]>
Commit | Line | Data |
---|---|---|
4a35548b MS |
1 | /* |
2 | * This file is part of the libsigrok project. | |
3 | * | |
4 | * Copyright (C) 2013 Marc Schink <sigrok-dev@marcschink.de> | |
e53f32d2 | 5 | * Copyright (C) 2019 Gerhard Sittig <gerhard.sittig@gmx.net> |
4a35548b MS |
6 | * |
7 | * This program is free software: you can redistribute it and/or modify | |
8 | * it under the terms of the GNU General Public License as published by | |
9 | * the Free Software Foundation, either version 3 of the License, or | |
10 | * (at your option) any later version. | |
11 | * | |
12 | * This program is distributed in the hope that it will be useful, | |
13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | * GNU General Public License for more details. | |
16 | * | |
17 | * You should have received a copy of the GNU General Public License | |
18 | * along with this program. If not, see <http://www.gnu.org/licenses/>. | |
19 | */ | |
20 | ||
e05f1827 GS |
21 | #include "config.h" |
22 | ||
cd7c5f96 | 23 | #include <ctype.h> |
e05f1827 | 24 | #include <glib.h> |
4a35548b MS |
25 | #include <stdlib.h> |
26 | #include <string.h> | |
cd7c5f96 | 27 | #include <strings.h> |
e05f1827 | 28 | |
c1aae900 | 29 | #include <libsigrok/libsigrok.h> |
4a35548b | 30 | #include "libsigrok-internal.h" |
f6dcb320 | 31 | #include "scpi.h" /* String un-quote for channel name from header line. */ |
4a35548b | 32 | |
3544f848 | 33 | #define LOG_PREFIX "input/csv" |
4a35548b | 34 | |
9a4fd01a | 35 | #define CHUNK_SIZE (4 * 1024 * 1024) |
cd59e6ec | 36 | |
4a35548b MS |
37 | /* |
38 | * The CSV input module has the following options: | |
39 | * | |
72903e9d GS |
40 | * column_formats: Specifies the data formats and channel counts for the |
41 | * input file's text columns. Accepts a comma separated list of tuples | |
42 | * with: an optional column repeat count ('*' as a wildcard meaning | |
43 | * "all remaining columns", only applicable to the last field), a format | |
44 | * specifying character ('x' hexadecimal, 'o' octal, 'b' binary, 'l' | |
45 | * single-bit logic), and an optional bit count (translating to: logic | |
08eb955a GS |
46 | * channels communicated in that column). The 'a' format marks analog |
47 | * data, an optionally following number is the digits count (resolution). | |
7e4e65bf GS |
48 | * The 't' format marks timestamp values, which could help in automatic |
49 | * determination of the input stream's samplerate. This "column_formats" | |
50 | * option is most versatile, other forms of specifying the column layout | |
51 | * only exist for backwards compatibility, and are rather limited. They | |
52 | * exclusively support logic input data in strictly adjacent columns, | |
53 | * with further constraints on column layout for multi-bit data. | |
4a35548b | 54 | * |
72903e9d GS |
55 | * single_column: Specifies the column number which contains the logic data |
56 | * for single-column mode. All logic data is taken from several bits | |
57 | * which all are kept within that one column. Only exists for backwards | |
58 | * compatibility, see "column_formats" for more flexibility. | |
4a35548b | 59 | * |
72903e9d GS |
60 | * first_column: Specifies the number of the first column with logic data |
61 | * in simple multi-column mode. Only exists for backwards compatibility, | |
62 | * see "column_formats" for more flexibility. | |
4a35548b | 63 | * |
72903e9d GS |
64 | * logic_channels: Specifies the number of logic channels. Is required in |
65 | * simple single-column mode. Is optional in simple multi-column mode | |
66 | * (and defaults to all remaining columns). Only exists for backwards | |
67 | * compatibility, see "column_formats" for more flexibility. | |
4a35548b | 68 | * |
72903e9d GS |
69 | * single_format: Specifies the format of the input text in simple single- |
70 | * column mode. Available formats are: 'bin' (default), 'hex' and 'oct'. | |
71 | * Simple multi-column mode always uses single-bit data per column. | |
72 | * Only exists for backwards compatibility, see "column_formats" for | |
73 | * more flexibility. | |
4a35548b | 74 | * |
72903e9d GS |
75 | * start_line: Specifies at which line to start processing the input file. |
76 | * Allows to skip leading lines which neither are header nor data lines. | |
77 | * By default all of the input file gets processed. | |
4a35548b | 78 | * |
72903e9d GS |
79 | * header: Boolean option, controls whether the first processed line is used |
80 | * to determine channel names. Off by default. Generic channel names are | |
81 | * used in the absence of header line content. | |
4a35548b | 82 | * |
72903e9d GS |
83 | * samplerate: Specifies the samplerate of the input data. Defaults to 0. |
84 | * User specs take precedence over data which optionally gets derived | |
85 | * from input data. | |
4a35548b | 86 | * |
72903e9d GS |
87 | * column_separator: Specifies the sequence which separates the text file |
88 | * columns. Cannot be empty. Defaults to comma. | |
89 | * | |
90 | * comment_leader: Specifies the sequence which starts comments that run | |
91 | * up to the end of the current text line. Can be empty to disable | |
92 | * comment support. Defaults to semicolon. | |
93 | * | |
94 | * Typical examples of using these options: | |
95 | * - ... -I csv:column_formats=*l ... | |
96 | * All columns are single-bit logic data. Identical to the previous | |
97 | * multi-column mode (the default when no options were given at all). | |
98 | * - ... -I csv:column_formats=3-,*l ... | |
99 | * Ignore the first three columns, get single-bit logic data from all | |
100 | * remaining lines (multi-column mode with first-column above 1). | |
101 | * - ... -I csv:column_formats=3-,4l,x8 ... | |
102 | * Ignore the first three columns, get single-bit logic data from the | |
103 | * next four columns, then eight-bit data in hex format from the next | |
104 | * column. More columns may follow in the input text but won't get | |
105 | * processed. (Mix of previous multi-column as well as single-column | |
106 | * modes.) | |
107 | * - ... -I csv:column_formats=4x8,b16,5l ... | |
108 | * Get eight-bit data in hex format from the first four columns, then | |
109 | * sixteen-bit data in binary format, then five times single-bit data. | |
110 | * - ... -I csv:single_column=2:single_format=bin:logic_channels=8 ... | |
111 | * Get eight logic bits in binary format from column 2. (Simple | |
112 | * single-column mode, corresponds to the "-,b8" format.) | |
113 | * - ... -I csv:first_column=6:logic_channels=4 ... | |
114 | * Get four single-bit logic channels from columns 6 to 9 respectively. | |
115 | * (Simple multi-column mode, corresponds to the "5-,4b" format.) | |
116 | * - ... -I csv:start_line=20:header=yes:... | |
117 | * Skip the first 19 text lines. Use line 20 to derive channel names. | |
118 | * Data starts at line 21. | |
08eb955a GS |
119 | * - ... -I csv:column_formats=*a6 ... |
120 | * Each column contains an analog value with six significant digits | |
121 | * after the decimal period. | |
7e4e65bf GS |
122 | * - ... -I csv:column_formats=t,2a ... |
123 | * The first column contains timestamps, the next two columns contain | |
124 | * analog values. The capture's samplerate could get determined from | |
125 | * the timestamp values if not provided by the user by means of the | |
126 | * 'samplerate' option. This assumes a mere number in units of seconds, | |
127 | * and equidistant rows, there is no fancy support for textual unit | |
128 | * suffixes nor gaps in the stream of samples nor other non-linearity, | |
129 | * just '-' ignore the column if the format is not supported). | |
cb3b8051 GS |
130 | * |
131 | * IMPORTANT! Make sure the .format_match() logic matches the default | |
132 | * values for the input module's options. Ideally the format match test | |
133 | * shall pass for all input data that is supported by default. | |
4a35548b MS |
134 | */ |
135 | ||
ccff468b GS |
136 | /* |
137 | * TODO | |
138 | * | |
5a971176 GS |
139 | * - Add a test suite for input modules in general, and CSV in specific? |
140 | * Becomes more important with the multitude of options and their | |
141 | * interaction. Could cover edge cases (BOM presence, line termination | |
142 | * absence, etc) and auto-stuff as well (channel names, channel counts, | |
143 | * samplerates, etc). | |
ccff468b GS |
144 | */ |
145 | ||
ce3fe379 | 146 | typedef double csv_analog_t; |
43bdef26 | 147 | |
4a35548b | 148 | /* Single column formats. */ |
ad6a2bee | 149 | enum single_col_format { |
e53f32d2 GS |
150 | FORMAT_NONE, /* Ignore this column. */ |
151 | FORMAT_BIN, /* Bin digits for a set of bits (or just one bit). */ | |
152 | FORMAT_HEX, /* Hex digits for a set of bits. */ | |
153 | FORMAT_OCT, /* Oct digits for a set of bits. */ | |
43bdef26 | 154 | FORMAT_ANALOG, /* Floating point number for an analog channel. */ |
7e4e65bf | 155 | FORMAT_TIME, /* Timestamps. */ |
e53f32d2 GS |
156 | }; |
157 | ||
158 | static const char *col_format_text[] = { | |
159 | [FORMAT_NONE] = "unknown", | |
160 | [FORMAT_BIN] = "binary", | |
161 | [FORMAT_HEX] = "hexadecimal", | |
162 | [FORMAT_OCT] = "octal", | |
43bdef26 | 163 | [FORMAT_ANALOG] = "analog", |
7e4e65bf | 164 | [FORMAT_TIME] = "timestamp", |
e53f32d2 GS |
165 | }; |
166 | ||
1a920e33 GS |
167 | static const char col_format_char[] = { |
168 | [FORMAT_NONE] = '?', | |
169 | [FORMAT_BIN] = 'b', | |
170 | [FORMAT_HEX] = 'x', | |
171 | [FORMAT_OCT] = 'o', | |
43bdef26 | 172 | [FORMAT_ANALOG] = 'a', |
7e4e65bf | 173 | [FORMAT_TIME] = 't', |
1a920e33 GS |
174 | }; |
175 | ||
fc3b42e9 GS |
176 | static gboolean format_is_ignore(enum single_col_format fmt) |
177 | { | |
178 | return fmt == FORMAT_NONE; | |
179 | } | |
180 | ||
181 | static gboolean format_is_logic(enum single_col_format fmt) | |
182 | { | |
183 | return fmt >= FORMAT_BIN && fmt <= FORMAT_OCT; | |
184 | } | |
185 | ||
186 | static gboolean format_is_analog(enum single_col_format fmt) | |
187 | { | |
