X-Git-Url: https://sigrok.org/gitweb/?a=blobdiff_plain;f=src%2Finput%2Fcsv.c;h=6d3264902362f04363deb6e2b28e7c22748cf82c;hb=19267272d39e3ff16874f538b22b407ff242e87c;hp=cc721baf2427a3e89c3c50d12bf196569dfafe59;hpb=60107497fed4b307a7d64d23e3d8d6137f64c4e4;p=libsigrok.git
diff --git a/src/input/csv.c b/src/input/csv.c
index cc721baf..6d326490 100644
--- a/src/input/csv.c
+++ b/src/input/csv.c
@@ -17,14 +17,19 @@
* along with this program. If not, see .
*/
+#include "config.h"
+
+#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:
*
@@ -66,11 +71,44 @@
* 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 {
+enum single_col_format {
FORMAT_BIN,
FORMAT_HEX,
- FORMAT_OCT
+ FORMAT_OCT,
};
struct context {
@@ -80,7 +118,7 @@ struct context {
uint64_t samplerate;
/* Number of channels. */
- gsize num_channels;
+ size_t num_channels;
/* Column delimiter character(s). */
GString *delimiter;
@@ -88,30 +126,30 @@ 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. */
gboolean multi_column_mode;
/* Column number of the sample data in single column mode. */
- gsize 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.
*/
- gsize 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.
*/
- gsize first_channel;
+ size_t 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
@@ -120,28 +158,27 @@ 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. */
- gsize 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. */
- gsize line_number;
-};
+ 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;
+ /* List of previously created sigrok channels. */
+ GSList *prev_sr_channels;
+};
- return SR_ERR;
-}
+/*
+ * Primitive operations for text input: Strip comments off text lines.
+ * Split text lines into columns. Process input text for individual
+ * columns.
+ */
static void strip_comment(char *buf, const GString *prefix)
{
@@ -150,10 +187,26 @@ static void strip_comment(char *buf, const GString *prefix)
if (!prefix->len)
return;
- if ((ptr = strstr(buf, prefix->str)))
+ if ((ptr = strstr(buf, prefix->str))) {
*ptr = '\0';
+ g_strstrip(buf);
+ }
}
+/* TODO Move parse_line() here. */
+
+/**
+ * @brief Parse a text field into multiple bits, binary presentation.
+ *
+ * @param[in] str The input text, a run of 0/1 digits.
+ * @param[in] inc The input module's context.
+ *
+ * @retval SR_OK Success.
+ * @retval SR_ERR Invalid input data (empty, or format error).
+ *
+ * This routine modifies the logic levels in the current sample set,
+ * based on the text input which consists of binary digits.
+ */
static int parse_binstr(const char *str, struct context *inc)
{
gsize i, j, length;
@@ -167,7 +220,7 @@ static int parse_binstr(const char *str, struct context *inc)
}
/* 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;
@@ -184,6 +237,18 @@ static int parse_binstr(const char *str, struct context *inc)
return SR_OK;
}
+/**
+ * @brief Parse a text field into multiple bits, hexadecimal presentation.
+ *
+ * @param[in] str The input text, a run of hex digits.
+ * @param[in] inc The input module's context.
+ *
+ * @retval SR_OK Success.
+ * @retval SR_ERR Invalid input data (empty, or format error).
+ *
+ * This routine modifies the logic levels in the current sample set,
+ * based on the text input which consists of hexadecimal digits.
+ */
static int parse_hexstr(const char *str, struct context *inc)
{
gsize i, j, k, length;
@@ -199,7 +264,7 @@ static int parse_hexstr(const char *str, struct context *inc)
}
/* 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;
@@ -228,6 +293,18 @@ static int parse_hexstr(const char *str, struct context *inc)
return SR_OK;
}
+/**
+ * @brief Parse a text field into multiple bits, octal presentation.
+ *
+ * @param[in] str The input text, a run of oct digits.
+ * @param[in] inc The input module's context.
+ *
+ * @retval SR_OK Success.
+ * @retval SR_ERR Invalid input data (empty, or format error).
+ *
+ * This routine modifies the logic levels in the current sample set,
+ * based on the text input which consists of octal digits.
