X-Git-Url: https://sigrok.org/gitweb/?a=blobdiff_plain;f=src%2Finput%2Fcsv.c;h=2fffc50394fb6b39765f45561160b686f90c8d75;hb=affaf540128f83c4f3396043077a94ba888cde1f;hp=ab22f0754f14ac788e1de61f1d78bbb6b1b3249a;hpb=7f4c3a622405b409e579e8344192e8daf15cb817;p=libsigrok.git diff --git a/src/input/csv.c b/src/input/csv.c index ab22f075..2fffc503 100644 --- a/src/input/csv.c +++ b/src/input/csv.c @@ -26,6 +26,8 @@ #define LOG_PREFIX "input/csv" +#define CHUNK_SIZE (4 * 1024 * 1024) + /* * The CSV input module has the following options: * @@ -74,11 +76,7 @@ * 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 +87,7 @@ * 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 @@ -162,14 +158,18 @@ struct context { /* Format sample data is stored in single column mode. */ int 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) @@ -196,7 +196,7 @@ 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; @@ -228,7 +228,7 @@ 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; @@ -272,7 +272,7 @@ 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; @@ -355,7 +355,7 @@ 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 +397,47 @@ 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; } @@ -486,6 +507,44 @@ 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) @@ -597,13 +656,24 @@ static int initial_parse(const struct sr_input *in, GString *buf) 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) @@ -759,6 +829,7 @@ static int process_buffer(struct sr_input *in, gboolean is_eof) 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); @@ -766,6 +837,7 @@ static int process_buffer(struct sr_input *in, gboolean is_eof) sr_err("Column %u in line %zu is out of bounds.", inc->first_column, inc->line_number); g_strfreev(columns); + g_strfreev(lines); return SR_ERR; } /* @@ -776,6 +848,7 @@ static int process_buffer(struct sr_input *in, gboolean is_eof) 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; } @@ -785,16 +858,19 @@ static int process_buffer(struct sr_input *in, gboolean is_eof) 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); @@ -838,6 +914,12 @@ static int end(struct sr_input *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) @@ -850,16 +932,14 @@ 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) @@ -874,25 +954,32 @@ static int reset(struct sr_input *in) } 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 }, + { "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));