X-Git-Url: https://sigrok.org/gitweb/?a=blobdiff_plain;f=src%2Finput%2Fcsv.c;h=b8574d3e0542efee8233d4cd9571869075da13f6;hb=c6aa9870b4e44c94621ea124898b9b860acf86ce;hp=baf37188ad6427ffd903ab31b62197e651dd45b4;hpb=4439363aa02928098eab8dcd18b5ccb41737c882;p=libsigrok.git
diff --git a/src/input/csv.c b/src/input/csv.c
index baf37188..b8574d3e 100644
--- a/src/input/csv.c
+++ b/src/input/csv.c
@@ -17,15 +17,19 @@
* along with this program. If not, see .
*/
-#include
+#include "config.h"
+
+#include
#include
#include
-#include
+
#include
#include "libsigrok-internal.h"
#define LOG_PREFIX "input/csv"
+#define CHUNK_SIZE (4 * 1024 * 1024)
+
/*
* The CSV input module has the following options:
*
@@ -74,11 +78,7 @@
* 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.
- * - The current implementation insists in the presence of end-of-line
- * markers on _every_ line in the input stream. "Incomplete" text
- * files that are so typical on the Windows platform get rejected as
- * invalid.
+ * 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
@@ -89,9 +89,7 @@
* 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). Potential absence of the terminator for the last
- * line is orthogonal, and can get handled by a "force" flag when
- * the end() routine calls the process_buffer() routine.
+ * 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
@@ -107,10 +105,10 @@
*/
/* Single column formats. */
-enum {
+enum single_col_format {
FORMAT_BIN,
FORMAT_HEX,
- FORMAT_OCT
+ FORMAT_OCT,
};
struct context {
@@ -120,7 +118,7 @@ struct context {
uint64_t samplerate;
/* Number of channels. */
- unsigned int num_channels;
+ size_t num_channels;
/* Column delimiter character(s). */
GString *delimiter;
@@ -135,20 +133,20 @@ struct context {
gboolean multi_column_mode;
/* Column number of the sample data in single column mode. */
- unsigned int single_column;
+ size_t 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.
*/
- unsigned int first_column;
+ size_t 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.
*/
- unsigned int first_channel;
+ size_t first_channel;
/* Line number to start processing. */
size_t start_line;
@@ -160,16 +158,20 @@ struct context {
gboolean header;
/* Format sample data is stored in single column mode. */
- int format;
+ enum single_col_format format;
- /* Size of the sample buffer. */
- size_t sample_buffer_size;
+ size_t sample_unit_size; /**!< Byte count for a single sample. */
+ uint8_t *sample_buffer; /**!< Buffer for a single sample. */
- /* Buffer to store sample data. */
- uint8_t *sample_buffer;
+ uint8_t *datafeed_buffer; /**!< Queue for datafeed submission. */
+ size_t datafeed_buf_size;
+ size_t datafeed_buf_fill;
/* Current line number. */
size_t line_number;
+
+ /* List of previously created sigrok channels. */
+ GSList *prev_sr_channels;
};
static void strip_comment(char *buf, const GString *prefix)
@@ -190,13 +192,13 @@ static int parse_binstr(const char *str, struct context *inc)
length = strlen(str);
if (!length) {
- sr_err("Column %u in line %zu is empty.", inc->single_column,
+ sr_err("Column %zu in line %zu is empty.", inc->single_column,
inc->line_number);
return SR_ERR;
}
/* Clear buffer in order to set bits only. */
- memset(inc->sample_buffer, 0, (inc->num_channels + 7) >> 3);
+ memset(inc->sample_buffer, 0, inc->sample_unit_size);
i = inc->first_channel;
@@ -204,7 +206,7 @@ static int parse_binstr(const char *str, struct context *inc)
if (str[length - i - 1] == '1') {
