1 | // Micro Syntax
|
2 | //
|
3 | // See doctools/micro-syntax.md
|
4 |
|
5 | #include "micro_syntax.h" // requires -I $BASE_DIR
|
6 |
|
7 | #include <assert.h>
|
8 | #include <errno.h>
|
9 | #include <getopt.h>
|
10 | #include <stdarg.h> // va_list, etc.
|
11 | #include <stdbool.h>
|
12 | #include <stdio.h>
|
13 | #include <stdlib.h> // free
|
14 | #include <string.h>
|
15 |
|
16 | #include <string>
|
17 | #include <vector>
|
18 |
|
19 | const char* RESET = "\x1b[0;0m";
|
20 | const char* BOLD = "\x1b[1m";
|
21 | const char* UNDERLINE = "\x1b[4m";
|
22 | const char* REVERSE = "\x1b[7m"; // reverse video
|
23 |
|
24 | const char* BLACK = "\x1b[30m";
|
25 | const char* RED = "\x1b[31m";
|
26 | const char* GREEN = "\x1b[32m";
|
27 | const char* YELLOW = "\x1b[33m";
|
28 | const char* BLUE = "\x1b[34m";
|
29 | const char* PURPLE = "\x1b[35m";
|
30 | const char* CYAN = "\x1b[36m";
|
31 | const char* WHITE = "\x1b[37m";
|
32 |
|
33 | const char* BLACK2 = "\x1b[90m";
|
34 | const char* RED2 = "\x1b[91m";
|
35 | const char* BLUE2 = "\x1b[94m";
|
36 |
|
37 | void Log(const char* fmt, ...) {
|
38 | va_list args;
|
39 | va_start(args, fmt);
|
40 | vfprintf(stderr, fmt, args);
|
41 | va_end(args);
|
42 | fputs("\n", stderr);
|
43 | }
|
44 |
|
45 | enum class lang_e {
|
46 | PlainText,
|
47 |
|
48 | Cpp, // including C
|
49 | Py,
|
50 | Shell,
|
51 | Ysh, // ''' etc.
|
52 | Html,
|
53 | Asdl,
|
54 | R, // uses # comments
|
55 |
|
56 | // JS, // uses // comments
|
57 | };
|
58 |
|
59 | class Reader {
|
60 | // We don't care about internal NUL, so this interface doesn't allow it
|
61 |
|
62 | public:
|
63 | Reader(FILE* f, const char* filename)
|
64 | : f_(f), filename_(filename), line_(nullptr), allocated_size_(0) {
|
65 | }
|
66 |
|
67 | const char* Filename() { // for error messages only, nullptr for stdin
|
68 | return filename_;
|
69 | }
|
70 |
|
71 | bool NextLine() {
|
72 | // Returns false if there was an error, and sets err_num_.
|
73 | // Returns true if not error, and Current() can be checked.
|
74 |
|
75 | // Note: getline() frees the previous line, so we don't have to
|
76 | ssize_t len = getline(&line_, &allocated_size_, f_);
|
77 |
|
78 | if (len < 0) { // EOF is -1
|
79 | // man page says the buffer should be freed if getline() fails
|
80 | free(line_);
|
81 |
|
82 | line_ = nullptr; // tell the caller not to continue
|
83 |
|
84 | if (errno != 0) { // I/O error
|
85 | err_num_ = errno;
|
86 | return false;
|
87 | }
|
88 | }
|
89 | return true;
|
90 | }
|
91 |
|
92 | char* Current() {
|
93 | // Returns nullptr on EOF.
|
94 | return line_;
|
95 | }
|
96 |
|
97 | FILE* f_;
|
98 | const char* filename_;
|
99 |
|
100 | char* line_; // valid for one NextLine() call, nullptr on EOF or error
|
101 | size_t allocated_size_; // unused, but must pass address to getline()
|
102 | int err_num_; // set on error
|
103 | };
|
104 |
|
105 | class Printer {
|
106 | public:
|
107 | virtual void PrintLineNumber(int line_num) = 0;
|
108 | virtual void PrintLineEnd() {
|
109 | }
|
110 | virtual void PrintToken(const char* line, int line_num, int start_col,
|
111 | Token token) = 0;
|
112 | virtual void Swap(std::string* s) {
|
113 | assert(0);
|
114 | }
|
115 | virtual ~Printer() {
|
116 | }
|
117 | };
|
118 |
|
119 | class HtmlPrinter : public Printer {
|
120 | public:
|
121 | HtmlPrinter() : Printer(), out_() {
|
122 | }
|
123 |
|
124 | virtual void Swap(std::string* s) {
|
125 | // assert(s != nullptr);
|
126 | out_.swap(*s);
|
127 | }
|
128 |
|
129 | virtual void PrintLineNumber(int line_num) {
|
130 | char buf[16];
|
131 | snprintf(buf, 16, "%d", line_num);
|
132 |
|
133 | out_.append("<tr><td class=num>"); // <tr> closed by PrintLineEnd()
|
134 | out_.append(buf);
|
135 | out_.append("</td><td id=L"); // jump to line with foo.html#L32
|
136 | out_.append(buf);
|
137 | out_.append(" class=line>"); // <td> closed by PrintLineEnd()
|
138 | }
|
139 |
|
140 | virtual void PrintLineEnd() {
|
141 | out_.append("</td></tr>");
|
142 | }
|
143 |
|
144 | virtual void PrintToken(const char* line, int line_num, int start_col,
|
145 | Token tok) {
