1 | /* $Id: scmparser.cpp 93115 2022-01-01 11:31:46Z vboxsync $ */
|
---|
2 | /** @file
|
---|
3 | * IPRT Testcase / Tool - Source Code Massager, Code Parsers.
|
---|
4 | */
|
---|
5 |
|
---|
6 | /*
|
---|
7 | * Copyright (C) 2010-2022 Oracle Corporation
|
---|
8 | *
|
---|
9 | * This file is part of VirtualBox Open Source Edition (OSE), as
|
---|
10 | * available from http://www.virtualbox.org. This file is free software;
|
---|
11 | * you can redistribute it and/or modify it under the terms of the GNU
|
---|
12 | * General Public License (GPL) as published by the Free Software
|
---|
13 | * Foundation, in version 2 as it comes in the "COPYING" file of the
|
---|
14 | * VirtualBox OSE distribution. VirtualBox OSE is distributed in the
|
---|
15 | * hope that it will be useful, but WITHOUT ANY WARRANTY of any kind.
|
---|
16 | */
|
---|
17 |
|
---|
18 |
|
---|
19 | /*********************************************************************************************************************************
|
---|
20 | * Header Files *
|
---|
21 | *********************************************************************************************************************************/
|
---|
22 | #include <iprt/assert.h>
|
---|
23 | #include <iprt/ctype.h>
|
---|
24 | #include <iprt/dir.h>
|
---|
25 | #include <iprt/env.h>
|
---|
26 | #include <iprt/file.h>
|
---|
27 | #include <iprt/errcore.h>
|
---|
28 | #include <iprt/getopt.h>
|
---|
29 | #include <iprt/initterm.h>
|
---|
30 | #include <iprt/mem.h>
|
---|
31 | #include <iprt/message.h>
|
---|
32 | #include <iprt/param.h>
|
---|
33 | #include <iprt/path.h>
|
---|
34 | #include <iprt/process.h>
|
---|
35 | #include <iprt/stream.h>
|
---|
36 | #include <iprt/string.h>
|
---|
37 |
|
---|
38 | #include "scm.h"
|
---|
39 |
|
---|
40 |
|
---|
41 | /*********************************************************************************************************************************
|
---|
42 | * Structures and Typedefs *
|
---|
43 | *********************************************************************************************************************************/
|
---|
44 | typedef size_t (*PFNISCOMMENT)(const char *pchLine, size_t cchLine, bool fSecond);
|
---|
45 |
|
---|
46 |
|
---|
47 | /**
|
---|
48 | * Callback for checking if C++ line comment.
|
---|
49 | */
|
---|
50 | static size_t isCppLineComment(const char *pchLine, size_t cchLine, bool fSecond)
|
---|
51 | {
|
---|
52 | if ( cchLine >= 2
|
---|
53 | && pchLine[0] == '/'
|
---|
54 | && pchLine[1] == '/')
|
---|
55 | {
|
---|
56 | if (!fSecond)
|
---|
57 | return 2;
|
---|
58 | if (cchLine >= 3 && pchLine[2] == '/')
|
---|
59 | return 3;
|
---|
60 | }
|
---|
61 | return 0;
|
---|
62 | }
|
---|
63 |
|
---|
64 |
|
---|
65 | /**
|
---|
66 | * Callback for checking if hash comment.
|
---|
67 | */
|
---|
68 | static size_t isHashComment(const char *pchLine, size_t cchLine, bool fSecond)
|
---|
69 | {
|
---|
70 | if (cchLine >= 1 && *pchLine == '#')
|
---|
71 | {
|
---|
72 | if (!fSecond)
|
---|
73 | return 1;
|
---|
74 | if (cchLine >= 2 && pchLine[1] == '#')
|
---|
75 | return 2;
|
---|
76 | }
|
---|
77 | return 0;
|
---|
78 | }
|
---|
79 |
|
---|
80 |
|
---|
81 | /**
|
---|
82 | * Callback for checking if semicolon comment.
|
---|
83 | */
|
---|
84 | static size_t isSemicolonComment(const char *pchLine, size_t cchLine, bool fSecond)
|
---|
85 | {
|
---|
86 | if (cchLine >= 1 && *pchLine == ';')
|
---|
87 | {
|
---|
88 | if (!fSecond)
|
---|
89 | return 1;
|
---|
90 | if (cchLine >= 2 && pchLine[1] == ';')
|
---|
91 | return 2;
|
---|
92 | }
|
---|
93 | return 0;
|
---|
94 | }
|
---|
95 |
|
---|
96 |
|
---|
97 | /** Macro for checking for a batch file comment prefix. */
|
---|
98 | #define IS_REM(a_pch, a_off, a_cch) \
|
---|
99 | ( (a_off) + 3 <= (a_cch) \
|
---|
100 | && ((a_pch)[(a_off) ] == 'R' || (a_pch)[(a_off) ] == 'r') \
|
---|
101 | && ((a_pch)[(a_off) + 1] == 'E' || (a_pch)[(a_off) + 1] == 'e') \
|
---|
102 | && ((a_pch)[(a_off) + 2] == 'M' || (a_pch)[(a_off) + 2] == 'm') \
|
---|
103 | && ((a_off) + 3 == (a_cch) || RT_C_IS_SPACE((a_pch)[(a_off) + 3])) )
|
---|
104 |
|
---|
105 |
|
---|
106 | /**
|
---|
107 | * Callback for checking if batch comment.
|
---|
108 | */
|
---|
109 | static size_t isBatchComment(const char *pchLine, size_t cchLine, bool fSecond)
|
---|
110 | {
|
---|
111 | if (!fSecond)
|
---|
112 | {
|
---|
113 | if (IS_REM(pchLine, 0, cchLine))
|
---|
114 | return 3;
|
---|
115 | }
|
---|
116 | else
|
---|
117 | {
|
---|
118 | /* Check for the 2nd in "rem rem" lines. */
|
---|
119 | if ( cchLine >= 4
|
---|
120 | && RT_C_IS_SPACE(*pchLine)
|
---|
121 | && IS_REM(pchLine, 1, cchLine))
|
---|
122 | return 4;
|
---|
123 | }
|
---|
124 | return 0;
|
---|
125 | }
|
---|
126 |
|
---|
127 | /**
|
---|
128 | * Callback for checking if SQL comment.
