scmparser.cpp@ 100875

Last change on this file since 100875 was 98103, checked in by vboxsync, 22 months ago
Copyright year updates by scm.
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 43.4 KB

Line
1	/* $Id: scmparser.cpp 98103 2023-01-17 14:15:46Z vboxsync $ */
2	/** @file
3	* IPRT Testcase / Tool - Source Code Massager, Code Parsers.
4	*/
5
6	/*
7	* Copyright (C) 2010-2023 Oracle and/or its affiliates.
8	*
9	* This file is part of VirtualBox base platform packages, as
10	* available from https://www.virtualbox.org.
11	*
12	* This program is free software; you can redistribute it and/or
13	* modify it under the terms of the GNU General Public License
14	* as published by the Free Software Foundation, in version 3 of the
15	* License.
16	*
17	* This program is distributed in the hope that it will be useful, but
18	* WITHOUT ANY WARRANTY; without even the implied warranty of
19	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20	* General Public License for more details.
21	*
22	* You should have received a copy of the GNU General Public License
23	* along with this program; if not, see <https://www.gnu.org/licenses>.
24	*
25	* SPDX-License-Identifier: GPL-3.0-only
26	*/
27
28
29	/*********************************************************************************************************************************
30	* Header Files *
31	*********************************************************************************************************************************/
32	#include <iprt/assert.h>
33	#include <iprt/ctype.h>
34	#include <iprt/dir.h>
35	#include <iprt/env.h>
36	#include <iprt/file.h>
37	#include <iprt/errcore.h>
38	#include <iprt/getopt.h>
39	#include <iprt/initterm.h>
40	#include <iprt/mem.h>
41	#include <iprt/message.h>
42	#include <iprt/param.h>
43	#include <iprt/path.h>
44	#include <iprt/process.h>
45	#include <iprt/stream.h>
46	#include <iprt/string.h>
47
48	#include "scm.h"
49
50
51	/*********************************************************************************************************************************
52	* Structures and Typedefs *
53	*********************************************************************************************************************************/
54	typedef size_t (PFNISCOMMENT)(const char pchLine, size_t cchLine, bool fSecond);
55
56
57	/**
58	* Callback for checking if C++ line comment.
59	*/
60	static size_t isCppLineComment(const char *pchLine, size_t cchLine, bool fSecond)
61	{
62	if ( cchLine >= 2
63	&& pchLine[0] == '/'
64	&& pchLine[1] == '/')
65	{
66	if (!fSecond)
67	return 2;
68	if (cchLine >= 3 && pchLine[2] == '/')
69	return 3;
70	}
71	return 0;
72	}
73
74
75	/**
76	* Callback for checking if hash comment.
77	*/
78	static size_t isHashComment(const char *pchLine, size_t cchLine, bool fSecond)
79	{
80	if (cchLine >= 1 && *pchLine == '#')
81	{
82	if (!fSecond)
83	return 1;
84	if (cchLine >= 2 && pchLine[1] == '#')
85	return 2;
86	}
87	return 0;
88	}
89
90
91	/**
92	* Callback for checking if semicolon comment.
93	*/
94	static size_t isSemicolonComment(const char *pchLine, size_t cchLine, bool fSecond)
95	{
96	if (cchLine >= 1 && *pchLine == ';')
97	{
98	if (!fSecond)
99	return 1;
100	if (cchLine >= 2 && pchLine[1] == ';')
101	return 2;
102	}
103	return 0;
104	}
105
106
107	/** Macro for checking for a XML comment start. */
108	#define IS_XML_COMMENT_START(a_pch, a_off, a_cch) \
109	( (a_off) + 4 <= (a_cch) \
110	&& (a_pch)[(a_off) ] == '<' \
111	&& (a_pch)[(a_off) + 1] == '!' \
112	&& (a_pch)[(a_off) + 2] == '-' \
113	&& (a_pch)[(a_off) + 3] == '-' \
114	&& ((a_off) + 4 == (a_cch) \|\| RT_C_IS_SPACE((a_pch)[(a_off) + 4])) )
115
116	/** Macro for checking for a XML comment end. */
117	#define IS_XML_COMMENT_END(a_pch, a_off, a_cch) \
118	( (a_off) + 3 <= (a_cch) \
119	&& (a_pch)[(a_off) ] == '-' \
120	&& (a_pch)[(a_off) + 1] == '-' \
121	&& (a_pch)[(a_off) + 2] == '>')
122
123
124	/** Macro for checking for a batch file comment prefix. */
125	#define IS_REM(a_pch, a_off, a_cch) \
126	( (a_off) + 3 <= (a_cch) \
127	&& ((a_pch)[(a_off) ] == 'R' \|\| (a_pch)[(a_off) ] == 'r') \
128	&& ((a_pch)[(a_off) + 1] == 'E' \|\| (a_pch)[(a_off) + 1] == 'e') \
129	&& ((a_pch)[(a_off) + 2] == 'M' \|\| (a_pch)[(a_off) + 2] == 'm') \
130	&& ((a_off) + 3 == (a_cch) \|\| RT_C_IS_SPACE((a_pch)[(a_off) + 3])) )
131
132
133	/**
134	* Callback for checking if batch comment.