|
146 | const char* p_start = line + start_col;
|
147 | int num_bytes = tok.end_col - start_col;
|
148 |
|
149 | switch (tok.id) {
|
150 | case Id::Comm:
|
151 | PrintSpan("comm", p_start, num_bytes);
|
152 | break;
|
153 |
|
154 | case Id::Name:
|
155 | PrintEscaped(p_start, num_bytes);
|
156 | break;
|
157 |
|
158 | case Id::PreprocCommand:
|
159 | case Id::LineCont:
|
160 | PrintSpan("preproc", p_start, num_bytes);
|
161 | break;
|
162 |
|
163 | case Id::Re2c:
|
164 | PrintSpan("re2c", p_start, num_bytes);
|
165 | break;
|
166 |
|
167 | case Id::Other:
|
168 | // PrintSpan("other", p_start, num_bytes);
|
169 | PrintEscaped(p_start, num_bytes);
|
170 | break;
|
171 |
|
172 | // for now these are strings
|
173 | case Id::HereBegin:
|
174 | case Id::HereEnd:
|
175 | case Id::Str:
|
176 | PrintSpan("str", p_start, num_bytes);
|
177 | break;
|
178 |
|
179 | case Id::LBrace:
|
180 | case Id::RBrace:
|
181 | PrintSpan("brace", p_start, num_bytes);
|
182 | break;
|
183 |
|
184 | case Id::Unknown:
|
185 | PrintSpan("x", p_start, num_bytes);
|
186 | break;
|
187 |
|
188 | default:
|
189 | PrintEscaped(p_start, num_bytes);
|
190 | break;
|
191 | }
|
192 | }
|
193 |
|
194 | private:
|
195 | void PrintEscaped(const char* s, int len) {
|
196 | // HTML escape the code string
|
197 | for (int i = 0; i < len; ++i) {
|
198 | char c = s[i];
|
199 |
|
200 | switch (c) {
|
201 | case '<':
|
202 | out_.append("<");
|
203 | break;
|
204 | case '>':
|
205 | out_.append(">");
|
206 | break;
|
207 | case '&':
|
208 | out_.append("&");
|
209 | break;
|
210 | default:
|
211 | // Is this inefficient? Fill 1 char
|
212 | out_.append(1, s[i]);
|
213 | break;
|
214 | }
|
215 | }
|
216 | }
|
217 |
|
218 | void PrintSpan(const char* css_class, const char* s, int len) {
|
219 | out_.append("<span class=");
|
220 | out_.append(css_class);
|
221 | out_.append(">");
|
222 |
|
223 | PrintEscaped(s, len);
|
224 |
|
225 | out_.append("</span>");
|
226 | }
|
227 |
|
228 | std::string out_;
|
229 | };
|
230 |
|
231 | struct Flags {
|
232 | lang_e lang;
|
233 | bool tsv;
|
234 | bool web;
|
235 | bool more_color;
|
236 | bool comments_only;
|
237 |
|
238 | int argc;
|
239 | char** argv;
|
240 | };
|
241 |
|
242 | class AnsiPrinter : public Printer {
|
243 | public:
|
244 | AnsiPrinter(const Flags& flag) : Printer(), flag_(flag) {
|
245 | }
|
246 |
|
247 | virtual void PrintLineNumber(int line_num) {
|
248 | if (flag_.comments_only) {
|
249 | return;
|
250 | }
|
251 | printf("%s%5d%s ", BLACK2, line_num, RESET);
|
252 | }
|
253 |
|
254 | virtual void PrintToken(const char* line, int line_num, int start_col,
|
255 | Token tok) {
|
256 | const char* p_start = line + start_col;
|
257 | int num_bytes = tok.end_col - start_col;
|
258 | switch (tok.id) {
|
259 | case Id::Comm:
|
260 | if (flag_.comments_only) {
|
261 | PrintAlways(p_start, num_bytes);
|
262 | } else {
|
263 | PrintColor(BLUE, p_start, num_bytes);
|
264 | }
|
265 | break;
|
266 |
|
267 | case Id::Name:
|
268 | PrintText(p_start, num_bytes);
|
269 | break;
|
270 |
|
271 | case Id::PreprocCommand:
|
272 | case Id::LineCont:
|
273 | PrintColor(PURPLE, p_start, num_bytes);
|
274 | break;
|
275 |
|
276 | case Id::Re2c:
|
277 | PrintColor(PURPLE, p_start, num_bytes);
|
278 | break;
|
279 |
|
280 | case Id::Other:
|
281 | if (flag_.more_color) {
|
282 | PrintColor(PURPLE, p_start, num_bytes);
|
283 | } else {
|
284 | PrintText(p_start, num_bytes);
|
285 | }
|
286 | break;
|
287 |
|
288 | case Id::WS:
|
289 | if (flag_.more_color) {
|
290 | fputs(REVERSE, stdout);
|
291 | PrintColor(WHITE, p_start, num_bytes);
|
292 | } else {
|
293 | PrintText(p_start, num_bytes);
|
294 | }
|
295 | break;
|
296 |
|
297 | case Id::Str:
|
298 | PrintColor(RED, p_start, num_bytes);
|
299 | break;
|
300 |
|
301 | case Id::HereBegin:
|
302 | case Id::HereEnd: {
|
303 | PrintColor(RED2, p_start, num_bytes);
|
304 |
|
305 | // Debug submatch extraction
|
306 | #if 0
|
307 | fputs(RED, stdout);
|
308 | int n = tok.submatch_len;
|
309 | fwrite(tok.submatch_start, 1, n, stdout);
|
310 | fputs(RESET, stdout);
|
311 | #endif
