper support
+ * for CRLF (Windows) as well as LF (Unix), and allow for correct
+ * line number counts.
+ * - When support for CR-only line termination cannot get dropped,
+ * then the current implementation is inappropriate. Currently the
+ * input stream is scanned for the first occurance of either of the
+ * supported termination styles (which is good). For the remaining
+ * session a consistent encoding of the text lines is assumed (which
+ * is acceptable).
+ * - When line numbers need to be correct and reliable, _and_ the full
+ * set of previously supported line termination sequences are required,
+ * and potentially more are to get added for improved compatibility
+ * with more platforms or generators, then the current approach of
+ * splitting on runs of termination characters needs to get replaced,
+ * by the more expensive approach to scan for and count the initially
+ * determined termination sequence.
+ *
+ * - Add support for analog input data? (optional)
+ * - Needs a syntax first for user specs which channels (columns) are
+ * logic and which are analog. May need heuristics(?) to guess from
+ * input data in the absence of user provided specs.
+ */
+
/* Single column formats. */
enum {
FORMAT_BIN,
@@ -88,7 +122,7 @@ struct context {
/* Comment prefix character(s). */
GString *comment;
- /* Termination character(s) used in current stream. */
+ /* Termination character(s) used in current stream. */
char *termination;
/* Determines if sample data is stored in multiple columns. */
@@ -132,17 +166,6 @@ struct context {
size_t line_number;
};
-static int format_match(GHashTable *metadata)
-{
- char *buf;
-
- buf = g_hash_table_lookup(metadata, GINT_TO_POINTER(SR_INPUT_META_MIMETYPE));
- if (!strcmp(buf, "text/csv"))
- return SR_OK;
-
- return SR_ERR;
-}
-
static void strip_comment(char *buf, const GString *prefix)
{
char *ptr;
@@ -323,20 +346,22 @@ static char **parse_line(char *buf, struct context *inc, int max_columns)
static int parse_multi_columns(char **columns, struct context *inc)
{
gsize i;
+ char *column;
/* Clear buffer in order to set bits only. */
memset(inc->sample_buffer, 0, (inc->num_channels + 7) >> 3);
for (i = 0; i < inc->num_channels; i++) {
- if (columns[i][0] == '1') {
+ column = columns[i];
+ if (column[0] == '1') {
inc->sample_buffer[i / 8] |= (1 << (i % 8));
- } else if (!strlen(columns[i])) {
+ } else if (!strlen(column)) {
sr_err("Column %zu in line %zu is empty.",
inc->first_channel + i, inc->line_number);
return SR_ERR;
- } else if (columns[i][0] != '0') {
+ } else if (column[0] != '0') {
sr_err("Invalid value '%s' in column %zu in line %zu.",
- columns[i], inc->first_channel + i,
+ column, inc->first_channel + i,
inc->line_number);
return SR_ERR;
}
@@ -381,7 +406,8 @@ static int send_samples(const struct sr_dev_inst *sdi, uint8_t *buffer,
logic.data = buffer;
for (i = 0; i < count; i++) {
- if ((res = sr_session_send(sdi, &packet)) != SR_OK)
+ res = sr_session_send(sdi, &packet);
+ if (res != SR_OK)
return res;
}
@@ -454,6 +480,8 @@ static int init(struct sr_input *in, GHashTable *options)
return SR_OK;
}
+static const char *delim_set = "\r\n";
+
static const char *get_line_termination(GString *buf)
{
const char *term;
@@ -476,26 +504,27 @@ static int initial_parse(const struct sr_input *in, GString *buf)
unsigned int num_columns, i;
size_t line_number, l;
int ret;
- char **lines, **columns;
+ char **lines, *line, **columns, *column;
ret = SR_OK;
inc = in->priv;
columns = NULL;
line_number = 0;
- lines = g_strsplit_set(buf->str, "\r\n", 0);
+ lines = g_strsplit_set(buf->str, delim_set, 0);
for (l = 0; lines[l]; l++) {
line_number++;
+ line = lines[l];
if (inc->start_line > line_number) {
sr_spew("Line %zu skipped.", line_number);
continue;
}
- if (lines[l][0] == '\0') {
+ if (line[0] == '\0') {
sr_spew("Blank line %zu skipped.", line_number);
continue;
}
- strip_comment(lines[l], inc->comment);
- if (lines[l][0] == '\0') {
+ strip_comment(line, inc->comment);
+ if (line[0] == '\0') {
sr_spew("Comment-only line %zu skipped.", line_number);
continue;
}
@@ -513,7 +542,8 @@ static int initial_parse(const struct sr_input *in, GString *buf)
* In order to determine the number of columns parse the current line
* without limiting the number of columns.