188 | return fmt == FORMAT_ANALOG; | |
189 | } | |
190 | ||
7e4e65bf GS |
191 | static gboolean format_is_timestamp(enum single_col_format fmt) |
192 | { | |
193 | return fmt == FORMAT_TIME; | |
194 | } | |
195 | ||
e53f32d2 GS |
196 | struct column_details { |
197 | size_t col_nr; | |
198 | enum single_col_format text_format; | |
199 | size_t channel_offset; | |
200 | size_t channel_count; | |
a267bf45 | 201 | int analog_digits; |
05719d75 | 202 | GString **channel_names; |
4a35548b MS |
203 | }; |
204 | ||
205 | struct context { | |
41d214f6 BV |
206 | gboolean started; |
207 | ||
7e4e65bf | 208 | /* Current samplerate, optionally determined from input data. */ |
4a35548b | 209 | uint64_t samplerate; |
811dcf7e | 210 | uint64_t calc_samplerate; |
7e4e65bf | 211 | double prev_timestamp; |
246aca5f | 212 | gboolean samplerate_sent; |
4a35548b | 213 | |
a267bf45 | 214 | /* Number of channels. */ |
836fac9c | 215 | size_t logic_channels; |
43bdef26 | 216 | size_t analog_channels; |
4a35548b | 217 | |
836fac9c | 218 | /* Column delimiter (actually separator), comment leader, EOL sequence. */ |
4a35548b | 219 | GString *delimiter; |
4a35548b | 220 | GString *comment; |
41d214f6 BV |
221 | char *termination; |
222 | ||
1a920e33 GS |
223 | /* Format specs for input columns, and processing state. */ |
224 | size_t column_seen_count; | |
225 | const char *column_formats; | |
e53f32d2 GS |
226 | size_t column_want_count; |
227 | struct column_details *column_details; | |
228 | ||
4a35548b | 229 | /* Line number to start processing. */ |
6433156c | 230 | size_t start_line; |
4a35548b MS |
231 | |
232 | /* | |
233 | * Determines if the first line should be treated as header and used for | |
ba7dd8bb | 234 | * channel names in multi column mode. |
4a35548b | 235 | */ |
de8fe3b5 GS |
236 | gboolean use_header; |
237 | gboolean header_seen; | |
4a35548b | 238 | |
cd59e6ec GS |
239 | size_t sample_unit_size; /**!< Byte count for a single sample. */ |
240 | uint8_t *sample_buffer; /**!< Buffer for a single sample. */ | |
43bdef26 | 241 | csv_analog_t *analog_sample_buffer; /**!< Buffer for one set of analog values. */ |
4a35548b | 242 | |
cd59e6ec GS |
243 | uint8_t *datafeed_buffer; /**!< Queue for datafeed submission. */ |
244 | size_t datafeed_buf_size; | |
245 | size_t datafeed_buf_fill; | |
43bdef26 GS |
246 | /* "Striped" layout, M samples for N channels each. */ |
247 | csv_analog_t *analog_datafeed_buffer; /**!< Queue for analog datafeed. */ | |
248 | size_t analog_datafeed_buf_size; | |
249 | size_t analog_datafeed_buf_fill; | |
a267bf45 | 250 | int *analog_datafeed_digits; |
05719d75 | 251 | GSList **analog_datafeed_channels; |
4a35548b | 252 | |
4a35548b | 253 | /* Current line number. */ |
6433156c | 254 | size_t line_number; |
affaf540 GS |
255 | |
256 | /* List of previously created sigrok channels. */ | |
257 | GSList *prev_sr_channels; | |
307b2393 | 258 | GSList **prev_df_channels; |
4a35548b MS |
259 | }; |
260 | ||
626c388a GS |
261 | /* |
262 | * Primitive operations to handle sample sets: | |
263 | * - Keep a buffer for datafeed submission, capable of holding many | |
264 | * samples (reduces call overhead, improves throughput). | |
265 | * - Have a "current sample set" pointer reference one position in that | |
266 | * large samples buffer. | |
267 | * - Clear the current sample set before text line inspection, then set | |
268 | * the bits which are found active in the current line of text input. | |
269 | * Phrase the API such that call sites can be kept simple. Advance to | |
270 | * the next sample set between lines, flush the larger buffer as needed | |
271 | * (when it is full, or upon EOF). | |
272 | */ | |
273 | ||
43bdef26 GS |
274 | static int flush_samplerate(const struct sr_input *in) |
275 | { | |
276 | struct context *inc; | |
43bdef26 GS |
277 | |
278 | inc = in->priv; | |
811dcf7e GS |
279 | if (!inc->calc_samplerate && inc->samplerate) |
280 | inc->calc_samplerate = inc->samplerate; | |
281 | if (inc->calc_samplerate && !inc->samplerate_sent) { | |
f8a8d4bb GS |
282 | (void)sr_session_send_meta(in->sdi, SR_CONF_SAMPLERATE, |
283 | g_variant_new_uint64(inc->calc_samplerate)); | |
43bdef26 GS |
284 | inc->samplerate_sent = TRUE; |
285 | } | |
286 | ||
287 | return SR_OK; | |
288 | } | |
289 | ||
626c388a GS |
290 | static void clear_logic_samples(struct context *inc) |
291 | { | |
43bdef26 GS |
292 | if (!inc->logic_channels) |
293 | return; | |
626c388a GS |
294 | inc->sample_buffer = &inc->datafeed_buffer[inc->datafeed_buf_fill]; |
295 | memset(inc->sample_buffer, 0, inc->sample_unit_size); | |
296 | } | |
297 | ||
298 | static void set_logic_level(struct context *inc, size_t ch_idx, int on) | |
299 | { | |
300 | size_t byte_idx, bit_idx; | |
301 | uint8_t bit_mask; | |
302 | ||
836fac9c | 303 | if (ch_idx >= inc->logic_channels) |
626c388a GS |
304 | return; |
305 | if (!on) | |
306 | return; | |
307 | ||
308 | byte_idx = ch_idx / 8; | |
309 | bit_idx = ch_idx % 8; | |
310 | bit_mask = 1 << bit_idx; | |
311 | inc->sample_buffer[byte_idx] |= bit_mask; | |
312 | } | |
313 | ||
314 | static int flush_logic_samples(const struct sr_input *in) | |
315 | { | |
316 | struct context *inc; | |
317 | struct sr_datafeed_packet packet; | |
318 | struct sr_datafeed_logic logic; | |
319 | int rc; | |
320 | ||
321 | inc = in->priv; | |
322 | if (!inc->datafeed_buf_fill) | |
323 | return SR_OK; | |
324 | ||
43bdef26 GS |
325 | rc = flush_samplerate(in); |
326 | if (rc != SR_OK) | |
327 | return rc; | |
246aca5f | 328 | |
626c388a GS |
329 | memset(&packet, 0, sizeof(packet)); |
330 | memset(&logic, 0, sizeof(logic)); | |
331 | packet.type = SR_DF_LOGIC; | |
332 | packet.payload = &logic; | |
333 | logic.unitsize = inc->sample_unit_size; | |
334 | logic.length = inc->datafeed_buf_fill; | |
335 | logic.data = inc->datafeed_buffer; | |
336 | ||
337 | rc = sr_session_send(in->sdi, &packet); | |
338 | if (rc != SR_OK) | |
339 | return rc; | |
340 | ||
341 | inc->datafeed_buf_fill = 0; | |
212769c3 | 342 | |
626c388a GS |
343 | return SR_OK; |
344 | } | |
345 | ||
346 | static int queue_logic_samples(const struct sr_input *in) | |
347 | { | |
348 | struct context *inc; | |
349 | int rc; | |
350 | ||
351 | inc = in->priv; | |
836fac9c GS |
352 | if (!inc->logic_channels) |
353 | return SR_OK; | |
626c388a GS |
354 | |
355 | inc->datafeed_buf_fill += inc->sample_unit_size; | |
356 | if (inc->datafeed_buf_fill == inc->datafeed_buf_size) { | |
357 | rc = flush_logic_samples(in); | |
358 | if (rc != SR_OK) | |
359 | return rc; | |
360 | } | |
212769c3 | 361 | |
626c388a GS |
362 | return SR_OK; |
363 | } | |
364 | ||
43bdef26 GS |
365 | static void set_analog_value(struct context *inc, size_t ch_idx, csv_analog_t value); |
366 | ||
367 | static void clear_analog_samples(struct context *inc) | |
368 | { | |
369 | size_t idx; | |
370 | ||
371 | if (!inc->analog_channels) | |
372 | return; | |
373 | inc->analog_sample_buffer = &inc->analog_datafeed_buffer[inc->analog_datafeed_buf_fill]; | |
374 | for (idx = 0; idx < inc->analog_channels; idx++) | |
375 | set_analog_value(inc, idx, 0.0); | |
376 | } | |
377 | ||
378 | static void set_analog_value(struct context *inc, size_t ch_idx, csv_analog_t value) | |
379 | { | |
380 | if (ch_idx >= inc->analog_channels) | |
381 | return; | |
382 | if (!value) | |
383 | return; | |
384 | inc->analog_sample_buffer[ch_idx * inc->analog_datafeed_buf_size] = value; | |
385 | } | |
386 | ||
387 | static int flush_analog_samples(const struct sr_input *in) | |
388 | { | |
43bdef26 GS |
389 | struct context *inc; |
390 | struct sr_datafeed_packet packet; | |
391 | struct sr_datafeed_analog analog; | |
392 | struct sr_analog_encoding encoding; | |
393 | struct sr_analog_meaning meaning; | |
394 | struct sr_analog_spec spec; | |
395 | csv_analog_t *samples; | |
396 | size_t ch_idx; | |
a267bf45 | 397 | int digits; |
43bdef26 GS |
398 | int rc; |
399 | ||
400 | inc = in->priv; | |
401 | if (!inc->analog_datafeed_buf_fill) | |
402 | return SR_OK; | |
403 | ||
404 | rc = flush_samplerate(in); | |
405 | if (rc != SR_OK) | |
406 | return rc; | |
407 | ||
408 | samples = inc->analog_datafeed_buffer; | |
409 | for (ch_idx = 0; ch_idx < inc->analog_channels; ch_idx++) { | |
a267bf45 | 410 | digits = inc->analog_datafeed_digits[ch_idx]; |
43bdef26 GS |
411 | sr_analog_init(&analog, &encoding, &meaning, &spec, digits); |
412 | memset(&packet, 0, sizeof(packet)); | |
413 | packet.type = SR_DF_ANALOG; | |
414 | packet.payload = &analog; | |
415 | analog.num_samples = inc->analog_datafeed_buf_fill; | |
416 | analog.data = samples; | |
417 | analog.meaning->channels = inc->analog_datafeed_channels[ch_idx]; | |
418 | analog.meaning->mq = 0; | |
419 | analog.meaning->mqflags = 0; | |
420 | analog.meaning->unit = 0; | |
421 | analog.encoding->unitsize = sizeof(samples[0]); | |
422 | analog.encoding->is_signed = TRUE; | |
423 | analog.encoding->is_float = TRUE; | |
424 | #ifdef WORDS_BIGENDIAN | |
425 | analog.encoding->is_bigendian = TRUE; | |
426 | #else | |
427 | analog.encoding->is_bigendian = FALSE; | |
428 | #endif | |
429 | analog.encoding->digits = spec.spec_digits; | |
430 | rc = sr_session_send(in->sdi, &packet); | |
431 | if (rc != SR_OK) | |
432 | return rc; | |
433 | samples += inc->analog_datafeed_buf_size; | |
434 | } | |
435 | ||
436 | inc->analog_datafeed_buf_fill = 0; | |
212769c3 | 437 | |
43bdef26 GS |
438 | return SR_OK; |
439 | } | |
440 | ||
441 | static int queue_analog_samples(const struct sr_input *in) | |
442 | { | |
443 | struct context *inc; | |
444 | int rc; | |
445 | ||
446 | inc = in->priv; | |
447 | if (!inc->analog_channels) | |
448 | return SR_OK; | |
449 | ||
450 | inc->analog_datafeed_buf_fill++; | |
451 | if (inc->analog_datafeed_buf_fill == inc->analog_datafeed_buf_size) { | |
452 | rc = flush_analog_samples(in); | |
453 | if (rc != SR_OK) | |
454 | return rc; | |
455 | } | |
212769c3 | 456 | |
43bdef26 GS |
457 | return SR_OK; |
458 | } | |
459 | ||
2142a79b GS |
460 | /* Helpers for "column processing". */ |
461 | ||
