X-Git-Url: https://sigrok.org/gitweb/?a=blobdiff_plain;f=src%2Finput%2Fcsv.c;h=fdfc26419ae7e44e910549dafc4f8067cc985d40;hb=685ed709982f5edd84f7666d0a8e03c2124dff92;hp=c0316b62565f4bd84ac9966ca0bf9287ebd7c0a4;hpb=d4c937749a92ce6defa2f0095b34692181afe597;p=libsigrok.git
diff --git a/src/input/csv.c b/src/input/csv.c
index c0316b62..fdfc2641 100644
--- a/src/input/csv.c
+++ b/src/input/csv.c
@@ -17,14 +17,17 @@
* along with this program. If not, see .
*/
+#include
#include
#include
#include
-#include "libsigrok.h"
+#include
#include "libsigrok-internal.h"
#define LOG_PREFIX "input/csv"
+#define CHUNK_SIZE (4 * 1024 * 1024)
+
/*
* The CSV input module has the following options:
*
@@ -66,6 +69,39 @@
* 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,
@@ -74,11 +110,13 @@ enum {
};
struct context {
+ gboolean started;
+
/* Current selected samplerate. */
uint64_t samplerate;
/* Number of channels. */
- gsize num_channels;
+ unsigned int num_channels;
/* Column delimiter character(s). */
GString *delimiter;
@@ -86,27 +124,30 @@ struct context {
/* Comment prefix character(s). */
GString *comment;
+ /* Termination character(s) used in current stream. */
+ char *termination;
+
/* Determines if sample data is stored in multiple columns. */
gboolean multi_column_mode;
/* Column number of the sample data in single column mode. */
- gsize single_column;
+ unsigned int single_column;
/*
* Number of the first column to parse. Equivalent to the number of the
* first channel in multi column mode and the single column number in
* single column mode.
*/
- gsize first_column;
+ unsigned int first_column;
/*
* Column number of the first channel in multi column mode and position of
* the bit for the first channel in single column mode.
*/
- gsize first_channel;
+ unsigned int first_channel;
/* Line number to start processing. */
- gsize start_line;
+ size_t start_line;
/*
* Determines if the first line should be treated as header and used for
@@ -117,106 +158,51 @@ struct context {
/* Format sample data is stored in single column mode. */
int format;
- /* Size of the sample buffer. */
- gsize sample_buffer_size;
-
- /* Buffer to store sample data. */
- uint8_t *sample_buffer;
-
- GIOChannel *channel;
+ size_t sample_unit_size; /**!< Byte count for a single sample. */
+ uint8_t *sample_buffer; /**!< Buffer for a single sample. */
- /* Buffer for the current line. */
- GString *buffer;
+ uint8_t *datafeed_buffer; /**!< Queue for datafeed submission. */
+ size_t datafeed_buf_size;
+ size_t datafeed_buf_fill;
/* Current line number. */
- gsize line_number;
+ size_t line_number;
};
-static int format_match(const char *filename)
-{
- /* Require .csv extension. */
- if (strcmp(filename + strlen(filename) - 4, ".csv"))
- return FALSE;
-
- if (!filename) {
- sr_err("%s: filename was NULL.", __func__);
- return FALSE;
- }
-
- if (!g_file_test(filename, G_FILE_TEST_EXISTS)) {
- sr_err("Input file '%s' does not exist.", filename);
- return FALSE;
- }
-
- if (!g_file_test(filename, G_FILE_TEST_IS_REGULAR)) {
- sr_err("Input file '%s' not a regular file.", filename);
- return FALSE;
- }
-
- return TRUE;
-}
-
-static void free_context(struct context *ctx)
-{
- if (!ctx)
- return;
-
- if (ctx->delimiter)
- g_string_free(ctx->delimiter, TRUE);
-
- if (ctx->comment)
- g_string_free(ctx->comment, TRUE);
-
- if (ctx->channel) {
- g_io_channel_shutdown(ctx->channel, FALSE, NULL);
- g_io_channel_unref(ctx->channel);
- }
-
- if (ctx->sample_buffer)
- g_free(ctx->sample_buffer);
-
- if (ctx->buffer)
- g_string_free(ctx->buffer, TRUE);
-
- g_free(ctx);
-}
-
-static void strip_comment(GString *string, const GString *prefix)
+static void strip_comment(char *buf, const GString *prefix)
{
char *ptr;
if (!prefix->len)
return;
- if (!(ptr = strstr(string->str, prefix->str)))
- return;
-
- g_string_truncate(string, ptr - string->str);
+ if ((ptr = strstr(buf, prefix->str)))
+ *ptr = '\0';
}
-static int parse_binstr(const char *str, struct context *ctx)
+static int parse_binstr(const char *str, struct context *inc)
{
gsize i, j, length;
length = strlen(str);
if (!length) {
- sr_err("Column %zu in line %zu is empty.", ctx->single_column,
- ctx->line_number);
+ sr_err("Column %u in line %zu is empty.", inc->single_column,
+ inc->line_number);
