driver: -s <replstr> support: implementation and integration.

7 files changed, 346 insertions, 4 deletions

diff --git a/include/tpax/tpax.h b/include/tpax/tpax.h
index 00271cb..fa685c3 100644
--- a/include/tpax/tpax.h
+++ b/include/tpax/tpax.h
@@ -1,6 +1,7 @@
 #ifndef TPAX_H
 #define TPAX_H
 
+#include <regex.h>
 #include <stdint.h>
 #include <stddef.h>
 #include <sys/stat.h>
@@ -172,6 +173,9 @@ tpax_api int  tpax_archive_seal             (const struct tpax_driver_ctx *);
 tpax_api int  tpax_util_path_copy           (char *, const char *, size_t, uint32_t, size_t *);
 tpax_api int  tpax_util_stat_compare        (const struct stat *, const struct stat *);
 
+tpax_api int  tpax_util_path_replstr        (char * dstpath, const char * srcpath, const char * replstr,
+                                             const regex_t * regex, size_t buflen, int flags);
+
 /* utility api */
 tpax_api int  tpax_main                     (char **, char **,
                                              const struct tpax_fd_ctx *);
diff --git a/project/common.mk b/project/common.mk
index 26e171c..03e35ee 100644
--- a/project/common.mk
+++ b/project/common.mk
@@ -12,6 +12,7 @@ API_SRCS = \
 	src/output/tpax_output_error.c \
 	src/skin/tpax_skin_default.c \
 	src/util/tpax_path_copy.c \
+	src/util/tpax_path_replstr.c \
 	src/util/tpax_stat_compare.c \
 
 INTERNAL_SRCS = \
diff --git a/src/driver/tpax_driver_ctx.c b/src/driver/tpax_driver_ctx.c
index dff6ef6..5593ec6 100644
--- a/src/driver/tpax_driver_ctx.c
+++ b/src/driver/tpax_driver_ctx.c
@@ -6,6 +6,7 @@
 
 #define _DEFAULT_SOURCE 1
 
+#include <regex.h>
 #include <stdint.h>
 #include <stdlib.h>
 #include <string.h>
@@ -322,6 +323,109 @@ static void tpax_set_archive_block_size(struct tpax_common_ctx * cctx)
 		cctx->blksize = TPAX_USTAR_BLOCK_SIZE;
 }
 
+static int tpax_add_replstr(
+	struct argv_entry *             entry,
+	struct tpax_replstr *           replstr,
+	char **                         mark)
+{
+	const char *                    src;
+	char *                          dst;
+	char                            sep;
+	int                             nsep;
+
+	/* non-null separator character */
+	if (!(sep = entry->arg[0]))
+		return -1;
+
+	/* exactly three separator characters */
+	for (nsep=1,src=&entry->arg[1]; *src; src++) {
+		if ((src[0] == '\\') && (src[1] == sep)) {
+			src++;
+
+		} else if (src[0] == sep) {
+			nsep++;
+		}
+	}
+
+	if (nsep != 3)
+		return -1;
+
+	/* regexp */
+	for (src=&entry->arg[1],dst=*mark; (*src != sep); src++) {
+		if ((src[0] == '\\') && (src[1] == sep))
+			src++;
+
+		*dst++ = *src;
+	}
+
+	replstr->replarg = entry->arg;
+	replstr->replstr = ++dst;
+	replstr->regexp  = *mark;
+
+	/* replstr */
+	for (++src; (*src != sep); src++) {
+		if ((src[0] == '\\') && (src[1] == sep))
+			src++;
+
+		*dst++ = *src;
+	}
+
+	src++;
+	dst++;
+
+	*mark = dst;
+
+	/* flags */
+	if (src[0] && src[1] && src[2])
+		return -1;
+
+	if (src[0] && (src[0] == src[1]))
+		return -1;
+
+	if (src[0] && (src[0] != 'g') && (src[0] != 'p'))
+		return -1;
+
+	if (src[0] && src[1] && (src[1] != 'g') && (src[1] != 'p'))
+		return -1;
+
+	if (src[0] && ((src[0] == 'g') || (src[1] == 'g')))
+		replstr->flags |= TPAX_REPL_GLOBAL;
+
+	if (src[0] && ((src[0] == 'p') || (src[1] == 'p')))
+		replstr->flags |= TPAX_REPL_PRINT;
+
+	/* regex */
+	if (regcomp(&replstr->regex,replstr->regexp,0)) {
+		replstr->regexp = 0;
+		return -1;
+	}
+
+	return 0;
+}
+
+static int tpax_init_replstr_vector(
+	struct tpax_driver_ctx_impl *   ctx,
+	struct argv_meta *              meta)
+{
+	struct argv_entry *             entry;
+	struct tpax_replstr *           replstr;
+	char *                          mark;
+
+	if (!(replstr = ctx->replstrv))
+		return 0;
+
+	for (entry=meta->entries,mark=ctx->replstrs; entry->fopt || entry->arg; entry++) {
+		if (entry->tag == TAG_REPLSTR) {
+			if (tpax_add_replstr(entry,replstr,&mark) < 0)
+				return -1;
+
+			replstr++;
+		}
+	}
+
+	return 0;
+}
+
 static int tpax_driver_is_valid_keyval(struct argv_keyval * keyval)
 {
 	(void)keyval;
@@ -332,7 +436,9 @@ static struct tpax_driver_ctx_impl * tpax_driver_ctx_alloc(
 	struct argv_meta *		meta,
 	const struct tpax_fd_ctx *	fdctx,
 	const struct tpax_common_ctx *	cctx,
-	size_t				nunits)
+	size_t				nunits,
+	size_t                          nreplstr,
+	size_t                          sreplstr)
 {
 	struct tpax_driver_ctx_alloc *	ictx;
 	size_t				size;
@@ -379,6 +485,19 @@ static struct tpax_driver_ctx_impl * tpax_driver_ctx_alloc(
 		return 0;
 	}
 
