bldprog-strtab-template.cpp.h@ 104212

Last change on this file since 104212 was 104010, checked in by vboxsync, 8 months ago
*: Fixed various calloc argument order warnings from gcc 14.
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 42.9 KB

Line
1	/* $Id: bldprog-strtab-template.cpp.h 104010 2024-03-23 01:24:05Z vboxsync $ */
2	/** @file
3	* IPRT - Build Program - String Table Generator.
4	*/
5
6	/*
7	* Copyright (C) 2006-2023 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	* The contents of this file may alternatively be used under the terms
26	* of the Common Development and Distribution License Version 1.0
27	* (CDDL), a copy of it is provided in the "COPYING.CDDL" file included
28	* in the VirtualBox distribution, in which case the provisions of the
29	* CDDL are applicable instead of those of the GPL.
30	*
31	* You may elect to license modified versions of this file under the
32	* terms and conditions of either the GPL or the CDDL or both.
33	*
34	* SPDX-License-Identifier: GPL-3.0-only OR CDDL-1.0
35	*/
36
37
38	/*
39	* (Avoid include sanity checks as this is ring-3 only, C++ code.)
40	*/
41	#if defined(__cplusplus) && defined(IN_RING3)
42
43
44	/*********************************************************************************************************************************
45	* Defined Constants And Macros *
46	*********************************************************************************************************************************/
47	/** @def BLDPROG_STRTAB_MAX_STRLEN
48	* The max length of strings in the table. */
49	#if !defined(BLDPROG_STRTAB_MAX_STRLEN) \|\| defined(DOXYGEN_RUNNING)
50	# define BLDPROG_STRTAB_MAX_STRLEN 256
51	#endif
52
53	/** @def BLDPROG_STRTAB_WITH_COMPRESSION
54	* Enables very simple string compression.
55	*/
56	#if defined(DOXYGEN_RUNNING)
57	# define BLDPROG_STRTAB_WITH_COMPRESSION
58	#endif
59
60	/** @def BLDPROG_STRTAB_WITH_CAMEL_WORDS
61	* Modifies the string compression to look for camel case words.
62	*/
63	#if defined(DOXYGEN_RUNNING)
64	# define BLDPROG_STRTAB_WITH_CAMEL_WORDS
65	#endif
66
67	/** @def BLDPROG_STRTAB_PURE_ASCII
68	* String compression assumes pure 7-bit ASCII and will fail on UTF-8 when this
69	* is defined. Otherwise, the compression code will require IPRT to link.
70	*/
71	#if defined(DOXYGEN_RUNNING)
72	# define BLDPROG_STRTAB_PURE_ASCII
73	#endif
74
75
76
77	/*********************************************************************************************************************************
78	* Header Files *
79	*********************************************************************************************************************************/
80	#include <iprt/stdarg.h>
81	#include <iprt/ctype.h>
82	#include <stdlib.h>
83	#include <stdio.h>
84	#include <string.h>
85
86	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
87	# include <map>
88	# include <iprt/sanitized/string>
89
90	# define BLDPROG_STRTAB_WITH_WORD_SEP_ALTERNATIVE
91	typedef struct BLDPROGWORDFREQSTATS
92	{
93	uint32_t cWithoutSep; /*< Number of occurances without a separator. /
94	# ifdef BLDPROG_STRTAB_WITH_WORD_SEP_ALTERNATIVE
95	uint32_t cWithSep; /*< Number of occurance with a separator. /
96	char chSep; /*< The separator. First come basis. /
97	# endif
98	} BLDPROGWORDFREQSTATS;
99
100	typedef std::map<std::string, BLDPROGWORDFREQSTATS> BLDPROGWORDFREQMAP;
101
102	#endif
103
104	#include "../../src/VBox/Runtime/include/internal/strhash.h" /** @todo make this one public */
105
106
107	/*********************************************************************************************************************************
108	* Structures and Typedefs *
109	*********************************************************************************************************************************/
110	/**
111	* Build table string.
112	*/
113	typedef struct BLDPROGSTRING
114	{
115	/** The string.
116	* @note This may be modified or replaced (allocated from heap) when
117	* compressing the string table. */
118	char *pszString;
119	/** The string hash value. */
120	uint32_t uHash;
121	/** The strint table offset. */
122	uint32_t offStrTab;
123	/** The string length. */
124	size_t cchString;
125	/** Pointer to the next string reference (same string table entry). */
126	struct BLDPROGSTRING *pNextRef;
127	/** Pointer to the next string with the same hash value (collision). */
128	struct BLDPROGSTRING *pNextCollision;
129
130	} BLDPROGSTRING;
131	/** Pointer to a string table string. */
132	typedef BLDPROGSTRING *PBLDPROGSTRING;
133
134
135	/** String table data. */
136	typedef struct BLDPROGSTRTAB
137	{
138	/** The size of g_papStrHash. */
139	size_t cStrHash;
140	/** String hash table. */
141	PBLDPROGSTRING *papStrHash;
142	/** Duplicate strings found by AddString. */
143	size_t cDuplicateStrings;
144	/** Total length of the unique strings (no terminators). */
145	size_t cchUniqueStrings;
146	/** Number of unique strings after AddString. */
147	size_t cUniqueStrings;
148	/** Number of collisions. */
149	size_t cCollisions;
150
151	/** Number of entries in apSortedStrings. */
152	size_t cSortedStrings;
153	/** The sorted string table. */
154	PBLDPROGSTRING *papSortedStrings;
155
156	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
157	/** The 256 words we've picked to be indexed by reference. */
158	BLDPROGSTRING aCompDict[256];
159	/** The frequency of the 256 dictionary entries. */
160	size_t auCompDictFreq[256];
161	/** Incoming strings pending compression. */
162	PBLDPROGSTRING *papPendingStrings;
163	/** Current number of entries in papStrPending. */
164	size_t cPendingStrings;
165	/** The allocated size of papPendingStrings. */
166	size_t cMaxPendingStrings;
167	/** Work frequency map.
168	* @todo rewrite in plain C. */
169	BLDPROGWORDFREQMAP Frequencies;
170	/** Map of characters used by input strings. */
171	uint64_t bmUsedChars[256/64];
172	#endif
173
174	/** The string table. */
175	char *pachStrTab;
176	/** The actual string table size. */
177	size_t cchStrTab;
178	} BLDPROGSTRTAB;
179	typedef BLDPROGSTRTAB *PBLDPROGSTRTAB;
180
181	#if RT_CLANG_PREREQ(4, 0) \|\| RT_GNUC_PREREQ(4, 6)
182	# pragma GCC diagnostic push
183	# pragma GCC diagnostic ignored "-Wunused-function"
184	#endif
185
186
187	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
188
189	/**
190	* Same as ASMBitTest.