+ */
static int parse_octstr(const char *str, struct context *inc)
{
gsize i, j, k, length;
@@ -243,7 +320,7 @@ static int parse_octstr(const char *str, struct context *inc)
}
/* 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;
@@ -272,71 +349,112 @@ 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)
+static int parse_single_column(const char *column, struct context *inc)
+{
+ switch (inc->format) {
+ case FORMAT_BIN:
+ return parse_binstr(column, inc);
+ case FORMAT_HEX:
+ return parse_hexstr(column, inc);
+ case FORMAT_OCT:
+ return parse_octstr(column, inc);
+ }
+
+ return SR_ERR;
+}
+
+/**
+ * @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.
+ *
+ * This routine splits a text line on previously determined separators.
+ * A previously determined set of columns gets isolated (starting at a
+ * first position and spanning a given number of columns). A negative
+ * value for the maximum number of columns results in no restriction on
+ * the result set's length (the first columns still get trimmed off).
+ */
+static char **parse_line(char *buf, struct context *inc, ssize_t max_cols)
{
const char *str, *remainder;
GSList *list, *l;
char **columns;
char *column;
- gsize n, k;
+ gsize seen, taken;
- n = 0;
- k = 0;
+ seen = 0;
+ taken = 0;
list = NULL;
remainder = buf;
str = strstr(remainder, inc->delimiter->str);
-
- while (str && max_columns) {
- if (n >= inc->first_column) {
+ while (str && max_cols) {
+ if (seen >= inc->first_column) {
column = g_strndup(remainder, str - remainder);
list = g_slist_prepend(list, g_strstrip(column));
- max_columns--;
- k++;
+ max_cols--;
+ taken++;
}
remainder = str + inc->delimiter->len;
str = strstr(remainder, inc->delimiter->str);
- n++;
+ seen++;
}
- if (buf[0] && max_columns && n >= inc->first_column) {
+ if (buf[0] && max_cols && seen >= inc->first_column) {
column = g_strdup(remainder);
list = g_slist_prepend(list, g_strstrip(column));
- k++;
+ taken++;
}
- if (!(columns = g_try_new(char *, k + 1)))
+ if (!(columns = g_try_new(char *, taken + 1)))
return NULL;
-
- columns[k--] = NULL;
-
+ columns[taken--] = NULL;
for (l = list; l; l = l->next)
- columns[k--] = l->data;
+ columns[taken--] = l->data;
g_slist_free(list);
return columns;
}
+/**
+ * @brief Picks logic levels from multiple binary colomns, one channel per column.
+ *
+ * @param[in] columns The text fields which are kept in the columns.
+ * @param[in] inc The input module's context.
+ *
+ * @retval SR_OK Success.
+ * @retval SR_ERR Insufficient input, or syntax errors.
+ *
+ * This routine exclusively handles binary input where one logic channel
+ * occupies one column each. All channels are expected to reside in one
+ * consequtive run of 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);
+ memset(inc->sample_buffer, 0, inc->sample_unit_size);
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;
}
@@ -345,46 +463,47 @@ static int parse_multi_columns(char **columns, struct context *inc)
return SR_OK;
}
-static int parse_single_column(const char *column, struct context *inc)
+static int flush_samples(const struct sr_input *in)
{
- int res;
+ struct context *inc;
+ struct sr_datafeed_packet packet;
+ struct sr_datafeed_logic logic;
+ int rc;
- res = SR_ERR;
+ inc = in->priv;
+ if (!inc->datafeed_buf_fill)
+ return SR_OK;
- switch(inc->format) {
- case FORMAT_BIN:
- res = parse_binstr(column, inc);
- break;
- case FORMAT_HEX:
- res = parse_hexstr(column, inc);
- break;
- case FORMAT_OCT:
- res = parse_octstr(column, inc);
- break;
- }
+ memset(&packet, 0, sizeof(packet));
+ memset(&logic, 0, sizeof(logic));
+ packet.type = SR_DF_LOGIC;
+ packet.payload = &logic;
+ 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;
- return res;
+ inc->datafeed_buf_fill = 0;
+ return SR_OK;
}
-static int send_samples(const struct sr_dev_inst *sdi, uint8_t *buffer,
- gsize buffer_size, gsize count)
+static int queue_samples(const struct sr_input *in)
{
- struct sr_datafeed_packet packet;
- struct sr_datafeed_logic logic;
- int res;
- gsize i;
+ struct context *inc;
+ int rc;
- packet.type = SR_DF_LOGIC;
- packet.payload = &logic;
- logic.unitsize = buffer_size;
- logic.length = buffer_size;
- logic.data = buffer;
+ inc = in->priv;
- for (i = 0; i < count; i++) {
- if ((res = sr_session_send(sdi, &packet)) != SR_OK)
- return res;
+ 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;
}
@@ -393,13 +512,13 @@ static int init(struct sr_input *in, GHashTable *options)