return SR_ERR;
}
/* Clear buffer in order to set bits only. */
- memset(ctx->sample_buffer, 0, (ctx->num_channels + 7) >> 3);
+ memset(inc->sample_buffer, 0, inc->sample_unit_size);
- i = ctx->first_channel;
+ i = inc->first_channel;
- for (j = 0; i < length && j < ctx->num_channels; i++, j++) {
+ for (j = 0; i < length && j < inc->num_channels; i++, j++) {
if (str[length - i - 1] == '1') {
- ctx->sample_buffer[j / 8] |= (1 << (j % 8));
+ inc->sample_buffer[j / 8] |= (1 << (j % 8));
} else if (str[length - i - 1] != '0') {
- sr_err("Invalid value '%s' in column %zu in line %zu.",
- str, ctx->single_column, ctx->line_number);
+ sr_err("Invalid value '%s' in column %u in line %zu.",
+ str, inc->single_column, inc->line_number);
return SR_ERR;
}
}
@@ -224,7 +210,7 @@ static int parse_binstr(const char *str, struct context *ctx)
return SR_OK;
}
-static int parse_hexstr(const char *str, struct context *ctx)
+static int parse_hexstr(const char *str, struct context *inc)
{
gsize i, j, k, length;
uint8_t value;
@@ -233,33 +219,33 @@ static int parse_hexstr(const char *str, struct context *ctx)
length = strlen(str);
if (!length) {
- sr_err("Column %zu in line %zu is empty.", ctx->single_column,
- ctx->line_number);
+ sr_err("Column %u in line %zu is empty.", inc->single_column,
+ inc->line_number);
return SR_ERR;
}
/* Clear buffer in order to set bits only. */
- memset(ctx->sample_buffer, 0, (ctx->num_channels + 7) >> 3);
+ memset(inc->sample_buffer, 0, inc->sample_unit_size);
/* Calculate the position of the first hexadecimal digit. */
- i = ctx->first_channel / 4;
+ i = inc->first_channel / 4;
- for (j = 0; i < length && j < ctx->num_channels; i++) {
+ for (j = 0; i < length && j < inc->num_channels; i++) {
c = str[length - i - 1];
if (!g_ascii_isxdigit(c)) {
- sr_err("Invalid value '%s' in column %zu in line %zu.",
- str, ctx->single_column, ctx->line_number);
+ sr_err("Invalid value '%s' in column %u in line %zu.",
+ str, inc->single_column, inc->line_number);
return SR_ERR;
}
value = g_ascii_xdigit_value(c);
- k = (ctx->first_channel + j) % 4;
+ k = (inc->first_channel + j) % 4;
- for (; j < ctx->num_channels && k < 4; k++) {
+ for (; j < inc->num_channels && k < 4; k++) {
if (value & (1 << k))
- ctx->sample_buffer[j / 8] |= (1 << (j % 8));
+ inc->sample_buffer[j / 8] |= (1 << (j % 8));
j++;
}
@@ -268,7 +254,7 @@ static int parse_hexstr(const char *str, struct context *ctx)
return SR_OK;
}
-static int parse_octstr(const char *str, struct context *ctx)
+static int parse_octstr(const char *str, struct context *inc)
{
gsize i, j, k, length;
uint8_t value;
@@ -277,33 +263,33 @@ static int parse_octstr(const char *str, struct context *ctx)
length = strlen(str);
if (!length) {
- sr_err("Column %zu in line %zu is empty.", ctx->single_column,
- ctx->line_number);
+ sr_err("Column %u in line %zu is empty.", inc->single_column,
+ inc->line_number);
return SR_ERR;
}
/* Clear buffer in order to set bits only. */
- memset(ctx->sample_buffer, 0, (ctx->num_channels + 7) >> 3);
+ memset(inc->sample_buffer, 0, inc->sample_unit_size);
/* Calculate the position of the first octal digit. */
- i = ctx->first_channel / 3;
+ i = inc->first_channel / 3;
- for (j = 0; i < length && j < ctx->num_channels; i++) {
+ for (j = 0; i < length && j < inc->num_channels; i++) {
c = str[length - i - 1];
if (c < '0' || c > '7') {
- sr_err("Invalid value '%s' in column %zu in line %zu.",
- str, ctx->single_column, ctx->line_number);
+ sr_err("Invalid value '%s' in column %u in line %zu.",
+ str, inc->single_column, inc->line_number);
return SR_ERR;
}
value = g_ascii_xdigit_value(c);
- k = (ctx->first_channel + j) % 3;
+ k = (inc->first_channel + j) % 3;
- for (; j < ctx->num_channels && k < 3; k++) {
+ for (; j < inc->num_channels && k < 3; k++) {
if (value & (1 << k))
- ctx->sample_buffer[j / 8] |= (1 << (j % 8));
+ inc->sample_buffer[j / 8] |= (1 << (j % 8));
j++;
}
@@ -312,7 +298,7 @@ static int parse_octstr(const char *str, struct context *ctx)
return SR_OK;
}
-static char **parse_line(const struct context *ctx, int max_columns)
+static char **parse_line(char *buf, struct context *inc, int max_columns)
{
const char *str, *remainder;
GSList *list, *l;
@@ -324,11 +310,11 @@ static char **parse_line(const struct context *ctx, int max_columns)