191	*
192	* We cannot safely use ASMBitTest here because it must be inline, as this code
193	* is used to build RuntimeBldProg. */
194	DECLINLINE(bool) BldProgBitIsSet(uint64_t const *pbmBitmap, size_t iBit)
195	{
196	return RT_BOOL(pbmBitmap[iBit / 64] & RT_BIT_64(iBit % 64));
197	}
198
199
200	/**
201	* Same as ASMBitSet.
202	*
203	* We cannot safely use ASMBitSet here because it must be inline, as this code
204	* is used to build RuntimeBldProg.
205	*/
206	DECLINLINE(void) BldProgBitSet(uint64_t *pbmBitmap, size_t iBit)
207	{
208	pbmBitmap[iBit / 64] \|= RT_BIT_64(iBit % 64);
209	}
210
211	#endif
212
213
214	/**
215	* Initializes the strint table compiler.
216	*
217	* @returns success indicator (out of memory if false).
218	* @param pThis The strint table compiler instance.
219	* @param cMaxStrings The max number of strings we'll be adding.
220	*/
221	static bool BldProgStrTab_Init(PBLDPROGSTRTAB pThis, size_t cMaxStrings)
222	{
223	pThis->cStrHash = 0;
224	pThis->papStrHash = NULL;
225	pThis->cDuplicateStrings = 0;
226	pThis->cchUniqueStrings = 0;
227	pThis->cUniqueStrings = 0;
228	pThis->cCollisions = 0;
229	pThis->cSortedStrings = 0;
230	pThis->papSortedStrings = NULL;
231	pThis->pachStrTab = NULL;
232	pThis->cchStrTab = 0;
233	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
234	memset(pThis->aCompDict, 0, sizeof(pThis->aCompDict));
235	pThis->papPendingStrings = NULL;
236	pThis->cPendingStrings = 0;
237	pThis->cMaxPendingStrings = cMaxStrings;
238	memset(pThis->bmUsedChars, 0, sizeof(pThis->bmUsedChars));
239	BldProgBitSet(pThis->bmUsedChars, 0); /* Some parts of the code still thinks zero is a terminator, so don't use it for now. */
240	# ifndef BLDPROG_STRTAB_PURE_ASCII
241	BldProgBitSet(pThis->bmUsedChars, 0xff); /* Reserve escape byte for codepoints above 127. */
242	# endif
243	#endif
244
245	/*
246	* Allocate a hash table double the size of all strings (to avoid too
247	* many collisions). Add all strings to it, finding duplicates in the
248	* process.
249	*/
250	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
251	cMaxStrings += RT_ELEMENTS(pThis->aCompDict);
252	#endif
253	cMaxStrings *= 2;
254	pThis->papStrHash = (PBLDPROGSTRING *)calloc(cMaxStrings, sizeof(pThis->papStrHash[0]));
255	if (pThis->papStrHash)
256	{
257	pThis->cStrHash = cMaxStrings;
258	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
259	pThis->papPendingStrings = (PBLDPROGSTRING *)calloc(pThis->cMaxPendingStrings, sizeof(pThis->papPendingStrings[0]));
260	if (pThis->papPendingStrings)
261	#endif
262	return true;
263
264	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
265	free(pThis->papStrHash);
266	pThis->papStrHash = NULL;
267	#endif
268	}
269	return false;
270	}
271
272
273	#if 0 /* unused */
274	static void BldProgStrTab_Delete(PBLDPROGSTRTAB pThis)
275	{
276	free(pThis->papStrHash);
277	free(pThis->papSortedStrings);
278	free(pThis->pachStrTab);
279	# ifdef BLDPROG_STRTAB_WITH_COMPRESSION
280	free(pThis->papPendingStrings);
281	# endif
282	memset(pThis, 0, sizeof(*pThis));
283	}
284	#endif
285
286	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
287
288	DECLINLINE(size_t) bldProgStrTab_compressorFindNextWord(const char pszSrc, char ch, const char *ppszSrc)
289	{
290	/*
291	* Skip leading word separators.
292	*/
293	# ifdef BLDPROG_STRTAB_WITH_CAMEL_WORDS
294	while ( ch == ' '
295	\|\| ch == '-'
296	\|\| ch == '+'
297	\|\| ch == '_')
298	# else
299	while (ch == ' ')
300	# endif
301	ch = *++pszSrc;
302	if (ch)
303	{
304	/*
305	* Find end of word.
306	*/
307	size_t cchWord = 1;
308	# ifdef BLDPROG_STRTAB_WITH_CAMEL_WORDS
309	char chPrev = ch;
310	while ( (ch = pszSrc[cchWord]) != ' '
311	&& ch != '\0'
312	&& ch != '-'
313	&& ch != '+'
314	&& ch != '_'
315	&& ( ch == chPrev
316	\|\| !RT_C_IS_UPPER(ch)
317	\|\| RT_C_IS_UPPER(chPrev)) )
318	{
319	chPrev = ch;
320	cchWord++;
321	}
322	# else
323	while ((ch = pszSrc[cchWord]) != ' ' && ch != '\0')
324	cchWord++;
325	# endif
326	*ppszSrc = pszSrc;
327	return cchWord;
328	}
329
330	*ppszSrc = pszSrc;
331	return 0;
332	}
333
334
335	/**
336	* Analyzes a string.
337	*
338	* @param pThis The strint table compiler instance.
339	* @param pStr The string to analyze.
340	*/
341	static void bldProgStrTab_compressorAnalyzeString(PBLDPROGSTRTAB pThis, PBLDPROGSTRING pStr)
342	{
343	/*
344	* Mark all the string characters as used.
345	*/
346	const char *psz = pStr->pszString;
347	char ch;
348	while ((ch = *psz++) != '\0')
349	BldProgBitSet(pThis->bmUsedChars, (uint8_t)ch);
350
351	/*
352	* For now we just consider words.