|
312 | } break;
|
313 |
|
314 | case Id::DelimStrBegin:
|
315 | case Id::DelimStrEnd: {
|
316 | PrintColor(RED2, p_start, num_bytes);
|
317 |
|
318 | // Debug submatch extraction
|
319 | #if 0
|
320 | fputs(RED, stdout);
|
321 | int n = tok.submatch_len;
|
322 | fwrite(tok.submatch_start, 1, n, stdout);
|
323 | fputs(RESET, stdout);
|
324 | #endif
|
325 | } break;
|
326 |
|
327 | case Id::LBrace:
|
328 | case Id::RBrace:
|
329 | PrintColor(GREEN, p_start, num_bytes);
|
330 | break;
|
331 |
|
332 | case Id::TagNameLeft:
|
333 | case Id::TagNameRight:
|
334 | PrintColor(PURPLE, p_start, num_bytes);
|
335 | break;
|
336 |
|
337 | case Id::SelfClose:
|
338 | case Id::EndTag:
|
339 | PrintColor(RED2, p_start, num_bytes);
|
340 | break;
|
341 |
|
342 | case Id::CharEscape:
|
343 | PrintColor(BLUE, p_start, num_bytes);
|
344 | break;
|
345 |
|
346 | case Id::Unknown:
|
347 | case Id::BadAmpersand:
|
348 | case Id::BadGreaterThan:
|
349 | case Id::BadLessThan:
|
350 | // Make errors red
|
351 | fputs(REVERSE, stdout);
|
352 | PrintColor(RED, p_start, num_bytes);
|
353 | break;
|
354 |
|
355 | default:
|
356 | PrintText(p_start, num_bytes);
|
357 | break;
|
358 | }
|
359 | }
|
360 |
|
361 | private:
|
362 | void PrintColor(const char* color, const char* s, int n) {
|
363 | fputs(color, stdout);
|
364 | PrintText(s, n);
|
365 | fputs(RESET, stdout);
|
366 | }
|
367 |
|
368 | void PrintText(const char* s, int n) {
|
369 | if (flag_.comments_only) {
|
370 | for (int i = 0; i < n; ++i) {
|
371 | // Replace everything but newline with space
|
372 | // TODO: I think we always want a newline token, including in comments.
|
373 | // That will simplify this.
|
374 | char c = (s[i] == '\n') ? '\n' : ' ';
|
375 | fwrite(&c, 1, 1, stdout);
|
376 | }
|
377 | } else {
|
378 | fwrite(s, 1, n, stdout);
|
379 | }
|
380 | }
|
381 |
|
382 | void PrintAlways(const char* s, int n) {
|
383 | fwrite(s, 1, n, stdout);
|
384 | }
|
385 |
|
386 | const Flags& flag_;
|
387 | };
|
388 |
|
389 | const char* Id_str(Id id) {
|
390 | switch (id) {
|
391 | case Id::Comm:
|
392 | return "Comm";
|
393 | case Id::MaybeComment: // fix-up doesn't guarantee this is gone
|
394 | return "MaybeComment";
|
395 | case Id::WS:
|
396 | return "WS";
|
397 | case Id::Re2c:
|
398 | return "Re2c";
|
399 |
|
400 | case Id::MaybePreproc: // fix-up doesn't guarantee this is gone
|
401 | return "MaybePreproc";
|
402 | case Id::PreprocCommand:
|
403 | return "PreprocCommand";
|
404 | case Id::PreprocOther:
|
405 | return "PreprocOther";
|
406 | case Id::LineCont:
|
407 | return "LineCont";
|
408 |
|
409 | case Id::Name:
|
410 | return "Name";
|
411 | case Id::Other:
|
412 | return "Other";
|
413 |
|
414 | case Id::Str:
|
415 | return "Str";
|
416 |
|
417 | case Id::HereBegin:
|
418 | return "HereBegin";
|
419 | case Id::HereEnd:
|
420 | return "HereEnd";
|
421 | case Id::DelimStrBegin:
|
422 | return "DelimStrBegin";
|
423 | case Id::DelimStrEnd:
|
424 | return "DelimStrEnd";
|
425 |
|
426 | case Id::LBrace:
|
427 | return "LBrace";
|
428 | case Id::RBrace:
|
429 | return "RBrace";
|
430 |
|
431 | case Id::Unknown:
|
432 | return "Unknown";
|
433 | default:
|
434 | assert(0);
|
435 | }
|
436 | }
|
437 |
|
438 | class TsvPrinter : public Printer {
|
439 | public:
|
440 | virtual void PrintLineNumber(int line_num) {
|
441 | ;
|
442 | }
|
443 |
|
444 | virtual void Swap(std::string* s) {
|
445 | // out_.swap(*s);
|
446 | }
|
447 |
|
448 | virtual void PrintToken(const char* line, int line_num, int start_col,
|
449 | Token tok) {
|
450 | printf("%d\t%s\t%d\t%d\n", line_num, Id_str(tok.id), start_col,
|
451 | tok.end_col);
|
452 | // printf(" -> mode %d\n", lexer.line_mode);
|
453 | }
|
454 | virtual ~TsvPrinter() {
|
455 | }
|
456 | };
|
457 |
|
458 | bool TokenIsSignificant(Id id) {
|
459 | switch (id) {
|
460 | case Id::Name:
|
461 | case Id::Other:
|
462 | case Id::PreprocCommand:
|
463 | case Id::PreprocOther:
|
464 | case Id::Re2c:
|
465 | return true;
|
466 |
|
467 | // Comments, whitespace, and string literals aren't significant
|
468 | // TODO: can abort on Id::Unknown?