k = 0;
list = NULL;
- remainder = ctx->buffer->str;
- str = strstr(remainder, ctx->delimiter->str);
+ remainder = buf;
+ str = strstr(remainder, inc->delimiter->str);
while (str && max_columns) {
- if (n >= ctx->first_column) {
+ if (n >= inc->first_column) {
column = g_strndup(remainder, str - remainder);
list = g_slist_prepend(list, g_strstrip(column));
@@ -336,12 +322,12 @@ static char **parse_line(const struct context *ctx, int max_columns)
k++;
}
- remainder = str + ctx->delimiter->len;
- str = strstr(remainder, ctx->delimiter->str);
+ remainder = str + inc->delimiter->len;
+ str = strstr(remainder, inc->delimiter->str);
n++;
}
- if (ctx->buffer->len && max_columns && n >= ctx->first_column) {
+ if (buf[0] && max_columns && n >= inc->first_column) {
column = g_strdup(remainder);
list = g_slist_prepend(list, g_strstrip(column));
k++;
@@ -360,24 +346,26 @@ static char **parse_line(const struct context *ctx, int max_columns)
return columns;
}
-static int parse_multi_columns(char **columns, struct context *ctx)
+static int parse_multi_columns(char **columns, struct context *inc)
{
gsize i;
+ char *column;
/* Clear buffer in order to set bits only. */
- memset(ctx->sample_buffer, 0, (ctx->num_channels + 7) >> 3);
+ memset(inc->sample_buffer, 0, inc->sample_unit_size);
- for (i = 0; i < ctx->num_channels; i++) {
- if (columns[i][0] == '1') {
- ctx->sample_buffer[i / 8] |= (1 << (i % 8));
- } else if (!strlen(columns[i])) {
+ for (i = 0; i < inc->num_channels; i++) {
+ column = columns[i];
+ if (column[0] == '1') {
+ inc->sample_buffer[i / 8] |= (1 << (i % 8));
+ } else if (!strlen(column)) {
sr_err("Column %zu in line %zu is empty.",
- ctx->first_channel + i, ctx->line_number);
+ 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], ctx->first_channel + i,
- ctx->line_number);
+ column, inc->first_channel + i,
+ inc->line_number);
return SR_ERR;
}
}
@@ -385,488 +373,589 @@ static int parse_multi_columns(char **columns, struct context *ctx)
return SR_OK;
}
-static int parse_single_column(const char *column, struct context *ctx)
+static int parse_single_column(const char *column, struct context *inc)
{
int res;
res = SR_ERR;
- switch(ctx->format) {
+ switch (inc->format) {
case FORMAT_BIN:
- res = parse_binstr(column, ctx);
+ res = parse_binstr(column, inc);
break;
case FORMAT_HEX:
- res = parse_hexstr(column, ctx);
+ res = parse_hexstr(column, inc);
break;
case FORMAT_OCT:
- res = parse_octstr(column, ctx);
+ res = parse_octstr(column, inc);
break;
}
return res;
}
-static int send_samples(const struct sr_dev_inst *sdi, uint8_t *buffer,
- gsize buffer_size, gsize count)
+static int flush_samples(const struct sr_input *in)
{
- int res;
+ struct context *inc;
struct sr_datafeed_packet packet;
struct sr_datafeed_logic logic;
- gsize i;
+ int rc;
+
+ inc = in->priv;
+ if (!inc->datafeed_buf_fill)
+ return SR_OK;
+ memset(&packet, 0, sizeof(packet));
+ memset(&logic, 0, sizeof(logic));
packet.type = SR_DF_LOGIC;
packet.payload = &logic;
- logic.unitsize = buffer_size;
- logic.length = buffer_size;
- logic.data = buffer;
+ logic.unitsize = inc->sample_unit_size;
+ logic.length = inc->datafeed_buf_fill;
+ logic.data = inc->datafeed_buffer;
- for (i = 0; i < count; i++) {
- if ((res = sr_session_send(sdi, &packet)) != SR_OK)
- return res;
- }
+ rc = sr_session_send(in->sdi, &packet);
+ if (rc != SR_OK)
+ return rc;
+ inc->datafeed_buf_fill = 0;
return SR_OK;
}
-static int init(struct sr_input *in, const char *filename)
+static int queue_samples(const struct sr_input *in)
{
- int res;
- struct context *ctx;
- const char *param;
- GIOStatus status;
- gsize i, term_pos;
- char channel_name[SR_MAX_CHANNELNAME_LEN + 1];
- struct sr_channel *ch;
- char **columns;
- gsize num_columns;
- char *ptr;
-
- if (!(ctx = g_try_malloc0(sizeof(struct context)))) {
- sr_err("Context malloc failed.");
- return SR_ERR_MALLOC;
- }
-
- /* Create a virtual device. */
- in->sdi = sr_dev_inst_new(0, SR_ST_ACTIVE, NULL, NULL, NULL);
- in->internal = ctx;
+ struct context *inc;
+ int rc;
- /* Set default samplerate. */
- ctx->samplerate = 0;
+ inc = in->priv;
- /*
- * Enable auto-detection of the number of channels in multi column mode
- * and enforce the specification of the number of channels in single
- * column mode.