inc->sample_buffer[j / 8] |= (1 << (j % 8));
} else if (str[length - i - 1] != '0') {
- sr_err("Invalid value '%s' in column %u in line %zu.",
+ sr_err("Invalid value '%s' in column %zu in line %zu.",
str, inc->single_column, inc->line_number);
return SR_ERR;
}
@@ -222,13 +224,13 @@ static int parse_hexstr(const char *str, struct context *inc)
length = strlen(str);
if (!length) {
- sr_err("Column %u in line %zu is empty.", inc->single_column,
+ sr_err("Column %zu in line %zu is empty.", inc->single_column,
inc->line_number);
return SR_ERR;
}
/* Clear buffer in order to set bits only. */
- memset(inc->sample_buffer, 0, (inc->num_channels + 7) >> 3);
+ memset(inc->sample_buffer, 0, inc->sample_unit_size);
/* Calculate the position of the first hexadecimal digit. */
i = inc->first_channel / 4;
@@ -237,7 +239,7 @@ static int parse_hexstr(const char *str, struct context *inc)
c = str[length - i - 1];
if (!g_ascii_isxdigit(c)) {
- sr_err("Invalid value '%s' in column %u in line %zu.",
+ sr_err("Invalid value '%s' in column %zu in line %zu.",
str, inc->single_column, inc->line_number);
return SR_ERR;
}
@@ -266,13 +268,13 @@ static int parse_octstr(const char *str, struct context *inc)
length = strlen(str);
if (!length) {
- sr_err("Column %u in line %zu is empty.", inc->single_column,
+ sr_err("Column %zu in line %zu is empty.", inc->single_column,
inc->line_number);
return SR_ERR;
}
/* Clear buffer in order to set bits only. */
- memset(inc->sample_buffer, 0, (inc->num_channels + 7) >> 3);
+ memset(inc->sample_buffer, 0, inc->sample_unit_size);
/* Calculate the position of the first octal digit. */
i = inc->first_channel / 3;
@@ -281,7 +283,7 @@ static int parse_octstr(const char *str, struct context *inc)
c = str[length - i - 1];
if (c < '0' || c > '7') {
- sr_err("Invalid value '%s' in column %u in line %zu.",
+ sr_err("Invalid value '%s' in column %zu in line %zu.",
str, inc->single_column, inc->line_number);
return SR_ERR;
}
@@ -301,7 +303,16 @@ static int parse_octstr(const char *str, struct context *inc)
return SR_OK;
}
-static char **parse_line(char *buf, struct context *inc, int max_columns)
+/**
+ * @brief Splits a text line into a set of columns.
+ *
+ * @param[in] buf The input text line to split.
+ * @param[in] inc The input module's context.
+ * @param[in] max_cols The maximum column count, negative to get all of them.
+ *
+ * @returns An array of strings, representing the columns' text.
+ */
+static char **parse_line(char *buf, struct context *inc, ssize_t max_cols)
{
const char *str, *remainder;
GSList *list, *l;
@@ -316,12 +327,12 @@ static char **parse_line(char *buf, struct context *inc, int max_columns)
remainder = buf;
str = strstr(remainder, inc->delimiter->str);
- while (str && max_columns) {
+ while (str && max_cols) {
if (n >= inc->first_column) {
column = g_strndup(remainder, str - remainder);
list = g_slist_prepend(list, g_strstrip(column));
- max_columns--;
+ max_cols--;
k++;
}
@@ -330,7 +341,7 @@ static char **parse_line(char *buf, struct context *inc, int max_columns)
n++;
}
- if (buf[0] && max_columns && n >= inc->first_column) {
+ if (buf[0] && max_cols && n >= inc->first_column) {
column = g_strdup(remainder);
list = g_slist_prepend(list, g_strstrip(column));
k++;
@@ -355,7 +366,7 @@ static int parse_multi_columns(char **columns, struct context *inc)
char *column;
/* Clear buffer in order to set bits only. */
- memset(inc->sample_buffer, 0, (inc->num_channels + 7) >> 3);
+ memset(inc->sample_buffer, 0, inc->sample_unit_size);