353	*/
354	psz = pStr->pszString;
355	while ((ch = *psz) != '\0')
356	{
357	size_t cchWord = bldProgStrTab_compressorFindNextWord(psz, ch, &psz);
358	if (cchWord > 1)
359	{
360	std::string strWord(psz, cchWord);
361	BLDPROGWORDFREQMAP::iterator it = pThis->Frequencies.find(strWord);
362	if (it != pThis->Frequencies.end())
363	{
364	# ifdef BLDPROG_STRTAB_WITH_WORD_SEP_ALTERNATIVE
365	char const chSep = psz[cchWord];
366	if (chSep != '\0' && (it->second.chSep == chSep \|\| it->second.chSep == '\0'))
367	{
368	it->second.chSep = chSep;
369	it->second.cWithSep++;
370	}
371	else
372	# endif
373	it->second.cWithoutSep++;
374	}
375	else
376	{
377	# ifdef BLDPROG_STRTAB_WITH_WORD_SEP_ALTERNATIVE
378	char const chSep = psz[cchWord];
379	if (chSep != '\0')
380	{
381	BLDPROGWORDFREQSTATS const NewWord = { 0, 0, chSep };
382	pThis->Frequencies[strWord] = NewWord;
383	}
384	else
385	{
386	static BLDPROGWORDFREQSTATS const s_NewWord = { 0, 0, 0 };
387	pThis->Frequencies[strWord] = s_NewWord;
388	}
389	# else
390	pThis->Frequencies[strWord].cWithoutSep = 0;
391	# endif
392	}
393
394	# if 0 /** @todo need better accounting for overlapping alternatives before this can be enabled. */
395	/* Two words - immediate yields calc may lie when this enabled and we may pick the wrong words. */
396	if (ch == ' ')
397	{
398	ch = psz[++cchWord];
399	if (ch != ' ' && ch != '\0')
400	{
401	size_t const cchSaved = cchWord;
402
403	do
404	cchWord++;
405	while ((ch = psz[cchWord]) != ' ' && ch != '\0');
406
407	strWord = std::string(psz, cchWord);
408	BLDPROGWORDFREQMAP::iterator it = pThis->Frequencies.find(strWord);
409	if (it != pThis->Frequencies.end())
410	it->second += cchWord - 1;
411	else
412	pThis->Frequencies[strWord] = 0;
413
414	cchWord = cchSaved;
415	}
416	}
417	# endif
418	}
419	else if (!cchWord)
420	break;
421
422	/* Advance. */
423	psz += cchWord;
424	}
425	pStr->cchString = psz - pStr->pszString;
426	if (pStr->cchString > BLDPROG_STRTAB_MAX_STRLEN)
427	{
428	fprintf(stderr, "error: String to long (%u)\n", (unsigned)pStr->cchString);
429	abort();
430	}
431	}
432
433	#endif /* BLDPROG_STRTAB_WITH_COMPRESSION */
434
435	/**
436	* Adds a string to the hash table.
437	* @param pThis The strint table compiler instance.
438	* @param pStr The string.
439	*/
440	static void bldProgStrTab_AddStringToHashTab(PBLDPROGSTRTAB pThis, PBLDPROGSTRING pStr)
441	{
442	pStr->pNextRef = NULL;
443	pStr->pNextCollision = NULL;
444	pStr->offStrTab = 0;
445	pStr->uHash = sdbm(pStr->pszString, &pStr->cchString);
446	if (pStr->cchString > BLDPROG_STRTAB_MAX_STRLEN)
447	{
448	fprintf(stderr, "error: String to long (%u)\n", (unsigned)pStr->cchString);
449	exit(RTEXITCODE_FAILURE);
450	}
451
452	size_t idxHash = pStr->uHash % pThis->cStrHash;
453	PBLDPROGSTRING pCur = pThis->papStrHash[idxHash];
454	if (!pCur)
455	pThis->papStrHash[idxHash] = pStr;
456	else
457	{
458	/* Look for matching string. */
459	do
460	{
461	if ( pCur->uHash == pStr->uHash
462	&& pCur->cchString == pStr->cchString
463	&& memcmp(pCur->pszString, pStr->pszString, pStr->cchString) == 0)
464	{
465	pStr->pNextRef = pCur->pNextRef;
466	pCur->pNextRef = pStr;
467	pThis->cDuplicateStrings++;
468	return;
469	}
470	pCur = pCur->pNextCollision;
471	} while (pCur != NULL);
472
473	/* No matching string, insert. */
474	pThis->cCollisions++;
475	pStr->pNextCollision = pThis->papStrHash[idxHash];
476	pThis->papStrHash[idxHash] = pStr;
477	}
478
479	pThis->cUniqueStrings++;
480	pThis->cchUniqueStrings += pStr->cchString;
481	}
482
483
484	/**
485	* Adds a string to the string table.
486	*
487	* @param pThis The strint table compiler instance.
488	* @param pStr The string.
489	*/
490	static void BldProgStrTab_AddString(PBLDPROGSTRTAB pThis, PBLDPROGSTRING pStr)
491	{
492	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
493	bldProgStrTab_compressorAnalyzeString(pThis, pStr);
494	if (pThis->cPendingStrings < pThis->cMaxPendingStrings)
495	pThis->papPendingStrings[pThis->cPendingStrings++] = pStr;
496	else
497	abort();
498	#else
499	bldProgStrTab_AddStringToHashTab(pThis, pStr);
500	#endif
501	}
502
503
504	/**
505	* Adds a string to the string table.
506	*
507	* @param pThis The strint table compiler instance.
508	* @param pStr The string entry (uninitialized).
509	* @param psz The string, will be duplicated if compression is enabled.
510	*/
511	DECLINLINE(void) BldProgStrTab_AddStringDup(PBLDPROGSTRTAB pThis, PBLDPROGSTRING pStr, const char *psz)
512	{
513	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
514	pStr->pszString = strdup(psz);
515	if (!pStr->pszString)
516	abort();
517	#else
518	pStr->pszString = (char *)psz;
519	#endif
520	BldProgStrTab_AddString(pThis, pStr);
521	}
522
523	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
524
525	/**
526	* Copies @a cchSrc chars from @a pchSrc to @a pszDst, escaping special
527	* sequences.
528	*
529	* @returns New @a pszDst position, NULL if invalid source encoding.
530	* @param pszDst The destination buffer.
531	* @param pszSrc The source buffer.
532	* @param cchSrc How much to copy.