- */
- ctx->num_channels = 0;
-
- /* Set default delimiter. */
- if (!(ctx->delimiter = g_string_new(","))) {
- sr_err("Delimiter malloc failed.");
- free_context(ctx);
- return SR_ERR_MALLOC;
- }
-
- /*
- * Set default comment prefix. Note that an empty comment prefix
- * disables removing of comments.
- */
- if (!(ctx->comment = g_string_new(""))) {
- sr_err("Comment malloc failed.");
- free_context(ctx);
- return SR_ERR_MALLOC;
+ inc->datafeed_buf_fill += inc->sample_unit_size;
+ if (inc->datafeed_buf_fill == inc->datafeed_buf_size) {
+ rc = flush_samples(in);
+ if (rc != SR_OK)
+ return rc;
}
+ inc->sample_buffer = &inc->datafeed_buffer[inc->datafeed_buf_fill];
+ return SR_OK;
+}
- /* Enable multi column mode by default. */
- ctx->multi_column_mode = TRUE;
-
- /* Use first column as default single column number. */
- ctx->single_column = 0;
-
- /*
- * In multi column mode start parsing sample data at the first column
- * and in single column mode at the first bit.
- */
- ctx->first_channel = 0;
+static int init(struct sr_input *in, GHashTable *options)
+{
+ struct context *inc;
+ const char *s;
- /* Start at the beginning of the file. */
- ctx->start_line = 1;
+ in->sdi = g_malloc0(sizeof(struct sr_dev_inst));
+ in->priv = inc = g_malloc0(sizeof(struct context));
- /* Disable the usage of the first line as header by default. */
- ctx->header = FALSE;
+ inc->single_column = g_variant_get_int32(g_hash_table_lookup(options, "single-column"));
+ inc->multi_column_mode = inc->single_column == 0;
- /* Set default format for single column mode. */
- ctx->format = FORMAT_BIN;
+ inc->num_channels = g_variant_get_int32(g_hash_table_lookup(options, "numchannels"));
- if (!(ctx->buffer = g_string_new(""))) {
- sr_err("Line buffer malloc failed.");
- free_context(ctx);
- return SR_ERR_MALLOC;
+ inc->delimiter = g_string_new(g_variant_get_string(
+ g_hash_table_lookup(options, "delimiter"), NULL));
+ if (inc->delimiter->len == 0) {
+ sr_err("Delimiter must be at least one character.");
+ return SR_ERR_ARG;
}
- if (in->param) {
- if ((param = g_hash_table_lookup(in->param, "samplerate"))) {
- res = sr_parse_sizestring(param, &ctx->samplerate);
-
- if (res != SR_OK) {
- sr_err("Invalid samplerate: %s.", param);
- free_context(ctx);
- return SR_ERR_ARG;
- }
- }
-
- if ((param = g_hash_table_lookup(in->param, "numchannels")))
- ctx->num_channels = g_ascii_strtoull(param, NULL, 10);
-
- if ((param = g_hash_table_lookup(in->param, "delimiter"))) {
- if (!strlen(param)) {
- sr_err("Delimiter must be at least one character.");
- free_context(ctx);
- return SR_ERR_ARG;
- }
-
- if (!g_ascii_strcasecmp(param, "\\t"))
- g_string_assign(ctx->delimiter, "\t");
- else
- g_string_assign(ctx->delimiter, param);
- }
-
- if ((param = g_hash_table_lookup(in->param, "comment")))
- g_string_assign(ctx->comment, param);
-
- if ((param = g_hash_table_lookup(in->param, "single-column"))) {
- ctx->single_column = g_ascii_strtoull(param, &ptr, 10);
- ctx->multi_column_mode = FALSE;
+ s = g_variant_get_string(g_hash_table_lookup(options, "format"), NULL);
+ if (!g_ascii_strncasecmp(s, "bin", 3)) {
+ inc->format = FORMAT_BIN;
+ } else if (!g_ascii_strncasecmp(s, "hex", 3)) {
+ inc->format = FORMAT_HEX;
+ } else if (!g_ascii_strncasecmp(s, "oct", 3)) {
+ inc->format = FORMAT_OCT;
+ } else {
+ sr_err("Invalid format: '%s'", s);
+ return SR_ERR_ARG;
+ }
- if (param == ptr) {
- sr_err("Invalid single-colum number: %s.",
- param);
- free_context(ctx);
- return SR_ERR_ARG;
- }
- }
+ inc->comment = g_string_new(g_variant_get_string(
+ g_hash_table_lookup(options, "comment"), NULL));
+ if (g_string_equal(inc->comment, inc->delimiter)) {
+ /* That's never going to work. Likely the result of the user
+ * setting the delimiter to ; -- the default comment. Clearing
+ * the comment setting will work in that case. */
+ g_string_truncate(inc->comment, 0);
+ }
- if ((param = g_hash_table_lookup(in->param, "first-channel")))