135	*/
136	static size_t isBatchComment(const char *pchLine, size_t cchLine, bool fSecond)
137	{
138	if (!fSecond)
139	{
140	if (IS_REM(pchLine, 0, cchLine))
141	return 3;
142	}
143	else
144	{
145	/* Check for the 2nd in "rem rem" lines. */
146	if ( cchLine >= 4
147	&& RT_C_IS_SPACE(*pchLine)
148	&& IS_REM(pchLine, 1, cchLine))
149	return 4;
150	}
151	return 0;
152	}
153
154	/**
155	* Callback for checking if SQL comment.
156	*/
157	static size_t isSqlComment(const char *pchLine, size_t cchLine, bool fSecond)
158	{
159	if ( cchLine >= 2
160	&& pchLine[0] == '-'
161	&& pchLine[1] == '-')
162	{
163	if (!fSecond)
164	return 2;
165	if ( cchLine >= 3
166	&& pchLine[2] == '-')
167	return 3;
168	}
169	return 0;
170	}
171
172	/**
173	* Callback for checking if tick comment.
174	*/
175	static size_t isTickComment(const char *pchLine, size_t cchLine, bool fSecond)
176	{
177	if (cchLine >= 1 && *pchLine == '\'')
178	{
179	if (!fSecond)
180	return 1;
181	if (cchLine >= 2 && pchLine[1] == '\'')
182	return 2;
183	}
184	return 0;
185	}
186
187
188	/**
189	* Common worker for enumeratePythonComments and enumerateSimpleLineComments.
190	*
191	* @returns IPRT status code.
192	* @param pIn The input stream.
193	* @param pfnIsComment Comment tester function.
194	* @param pfnCallback The callback.
195	* @param pvUser The user argument for the callback.
196	* @param ppchLine Pointer to the line variable.
197	* @param pcchLine Pointer to the line length variable.
198	* @param penmEol Pointer to the line ending type variable.
199	* @param piLine Pointer to the line number variable.
200	* @param poff Pointer to the line offset variable. On input this
201	* is positioned at the start of the comment.
202	*/
203	static int handleLineComment(PSCMSTREAM pIn, PFNISCOMMENT pfnIsComment,
204	PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser,
205	const char *ppchLine, size_t pcchLine, PSCMEOL penmEol,
206	uint32_t piLine, size_t poff)
207	{
208	/* Unpack input/output variables. */
209	uint32_t iLine = *piLine;
210	const char pchLine = ppchLine;
211	size_t cchLine = *pcchLine;
212	size_t off = *poff;
213	SCMEOL enmEol = *penmEol;
214
215	/*
216	* Take down the basic info about the comment.
217	*/
218	SCMCOMMENTINFO Info;
219	Info.iLineStart = iLine;
220	Info.iLineEnd = iLine;
221	Info.offStart = (uint32_t)off;
222	Info.offEnd = (uint32_t)cchLine;
223
224	size_t cchSkip = pfnIsComment(&pchLine[off], cchLine - off, false);
225	Assert(cchSkip > 0);
226	off += cchSkip;
227
228	/* Determine comment type. */
229	Info.enmType = kScmCommentType_Line;
230	char ch;
231	cchSkip = 1;
232	if ( off < cchLine
233	&& ( (ch = pchLine[off]) == '!'
234	\|\| (cchSkip = pfnIsComment(&pchLine[off], cchLine - off, true)) > 0) )
235	{
236	unsigned ch2;
237	if ( off + cchSkip == cchLine
238	\|\| RT_C_IS_SPACE(ch2 = pchLine[off + cchSkip]) )
239	{
240	Info.enmType = ch != '!' ? kScmCommentType_Line_JavaDoc : kScmCommentType_Line_Qt;
241	off += cchSkip;
242	}
243	else if ( ch2 == '<'
244	&& ( off + cchSkip + 1 == cchLine
245	\|\| RT_C_IS_SPACE(pchLine[off + cchSkip + 1]) ))
246	{
247	Info.enmType = ch == '!' ? kScmCommentType_Line_JavaDoc_After : kScmCommentType_Line_Qt_After;
248	off += cchSkip + 1;
249	}
250	}
251
252	/*
253	* Copy body of the first line. Like for C, we ignore a single space in the first comment line.
254	*/
255	if (off < cchLine && RT_C_IS_SPACE(pchLine[off]))
256	off++;
257	size_t cchBody = cchLine;
258	while (cchBody > off && RT_C_IS_SPACE(pchLine[cchBody - 1]))
259	cchBody--;
260	cchBody -= off;
261	size_t cbBodyAlloc = RT_MAX(_1K, RT_ALIGN_Z(cchBody + 64, 128));
262	char pszBody = (char )RTMemAlloc(cbBodyAlloc);
263	if (!pszBody)
264	return VERR_NO_MEMORY;
265	memcpy(pszBody, &pchLine[off], cchBody);
266	pszBody[cchBody] = '\0';
267
268	Info.cBlankLinesBefore = cchBody == 0;
269
270	/*
271	* Look for more comment lines and append them to the body.