533	*/
534	static char bldProgStrTab_compressorCopyAndEscape(char pszDst, const char *pszSrc, size_t cchSrc)
535	{
536	while (cchSrc-- > 0)
537	{
538	char ch = *pszSrc;
539	if (!((unsigned char)ch & 0x80))
540	{
541	*pszDst++ = ch;
542	pszSrc++;
543	}
544	else
545	{
546	# ifdef BLDPROG_STRTAB_PURE_ASCII
547	fprintf(stderr, "error: unexpected char value %#x\n", ch);
548	return NULL;
549	# else
550	RTUNICP uc;
551	int rc = RTStrGetCpEx(&pszSrc, &uc);
552	if (RT_SUCCESS(rc))
553	{
554	pszDst++ = (unsigned char)0xff; / escape single code point. */
555	pszDst = RTStrPutCp(pszDst, uc);
556	}
557	else
558	{
559	fprintf(stderr, "Error: RTStrGetCpEx failed with rc=%d\n", rc);
560	return NULL;
561	}
562	# endif
563	}
564	}
565	return pszDst;
566	}
567
568
569	/**
570	* Replaces the dictionary words and escapes non-ascii chars in a string.
571	*
572	* @param pThis The strint table compiler instance.
573	* @param pString The string to fixup.
574	*/
575	static bool bldProgStrTab_compressorFixupString(PBLDPROGSTRTAB pThis, PBLDPROGSTRING pStr)
576	{
577	char szNew[BLDPROG_STRTAB_MAX_STRLEN * 2];
578	char *pszDst = szNew;
579	const char *pszSrc = pStr->pszString;
580	const char *pszSrcEnd = pszSrc + pStr->cchString;
581
582	char ch;
583	while ((ch = *pszSrc) != '\0')
584	{
585	const char * const pszSrcUncompressed = pszSrc;
586	size_t cchWord = bldProgStrTab_compressorFindNextWord(pszSrc, ch, &pszSrc);
587	size_t cchSrcUncompressed = pszSrc - pszSrcUncompressed;
588	if (cchSrcUncompressed > 0)
589	{
590	pszDst = bldProgStrTab_compressorCopyAndEscape(pszDst, pszSrcUncompressed, cchSrcUncompressed);
591	if (!pszDst)
592	return false;
593	}
594	if (!cchWord)
595	break;
596
597	/* Check for g_aWord matches. */
598	if (cchWord > 1)
599	{
600	size_t cchMax = pszSrcEnd - pszSrc;
601	for (unsigned i = 0; i < RT_ELEMENTS(pThis->aCompDict); i++)
602	{
603	size_t cchLen = pThis->aCompDict[i].cchString;
604	if ( cchLen >= cchWord
605	&& cchLen <= cchMax
606	&& memcmp(pThis->aCompDict[i].pszString, pszSrc, cchLen) == 0)
607	{
608	*pszDst++ = (unsigned char)i;
609	pszSrc += cchLen;
610	cchWord = 0;
611	break;
612	}
613	}
614	}
615
616	if (cchWord > 0)
617	{
618	/* Copy the current word. */
619	pszDst = bldProgStrTab_compressorCopyAndEscape(pszDst, pszSrc, cchWord);
620	if (!pszDst)
621	return false;
622	pszSrc += cchWord;
623	}
624	}
625
626	/* Just terminate it now. */
627	*pszDst = '\0';
628
629	/*
630	* Update the string.
631	*/
632	size_t cchNew = pszDst - &szNew[0];
633	if (cchNew > pStr->cchString)
634	{
635	pStr->pszString = (char *)malloc(cchNew + 1);
636	if (!pStr->pszString)
637	{
638	fprintf(stderr, "Out of memory!\n");
639	return false;
640	}
641	}
642	memcpy(pStr->pszString, szNew, cchNew + 1);
643	pStr->cchString = cchNew;
644
645	return true;
646	}
647
648
649	/**
650	* Entry in SortedDictionary.
651	*
652	* Uses variable length string member, so not class and allocated via malloc.
653	*/
654	struct SortedDictionaryEntry
655	{
656	size_t m_cchGain;
657	size_t m_cchString;
658	RT_FLEXIBLE_ARRAY_EXTENSION
659	char m_szString[RT_FLEXIBLE_ARRAY];
660
661	/** Allocates and initializes a new entry. */
662	static SortedDictionaryEntry allocate(const char a_pch, size_t a_cch, size_t a_cchGain, char a_chSep)
663	{
664	size_t cbString = a_cch + !!a_chSep + 1;
665	SortedDictionaryEntry pNew = (SortedDictionaryEntry )malloc(RT_UOFFSETOF(SortedDictionaryEntry, m_szString) + cbString);
666	if (pNew)
667	{
668	pNew->m_cchGain = a_cchGain;
669	memcpy(pNew->m_szString, a_pch, a_cch);
670	if (a_chSep)
671	pNew->m_szString[a_cch++] = a_chSep;
672	pNew->m_szString[a_cch] = '\0';
673	pNew->m_cchString = a_cch;
674	}
675	return pNew;
676	}
677
678
679	/** Compares this dictionary entry with an incoming one.
680	* @retval -1 if this entry is of less worth than the new one.
681	* @retval 0 if this entry is of equal worth to the new one.
682	* @retval +1 if this entry is of more worth than the new one.
683	*/
684	int compare(size_t a_cchGain, size_t a_cchString)
685	{
686	/* Higher gain is preferred of course: */
687	if (m_cchGain < a_cchGain)
688	return -1;
689	if (m_cchGain > a_cchGain)
690	return 1;
691
692	/* Gain is the same. Prefer the shorter string, as it will result in a shorter string table: */
693	if (m_cchString > a_cchString)
694	return -1;
695	if (m_cchString < a_cchString)
696	return 1;
697	return 0;
698	}
699	};
700
701
702	/**
703	* Insertion sort dictionary that keeps the 256 best words.
704	*
705	* Used by bldProgStrTab_compressorDoStringCompression to pick the dictionary
706	* words.
707	*/
708	class SortedDictionary
709	{
710	public:
711	size_t m_cEntries;
712	SortedDictionaryEntry *m_apEntries[256];
713
714	SortedDictionary()
715	: m_cEntries(0)
716	{
717	for (size_t i = 0; i < RT_ELEMENTS(m_apEntries); i++)
718	m_apEntries[i] = NULL;
719	}
720
721	~SortedDictionary()
722	{
723	while (m_cEntries > 0)
724	{
725	free(m_apEntries[--m_cEntries]);
726	m_apEntries[m_cEntries] = NULL;
727	}
728	}
729
730
731	/**
732	* Inserts a new entry, if it's worth it.