*/
- if (!(columns = parse_line(lines[l], inc, -1))) {
+ columns = parse_line(line, inc, -1);
+ if (!columns) {
sr_err("Error while parsing line %zu.", line_number);
ret = SR_ERR;
goto out;
@@ -553,8 +583,9 @@ static int initial_parse(const struct sr_input *in, GString *buf)
channel_name = g_string_sized_new(64);
for (i = 0; i < inc->num_channels; i++) {
- if (inc->header && inc->multi_column_mode && columns[i][0] != '\0')
- g_string_assign(channel_name, columns[i]);
+ column = columns[i];
+ if (inc->header && inc->multi_column_mode && column[0] != '\0')
+ g_string_assign(channel_name, column);
else
g_string_printf(channel_name, "%u", i);
sr_channel_new(in->sdi, i, SR_CHANNEL_LOGIC, TRUE, channel_name->str);
@@ -576,6 +607,27 @@ out:
return ret;
}
+/*
+ * Gets called from initial_receive(), which runs until the end-of-line
+ * encoding of the input stream could get determined. Assumes that this
+ * routine receives enough buffered initial input data to either see the
+ * BOM when there is one, or that no BOM will follow when a text line
+ * termination sequence was seen. Silently drops the UTF-8 BOM sequence
+ * from the input buffer if one was seen. Does not care to protect
+ * against multiple execution or dropping the BOM multiple times --
+ * there should be at most one in the input stream.
+ */
+static void initial_bom_check(const struct sr_input *in)
+{
+ static const char *utf8_bom = "\xef\xbb\xbf";
+
+ if (in->buf->len < strlen(utf8_bom))
+ return;
+ if (strncmp(in->buf->str, utf8_bom, strlen(utf8_bom)) != 0)
+ return;
+ g_string_erase(in->buf, 0, strlen(utf8_bom));
+}
+
static int initial_receive(const struct sr_input *in)
{
struct context *inc;
@@ -584,13 +636,17 @@ static int initial_receive(const struct sr_input *in)
char *p;
const char *termination;
+ initial_bom_check(in);
+
inc = in->priv;
- if (!(termination = get_line_termination(in->buf)))
+ termination = get_line_termination(in->buf);
+ if (!termination)
/* Don't have a full line yet. */
return SR_ERR_NA;
- if (!(p = g_strrstr_len(in->buf->str, in->buf->len, termination)))
+ p = g_strrstr_len(in->buf->str, in->buf->len, termination);
+ if (!p)
/* Don't have a full line yet. */
return SR_ERR_NA;
len = p - in->buf->str - 1;
@@ -609,7 +665,7 @@ static int initial_receive(const struct sr_input *in)
return ret;
}
-static int process_buffer(struct sr_input *in)
+static int process_buffer(struct sr_input *in, gboolean is_eof)
{
struct sr_datafeed_packet packet;
struct sr_datafeed_meta meta;
@@ -618,11 +674,11 @@ static int process_buffer(struct sr_input *in)
gsize num_columns;
uint64_t samplerate;
int max_columns, ret, l;
- char *p, **lines, **columns;
+ char *p, **lines, *line, **columns;
inc = in->priv;
if (!inc->started) {
- std_session_send_df_header(in->sdi, LOG_PREFIX);
+ std_session_send_df_header(in->sdi);
if (inc->samplerate) {
packet.type = SR_DF_META;
@@ -631,37 +687,58 @@ static int process_buffer(struct sr_input *in)
src = sr_config_new(SR_CONF_SAMPLERATE, g_variant_new_uint64(samplerate));
meta.config = g_slist_append(NULL, src);
sr_session_send(in->sdi, &packet);
+ g_slist_free(meta.config);
sr_config_free(src);
}
inc->started = TRUE;
}
- if (!(p = g_strrstr_len(in->buf->str, in->buf->len, inc->termination)))
- /* Don't have a full line. */
- return SR_ERR;
-
- *p = '\0';
- g_strstrip(in->buf->str);
-
/* Limit the number of columns to parse. */
if (inc->multi_column_mode)
max_columns = inc->num_channels;
else
max_columns = 1;
+ /*
+ * Consider empty input non-fatal. Keep accumulating input until
+ * at least one full text line has become available. Grab the
+ * maximum amount of accumulated data that consists of full text
+ * lines, and process what has been received so far, leaving not
+ * yet complete lines for the next invocation.