+	if (nreplstr && !(ictx->ctx.replstrv = calloc(++nreplstr,sizeof(*ictx->ctx.replstrv)))) {
+		free(ictx->ctx.keyvalv);
+		free(ictx);
+		return 0;
+	}
+
+	if (sreplstr && !(ictx->ctx.replstrs = calloc(sreplstr,1))) {
+		free(ictx->ctx.replstrv);
+		free(ictx->ctx.keyvalv);
+		free(ictx);
+		return 0;
+	}
+
 	if ((pkeyval = ictx->ctx.keyvalv))
 		for (entry=meta->entries; entry->fopt || entry->arg; entry++)
 			if (entry->keyv)
@@ -474,6 +593,8 @@ int tpax_lib_get_driver_ctx(
 	struct argv_keyval **           pkeyval;
 	struct tpax_fd_ctx              lfdctx;
 	size_t				nunits;
+	size_t                          nreplstr;
+	size_t                          sreplstr;
 	const char *			program;
 	int				fddst;
 	const char *			ch;
@@ -497,6 +618,9 @@ int tpax_lib_get_driver_ctx(
 	program = argv_program_name(argv[0]);
 	memset(&cctx,0,sizeof(cctx));
 
+	nreplstr = 0;
+	sreplstr = 0;
+
 	cctx.drvflags = flags;
 	fddst         = fdctx->fddst;
 
@@ -581,6 +705,12 @@ int tpax_lib_get_driver_ctx(
 							meta);
 					break;
 
+				case TAG_REPLSTR:
+					sreplstr += strlen(entry->arg);
+					sreplstr++;
+					nreplstr++;
+					break;
+
 				case TAG_RECURSE:
 					cctx.drvflags |= TPAX_DRIVER_DIR_MEMBER_RECURSE;
 					break;
@@ -751,13 +881,19 @@ int tpax_lib_get_driver_ctx(
 	}
 
 	/* driver ctx */
-	if (!(ctx = tpax_driver_ctx_alloc(meta,fdctx,&cctx,nunits))) {
+	if (!(ctx = tpax_driver_ctx_alloc(meta,fdctx,&cctx,nunits,nreplstr,sreplstr))) {
 		if (cctx.drvflags & TPAX_DRIVER_EXEC_MODE_COPY)
 			close(fddst);
 
 		return tpax_get_driver_ctx_fail(meta);
 	}
 
+	/* replstr validation and vector initialization */
+	if (tpax_init_replstr_vector(ctx,meta) < 0) {
+		tpax_lib_free_driver_ctx(&ctx->ctx);
+		return TPAX_ERROR;
+	}
+
 	/* keyval validation */
 	for (pkeyval=ctx->keyvalv; pkeyval && *pkeyval; pkeyval++)
 		if (!tpax_driver_is_valid_keyval(*pkeyval))
@@ -781,6 +917,8 @@ static void tpax_free_driver_ctx_impl(struct tpax_driver_ctx_alloc * ictx)
 	size_t  size;
 	char ** ppref;
 
+	struct tpax_replstr * replstrv;
+
 	for (; ictx->ctx.dirents; ) {
 		next = ictx->ctx.dirents->next;
 		size = ictx->ctx.dirents->size;
@@ -789,6 +927,15 @@ static void tpax_free_driver_ctx_impl(struct tpax_driver_ctx_alloc * ictx)
 		ictx->ctx.dirents = (struct tpax_dirent_buffer *)next;
 	}
 
+	for (replstrv=ictx->ctx.replstrv; replstrv && replstrv->regexp; replstrv++)
+		regfree(&replstrv->regex);
+
+	if (ictx->ctx.replstrv)
+		free(ictx->ctx.replstrv);
+
+	if (ictx->ctx.replstrs)
+		free(ictx->ctx.replstrs);
+
 	if (ictx->ctx.keyvalv)
 		free(ictx->ctx.keyvalv);
 
diff --git a/src/internal/tpax_driver_impl.h b/src/internal/tpax_driver_impl.h
index d340748..3df8244 100644
--- a/src/internal/tpax_driver_impl.h
+++ b/src/internal/tpax_driver_impl.h
@@ -7,6 +7,7 @@
 #ifndef TPAX_DRIVER_IMPL_H
 #define TPAX_DRIVER_IMPL_H
 
+#include <regex.h>
 #include <stdint.h>
 #include <dirent.h>
 #include <stdio.h>
@@ -31,6 +32,9 @@
 #define TPAX_ITEM_SYMLINK  0X4
 #define TPAX_ITEM_NAMEREF  0x8
 
+#define TPAX_REPL_GLOBAL   0x01
+#define TPAX_REPL_PRINT    0x02
+
 extern const struct argv_option tpax_default_options[];
 
 enum app_tags {
@@ -45,6 +49,7 @@ enum app_tags {
 	TAG_FORMAT,
 	TAG_BLKSIZE,
 	TAG_OPTIONS,
+	TAG_REPLSTR,
 	TAG_RECURSE,
 	TAG_NORECURSE,
 	TAG_STRICT_PATH,
@@ -74,6 +79,14 @@ struct tpax_dirent_buffer {
 	struct tpax_dirent            dbuf[];
 };
 
+struct tpax_replstr {
+	const char *                  replarg;
+	const char *                  replstr;
+	const char *                  regexp;
+	regex_t                       regex;
+	uint32_t                      flags;
+};
+
 struct tpax_driver_ctx_impl {
 	const char *                  file;
 	struct tpax_common_ctx        cctx;
@@ -82,6 +95,8 @@ struct tpax_driver_ctx_impl {
 	const struct tpax_unit_ctx *  euctx;
 	const char *                  eunit;
 	struct argv_keyval **         keyvalv;
+	struct tpax_replstr *         replstrv;
+	char *                        replstrs;
 	struct tpax_error_info **     errinfp;
 	struct tpax_error_info **     erricap;
 	struct tpax_error_info *      erriptr[64];
diff --git a/src/logic/tpax_archive_write.c b/src/logic/tpax_archive_write.c
index 32a5f2c..96af625 100644
--- a/src/logic/tpax_archive_write.c
+++ b/src/logic/tpax_archive_write.c
@@ -75,6 +75,35 @@ static int tpax_archive_write_ret(
 	return ret;
 }
 