462 | static int split_column_format(const char *spec, | |
463 | size_t *column_count, enum single_col_format *format, size_t *bit_count) | |
464 | { | |
465 | size_t count; | |
466 | char *endp, format_char; | |
467 | enum single_col_format format_code; | |
468 | ||
469 | if (!spec || !*spec) | |
470 | return SR_ERR_ARG; | |
471 | ||
1a920e33 | 472 | /* Get the (optional, decimal, default 1) column count. Accept '*'. */ |
2142a79b | 473 | endp = NULL; |
1a920e33 | 474 | if (*spec == '*') { |
5ada72fc | 475 | /* Workaround, strtoul("*") won't always yield expected endp. */ |
1a920e33 GS |
476 | count = 0; |
477 | endp = (char *)&spec[1]; | |
478 | } else { | |
479 | count = strtoul(spec, &endp, 10); | |
480 | } | |
2142a79b GS |
481 | if (!endp) |
482 | return SR_ERR_ARG; | |
483 | if (endp == spec) | |
484 | count = 1; | |
485 | if (column_count) | |
486 | *column_count = count; | |
487 | spec = endp; | |
488 | ||
489 | /* Get the (mandatory, single letter) type spec (-/xob/l). */ | |
490 | format_char = *spec++; | |
491 | switch (format_char) { | |
5ada72fc | 492 | case '-': |
2142a79b GS |
493 | case '/': |
494 | format_char = '-'; | |
495 | format_code = FORMAT_NONE; | |
496 | break; | |
497 | case 'x': | |
498 | format_code = FORMAT_HEX; | |
499 | break; | |
500 | case 'o': | |
501 | format_code = FORMAT_OCT; | |
502 | break; | |
503 | case 'b': | |
504 | case 'l': | |
505 | format_code = FORMAT_BIN; | |
506 | break; | |
43bdef26 GS |
507 | case 'a': |
508 | format_code = FORMAT_ANALOG; | |
509 | break; | |
7e4e65bf GS |
510 | case 't': |
511 | format_code = FORMAT_TIME; | |
512 | break; | |
2142a79b GS |
513 | default: /* includes NUL */ |
514 | return SR_ERR_ARG; | |
515 | } | |
516 | if (format) | |
517 | *format = format_code; | |
518 | ||
519 | /* Get the (optional, decimal, default 1) bit count. */ | |
520 | endp = NULL; | |
521 | count = strtoul(spec, &endp, 10); | |
522 | if (!endp) | |
523 | return SR_ERR_ARG; | |
524 | if (endp == spec) | |
fc3b42e9 GS |
525 | count = format_is_analog(format_code) ? 3 : 1; |
526 | if (format_is_ignore(format_code)) | |
2142a79b GS |
527 | count = 0; |
528 | if (format_char == 'l') | |
529 | count = 1; | |
530 | if (bit_count) | |
531 | *bit_count = count; | |
532 | spec = endp; | |
533 | ||
534 | /* Input spec must have been exhausted. */ | |
535 | if (*spec) | |
536 | return SR_ERR_ARG; | |
537 | ||
538 | return SR_OK; | |
539 | } | |
540 | ||
9e7af34e GS |
541 | static int make_column_details_from_format(const struct sr_input *in, |
542 | const char *column_format, char **column_texts) | |
2142a79b | 543 | { |
9e7af34e | 544 | struct context *inc; |
2142a79b | 545 | char **formats, *format; |
43bdef26 | 546 | size_t format_count, column_count, logic_count, analog_count; |
1a920e33 | 547 | size_t auto_column_count; |
43bdef26 | 548 | size_t format_idx, c, b, column_idx, channel_idx, analog_idx; |
2142a79b GS |
549 | enum single_col_format f; |
550 | struct column_details *detail; | |
9e7af34e GS |
551 | GString *channel_name; |
552 | size_t create_idx; | |
553 | char *column; | |
554 | const char *caption; | |
43bdef26 | 555 | int channel_type, channel_sdi_nr; |
05719d75 | 556 | void *channel; |
2142a79b GS |
557 | int ret; |
558 | ||
9e7af34e GS |
559 | inc = in->priv; |
560 | inc->column_seen_count = g_strv_length(column_texts); | |
561 | ||
05719d75 | 562 | /* Split the input spec, count involved columns and channels. */ |
2142a79b GS |
563 | formats = g_strsplit(column_format, ",", 0); |
564 | if (!formats) { | |
565 | sr_err("Cannot parse columns format %s (comma split).", column_format); | |
566 | return SR_ERR_ARG; | |
567 | } | |
568 | format_count = g_strv_length(formats); | |
569 | if (!format_count) { | |
570 | sr_err("Cannot parse columns format %s (field count).", column_format); | |
571 | g_strfreev(formats); | |
572 | return SR_ERR_ARG; | |
573 | } | |
43bdef26 | 574 | column_count = logic_count = analog_count = 0; |
1a920e33 | 575 | auto_column_count = 0; |
2142a79b GS |
576 | for (format_idx = 0; format_idx < format_count; format_idx++) { |
577 | format = formats[format_idx]; | |
578 | ret = split_column_format(format, &c, &f, &b); | |
579 | sr_dbg("fmt %s -> %zu cols, %s fmt, %zu bits, rc %d", format, c, col_format_text[f], b, ret); | |
580 | if (ret != SR_OK) { | |
581 | sr_err("Cannot parse columns format %s (field split, %s).", column_format, format); | |
582 | g_strfreev(formats); | |
583 | return SR_ERR_ARG; | |
584 | } | |
1a920e33 GS |
585 | if (f && !c) { |
586 | /* User requested "auto-count", must be last format. */ | |
587 | if (formats[format_idx + 1]) { | |
588 | sr_err("Auto column count must be last format field."); | |
589 | g_strfreev(formats); | |
590 | return SR_ERR_ARG; | |
591 | } | |
592 | auto_column_count = inc->column_seen_count - column_count; | |
593 | c = auto_column_count; | |
594 | } | |
2142a79b | 595 | column_count += c; |
fc3b42e9 | 596 | if (format_is_analog(f)) |
43bdef26 | 597 | analog_count += c; |
fc3b42e9 | 598 | else if (format_is_logic(f)) |
43bdef26 | 599 | logic_count += c * b; |
2142a79b | 600 | } |
43bdef26 GS |
601 | sr_dbg("Column format %s -> %zu columns, %zu logic, %zu analog channels.", |
602 | column_format, column_count, logic_count, analog_count); | |
2142a79b | 603 | |
05719d75 | 604 | /* Allocate and fill in "column processing" details. */ |
2142a79b | 605 | inc->column_want_count = column_count; |
9e7af34e GS |
606 | if (inc->column_seen_count < inc->column_want_count) { |
607 | sr_err("Insufficient input text width for desired data amount, got %zu but want %zu columns.", | |
608 | inc->column_seen_count, inc->column_want_count); | |
609 | g_strfreev(formats); | |
610 | return SR_ERR_ARG; | |
611 | } | |
05719d75 GS |
612 | inc->logic_channels = logic_count; |
613 | inc->analog_channels = analog_count; | |
614 | inc->analog_datafeed_digits = g_malloc0(inc->analog_channels * sizeof(inc->analog_datafeed_digits[0])); | |
615 | inc->analog_datafeed_channels = g_malloc0(inc->analog_channels * sizeof(inc->analog_datafeed_channels[0])); | |
2142a79b | 616 | inc->column_details = g_malloc0_n(column_count, sizeof(inc->column_details[0])); |
43bdef26 | 617 | column_idx = channel_idx = analog_idx = 0; |
9e7af34e | 618 | channel_name = g_string_sized_new(64); |
2142a79b | 619 | for (format_idx = 0; format_idx < format_count; format_idx++) { |
9e7af34e | 620 | /* Process a format field, which can span multiple columns. */ |
2142a79b GS |
621 | format = formats[format_idx]; |
622 | (void)split_column_format(format, &c, &f, &b); | |
1a920e33 GS |
623 | if (f && !c) |
624 | c = auto_column_count; | |
2142a79b | 625 | while (c-- > 0) { |
9e7af34e | 626 | /* Fill in a column's processing details. */ |
2142a79b GS |
627 | detail = &inc->column_details[column_idx++]; |
628 | detail->col_nr = column_idx; | |
629 | detail->text_format = f; | |
fc3b42e9 | 630 | if (format_is_analog(detail->text_format)) { |
43bdef26 GS |
631 | detail->channel_offset = analog_idx; |
632 | detail->channel_count = 1; | |
a267bf45 | 633 | detail->analog_digits = b; |
43bdef26 | 634 | analog_idx += detail->channel_count; |
fc3b42e9 | 635 | } else if (format_is_logic(detail->text_format)) { |
2142a79b GS |
636 | detail->channel_offset = channel_idx; |
637 | detail->channel_count = b; | |
43bdef26 | 638 | channel_idx += detail->channel_count; |
fc3b42e9 GS |
639 | } else if (format_is_ignore(detail->text_format)) { |
640 | /* EMPTY */ | |
9e7af34e | 641 | continue; |
fc3b42e9 GS |
642 | } else { |
643 | /* | |
644 | * Neither logic nor analog data, nor ignore. | |
645 | * Format was noted. No channel creation involved. | |
646 | */ | |
647 | continue; | |
648 | } | |
9e7af34e | 649 | /* |
08eb955a | 650 | * Pick most appropriate channel names. Optionally |
9e7af34e GS |
651 | * use text from a header line (when requested by the |
652 | * user). In the absence of header text, channels are | |
653 | * assigned rather generic names. | |
654 | * | |
655 | * Manipulation of the column's caption (when a header | |
656 | * line is seen) is acceptable, because this header | |
657 | * line won't get processed another time. | |
658 | */ | |
659 | column = column_texts[detail->col_nr - 1]; | |
4b834894 GS |
660 | if (inc->use_header && column && *column) { |
661 | column = g_strstrip(column); | |
9e7af34e | 662 | caption = sr_scpi_unquote_string(column); |
4b834894 | 663 | } else { |
9e7af34e | 664 | caption = NULL; |
4b834894 | 665 | } |
9e7af34e GS |
666 | if (!caption || !*caption) |
667 | caption = NULL; | |
3f1f63f0 | 668 | /* |
05719d75 GS |
669 | * Collect channel creation details here, but defer |
670 | * actual creation of the channels such that all | |
671 | * logic channels can get created first and analog | |
672 | * channels only get created afterwards. | |
3f1f63f0 | 673 | */ |
05719d75 | 674 | detail->channel_names = g_malloc0(detail->channel_count * sizeof(detail->channel_names[0])); |
9e7af34e GS |
675 | for (create_idx = 0; create_idx < detail->channel_count; create_idx++) { |
676 | if (caption && detail->channel_count == 1) { | |
677 | g_string_assign(channel_name, caption); | |
678 | } else if (caption) { | |
679 | g_string_printf(channel_name, "%s[%zu]", | |
680 | caption, create_idx); | |
681 | } else { | |
682 | g_string_printf(channel_name, "%zu", | |
683 | detail->channel_offset + create_idx); | |
684 | } | |
05719d75 | 685 | detail->channel_names[create_idx] = g_string_new_len(channel_name->str, channel_name->len); |
9e7af34e | 686 | } |
2142a79b GS |
687 | } |
688 | } | |
9e7af34e | 689 | g_string_free(channel_name, TRUE); |
2142a79b GS |
690 | g_strfreev(formats); |
691 | ||
05719d75 GS |
692 | /* Create channels in strict logic to analog order. */ |
693 | channel_type = SR_CHANNEL_LOGIC; | |
694 | for (column_idx = 0; column_idx < inc->column_want_count; column_idx++) { | |
695 | detail = &inc->column_details[column_idx]; | |
696 | if (!format_is_logic(detail->text_format)) | |
697 | continue; | |
698 | for (create_idx = 0; create_idx < detail->channel_count; create_idx++) { | |