272	*/
273	while ((pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol)) != NULL)
274	{
275	iLine++;
276
277	/* Skip leading spaces. */
278	off = 0;
279	while (off < cchLine && RT_C_IS_SPACE(pchLine[off]))
280	off++;
281
282	/* Check if it's a comment. */
283	if ( off >= cchLine
284	\|\| (cchSkip = pfnIsComment(&pchLine[off], cchLine - off, false)) == 0)
285	break;
286	off += cchSkip;
287
288	/* Split on doxygen comment start (if not already in one). */
289	if ( Info.enmType == kScmCommentType_Line
290	&& off + 1 < cchLine
291	&& ( pfnIsComment(&pchLine[off], cchLine - off, true) > 0
292	\|\| ( pchLine[off + 1] == '!'
293	&& ( off + 2 == cchLine
294	\|\| pchLine[off + 2] != '!') ) ) )
295	{
296	off -= cchSkip;
297	break;
298	}
299
300	/* Append the body w/o trailing spaces and some leading ones. */
301	if (off < cchLine && RT_C_IS_SPACE(pchLine[off]))
302	off++;
303	while (off < cchLine && off < Info.offStart + 3 && RT_C_IS_SPACE(pchLine[off]))
304	off++;
305	size_t cchAppend = cchLine;
306	while (cchAppend > off && RT_C_IS_SPACE(pchLine[cchAppend - 1]))
307	cchAppend--;
308	cchAppend -= off;
309
310	size_t cchNewBody = cchBody + 1 + cchAppend;
311	if (cchNewBody >= cbBodyAlloc)
312	{
313	cbBodyAlloc = RT_MAX(cbBodyAlloc ? cbBodyAlloc * 2 : _1K, RT_ALIGN_Z(cchNewBody + 64, 128));
314	void *pvNew = RTMemRealloc(pszBody, cbBodyAlloc);
315	if (pvNew)
316	pszBody = (char *)pvNew;
317	else
318	{
319	RTMemFree(pszBody);
320	return VERR_NO_MEMORY;
321	}
322	}
323
324	if ( cchBody > 0
325	\|\| cchAppend > 0)
326	{
327	if (cchBody > 0)
328	pszBody[cchBody++] = '\n';
329	memcpy(&pszBody[cchBody], &pchLine[off], cchAppend);
330	cchBody += cchAppend;
331	pszBody[cchBody] = '\0';
332	}
333	else
334	Info.cBlankLinesBefore++;
335
336	/* Advance. */
337	Info.offEnd = (uint32_t)cchLine;
338	Info.iLineEnd = iLine;
339	}
340
341	/*
342	* Strip trailing empty lines in the body.
343	*/
344	Info.cBlankLinesAfter = 0;
345	while (cchBody >= 1 && pszBody[cchBody - 1] == '\n')
346	{
347	Info.cBlankLinesAfter++;
348	pszBody[--cchBody] = '\0';
349	}
350
351	/*
352	* Do the callback and return.
353	*/
354	int rc = pfnCallback(&Info, pszBody, cchBody, pvUser);
355
356	RTMemFree(pszBody);
357
358	*piLine = iLine;
359	*ppchLine = pchLine;
360	*pcchLine = cchLine;
361	*poff = off;
362	*penmEol = enmEol;
363	return rc;
364	}
365
366
367
368	/**
369	* Common string literal handler.
370	*
371	* @returns new pchLine value.
372	* @param pIn The input string.
373	* @param chType The quotation type.
374	* @param pchLine The current line.
375	* @param ppchLine Pointer to the line variable.
376	* @param pcchLine Pointer to the line length variable.
377	* @param penmEol Pointer to the line ending type variable.
378	* @param piLine Pointer to the line number variable.
379	* @param poff Pointer to the line offset variable.
380	*/
381	static const char handleStringLiteral(PSCMSTREAM pIn, char chType, const char pchLine, size_t *pcchLine, PSCMEOL penmEol,
382	uint32_t piLine, size_t poff)
383	{
384	size_t off = *poff;
385	for (;;)
386	{
387	bool fEnd = false;
388	bool fEscaped = false;
389	size_t const cchLine = *pcchLine;
390	while (off < cchLine)
391	{
392	char ch = pchLine[off++];
393	if (!fEscaped)
394	{
395	if (ch != chType)
396	{
397	if (ch != '\\')
398	{ /* likely */ }
399	else
400	fEscaped = true;
401	}
402	else
403	{
404	fEnd = true;
405	break;
406	}
407	}
408	else
409	fEscaped = false;
410	}
411	if (fEnd)
412	break;
413
414	/* next line */
415	pchLine = ScmStreamGetLine(pIn, pcchLine, penmEol);
416	if (!pchLine)
417	break;
418	*piLine += 1;
419	off = 0;
420	}
421
422	*poff = off;
423	return pchLine;
424	}
425
426
427	/**
428	* Deals with comments in C and C++ code.
429	*
430	* @returns VBox status code / callback return code.
431	* @param pIn The stream to parse.
432	* @param pfnCallback The callback.
433	* @param pvUser The user parameter for the callback.
434	*/
435	static int enumerateCStyleComments(PSCMSTREAM pIn, PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
436	{
437	AssertCompile('\'' < '/');
438	AssertCompile('"' < '/');
439
440	int rcRet = VINF_SUCCESS;
441	uint32_t iLine = 0;
442	SCMEOL enmEol;
443	size_t cchLine;
444	const char *pchLine;
445	while ((pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol)) != NULL)
446	{
447	size_t off = 0;
448	while (off < cchLine)
449	{
450	unsigned ch = pchLine[off++];
451	if (ch > (unsigned)'/')
452	{ /* not interesting */ }
453	else if (ch == '/')
454	{
455	if (off < cchLine)
456	{
457	ch = pchLine[off++];
458	if (ch == '*')
459	{
460	/*
461	* Multiline comment. Find the end.