+static int tpax_apply_string_replacement(
+	const struct tpax_driver_ctx *  dctx,
+	const char *                    path,
+	char *                          replbuf,
+	size_t                          buflen)
+{
+	int                            ret;
+	struct tpax_driver_ctx_impl *  ictx;
+	struct tpax_replstr *          replstrv;
+
+	ictx = tpax_get_driver_ictx(dctx);
+
+	if (!(replstrv = ictx->replstrv))
+		return 0;
+
+	for (ret=0; !ret && replstrv->regexp; replstrv++) {
+		ret = tpax_util_path_replstr(
+			replbuf,path,
+			replstrv->replstr,
+			&replstrv->regex,
+			buflen,replstrv->flags);
+
+		if ((ret > 0) && (replstrv->flags & TPAX_REPL_PRINT))
+			tpax_dprintf(tpax_driver_fderr(dctx),"%s >> %s\n",path,replbuf);
+	}
+
+	return ret;
+}
+
 static int tpax_archive_write_impl(
 	const struct tpax_driver_ctx *  dctx,
 	const struct tpax_dirent *      cdent,
@@ -85,12 +114,14 @@ static int tpax_archive_write_impl(
 	struct tpax_ustar_header        uhdr;
 	const struct stat *             st;
 	struct stat                     stbuf;
+	const  char *                   apath;
 	const  char *                   path;
 	const  char *                   slnk;
 	const  char *                   mlnk;
 	off_t                           hpos;
 	off_t                           dpos;
 	int                             fdtmp;
+	int                             slen;
 	ssize_t                         nread;
 	ssize_t                         nbytes;
 	void *                          buf;
@@ -98,6 +129,7 @@ static int tpax_archive_write_impl(
 	size_t                          cmplen;
 	void *				membuf;
 	char *                          ch;
+	char                           	replbuf[PATH_MAX];
 	char                           	pathbuf[PATH_MAX];
 
 	/* followed symlink? */
@@ -110,9 +142,17 @@ static int tpax_archive_write_impl(
 			dctx,
 			TPAX_ERR_FLOW_ERROR);
 
+	/* regex matching and patter substitution */
+	if ((slen = tpax_apply_string_replacement(dctx,path,replbuf,buflen)) < 0)
+		return TPAX_CUSTOM_ERROR(
+			dctx,
+			TPAX_ERR_FLOW_ERROR);
+
+	apath = slen ? replbuf : path;
+
 	/* verbose mode */
 	if (dctx->cctx->drvflags & TPAX_DRIVER_VERBOSE)
-		tpax_dprintf(tpax_driver_fderr(dctx),"%s",path);
+		tpax_dprintf(tpax_driver_fderr(dctx),"%s",apath);
 
 	/* uctx */
 	if (tpax_lib_get_unit_ctx(dctx,fdcwd,path,&uctx) < 0)
@@ -171,7 +211,7 @@ static int tpax_archive_write_impl(
 
 	/* header */
 	if (tpax_meta_init_ustar_header(
-			dctx,path,st,
+			dctx,apath,st,
 			slnk,&uhdr) < 0)
 		return tpax_archive_write_ret(
 			TPAX_NESTED_ERROR(dctx),
diff --git a/src/skin/tpax_skin_default.c b/src/skin/tpax_skin_default.c
index a618645..dd62b86 100644
--- a/src/skin/tpax_skin_default.c
+++ b/src/skin/tpax_skin_default.c
@@ -93,6 +93,21 @@ const tpax_hidden struct argv_option tpax_default_options[] = {
 			"a user-provided, format-specific keyval array of the form "
 			"keyword[[:]=value][,keyword[[:]=value], ...]"},
 