+ *
+ * Enforce that all previously buffered data gets processed in
+ * the "EOF" condition. Do not insist in the presence of the
+ * termination sequence for the last line (may often be missing
+ * on Windows). A present termination sequence will just result
+ * in the "execution of an empty line", and does not harm.
+ */
+ if (!in->buf->len)
+ return SR_OK;
+ if (is_eof) {
+ p = in->buf->str + in->buf->len;
+ } else {
+ p = g_strrstr_len(in->buf->str, in->buf->len, inc->termination);
+ if (!p)
+ return SR_ERR;
+ *p = '\0';
+ p += strlen(inc->termination);
+ }
+ g_strstrip(in->buf->str);
+
ret = SR_OK;
- lines = g_strsplit_set(in->buf->str, "\r\n", 0);
+ lines = g_strsplit_set(in->buf->str, delim_set, 0);
for (l = 0; lines[l]; l++) {
inc->line_number++;
- if (lines[l][0] == '\0') {
+ line = lines[l];
+ if (line[0] == '\0') {
sr_spew("Blank line %zu skipped.", inc->line_number);
continue;
}
/* Remove trailing comment. */
- strip_comment(lines[l], inc->comment);
- if (lines[l][0] == '\0') {
+ strip_comment(line, inc->comment);
+ if (line[0] == '\0') {
sr_spew("Comment-only line %zu skipped.", inc->line_number);
continue;
}
@@ -673,7 +750,8 @@ static int process_buffer(struct sr_input *in)
continue;
}
- if (!(columns = parse_line(lines[l], inc, max_columns))) {
+ columns = parse_line(line, inc, max_columns);
+ if (!columns) {
sr_err("Error while parsing line %zu.", inc->line_number);
return SR_ERR;
}
@@ -714,7 +792,7 @@ static int process_buffer(struct sr_input *in)
g_strfreev(columns);
}
g_strfreev(lines);
- g_string_erase(in->buf, 0, p - in->buf->str + 1);
+ g_string_erase(in->buf, 0, p - in->buf->str);
return ret;
}
@@ -728,7 +806,8 @@ static int receive(struct sr_input *in, GString *buf)
inc = in->priv;
if (!inc->termination) {
- if ((ret = initial_receive(in)) == SR_ERR_NA)
+ ret = initial_receive(in);
+ if (ret == SR_ERR_NA)
/* Not enough data yet. */
return SR_OK;
else if (ret != SR_OK)
@@ -739,7 +818,7 @@ static int receive(struct sr_input *in, GString *buf)
return SR_OK;
}
- ret = process_buffer(in);
+ ret = process_buffer(in, FALSE);
return ret;
}
@@ -747,20 +826,16 @@ static int receive(struct sr_input *in, GString *buf)
static int end(struct sr_input *in)
{
struct context *inc;
- struct sr_datafeed_packet packet;
int ret;
if (in->sdi_ready)
- ret = process_buffer(in);
+ ret = process_buffer(in, TRUE);
else
ret = SR_OK;
inc = in->priv;
- if (inc->started) {
- /* End of stream. */
- packet.type = SR_DF_END;
- sr_session_send(in->sdi, &packet);
- }
+ if (inc->started)
+ std_session_send_df_end(in->sdi);
return ret;
}
@@ -781,6 +856,17 @@ static void cleanup(struct sr_input *in)
g_free(inc->sample_buffer);
}
+static int reset(struct sr_input *in)
+{
+ struct context *inc = in->priv;
+
+ cleanup(in);
+ inc->started = FALSE;
+ g_string_truncate(in->buf, 0);
+
+ return SR_OK;
+}
+
static struct sr_option options[] = {
{ "single-column", "Single column", "Enable/specify single column", NULL, NULL },
{ "numchannels", "Max channels", "Number of channels", NULL, NULL },
@@ -794,7 +880,7 @@ static struct sr_option options[] = {
ALL_ZERO
};
-static struct sr_option *get_options(void)
+static const struct sr_option *get_options(void)
{
if (!options[0].def) {
options[0].def = g_variant_ref_sink(g_variant_new_int32(0));
@@ -816,11 +902,10 @@ SR_PRIV struct sr_input_module input_csv = {
.name = "CSV",
.desc = "Comma-separated values",
.exts = (const char*[]){"csv", NULL},
- .metadata = { SR_INPUT_META_MIMETYPE },
.options = get_options,
- .format_match = format_match,
.init = init,
.receive = receive,
.end = end,
.cleanup = cleanup,
+ .reset = reset,
};