462	*
463	* Note! This is very similar to the python doc string handling further down.
464	*/
465	SCMCOMMENTINFO Info;
466	Info.iLineStart = iLine;
467	Info.offStart = (uint32_t)off - 2;
468	Info.iLineEnd = UINT32_MAX;
469	Info.offEnd = UINT32_MAX;
470	Info.cBlankLinesBefore = 0;
471
472	/* Determine comment type (same as for line-comments). */
473	Info.enmType = kScmCommentType_MultiLine;
474	if ( off < cchLine
475	&& ( (ch = pchLine[off]) == '*'
476	\|\| ch == '!') )
477	{
478	unsigned ch2;
479	if ( off + 1 == cchLine
480	\|\| RT_C_IS_SPACE(ch2 = pchLine[off + 1]) )
481	{
482	Info.enmType = ch == '*' ? kScmCommentType_MultiLine_JavaDoc : kScmCommentType_MultiLine_Qt;
483	off += 1;
484	}
485	else if ( ch2 == '<'
486	&& ( off + 2 == cchLine
487	\|\| RT_C_IS_SPACE(pchLine[off + 2]) ))
488	{
489	Info.enmType = ch == '*' ? kScmCommentType_MultiLine_JavaDoc_After
490	: kScmCommentType_MultiLine_Qt_After;
491	off += 2;
492	}
493	}
494
495	/*
496	* Copy the body and find the end of the multiline comment.
497	*/
498	size_t cbBodyAlloc = 0;
499	size_t cchBody = 0;
500	char *pszBody = NULL;
501	for (;;)
502	{
503	/* Parse the line up to the end-of-comment or end-of-line. */
504	size_t offLineStart = off;
505	size_t offLastNonBlank = off;
506	size_t offFirstNonBlank = ~(size_t)0;
507	while (off < cchLine)
508	{
509	ch = pchLine[off++];
510	if (ch != '*' \|\| off >= cchLine \|\| pchLine[off] != '/')
511	{
512	if (RT_C_IS_BLANK(ch))
513	{/* kind of likely */}
514	else
515	{
516	offLastNonBlank = off - 1;
517	if (offFirstNonBlank != ~(size_t)0)
518	{/* likely */}
519	else if ( ch != '' / ignore continuation-asterisks */
520	\|\| off > Info.offStart + 1 + 1
521	\|\| off > cchLine
522	\|\| ( off < cchLine
523	&& !RT_C_IS_SPACE(pchLine[off]))
524	\|\| pszBody == NULL)
525	offFirstNonBlank = off - 1;
526	}
527	}
528	else
529	{
530	Info.offEnd = (uint32_t)++off;
531	Info.iLineEnd = iLine;
532	break;
533	}
534	}
535
536	/* Append line content to the comment body string. */
537	size_t cchAppend;
538	if (offFirstNonBlank == ~(size_t)0)
539	cchAppend = 0; /* empty line */
540	else
541	{
542	if (pszBody)
543	offLineStart = RT_MIN(Info.offStart + 3, offFirstNonBlank);
544	else if (offFirstNonBlank > Info.offStart + 2) /* Skip one leading blank at the start of the comment. */
545	offLineStart++;
546	cchAppend = offLastNonBlank + 1 - offLineStart;
547	Assert(cchAppend <= cchLine);
548	}
549
550	size_t cchNewBody = cchBody + (cchBody > 0) + cchAppend;
551	if (cchNewBody >= cbBodyAlloc)
552	{
553	cbBodyAlloc = RT_MAX(cbBodyAlloc ? cbBodyAlloc * 2 : _1K, RT_ALIGN_Z(cchNewBody + 64, 128));
554	void *pvNew = RTMemRealloc(pszBody, cbBodyAlloc);
555	if (pvNew)
556	pszBody = (char *)pvNew;
557	else
558	{
559	RTMemFree(pszBody);
560	return VERR_NO_MEMORY;
561	}
562	}
563
564	if (cchBody > 0) /* no leading blank lines */
565	pszBody[cchBody++] = '\n';
566	else if (cchAppend == 0)
567	Info.cBlankLinesBefore++;
568	memcpy(&pszBody[cchBody], &pchLine[offLineStart], cchAppend);
569	cchBody += cchAppend;
570	pszBody[cchBody] = '\0';
571
572	/* Advance to the next line, if we haven't yet seen the end of this comment. */
573	if (Info.iLineEnd != UINT32_MAX)
574	break;
575	pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
576	if (!pchLine)
577	{
578	Info.offEnd = (uint32_t)cchLine;
579	Info.iLineEnd = iLine;
580	break;
581	}
582	iLine++;
583	off = 0;
584	}
585
586	/* Strip trailing empty lines in the body. */
587	Info.cBlankLinesAfter = 0;
588	while (cchBody >= 1 && pszBody[cchBody - 1] == '\n')
589	{
590	Info.cBlankLinesAfter++;
591	pszBody[--cchBody] = '\0';
592	}
593
594	/* Do the callback. */
595	int rc = pfnCallback(&Info, pszBody, cchBody, pvUser);
596	RTMemFree(pszBody);
597	if (RT_FAILURE(rc))
598	return rc;
599	if (rc > VINF_SUCCESS && rcRet == VINF_SUCCESS)
600	rcRet = rc;
601	}
602	else if (ch == '/')
603	{
604	/*
605	* Line comment. Join the other line comment guys.