|
---|
129 | */
|
---|
130 | static size_t isSqlComment(const char *pchLine, size_t cchLine, bool fSecond)
|
---|
131 | {
|
---|
132 | if ( cchLine >= 2
|
---|
133 | && pchLine[0] == '-'
|
---|
134 | && pchLine[1] == '-')
|
---|
135 | {
|
---|
136 | if (!fSecond)
|
---|
137 | return 2;
|
---|
138 | if ( cchLine >= 3
|
---|
139 | && pchLine[2] == '-')
|
---|
140 | return 3;
|
---|
141 | }
|
---|
142 | return 0;
|
---|
143 | }
|
---|
144 |
|
---|
145 | /**
|
---|
146 | * Callback for checking if tick comment.
|
---|
147 | */
|
---|
148 | static size_t isTickComment(const char *pchLine, size_t cchLine, bool fSecond)
|
---|
149 | {
|
---|
150 | if (cchLine >= 1 && *pchLine == '\'')
|
---|
151 | {
|
---|
152 | if (!fSecond)
|
---|
153 | return 1;
|
---|
154 | if (cchLine >= 2 && pchLine[1] == '\'')
|
---|
155 | return 2;
|
---|
156 | }
|
---|
157 | return 0;
|
---|
158 | }
|
---|
159 |
|
---|
160 |
|
---|
161 | /**
|
---|
162 | * Common worker for enumeratePythonComments and enumerateSimpleLineComments.
|
---|
163 | *
|
---|
164 | * @returns IPRT status code.
|
---|
165 | * @param pIn The input stream.
|
---|
166 | * @param pfnIsComment Comment tester function.
|
---|
167 | * @param pfnCallback The callback.
|
---|
168 | * @param pvUser The user argument for the callback.
|
---|
169 | * @param ppchLine Pointer to the line variable.
|
---|
170 | * @param pcchLine Pointer to the line length variable.
|
---|
171 | * @param penmEol Pointer to the line ending type variable.
|
---|
172 | * @param piLine Pointer to the line number variable.
|
---|
173 | * @param poff Pointer to the line offset variable. On input this
|
---|
174 | * is positioned at the start of the comment.
|
---|
175 | */
|
---|
176 | static int handleLineComment(PSCMSTREAM pIn, PFNISCOMMENT pfnIsComment,
|
---|
177 | PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser,
|
---|
178 | const char **ppchLine, size_t *pcchLine, PSCMEOL penmEol,
|
---|
179 | uint32_t *piLine, size_t *poff)
|
---|
180 | {
|
---|
181 | /* Unpack input/output variables. */
|
---|
182 | uint32_t iLine = *piLine;
|
---|
183 | const char *pchLine = *ppchLine;
|
---|
184 | size_t cchLine = *pcchLine;
|
---|
185 | size_t off = *poff;
|
---|
186 | SCMEOL enmEol = *penmEol;
|
---|
187 |
|
---|
188 | /*
|
---|
189 | * Take down the basic info about the comment.
|
---|
190 | */
|
---|
191 | SCMCOMMENTINFO Info;
|
---|
192 | Info.iLineStart = iLine;
|
---|
193 | Info.iLineEnd = iLine;
|
---|
194 | Info.offStart = (uint32_t)off;
|
---|
195 | Info.offEnd = (uint32_t)cchLine;
|
---|
196 |
|
---|
197 | size_t cchSkip = pfnIsComment(&pchLine[off], cchLine - off, false);
|
---|
198 | Assert(cchSkip > 0);
|
---|
199 | off += cchSkip;
|
---|
200 |
|
---|
201 | /* Determin comment type. */
|
---|
202 | Info.enmType = kScmCommentType_Line;
|
---|
203 | char ch;
|
---|
204 | cchSkip = 1;
|
---|
205 | if ( off < cchLine
|
---|
206 | && ( (ch = pchLine[off]) == '!'
|
---|
207 | || (cchSkip = pfnIsComment(&pchLine[off], cchLine - off, true)) > 0) )
|
---|
208 | {
|
---|
209 | unsigned ch2;
|
---|
210 | if ( off + cchSkip == cchLine
|
---|
211 | || RT_C_IS_SPACE(ch2 = pchLine[off + cchSkip]) )
|
---|
212 | {
|
---|
213 | Info.enmType = ch != '!' ? kScmCommentType_Line_JavaDoc : kScmCommentType_Line_Qt;
|
---|
214 | off += cchSkip;
|
---|
215 | }
|
---|
216 | else if ( ch2 == '<'
|
---|
217 | && ( off + cchSkip + 1 == cchLine
|
---|
218 | || RT_C_IS_SPACE(pchLine[off + cchSkip + 1]) ))
|
---|
219 | {
|
---|
220 | Info.enmType = ch == '!' ? kScmCommentType_Line_JavaDoc_After : kScmCommentType_Line_Qt_After;
|
---|
221 | off += cchSkip + 1;
|
---|
222 | }
|
---|
223 | }
|
---|
224 |
|
---|
225 | /*
|
---|
226 | * Copy body of the first line. Like for C, we ignore a single space in the first comment line.
|
---|
227 | */
|
---|
228 | if (off < cchLine && RT_C_IS_SPACE(pchLine[off]))
|
---|
229 | off++;
|
---|
230 | size_t cchBody = cchLine;
|
---|
231 | while (cchBody > off && RT_C_IS_SPACE(pchLine[cchBody - 1]))
|
---|
232 | cchBody--;
|
---|
233 | cchBody -= off;
|
---|
234 | size_t cbBodyAlloc = RT_MAX(_1K, RT_ALIGN_Z(cchBody + 64, 128));
|
---|
235 | char *pszBody = (char *)RTMemAlloc(cbBodyAlloc);
|
---|
236 | if (!pszBody)
|
---|
237 | return VERR_NO_MEMORY;
|
---|
238 | memcpy(pszBody, &pchLine[off], cchBody);
|
---|
239 | pszBody[cchBody] = '\0';
|
---|
240 |
|
---|
241 | Info.cBlankLinesBefore = cchBody == 0;
|
---|
242 |
|
---|
243 | /*
|
---|
244 | * Look for more comment lines and append them to the body.