|
469 | default:
|
470 | break;
|
471 | }
|
472 | return false;
|
473 | }
|
474 |
|
475 | class OutputStream {
|
476 | // stdout contains either
|
477 | // - netstrings of HTML, or TSV Token structs
|
478 | // - ANSI text
|
479 |
|
480 | public:
|
481 | OutputStream(Printer* pr) : pr_(pr) {
|
482 | }
|
483 | virtual void PathBegin(const char* path) = 0;
|
484 | virtual void Line(int line_num, const char* line,
|
485 | const std::vector<Token>& tokens) = 0;
|
486 | virtual void PathEnd(int num_lines, int num_sig_lines) = 0;
|
487 | virtual ~OutputStream() {
|
488 | }
|
489 |
|
490 | protected:
|
491 | Printer* pr_; // how to print each file
|
492 | };
|
493 |
|
494 | class NetStringOutput : public OutputStream {
|
495 | public:
|
496 | NetStringOutput(Printer* pr) : OutputStream(pr) {
|
497 | }
|
498 |
|
499 | virtual void PathBegin(const char* path) {
|
500 | if (path == nullptr) {
|
501 | path = "<stdin>";
|
502 | }
|
503 | PrintNetString(path, strlen(path));
|
504 | }
|
505 |
|
506 | virtual void Line(int line_num, const char* line,
|
507 | const std::vector<Token>& tokens) {
|
508 | pr_->PrintLineNumber(line_num);
|
509 |
|
510 | int start_col = 0;
|
511 | for (auto tok : tokens) {
|
512 | pr_->PrintToken(line, line_num, start_col, tok);
|
513 | start_col = tok.end_col;
|
514 | }
|
515 |
|
516 | pr_->PrintLineEnd();
|
517 | }
|
518 |
|
519 | virtual void PathEnd(int num_lines, int num_sig_lines) {
|
520 | std::string string_for_file;
|
521 | pr_->Swap(&string_for_file);
|
522 |
|
523 | PrintNetString(string_for_file.c_str(), string_for_file.size());
|
524 |
|
525 | // Output summary in JSON
|
526 | // TODO: change this to a 4th column
|
527 | char buf[64];
|
528 | int n = snprintf(buf, 64, "{\"num_lines\": %d, \"num_sig_lines\": %d}",
|
529 | num_lines, num_sig_lines);
|
530 | PrintNetString(buf, n);
|
531 | }
|
532 |
|
533 | private:
|
534 | void PrintNetString(const char* s, int len) {
|
535 | fprintf(stdout, "%d:%*s,", len, len, s);
|
536 | }
|
537 | };
|
538 |
|
539 | class AnsiOutput : public OutputStream {
|
540 | public:
|
541 | AnsiOutput(Printer* pr) : OutputStream(pr) {
|
542 | }
|
543 |
|
544 | // TODO: Can respect --comments-only
|
545 | virtual void PathBegin(const char* path) {
|
546 | if (path == nullptr) {
|
547 | path = "<stdin>";
|
548 | }
|
549 | // diff uses +++ ---
|
550 | printf("\n");
|
551 | printf("=== %s%s%s%s ===\n", BOLD, PURPLE, path, RESET);
|
552 | printf("\n");
|
553 | }
|
554 |
|
555 | virtual void Line(int line_num, const char* line,
|
556 | const std::vector<Token>& tokens) {
|
557 | pr_->PrintLineNumber(line_num);
|
558 |
|
559 | int start_col = 0;
|
560 | for (auto tok : tokens) {
|
561 | pr_->PrintToken(line, line_num, start_col, tok);
|
562 | start_col = tok.end_col;
|
563 | }
|
564 |
|
565 | pr_->PrintLineEnd();
|
566 | };
|
567 |
|
568 | // TODO: Can respect --comments-only
|
569 | virtual void PathEnd(int num_lines, int num_sig_lines) {
|
570 | fprintf(stdout, "%s%d lines, %d significant%s\n", GREEN, num_lines,
|
571 | num_sig_lines, RESET);
|
572 | };
|
573 | };
|
574 |
|
575 | void PrintTokens(std::vector<Token>& toks) {
|
576 | int start_col = 0;
|
577 | int i = 0;
|
578 | Log("===");
|
579 | for (auto tok : toks) {
|
580 | Log("%2d %10s %2d %2d", i, Id_str(tok.id), start_col, tok.end_col);
|
581 | start_col = tok.end_col;
|
582 | ++i;
|
583 | }
|
584 | Log("===");
|
585 | }
|
586 |
|
587 | // BUGGY, needs unit tests
|
588 |
|
589 | // Fiddly function, reduces the size of the output a bit
|
590 | // "hi" becomes 1 Id::DQ token instead of 3 separate Id::DQ tokens
|
591 | void Optimize(std::vector<Token>* tokens) {
|
592 | std::vector<Token>& toks = *tokens; // alias
|
593 |
|
594 | // PrintTokens(toks);
|
595 |
|
596 | int n = toks.size();
|
597 | if (n < 1) { // nothing to de-duplicate
|
598 | return;
|
599 | }
|
600 |
|
601 | int left = 0;
|
602 | int right = 1;
|
603 | while (right < n) {
|
604 | Log("right ID = %s, end %d", Id_str(toks[right].id), toks[right].end_col);
|
605 |
|
606 | if (toks[left].id == toks[right].id) {
|
607 | // Join the tokens together
|
608 | toks[left].end_col = toks[right].end_col;
|
609 | } else {
|
610 | toks[left] = toks[right];
|
611 | left++;
|
612 | Log(" not eq, left = %d", left);
|
613 | }
|
614 | right++;
|
615 | }
|
616 | Log("left = %d, right = %d", left, right);
|
617 |
|
618 | // Fiddly condition: one more iteration. Need some unit tests for this.