- ctx->first_channel = g_ascii_strtoull(param, NULL, 10);
+ inc->samplerate = g_variant_get_uint64(g_hash_table_lookup(options, "samplerate"));
- if ((param = g_hash_table_lookup(in->param, "startline"))) {
- ctx->start_line = g_ascii_strtoull(param, NULL, 10);
+ inc->first_channel = g_variant_get_int32(g_hash_table_lookup(options, "first-channel"));
- if (ctx->start_line < 1) {
- sr_err("Invalid start line: %s.", param);
- free_context(ctx);
- return SR_ERR_ARG;
- }
- }
+ inc->header = g_variant_get_boolean(g_hash_table_lookup(options, "header"));
- if ((param = g_hash_table_lookup(in->param, "header")))
- ctx->header = sr_parse_boolstring(param);
-
- if ((param = g_hash_table_lookup(in->param, "format"))) {
- if (!g_ascii_strncasecmp(param, "bin", 3)) {
- ctx->format = FORMAT_BIN;
- } else if (!g_ascii_strncasecmp(param, "hex", 3)) {
- ctx->format = FORMAT_HEX;
- } else if (!g_ascii_strncasecmp(param, "oct", 3)) {
- ctx->format = FORMAT_OCT;
- } else {
- sr_err("Invalid format: %s.", param);
- free_context(ctx);
- return SR_ERR;
- }
- }
+ inc->start_line = g_variant_get_int32(g_hash_table_lookup(options, "startline"));
+ if (inc->start_line < 1) {
+ sr_err("Invalid start line %zu.", inc->start_line);
+ return SR_ERR_ARG;
}
- if (ctx->multi_column_mode)
- ctx->first_column = ctx->first_channel;
+ if (inc->multi_column_mode)
+ inc->first_column = inc->first_channel;
else
- ctx->first_column = ctx->single_column;
+ inc->first_column = inc->single_column;
- if (!ctx->multi_column_mode && !ctx->num_channels) {
+ if (!inc->multi_column_mode && !inc->num_channels) {
sr_err("Number of channels needs to be specified in single column mode.");
- free_context(ctx);
- return SR_ERR;
+ return SR_ERR_ARG;
}
- if (!(ctx->channel = g_io_channel_new_file(filename, "r", NULL))) {
- sr_err("Input file '%s' could not be opened.", filename);
- free_context(ctx);
- return SR_ERR;
- }
+ return SR_OK;
+}
- while (TRUE) {
- ctx->line_number++;
- status = g_io_channel_read_line_string(ctx->channel,
- ctx->buffer, &term_pos, NULL);
+static const char *delim_set = "\r\n";
- if (status == G_IO_STATUS_EOF) {
- sr_err("Input file is empty.");
- free_context(ctx);
- return SR_ERR;
- }
+static const char *get_line_termination(GString *buf)
+{
+ const char *term;
- if (status != G_IO_STATUS_NORMAL) {
- sr_err("Error while reading line %zu.",
- ctx->line_number);
- free_context(ctx);
- return SR_ERR;
- }
+ term = NULL;
+ if (g_strstr_len(buf->str, buf->len, "\r\n"))
+ term = "\r\n";
+ else if (memchr(buf->str, '\n', buf->len))
+ term = "\n";
+ else if (memchr(buf->str, '\r', buf->len))
+ term = "\r";
- if (ctx->start_line > ctx->line_number) {
- sr_spew("Line %zu skipped.", ctx->line_number);
+ return term;
+}
+
+static int initial_parse(const struct sr_input *in, GString *buf)
+{
+ struct context *inc;
+ GString *channel_name;
+ unsigned int num_columns, i;
+ size_t line_number, l;
+ int ret;
+ char **lines, *line, **columns, *column;
+
+ ret = SR_OK;
+ inc = in->priv;
+ columns = NULL;
+
+ line_number = 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;
}
-
- /* Remove line termination character(s). */
- g_string_truncate(ctx->buffer, term_pos);
-
- if (!ctx->buffer->len) {
- sr_spew("Blank line %zu skipped.", ctx->line_number);
+ if (line[0] == '\0') {
+ sr_spew("Blank line %zu skipped.", line_number);
+ continue;
+ }
+ strip_comment(line, inc->comment);
+ if (line[0] == '\0') {
+ sr_spew("Comment-only line %zu skipped.", line_number);
continue;
}
- /* Remove trailing comment. */
- strip_comment(ctx->buffer, ctx->comment);
-
- if (ctx->buffer->len)
- break;
-
- sr_spew("Comment-only line %zu skipped.", ctx->line_number);
+ /* Reached first proper line. */
+ break;
+ }
+ if (!lines[l]) {
+ /* Not enough data for a proper line yet. */
+ ret = SR_ERR_NA;
+ goto out;
}
/*
* In order to determine the number of columns parse the current line
* without limiting the number of columns.