|
---|
245 | */
|
---|
246 | while ((pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol)) != NULL)
|
---|
247 | {
|
---|
248 | iLine++;
|
---|
249 |
|
---|
250 | /* Skip leading spaces. */
|
---|
251 | off = 0;
|
---|
252 | while (off < cchLine && RT_C_IS_SPACE(pchLine[off]))
|
---|
253 | off++;
|
---|
254 |
|
---|
255 | /* Check if it's a comment. */
|
---|
256 | if ( off >= cchLine
|
---|
257 | || (cchSkip = pfnIsComment(&pchLine[off], cchLine - off, false)) == 0)
|
---|
258 | break;
|
---|
259 | off += cchSkip;
|
---|
260 |
|
---|
261 | /* Split on doxygen comment start (if not already in one). */
|
---|
262 | if ( Info.enmType == kScmCommentType_Line
|
---|
263 | && off + 1 < cchLine
|
---|
264 | && ( pfnIsComment(&pchLine[off], cchLine - off, true) > 0
|
---|
265 | || ( pchLine[off + 1] == '!'
|
---|
266 | && ( off + 2 == cchLine
|
---|
267 | || pchLine[off + 2] != '!') ) ) )
|
---|
268 | {
|
---|
269 | off -= cchSkip;
|
---|
270 | break;
|
---|
271 | }
|
---|
272 |
|
---|
273 | /* Append the body w/o trailing spaces and some leading ones. */
|
---|
274 | if (off < cchLine && RT_C_IS_SPACE(pchLine[off]))
|
---|
275 | off++;
|
---|
276 | while (off < cchLine && off < Info.offStart + 3 && RT_C_IS_SPACE(pchLine[off]))
|
---|
277 | off++;
|
---|
278 | size_t cchAppend = cchLine;
|
---|
279 | while (cchAppend > off && RT_C_IS_SPACE(pchLine[cchAppend - 1]))
|
---|
280 | cchAppend--;
|
---|
281 | cchAppend -= off;
|
---|
282 |
|
---|
283 | size_t cchNewBody = cchBody + 1 + cchAppend;
|
---|
284 | if (cchNewBody >= cbBodyAlloc)
|
---|
285 | {
|
---|
286 | cbBodyAlloc = RT_MAX(cbBodyAlloc ? cbBodyAlloc * 2 : _1K, RT_ALIGN_Z(cchNewBody + 64, 128));
|
---|
287 | void *pvNew = RTMemRealloc(pszBody, cbBodyAlloc);
|
---|
288 | if (pvNew)
|
---|
289 | pszBody = (char *)pvNew;
|
---|
290 | else
|
---|
291 | {
|
---|
292 | RTMemFree(pszBody);
|
---|
293 | return VERR_NO_MEMORY;
|
---|
294 | }
|
---|
295 | }
|
---|
296 |
|
---|
297 | if ( cchBody > 0
|
---|
298 | || cchAppend > 0)
|
---|
299 | {
|
---|
300 | if (cchBody > 0)
|
---|
301 | pszBody[cchBody++] = '\n';
|
---|
302 | memcpy(&pszBody[cchBody], &pchLine[off], cchAppend);
|
---|
303 | cchBody += cchAppend;
|
---|
304 | pszBody[cchBody] = '\0';
|
---|
305 | }
|
---|
306 | else
|
---|
307 | Info.cBlankLinesBefore++;
|
---|
308 |
|
---|
309 | /* Advance. */
|
---|
310 | Info.offEnd = (uint32_t)cchLine;
|
---|
311 | Info.iLineEnd = iLine;
|
---|
312 | }
|
---|
313 |
|
---|
314 | /*
|
---|
315 | * Strip trailing empty lines in the body.
|
---|
316 | */
|
---|
317 | Info.cBlankLinesAfter = 0;
|
---|
318 | while (cchBody >= 1 && pszBody[cchBody - 1] == '\n')
|
---|
319 | {
|
---|
320 | Info.cBlankLinesAfter++;
|
---|
321 | pszBody[--cchBody] = '\0';
|
---|
322 | }
|
---|
323 |
|
---|
324 | /*
|
---|
325 | * Do the callback and return.
|
---|
326 | */
|
---|
327 | int rc = pfnCallback(&Info, pszBody, cchBody, pvUser);
|
---|
328 |
|
---|
329 | RTMemFree(pszBody);
|
---|
330 |
|
---|
331 | *piLine = iLine;
|
---|
332 | *ppchLine = pchLine;
|
---|
333 | *pcchLine = cchLine;
|
---|
334 | *poff = off;
|
---|
335 | *penmEol = enmEol;
|
---|
336 | return rc;
|
---|
337 | }
|
---|
338 |
|
---|
339 |
|
---|
340 |
|
---|
341 | /**
|
---|
342 | * Common string litteral handler.
|
---|
343 | *
|
---|
344 | * @returns new pchLine value.
|
---|
345 | * @param pIn The input string.
|
---|
346 | * @param chType The quotation type.
|
---|
347 | * @param pchLine The current line.
|
---|
348 | * @param ppchLine Pointer to the line variable.
|
---|
349 | * @param pcchLine Pointer to the line length variable.
|
---|
350 | * @param penmEol Pointer to the line ending type variable.
|
---|
351 | * @param piLine Pointer to the line number variable.
|
---|
352 | * @param poff Pointer to the line offset variable.