606	*/
607	off -= 2;
608	int rc = handleLineComment(pIn, isCppLineComment, pfnCallback, pvUser,
609	&pchLine, &cchLine, &enmEol, &iLine, &off);
610	if (RT_FAILURE(rc))
611	return rc;
612	if (rcRet == VINF_SUCCESS)
613	rcRet = rc;
614	}
615
616	if (!pchLine)
617	break;
618	}
619	}
620	else if (ch == '"')
621	{
622	/*
623	* String literal may include sequences that looks like comments. So,
624	* they needs special handling to avoid confusion.
625	*/
626	pchLine = handleStringLiteral(pIn, '"', pchLine, &cchLine, &enmEol, &iLine, &off);
627	}
628	/* else: We don't have to deal with character literal as these shouldn't
629	include comment-like sequences. */
630	} /* for each character in the line */
631
632	iLine++;
633	} /* for each line in the stream */
634
635	int rcStream = ScmStreamGetStatus(pIn);
636	if (RT_SUCCESS(rcStream))
637	return rcRet;
638	return rcStream;
639	}
640
641
642	/**
643	* Deals with comments in Python code.
644	*
645	* @returns VBox status code / callback return code.
646	* @param pIn The stream to parse.
647	* @param pfnCallback The callback.
648	* @param pvUser The user parameter for the callback.
649	*/
650	static int enumeratePythonComments(PSCMSTREAM pIn, PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
651	{
652	AssertCompile('#' < '\'');
653	AssertCompile('"' < '\'');
654
655	int rcRet = VINF_SUCCESS;
656	uint32_t iLine = 0;
657	SCMEOL enmEol;
658	size_t cchLine;
659	const char *pchLine;
660	SCMCOMMENTINFO Info;
661	while ((pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol)) != NULL)
662	{
663	size_t off = 0;
664	while (off < cchLine)
665	{
666	char ch = pchLine[off++];
667	if ((unsigned char)ch > (unsigned char)'\'')
668	{ /* not interesting */ }
669	else if (ch == '#')
670	{
671	/*
672	* Line comment. Join paths with the others.
673	*/
674	off -= 1;
675	int rc = handleLineComment(pIn, isHashComment, pfnCallback, pvUser,
676	&pchLine, &cchLine, &enmEol, &iLine, &off);
677	if (RT_FAILURE(rc))
678	return rc;
679	if (rcRet == VINF_SUCCESS)
680	rcRet = rc;
681
682	if (!pchLine)
683	break;
684	}
685	else if (ch == '"' \|\| ch == '\'')
686	{
687	/*
688	* String literal may be doc strings and they may legally include hashes.
689	*/
690	const char chType = ch;
691	if ( off + 1 >= cchLine
692	\|\| pchLine[off] != chType
693	\|\| pchLine[off + 1] != chType)
694	pchLine = handleStringLiteral(pIn, chType, pchLine, &cchLine, &enmEol, &iLine, &off);
695	else
696	{
697	/*
698	* Doc string (/ long string).
699	*
700	* Note! This is very similar to the multiline C comment handling above.
701	*/
702	Info.iLineStart = iLine;
703	Info.offStart = (uint32_t)off - 1;
704	Info.iLineEnd = UINT32_MAX;
705	Info.offEnd = UINT32_MAX;
706	Info.cBlankLinesBefore = 0;
707	Info.enmType = kScmCommentType_DocString;
708
709	off += 2;
710
711	/* Copy the body and find the end of the doc string comment. */
712	size_t cbBodyAlloc = 0;
713	size_t cchBody = 0;
714	char *pszBody = NULL;
715	for (;;)
716	{
717	/* Parse the line up to the end-of-comment or end-of-line. */
718	size_t offLineStart = off;
719	size_t offLastNonBlank = off;
720	size_t offFirstNonBlank = ~(size_t)0;
721	bool fEscaped = false;
722	while (off < cchLine)
723	{
724	ch = pchLine[off++];
725	if (!fEscaped)
726	{
727	if ( off + 1 >= cchLine
728	\|\| ch != chType
729	\|\| pchLine[off] != chType
730	\|\| pchLine[off + 1] != chType)
731	{
732	if (RT_C_IS_BLANK(ch))
733	{/* kind of likely */}
734	else
735	{
736	offLastNonBlank = off - 1;
737	if (offFirstNonBlank != ~(size_t)0)
738	{/* likely */}
739	else if ( ch != '' / ignore continuation-asterisks */
740	\|\| off > Info.offStart + 1 + 1
741	\|\| off > cchLine
742	\|\| ( off < cchLine
743	&& !RT_C_IS_SPACE(pchLine[off]))
744	\|\| pszBody == NULL)
745	offFirstNonBlank = off - 1;
746