699 | caption = detail->channel_names[create_idx]->str; | |
700 | channel_sdi_nr = g_slist_length(in->sdi->channels); | |
701 | sr_channel_new(in->sdi, channel_sdi_nr, channel_type, TRUE, caption); | |
702 | } | |
703 | } | |
704 | channel_type = SR_CHANNEL_ANALOG; | |
705 | for (column_idx = 0; column_idx < inc->column_want_count; column_idx++) { | |
706 | detail = &inc->column_details[column_idx]; | |
707 | if (!format_is_analog(detail->text_format)) | |
708 | continue; | |
709 | caption = detail->channel_names[0]->str; | |
710 | channel_sdi_nr = g_slist_length(in->sdi->channels); | |
711 | channel = sr_channel_new(in->sdi, channel_sdi_nr, channel_type, TRUE, caption); | |
712 | channel_idx = channel_sdi_nr - inc->logic_channels; | |
713 | inc->analog_datafeed_digits[channel_idx] = detail->analog_digits; | |
714 | inc->analog_datafeed_channels[channel_idx] = g_slist_append(NULL, channel); | |
715 | } | |
716 | ||
2142a79b GS |
717 | return SR_OK; |
718 | } | |
719 | ||
e53f32d2 GS |
720 | static const struct column_details *lookup_column_details(struct context *inc, size_t nr) |
721 | { | |
722 | if (!inc || !inc->column_details) | |
723 | return NULL; | |
724 | if (!nr || nr > inc->column_want_count) | |
725 | return NULL; | |
212769c3 | 726 | |
e53f32d2 GS |
727 | return &inc->column_details[nr - 1]; |
728 | } | |
729 | ||
19267272 GS |
730 | /* |
731 | * Primitive operations for text input: Strip comments off text lines. | |
732 | * Split text lines into columns. Process input text for individual | |
733 | * columns. | |
734 | */ | |
735 | ||
41d214f6 | 736 | static void strip_comment(char *buf, const GString *prefix) |
4a35548b MS |
737 | { |
738 | char *ptr; | |
739 | ||
740 | if (!prefix->len) | |
741 | return; | |
742 | ||
b2c4dde2 | 743 | if ((ptr = strstr(buf, prefix->str))) { |
41d214f6 | 744 | *ptr = '\0'; |
b2c4dde2 GS |
745 | g_strstrip(buf); |
746 | } | |
4a35548b MS |
747 | } |
748 | ||
19267272 | 749 | /** |
51e60cde | 750 | * Splits a text line into a set of columns. |
19267272 | 751 | * |
e53f32d2 | 752 | * @param[in] buf The input text line to split. |
19267272 GS |
753 | * @param[in] inc The input module's context. |
754 | * | |
e53f32d2 | 755 | * @returns An array of strings, representing the columns' text. |
19267272 | 756 | * |
e53f32d2 | 757 | * This routine splits a text line on previously determined separators. |
19267272 | 758 | */ |
e53f32d2 | 759 | static char **split_line(char *buf, struct context *inc) |
4a35548b | 760 | { |
cd11e33c GS |
761 | char **fields, *f; |
762 | size_t l; | |
763 | ||
764 | fields = g_strsplit(buf, inc->delimiter->str, 0); | |
765 | if (!fields) | |
766 | return NULL; | |
767 | ||
768 | l = g_strv_length(fields); | |
769 | while (l--) { | |
770 | f = fields[l]; | |
771 | g_strchomp(f); | |
772 | } | |
773 | ||
774 | return fields; | |
4a35548b MS |
775 | } |
776 | ||
19267272 | 777 | /** |
51e60cde | 778 | * Parse a multi-bit field into several logic channels. |
19267272 | 779 | * |
e53f32d2 | 780 | * @param[in] column The input text, a run of bin/hex/oct digits. |
19267272 | 781 | * @param[in] inc The input module's context. |
836fac9c | 782 | * @param[in] details The column processing details. |
19267272 GS |
783 | * |
784 | * @retval SR_OK Success. | |
785 | * @retval SR_ERR Invalid input data (empty, or format error). | |
786 | * | |
787 | * This routine modifies the logic levels in the current sample set, | |
e53f32d2 | 788 | * based on the text input and a user provided format spec. |
19267272 | 789 | */ |
836fac9c GS |
790 | static int parse_logic(const char *column, struct context *inc, |
791 | const struct column_details *details) | |
4a35548b | 792 | { |
e53f32d2 GS |
793 | size_t length, ch_rem, ch_idx, ch_inc; |
794 | const char *rdptr; | |
4a35548b | 795 | char c; |
e53f32d2 GS |
796 | gboolean valid; |
797 | const char *type_text; | |
798 | uint8_t bits; | |
799 | ||
e53f32d2 GS |
800 | /* |
801 | * Prepare to read the digits from the text end towards the start. | |
802 | * A digit corresponds to a variable number of channels (depending | |
803 | * on the value's radix). Prepare the mapping of text digits to | |
804 | * (a number of) logic channels. | |
805 | */ | |
806 | length = strlen(column); | |
4a35548b | 807 | if (!length) { |
836fac9c | 808 | sr_err("Column %zu in line %zu is empty.", details->col_nr, |
41d214f6 | 809 | inc->line_number); |
4a35548b MS |
810 | return SR_ERR; |
811 | } | |
e53f32d2 | 812 | rdptr = &column[length]; |
836fac9c GS |
813 | ch_idx = details->channel_offset; |
814 | ch_rem = details->channel_count; | |
4a35548b | 815 | |
e53f32d2 GS |
816 | /* |
817 | * Get another digit and derive up to four logic channels' state from | |
818 | * it. Make sure to not process more bits than the column has channels | |
819 | * associated with it. | |
820 | */ | |
821 | while (rdptr > column && ch_rem) { | |
822 | /* Check for valid digits according to the input radix. */ | |
823 | c = *(--rdptr); | |
836fac9c | 824 | switch (details->text_format) { |
e53f32d2 GS |
825 | case FORMAT_BIN: |
826 | valid = g_ascii_isxdigit(c) && c < '2'; | |
827 | ch_inc = 1; | |
828 | break; | |
829 | case FORMAT_OCT: | |
830 | valid = g_ascii_isxdigit(c) && c < '8'; | |
831 | ch_inc = 3; | |
832 | break; | |
833 | case FORMAT_HEX: | |
834 | valid = g_ascii_isxdigit(c); | |
835 | ch_inc = 4; | |
836 | break; | |
837 | default: | |
838 | valid = FALSE; | |
839 | break; | |
4a35548b | 840 | } |
e53f32d2 | 841 | if (!valid) { |
836fac9c | 842 | type_text = col_format_text[details->text_format]; |
e53f32d2 | 843 | sr_err("Invalid text '%s' in %s type column %zu in line %zu.", |
836fac9c | 844 | column, type_text, details->col_nr, inc->line_number); |
4a35548b | 845 | return SR_ERR; |
e53f32d2 GS |
846 | } |
847 | /* Use the digit's bits for logic channels' data. */ | |
848 | bits = g_ascii_xdigit_value(c); | |
836fac9c | 849 | switch (details->text_format) { |
e53f32d2 GS |
850 | case FORMAT_HEX: |
851 | if (ch_rem >= 4) { | |
852 | ch_rem--; | |
853 | set_logic_level(inc, ch_idx + 3, bits & (1 << 3)); | |
854 | } | |
855 | /* FALLTHROUGH */ | |
856 | case FORMAT_OCT: | |
857 | if (ch_rem >= 3) { | |
858 | ch_rem--; | |
859 | set_logic_level(inc, ch_idx + 2, bits & (1 << 2)); | |
860 | } | |
861 | if (ch_rem >= 2) { | |
862 | ch_rem--; | |
863 | set_logic_level(inc, ch_idx + 1, bits & (1 << 1)); | |
864 | } | |
865 | /* FALLTHROUGH */ | |
866 | case FORMAT_BIN: | |
867 | ch_rem--; | |
868 | set_logic_level(inc, ch_idx + 0, bits & (1 << 0)); | |
869 | break; | |
fc3b42e9 | 870 | default: |
836fac9c GS |
871 | /* ShouldNotHappen(TM), but silences compiler warning. */ |
872 | return SR_ERR; | |
4a35548b | 873 | } |
e53f32d2 | 874 | ch_idx += ch_inc; |
4a35548b | 875 | } |
e53f32d2 GS |
876 | /* |
877 | * TODO Determine whether the availability of extra input data | |
878 | * for unhandled logic channels is worth warning here. In this | |
879 | * implementation users are in control, and can have the more | |
880 | * significant bits ignored (which can be considered a feature | |
881 | * and not really a limitation). | |
882 | */ | |
4a35548b MS |
883 | |
884 | return SR_OK; | |
885 | } | |
886 | ||
43bdef26 | 887 | /** |
51e60cde | 888 | * Parse a floating point text into an analog value. |
43bdef26 GS |
889 | * |
890 | * @param[in] column The input text, a floating point number. | |
891 | * @param[in] inc The input module's context. | |
892 | * @param[in] details The column processing details. | |
893 | * | |
894 | * @retval SR_OK Success. | |
895 | * @retval SR_ERR Invalid input data (empty, or format error). | |
896 | * | |
897 | * This routine modifies the analog values in the current sample set, | |
898 | * based on the text input and a user provided format spec. | |
899 | */ | |
900 | static int parse_analog(const char *column, struct context *inc, | |
901 | const struct column_details *details) | |
902 | { | |
903 | size_t length; | |
904 | double dvalue; float fvalue; | |
905 | csv_analog_t value; | |
906 | int ret; | |
907 | ||
fc3b42e9 | 908 | if (!format_is_analog(details->text_format)) |
43bdef26 GS |
909 | return SR_ERR_BUG; |
910 | ||
911 | length = strlen(column); | |
912 | if (!length) { | |
913 | sr_err("Column %zu in line %zu is empty.", details->col_nr, | |
914 | inc->line_number); | |
915 | return SR_ERR; | |
916 | } | |
917 | if (sizeof(value) == sizeof(double)) { | |
918 | ret = sr_atod_ascii(column, &dvalue); | |
919 | value = dvalue; | |
920 | } else if (sizeof(value) == sizeof(float)) { | |
921 | ret = sr_atof_ascii(column, &fvalue); | |
922 | value = fvalue; | |
923 | } else { | |
924 | ret = SR_ERR_BUG; | |
925 | } | |
926 | if (ret != SR_OK) { | |
927 | sr_err("Cannot parse analog text %s in column %zu in line %zu.", | |
928 | column, details->col_nr, inc->line_number); | |
929 | return SR_ERR_DATA; | |
930 | } | |
931 | set_analog_value(inc, details->channel_offset, value); | |
932 | ||
933 | return SR_OK; | |
934 | } | |
935 | ||
7e4e65bf | 936 | /** |
51e60cde | 937 | * Parse a timestamp text, auto-determine samplerate. |
7e4e65bf GS |
938 | * |
939 | * @param[in] column The input text, a floating point number. | |
940 | * @param[in] inc The input module's context. | |
941 | * @param[in] details The column processing details. | |
942 | * | |
943 | * @retval SR_OK Success. | |
944 | * @retval SR_ERR Invalid input data (empty, or format error). | |
945 | * | |
946 | * This routine attempts to automatically determine the input data's | |
947 | * samplerate from text rows' timestamp values. Only simple formats are | |
948 | * supported, user provided values always take precedence. | |
949 | */ | |
950 | static int parse_timestamp(const char *column, struct context *inc, | |
951 | const struct column_details *details) | |
952 | { | |
953 | double ts, rate; | |
954 | int ret; | |
955 | ||
956 | if (!format_is_timestamp(details->text_format)) | |
957 | return SR_ERR_BUG; | |
958 | ||
959 | /* | |
960 | * Implementor's notes on timestamp interpretation. Use a simple | |
961 | * approach for improved maintainability which covers most cases | |
962 | * of input data. There is not much gain in adding complexity, | |
963 | * users can easily provide the rate when auto-detection fails. | |
964 | * - Bail out if samplerate is known already. | |
965 | * - Try to interpret the timestamp (simple float conversion). | |
966 | * If conversion fails then clear all previous knowledge and | |
967 | * bail out (non-fatal, perhaps warn). Silently ignore values | |
968 | * of zero since those could be silent fails -- assume that | |
969 | * genuine data contains at least two adjacent rows with useful | |
970 | * timestamps for the feature to work reliably. Annoying users | |
971 | * with "failed to detect" messages is acceptable here, since | |
972 | * users expecting the feature to work should provide useful | |
973 | * data, and there are easy ways to disable the detection or | |
974 | * ignore the column. | |
975 | * - If there is no previous timestamp, keep the current value | |
976 | * for later reference and bail out. | |
977 | * - If a previous timestamp was seen, determine the difference | |
978 | * between them, and derive the samplerate. Update internal | |
979 | * state (the value automatically gets sent to the datafeed), | |
980 | * and clear previous knowledge. Subsequent calls will ignore | |
981 | * following input data (see above, rate is known). | |
982 | * | |
983 | * TODO Potential future improvements: | |
984 | * - Prefer rationals over floats for improved precision and | |
985 | * reduced rounding errors which result in odd rates. | |
986 | * - Support other formats ("2 ms" or similar)? | |
987 | */ | |
811dcf7e | 988 | if (inc->calc_samplerate) |
7e4e65bf GS |
989 | return SR_OK; |
990 | ret = sr_atod_ascii(column, &ts); | |
991 | if (ret != SR_OK) | |
992 | ts = 0.0; | |
993 | if (!ts) { | |
811dcf7e | 994 | sr_info("Cannot convert timestamp text %s in line %zu (or zero value).", |
7e4e65bf GS |
995 | column, inc->line_number); |
996 | inc->prev_timestamp = 0.0; | |
997 | return SR_OK; | |
998 | } | |
999 | if (!inc->prev_timestamp) { | |
1000 | sr_dbg("First timestamp value %g in line %zu.", | |
1001 | ts, inc->line_number); | |
1002 | inc->prev_timestamp = ts; | |
1003 | return SR_OK; | |
1004 | } | |
1005 | sr_dbg("Second timestamp value %g in line %zu.", ts, inc->line_number); | |
1006 | ts -= inc->prev_timestamp; | |
1007 | sr_dbg("Timestamp difference %g in line %zu.", | |
1008 | ts, inc->line_number); | |
1009 | if (!ts) { | |
1010 | sr_warn("Zero timestamp difference in line %zu.", | |
1011 | inc->line_number); | |
1012 | inc->prev_timestamp = ts; | |
1013 | return SR_OK; | |
1014 | } | |
1015 | rate = 1.0 / ts; | |
1016 | rate += 0.5; | |
1017 | rate = (uint64_t)rate; | |
1018 | sr_dbg("Rate from timestamp %g in line %zu.", rate, inc->line_number); | |
811dcf7e | 1019 | inc->calc_samplerate = rate; |
7e4e65bf GS |
1020 | inc->prev_timestamp = 0.0; |
1021 | ||
1022 | return SR_OK; | |
1023 | } | |
1024 | ||
836fac9c | 1025 | /** |
51e60cde | 1026 | * Parse routine which ignores the input text. |
836fac9c GS |
1027 | * |
1028 | * This routine exists to unify dispatch code paths, mapping input file | |
1029 | * columns' data types to their respective parse routines. | |
1030 | */ | |
1031 | static int parse_ignore(const char *column, struct context *inc, | |
1032 | const struct column_details *details) | |
1033 | { | |
1034 | (void)column; | |
1035 | (void)inc; | |
1036 | (void)details; | |
212769c3 | 1037 | |
836fac9c GS |
1038 | return SR_OK; |
1039 | } | |
1040 | ||
1041 | typedef int (*col_parse_cb)(const char *column, struct context *inc, | |
1042 | const struct column_details *details); | |
1043 | ||
1044 | static const col_parse_cb col_parse_funcs[] = { | |
1045 | [FORMAT_NONE] = parse_ignore, | |
1046 | [FORMAT_BIN] = parse_logic, | |
1047 | [FORMAT_OCT] = parse_logic, | |
1048 | [FORMAT_HEX] = parse_logic, | |
43bdef26 | 1049 | [FORMAT_ANALOG] = parse_analog, |
7e4e65bf | 1050 | [FORMAT_TIME] = parse_timestamp, |
836fac9c GS |
1051 | }; |
1052 | ||
cd7c5f96 GS |
1053 | /* |
1054 | * BEWARE! Implementor's notes. Sync with feature set and default option | |
1055 | * values required during maintenance of the input module implementation. | |
1056 | * | |
1057 | * When applications invoke .format_match() routines, trying automatic | |
1058 | * determination of an input file's format handler, then no options are | |
1059 | * in effect. Because specifying options requires selection of an input | |
1060 | * module to pass the options to, which obsoletes the format-match check. | |
1061 | * | |
1062 | * Which means that we only need to deal with the default format here, | |
1063 | * which happens to be the simple multi-column format without header | |
1064 | * lines or leading garbage. Which means that the check can be rather | |
1065 | * strict, resulting in high levels of confidence upon match, never | |
1066 | * "accidently" winning for unreadable or unsupported-by-default formats. | |
1067 | * | |
1068 | * This .format_match() logic only needs to become more involved when | |
1069 | * default option values change, or when automatic detection of column | |
1070 | * data types improves. Then the supported-by-default types of input | |
1071 | * data must be considered acceptable here in the format-match check | |
1072 | * as well. | |
1073 | * | |
1074 | * Notice that the format check cannot re-use regular processing logic | |
1075 | * when their implementation assumes proper input data and wll generate | |
1076 | * diagnostics for unexpected input data. Failure to match the format is | |
1077 | * non-fatal here, mismatch must remain silent. It's up to applications | |
1078 | * how large a chunk of data gets passed here (start of the file's | |
1079 | * content). But inspection of the first few hundred bytes will usually | |
1080 | * be GoodEnough(TM) for the format-match purpose. Notice that filenames | |
1081 | * need not necessarily be available to the format-match routine. | |
1082 | * | |
1083 | * This implementation errs on the safe side. Users can always select | |
1084 | * the CSV input module when automatic format detection fails. | |
1085 | */ | |
1086 | static int format_match(GHashTable *metadata, unsigned int *confidence) | |
1087 | { | |
1088 | const int match_confidence = 100; | |
1089 | const char *default_extension = ".csv"; | |
1090 | const char *line_termination = "\n"; | |
1091 | const char *comment_leader = ";"; | |
1092 | const char *column_separator = ","; | |
1093 | const char *binary_charset = "01"; | |
1094 | ||
1095 | const char *fn; | |
1096 | GString *buf; | |
1097 | size_t fn_len; | |
1098 | GString *tmpbuf; | |
1099 | gboolean status; | |
1100 | size_t line_idx, col_idx; | |
1101 | char *rdptr, **lines, *line; | |
1102 | char **cols, *col; | |
1103 | ||
1104 | /* Get the application provided input data properties. */ | |
1105 | fn = g_hash_table_lookup(metadata, GINT_TO_POINTER(SR_INPUT_META_FILENAME)); | |
1106 | buf = g_hash_table_lookup(metadata, GINT_TO_POINTER(SR_INPUT_META_HEADER)); | |
1107 | ||
1108 | /* Filenames are a strong hint. Use then when available. */ | |
1109 | if (fn && *fn && (fn_len = strlen(fn)) >= strlen(default_extension)) { | |
1110 | if (strcasecmp(&fn[fn_len - strlen(default_extension)], default_extension) == 0) { | |
1111 | *confidence = 10; | |
1112 | return SR_OK; | |
1113 | } | |
1114 | } | |
1115 | ||
1116 | /* | |
1117 | * Check file content for compatibility with the input module's | |
1118 | * default format. Which translates to: | |
1119 | * - Must be at least one text line worth of input data. Ignore | |
1120 | * incomplete lines at the end of the available buffer. | |
1121 | * - Must be LF terminated text lines, optional CR-LF sequence. | |
1122 | * (Drop CR-only for simplicity since that's rare and users | |
1123 | * can override the automatic detection.) | |
1124 | * - Strip comments and skip empty lines. | |
1125 | * - Data lines must be binary input (potentially multiple bits | |
1126 | * per column which then get ignored). Presence of comma is | |
1127 | * optional but then must be followed by another data column. | |
1128 | * - No other content is acceptable, there neither are ignored | |
1129 | * columns nor analog data nor timestamps in the default layout. | |
1130 | * (See the above "sync format match with default options" | |
1131 | * comment though during maintenance!) | |
1132 | * Run the check on a copy to not affect the caller's buffer. | |
1133 | */ | |
1134 | if (!buf || !buf->len || !buf->str || !*buf->str) | |
1135 | return SR_ERR; | |
1136 | rdptr = g_strstr_len(buf->str, buf->len, line_termination); | |
1137 | if (!rdptr) | |
1138 | return SR_ERR; | |
1139 | tmpbuf = g_string_new_len(buf->str, rdptr + 1 - buf->str); | |
1140 | tmpbuf->str[tmpbuf->len - 1] = '\0'; | |
1141 | status = TRUE; | |
1142 | *confidence = match_confidence; | |
1143 | lines = g_strsplit(tmpbuf->str, line_termination, 0); | |
1144 | for (line_idx = 0; status && (line = lines[line_idx]); line_idx++) { | |
1145 | rdptr = strstr(line, comment_leader); | |
1146 | if (rdptr) | |
1147 | *rdptr = '\0'; | |
1148 | line = g_strstrip(line); | |
1149 | if (!line || !*line) | |
1150 | continue; | |
1151 | cols = g_strsplit(line, column_separator, 0); | |
1152 | if (!cols) { | |
1153 | status = FALSE; | |
1154 | break; | |
1155 | } | |
1156 | for (col_idx = 0; status && (col = cols[col_idx]); col_idx++) { | |
1157 | if (strspn(col, binary_charset) != strlen(col)) { | |
1158 | status = FALSE; | |
1159 | break; | |
1160 | } | |
1161 | } | |
1162 | g_strfreev(cols); | |
1163 | } | |
1164 | g_strfreev(lines); | |
1165 | g_string_free(tmpbuf, TRUE); | |
1166 | ||
1167 | if (!status) | |
1168 | return SR_ERR; | |
212769c3 | 1169 | |
cd7c5f96 GS |
1170 | return SR_OK; |
1171 | } | |
1172 | ||
41d214f6 | 1173 | static int init(struct sr_input *in, GHashTable *options) |
4a35548b | 1174 | { |