|
---|
353 | */
|
---|
354 | static const char *handleStringLitteral(PSCMSTREAM pIn, char chType, const char *pchLine, size_t *pcchLine, PSCMEOL penmEol,
|
---|
355 | uint32_t *piLine, size_t *poff)
|
---|
356 | {
|
---|
357 | size_t off = *poff;
|
---|
358 | for (;;)
|
---|
359 | {
|
---|
360 | bool fEnd = false;
|
---|
361 | bool fEscaped = false;
|
---|
362 | size_t const cchLine = *pcchLine;
|
---|
363 | while (off < cchLine)
|
---|
364 | {
|
---|
365 | char ch = pchLine[off++];
|
---|
366 | if (!fEscaped)
|
---|
367 | {
|
---|
368 | if (ch != chType)
|
---|
369 | {
|
---|
370 | if (ch != '\\')
|
---|
371 | { /* likely */ }
|
---|
372 | else
|
---|
373 | fEscaped = true;
|
---|
374 | }
|
---|
375 | else
|
---|
376 | {
|
---|
377 | fEnd = true;
|
---|
378 | break;
|
---|
379 | }
|
---|
380 | }
|
---|
381 | else
|
---|
382 | fEscaped = false;
|
---|
383 | }
|
---|
384 | if (fEnd)
|
---|
385 | break;
|
---|
386 |
|
---|
387 | /* next line */
|
---|
388 | pchLine = ScmStreamGetLine(pIn, pcchLine, penmEol);
|
---|
389 | if (!pchLine)
|
---|
390 | break;
|
---|
391 | *piLine += 1;
|
---|
392 | off = 0;
|
---|
393 | }
|
---|
394 |
|
---|
395 | *poff = off;
|
---|
396 | return pchLine;
|
---|
397 | }
|
---|
398 |
|
---|
399 |
|
---|
400 | /**
|
---|
401 | * Deals with comments in C and C++ code.
|
---|
402 | *
|
---|
403 | * @returns VBox status code / callback return code.
|
---|
404 | * @param pIn The stream to parse.
|
---|
405 | * @param pfnCallback The callback.
|
---|
406 | * @param pvUser The user parameter for the callback.
|
---|
407 | */
|
---|
408 | static int enumerateCStyleComments(PSCMSTREAM pIn, PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
|
---|
409 | {
|
---|
410 | AssertCompile('\'' < '/');
|
---|
411 | AssertCompile('"' < '/');
|
---|
412 |
|
---|
413 | int rcRet = VINF_SUCCESS;
|
---|
414 | uint32_t iLine = 0;
|
---|
415 | SCMEOL enmEol;
|
---|
416 | size_t cchLine;
|
---|
417 | const char *pchLine;
|
---|
418 | while ((pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol)) != NULL)
|
---|
419 | {
|
---|
420 | size_t off = 0;
|
---|
421 | while (off < cchLine)
|
---|
422 | {
|
---|
423 | unsigned ch = pchLine[off++];
|
---|
424 | if (ch > (unsigned)'/')
|
---|
425 | { /* not interesting */ }
|
---|
426 | else if (ch == '/')
|
---|
427 | {
|
---|
428 | if (off < cchLine)
|
---|
429 | {
|
---|
430 | ch = pchLine[off++];
|
---|
431 | if (ch == '*')
|
---|
432 | {
|
---|
433 | /*
|
---|
434 | * Multiline comment. Find the end.
|
---|
435 | *
|
---|
436 | * Note! This is very similar to the python doc string handling further down.
|
---|
437 | */
|
---|
438 | SCMCOMMENTINFO Info;
|
---|
439 | Info.iLineStart = iLine;
|
---|
440 | Info.offStart = (uint32_t)off - 2;
|
---|
441 | Info.iLineEnd = UINT32_MAX;
|
---|
442 | Info.offEnd = UINT32_MAX;
|
---|
443 | Info.cBlankLinesBefore = 0;
|
---|
444 |
|
---|
445 | /* Determin comment type (same as for line-comments). */
|
---|
446 | Info.enmType = kScmCommentType_MultiLine;
|
---|
447 | if ( off < cchLine
|
---|
448 | && ( (ch = pchLine[off]) == '*'
|
---|
449 | || ch == '!') )
|
---|
450 | {
|
---|
451 | unsigned ch2;
|
---|
452 | if ( off + 1 == cchLine
|
---|
453 | || RT_C_IS_SPACE(ch2 = pchLine[off + 1]) )
|
---|
454 | {
|
---|
455 | Info.enmType = ch == '*' ? kScmCommentType_MultiLine_JavaDoc : kScmCommentType_MultiLine_Qt;
|
---|
456 | off += 1;
|
---|
457 | }
|
---|
458 | else if ( ch2 == '<'
|
---|
459 | && ( off + 2 == cchLine
|
---|
460 | || RT_C_IS_SPACE(pchLine[off + 2]) ))
|
---|
461 | {
|
---|
462 | Info.enmType = ch == '*' ? kScmCommentType_MultiLine_JavaDoc_After
|
---|
463 | : kScmCommentType_MultiLine_Qt_After;
|
---|
464 | off += 2;
|
---|
465 | }
|
---|
466 | }
|
---|
467 |
|
---|
468 | /*
|
---|
469 | * Copy the body and find the end of the multiline comment.