747	if (ch != '\\')
748	{/* likely */ }
749	else
750	fEscaped = true;
751	}
752	}
753	else
754	{
755	off += 2;
756	Info.offEnd = (uint32_t)off;
757	Info.iLineEnd = iLine;
758	break;
759	}
760	}
761	else
762	fEscaped = false;
763	}
764
765	/* Append line content to the comment body string. */
766	size_t cchAppend;
767	if (offFirstNonBlank == ~(size_t)0)
768	cchAppend = 0; /* empty line */
769	else
770	{
771	if (pszBody)
772	offLineStart = RT_MIN(Info.offStart + 3, offFirstNonBlank);
773	else if (offFirstNonBlank > Info.offStart + 2) /* Skip one leading blank at the start of the comment. */
774	offLineStart++;
775	cchAppend = offLastNonBlank + 1 - offLineStart;
776	Assert(cchAppend <= cchLine);
777	}
778
779	size_t cchNewBody = cchBody + (cchBody > 0) + cchAppend;
780	if (cchNewBody >= cbBodyAlloc)
781	{
782	cbBodyAlloc = RT_MAX(cbBodyAlloc ? cbBodyAlloc * 2 : _1K, RT_ALIGN_Z(cchNewBody + 64, 128));
783	void *pvNew = RTMemRealloc(pszBody, cbBodyAlloc);
784	if (pvNew)
785	pszBody = (char *)pvNew;
786	else
787	{
788	RTMemFree(pszBody);
789	return VERR_NO_MEMORY;
790	}
791	}
792
793	if (cchBody > 0) /* no leading blank lines */
794	pszBody[cchBody++] = '\n';
795	else if (cchAppend == 0)
796	Info.cBlankLinesBefore++;
797	memcpy(&pszBody[cchBody], &pchLine[offLineStart], cchAppend);
798	cchBody += cchAppend;
799	pszBody[cchBody] = '\0';
800
801	/* Advance to the next line, if we haven't yet seen the end of this comment. */
802	if (Info.iLineEnd != UINT32_MAX)
803	break;
804	pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
805	if (!pchLine)
806	{
807	Info.offEnd = (uint32_t)cchLine;
808	Info.iLineEnd = iLine;
809	break;
810	}
811	iLine++;
812	off = 0;
813	}
814
815	/* Strip trailing empty lines in the body. */
816	Info.cBlankLinesAfter = 0;
817	while (cchBody >= 1 && pszBody[cchBody - 1] == '\n')
818	{
819	Info.cBlankLinesAfter++;
820	pszBody[--cchBody] = '\0';
821	}
822
823	/* Do the callback. */
824	int rc = pfnCallback(&Info, pszBody, cchBody, pvUser);
825	RTMemFree(pszBody);
826	if (RT_FAILURE(rc))
827	return rc;
828	if (rc > VINF_SUCCESS && rcRet == VINF_SUCCESS)
829	rcRet = rc;
830	}
831
832	if (!pchLine)
833	break;
834	}
835	/* else: We don't have to deal with character literal as these shouldn't
836	include comment-like sequences. */
837	} /* for each character in the line */
838
839	iLine++;
840	} /* for each line in the stream */
841
842	int rcStream = ScmStreamGetStatus(pIn);
843	if (RT_SUCCESS(rcStream))
844	return rcRet;
845	return rcStream;
846	}
847
848
849	/**
850	* Deals with XML comments.
851	*
852	* @returns VBox status code / callback return code.
853	* @param pIn The stream to parse.
854	* @param pfnCallback The callback.
855	* @param pvUser The user parameter for the callback.
856	*/
857	static int enumerateXmlComments(PSCMSTREAM pIn, PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
858	{
859	int rcRet = VINF_SUCCESS;
860	uint32_t iLine = 0;
861	SCMEOL enmEol;
862	size_t cchLine;
863	const char *pchLine;
864	while ((pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol)) != NULL)
865	{
866	size_t off = 0;
867	while (off < cchLine)
868	{
869	/*
870	* Skip leading blanks and check for start of XML comment.
871	*/
872	while (off + 3 < cchLine && RT_C_IS_SPACE(pchLine[off]))
873	off++;
874	if (IS_XML_COMMENT_START(pchLine, off, cchLine))
875	{
876	/*
877	* XML comment. Find the end.
878	*
879	* Note! This is very similar to the python doc string handling above.
880	*/
881	SCMCOMMENTINFO Info;
882	Info.iLineStart = iLine;
883	Info.offStart = (uint32_t)off;
884	Info.iLineEnd = UINT32_MAX;
885	Info.offEnd = UINT32_MAX;
886	Info.cBlankLinesBefore = 0;
887	Info.enmType = kScmCommentType_Xml;
888
889	off += 4;
890
891	/*
892	* Copy the body and find the end of the XML comment.