+	{"Wreplstr",    's',TAG_REPLSTR,ARGV_OPTARG_REQUIRED,
+			ARGV_OPTION_HYBRID_ONLY|ARGV_OPTION_HYBRID_SPACE,0,0,
+			"rename files and archive members as they are being added to "
+			"or extracted from the archive according to the specified "
+			"ed(1) style replacement string, which should be in the format "
+			"<sep><regex><sep><replstr><sep>[gp]; as an example, "
+			"-s ',^/git/tpax/,tpax-1.2.3/,' uses <comma> as the separator "
+			"character, and instructs pax to prefix all files rooted in "
+			"'/git/tpax/' with 'tpax-1.2.3/' while leaving the names of files which "
+			"do not match the regex expression unchanged. "
+			"When this option is repeated, pax shall attempt to match each file or "
+			"member name against all of the provided repalcement-string arguments "
+			"in the order of appearnce on the command line until the first "
+			"successful match."},
+
 	{"Wstrict-device-id",
 			'X',TAG_STRICT_DEVICE_ID,ARGV_OPTARG_NONE,
 			ARGV_OPTION_HYBRID_ONLY,0,0,
diff --git a/src/util/tpax_path_replstr.c b/src/util/tpax_path_replstr.c
new file mode 100644
index 0000000..1935628
--- /dev/null
+++ b/src/util/tpax_path_replstr.c
@@ -0,0 +1,120 @@
+/**************************************************************/
+/*  tpax: a topological pax implementation                    */
+/*  Copyright (C) 2020--2024  SysDeer Technologies, LLC       */
+/*  Released under GPLv2 and GPLv3; see COPYING.TPAX.         */
+/**************************************************************/
+
+#include <regex.h>
+#include <stdlib.h>
+#include <string.h>
+#include <errno.h>
+
+#include <tpax/tpax.h>
+#include "tpax_driver_impl.h"
+
+static int tpax_backref_idx(const char c)
+{
+	return ((c >= '1') && (c <= '9')) ? c - '0' : 0;
+}
+
+int tpax_util_path_replstr(
+	char *          dstpath,
+	const char *    srcpath,
+	const char *    replstr,
+	const regex_t * regex,
+	size_t          buflen,
+	int             flags)
+{
+	int             ret;
+	int             idx;
+	regoff_t        ro;
+	const char *    ch;
+	char *          dst;
+	size_t          explen;
+	regmatch_t      pmatch[11];
+
+	/* attempt to match */
+	switch (regexec(regex,srcpath,11,pmatch,0)) {
+		case 0:
+			break;
+
+		case REG_NOMATCH:
+			return 0;
+
+		default:
+			return -1;
+	}
+
+	/* copy bytes leading up to match */
+	if (buflen <= (explen = pmatch[0].rm_so)) {
+		errno = ENOBUFS;
+		return -1;
+	}
+
+	for (ro=0,dst=dstpath; ro<pmatch[0].rm_so; ro++)
+		*dst++ = srcpath[ro];
+
+	buflen -= explen;
+
+	/* copy replacement string */
+	for (ch=replstr,ret=0; buflen && *ch; ch++) {
+		/* <ampersand> stands for the entire matched string */
+		if (ch[0] == '&') {
+			idx = 0;
+
+		/* back-reference semantics: a matched subexpression or an empty string */
+		} else if ((ch[0] == '\\') && (idx = tpax_backref_idx(ch[1]))) {
+			if (pmatch[idx].rm_so < 0)
+				idx = -1;
+
+			ch++;
+
+		/* all other escaped characters */
+		} else if (ch[0] == '\\') {
+			*dst++ = *++ch;
+			idx    = -1;
+			buflen--;
+
+		/* all other characters */
+		} else {
+			*dst++ = *ch;
+			idx    = -1;
+			buflen--;
+		}
+
+		/* copy matched string or matched subexpression, if any */
+		if (idx >= 0) {
+			if (buflen <= (explen = (pmatch[idx].rm_eo - pmatch[idx].rm_so))) {
+				errno = ENOBUFS;
+				return -1;
+			}
+
+			for (ro=pmatch[idx].rm_so; ro<pmatch[idx].rm_eo; ro++)
+				*dst++ = srcpath[ro];
+
+			buflen -= explen;
+		}
+	}
+
+	/* replace further occurrences as needed */
+	if ((flags & TPAX_REPL_GLOBAL) && srcpath[pmatch[0].rm_eo])
+		ret = tpax_util_path_replstr(
+			dst,&srcpath[pmatch[0].rm_eo],replstr,
+			regex,buflen,flags);
+
+	if (ret < 0)
+		return -1;
+
+	/* copy remaining, non-matching bytes as needed */
+	if (ret == 0) {
+		for (ch=&srcpath[pmatch[0].rm_eo]; *ch; ch++)
+			*dst++ = *ch;
+
+		*dst = '\0';
+	}
+
+	/* all done */
+	ret += (dst - dstpath);
+
+	return ret;
+}