|
---|
470 | */
|
---|
471 | size_t cbBodyAlloc = 0;
|
---|
472 | size_t cchBody = 0;
|
---|
473 | char *pszBody = NULL;
|
---|
474 | for (;;)
|
---|
475 | {
|
---|
476 | /* Parse the line up to the end-of-comment or end-of-line. */
|
---|
477 | size_t offLineStart = off;
|
---|
478 | size_t offLastNonBlank = off;
|
---|
479 | size_t offFirstNonBlank = ~(size_t)0;
|
---|
480 | while (off < cchLine)
|
---|
481 | {
|
---|
482 | ch = pchLine[off++];
|
---|
483 | if (ch != '*' || off >= cchLine || pchLine[off] != '/')
|
---|
484 | {
|
---|
485 | if (RT_C_IS_BLANK(ch))
|
---|
486 | {/* kind of likely */}
|
---|
487 | else
|
---|
488 | {
|
---|
489 | offLastNonBlank = off - 1;
|
---|
490 | if (offFirstNonBlank != ~(size_t)0)
|
---|
491 | {/* likely */}
|
---|
492 | else if ( ch != '*' /* ignore continuation-asterisks */
|
---|
493 | || off > Info.offStart + 1 + 1
|
---|
494 | || off > cchLine
|
---|
495 | || ( off < cchLine
|
---|
496 | && !RT_C_IS_SPACE(pchLine[off]))
|
---|
497 | || pszBody == NULL)
|
---|
498 | offFirstNonBlank = off - 1;
|
---|
499 | }
|
---|
500 | }
|
---|
501 | else
|
---|
502 | {
|
---|
503 | Info.offEnd = (uint32_t)++off;
|
---|
504 | Info.iLineEnd = iLine;
|
---|
505 | break;
|
---|
506 | }
|
---|
507 | }
|
---|
508 |
|
---|
509 | /* Append line content to the comment body string. */
|
---|
510 | size_t cchAppend;
|
---|
511 | if (offFirstNonBlank == ~(size_t)0)
|
---|
512 | cchAppend = 0; /* empty line */
|
---|
513 | else
|
---|
514 | {
|
---|
515 | if (pszBody)
|
---|
516 | offLineStart = RT_MIN(Info.offStart + 3, offFirstNonBlank);
|
---|
517 | else if (offFirstNonBlank > Info.offStart + 2) /* Skip one leading blank at the start of the comment. */
|
---|
518 | offLineStart++;
|
---|
519 | cchAppend = offLastNonBlank + 1 - offLineStart;
|
---|
520 | Assert(cchAppend <= cchLine);
|
---|
521 | }
|
---|
522 |
|
---|
523 | size_t cchNewBody = cchBody + (cchBody > 0) + cchAppend;
|
---|
524 | if (cchNewBody >= cbBodyAlloc)
|
---|
525 | {
|
---|
526 | cbBodyAlloc = RT_MAX(cbBodyAlloc ? cbBodyAlloc * 2 : _1K, RT_ALIGN_Z(cchNewBody + 64, 128));
|
---|
527 | void *pvNew = RTMemRealloc(pszBody, cbBodyAlloc);
|
---|
528 | if (pvNew)
|
---|
529 | pszBody = (char *)pvNew;
|
---|
530 | else
|
---|
531 | {
|
---|
532 | RTMemFree(pszBody);
|
---|
533 | return VERR_NO_MEMORY;
|
---|
534 | }
|
---|
535 | }
|
---|
536 |
|
---|
537 | if (cchBody > 0) /* no leading blank lines */
|
---|
538 | pszBody[cchBody++] = '\n';
|
---|
539 | else if (cchAppend == 0)
|
---|
540 | Info.cBlankLinesBefore++;
|
---|
541 | memcpy(&pszBody[cchBody], &pchLine[offLineStart], cchAppend);
|
---|
542 | cchBody += cchAppend;
|
---|
543 | pszBody[cchBody] = '\0';
|
---|
544 |
|
---|
545 | /* Advance to the next line, if we haven't yet seen the end of this comment. */
|
---|
546 | if (Info.iLineEnd != UINT32_MAX)
|
---|
547 | break;
|
---|
548 | pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
|
---|
549 | if (!pchLine)
|
---|
550 | {
|
---|
551 | Info.offEnd = (uint32_t)cchLine;
|
---|
552 | Info.iLineEnd = iLine;
|
---|
553 | break;
|
---|
554 | }
|
---|
555 | iLine++;
|
---|
556 | off = 0;
|
---|
557 | }
|
---|
558 |
|
---|
559 | /* Strip trailing empty lines in the body. */
|
---|
560 | Info.cBlankLinesAfter = 0;
|
---|
561 | while (cchBody >= 1 && pszBody[cchBody - 1] == '\n')
|
---|
562 | {
|
---|
563 | Info.cBlankLinesAfter++;
|
---|
564 | pszBody[--cchBody] = '\0';
|
---|
565 | }
|
---|
566 |
|
---|
567 | /* Do the callback. */
|
---|
568 | int rc = pfnCallback(&Info, pszBody, cchBody, pvUser);
|
---|
569 | RTMemFree(pszBody);
|
---|
570 | if (RT_FAILURE(rc))
|
---|
571 | return rc;
|
---|
572 | if (rc > VINF_SUCCESS && rcRet == VINF_SUCCESS)
|
---|
573 | rcRet = rc;
|
---|
574 | }
|
---|
575 | else if (ch == '/')
|
---|
576 | {
|
---|
577 | /*
|
---|
578 | * Line comment. Join the other line comment guys.
|
---|
579 | */
|
---|
580 | off -= 2;
|
---|
581 | int rc = handleLineComment(pIn, isCppLineComment, pfnCallback, pvUser,
|
---|
582 | &pchLine, &cchLine, &enmEol, &iLine, &off);
|
---|
583 | if (RT_FAILURE(rc))
|
---|
584 | return rc;
|
---|
585 | if (rcRet == VINF_SUCCESS)
|
---|
586 | rcRet = rc;
|
---|
587 | }
|
---|
588 |
|
---|
589 | if (!pchLine)
|
---|
590 | break;
|
---|
591 | }
|
---|
592 | }
|
---|
593 | else if (ch == '"')
|
---|
594 | {
|
---|
595 | /*
|
---|
596 | * String litterals may include sequences that looks like comments. So,
|
---|
597 | * they needs special handling to avoid confusion.
|
---|
598 | */
|
---|
599 | pchLine = handleStringLitteral(pIn, '"', pchLine, &cchLine, &enmEol, &iLine, &off);
|
---|
600 | }
|
---|
601 | /* else: We don't have to deal with character litterals as these shouldn't
|
---|
602 | include comment-like sequences. */
|
---|
603 | } /* for each character in the line */
|
---|
604 |
|
---|
605 | iLine++;
|
---|
606 | } /* for each line in the stream */
|
---|
607 |
|
---|
608 | int rcStream = ScmStreamGetStatus(pIn);
|
---|
609 | if (RT_SUCCESS(rcStream))
|
---|
610 | return rcRet;
|
---|
611 | return rcStream;
|
---|
612 | }
|
---|
613 |
|
---|
614 |
|
---|
615 | /**
|
---|
616 | * Deals with comments in Python code.
|
---|
617 | *
|
---|
618 | * @returns VBox status code / callback return code.
|
---|
619 | * @param pIn The stream to parse.
|
---|
620 | * @param pfnCallback The callback.
|
---|
621 | * @param pvUser The user parameter for the callback.
|
---|
622 | */
|
---|
623 | static int enumeratePythonComments(PSCMSTREAM pIn, PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
|
---|
624 | {
|
---|
625 | AssertCompile('#' < '\'');
|
---|
626 | AssertCompile('"' < '\'');
|
---|
627 |
|
---|
628 | int rcRet = VINF_SUCCESS;
|
---|
629 | uint32_t iLine = 0;
|
---|
630 | SCMEOL enmEol;
|
---|
631 | size_t cchLine;
|
---|
632 | const char *pchLine;
|
---|
633 | SCMCOMMENTINFO Info;
|
---|
634 | while ((pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol)) != NULL)
|
---|
635 | {
|
---|
636 | size_t off = 0;
|
---|
637 | while (off < cchLine)
|
---|
638 | {
|
---|
639 | char ch = pchLine[off++];
|
---|
640 | if ((unsigned char)ch > (unsigned char)'\'')
|
---|
641 | { /* not interesting */ }
|
---|
642 | else if (ch == '#')
|
---|
643 | {
|
---|
644 | /*
|
---|
645 | * Line comment. Join paths with the others.
|
---|
646 | */
|
---|
647 | off -= 1;
|
---|
648 | int rc = handleLineComment(pIn, isHashComment, pfnCallback, pvUser,
|
---|
649 | &pchLine, &cchLine, &enmEol, &iLine, &off);
|
---|
650 | if (RT_FAILURE(rc))
|
---|
651 | return rc;
|
---|
652 | if (rcRet == VINF_SUCCESS)
|
---|
653 | rcRet = rc;
|
---|
654 |
|
---|
655 | if (!pchLine)
|
---|
656 | break;
|
---|
657 | }
|
---|
658 | else if (ch == '"' || ch == '\'')
|
---|
659 | {
|
---|
660 | /*
|
---|
661 | * String litterals may be doc strings and they may legally include hashes.
|
---|
662 | */
|
---|
663 | const char chType = ch;
|
---|
664 | if ( off + 1 >= cchLine
|
---|
665 | || pchLine[off] != chType
|
---|
666 | || pchLine[off + 1] != chType)
|
---|
667 | pchLine = handleStringLitteral(pIn, chType, pchLine, &cchLine, &enmEol, &iLine, &off);
|
---|
668 | else
|
---|
669 | {
|
---|
670 | /*
|
---|
671 | * Doc string (/ long string).
|
---|
672 | *
|
---|
673 | * Note! This is very similar to the multiline C comment handling above.
|
---|
674 | */
|
---|
675 | Info.iLineStart = iLine;
|
---|
676 | Info.offStart = (uint32_t)off - 1;
|
---|
677 | Info.iLineEnd = UINT32_MAX;
|
---|
678 | Info.offEnd = UINT32_MAX;
|
---|
679 | Info.cBlankLinesBefore = 0;
|
---|
680 | Info.enmType = kScmCommentType_DocString;
|
---|
681 |
|
---|
682 | off += 2;
|
---|
683 |
|
---|
684 | /* Copy the body and find the end of the doc string comment. */
|
---|
685 | size_t cbBodyAlloc = 0;
|
---|
686 | size_t cchBody = 0;
|
---|
687 | char *pszBody = NULL;
|
---|
688 | for (;;)
|
---|
689 | {
|
---|
690 | /* Parse the line up to the end-of-comment or end-of-line. */
|
---|
691 | size_t offLineStart = off;
|
---|
692 | size_t offLastNonBlank = off;
|
---|
693 | size_t offFirstNonBlank = ~(size_t)0;
|
---|
694 | bool fEscaped = false;
|
---|
695 | while (off < cchLine)
|
---|
696 | {
|
---|
697 | ch = pchLine[off++];
|
---|
698 | if (!fEscaped)
|
---|
699 | {
|
---|
700 | if ( off + 1 >= cchLine
|
---|
701 | || ch != chType
|
---|
702 | || pchLine[off] != chType
|
---|
703 | || pchLine[off + 1] != chType)
|
---|
704 | {
|
---|
705 | if (RT_C_IS_BLANK(ch))
|
---|
706 | {/* kind of likely */}
|
---|
707 | else
|
---|
708 | {
|
---|
709 | offLastNonBlank = off - 1;
|
---|
710 | if (offFirstNonBlank != ~(size_t)0)
|
---|
711 | {/* likely */}
|
---|
712 | else if ( ch != '*' /* ignore continuation-asterisks */
|
---|
713 | || off > Info.offStart + 1 + 1
|
---|
714 | || off > cchLine
|
---|
715 | || ( off < cchLine
|
---|
716 | && !RT_C_IS_SPACE(pchLine[off]))
|
---|
717 | || pszBody == NULL)
|
---|
718 | offFirstNonBlank = off - 1;
|
---|
719 |
|
---|
720 | if (ch != '\\')
|
---|
721 | {/* likely */ }
|
---|
722 | else
|
---|
723 | fEscaped = true;
|
---|
724 | }
|
---|
725 | }
|
---|
726 | else
|
---|
727 | {
|
---|
728 | off += 2;
|
---|
729 | Info.offEnd = (uint32_t)off;
|
---|
730 | Info.iLineEnd = iLine;
|
---|
731 | break;
|
---|
732 | }
|
---|
733 | }
|
---|
734 | else
|
---|
735 | fEscaped = false;
|
---|
736 | }
|
---|
737 |
|
---|
738 | /* Append line content to the comment body string. */
|
---|
739 | size_t cchAppend;
|
---|
740 | if (offFirstNonBlank == ~(size_t)0)
|
---|
741 | cchAppend = 0; /* empty line */
|
---|
742 | else
|
---|
743 | {
|
---|
744 | if (pszBody)
|
---|
745 | offLineStart = RT_MIN(Info.offStart + 3, offFirstNonBlank);
|
---|
746 | else if (offFirstNonBlank > Info.offStart + 2) /* Skip one leading blank at the start of the comment. */
|
---|
747 | offLineStart++;
|
---|
748 | cchAppend = offLastNonBlank + 1 - offLineStart;
|
---|
749 | Assert(cchAppend <= cchLine);
|
---|
750 | }
|
---|
751 |
|
---|
752 | size_t cchNewBody = cchBody + (cchBody > 0) + cchAppend;
|
---|
753 | if (cchNewBody >= cbBodyAlloc)
|
---|
754 | {
|
---|
755 | cbBodyAlloc = RT_MAX(cbBodyAlloc ? cbBodyAlloc * 2 : _1K, RT_ALIGN_Z(cchNewBody + 64, 128));
|
---|
756 | void *pvNew = RTMemRealloc(pszBody, cbBodyAlloc);
|
---|
757 | if (pvNew)
|
---|
758 | pszBody = (char *)pvNew;
|
---|
759 | else
|
---|
760 | {
|
---|
761 | RTMemFree(pszBody);
|
---|
762 | return VERR_NO_MEMORY;
|
---|
763 | }
|
---|
764 | }
|
---|
765 |
|
---|
766 | if (cchBody > 0) /* no leading blank lines */
|
---|
767 | pszBody[cchBody++] = '\n';
|
---|
768 | else if (cchAppend == 0)
|
---|
769 | Info.cBlankLinesBefore++;
|
---|
770 | memcpy(&pszBody[cchBody], &pchLine[offLineStart], cchAppend);
|
---|
771 | cchBody += cchAppend;
|
---|
772 | pszBody[cchBody] = '\0';
|
---|
773 |
|
---|
774 | /* Advance to the next line, if we haven't yet seen the end of this comment. */
|
---|
775 | if (Info.iLineEnd != UINT32_MAX)
|
---|
776 | break;
|
---|
777 | pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
|
---|
778 | if (!pchLine)
|
---|
779 | {
|
---|
780 | Info.offEnd = (uint32_t)cchLine;
|
---|
781 | Info.iLineEnd = iLine;
|
---|
782 | break;
|
---|
783 | }
|
---|
784 | iLine++;
|
---|
785 | off = 0;
|
---|
786 | }
|
---|
787 |
|
---|
788 | /* Strip trailing empty lines in the body. */
|
---|
789 | Info.cBlankLinesAfter = 0;
|
---|
790 | while (cchBody >= 1 && pszBody[cchBody - 1] == '\n')
|
---|
791 | {
|
---|
792 | Info.cBlankLinesAfter++;
|
---|
793 | pszBody[--cchBody] = '\0';
|
---|
794 | }
|
---|
795 |
|
---|
796 | /* Do the callback. */
|
---|
797 | int rc = pfnCallback(&Info, pszBody, cchBody, pvUser);
|
---|
798 | RTMemFree(pszBody);
|
---|
799 | if (RT_FAILURE(rc))
|
---|
800 | return rc;
|
---|
801 | if (rc > VINF_SUCCESS && rcRet == VINF_SUCCESS)
|
---|
802 | rcRet = rc;
|
---|
803 | }
|
---|
804 |
|
---|
805 | if (!pchLine)
|
---|
806 | break;
|
---|
807 | }
|
---|
808 | /* else: We don't have to deal with character litterals as these shouldn't
|
---|
809 | include comment-like sequences. */
|
---|
810 | } /* for each character in the line */
|
---|
811 |
|
---|
812 | iLine++;
|
---|
813 | } /* for each line in the stream */
|
---|
814 |
|
---|
815 | int rcStream = ScmStreamGetStatus(pIn);
|
---|
816 | if (RT_SUCCESS(rcStream))
|
---|
817 | return rcRet;
|
---|
818 | return rcStream;
|
---|
819 | }
|
---|
820 |
|
---|
821 |
|
---|
822 | /**
|
---|
823 | * Deals with comments in DOS batch files.
|
---|
824 | *
|
---|
825 | * @returns VBox status code / callback return code.
|
---|
826 | * @param pIn The stream to parse.
|
---|
827 | * @param pfnCallback The callback.
|
---|
828 | * @param pvUser The user parameter for the callback.
|
---|
829 | */
|
---|
830 | static int enumerateBatchComments(PSCMSTREAM pIn, PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
|
---|
831 | {
|
---|
832 | int rcRet = VINF_SUCCESS;
|
---|
833 | uint32_t iLine = 0;
|
---|
834 | SCMEOL enmEol;
|
---|
835 | size_t cchLine;
|
---|
836 | const char *pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
|
---|
837 | while (pchLine != NULL)
|
---|
838 | {
|
---|
839 | /*
|
---|
840 | * Skip leading blanks and check for 'rem'.
|
---|
841 | * At the moment we do not parse '::lable-comments'.
|
---|
842 | */
|
---|
843 | size_t off = 0;
|
---|
844 | while (off + 3 < cchLine && RT_C_IS_SPACE(pchLine[off]))
|
---|
845 | off++;
|
---|
846 | if (!IS_REM(pchLine, off, cchLine))
|
---|
847 | {
|
---|
848 | iLine++;
|
---|
849 | pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
|
---|
850 | }
|
---|
851 | else
|
---|
852 | {
|
---|
853 | int rc = handleLineComment(pIn, isBatchComment, pfnCallback, pvUser,
|
---|
854 | &pchLine, &cchLine, &enmEol, &iLine, &off);
|
---|
855 | if (RT_FAILURE(rc))
|
---|
856 | return rc;
|
---|
857 | if (rcRet == VINF_SUCCESS)
|
---|
858 | rcRet = rc;
|
---|
859 | }
|
---|
860 | }
|
---|
861 |
|
---|
862 | int rcStream = ScmStreamGetStatus(pIn);
|
---|
863 | if (RT_SUCCESS(rcStream))
|
---|
864 | return rcRet;
|
---|
865 | return rcStream;
|
---|
866 | }
|
---|
867 |
|
---|
868 |
|
---|
869 | /**
|
---|
870 | * Deals with comments in SQL files.
|
---|
871 | *
|
---|
872 | * @returns VBox status code / callback return code.
|
---|
873 | * @param pIn The stream to parse.
|
---|
874 | * @param pfnCallback The callback.
|
---|
875 | * @param pvUser The user parameter for the callback.
|
---|
876 | */
|
---|
877 | static int enumerateSqlComments(PSCMSTREAM pIn, PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
|
---|
878 | {
|
---|
879 | int rcRet = VINF_SUCCESS;
|
---|
880 | uint32_t iLine = 0;
|
---|
881 | SCMEOL enmEol;
|
---|
882 | size_t cchLine;
|
---|
883 | const char *pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
|
---|
884 | while (pchLine != NULL)
|
---|
885 | {
|
---|
886 | /*
|
---|
887 | * Skip leading blanks and check for '--'.