|
619 | toks[left] = toks[right - 1];
|
620 | left++;
|
621 | assert(left <= n);
|
622 |
|
623 | // Erase the remaining ones
|
624 | toks.resize(left);
|
625 |
|
626 | // PrintTokens(toks);
|
627 | }
|
628 |
|
629 | // Version of the above that's not in-place, led to a bug fix
|
630 | void Optimize2(std::vector<Token>* tokens) {
|
631 | std::vector<Token> optimized;
|
632 |
|
633 | int n = tokens->size();
|
634 | if (n < 1) {
|
635 | return;
|
636 | }
|
637 |
|
638 | optimized.reserve(n);
|
639 |
|
640 | int left = 0;
|
641 | int right = 1;
|
642 | while (right < n) {
|
643 | optimized.push_back((*tokens)[left]);
|
644 | left++;
|
645 | right++;
|
646 | }
|
647 | optimized.push_back((*tokens)[left]);
|
648 | left++;
|
649 |
|
650 | tokens->swap(optimized);
|
651 | }
|
652 |
|
653 | bool LineEqualsHereDelim(const char* line, std::string& here_delim) {
|
654 | // Compare EOF vs. EOF\n or EOF\t\n or x\n
|
655 |
|
656 | // Hack: skip leading tab unconditionally, even though that's only alowed in
|
657 | // <<- Really we should capture the operator and the delim?
|
658 | if (*line == '\t') {
|
659 | line++;
|
660 | }
|
661 |
|
662 | int n = strlen(line);
|
663 | int h = here_delim.size();
|
664 |
|
665 | // Log("Here delim=%s line=%s", here_delim.c_str(), line);
|
666 |
|
667 | // Line should be at least one longer, EOF\n
|
668 | if (n <= h) {
|
669 | // Log(" [0] line too short");
|
670 | return false;
|
671 | }
|
672 |
|
673 | int i = 0;
|
674 | for (; i < h; ++i) {
|
675 | if (here_delim[i] != line[i]) {
|
676 | // Log(" [1] byte %d not equal", i);
|
677 | return false;
|
678 | }
|
679 | }
|
680 |
|
681 | while (i < n) {
|
682 | switch (line[i]) {
|
683 | case ' ':
|
684 | case '\t':
|
685 | case '\r':
|
686 | case '\n':
|
687 | break;
|
688 | default:
|
689 | // Log(" [2] byte %d not whitespace", i);
|
690 | return false; // line can't have whitespace on the end
|
691 | }
|
692 | ++i;
|
693 | }
|
694 |
|
695 | return true;
|
696 | }
|
697 |
|
698 | void CppHook::TryPreprocess(char* line, std::vector<Token>* tokens) {
|
699 | // Fills tokens, which can be checked for beginning and end tokens
|
700 |
|
701 | Lexer<pp_mode_e> lexer(line);
|
702 | Matcher<pp_mode_e> matcher;
|
703 |
|
704 | while (true) { // tokens on each line
|
705 | Token tok;
|
706 | // Log("Match %d", lexer.p_current - lexer.line_);
|
707 | bool eol = matcher.Match(&lexer, &tok);
|
708 | // Log("EOL %d", eol);
|
709 | if (eol) {
|
710 | break;
|
711 | }
|
712 | // Log("TOK %s %d", Id_str(tok.id), tok.end_col);
|
713 | tokens->push_back(tok); // make a copy
|
714 | }
|
715 | }
|
716 |
|
717 | void FixShellComments(std::vector<Token>& tokens) {
|
718 | int n = tokens.size();
|
719 | for (int i = 0; i < n; ++i) {
|
720 | // # comment at start of line
|
721 | if (tokens[i].id == Id::MaybeComment) {
|
722 | if (i == 0) {
|
723 | tokens[i].id = Id::Comm;
|
724 | }
|
725 | if (i != 0 and tokens[i - 1].id == Id::WS) {
|
726 | tokens[i].id = Id::Comm;
|
727 | }
|
728 | }
|
729 | }
|
730 | }
|
731 |
|
732 | // This templated method causes some code expansion, but not too much. The
|
733 | // binary went from 38 KB to 42 KB, after being stripped.