733	* @returns true on succes, false if out of memory.
734	*/
735	bool insert(const char *a_pchString, size_t a_cchStringBase, size_t a_cchGain, char a_chSep = 0)
736	{
737	size_t const cchString = a_cchStringBase + (a_chSep + 1);
738
739	/*
740	* Drop the insert if the symbol table is full and the insert is less worth the last entry:
741	*/
742	if ( m_cEntries >= RT_ELEMENTS(m_apEntries)
743	&& m_apEntries[RT_ELEMENTS(m_apEntries) - 1]->compare(a_cchGain, cchString) >= 0)
744	return true;
745
746	/*
747	* Create a new entry to insert.
748	*/
749	SortedDictionaryEntry *pNewEntry = SortedDictionaryEntry::allocate(a_pchString, a_cchStringBase, a_cchGain, a_chSep);
750	if (!pNewEntry)
751	return false;
752
753	/*
754	* Find the insert point.
755	*/
756	if (m_cEntries == 0)
757	{
758	m_apEntries[0] = pNewEntry;
759	m_cEntries = 1;
760	}
761	else
762	{
763	/* If table is full, drop the last entry before we start (already made
764	sure the incoming entry is preferable to the one were dropping): */
765	if (m_cEntries >= RT_ELEMENTS(m_apEntries))
766	{
767	free(m_apEntries[RT_ELEMENTS(m_apEntries) - 1]);
768	m_apEntries[RT_ELEMENTS(m_apEntries) - 1] = NULL;
769	m_cEntries = RT_ELEMENTS(m_apEntries) - 1;
770	}
771
772	/* Find where to insert the new entry: */
773	/** @todo use binary search. */
774	size_t i = m_cEntries;
775	while (i > 0 && m_apEntries[i - 1]->compare(a_cchGain, cchString) < 0)
776	i--;
777
778	/* Shift entries to make room and insert the new entry. */
779	if (i < m_cEntries)
780	memmove(&m_apEntries[i + 1], &m_apEntries[i], (m_cEntries - i) * sizeof(m_apEntries[0]));
781	m_apEntries[i] = pNewEntry;
782	m_cEntries++;
783	}
784	return true;
785	}
786	};
787
788
789	/**
790	* Compresses the vendor and product strings.
791	*
792	* This is very very simple (a lot less work than the string table for
793	* instance).
794	*/
795	static bool bldProgStrTab_compressorDoStringCompression(PBLDPROGSTRTAB pThis, bool fVerbose)
796	{
797	/*
798	* Sort the frequency analyzis result and pick the top entries for any
799	* available dictionary slots.
800	*/
801	SortedDictionary SortedDict;
802	for (BLDPROGWORDFREQMAP::iterator it = pThis->Frequencies.begin(); it != pThis->Frequencies.end(); ++it)
803	{
804	bool fInsert;
805	size_t const cchString = it->first.length();
806	# ifndef BLDPROG_STRTAB_WITH_WORD_SEP_ALTERNATIVE
807	size_t const cchGainWithout = it->second.cWithoutSep * cchString;
808	# else
809	size_t const cchGainWithout = (it->second.cWithoutSep + it->second.cWithSep) * cchString;
810	size_t const cchGainWith = it->second.cWithSep * (cchString + 1);
811	if (cchGainWith > cchGainWithout)
812	fInsert = SortedDict.insert(it->first.c_str(), cchString, cchGainWith, it->second.chSep);
813	else
814	# endif
815	fInsert = SortedDict.insert(it->first.c_str(), cchString, cchGainWithout);
816	if (!fInsert)
817	return false;
818	}
819
820	size_t cb = 0;
821	size_t cWords = 0;
822	size_t iDict = 0;
823	# ifdef BLDPROG_STRTAB_PURE_ASCII
824	for (size_t i = 0; i < RT_ELEMENTS(pThis->aCompDict); i++)
825	# else
826	for (size_t i = 0; i < RT_ELEMENTS(pThis->aCompDict) - 1; i++)
827	# endif
828	{
829	char szTmp[2] = { (char)i, '\0' };
830	const char *psz = szTmp;
831	if ( BldProgBitIsSet(pThis->bmUsedChars, i)
832	\|\| iDict >= SortedDict.m_cEntries)
833	{
834	/* character entry */
835	pThis->auCompDictFreq[i] = 0;
836	pThis->aCompDict[i].cchString = 1;
837	}
838	else
839	{
840	/* word entry */
841	cb += SortedDict.m_apEntries[iDict]->m_cchGain;
842	pThis->auCompDictFreq[i] = SortedDict.m_apEntries[iDict]->m_cchGain;
843	pThis->aCompDict[i].cchString = SortedDict.m_apEntries[iDict]->m_cchString;
844	psz = SortedDict.m_apEntries[iDict]->m_szString;
845	cWords++;
846	iDict++;
847	}
848	pThis->aCompDict[i].pszString = (char *)malloc(pThis->aCompDict[i].cchString + 1);
849	if (pThis->aCompDict[i].pszString)
850	memcpy(pThis->aCompDict[i].pszString, psz, pThis->aCompDict[i].cchString + 1);
851	else
852	return false;
853	}
854
855	if (fVerbose)
856	printf("debug: Estimated string compression saving: %u bytes\n"
857	"debug: %u words, %u characters\n"
858	, (unsigned)cb, (unsigned)cWords, (unsigned)(RT_ELEMENTS(pThis->aCompDict) - cWords));
859
860	/*
861	* Rework the strings.
862	*/
863	size_t cchOld = 0;
864	size_t cchOldMax = 0;
865	size_t cchOldMin = BLDPROG_STRTAB_MAX_STRLEN;
866	size_t cchNew = 0;
867	size_t cchNewMax = 0;
868	size_t cchNewMin = BLDPROG_STRTAB_MAX_STRLEN;
869	size_t i = pThis->cPendingStrings;
870	while (i-- > 0)
871	{
872	PBLDPROGSTRING pCurStr = pThis->papPendingStrings[i];
873	cchOld += pCurStr->cchString;
874	if (pCurStr->cchString > cchOldMax)
875	cchOldMax = pCurStr->cchString;
876	if (pCurStr->cchString < cchOldMin)
877	cchOldMin = pCurStr->cchString;
878
879	if (!bldProgStrTab_compressorFixupString(pThis, pCurStr))
880	return false;
881
882	cchNew += pCurStr->cchString;
883	if (pCurStr->cchString > cchNewMax)
884	cchNewMax = pCurStr->cchString;
885	if (pCurStr->cchString < cchNewMin)
886	cchNewMin = pCurStr->cchString;
887
888	bldProgStrTab_AddStringToHashTab(pThis, pCurStr);
889	}
890
891	/*
892	* Do debug stats.
893	*/
894	if (fVerbose)
895	{
896	for (i = 0; i < RT_ELEMENTS(pThis->aCompDict); i++)
897	cchNew += pThis->aCompDict[i].cchString + 1;
898
899	printf("debug: Strings: original: %u bytes; compressed: %u bytes;", (unsigned)cchOld, (unsigned)cchNew);
900	if (cchNew < cchOld)
901	printf(" saving %u bytes (%u%%)\n", (unsigned)(cchOld - cchNew), (unsigned)((cchOld - cchNew) * 100 / cchOld));
902	else
903	printf(" wasting %u bytes!\n", (unsigned)(cchOld - cchNew));
904	printf("debug: Original string lengths: average %u; min %u; max %u\n",
905	(unsigned)(cchOld / pThis->cPendingStrings), (unsigned)cchOldMin, (unsigned)cchOldMax);
906	printf("debug: Compressed string lengths: average %u; min %u; max %u\n",
907	(unsigned)(cchNew / pThis->cPendingStrings), (unsigned)cchNewMin, (unsigned)cchNewMax);
908	}
909
910	return true;
911	}
912
913	#endif /* BLDPROG_STRTAB_WITH_COMPRESSION */
914
915	/**
916	* Inserts a string into g_apUniqueStrings.
917	* @param pThis The strint table compiler instance.
918	* @param pStr The string.
919	*/
920	static void bldProgStrTab_InsertUniqueString(PBLDPROGSTRTAB pThis, PBLDPROGSTRING pStr)
921	{
922	size_t iIdx;
923	size_t iEnd = pThis->cSortedStrings;
924	if (iEnd)
925	{
926	size_t iStart = 0;
927	for (;;)
928	{
929	iIdx = iStart + (iEnd - iStart) / 2;
930	if (pThis->papSortedStrings[iIdx]->cchString < pStr->cchString)
931	{
932	if (iIdx <= iStart)
933	break;
934	iEnd = iIdx;
935	}
936	else if (pThis->papSortedStrings[iIdx]->cchString > pStr->cchString)
937	{
938	if (++iIdx >= iEnd)
939	break;
940	iStart = iIdx;
941	}
942	else
943	break;
944	}
945
946	if (iIdx != pThis->cSortedStrings)
947	memmove(&pThis->papSortedStrings[iIdx + 1], &pThis->papSortedStrings[iIdx],
948	(pThis->cSortedStrings - iIdx) * sizeof(pThis->papSortedStrings[iIdx]));
949	}
950	else
951	iIdx = 0;
952
953	pThis->papSortedStrings[iIdx] = pStr;
954	pThis->cSortedStrings++;
955	}
956
957
958	/**
959	* Compiles the string table after the string has been added.
960	*
961	* This will save space by dropping string terminators, eliminating duplicates
962	* and try find strings that are sub-strings of others.
963	*
964	* Will initialize the StrRef of all StrTabString instances.
965	*
966	* @returns success indicator (error flagged).
967	* @param pThis The strint table compiler instance.
968	*/
969	static bool BldProgStrTab_CompileIt(PBLDPROGSTRTAB pThis, bool fVerbose)
970	{
971	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
972	/*
973	* Do the compression and add all the compressed strings to the table.
974	*/
975	if (!bldProgStrTab_compressorDoStringCompression(pThis, fVerbose))
976	return false;
977
978	/*
979	* Add the dictionary strings.
980	*/
981	for (unsigned i = 0; i < RT_ELEMENTS(pThis->aCompDict); i++)
982	if (pThis->aCompDict[i].cchString > 1)
983	bldProgStrTab_AddStringToHashTab(pThis, &pThis->aCompDict[i]);
984	# ifdef RT_STRICT
985	else if ( pThis->aCompDict[i].cchString != 1
986	# ifndef BLDPROG_STRTAB_PURE_ASCII
987	&& i != 0xff
988	# endif
989	)
990	abort();
991	# endif
992	#endif
993	if (fVerbose)
994	printf("debug: %u unique strings (%u bytes), %u duplicates, %u collisions\n",
995	(unsigned)pThis->cUniqueStrings, (unsigned)pThis->cchUniqueStrings,
996	(unsigned)pThis->cDuplicateStrings, (unsigned)pThis->cCollisions);
997
998	/*
999	* Create papSortedStrings from the hash table. The table is sorted by
1000	* string length, with the longer strings first, so that we increase our
1001	* chances of locating duplicate substrings.
1002	*/
1003	pThis->papSortedStrings = (PBLDPROGSTRING )malloc(sizeof(pThis->papSortedStrings[0]) pThis->cUniqueStrings);
1004	if (!pThis->papSortedStrings)
1005	return false;
1006	pThis->cSortedStrings = 0;
1007	size_t idxHash = pThis->cStrHash;
1008	while (idxHash-- > 0)
1009	{
1010	PBLDPROGSTRING pCur = pThis->papStrHash[idxHash];
1011	if (pCur)
1012	{
1013	do
1014	{
1015	bldProgStrTab_InsertUniqueString(pThis, pCur);
1016	pCur = pCur->pNextCollision;
1017	} while (pCur);
1018	}
1019	}
1020
1021	/*
1022	* Create the actual string table.
1023	*/
1024	pThis->pachStrTab = (char *)malloc(pThis->cchUniqueStrings + 1);
1025	if (!pThis->pachStrTab)
1026	return false;
1027	pThis->cchStrTab = 0;
1028	for (size_t i = 0; i < pThis->cSortedStrings; i++)
1029	{
1030	PBLDPROGSTRING pCur = pThis->papSortedStrings[i];
1031	const char * const pszCur = pCur->pszString;
1032	size_t const cchCur = pCur->cchString;
1033	size_t offStrTab = pThis->cchStrTab;
1034
1035	/*
1036	* See if the string is a substring already found in the string table.
1037	* Excluding the zero terminator increases the chances for this.
1038	*/
1039	size_t cchLeft = pThis->cchStrTab >= cchCur ? pThis->cchStrTab - cchCur : 0;
1040	const char *pchLeft = pThis->pachStrTab;
1041	char const chFirst = *pszCur;
1042	while (cchLeft > 0)
1043	{
1044	const char pchCandidate = (const char )memchr(pchLeft, chFirst, cchLeft);
1045	if (!pchCandidate)
1046	break;
1047	if (memcmp(pchCandidate, pszCur, cchCur) == 0)
1048	{
1049	offStrTab = pchCandidate - pThis->pachStrTab;
1050	break;
1051	}
1052
1053	cchLeft -= pchCandidate + 1 - pchLeft;
1054	pchLeft = pchCandidate + 1;
1055	}
1056
1057	if (offStrTab == pThis->cchStrTab)
1058	{
1059	/*
1060	* See if the start of the string overlaps the end of the string table.
1061	*/
1062	if (pThis->cchStrTab && cchCur > 1)
1063	{
1064	cchLeft = RT_MIN(pThis->cchStrTab, cchCur - 1);
1065	pchLeft = &pThis->pachStrTab[pThis->cchStrTab - cchLeft];
1066	while (cchLeft > 0)
1067	{
1068	const char pchCandidate = (const char )memchr(pchLeft, chFirst, cchLeft);
1069	if (!pchCandidate)
1070	break;
1071	cchLeft -= pchCandidate - pchLeft;
1072	pchLeft = pchCandidate;
1073	if (memcmp(pchLeft, pszCur, cchLeft) == 0)
1074	{
1075	size_t cchToCopy = cchCur - cchLeft;
1076	memcpy(&pThis->pachStrTab[offStrTab], &pszCur[cchLeft], cchToCopy);
1077	pThis->cchStrTab += cchToCopy;
1078	offStrTab = pchCandidate - pThis->pachStrTab;
1079	break;
1080	}
1081	cchLeft--;
1082	pchLeft++;
1083	}
1084	}
1085
1086	/*
1087	* If we didn't have any luck above, just append the string.
1088	*/
1089	if (offStrTab == pThis->cchStrTab)
1090	{
1091	memcpy(&pThis->pachStrTab[offStrTab], pszCur, cchCur);
1092	pThis->cchStrTab += cchCur;
1093	}
1094	}
1095
1096	/*
1097	* Set the string table offset for all the references to this string.
1098	*/
1099	do
1100	{
1101	pCur->offStrTab = (uint32_t)offStrTab;
1102	pCur = pCur->pNextRef;
1103	} while (pCur != NULL);
1104	}
1105
1106	if (fVerbose)
1107	printf("debug: String table: %u bytes\n", (unsigned)pThis->cchStrTab);
1108	return true;
1109	}
1110
1111
1112	#ifdef RT_STRICT
1113	/**
1114	* Sanity checks a string table string.
1115	* @param pThis The strint table compiler instance.
1116	* @param pStr The string to check.
1117	*/
1118	static void BldProgStrTab_CheckStrTabString(PBLDPROGSTRTAB pThis, PBLDPROGSTRING pStr)
1119	{
1120	if (pStr->cchString != strlen(pStr->pszString))
1121	abort();
1122	if (pStr->offStrTab >= pThis->cchStrTab)
1123	abort();
1124	if (pStr->offStrTab + pStr->cchString > pThis->cchStrTab)
1125	abort();
1126	if (memcmp(pStr->pszString, &pThis->pachStrTab[pStr->offStrTab], pStr->cchString) != 0)
1127	abort();
1128	}
1129	#endif
1130
1131
1132	/**
1133	* Output the string table string in C string litteral fashion.
1134	*
1135	* @param pThis The strint table instance.
1136	* @param pString The string to print.
1137	* @param pOut The output stream.
1138	*/
1139	static void BldProgStrTab_PrintCStringLitteral(PBLDPROGSTRTAB pThis, PBLDPROGSTRING pString, FILE *pOut)
1140	{
1141	const unsigned char psz = (const unsigned char )pString->pszString;
1142	unsigned char uch;
1143	while ((uch = *psz++) != '\0')
1144	{
1145	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
1146	if (pThis->aCompDict[uch].cchString == 1)
1147	#else
1148	if (!(uch & 0x80))
1149	#endif
1150	{
1151	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
1152	if (uch == 0xff)
1153	abort();
1154	#endif
1155	if (uch != '\'' && uch != '\\')
1156	fputc((char)uch, pOut);
1157	else
1158	{
1159	fputc('\\', pOut);
1160	fputc((char)uch, pOut);
1161	}
1162	}
1163	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
1164	# ifndef BLDPROG_STRTAB_PURE_ASCII
1165	else if (uch == 0xff)
1166	{
1167	RTUNICP uc = RTStrGetCp((const char *)psz);
1168	psz += RTStrCpSize(uc);
1169	fprintf(pOut, "\\u%04x", uc);
1170	}
1171	# endif
1172	else
1173	fputs(pThis->aCompDict[uch].pszString, pOut);
1174	#else
1175	else
1176	fprintf(pOut, "\\x%02x", (unsigned)uch);
1177	NOREF(pThis);
1178	#endif
1179	}
1180	}
1181
1182
1183	/**
1184	* Writes the string table code to the output stream.
1185	*
1186	* @param pThis The strint table compiler instance.
1187	* @param pOut The output stream.
1188	* @param pszScope The scoping ("static " or empty string),
1189	* including trailing space.
1190	* @param pszPrefix The variable prefix. Typically "g_" for C programs,
1191	* whereas C++ classes normally use "class::s_g".
1192	* @param pszBaseName The base name for the variables. The user
1193	* accessible variable will end up as just the
1194	* prefix and basename concatenated.
1195	*/
1196	static void BldProgStrTab_WriteStringTable(PBLDPROGSTRTAB pThis, FILE *pOut,
1197	const char pszScope, const char pszPrefix, const char *pszBaseName)
1198	{
1199	#ifdef RT_STRICT
1200	/*
1201	* Do some quick sanity checks while we're here.
1202	*/
1203	# ifdef BLDPROG_STRTAB_WITH_COMPRESSION
1204	for (unsigned i = 0; i < RT_ELEMENTS(pThis->aCompDict); i++)
1205	{
1206	if (BldProgBitIsSet(pThis->bmUsedChars, i)
1207	# ifdef BLDPROG_STRTAB_PURE_ASCII
1208	? pThis->aCompDict[i].cchString != 1U
1209	# else
1210	? pThis->aCompDict[i].cchString != (i != 0xffU ? 1U : 0U)
1211	# endif
1212	: pThis->aCompDict[i].cchString < 1U)
1213	abort();
1214	if (pThis->aCompDict[i].cchString > 1U)
1215	BldProgStrTab_CheckStrTabString(pThis, &pThis->aCompDict[i]);
1216	}
1217	# endif
1218	#endif
1219
1220	/*
1221	* Create a table for speeding up the character categorization.
1222	*/
1223	uint8_t abCharCat[256];
1224	memset(abCharCat, 0, sizeof(abCharCat));
1225	abCharCat[(unsigned char)'\\'] = 1;
1226	abCharCat[(unsigned char)'\''] = 1;
1227	for (unsigned i = 0; i < 0x20; i++)
1228	abCharCat[i] = 2;
1229	for (unsigned i = 0x7f; i < 0x100; i++)
1230	abCharCat[i] = 2;
1231	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
1232	for (unsigned i = 0; i < 0x100; i++)
1233	if (!BldProgBitIsSet(pThis->bmUsedChars, i)) /* Encode table references using '\xYY'. */
1234	abCharCat[i] = 2;
1235	#endif
1236
1237	/*
1238	* We follow the sorted string table, one string per line.
1239	*/
1240	fprintf(pOut,
1241	"#include <iprt/bldprog-strtab.h>\n"
1242	"\n"
1243	"static const char g_achStrTab%s[] =\n"
1244	"{\n",
1245	pszBaseName);
1246
1247	uint32_t off = 0;
1248	for (uint32_t i = 0; i < pThis->cSortedStrings; i++)
1249	{
1250	PBLDPROGSTRING pCur = pThis->papSortedStrings[i];
1251	uint32_t offEnd = pCur->offStrTab + (uint32_t)pCur->cchString;
1252	if (offEnd > off)
1253	{
1254	/* Comment with an uncompressed and more readable version of the string. */
1255	if (off == pCur->offStrTab)
1256	fprintf(pOut, "/* 0x%05x = \"", off);
1257	else
1258	fprintf(pOut, "/* 0X%05x = \"", off);
1259	BldProgStrTab_PrintCStringLitteral(pThis, pCur, pOut);
1260	fputs("\" */\n", pOut);
1261
1262	/* Must use char by char here or we may trigger the max string
1263	length limit in the compiler, */
1264	fputs(" ", pOut);
1265	for (; off < offEnd; off++)
1266	{
1267	unsigned char uch = pThis->pachStrTab[off];
1268	fputc('\'', pOut);
1269	if (abCharCat[uch] == 0)
1270	fputc(uch, pOut);
1271	else if (abCharCat[uch] != 1)
1272	fprintf(pOut, "\\x%02x", (unsigned)uch);
1273	else
1274	{
1275	fputc('\\', pOut);
1276	fputc(uch, pOut);
1277	}
1278	fputc('\'', pOut);
1279	fputc(',', pOut);
1280	}
1281	fputc('\n', pOut);
1282	}
1283	}
1284
1285	fprintf(pOut,
1286	"};\n"
1287	"AssertCompile(sizeof(g_achStrTab%s) == %#x);\n\n",
1288	pszBaseName, (unsigned)pThis->cchStrTab);
1289
1290	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
1291	/*
1292	* Write the compression dictionary.
1293	*/
1294	fprintf(pOut,
1295	"static const RTBLDPROGSTRREF g_aCompDict%s[%u] = \n"
1296	"{\n",
1297	pszBaseName, (unsigned)RT_ELEMENTS(pThis->aCompDict));
1298	for (unsigned i = 0; i < RT_ELEMENTS(pThis->aCompDict); i++)
1299	if (pThis->aCompDict[i].cchString > 1)
1300	fprintf(pOut, " /[%3u]=/ { %#08x, %#04x }, // %6lu - %s\n", i,
1301	pThis->aCompDict[i].offStrTab, (unsigned)pThis->aCompDict[i].cchString,
1302	(unsigned long)pThis->auCompDictFreq[i], pThis->aCompDict[i].pszString);
1303	# ifndef BLDPROG_STRTAB_PURE_ASCII
1304	else if (i == 0xff)
1305	fprintf(pOut, " /[%3u]=/ { 0x000000, 0x00 }, // UTF-8 escape\n", i);
1306	# endif
1307	else if (i == 0)
1308	fprintf(pOut, " /[%3u]=/ { 0x000000, 0x00 }, // unused, because zero terminator\n", i);
1309	else if (i < 0x20 \|\| i >= 0x7f)
1310	fprintf(pOut, " /[%3u]=/ { 0x000000, 0x00 }, // %02x\n", i, i);
1311	else
1312	fprintf(pOut, " /[%3u]=/ { 0x000000, 0x00 }, // '%c'\n", i, (char)i);
1313	fprintf(pOut, "};\n\n");
1314	#endif
1315
1316
1317	/*
1318	* Write the string table data structure.
1319	*/
1320	fprintf(pOut,
1321	"%sconst RTBLDPROGSTRTAB %s%s = \n"
1322	"{\n"
1323	" /.pchStrTab = / &g_achStrTab%s[0],\n"
1324	" /.cchStrTab = / sizeof(g_achStrTab%s),\n"
1325	,
1326	pszScope, pszPrefix, pszBaseName, pszBaseName, pszBaseName);
1327	#ifdef BLDPROG_STRTAB_WITH_COMPRESSION
1328	fprintf(pOut,
1329	" /.cCompDict = / %u,\n"
1330	" /.paCompDict = / &g_aCompDict%s[0]\n"
1331	"};\n"
1332	# ifndef BLDPROG_STRTAB_PURE_ASCII /* 255 or 256 entries is how the decoder knows */
1333	, (unsigned)RT_ELEMENTS(pThis->aCompDict) - 1,
1334	# else
1335	, (unsigned)RT_ELEMENTS(pThis->aCompDict),
1336	# endif
1337	pszBaseName);
1338	#else
1339	fprintf(pOut,
1340	" /.cCompDict = / 0,\n"
1341	" /.paCompDict = / NULL\n"
1342	"};\n");
1343	#endif
1344	}
1345
1346	#if RT_CLANG_PREREQ(4, 0) \|\| RT_GNUC_PREREQ(4, 6)
1347	# pragma GCC diagnostic pop
1348	#endif
1349
1350	#endif /* __cplusplus && IN_RING3 */
1351

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

source: vbox/trunk/include/iprt/bldprog-strtab-template.cpp.h@ 104212

Download in other formats: