Add a more usefull getline-like function to libfstream

Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>

9 files changed, 343 insertions, 7 deletions

diff --git a/include/fstream.h b/include/fstream.h
index 8693fff..07ee5a7 100644
--- a/include/fstream.h
+++ b/include/fstream.h
@@ -56,6 +56,12 @@ enum {
 };
 
 enum {
+	ISTREAM_LINE_LTRIM = 0x01,
+	ISTREAM_LINE_RTRIM = 0x02,
+	ISTREAM_LINE_SKIP_EMPTY = 0x04,
+};
+
+enum {
 	/**
 	 * @brief Deflate compressor with gzip headers.
 	 *
@@ -258,6 +264,35 @@ SQFS_INTERNAL const char *ostream_get_filename(ostream_t *strm);
 SQFS_INTERNAL int ostream_printf(ostream_t *strm, const char *fmt, ...);
 
 /**
+ * @brief Read a line of text from an input stream
+ *
+ * @memberof istream_t
+ *
+ * The line returned is allocated using malloc and must subsequently be
+ * freed when it is no longer needed. The line itself is always null-terminated
+ * and never includes the line break characters (LF or CR-LF).
+ *
+ * If the flag @ref ISTREAM_LINE_LTRIM is set, leading white space characters
+ * are removed. If the flag @ref ISTREAM_LINE_RTRIM is set, trailing white space
+ * characters are remvoed.
+ *
+ * If the flag @ref ISTREAM_LINE_SKIP_EMPTY is set and a line is discovered to
+ * be empty (after the optional trimming), the function discards the empty line
+ * and retries. The given line_num pointer is used to increment the line
+ * number.
+ *
+ * @param strm A pointer to an input stream.
+ * @param out Returns a pointer to a line on success.
+ * @param line_num This is incremented if lines are skipped.
+ * @param flags A combination of flags controling the functions behaviour.
+ *
+ * @return Zero on success, a negative value on error, a positive value if
+ *         end-of-file was reached without reading any data.
+ */
+SQFS_INTERNAL int istream_get_line(istream_t *strm, char **out,
+				   size_t *line_num, int flags);
+
+/**
  * @brief Read data from an input stream
  *
  * @memberof istream_t
diff --git a/lib/fstream/Makemodule.am b/lib/fstream/Makemodule.am
index 0c24c6f..ad5f426 100644
--- a/lib/fstream/Makemodule.am
+++ b/lib/fstream/Makemodule.am
@@ -1,7 +1,7 @@
 libfstream_a_SOURCES = include/fstream.h
 libfstream_a_SOURCES += lib/fstream/internal.h
 libfstream_a_SOURCES += lib/fstream/ostream.c lib/fstream/printf.c
-libfstream_a_SOURCES += lib/fstream/istream.c
+libfstream_a_SOURCES += lib/fstream/istream.c lib/fstream/get_line.c
 libfstream_a_SOURCES += lib/fstream/compressor.c
 libfstream_a_SOURCES += lib/fstream/compress/ostream_compressor.c
 libfstream_a_SOURCES += lib/fstream/uncompress/istream_compressor.c
diff --git a/lib/fstream/get_line.c b/lib/fstream/get_line.c
new file mode 100644
index 0000000..f7e0b59
--- /dev/null
+++ b/lib/fstream/get_line.c
@@ -0,0 +1,118 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * get_line.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "internal.h"
+
+static void ltrim(char *buffer)
+{
+	size_t i = 0;
+
+	while (isspace(buffer[i]))
+		++i;
+
+	if (i > 0)
+		memmove(buffer, buffer + i, strlen(buffer + i) + 1);
+}
+
+static void rtrim(char *buffer)
+{
+	size_t i = strlen(buffer);
+
+	while (i > 0 && isspace(buffer[i - 1]))
+		--i;
+
+	buffer[i] = '\0';
+}
+
+static size_t trim(char *buffer, int flags)
+{
+	if (flags & ISTREAM_LINE_LTRIM)
+		ltrim(buffer);
+
+	if (flags & ISTREAM_LINE_RTRIM)
+		rtrim(buffer);
+
+	return strlen(buffer);
+}
+
+int istream_get_line(istream_t *strm, char **out,
+		     size_t *line_num, int flags)
+{
+	char *line = NULL, *new;
+	size_t i, line_len = 0;
+	bool have_line = false;
+
+	*out = NULL;
+
+	for (;;) {
+		if (istream_precache(strm))
+			return -1;
+
+		if (strm->buffer_used == 0) {
+			if (line_len == 0)
+				goto out_eof;
+
+			line_len = trim(line, flags);
+
+			if (line_len == 0 &&
+			    (flags & ISTREAM_LINE_SKIP_EMPTY)) {
+				goto out_eof;
+			}
+			break;
+		}
+
+		for (i = 0; i < strm->buffer_used; ++i) {
+			if (strm->buffer[i] == '\n')
+				break;
+		}
+
+		if (i < strm->buffer_used) {
+			have_line = true;
+			strm->buffer_offset = i + 1;
+
+			if (i > 0 && strm->buffer[i - 1] == '\r')
+				--i;
+		} else {
+			strm->buffer_offset = i;
+		}
+
+		new = realloc(line, line_len + i + 1);
+		if (new == NULL)
+			goto fail_errno;
+
+		line = new;
+		memcpy(line + line_len, strm->buffer, i);
+		line_len += i;
+		line[line_len] = '\0';
+
+		if (have_line) {
+			line_len = trim(line, flags);
+
+			if (line_len == 0 &&
+			    (flags & ISTREAM_LINE_SKIP_EMPTY)) {
+				free(line);
+				line = NULL;
+				have_line = false;
+				*line_num += 1;
+				continue;
+			}
+			break;
+		}
+	}
+
+	*out = line;
+	return 0;
+fail_errno:
+	fprintf(stderr, "%s: " PRI_SZ ": %s.\n", strm->get_filename(strm),
+		*line_num, strerror(errno));
+	free(line);
+	*out = NULL;
+	return -1;
+out_eof:
+	free(line);
+	*out = NULL;
+	return 1;
+}
diff --git a/lib/fstream/internal.h b/lib/fstream/internal.h
index 2dc81e4..4f02f8c 100644
--- a/lib/fstream/internal.h
+++ b/lib/fstream/internal.h
@@ -16,6 +16,7 @@
 #include <stdarg.h>
 #include <stdlib.h>
 #include <unistd.h>
+#include <ctype.h>
 #include <fcntl.h>
 #include <errno.h>
 #include <stdio.h>
diff --git a/tests/Makemodule.am b/tests/Makemodule.am
index 94bfd4b..58cd27e 100644
--- a/tests/Makemodule.am
+++ b/tests/Makemodule.am
@@ -230,6 +230,11 @@ fstree_fuzz_LDADD = libfstree.a libcompat.a
 tar_fuzz_SOURCES = tests/tar_fuzz.c
 tar_fuzz_LDADD = libtar.a libfstream.a libcompat.a
 
+test_get_line_SOURCES = tests/get_line.c tests/test.h
+test_get_line_LDADD = libfstream.a libcompat.a
+test_get_line_CPPFLAGS = $(AM_CPPFLAGS)
+test_get_line_CPPFLAGS += -DTESTFILE=$(top_srcdir)/tests/get_line.txt
+
 check_PROGRAMS += test_mknode_simple test_mknode_slink test_mknode_reg
 check_PROGRAMS += test_mknode_dir test_gen_inode_numbers test_add_by_path
 check_PROGRAMS += test_get_path test_fstree_sort test_fstree_from_file
@@ -245,6 +250,7 @@ check_PROGRAMS += test_tar_sparse_gnu test_tar_sparse_gnu0 test_tar_sparse_gnu1
 check_PROGRAMS += test_tar_sparse_gnu2 test_tar_sparse_gnu3
 check_PROGRAMS += test_tar_xattr_bsd test_tar_xattr_schily
 check_PROGRAMS += test_tar_xattr_schily_bin test_tar_target_filled
+check_PROGRAMS += test_get_line
 