|
734 | // We get a little type safety with py_mode_e vs cpp_mode_e.
|
735 |
|
736 | template <typename T>
|
737 | int ScanOne(Reader* reader, OutputStream* out, Hook* hook) {
|
738 | Lexer<T> lexer(nullptr);
|
739 | Matcher<T> matcher;
|
740 |
|
741 | int line_num = 1;
|
742 | int num_sig = 0;
|
743 |
|
744 | std::vector<std::string> here_list; // delimiters to pop
|
745 | std::vector<int> here_start_num;
|
746 |
|
747 | // For multi-line strings. This has 0 or 1 entries, and the 1 entry can be
|
748 | // the empty string.
|
749 | std::vector<std::string> delim_begin;
|
750 |
|
751 | while (true) { // read each line, handling errors
|
752 | if (!reader->NextLine()) {
|
753 | const char* name = reader->Filename() ?: "<stdin>";
|
754 | Log("micro-syntax: getline() error on %s: %s", name,
|
755 | strerror(reader->err_num_));
|
756 | return 1;
|
757 | }
|
758 | char* line = reader->Current();
|
759 | if (line == nullptr) {
|
760 | break; // EOF
|
761 | }
|
762 |
|
763 | std::vector<Token> pre_tokens;
|
764 |
|
765 | hook->TryPreprocess(line, &pre_tokens);
|
766 |
|
767 | // e.g #define at beginning of line
|
768 | if (pre_tokens.size() && pre_tokens[0].id == Id::MaybePreproc) {
|
769 | pre_tokens[0].id = Id::PreprocCommand;
|
770 |
|
771 | out->Line(line_num, line, pre_tokens);
|
772 |
|
773 | line_num += 1;
|
774 | num_sig += 1;
|
775 |
|
776 | Token last = pre_tokens.back();
|
777 | while (last.id == Id::LineCont) {
|
778 | const char* blame = reader->Filename() ?: "<stdin>";
|
779 | if (!reader->NextLine()) {
|
780 | Log("micro-syntax: getline() error on %s: %s", blame,
|
781 | strerror(reader->err_num_));
|
782 | return 1;
|
783 | }
|
784 | char* line = reader->Current();
|
785 | if (line == nullptr) {
|
786 | Log("Unexpected end-of-file in preprocessor in %s", blame);
|
787 | return 1;
|
788 | }
|
789 |
|
790 | pre_tokens.clear();
|
791 | hook->TryPreprocess(line, &pre_tokens);
|
792 |
|
793 | out->Line(line_num, line, pre_tokens);
|
794 |
|
795 | line_num += 1;
|
796 | num_sig += 1;
|
797 |
|
798 | last = pre_tokens.back();
|
799 | }
|
800 | continue; // Skip the rest of the loop
|
801 | }
|
802 |
|
803 | //
|
804 | // Main Loop for "normal" lines (not preprocessor or here doc)
|
805 | //
|
806 |
|
807 | std::vector<Token> tokens;
|
808 | lexer.SetLine(line);
|
809 |
|
810 | bool line_is_sig = false;
|
811 | while (true) { // tokens on each line
|
812 | Token tok;
|
813 | bool eol = matcher.Match(&lexer, &tok);
|
814 | if (eol) {
|
815 | break;
|
816 | }
|
817 |
|
818 | switch (tok.id) {
|
819 | case Id::HereBegin: {
|
820 | // Put a copy on the stack
|
821 | int n = tok.submatch_end - tok.submatch_start;
|
822 | here_list.emplace_back(line + tok.submatch_start, n);
|
823 | here_start_num.push_back(line_num);
|
824 | } break;
|
825 |
|
826 | case Id::DelimStrBegin: {
|
827 | if (delim_begin.empty()) {
|
828 | int n = tok.submatch_end - tok.submatch_start;
|
829 | delim_begin.emplace_back(line + tok.submatch_start, n);
|
830 | } else {
|
831 | // We have entered cpp_mode_e::DelimStr, which means we should never
|
832 | // return another DelimStrBegin
|
833 | assert(0);
|
834 | }
|
835 | } break;
|
836 |
|
837 | case Id::DelimStrEnd: {
|
838 | if (delim_begin.empty()) {
|
839 | // We should never get this unless we got a DelimStrBegin first
|
840 | assert(0);
|
841 | } else {
|
842 | size_t n = tok.submatch_end - tok.submatch_start;
|
843 | std::string end_delim(line + tok.submatch_start, n);
|
844 |
|
845 | if (end_delim == delim_begin.back()) {
|
846 | lexer.line_mode = T::Outer; // the string is ended
|
847 | delim_begin.pop_back();
|
848 | } else {
|
849 | tok.id = Id::Str; // mismatched delimiter is just a string
|
850 | }
|
851 | }
|
852 | } break;
|
853 |
|
854 | // TODO: I think we need a mode to escape into strstr(), for
|
855 | // C++ - ending */
|
856 | // HTML - ending --> ?> ]]> </SCRipt>
|
857 | //
|
858 | // So instead of returning 'eol', we can return a string to search for?
|
859 | // Then we keep looking for more lines.
|
860 | //
|
861 | // This is similar to the problems of here doc and C++ multi-line
|
862 | // strings. The main difference is that we're not using a submatch.
|
863 | default:
|
864 | break;
|
865 | }
|
866 |
|
867 | tokens.push_back(tok); // make a copy
|
868 |
|
869 | if (TokenIsSignificant(tok.id)) {
|
870 | line_is_sig = true;
|
871 | }
|
872 | }
|
873 |
|
874 | #if 0
|
875 | PrintTokens(tokens);
|
876 | Log("%d tokens before", tokens.size());
|
877 | Optimize(&tokens);
|
878 | Log("%d tokens after", tokens.size());
|
879 | PrintTokens(tokens);
|
880 | #endif
|
881 |
|
882 | FixShellComments(tokens);
|
883 |
|
884 | out->Line(line_num, line, tokens);
|
885 | tokens.clear();
|
886 |
|
887 | // Potentially multiple here docs for this line
|
888 | int here_index = 0;
|
889 | for (auto here_delim : here_list) {
|
890 | // Log("HERE %s", here_delim.c_str());
|
891 |
|
892 | while (true) {
|
893 | const char* blame = reader->Filename() ?: "<stdin>";
|
894 | if (!reader->NextLine()) {
|
895 | Log("micro-syntax: getline() error on %s: %s", blame,
|
896 | strerror(reader->err_num_));
|
897 | return 1;
|
898 | }
|
899 | char* line = reader->Current();
|
900 | if (line == nullptr) {
|
901 | int start_line = here_start_num[here_index];
|
902 | Log("Unexpected end-of-file in here doc in %s, start line %d", blame,
|
903 | start_line);
|
904 | return 1;
|
905 | }
|
906 |
|
907 | line_num++;
|
908 |
|
909 | if (LineEqualsHereDelim(line, here_delim)) {
|
910 | int n = strlen(line);
|
911 | Token whole_line(Id::HereEnd, n);
|
912 | tokens.push_back(whole_line);
|
913 | out->Line(line_num, line, tokens);
|
914 | tokens.clear();
|
915 | break;
|
916 |
|
917 | } else {
|
918 | int n = strlen(line);
|
919 | Token whole_line(Id::Str, n);
|
920 | tokens.push_back(whole_line);
|
921 | out->Line(line_num, line, tokens);
|
922 | tokens.clear();
|
923 |
|
924 | // Log(" not equal: %s", line);
|
925 | }
|
926 | }
|
927 | here_index++;
|
928 | }
|
929 | here_list.clear();
|
930 | here_start_num.clear();
|
931 |
|
932 | line_num++;
|
933 | num_sig += line_is_sig;
|
934 | }
|
935 |
|
936 | out->PathEnd(line_num - 1, num_sig);
|
937 | return 0;
|
938 | }
|
939 |
|
940 | int ScanFiles(const Flags& flag, std::vector<char*> files, OutputStream* out,
|
941 | Hook* hook) {
|
942 | Reader* reader = nullptr;
|
943 |
|
944 | int status = 0;
|
945 | for (auto path : files) {
|
946 | FILE* f;
|
947 | if (path == nullptr) {
|
948 | f = stdin;
|
949 | } else {
|
950 | f = fopen(path, "r");
|
951 | if (f == nullptr) {
|
952 | Log("Error opening %s: %s", path, strerror(errno));
|
953 | return 1;
|
954 | }
|
955 | }
|
956 | out->PathBegin(path);
|
957 |
|
958 | reader = new Reader(f, path);
|
959 |
|
960 | switch (flag.lang) {
|
961 | case lang_e::PlainText:
|
962 | status = ScanOne<text_mode_e>(reader, out, hook);
|
963 | break;
|
964 |
|
965 | case lang_e::Py:
|
966 | status = ScanOne<py_mode_e>(reader, out, hook);
|
967 | break;
|
968 |
|
969 | case lang_e::Cpp:
|
970 | status = ScanOne<cpp_mode_e>(reader, out, hook);
|
971 | break;
|
972 |
|
973 | case lang_e::Shell:
|
974 | status = ScanOne<sh_mode_e>(reader, out, hook);
|
975 | break;
|
976 |
|
977 | case lang_e::Asdl:
|
978 | status = ScanOne<asdl_mode_e>(reader, out, hook);
|
979 | break;
|
980 |
|
981 | case lang_e::R:
|
982 | status = ScanOne<R_mode_e>(reader, out, hook);
|
983 | break;
|
984 |
|
985 | case lang_e::Html:
|
986 | status = ScanOne<html_mode_e>(reader, out, hook);
|
987 | break;
|
988 |
|
989 | default:
|
990 | assert(0);
|
991 | }
|
992 |
|
993 | delete reader;
|
994 |
|
995 | if (path == nullptr) {
|
996 | ;
|
997 | } else {
|
998 | fclose(f);
|
999 | }
|
1000 |
|
1001 | if (status != 0) {
|
1002 | break;
|
1003 | }
|
1004 | }
|
1005 |
|
1006 | return status;
|
1007 | }
|
1008 |
|
1009 | void PrintHelp() {
|
1010 | puts(R"(Usage: micro-syntax FLAGS* FILE*
|
1011 |
|
1012 | Recognizes the syntax of each file,, and prints it to stdout.
|
1013 |
|
1014 | If there are no files, reads stdin.
|
1015 |
|
1016 | Flags:
|
1017 | -h --help This help
|
1018 |
|
1019 | -l --lang Language: py|cpp|shell|...
|
1020 | -t Print tokens as TSV, instead of ANSI color
|
1021 | -w Print HTML for the web
|
1022 |
|
1023 | -m More color, useful for debugging tokens
|
1024 |
|
1025 | -n --no-comments Omit comments
|
1026 | -o --comments-only Only print comments
|
1027 | -e --empty-strs Substitute string literals for empty strings
|
1028 | --color on off always more
|
1029 |
|
1030 | )");
|
1031 | }
|
1032 |
|
1033 | int main(int argc, char** argv) {
|
1034 | Flags flag = {lang_e::PlainText};
|
1035 |
|
1036 | // http://www.gnu.org/software/libc/manual/html_node/Example-of-Getopt.html
|
1037 | // + means to be strict about flag parsing.
|
1038 | int c;
|
1039 | while ((c = getopt(argc, argv, "+hl:motw")) != -1) {
|
1040 | switch (c) {
|
1041 | case 'h':
|
1042 | PrintHelp();
|
1043 | return 0;
|
1044 |
|
1045 | case 'l':
|
1046 | if (strcmp(optarg, "cpp") == 0) {
|
1047 | flag.lang = lang_e::Cpp;
|
1048 |
|
1049 | } else if (strcmp(optarg, "py") == 0) {
|
1050 | flag.lang = lang_e::Py;
|
1051 |
|
1052 | } else if (strcmp(optarg, "shell") == 0) {
|
1053 | flag.lang = lang_e::Shell;
|
1054 |
|
1055 | } else if (strcmp(optarg, "asdl") == 0) {
|
1056 | flag.lang = lang_e::Asdl;
|
1057 |
|
1058 | } else if (strcmp(optarg, "R") == 0) {
|
1059 | flag.lang = lang_e::R;
|
1060 |
|
1061 | // TODO: implement all of these
|
1062 | } else if (strcmp(optarg, "js") == 0) {
|
1063 | flag.lang = lang_e::PlainText;
|
1064 |
|
1065 | } else if (strcmp(optarg, "css") == 0) {
|
1066 | flag.lang = lang_e::PlainText;
|
1067 |
|
1068 | } else if (strcmp(optarg, "md") == 0) {
|
1069 | flag.lang = lang_e::PlainText;
|
1070 |
|
1071 | } else if (strcmp(optarg, "yaml") == 0) {
|
1072 | flag.lang = lang_e::PlainText;
|
1073 |
|
1074 | } else if (strcmp(optarg, "html") == 0) {
|
1075 | flag.lang = lang_e::Html;
|
1076 |
|
1077 | } else if (strcmp(optarg, "txt") == 0) {
|
1078 | flag.lang = lang_e::PlainText;
|
1079 |
|
1080 | } else if (strcmp(optarg, "other") == 0) {
|
1081 | flag.lang = lang_e::PlainText;
|
1082 |
|
1083 | } else {
|
1084 | Log("Expected -l LANG to be "
|
1085 | "cpp|py|shell|asdl|R|js|css|md|yaml|html|txt, "
|
1086 | "got %s",
|
1087 | optarg);
|
1088 | return 2;
|
1089 | }
|
1090 | break;
|
1091 |
|
1092 | case 'm':
|
1093 | flag.more_color = true;
|
1094 | break;
|
1095 |
|
1096 | case 'o':
|
1097 | flag.comments_only = true;
|
1098 | break;
|
1099 |
|
1100 | case 't':
|
1101 | flag.tsv = true;
|
1102 | break;
|
1103 |
|
1104 | case 'w':
|
1105 | flag.web = true;
|
1106 | break;
|
1107 |
|
1108 | case '?': // getopt library will print error
|
1109 | return 2;
|
1110 |
|
1111 | default:
|
1112 | abort(); // should never happen
|
1113 | }
|
1114 | }
|
1115 |
|
1116 | int a = optind; // index into argv
|
1117 | flag.argv = argv + a;
|
1118 | flag.argc = argc - a;
|
1119 |
|
1120 | std::vector<char*> files; // filename, or nullptr for stdin
|
1121 | if (flag.argc != 0) {
|
1122 | for (int i = 0; i < flag.argc; ++i) {
|
1123 | files.push_back(flag.argv[i]);
|
1124 | }
|
1125 | } else {
|
1126 | files.push_back(nullptr); // stands for stdin
|
1127 | }
|
1128 |
|
1129 | Printer* pr; // for each file
|
1130 | OutputStream* out; // the entire stream
|
1131 |
|
1132 | if (flag.tsv) {
|
1133 | pr = new TsvPrinter();
|
1134 | out = new NetStringOutput(pr);
|
1135 | } else if (flag.web) {
|
1136 | pr = new HtmlPrinter();
|
1137 | out = new NetStringOutput(pr);
|
1138 | } else {
|
1139 | pr = new AnsiPrinter(flag);
|
1140 | out = new AnsiOutput(pr);
|
1141 | }
|
1142 |
|
1143 | Hook* hook = nullptr;
|
1144 | if (flag.lang == lang_e::Cpp) {
|
1145 | hook = new CppHook();
|
1146 | } else {
|
1147 | hook = new Hook(); // default hook
|
1148 | }
|
1149 |
|
1150 | int status = ScanFiles(flag, files, out, hook);
|
1151 |
|
1152 | delete hook;
|
1153 | delete pr;
|
1154 | delete out;
|
1155 |
|
1156 | return status;
|
1157 | }
|