struct context *inc;
const char *s;
- in->sdi = sr_dev_inst_new(SR_ST_ACTIVE, NULL, NULL, NULL);
+ 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));
@@ -431,13 +550,13 @@ 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 %d.", inc->start_line);
+ sr_err("Invalid start line %zu.", inc->start_line);
return SR_ERR_ARG;
}
@@ -454,9 +573,49 @@ static int init(struct sr_input *in, GHashTable *options)
return SR_OK;
}
-static char *get_line_termination(GString *buf)
+/*
+ * 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)
{
- char *term;
+ 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)
+{
+ const char *term;
term = NULL;
if (g_strstr_len(buf->str, buf->len, "\r\n"))
@@ -472,31 +631,31 @@ static char *get_line_termination(GString *buf)
static int initial_parse(const struct sr_input *in, GString *buf)
{
struct context *inc;
- struct sr_channel *ch;
GString *channel_name;
- gsize num_columns, l, i;
- unsigned int line_number;
+ size_t 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;
}
@@ -514,7 +673,8 @@ 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;
@@ -554,21 +714,32 @@ 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 && strlen(columns[i]))
- 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, "%lu", i);
- ch = sr_channel_new(i, SR_CHANNEL_LOGIC, TRUE, channel_name->str);
- in->sdi->channels = g_slist_append(in->sdi->channels, ch);
+ 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)
@@ -578,20 +749,46 @@ 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;
GString *new_buf;
int len, ret;
- char *termination, *p;
+ 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;
@@ -610,7 +807,7 @@ 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;
@@ -618,12 +815,13 @@ static int process_buffer(struct sr_input *in)
struct context *inc;
gsize num_columns;
uint64_t samplerate;
- int max_columns, ret, l;
- char *p, **lines, **columns;
+ size_t max_columns, l;
+ int ret;
+ 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;
@@ -632,42 +830,73 @@ 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;
- lines = g_strsplit_set(in->buf->str, "\r\n", 0);
+ /*
+ * 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++;
- 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;
}
- if (!(columns = parse_line(lines[l], inc, max_columns))) {
+ /* 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;
+ }
+
+ 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);
@@ -675,6 +904,7 @@ static int process_buffer(struct sr_input *in)
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;
}
/*
@@ -685,6 +915,7 @@ 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;
}
@@ -694,20 +925,23 @@ 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);
- g_string_erase(in->buf, 0, p - in->buf->str + 1);
+ g_string_erase(in->buf, 0, p - in->buf->str);
return ret;
}
@@ -721,7 +955,8 @@ 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)
@@ -732,7 +967,7 @@ static int receive(struct sr_input *in, GString *buf)
return SR_OK;
}
- ret = process_buffer(in);
+ ret = process_buffer(in, FALSE);
return ret;
}
@@ -740,20 +975,22 @@ 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;
+ if (ret != SR_OK)
+ return ret;
+
+ ret = flush_samples(in);
+ if (ret != SR_OK)
+ return ret;
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;
}
@@ -762,46 +999,72 @@ static void cleanup(struct sr_input *in)
{
struct context *inc;
+ keep_header_for_reread(in);
+
inc = in->priv;
- if (inc->delimiter)
- g_string_free(inc->delimiter, TRUE);
+ g_free(inc->termination);
+ inc->termination = NULL;
+ g_free(inc->datafeed_buffer);
+ inc->datafeed_buffer = NULL;
+}
- if (inc->comment)
- g_string_free(inc->comment, TRUE);
+static int reset(struct sr_input *in)
+{
+ struct context *inc = in->priv;
- if (inc->termination)
- g_free(inc->termination);
+ cleanup(in);
+ inc->started = FALSE;
+ g_string_truncate(in->buf, 0);
- if (inc->sample_buffer)
- g_free(inc->sample_buffer);
+ 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 struct sr_option *get_options(void)
+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;
@@ -811,11 +1074,11 @@ SR_PRIV struct sr_input_module input_csv = {
.id = "csv",
.name = "CSV",
.desc = "Comma-separated values",
- .metadata = { SR_INPUT_META_MIMETYPE },
+ .exts = (const char*[]){"csv", NULL},
.options = get_options,
- .format_match = format_match,
.init = init,
.receive = receive,
.end = end,
.cleanup = cleanup,
+ .reset = reset,
};