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

1 files changed, 120 insertions, 0 deletions

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;
+}