*/
- if (!(columns = parse_line(ctx, -1))) {
- sr_err("Error while parsing line %zu.", ctx->line_number);
- free_context(ctx);
- return SR_ERR;
+ columns = parse_line(line, inc, -1);
+ if (!columns) {
+ sr_err("Error while parsing line %zu.", line_number);
+ ret = SR_ERR;
+ goto out;
}
-
num_columns = g_strv_length(columns);
/* Ensure that the first column is not out of bounds. */
if (!num_columns) {
- sr_err("Column %zu in line %zu is out of bounds.",
- ctx->first_column, ctx->line_number);
- g_strfreev(columns);
- free_context(ctx);
- return SR_ERR;
+ sr_err("Column %u in line %zu is out of bounds.",
+ inc->first_column, line_number);
+ ret = SR_ERR;
+ goto out;
}
- if (ctx->multi_column_mode) {
+ if (inc->multi_column_mode) {
/*
* Detect the number of channels in multi column mode
* automatically if not specified.
*/
- if (!ctx->num_channels) {
- ctx->num_channels = num_columns;
- sr_info("Number of auto-detected channels: %zu.",
- ctx->num_channels);
+ if (!inc->num_channels) {
+ inc->num_channels = num_columns;
+ sr_dbg("Number of auto-detected channels: %u.",
+ inc->num_channels);
}
/*
* Ensure that the number of channels does not exceed the number
* of columns in multi column mode.
*/
- if (num_columns < ctx->num_channels) {
+ if (num_columns < inc->num_channels) {
sr_err("Not enough columns for desired number of channels in line %zu.",
- ctx->line_number);
- g_strfreev(columns);
- free_context(ctx);
- return SR_ERR;
+ line_number);
+ ret = SR_ERR;
+ goto out;
}
}
- for (i = 0; i < ctx->num_channels; i++) {
- if (ctx->header && ctx->multi_column_mode && strlen(columns[i]))
- snprintf(channel_name, sizeof(channel_name), "%s",
- columns[i]);
+ channel_name = g_string_sized_new(64);
+ for (i = 0; i < inc->num_channels; i++) {
+ column = columns[i];
+ if (inc->header && inc->multi_column_mode && column[0] != '\0')
+ g_string_assign(channel_name, column);
else
- snprintf(channel_name, sizeof(channel_name), "%zu", i);
-
- ch = sr_channel_new(i, SR_CHANNEL_LOGIC, TRUE, channel_name);
-
- if (!ch) {
- sr_err("Channel creation failed.");
- free_context(ctx);
- g_strfreev(columns);
- return SR_ERR;
- }
-
- in->sdi->channels = g_slist_append(in->sdi->channels, ch);
+ g_string_printf(channel_name, "%u", i);
+ sr_channel_new(in->sdi, i, SR_CHANNEL_LOGIC, TRUE, channel_name->str);
}
-
- g_strfreev(columns);
+ g_string_free(channel_name, TRUE);
/*
- * Calculate the minimum buffer size to store the sample data of the
- * channels.
+ * Calculate the minimum buffer size to store the set of samples
+ * of all channels (unit size). Determine a larger buffer size
+ * for datafeed submission that is a multiple of the unit size.
+ * Allocate the larger buffer, and have the "sample buffer" point
+ * to a location within that large buffer.
*/
- ctx->sample_buffer_size = (ctx->num_channels + 7) >> 3;
+ inc->sample_unit_size = (inc->num_channels + 7) / 8;
+ inc->datafeed_buf_size = CHUNK_SIZE;
+ inc->datafeed_buf_size *= inc->sample_unit_size;
+ inc->datafeed_buffer = g_malloc(inc->datafeed_buf_size);
+ inc->datafeed_buf_fill = 0;
+ inc->sample_buffer = &inc->datafeed_buffer[inc->datafeed_buf_fill];
+
+out:
+ if (columns)
+ g_strfreev(columns);
+ g_strfreev(lines);
- if (!(ctx->sample_buffer = g_try_malloc(ctx->sample_buffer_size))) {
- sr_err("Sample buffer malloc failed.");
- free_context(ctx);
- return SR_ERR_MALLOC;
- }
+ return ret;
+}
- return SR_OK;
+/*
+ * 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 loadfile(struct sr_input *in, const char *filename)
+static int initial_receive(const struct sr_input *in)
{
- int res;
- struct context *ctx;
- struct sr_datafeed_packet packet;
- struct sr_datafeed_meta meta;
- struct sr_config *cfg;
- GIOStatus status;
- gboolean read_new_line;
- gsize term_pos;
- char **columns;
- gsize num_columns;
- int max_columns;
+ struct context *inc;
+ GString *new_buf;
+ int len, ret;
+ char *p;
+ const char *termination;
- (void)filename;
+ initial_bom_check(in);
- ctx = in->internal;
+ inc = in->priv;
- /* Send header packet to the session bus. */
- std_session_send_df_header(in->sdi, LOG_PREFIX);
+ termination = get_line_termination(in->buf);
+ if (!termination)
+ /* Don't have a full line yet. */
+ return SR_ERR_NA;
- if (ctx->samplerate) {
- packet.type = SR_DF_META;
- packet.payload = &meta;
- cfg = sr_config_new(SR_CONF_SAMPLERATE,
- g_variant_new_uint64(ctx->samplerate));
- meta.config = g_slist_append(NULL, cfg);
- sr_session_send(in->sdi, &packet);
- sr_config_free(cfg);
- }
+ 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;
+ new_buf = g_string_new_len(in->buf->str, len);
+ g_string_append_c(new_buf, '\0');
- read_new_line = FALSE;
+ inc->termination = g_strdup(termination);
- /* Limit the number of columns to parse. */
- if (ctx->multi_column_mode)
- max_columns = ctx->num_channels;
+ if (in->buf->str[0] != '\0')
+ ret = initial_parse(in, new_buf);
else
- max_columns = 1;
+ ret = SR_OK;
- while (TRUE) {
- /*
- * Skip reading a new line for the first time if the last read
- * line was not a header because the sample data is not parsed
- * yet.