41d214f6 | 1175 | struct context *inc; |
1a920e33 | 1176 | size_t single_column, first_column, logic_channels; |
41d214f6 | 1177 | const char *s; |
836fac9c | 1178 | enum single_col_format format; |
1a920e33 | 1179 | char format_char; |
4a35548b | 1180 | |
836fac9c GS |
1181 | in->sdi = g_malloc0(sizeof(*in->sdi)); |
1182 | in->priv = inc = g_malloc0(sizeof(*inc)); | |
4a35548b | 1183 | |
72903e9d | 1184 | single_column = g_variant_get_uint32(g_hash_table_lookup(options, "single_column")); |
72903e9d | 1185 | logic_channels = g_variant_get_uint32(g_hash_table_lookup(options, "logic_channels")); |
41d214f6 | 1186 | inc->delimiter = g_string_new(g_variant_get_string( |
72903e9d | 1187 | g_hash_table_lookup(options, "column_separator"), NULL)); |
836fac9c | 1188 | if (!inc->delimiter->len) { |
72903e9d | 1189 | sr_err("Column separator cannot be empty."); |
41d214f6 | 1190 | return SR_ERR_ARG; |
4a35548b | 1191 | } |
72903e9d | 1192 | s = g_variant_get_string(g_hash_table_lookup(options, "single_format"), NULL); |
836fac9c GS |
1193 | if (g_ascii_strncasecmp(s, "bin", 3) == 0) { |
1194 | format = FORMAT_BIN; | |
1195 | } else if (g_ascii_strncasecmp(s, "hex", 3) == 0) { | |
1196 | format = FORMAT_HEX; | |
1197 | } else if (g_ascii_strncasecmp(s, "oct", 3) == 0) { | |
1198 | format = FORMAT_OCT; | |
41d214f6 | 1199 | } else { |
72903e9d | 1200 | sr_err("Invalid single-column format: '%s'", s); |
41d214f6 | 1201 | return SR_ERR_ARG; |
4a35548b | 1202 | } |
41d214f6 | 1203 | inc->comment = g_string_new(g_variant_get_string( |
72903e9d | 1204 | g_hash_table_lookup(options, "comment_leader"), NULL)); |
41d214f6 | 1205 | if (g_string_equal(inc->comment, inc->delimiter)) { |
e53f32d2 GS |
1206 | /* |
1207 | * Using the same sequence as comment leader and column | |
72903e9d GS |
1208 | * separator won't work. The user probably specified ';' |
1209 | * as the column separator but did not adjust the comment | |
e53f32d2 GS |
1210 | * leader. Try DWIM, drop comment strippin support here. |
1211 | */ | |
72903e9d | 1212 | sr_warn("Comment leader and column separator conflict, disabling comment support."); |
41d214f6 | 1213 | g_string_truncate(inc->comment, 0); |
4a35548b | 1214 | } |
6e8d95a5 | 1215 | inc->samplerate = g_variant_get_uint64(g_hash_table_lookup(options, "samplerate")); |
72903e9d | 1216 | first_column = g_variant_get_uint32(g_hash_table_lookup(options, "first_column")); |
de8fe3b5 | 1217 | inc->use_header = g_variant_get_boolean(g_hash_table_lookup(options, "header")); |
72903e9d | 1218 | inc->start_line = g_variant_get_uint32(g_hash_table_lookup(options, "start_line")); |
41d214f6 | 1219 | if (inc->start_line < 1) { |
6433156c | 1220 | sr_err("Invalid start line %zu.", inc->start_line); |
41d214f6 | 1221 | return SR_ERR_ARG; |
4a35548b MS |
1222 | } |
1223 | ||
e53f32d2 | 1224 | /* |
1a920e33 GS |
1225 | * Scan flexible, to get prefered format specs which describe |
1226 | * the input file's data formats. As well as some simple specs | |
1227 | * for backwards compatibility and user convenience. | |
1228 | * | |
1229 | * This logic ends up with a copy of the format string, either | |
1230 | * user provided or internally derived. Actual creation of the | |
1231 | * column processing details gets deferred until the first line | |
1232 | * of input data was seen. To support automatic determination of | |
1233 | * e.g. channel counts from column counts. | |
e53f32d2 | 1234 | */ |
72903e9d | 1235 | s = g_variant_get_string(g_hash_table_lookup(options, "column_formats"), NULL); |
2142a79b | 1236 | if (s && *s) { |
1a920e33 | 1237 | inc->column_formats = g_strdup(s); |
72903e9d | 1238 | sr_dbg("User specified column_formats: %s.", s); |
1a920e33 GS |
1239 | } else if (single_column && logic_channels) { |
1240 | format_char = col_format_char[format]; | |
1241 | if (single_column == 1) { | |
1242 | inc->column_formats = g_strdup_printf("%c%zu", | |
1243 | format_char, logic_channels); | |
e53f32d2 | 1244 | } else { |
1a920e33 GS |
1245 | inc->column_formats = g_strdup_printf("%zu-,%c%zu", |
1246 | single_column - 1, | |
1247 | format_char, logic_channels); | |
e53f32d2 | 1248 | } |
72903e9d | 1249 | sr_dbg("Backwards compat single_column, col %zu, fmt %s, bits %zu -> %s.", |
1a920e33 GS |
1250 | single_column, col_format_text[format], logic_channels, |
1251 | inc->column_formats); | |
1252 | } else if (!single_column) { | |
1253 | if (first_column > 1) { | |
1254 | inc->column_formats = g_strdup_printf("%zu-,%zul", | |
1255 | first_column - 1, logic_channels); | |
1256 | } else { | |
1257 | inc->column_formats = g_strdup_printf("%zul", | |
1258 | logic_channels); | |
1259 | } | |
1260 | sr_dbg("Backwards compat multi-column, col %zu, chans %zu -> %s.", | |
1261 | first_column, logic_channels, | |
1262 | inc->column_formats); | |
e53f32d2 | 1263 | } else { |
72903e9d | 1264 | sr_warn("Unknown or unsupported columns layout spec, assuming simple multi-column mode."); |
1a920e33 | 1265 | inc->column_formats = g_strdup("*l"); |
4a35548b MS |
1266 | } |
1267 | ||
41d214f6 BV |
1268 | return SR_OK; |
1269 | } | |
4a35548b | 1270 | |
affaf540 GS |
1271 | /* |
1272 | * Check the channel list for consistency across file re-import. See | |
1273 | * the VCD input module for more details and motivation. | |
1274 | */ | |
307b2393 GS |
1275 | static void release_df_channels(struct context *inc, GSList **l) |
1276 | { | |
1277 | size_t idx; | |
1278 | ||
1279 | if (!inc->analog_channels || !l) | |
1280 | return; | |
1281 | for (idx = 0; idx < inc->analog_channels; idx++) | |
1282 | g_slist_free(l[idx]); | |
1283 | g_free(l); | |
1284 | } | |
affaf540 GS |
1285 | |
1286 | static void keep_header_for_reread(const struct sr_input *in) | |
1287 | { | |
1288 | struct context *inc; | |
1289 | ||
1290 | inc = in->priv; | |
307b2393 | 1291 | |
affaf540 GS |
1292 | g_slist_free_full(inc->prev_sr_channels, sr_channel_free_cb); |
1293 | inc->prev_sr_channels = in->sdi->channels; | |
1294 | in->sdi->channels = NULL; | |
307b2393 GS |
1295 | |
1296 | release_df_channels(inc, inc->prev_df_channels); | |
1297 | inc->prev_df_channels = inc->analog_datafeed_channels; | |
1298 | inc->analog_datafeed_channels = NULL; | |
affaf540 GS |
1299 | } |
1300 | ||
1301 | static int check_header_in_reread(const struct sr_input *in) | |
1302 | { | |
1303 | struct context *inc; | |
1304 | ||
1305 | if (!in) | |
1306 | return FALSE; | |
1307 | inc = in->priv; | |
1308 | if (!inc) | |
1309 | return FALSE; | |
1310 | if (!inc->prev_sr_channels) | |
1311 | return TRUE; | |
1312 | ||
1313 | if (sr_channel_lists_differ(inc->prev_sr_channels, in->sdi->channels)) { | |
1314 | sr_err("Channel list change not supported for file re-read."); | |
1315 | return FALSE; | |
1316 | } | |
307b2393 | 1317 | |
affaf540 GS |
1318 | g_slist_free_full(in->sdi->channels, sr_channel_free_cb); |
1319 | in->sdi->channels = inc->prev_sr_channels; | |
1320 | inc->prev_sr_channels = NULL; | |
1321 | ||
307b2393 GS |
1322 | release_df_channels(inc, inc->analog_datafeed_channels); |
1323 | inc->analog_datafeed_channels = inc->prev_df_channels; | |
1324 | inc->prev_df_channels = NULL; | |
1325 | ||
affaf540 GS |
1326 | return TRUE; |
1327 | } | |
1328 | ||
492dfa90 GS |
1329 | static const char *delim_set = "\r\n"; |
1330 | ||
329733d9 | 1331 | static const char *get_line_termination(GString *buf) |
41d214f6 | 1332 | { |
329733d9 | 1333 | const char *term; |
4a35548b | 1334 | |
41d214f6 BV |
1335 | term = NULL; |
1336 | if (g_strstr_len(buf->str, buf->len, "\r\n")) | |
1337 | term = "\r\n"; | |
1338 | else if (memchr(buf->str, '\n', buf->len)) | |
1339 | term = "\n"; | |
1340 | else if (memchr(buf->str, '\r', buf->len)) | |
1341 | term = "\r"; | |
4a35548b | 1342 | |
41d214f6 BV |
1343 | return term; |
1344 | } | |
4a35548b | 1345 | |
41d214f6 BV |
1346 | static int initial_parse(const struct sr_input *in, GString *buf) |
1347 | { | |
1348 | struct context *inc; | |
9e7af34e | 1349 | size_t num_columns; |
3f1f63f0 | 1350 | size_t line_number, line_idx; |
41d214f6 | 1351 | int ret; |
9e7af34e | 1352 | char **lines, *line, **columns; |
41d214f6 BV |
1353 | |
1354 | ret = SR_OK; | |
1355 | inc = in->priv; | |
1356 | columns = NULL; | |
1357 | ||
9e7af34e | 1358 | /* Search for the first line to process (header or data). */ |
41d214f6 | 1359 | line_number = 0; |
ef0b9935 GS |
1360 | if (inc->termination) |
1361 | lines = g_strsplit(buf->str, inc->termination, 0); | |
1362 | else | |
1363 | lines = g_strsplit_set(buf->str, delim_set, 0); | |
e53f32d2 | 1364 | for (line_idx = 0; (line = lines[line_idx]); line_idx++) { |
41d214f6 BV |
1365 | line_number++; |
1366 | if (inc->start_line > line_number) { | |
e53f32d2 | 1367 | sr_spew("Line %zu skipped (before start).", line_number); |
4a35548b MS |
1368 | continue; |
1369 | } | |
df0db9fd | 1370 | if (line[0] == '\0') { |
41d214f6 BV |
1371 | sr_spew("Blank line %zu skipped.", line_number); |
1372 | continue; | |
1373 | } | |
df0db9fd GS |
1374 | strip_comment(line, inc->comment); |
1375 | if (line[0] == '\0') { | |
41d214f6 | 1376 | sr_spew("Comment-only line %zu skipped.", line_number); |
4a35548b MS |
1377 | continue; |
1378 | } | |
1379 | ||
41d214f6 BV |
1380 | /* Reached first proper line. */ |
1381 | break; | |
1382 | } | |
e53f32d2 | 1383 | if (!line) { |
41d214f6 | 1384 | /* Not enough data for a proper line yet. */ |
60107497 | 1385 | ret = SR_ERR_NA; |
41d214f6 | 1386 | goto out; |
4a35548b MS |
1387 | } |
1388 | ||
9e7af34e | 1389 | /* Get the number of columns in the line. */ |
e53f32d2 | 1390 | columns = split_line(line, inc); |
df0db9fd | 1391 | if (!columns) { |
41d214f6 BV |
1392 | sr_err("Error while parsing line %zu.", line_number); |
1393 | ret = SR_ERR; | |
1394 | goto out; | |
4a35548b | 1395 | } |
4a35548b | 1396 | num_columns = g_strv_length(columns); |
4a35548b | 1397 | if (!num_columns) { |
e53f32d2 | 1398 | sr_err("Error while parsing line %zu.", line_number); |
41d214f6 BV |
1399 | ret = SR_ERR; |
1400 | goto out; | |
4a35548b | 1401 | } |
51e60cde | 1402 | sr_dbg("Got %zu columns in text line: %s.", num_columns, line); |
e53f32d2 | 1403 | |
1a920e33 | 1404 | /* |
9e7af34e GS |
1405 | * Interpret the user provided column format specs. This might |
1406 | * involve inspection of the now received input text, to support | |
1407 | * e.g. automatic detection of channel counts in the absence of | |
1408 | * user provided specs. Optionally a header line is used to get | |
1409 | * channels' names. | |
1410 | * | |
1411 | * Check the then created channels for consistency across .reset | |
1412 | * and .receive sequences (file re-load). | |
1a920e33 | 1413 | */ |
9e7af34e | 1414 | ret = make_column_details_from_format(in, inc->column_formats, columns); |
1a920e33 GS |
1415 | if (ret != SR_OK) { |
1416 | sr_err("Cannot parse columns format using line %zu.", line_number); | |
1417 | goto out; | |
4a35548b | 1418 | } |
affaf540 GS |
1419 | if (!check_header_in_reread(in)) { |
1420 | ret = SR_ERR_DATA; | |
1421 | goto out; | |
1422 | } | |
4a35548b MS |
1423 | |
1424 | /* | |
9e7af34e | 1425 | * Allocate buffer memory for datafeed submission of sample data. |
cd59e6ec GS |
1426 | * Calculate the minimum buffer size to store the set of samples |
1427 | * of all channels (unit size). Determine a larger buffer size | |
1428 | * for datafeed submission that is a multiple of the unit size. | |
626c388a GS |
1429 | * Allocate the larger buffer, the "sample buffer" will point |
1430 | * to a location within that large buffer later. | |
fc3b42e9 GS |
1431 | * |
1432 | * TODO Move channel creation here, and just store required | |
1433 | * parameters in the format parser above? Could simplify the | |
1434 | * arrangement that logic and analog channels get created in | |
1435 | * strict sequence in their respective groups. | |
4a35548b | 1436 | */ |
43bdef26 GS |
1437 | if (inc->logic_channels) { |
1438 | inc->sample_unit_size = (inc->logic_channels + 7) / 8; | |
1439 | inc->datafeed_buf_size = CHUNK_SIZE; | |
1440 | inc->datafeed_buf_size *= inc->sample_unit_size; | |
1441 | inc->datafeed_buffer = g_malloc(inc->datafeed_buf_size); | |
1442 | if (!inc->datafeed_buffer) { | |
1443 | sr_err("Cannot allocate datafeed send buffer (logic)."); | |
1444 | ret = SR_ERR_MALLOC; | |
1445 | goto out; | |
1446 | } | |
1447 | inc->datafeed_buf_fill = 0; | |
1448 | } | |
1449 | ||
1450 | if (inc->analog_channels) { | |
1451 | size_t sample_size, sample_count; | |
1452 | sample_size = sizeof(inc->analog_datafeed_buffer[0]); | |
1453 | inc->analog_datafeed_buf_size = CHUNK_SIZE; | |
1454 | inc->analog_datafeed_buf_size /= sample_size; | |
1455 | inc->analog_datafeed_buf_size /= inc->analog_channels; | |
1456 | sample_count = inc->analog_channels * inc->analog_datafeed_buf_size; | |
1457 | inc->analog_datafeed_buffer = g_malloc0(sample_count * sample_size); | |
1458 | if (!inc->analog_datafeed_buffer) { | |
1459 | sr_err("Cannot allocate datafeed send buffer (analog)."); | |
1460 | ret = SR_ERR_MALLOC; | |
1461 | goto out; | |
1462 | } | |
1463 | inc->analog_datafeed_buf_fill = 0; | |
43bdef26 | 1464 | } |
4a35548b | 1465 | |
41d214f6 BV |
1466 | out: |
1467 | if (columns) | |
1468 | g_strfreev(columns); | |
1469 | g_strfreev(lines); | |
4a35548b | 1470 | |
41d214f6 | 1471 | return ret; |
4a35548b MS |
1472 | } |
1473 | ||
4439363a GS |
1474 | /* |
1475 | * Gets called from initial_receive(), which runs until the end-of-line | |
1476 | * encoding of the input stream could get determined. Assumes that this | |
1477 | * routine receives enough buffered initial input data to either see the | |
1478 | * BOM when there is one, or that no BOM will follow when a text line | |
1479 | * termination sequence was seen. Silently drops the UTF-8 BOM sequence | |
1480 | * from the input buffer if one was seen. Does not care to protect | |
1481 | * against multiple execution or dropping the BOM multiple times -- | |
1482 | * there should be at most one in the input stream. | |
1483 | */ | |
1484 | static void initial_bom_check(const struct sr_input *in) | |
1485 | { | |
1486 | static const char *utf8_bom = "\xef\xbb\xbf"; | |
1487 | ||
1488 | if (in->buf->len < strlen(utf8_bom)) | |
1489 | return; | |
1490 | if (strncmp(in->buf->str, utf8_bom, strlen(utf8_bom)) != 0) | |
1491 | return; | |
1492 | g_string_erase(in->buf, 0, strlen(utf8_bom)); | |
1493 | } | |
1494 | ||
41d214f6 | 1495 | static int initial_receive(const struct sr_input *in) |
4a35548b | 1496 | { |
41d214f6 BV |
1497 | struct context *inc; |
1498 | GString *new_buf; | |
1499 | int len, ret; | |
329733d9 UH |
1500 | char *p; |
1501 | const char *termination; | |
4a35548b | 1502 | |
4439363a GS |
1503 | initial_bom_check(in); |
1504 | ||
41d214f6 | 1505 | inc = in->priv; |
4a35548b | 1506 | |
df0db9fd GS |
1507 | termination = get_line_termination(in->buf); |
1508 | if (!termination) | |
41d214f6 | 1509 | /* Don't have a full line yet. */ |
d0181813 | 1510 | return SR_ERR_NA; |
4a35548b | 1511 | |
df0db9fd GS |
1512 | p = g_strrstr_len(in->buf->str, in->buf->len, termination); |
1513 | if (!p) | |
41d214f6 | 1514 | /* Don't have a full line yet. */ |
d0181813 | 1515 | return SR_ERR_NA; |
41d214f6 BV |
1516 | len = p - in->buf->str - 1; |
1517 | new_buf = g_string_new_len(in->buf->str, len); | |
1518 | g_string_append_c(new_buf, '\0'); | |
4a35548b | 1519 | |
41d214f6 BV |
1520 | inc->termination = g_strdup(termination); |
1521 | ||
1522 | if (in->buf->str[0] != '\0') | |
1523 | ret = initial_parse(in, new_buf); | |
1524 | else | |
1525 | ret = SR_OK; | |
1526 | ||
1527 | g_string_free(new_buf, TRUE); | |
1528 | ||
1529 | return ret; | |
1530 | } | |
1531 | ||
7f4c3a62 | 1532 | static int process_buffer(struct sr_input *in, gboolean is_eof) |
41d214f6 | 1533 | { |
41d214f6 BV |
1534 | struct context *inc; |
1535 | gsize num_columns; | |
e53f32d2 | 1536 | size_t line_idx, col_idx, col_nr; |
836fac9c GS |
1537 | const struct column_details *details; |
1538 | col_parse_cb parse_func; | |
ad6a2bee | 1539 | int ret; |
fbefa03f GS |
1540 | char *processed_up_to; |
1541 | char **lines, *line, **columns, *column; | |
41d214f6 | 1542 | |
41d214f6 | 1543 | inc = in->priv; |
d0181813 | 1544 | if (!inc->started) { |
bee2b016 | 1545 | std_session_send_df_header(in->sdi); |
d0181813 | 1546 | inc->started = TRUE; |
4a35548b MS |
1547 | } |
1548 | ||
4555d3bd GS |
1549 | /* |
1550 | * Consider empty input non-fatal. Keep accumulating input until | |
1551 | * at least one full text line has become available. Grab the | |
1552 | * maximum amount of accumulated data that consists of full text | |
1553 | * lines, and process what has been received so far, leaving not | |
1554 | * yet complete lines for the next invocation. | |
7f4c3a62 GS |
1555 | * |
1556 | * Enforce that all previously buffered data gets processed in | |
1557 | * the "EOF" condition. Do not insist in the presence of the | |
1558 | * termination sequence for the last line (may often be missing | |
1559 | * on Windows). A present termination sequence will just result | |
1560 | * in the "execution of an empty line", and does not harm. | |
4555d3bd GS |
1561 | */ |
1562 | if (!in->buf->len) | |
1563 | return SR_OK; | |
7f4c3a62 | 1564 | if (is_eof) { |
fbefa03f | 1565 | processed_up_to = in->buf->str + in->buf->len; |
7f4c3a62 | 1566 | } else { |
fbefa03f GS |
1567 | processed_up_to = g_strrstr_len(in->buf->str, in->buf->len, |
1568 | inc->termination); | |
1569 | if (!processed_up_to) | |
1570 | return SR_OK; | |
1571 | *processed_up_to = '\0'; | |
1572 | processed_up_to += strlen(inc->termination); | |
7f4c3a62 | 1573 | } |
4a35548b | 1574 | |
fbefa03f | 1575 | /* Split input text lines and process their columns. */ |
18078d05 | 1576 | ret = SR_OK; |
ef0b9935 | 1577 | lines = g_strsplit(in->buf->str, inc->termination, 0); |
e53f32d2 | 1578 | for (line_idx = 0; (line = lines[line_idx]); line_idx++) { |
41d214f6 | 1579 | inc->line_number++; |
ef0b9935 GS |
1580 | if (inc->line_number < inc->start_line) { |
1581 | sr_spew("Line %zu skipped (before start).", inc->line_number); | |
1582 | continue; | |
1583 | } | |
df0db9fd | 1584 | if (line[0] == '\0') { |
41d214f6 | 1585 | sr_spew("Blank line %zu skipped.", inc->line_number); |
4a35548b MS |
1586 | continue; |
1587 | } | |
1588 | ||
1589 | /* Remove trailing comment. */ | |
df0db9fd GS |
1590 | strip_comment(line, inc->comment); |
1591 | if (line[0] == '\0') { | |
41d214f6 | 1592 | sr_spew("Comment-only line %zu skipped.", inc->line_number); |
4a35548b MS |
1593 | continue; |
1594 | } | |
1595 | ||
160691b9 | 1596 | /* Skip the header line, its content was used as the channel names. */ |
de8fe3b5 | 1597 | if (inc->use_header && !inc->header_seen) { |
160691b9 | 1598 | sr_spew("Header line %zu skipped.", inc->line_number); |
de8fe3b5 | 1599 | inc->header_seen = TRUE; |
160691b9 JS |
1600 | continue; |
1601 | } | |
1602 | ||
e53f32d2 GS |
1603 | /* Split the line into columns, check for minimum length. */ |
1604 | columns = split_line(line, inc); | |
df0db9fd | 1605 | if (!columns) { |
41d214f6 | 1606 | sr_err("Error while parsing line %zu.", inc->line_number); |
2355d229 | 1607 | g_strfreev(lines); |
4a35548b MS |
1608 | return SR_ERR; |
1609 | } | |
4a35548b | 1610 | num_columns = g_strv_length(columns); |
e53f32d2 GS |
1611 | if (num_columns < inc->column_want_count) { |
1612 | sr_err("Insufficient column count %zu in line %zu.", | |
1613 | num_columns, inc->line_number); | |
4a35548b | 1614 | g_strfreev(columns); |
2355d229 | 1615 | g_strfreev(lines); |
4a35548b MS |
1616 | return SR_ERR; |
1617 | } | |
1618 | ||
836fac9c | 1619 | /* Have the columns of the current text line processed. */ |
626c388a | 1620 | clear_logic_samples(inc); |
43bdef26 | 1621 | clear_analog_samples(inc); |
e53f32d2 GS |
1622 | for (col_idx = 0; col_idx < inc->column_want_count; col_idx++) { |
1623 | column = columns[col_idx]; | |
1624 | col_nr = col_idx + 1; | |
836fac9c GS |
1625 | details = lookup_column_details(inc, col_nr); |
1626 | if (!details || !details->text_format) | |
1627 | continue; | |
1628 | parse_func = col_parse_funcs[details->text_format]; | |
1629 | if (!parse_func) | |
1630 | continue; | |
1631 | ret = parse_func(column, inc, details); | |
e53f32d2 GS |
1632 | if (ret != SR_OK) { |
1633 | g_strfreev(columns); | |
1634 | g_strfreev(lines); | |
1635 | return SR_ERR; | |
1636 | } | |
4a35548b MS |
1637 | } |
1638 | ||
626c388a GS |
1639 | /* Send sample data to the session bus (buffered). */ |
1640 | ret = queue_logic_samples(in); | |
43bdef26 | 1641 | ret += queue_analog_samples(in); |
41d214f6 | 1642 | if (ret != SR_OK) { |
4a35548b | 1643 | sr_err("Sending samples failed."); |
cd59e6ec | 1644 | g_strfreev(columns); |
2355d229 | 1645 | g_strfreev(lines); |
4a35548b MS |
1646 | return SR_ERR; |
1647 | } | |
cd59e6ec | 1648 | |
41d214f6 BV |
1649 | g_strfreev(columns); |
1650 | } | |
1651 | g_strfreev(lines); | |
fbefa03f | 1652 | g_string_erase(in->buf, 0, processed_up_to - in->buf->str); |
41d214f6 | 1653 | |
7066fd46 | 1654 | return ret; |
41d214f6 BV |
1655 | } |
1656 | ||
7066fd46 | 1657 | static int receive(struct sr_input *in, GString *buf) |
41d214f6 BV |
1658 | { |
1659 | struct context *inc; | |
7066fd46 BV |
1660 | int ret; |
1661 | ||
1662 | g_string_append_len(in->buf, buf->str, buf->len); | |
41d214f6 BV |
1663 | |
1664 | inc = in->priv; | |
1a920e33 | 1665 | if (!inc->column_seen_count) { |
df0db9fd GS |
1666 | ret = initial_receive(in); |
1667 | if (ret == SR_ERR_NA) | |
7066fd46 BV |
1668 | /* Not enough data yet. */ |
1669 | return SR_OK; | |
1670 | else if (ret != SR_OK) | |
1671 | return SR_ERR; | |
1672 | ||
1673 | /* sdi is ready, notify frontend. */ | |
1674 | in->sdi_ready = TRUE; | |
41d214f6 | 1675 | return SR_OK; |
7066fd46 BV |
1676 | } |
1677 | ||
7f4c3a62 | 1678 | ret = process_buffer(in, FALSE); |
7066fd46 BV |
1679 | |
1680 | return ret; | |
1681 | } | |
1682 | ||
1683 | static int end(struct sr_input *in) | |
1684 | { | |
1685 | struct context *inc; | |
7066fd46 | 1686 | int ret; |
41d214f6 | 1687 | |
7066fd46 | 1688 | if (in->sdi_ready) |
7f4c3a62 | 1689 | ret = process_buffer(in, TRUE); |
7066fd46 BV |
1690 | else |
1691 | ret = SR_OK; | |
cd59e6ec GS |
1692 | if (ret != SR_OK) |
1693 | return ret; | |
1694 | ||
626c388a | 1695 | ret = flush_logic_samples(in); |
43bdef26 | 1696 | ret += flush_analog_samples(in); |
cd59e6ec GS |
1697 | if (ret != SR_OK) |
1698 | return ret; | |
7066fd46 BV |
1699 | |
1700 | inc = in->priv; | |
3be42bc2 | 1701 | if (inc->started) |
bee2b016 | 1702 | std_session_send_df_end(in->sdi); |
4a35548b | 1703 | |
7066fd46 BV |
1704 | return ret; |
1705 | } | |
1706 | ||
d5cc282f | 1707 | static void cleanup(struct sr_input *in) |
7066fd46 | 1708 | { |
307b2393 | 1709 | struct context *inc, save_ctx; |
7066fd46 | 1710 | |
307b2393 | 1711 | /* Keep channel references between file re-imports. */ |
affaf540 GS |
1712 | keep_header_for_reread(in); |
1713 | ||
307b2393 | 1714 | /* Release dynamically allocated resources. */ |
7066fd46 BV |
1715 | inc = in->priv; |
1716 | ||
b1f83103 | 1717 | g_free(inc->termination); |
539188e5 | 1718 | inc->termination = NULL; |
cd59e6ec | 1719 | g_free(inc->datafeed_buffer); |
539188e5 | 1720 | inc->datafeed_buffer = NULL; |
43bdef26 GS |
1721 | g_free(inc->analog_datafeed_buffer); |
1722 | inc->analog_datafeed_buffer = NULL; | |
307b2393 GS |
1723 | g_free(inc->analog_datafeed_digits); |
1724 | inc->analog_datafeed_digits = NULL; | |
1725 | /* analog_datafeed_channels was released in keep_header_for_reread() */ | |
1726 | /* TODO Release channel names (before releasing details). */ | |
1727 | g_free(inc->column_details); | |
1728 | inc->column_details = NULL; | |
1729 | ||
1730 | /* Clear internal state, but keep what .init() has provided. */ | |
1731 | save_ctx = *inc; | |
1732 | memset(inc, 0, sizeof(*inc)); | |
1733 | inc->samplerate = save_ctx.samplerate; | |
1734 | inc->delimiter = save_ctx.delimiter; | |
1735 | inc->comment = save_ctx.comment; | |
1736 | inc->column_formats = save_ctx.column_formats; | |
1737 | inc->start_line = save_ctx.start_line; | |
1738 | inc->use_header = save_ctx.use_header; | |
1739 | inc->prev_sr_channels = save_ctx.prev_sr_channels; | |
1740 | inc->prev_df_channels = save_ctx.prev_df_channels; | |
4a35548b MS |
1741 | } |
1742 | ||
ad93bfb0 SA |
1743 | static int reset(struct sr_input *in) |
1744 | { | |
307b2393 | 1745 | struct context *inc; |
ad93bfb0 | 1746 | |
307b2393 | 1747 | inc = in->priv; |
ad93bfb0 SA |
1748 | cleanup(in); |
1749 | inc->started = FALSE; | |
1750 | g_string_truncate(in->buf, 0); | |
1751 | ||
1752 | return SR_OK; | |
1753 | } | |
1754 | ||
c6aa9870 | 1755 | enum option_index { |
2142a79b | 1756 | OPT_COL_FMTS, |
c6aa9870 | 1757 | OPT_SINGLE_COL, |
72903e9d | 1758 | OPT_FIRST_COL, |
c6aa9870 | 1759 | OPT_NUM_LOGIC, |
43e1e23a GS |
1760 | OPT_SINGLE_FMT, |
1761 | OPT_START_LINE, | |
72903e9d | 1762 | OPT_HEADER, |
43e1e23a GS |
1763 | OPT_SAMPLERATE, |
1764 | OPT_COL_SEP, | |
72903e9d | 1765 | OPT_COMMENT, |
c6aa9870 GS |
1766 | OPT_MAX, |
1767 | }; | |
1768 | ||
41d214f6 | 1769 | static struct sr_option options[] = { |
72903e9d GS |
1770 | [OPT_COL_FMTS] = { |
1771 | "column_formats", "Column format specs", | |
43e1e23a | 1772 | "Text columns data types. A comma separated list of [<cols>]<fmt>[<bits>] items. * for all remaining columns. - ignores columns, x/o/b/l logic data, a (and digits) analog data, t timestamps.", |
72903e9d GS |
1773 | NULL, NULL, |
1774 | }, | |
1775 | [OPT_SINGLE_COL] = { | |
1776 | "single_column", "Single column", | |
43e1e23a | 1777 | "Simple single-column mode, exclusively use text from the specified column (number starting at 1). Obsoleted by 'column_formats=4-,x16'.", |
72903e9d GS |
1778 | NULL, NULL, |
1779 | }, | |
1780 | [OPT_FIRST_COL] = { | |
1781 | "first_column", "First column", | |
43e1e23a | 1782 | "First column with logic data in simple multi-column mode (number starting at 1, default 1). Obsoleted by 'column_formats=4-,*l'.", |
72903e9d GS |
1783 | NULL, NULL, |
1784 | }, | |
1785 | [OPT_NUM_LOGIC] = { | |
1786 | "logic_channels", "Number of logic channels", | |
43e1e23a | 1787 | "Logic channel count, required in simple single-column mode, defaults to \"all remaining columns\" in simple multi-column mode. Obsoleted by 'column_formats=8l'.", |
72903e9d GS |
1788 | NULL, NULL, |
1789 | }, | |
43e1e23a | 1790 | [OPT_SINGLE_FMT] = { |
72903e9d | 1791 | "single_format", "Data format for simple single-column mode.", |
43e1e23a | 1792 | "The input text number format of simple single-column mode: bin, hex, oct. Obsoleted by 'column_formats=x8'.", |
72903e9d GS |
1793 | NULL, NULL, |
1794 | }, | |
43e1e23a | 1795 | [OPT_START_LINE] = { |
72903e9d GS |
1796 | "start_line", "Start line", |
1797 | "The line number at which to start processing input text (default: 1).", | |
1798 | NULL, NULL, | |
1799 | }, | |
1800 | [OPT_HEADER] = { | |
1801 | "header", "Get channel names from first line.", | |
31b4a9a2 | 1802 | "Use the first processed line's column captions (when available) as channel names. Enabled by default.", |
72903e9d GS |
1803 | NULL, NULL, |
1804 | }, | |
43e1e23a | 1805 | [OPT_SAMPLERATE] = { |
72903e9d | 1806 | "samplerate", "Samplerate (Hz)", |
08eb955a | 1807 | "The input data's sample rate in Hz. No default value.", |
72903e9d GS |
1808 | NULL, NULL, |
1809 | }, | |
43e1e23a | 1810 | [OPT_COL_SEP] = { |
72903e9d GS |
1811 | "column_separator", "Column separator", |
1812 | "The sequence which separates text columns. Non-empty text, comma by default.", | |
1813 | NULL, NULL, | |
1814 | }, | |
1815 | [OPT_COMMENT] = { | |
1816 | "comment_leader", "Comment leader character", | |
08eb955a | 1817 | "The text which starts comments at the end of text lines, semicolon by default.", |
72903e9d GS |
1818 | NULL, NULL, |
1819 | }, | |
c6aa9870 | 1820 | [OPT_MAX] = ALL_ZERO, |
41d214f6 BV |
1821 | }; |
1822 | ||
2c240774 | 1823 | static const struct sr_option *get_options(void) |
41d214f6 | 1824 | { |
31c41782 UH |
1825 | GSList *l; |
1826 | ||
41d214f6 | 1827 | if (!options[0].def) { |
1a920e33 | 1828 | options[OPT_COL_FMTS].def = g_variant_ref_sink(g_variant_new_string("")); |
e53f32d2 | 1829 | options[OPT_SINGLE_COL].def = g_variant_ref_sink(g_variant_new_uint32(0)); |
72903e9d | 1830 | options[OPT_FIRST_COL].def = g_variant_ref_sink(g_variant_new_uint32(1)); |
e53f32d2 | 1831 | options[OPT_NUM_LOGIC].def = g_variant_ref_sink(g_variant_new_uint32(0)); |
43e1e23a | 1832 | options[OPT_SINGLE_FMT].def = g_variant_ref_sink(g_variant_new_string("bin")); |
31c41782 UH |
1833 | l = NULL; |
1834 | l = g_slist_append(l, g_variant_ref_sink(g_variant_new_string("bin"))); | |
1835 | l = g_slist_append(l, g_variant_ref_sink(g_variant_new_string("hex"))); | |
1836 | l = g_slist_append(l, g_variant_ref_sink(g_variant_new_string("oct"))); | |
43e1e23a GS |
1837 | options[OPT_SINGLE_FMT].values = l; |
1838 | options[OPT_START_LINE].def = g_variant_ref_sink(g_variant_new_uint32(1)); | |
31b4a9a2 | 1839 | options[OPT_HEADER].def = g_variant_ref_sink(g_variant_new_boolean(TRUE)); |
43e1e23a GS |
1840 | options[OPT_SAMPLERATE].def = g_variant_ref_sink(g_variant_new_uint64(0)); |
1841 | options[OPT_COL_SEP].def = g_variant_ref_sink(g_variant_new_string(",")); | |
72903e9d | 1842 | options[OPT_COMMENT].def = g_variant_ref_sink(g_variant_new_string(";")); |
41d214f6 BV |
1843 | } |
1844 | ||
1845 | return options; | |
1846 | } | |
1847 | ||
d4c93774 | 1848 | SR_PRIV struct sr_input_module input_csv = { |
4a35548b | 1849 | .id = "csv", |
41d214f6 BV |
1850 | .name = "CSV", |
1851 | .desc = "Comma-separated values", | |
c7bc82ff | 1852 | .exts = (const char*[]){"csv", NULL}, |
cd7c5f96 | 1853 | .metadata = { SR_INPUT_META_FILENAME, SR_INPUT_META_HEADER | SR_INPUT_META_REQUIRED }, |
41d214f6 | 1854 | .options = get_options, |
cd7c5f96 | 1855 | .format_match = format_match, |
4a35548b | 1856 | .init = init, |
41d214f6 | 1857 | .receive = receive, |
7066fd46 | 1858 | .end = end, |
41d214f6 | 1859 | .cleanup = cleanup, |
ad93bfb0 | 1860 | .reset = reset, |
4a35548b | 1861 | }; |