893	*/
894	size_t cbBodyAlloc = 0;
895	size_t cchBody = 0;
896	char *pszBody = NULL;
897	for (;;)
898	{
899	/* Parse the line up to the end-of-comment or end-of-line. */
900	size_t offLineStart = off;
901	size_t offLastNonBlank = off;
902	size_t offFirstNonBlank = ~(size_t)0;
903	while (off < cchLine)
904	{
905	if (!IS_XML_COMMENT_END(pchLine, off, cchLine))
906	{
907	char ch = pchLine[off++];
908	if (RT_C_IS_BLANK(ch))
909	{/* kind of likely */}
910	else
911	{
912	offLastNonBlank = off - 1;
913	if (offFirstNonBlank != ~(size_t)0)
914	{/* likely */}
915	else if ( (ch != '' && ch != '#') / ignore continuation-asterisks */
916	\|\| off > Info.offStart + 1 + 1
917	\|\| off > cchLine
918	\|\| ( off < cchLine
919	&& !RT_C_IS_SPACE(pchLine[off]))
920	\|\| pszBody == NULL)
921	offFirstNonBlank = off - 1;
922	}
923	}
924	else
925	{
926	off += 3;
927	Info.offEnd = (uint32_t)off;
928	Info.iLineEnd = iLine;
929	break;
930	}
931	}
932
933	/* Append line content to the comment body string. */
934	size_t cchAppend;
935	if (offFirstNonBlank == ~(size_t)0)
936	cchAppend = 0; /* empty line */
937	else
938	{
939	offLineStart = offFirstNonBlank;
940	cchAppend = offLastNonBlank + 1 - offLineStart;
941	Assert(cchAppend <= cchLine);
942	}
943
944	size_t cchNewBody = cchBody + (cchBody > 0) + cchAppend;
945	if (cchNewBody >= cbBodyAlloc)
946	{
947	cbBodyAlloc = RT_MAX(cbBodyAlloc ? cbBodyAlloc * 2 : _1K, RT_ALIGN_Z(cchNewBody + 64, 128));
948	void *pvNew = RTMemRealloc(pszBody, cbBodyAlloc);
949	if (pvNew)
950	pszBody = (char *)pvNew;
951	else
952	{
953	RTMemFree(pszBody);
954	return VERR_NO_MEMORY;
955	}
956	}
957
958	if (cchBody > 0) /* no leading blank lines */
959	pszBody[cchBody++] = '\n';
960	else if (cchAppend == 0)
961	Info.cBlankLinesBefore++;
962	memcpy(&pszBody[cchBody], &pchLine[offLineStart], cchAppend);
963	cchBody += cchAppend;
964	pszBody[cchBody] = '\0';
965
966	/* Advance to the next line, if we haven't yet seen the end of this comment. */
967	if (Info.iLineEnd != UINT32_MAX)
968	break;
969	pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
970	if (!pchLine)
971	{
972	Info.offEnd = (uint32_t)cchLine;
973	Info.iLineEnd = iLine;
974	break;
975	}
976	iLine++;
977	off = 0;
978	}
979
980	/* Strip trailing empty lines in the body. */
981	Info.cBlankLinesAfter = 0;
982	while (cchBody >= 1 && pszBody[cchBody - 1] == '\n')
983	{
984	Info.cBlankLinesAfter++;
985	pszBody[--cchBody] = '\0';
986	}
987
988	/* Do the callback. */
989	int rc = pfnCallback(&Info, pszBody, cchBody, pvUser);
990	RTMemFree(pszBody);
991	if (RT_FAILURE(rc))
992	return rc;
993	if (rc > VINF_SUCCESS && rcRet == VINF_SUCCESS)
994	rcRet = rc;
995	}
996	else
997	off++;
998	} /* for each character in the line */
999
1000	iLine++;
1001	} /* for each line in the stream */
1002
1003	int rcStream = ScmStreamGetStatus(pIn);
1004	if (RT_SUCCESS(rcStream))
1005	return rcRet;
1006	return rcStream;
1007	}
1008
1009
1010	/**
1011	* Deals with comments in DOS batch files.
1012	*
1013	* @returns VBox status code / callback return code.
1014	* @param pIn The stream to parse.
1015	* @param pfnCallback The callback.
1016	* @param pvUser The user parameter for the callback.
1017	*/
1018	static int enumerateBatchComments(PSCMSTREAM pIn, PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
1019	{
1020	int rcRet = VINF_SUCCESS;
1021	uint32_t iLine = 0;
1022	SCMEOL enmEol;
1023	size_t cchLine;
1024	const char *pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
1025	while (pchLine != NULL)
1026	{
1027	/*
1028	* Skip leading blanks and check for 'rem'.
1029	* At the moment we do not parse '::label-comments'.
1030	*/
1031	size_t off = 0;
1032	while (off + 3 < cchLine && RT_C_IS_SPACE(pchLine[off]))
1033	off++;
1034	if (!IS_REM(pchLine, off, cchLine))
1035	{
1036	iLine++;
1037	pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
1038	}
1039	else
1040	{
1041	int rc = handleLineComment(pIn, isBatchComment, pfnCallback, pvUser,
1042	&pchLine, &cchLine, &enmEol, &iLine, &off);
1043	if (RT_FAILURE(rc))
1044	return rc;
1045	if (rcRet == VINF_SUCCESS)
1046	rcRet = rc;
1047	}
1048	}
1049
1050	int rcStream = ScmStreamGetStatus(pIn);
1051	if (RT_SUCCESS(rcStream))
1052	return rcRet;
1053	return rcStream;
1054	}
1055
1056
1057	/**
1058	* Deals with comments in SQL files.
1059	*
1060	* @returns VBox status code / callback return code.
1061	* @param pIn The stream to parse.
1062	* @param pfnCallback The callback.
1063	* @param pvUser The user parameter for the callback.