- */
- if (read_new_line || ctx->header) {
- ctx->line_number++;
- status = g_io_channel_read_line_string(ctx->channel,
- ctx->buffer, &term_pos, NULL);
-
- if (status == G_IO_STATUS_EOF)
- break;
-
- if (status != G_IO_STATUS_NORMAL) {
- sr_err("Error while reading line %zu.",
- ctx->line_number);
- free_context(ctx);
- return SR_ERR;
- }
-
- /* Remove line termination character(s). */
- g_string_truncate(ctx->buffer, term_pos);
+ g_string_free(new_buf, TRUE);
+
+ return ret;
+}
+
+static int process_buffer(struct sr_input *in, gboolean is_eof)
+{
+ struct sr_datafeed_packet packet;
+ struct sr_datafeed_meta meta;
+ struct sr_config *src;
+ struct context *inc;
+ gsize num_columns;
+ uint64_t samplerate;
+ int max_columns, ret, l;
+ char *p, **lines, *line, **columns;
+
+ inc = in->priv;
+ if (!inc->started) {
+ std_session_send_df_header(in->sdi);
+
+ if (inc->samplerate) {
+ packet.type = SR_DF_META;
+ packet.payload = &meta;
+ samplerate = inc->samplerate;
+ 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);
}
- read_new_line = TRUE;
+ inc->started = TRUE;
+ }
+
+ /* Limit the number of columns to parse. */
+ if (inc->multi_column_mode)
+ max_columns = inc->num_channels;
+ else
+ max_columns = 1;
- if (!ctx->buffer->len) {
- sr_spew("Blank line %zu skipped.", ctx->line_number);
+ /*
+ * 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, delim_set, 0);
+ for (l = 0; lines[l]; l++) {
+ inc->line_number++;
+ line = lines[l];
+ if (line[0] == '\0') {
+ sr_spew("Blank line %zu skipped.", inc->line_number);
continue;
}
/* Remove trailing comment. */
- strip_comment(ctx->buffer, ctx->comment);
+ strip_comment(line, inc->comment);
+ if (line[0] == '\0') {
+ sr_spew("Comment-only line %zu skipped.", inc->line_number);
+ continue;
+ }
- if (!ctx->buffer->len) {
- sr_spew("Comment-only line %zu skipped.",
- ctx->line_number);
+ /* Skip the header line, its content was used as the channel names. */
+ if (inc->header) {
+ sr_spew("Header line %zu skipped.", inc->line_number);
+ inc->header = FALSE;
continue;
}
- if (!(columns = parse_line(ctx, max_columns))) {
- sr_err("Error while parsing line %zu.",
- ctx->line_number);
- free_context(ctx);
+ columns = parse_line(line, inc, max_columns);
+ if (!columns) {
+ sr_err("Error while parsing line %zu.", inc->line_number);
+ g_strfreev(lines);
return SR_ERR;
}
-
num_columns = g_strv_length(columns);
-
- /* Ensure that the first column is not out of bounds. */
if (!num_columns) {
- sr_err("Column %zu in line %zu is out of bounds.",
- ctx->first_column, ctx->line_number);
+ sr_err("Column %u in line %zu is out of bounds.",
+ inc->first_column, inc->line_number);
g_strfreev(columns);
- free_context(ctx);
+ g_strfreev(lines);
return SR_ERR;
}
-
/*
* Ensure that the number of channels does not exceed the number
* of columns in multi column mode.
*/
- if (ctx->multi_column_mode && num_columns < ctx->num_channels) {
+ if (inc->multi_column_mode && num_columns < inc->num_channels) {
sr_err("Not enough columns for desired number of channels in line %zu.",
- ctx->line_number);
+ inc->line_number);
g_strfreev(columns);
- free_context(ctx);
+ g_strfreev(lines);
return SR_ERR;
}
- if (ctx->multi_column_mode)
- res = parse_multi_columns(columns, ctx);
+ if (inc->multi_column_mode)
+ ret = parse_multi_columns(columns, inc);
else
- res = parse_single_column(columns[0], ctx);
+ ret = parse_single_column(columns[0], inc);
+ if (ret != SR_OK) {
+ g_strfreev(columns);
+ g_strfreev(lines);
+ return SR_ERR;
+ }
- if (res != SR_OK) {
+ /* Send sample data to the session bus. */
+ ret = queue_samples(in);
+ if (ret != SR_OK) {
+ sr_err("Sending samples failed.");
g_strfreev(columns);
- free_context(ctx);
+ g_strfreev(lines);
return SR_ERR;
}
g_strfreev(columns);
+ }
+ g_strfreev(lines);
+ g_string_erase(in->buf, 0, p - in->buf->str);
- /*
- * TODO: Parse sample numbers / timestamps and use it for
- * decompression.
- */
-
- /* Send sample data to the session bus. */
- res = send_samples(in->sdi, ctx->sample_buffer,
- ctx->sample_buffer_size, 1);
+ return ret;
+}
- if (res != SR_OK) {
- sr_err("Sending samples failed.");
- free_context(ctx);
+static int receive(struct sr_input *in, GString *buf)
+{
+ struct context *inc;
+ int ret;
+
+ g_string_append_len(in->buf, buf->str, buf->len);
+
+ inc = in->priv;
+ if (!inc->termination) {
+ ret = initial_receive(in);
+ if (ret == SR_ERR_NA)
+ /* Not enough data yet. */
+ return SR_OK;
+ else if (ret != SR_OK)
return SR_ERR;
- }
+
+ /* sdi is ready, notify frontend. */
+ in->sdi_ready = TRUE;
+ return SR_OK;
}
- /* Send end packet to the session bus. */
- packet.type = SR_DF_END;
- sr_session_send(in->sdi, &packet);
+ ret = process_buffer(in, FALSE);
+
+ return ret;
+}
+
+static int end(struct sr_input *in)
+{
+ struct context *inc;
+ int ret;
+
+ if (in->sdi_ready)
+ ret = process_buffer(in, TRUE);
+ else
+ ret = SR_OK;
+ if (ret != SR_OK)
+ return ret;
+
+ ret = flush_samples(in);
+ if (ret != SR_OK)
+ return ret;
+
+ inc = in->priv;
+ if (inc->started)
+ std_session_send_df_end(in->sdi);
+
+ return ret;
+}
+
+static void cleanup(struct sr_input *in)
+{
+ struct context *inc;
+
+ inc = in->priv;
+
+ if (inc->delimiter)
+ g_string_free(inc->delimiter, TRUE);
- free_context(ctx);
+ if (inc->comment)
+ g_string_free(inc->comment, TRUE);
+
+ g_free(inc->termination);
+ g_free(inc->datafeed_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 single-column mode, using the specified column (>= 1); 0: multi-col. mode", NULL, NULL },
+ { "numchannels", "Number of logic channels", "The number of (logic) channels (single-col. mode: number of bits beginning at 'first channel', LSB-first)", NULL, NULL },
+ { "delimiter", "Column delimiter", "The column delimiter (>= 1 characters)", NULL, NULL },
+ { "format", "Data format (single-col. mode)", "The numeric format of the data (single-col. mode): bin, hex, oct", NULL, NULL },
+ { "comment", "Comment character(s)", "The comment prefix character(s)", NULL, NULL },
+ { "samplerate", "Samplerate (Hz)", "The sample rate (used during capture) in Hz", NULL, NULL },
+ { "first-channel", "First channel", "The column number of the first channel (multi-col. mode); bit position for the first channel (single-col. mode)", NULL, NULL },
+ { "header", "Interpret first line as header (multi-col. mode)", "Treat the first line as header with channel names (multi-col. mode)", NULL, NULL },
+ { "startline", "Start line", "The line number at which to start processing samples (>= 1)", NULL, NULL },
+ ALL_ZERO
+};
+
+static const struct sr_option *get_options(void)
+{
+ GSList *l;
+
+ if (!options[0].def) {
+ options[0].def = g_variant_ref_sink(g_variant_new_int32(0));
+ options[1].def = g_variant_ref_sink(g_variant_new_int32(0));
+ options[2].def = g_variant_ref_sink(g_variant_new_string(","));
+ options[3].def = g_variant_ref_sink(g_variant_new_string("bin"));
+ l = NULL;
+ l = g_slist_append(l, g_variant_ref_sink(g_variant_new_string("bin")));
+ l = g_slist_append(l, g_variant_ref_sink(g_variant_new_string("hex")));
+ l = g_slist_append(l, g_variant_ref_sink(g_variant_new_string("oct")));
+ options[3].values = l;
+ options[4].def = g_variant_ref_sink(g_variant_new_string(";"));
+ options[5].def = g_variant_ref_sink(g_variant_new_uint64(0));
+ options[6].def = g_variant_ref_sink(g_variant_new_int32(0));
+ options[7].def = g_variant_ref_sink(g_variant_new_boolean(FALSE));
+ options[8].def = g_variant_ref_sink(g_variant_new_int32(1));
+ }
+
+ return options;
+}
+
SR_PRIV struct sr_input_module input_csv = {
.id = "csv",
- .description = "Comma-separated values (CSV)",
- .format_match = format_match,
+ .name = "CSV",
+ .desc = "Comma-separated values",
+ .exts = (const char*[]){"csv", NULL},
+ .options = get_options,
.init = init,
- .loadfile = loadfile,
+ .receive = receive,
+ .end = end,
+ .cleanup = cleanup,
+ .reset = reset,
};