for (i = 0; i < inc->num_channels; i++) {
column = columns[i];
@@ -397,26 +408,47 @@ static int parse_single_column(const char *column, struct context *inc)
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)
{
+ struct context *inc;
struct sr_datafeed_packet packet;
struct sr_datafeed_logic logic;
- int res;
- 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;
-
- for (i = 0; i < count; i++) {
- res = sr_session_send(sdi, &packet);
- if (res != SR_OK)
- return res;
- }
+ logic.unitsize = inc->sample_unit_size;
+ logic.length = inc->datafeed_buf_fill;
+ logic.data = inc->datafeed_buffer;
+
+ rc = sr_session_send(in->sdi, &packet);
+ if (rc != SR_OK)
+ return rc;
+
+ inc->datafeed_buf_fill = 0;
+ return SR_OK;
+}
+
+static int queue_samples(const struct sr_input *in)
+{
+ struct context *inc;
+ int rc;
+
+ inc = in->priv;
+ 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;
}
@@ -428,10 +460,10 @@ static int init(struct sr_input *in, GHashTable *options)
in->sdi = g_malloc0(sizeof(struct sr_dev_inst));
in->priv = inc = g_malloc0(sizeof(struct context));
- inc->single_column = g_variant_get_int32(g_hash_table_lookup(options, "single-column"));
+ inc->single_column = g_variant_get_uint32(g_hash_table_lookup(options, "single-column"));
inc->multi_column_mode = inc->single_column == 0;
- inc->num_channels = g_variant_get_int32(g_hash_table_lookup(options, "numchannels"));
+ inc->num_channels = g_variant_get_uint32(g_hash_table_lookup(options, "numchannels"));
inc->delimiter = g_string_new(g_variant_get_string(
g_hash_table_lookup(options, "delimiter"), NULL));
@@ -463,11 +495,11 @@ static int init(struct sr_input *in, GHashTable *options)
inc->samplerate = g_variant_get_uint64(g_hash_table_lookup(options, "samplerate"));
- inc->first_channel = g_variant_get_int32(g_hash_table_lookup(options, "first-channel"));
+ inc->first_channel = g_variant_get_uint32(g_hash_table_lookup(options, "first-channel"));
inc->header = g_variant_get_boolean(g_hash_table_lookup(options, "header"));
- inc->start_line = g_variant_get_int32(g_hash_table_lookup(options, "startline"));
+ inc->start_line = g_variant_get_uint32(g_hash_table_lookup(options, "startline"));
if (inc->start_line < 1) {
sr_err("Invalid start line %zu.", inc->start_line);
return SR_ERR_ARG;
@@ -486,6 +518,44 @@ static int init(struct sr_input *in, GHashTable *options)
return SR_OK;
}
+/*
+ * Check the channel list for consistency across file re-import. See
+ * the VCD input module for more details and motivation.
+ */
+
+static void keep_header_for_reread(const struct sr_input *in)
+{
+ struct context *inc;
+
+ inc = in->priv;
+ g_slist_free_full(inc->prev_sr_channels, sr_channel_free_cb);
+ inc->prev_sr_channels = in->sdi->channels;
+ in->sdi->channels = NULL;
+}
+
+static int check_header_in_reread(const struct sr_input *in)
+{
+ struct context *inc;
+
+ if (!in)
+ return FALSE;
+ inc = in->priv;
+ if (!inc)
+ return FALSE;
+ if (!inc->prev_sr_channels)
+ return TRUE;
+
+ if (sr_channel_lists_differ(inc->prev_sr_channels, in->sdi->channels)) {
+ sr_err("Channel list change not supported for file re-read.");
+ return FALSE;
+ }
+ g_slist_free_full(in->sdi->channels, sr_channel_free_cb);
+ in->sdi->channels = inc->prev_sr_channels;
+ inc->prev_sr_channels = NULL;
+
+ return TRUE;
+}
+
static const char *delim_set = "\r\n";
static const char *get_line_termination(GString *buf)
@@ -507,7 +577,7 @@ static int initial_parse(const struct sr_input *in, GString *buf)
{
struct context *inc;
GString *channel_name;
- unsigned int num_columns, i;
+ size_t num_columns, i;
size_t line_number, l;
int ret;
char **lines, *line, **columns, *column;
@@ -558,7 +628,7 @@ static int initial_parse(const struct sr_input *in, GString *buf)
/* Ensure that the first column is not out of bounds. */
if (!num_columns) {
- sr_err("Column %u in line %zu is out of bounds.",
+ sr_err("Column %zu in line %zu is out of bounds.",
inc->first_column, line_number);
ret = SR_ERR;
goto out;
@@ -571,7 +641,7 @@ static int initial_parse(const struct sr_input *in, GString *buf)
*/
if (!inc->num_channels) {
inc->num_channels = num_columns;
- sr_dbg("Number of auto-detected channels: %u.",
+ sr_dbg("Number of auto-detected channels: %zu.",
inc->num_channels);
}
@@ -593,17 +663,28 @@ static int initial_parse(const struct sr_input *in, GString *buf)
if (inc->header && inc->multi_column_mode && column[0] != '\0')
g_string_assign(channel_name, column);
else
- g_string_printf(channel_name, "%u", i);
+ g_string_printf(channel_name, "%zu", i);
sr_channel_new(in->sdi, i, SR_CHANNEL_LOGIC, TRUE, channel_name->str);
}
g_string_free(channel_name, TRUE);
+ if (!check_header_in_reread(in)) {
+ ret = SR_ERR_DATA;
+ goto out;
+ }
/*
- * 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.
*/
- inc->sample_buffer_size = (inc->num_channels + 7) >> 3;
- inc->sample_buffer = g_malloc(inc->sample_buffer_size);
+ 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)
@@ -671,7 +752,7 @@ 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;
@@ -679,7 +760,8 @@ static int process_buffer(struct sr_input *in)
struct context *inc;
gsize num_columns;
uint64_t samplerate;
- int max_columns, ret, l;
+ size_t max_columns, l;
+ int ret;
char *p, **lines, *line, **columns;
inc = in->priv;
@@ -712,14 +794,24 @@ static int process_buffer(struct sr_input *in)
* 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;
- p = g_strrstr_len(in->buf->str, in->buf->len, inc->termination);
- if (!p)
- return SR_ERR;
- *p = '\0';
- p += strlen(inc->termination);
+ 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;
@@ -749,13 +841,15 @@ static int process_buffer(struct sr_input *in)
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);
if (!num_columns) {
- sr_err("Column %u in line %zu is out of bounds.",
+ sr_err("Column %zu in line %zu is out of bounds.",
inc->first_column, inc->line_number);
g_strfreev(columns);
+ g_strfreev(lines);
return SR_ERR;
}
/*
@@ -766,6 +860,7 @@ static int process_buffer(struct sr_input *in)
sr_err("Not enough columns for desired number of channels in line %zu.",
inc->line_number);
g_strfreev(columns);
+ g_strfreev(lines);
return SR_ERR;
}
@@ -775,16 +870,19 @@ static int process_buffer(struct sr_input *in)
ret = parse_single_column(columns[0], inc);
if (ret != SR_OK) {
g_strfreev(columns);
+ g_strfreev(lines);
return SR_ERR;
}
/* Send sample data to the session bus. */
- ret = send_samples(in->sdi, inc->sample_buffer,
- inc->sample_buffer_size, 1);
+ ret = queue_samples(in);
if (ret != SR_OK) {
sr_err("Sending samples failed.");
+ g_strfreev(columns);
+ g_strfreev(lines);
return SR_ERR;
}
+
g_strfreev(columns);
}
g_strfreev(lines);
@@ -814,7 +912,7 @@ static int receive(struct sr_input *in, GString *buf)
return SR_OK;
}
- ret = process_buffer(in);
+ ret = process_buffer(in, FALSE);
return ret;
}
@@ -825,9 +923,15 @@ static int end(struct sr_input *in)
int ret;
if (in->sdi_ready)
- ret = process_buffer(in);
+ 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)
@@ -840,16 +944,14 @@ static void cleanup(struct sr_input *in)
{
struct context *inc;
- inc = in->priv;
-
- if (inc->delimiter)
- g_string_free(inc->delimiter, TRUE);
+ keep_header_for_reread(in);
- if (inc->comment)
- g_string_free(inc->comment, TRUE);
+ inc = in->priv;
g_free(inc->termination);
- g_free(inc->sample_buffer);
+ inc->termination = NULL;
+ g_free(inc->datafeed_buffer);
+ inc->datafeed_buffer = NULL;
}
static int reset(struct sr_input *in)
@@ -863,31 +965,51 @@ static int reset(struct sr_input *in)
return SR_OK;
}
+enum option_index {
+ OPT_SINGLE_COL,
+ OPT_NUM_LOGIC,
+ OPT_DELIM,
+ OPT_FORMAT,
+ OPT_COMMENT,
+ OPT_RATE,
+ OPT_FIRST_LOGIC,
+ OPT_HEADER,
+ OPT_START,
+ OPT_MAX,
+};
+
static struct sr_option options[] = {
- { "single-column", "Single column", "Enable/specify single column", NULL, NULL },
- { "numchannels", "Max channels", "Number of channels", NULL, NULL },
- { "delimiter", "Delimiter", "Column delimiter", NULL, NULL },
- { "format", "Format", "Numeric format", NULL, NULL },
- { "comment", "Comment", "Comment prefix character", NULL, NULL },
- { "samplerate", "Samplerate", "Samplerate used during capture", NULL, NULL },
- { "first-channel", "First channel", "Column number of first channel", NULL, NULL },
- { "header", "Header", "Treat first line as header with channel names", NULL, NULL },
- { "startline", "Start line", "Line number at which to start processing samples", NULL, NULL },
- ALL_ZERO
+ [OPT_SINGLE_COL] = { "single-column", "Single column", "Enable single-column mode, using the specified column (>= 1); 0: multi-col. mode", NULL, NULL },
+ [OPT_NUM_LOGIC] = { "numchannels", "Number of logic channels", "The number of (logic) channels (single-col. mode: number of bits beginning at 'first channel', LSB-first)", NULL, NULL },
+ [OPT_DELIM] = { "delimiter", "Column delimiter", "The column delimiter (>= 1 characters)", NULL, NULL },
+ [OPT_FORMAT] = { "format", "Data format (single-col. mode)", "The numeric format of the data (single-col. mode): bin, hex, oct", NULL, NULL },
+ [OPT_COMMENT] = { "comment", "Comment character(s)", "The comment prefix character(s)", NULL, NULL },
+ [OPT_RATE] = { "samplerate", "Samplerate (Hz)", "The sample rate (used during capture) in Hz", NULL, NULL },
+ [OPT_FIRST_LOGIC] = { "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 },
+ [OPT_HEADER] = { "header", "Interpret first line as header (multi-col. mode)", "Treat the first line as header with channel names (multi-col. mode)", NULL, NULL },
+ [OPT_START] = { "startline", "Start line", "The line number at which to start processing samples (>= 1)", NULL, NULL },
+ [OPT_MAX] = 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"));
- 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));
+ options[OPT_SINGLE_COL].def = g_variant_ref_sink(g_variant_new_int32(0));
+ options[OPT_NUM_LOGIC].def = g_variant_ref_sink(g_variant_new_int32(0));
+ options[OPT_DELIM].def = g_variant_ref_sink(g_variant_new_string(","));
+ options[OPT_FORMAT].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[OPT_FORMAT].values = l;
+ options[OPT_COMMENT].def = g_variant_ref_sink(g_variant_new_string(";"));
+ options[OPT_RATE].def = g_variant_ref_sink(g_variant_new_uint64(0));
+ options[OPT_FIRST_LOGIC].def = g_variant_ref_sink(g_variant_new_int32(0));
+ options[OPT_HEADER].def = g_variant_ref_sink(g_variant_new_boolean(FALSE));
+ options[OPT_START].def = g_variant_ref_sink(g_variant_new_int32(1));
}
return options;