1064	*/
1065	static int enumerateSqlComments(PSCMSTREAM pIn, PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
1066	{
1067	int rcRet = VINF_SUCCESS;
1068	uint32_t iLine = 0;
1069	SCMEOL enmEol;
1070	size_t cchLine;
1071	const char *pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
1072	while (pchLine != NULL)
1073	{
1074	/*
1075	* Skip leading blanks and check for '--'.
1076	*/
1077	size_t off = 0;
1078	while (off + 3 < cchLine && RT_C_IS_SPACE(pchLine[off]))
1079	off++;
1080	if ( cchLine < 2
1081	\|\| pchLine[0] != '-'
1082	\|\| pchLine[1] != '-')
1083	{
1084	iLine++;
1085	pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol);
1086	}
1087	else
1088	{
1089	int rc = handleLineComment(pIn, isSqlComment, pfnCallback, pvUser,
1090	&pchLine, &cchLine, &enmEol, &iLine, &off);
1091	if (RT_FAILURE(rc))
1092	return rc;
1093	if (rcRet == VINF_SUCCESS)
1094	rcRet = rc;
1095	}
1096	}
1097
1098	int rcStream = ScmStreamGetStatus(pIn);
1099	if (RT_SUCCESS(rcStream))
1100	return rcRet;
1101	return rcStream;
1102	}
1103
1104
1105	/**
1106	* Deals with simple line comments.
1107	*
1108	* @returns VBox status code / callback return code.
1109	* @param pIn The stream to parse.
1110	* @param chStart The start of comment character.
1111	* @param pfnIsComment Comment tester function.
1112	* @param pfnCallback The callback.
1113	* @param pvUser The user parameter for the callback.
1114	*/
1115	static int enumerateSimpleLineComments(PSCMSTREAM pIn, char chStart, PFNISCOMMENT pfnIsComment,
1116	PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
1117	{
1118	int rcRet = VINF_SUCCESS;
1119	uint32_t iLine = 0;
1120	SCMEOL enmEol;
1121	size_t cchLine;
1122	const char *pchLine;
1123	while ((pchLine = ScmStreamGetLine(pIn, &cchLine, &enmEol)) != NULL)
1124	{
1125	size_t off = 0;
1126	while (off < cchLine)
1127	{
1128	char ch = pchLine[off++];
1129	if (ch != chStart)
1130	{ /* not interesting */ }
1131	else
1132	{
1133	off -= 1;
1134	int rc = handleLineComment(pIn, pfnIsComment, pfnCallback, pvUser,
1135	&pchLine, &cchLine, &enmEol, &iLine, &off);
1136	if (RT_FAILURE(rc))
1137	return rc;
1138	if (rcRet == VINF_SUCCESS)
1139	rcRet = rc;
1140
1141	if (!pchLine)
1142	break;
1143	}
1144	} /* for each character in the line */
1145
1146	iLine++;
1147	} /* for each line in the stream */
1148
1149	int rcStream = ScmStreamGetStatus(pIn);
1150	if (RT_SUCCESS(rcStream))
1151	return rcRet;
1152	return rcStream;
1153	}
1154
1155
1156	/**
1157	* Enumerates the comments in the given stream, calling @a pfnCallback for each.
1158	*
1159	* @returns IPRT status code.
1160	* @param pIn The stream to parse.
1161	* @param enmCommentStyle The comment style of the source stream.
1162	* @param pfnCallback The function to call.
1163	* @param pvUser User argument to the callback.
1164	*/
1165	int ScmEnumerateComments(PSCMSTREAM pIn, SCMCOMMENTSTYLE enmCommentStyle, PFNSCMCOMMENTENUMERATOR pfnCallback, void *pvUser)
1166	{
1167	switch (enmCommentStyle)
1168	{
1169	case kScmCommentStyle_C:
1170	return enumerateCStyleComments(pIn, pfnCallback, pvUser);
1171
1172	case kScmCommentStyle_Python:
1173	return enumeratePythonComments(pIn, pfnCallback, pvUser);
1174
1175	case kScmCommentStyle_Semicolon:
1176	return enumerateSimpleLineComments(pIn, ';', isSemicolonComment, pfnCallback, pvUser);
1177
1178	case kScmCommentStyle_Hash:
1179	return enumerateSimpleLineComments(pIn, '#', isHashComment, pfnCallback, pvUser);
1180
1181	case kScmCommentStyle_Rem_Upper:
1182	case kScmCommentStyle_Rem_Lower:
1183	case kScmCommentStyle_Rem_Camel:
1184	return enumerateBatchComments(pIn, pfnCallback, pvUser);
1185
1186	case kScmCommentStyle_Sql:
1187	return enumerateSqlComments(pIn, pfnCallback, pvUser);
1188
1189	case kScmCommentStyle_Tick:
1190	return enumerateSimpleLineComments(pIn, '\'', isTickComment, pfnCallback, pvUser);
1191
1192	case kScmCommentStyle_Xml:
1193	return enumerateXmlComments(pIn, pfnCallback, pvUser);
1194
1195	default:
1196	AssertFailedReturn(VERR_INVALID_PARAMETER);
1197	}
1198	}
1199

Note: See TracBrowser for help on using the repository browser.

source: vbox/trunk/src/bldprogs/scmparser.cpp@ 100875

Download in other formats: