fileinfo: libmagic: Simplify libmagic patch

For the time being, the new args passing pre-compiled regex are cut out.
On the PHP side, PCRE patterns are cached and thus a duplicated
mechanism isn't necessary.

There might be still advantages to carry pre-compiled patterns for
libmagic separately, but reusing this mechanism needs a better
consideration, as it might require to retain from the PHP internal pattern
caching mechanism.

Signed-off-by: Anatol Belski <ab@php.net>

Author: weltling

ext/fileinfo/libmagic/apprentice.c

@@ -147,10 +147,7 @@ private uint64_t swap8(uint64_t);
 #endif
 
 private char *mkdbname(struct magic_set *, const char *, int);
-private struct magic_map *apprentice_buf(struct magic_set *, struct magic *,
-    size_t);
 private struct magic_map *apprentice_map(struct magic_set *, const char *);
-private int check_buffer(struct magic_set *, struct magic_map *, const char *);
 private void apprentice_unmap(struct magic_map *);
 private int apprentice_compile(struct magic_set *, struct magic_map *,
     const char *);
@@ -597,8 +594,6 @@ file_ms_alloc(int flags)
 private void
 apprentice_unmap(struct magic_map *map)
 {
-	size_t i;
-	char *p;
 	if (map == NULL)
 		return;
 	if (map->p != php_magic_database) {
@@ -1316,7 +1311,7 @@ load_1(struct magic_set *ms, int action, const char *fn, int *errs,
 					continue;
 				}
 				if ((*bang[i].fun)(ms, &me,
-				    line + bang[i].len + 2,
+				    line + bang[i].len + 2, 
 				    len - bang[i].len - 2) != 0) {
 					(*errs)++;
 					continue;
@@ -3648,7 +3643,6 @@ file_magicfind(struct magic_set *ms, const char *name, struct mlist *v)
 				continue;
 			if (strcmp(ma[i].value.s, name) == 0) {
 				v->magic = &ma[i];
-				v->magic_rxcomp = &(ml->magic_rxcomp[i]);
 				for (j = i + 1; j < ml->nmagic; j++)
 				    if (ma[j].cont_level == 0)
 					    break;

ext/fileinfo/libmagic/file.h

@@ -164,8 +164,6 @@
 #define FILE_COMPILE	2
 #define FILE_LIST	3
 
-typedef void* file_regex_t;
-
 struct buffer {
 	int fd;
 	zend_stat_t st;
@@ -405,7 +403,6 @@ struct magic {
 /* list of magic entries */
 struct mlist {
 	struct magic *magic;		/* array of magic entries */
-	file_regex_t **magic_rxcomp;	/* array of compiled regexps */
 	size_t nmagic;			/* number of entries in array */
 	void *map;			/* internal resources used by entry */
 	struct mlist *next, *prev;

ext/fileinfo/libmagic/funcs.c

@@ -132,7 +132,7 @@ file_checkfmt(char *msg, size_t mlen, const char *fmt)
 protected int
 file_vprintf(struct magic_set *ms, const char *fmt, va_list ap)
 {
-	int len;
+	size_t len;
 	char *buf, *newstr;
 	char tbuf[1024];
 
@@ -150,7 +150,7 @@ file_vprintf(struct magic_set *ms, const char *fmt, va_list ap)
 		size_t blen = ms->o.blen;
 		if (buf) efree(buf);
 		file_clearbuf(ms);
-		file_error(ms, 0, "Output buffer space exceeded %d+%"
+		file_error(ms, 0, "Output buffer space exceeded %" SIZE_T_FORMAT "u+%"
 		    SIZE_T_FORMAT "u", len, blen);
 		return -1;
 	}

ext/fileinfo/libmagic/readcdf.c

@@ -145,7 +145,7 @@ cdf_file_property_info(struct magic_set *ms, const cdf_property_info_t *info,
 
 	memset(&ts, 0, sizeof(ts));
 
-        if (!NOTMIME(ms) && root_storage)
+	if (!NOTMIME(ms) && root_storage)
 		str = cdf_clsid_to_mime(root_storage->d_storage_uuid,
 		    clsid2mime);
 

ext/fileinfo/libmagic/softmagic.c

@@ -43,7 +43,7 @@ FILE_RCSID("@(#)$File: softmagic.c,v 1.328 2022/09/13 18:46:07 christos Exp $")
 #include <time.h>
 #include "der.h"
 
-private int match(struct magic_set *, struct magic *, file_regex_t **, size_t,
+private int match(struct magic_set *, struct magic *, size_t,
     const struct buffer *, size_t, int, int, int, uint16_t *,
     uint16_t *, int *, int *, int *, int *);
 private int mget(struct magic_set *, struct magic *, const struct buffer *,
@@ -52,7 +52,7 @@ private int mget(struct magic_set *, struct magic *, const struct buffer *,
     uint16_t *, int *, int *, int *, int *);
 private int msetoffset(struct magic_set *, struct magic *, struct buffer *,
     const struct buffer *, size_t, unsigned int);
-private int magiccheck(struct magic_set *, struct magic *, file_regex_t **);
+private int magiccheck(struct magic_set *, struct magic *);
 private int mprint(struct magic_set *, struct magic *);
 private int moffset(struct magic_set *, struct magic *, const struct buffer *,
     int32_t *);
@@ -131,7 +131,7 @@ file_softmagic(struct magic_set *ms, const struct buffer *b,
 	}
 
 	for (ml = ms->mlist[0]->next; ml != ms->mlist[0]; ml = ml->next) {
-		int ret = match(ms, ml->magic, ml->magic_rxcomp, ml->nmagic, b,
+		int ret = match(ms, ml->magic, ml->nmagic, b,
 		    0, mode, text, 0, indir_count, name_count,
 		    &printed_something, &need_separator, NULL, NULL);
 		switch (ret) {
@@ -206,7 +206,7 @@ file_fmtcheck(struct magic_set *ms, const char *desc, const char *def,
  *	so that higher-level continuations are processed.
  */
 private int
-match(struct magic_set *ms, struct magic *magic, file_regex_t **magic_rxcomp,
+match(struct magic_set *ms, struct magic *magic,
     size_t nmagic, const struct buffer *b, size_t offset, int mode, int text,
     int flip, uint16_t *indir_count, uint16_t *name_count,
     int *printed_something, int *need_separator, int *returnval,
@@ -236,7 +236,6 @@ match(struct magic_set *ms, struct magic *magic, file_regex_t **magic_rxcomp,
 	for (magindex = 0; magindex < nmagic; magindex++) {
 		int flush = 0;
 		struct magic *m = &magic[magindex];
-		file_regex_t **m_rxcomp = &magic_rxcomp[magindex];
 
 		if (m->type != FILE_NAME)
 		if ((IS_LIBMAGIC_STRING(m->type) &&
@@ -274,7 +273,7 @@ match(struct magic_set *ms, struct magic *magic, file_regex_t **magic_rxcomp,
 				*returnval = 1;
 			}
 
-			switch (magiccheck(ms, m, m_rxcomp)) {
+			switch (magiccheck(ms, m)) {
 			case -1:
 				return -1;
 			case 0:
@@ -335,7 +334,6 @@ match(struct magic_set *ms, struct magic *magic, file_regex_t **magic_rxcomp,
 		while (magindex + 1 < nmagic &&
 		    magic[magindex + 1].cont_level != 0) {
 			m = &magic[++magindex];
-			m_rxcomp = &magic_rxcomp[magindex];
 			ms->line = m->lineno; /* for messages */
 
 			if (cont_level < m->cont_level)
@@ -389,7 +387,7 @@ match(struct magic_set *ms, struct magic *magic, file_regex_t **magic_rxcomp,
 				break;
 			}
 
-			switch (flush ? 1 : magiccheck(ms, m, m_rxcomp)) {
+			switch (flush ? 1 : magiccheck(ms, m)) {
 			case -1:
 				return -1;
 			case 0:
@@ -1870,7 +1868,7 @@ mget(struct magic_set *ms, struct magic *m, const struct buffer *b,
 		for (mlp = ms->mlist[0]->next; mlp != ms->mlist[0];
 		    mlp = mlp->next)
 		{
-			if ((rv = match(ms, mlp->magic, mlp->magic_rxcomp,
+			if ((rv = match(ms, mlp->magic,
 			    mlp->nmagic, &bb, 0, BINTEST, text, 0, indir_count,
 			    name_count, printed_something, need_separator,
 			    NULL, NULL)) != 0)
@@ -1923,7 +1921,7 @@ mget(struct magic_set *ms, struct magic *m, const struct buffer *b,
 		nfound_match = 0;
 		(*name_count)++;
 		eoffset = ms->eoffset;
-		rv = match(ms, ml.magic, ml.magic_rxcomp, ml.nmagic, b,
+		rv = match(ms, ml.magic, ml.nmagic, b,
 		    offset + o, mode, text, flip, indir_count, name_count,
 		    printed_something, need_separator, returnval,
 		    &nfound_match);
@@ -2047,7 +2045,7 @@ file_strncmp16(const char *a, const char *b, size_t len, size_t maxlen,
 }
 
 private int
-magiccheck(struct magic_set *ms, struct magic *m, file_regex_t **m_cache)
+magiccheck(struct magic_set *ms, struct magic *m)
 {
 	uint64_t l = m->value.q;
 	uint64_t v;