|
---|
888 | */
|
---|
889 | size_t off = 0;
|
---|
890 | while (off + 3 < cchLine && RT_C_IS_SPACE(pchLine[off]))
|
---|
891 | off++;
|
---|
892 | if ( cchLine < 2
|
---|
893 | || pchLine[0] != '-'
|
---|
894 | || pchLine[1] != '-')
|
---|
895 | {
|
---|
896 | iLine++;
|
---|
897 | pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
|
---|
898 | }
|
---|
899 | else
|
---|
900 | {
|
---|
901 | int rc = handleLineComment(pIn, isSqlComment, pfnCallback, pvUser,
|
---|
902 | &pchLine, &cchLine, &enmEol, &iLine, &off);
|
---|
903 | if (RT_FAILURE(rc))
|
---|
904 | return rc;
|
---|
905 | if (rcRet == VINF_SUCCESS)
|
---|
906 | rcRet = rc;
|
---|
907 | }
|
---|
908 | }
|
---|
909 |
|
---|
910 | int rcStream = ScmStreamGetStatus(pIn);
|
---|
911 | if (RT_SUCCESS(rcStream))
|
---|
912 | return rcRet;
|
---|
913 | return rcStream;
|
---|
914 | }
|
---|
915 |
|
---|
916 |
|
---|
917 | /**
|
---|
918 | * Deals with simple line comments.
|
---|
919 | *
|
---|
920 | * @returns VBox status code / callback return code.
|
---|
921 | * @param pIn The stream to parse.
|
---|
922 | * @param chStart The start of comment character.
|
---|
923 | * @param pfnIsComment Comment tester function.
|
---|
924 | * @param pfnCallback The callback.
|
---|
925 | * @param pvUser The user parameter for the callback.
|
---|
926 | */
|
---|
927 | static int enumerateSimpleLineComments(PSCMSTREAM pIn, char chStart, PFNISCOMMENT pfnIsComment,
|
---|
928 | PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
|
---|
929 | {
|
---|
930 | int rcRet = VINF_SUCCESS;
|
---|
931 | uint32_t iLine = 0;
|
---|
932 | SCMEOL enmEol;
|
---|
933 | size_t cchLine;
|
---|
934 | const char *pchLine;
|
---|
935 | while ((pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol)) != NULL)
|
---|
936 | {
|
---|
937 | size_t off = 0;
|
---|
938 | while (off < cchLine)
|
---|
939 | {
|
---|
940 | char ch = pchLine[off++];
|
---|
941 | if (ch != chStart)
|
---|
942 | { /* not interesting */ }
|
---|
943 | else
|
---|
944 | {
|
---|
945 | off -= 1;
|
---|
946 | int rc = handleLineComment(pIn, pfnIsComment, pfnCallback, pvUser,
|
---|
947 | &pchLine, &cchLine, &enmEol, &iLine, &off);
|
---|
948 | if (RT_FAILURE(rc))
|
---|
949 | return rc;
|
---|
950 | if (rcRet == VINF_SUCCESS)
|
---|
951 | rcRet = rc;
|
---|
952 |
|
---|
953 | if (!pchLine)
|
---|
954 | break;
|
---|
955 | }
|
---|
956 | } /* for each character in the line */
|
---|
957 |
|
---|
958 | iLine++;
|
---|
959 | } /* for each line in the stream */
|
---|
960 |
|
---|
961 | int rcStream = ScmStreamGetStatus(pIn);
|
---|
962 | if (RT_SUCCESS(rcStream))
|
---|
963 | return rcRet;
|
---|
964 | return rcStream;
|
---|
965 | }
|
---|
966 |
|
---|
967 |
|
---|
968 | /**
|
---|
969 | * Enumerates the comments in the given stream, calling @a pfnCallback for each.
|
---|
970 | *
|
---|
971 | * @returns IPRT status code.
|
---|
972 | * @param pIn The stream to parse.
|
---|
973 | * @param enmCommentStyle The comment style of the source stream.
|
---|
974 | * @param pfnCallback The function to call.
|
---|
975 | * @param pvUser User argument to the callback.
|
---|
976 | */
|
---|
977 | int ScmEnumerateComments(PSCMSTREAM pIn, SCMCOMMENTSTYLE enmCommentStyle, PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
|
---|
978 | {
|
---|
979 | switch (enmCommentStyle)
|
---|
980 | {
|
---|
981 | case kScmCommentStyle_C:
|
---|
982 | return enumerateCStyleComments(pIn, pfnCallback, pvUser);
|
---|
983 |
|
---|
984 | case kScmCommentStyle_Python:
|
---|
985 | return enumeratePythonComments(pIn, pfnCallback, pvUser);
|
---|
986 |
|
---|
987 | case kScmCommentStyle_Semicolon:
|
---|
988 | return enumerateSimpleLineComments(pIn, ';', isSemicolonComment, pfnCallback, pvUser);
|
---|
989 |
|
---|
990 | case kScmCommentStyle_Hash:
|
---|
991 | return enumerateSimpleLineComments(pIn, '#', isHashComment, pfnCallback, pvUser);
|
---|
992 |
|
---|
993 | case kScmCommentStyle_Rem_Upper:
|
---|
994 | case kScmCommentStyle_Rem_Lower:
|
---|
995 | case kScmCommentStyle_Rem_Camel:
|
---|
996 | return enumerateBatchComments(pIn, pfnCallback, pvUser);
|
---|
997 |
|
---|
998 | case kScmCommentStyle_Sql:
|
---|
999 | return enumerateSqlComments(pIn, pfnCallback, pvUser);
|
---|
1000 |
|
---|
1001 | case kScmCommentStyle_Tick:
|
---|
1002 | return enumerateSimpleLineComments(pIn, '\'', isTickComment, pfnCallback, pvUser);
|
---|
1003 |
|
---|
1004 | default:
|
---|
1005 | AssertFailedReturn(VERR_INVALID_PARAMETER);
|
---|
1006 | }
|
---|
1007 | }
|
---|
1008 |
|
---|