X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=serializer.cpp;h=f20c304de54e690a5c41852127da93b6b8c3c2bb;hb=cbecd483f57c465b2ad6d3867c760c2e5b5e79aa;hp=10a156411b5e551cec7e51be66907fef545a80c0;hpb=93d57f8f19e57efbb91e139bfca1064bd9e27bb3;p=plocate

diff --git a/serializer.cpp b/serializer.cpp
index 10a1564..f20c304 100644
--- a/serializer.cpp
+++ b/serializer.cpp
@@ -27,6 +27,121 @@ void apply_limit()
 	exit(0);
 }
 
+void print_possibly_escaped(const string &str)
+{
+	if (print_nul) {
+		printf("%s%c", str.c_str(), 0);
+		return;
+	} else if (!stdout_is_tty) {
+		printf("%s\n", str.c_str());
+		return;
+	}
+
+	// stdout is a terminal, so we should protect the user against
+	// escapes, stray newlines and the likes. First of all, check if
+	// all the characters are safe; we consider everything safe that
+	// isn't a control character, ', " or \. People could make
+	// filenames like "$(rm -rf)", but that's out-of-scope.
+	const char *ptr = str.data();
+	size_t len = str.size();
+
+	mbtowc(nullptr, 0, 0);
+	wchar_t pwc;
+	bool all_safe = true;
+	do {
+		int ret = mbtowc(&pwc, ptr, len);
+		if (ret == -1) {
+			all_safe = false;  // Malformed data.
+		} else if (ret == 0) {
+			break;  // EOF.
+		} else if (pwc < 32 || pwc == '\'' || pwc == '"' || pwc == '\\') {
+			all_safe = false;
+		} else {
+			ptr += ret;
+			len -= ret;
+		}
+	} while (all_safe);
+
+	if (all_safe) {
+		printf("%s\n", str.c_str());
+		return;
+	}
+
+	// Print escaped, but in such a way that the user can easily take the
+	// escaped output and paste into the shell. We print much like GNU ls does,
+	// ie., using the shell $'foo' construct whenever we need to print something
+	// escaped.
+	bool in_escaped_mode = false;
+	printf("'");
+
+	mbtowc(nullptr, 0, 0);
+	ptr = str.data();
+	len = str.size();
+	for (;;) {
+		int ret = mbtowc(nullptr, ptr, len);
+		if (ret == -1) {
+			// Malformed data.
+			printf("?");
+			++ptr;
+			--len;
+		} else if (ret == 0) {
+			break;  // EOF.
+		}
+		if (*ptr < 32 || *ptr == '\'' || *ptr == '"' || *ptr == '\\') {
+			if (!in_escaped_mode) {
+				printf("'$'");
+				in_escaped_mode = true;
+			}
+
+			// The list of allowed escapes is from bash(1).
+			switch (*ptr) {
+			case '\a':
+				printf("\\a");
+				break;
+			case '\b':
+				printf("\\b");
+				break;
+			case '\f':
+				printf("\\f");
+				break;
+			case '\n':
+				printf("\\n");
+				break;
+			case '\r':
+				printf("\\r");
+				break;
+			case '\t':
+				printf("\\t");
+				break;
+			case '\v':
+				printf("\\v");
+				break;
+			case '\\':
+				printf("\\\\");
+				break;
+			case '\'':
+				printf("\\'");
+				break;
+			case '"':
+				printf("\\\"");
+				break;
+			default:
+				printf("\\%03o", *ptr);
+				break;
+			}
+		} else {
+			if (in_escaped_mode) {
+				printf("''");
+				in_escaped_mode = false;
+			}
+			fwrite(ptr, ret, 1, stdout);
+		}
+		ptr += ret;
+		len -= ret;
+	}
+	printf("'\n");
+}
+
 void Serializer::print(uint64_t seq, uint64_t skip, const string msg)
 {
 	if (only_count) {
@@ -42,11 +157,7 @@ void Serializer::print(uint64_t seq, uint64_t skip, const string msg)
 	}
 
 	if (!msg.empty()) {
-		if (print_nul) {
-			printf("%s%c", msg.c_str(), 0);
-		} else {
-			printf("%s\n", msg.c_str());
-		}
+		print_possibly_escaped(msg);
 		apply_limit();
 	}
 	next_seq += skip;
@@ -54,11 +165,7 @@ void Serializer::print(uint64_t seq, uint64_t skip, const string msg)
 	// See if any delayed prints can now be dealt with.
 	while (!pending.empty() && pending.top().seq == next_seq) {
 		if (!pending.top().msg.empty()) {
-			if (print_nul) {
-				printf("%s%c", pending.top().msg.c_str(), 0);
-			} else {
-				printf("%s\n", pending.top().msg.c_str());
-			}
+			print_possibly_escaped(pending.top().msg);
 			apply_limit();
 		}
 		next_seq += pending.top().skip;