 noinst_PROGRAMS += fstree_fuzz tar_fuzz
 
@@ -261,7 +267,7 @@ TESTS += test_tar_pax5
 TESTS += test_tar_sparse_gnu test_tar_sparse_gnu0
 TESTS += test_tar_sparse_gnu1 test_tar_sparse_gnu2 test_tar_sparse_gnu3
 TESTS += test_tar_xattr_bsd test_tar_xattr_schily
-TESTS += test_tar_xattr_schily_bin test_tar_target_filled
+TESTS += test_tar_xattr_schily_bin test_tar_target_filled test_get_line
 
 if CORPORA_TESTS
 check_SCRIPTS += tests/cantrbry.sh tests/test_tar_sqfs.sh tests/pack_dir_root.sh
@@ -274,3 +280,4 @@ EXTRA_DIST += $(top_srcdir)/tests/fstree1.txt
 EXTRA_DIST += $(top_srcdir)/tests/corpus/cantrbry.tar.xz
 EXTRA_DIST += $(top_srcdir)/tests/corpus/cantrbry.sha512
 EXTRA_DIST += $(top_srcdir)/tests/pack_dir_root.txt.ref
+EXTRA_DIST += $(top_srcdir)/tests/get_line.txt
diff --git a/tests/get_line.c b/tests/get_line.c
new file mode 100644
index 0000000..c317c0e
--- /dev/null
+++ b/tests/get_line.c
@@ -0,0 +1,164 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * get_line.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+
+#include "fstream.h"
+#include "test.h"
+
+typedef struct {
+	size_t line_num;
+	const char *str;
+} line_t;
+
+static void run_test_case(const line_t *lines, size_t count,
+			  int flags)
+{
+	size_t i, line_num, old_line_num;
+	istream_t *fp;
+	char *line;
+	int ret;
+
+	fp = istream_open_file(STRVALUE(TESTFILE));
+	TEST_NOT_NULL(fp);
+
+	line_num = 1;
+	line = NULL;
+
+	for (i = 0; i < count; ++i) {
+		old_line_num = line_num;
+		ret = istream_get_line(fp, &line, &line_num, flags);
+
+		TEST_ASSERT(line_num >= old_line_num);
+		TEST_EQUAL_I(ret, 0);
+		TEST_NOT_NULL(line);
+
+		TEST_EQUAL_UI(line_num, lines[i].line_num);
+		TEST_STR_EQUAL(line, lines[i].str);
+
+		free(line);
+		line = NULL;
+		line_num += 1;
+	}
+
+	ret = istream_get_line(fp, &line, &line_num, flags);
+	TEST_ASSERT(ret > 0);
+
+	sqfs_destroy(fp);
+}
+
+static const line_t lines_raw[] = {
+	{ 1, "" },
+	{ 2, "The quick" },
+	{ 3, "  " },
+	{ 4, "  brown fox  " },
+	{ 5, "" },
+	{ 6, "jumps over" },
+	{ 7, "the" },
+	{ 8, "lazy" },
+	{ 9, "" },
+	{ 10, "dog" },
+	{ 11, "" },
+};
+
+static const line_t lines_ltrim[] = {
+	{ 1, "" },
+	{ 2, "The quick" },
+	{ 3, "" },
+	{ 4, "brown fox  " },
+	{ 5, "" },
+	{ 6, "jumps over" },
+	{ 7, "the" },
+	{ 8, "lazy" },
+	{ 9, "" },
+	{ 10, "dog" },
+	{ 11, "" },
+};
+
+static const line_t lines_rtrim[] = {
+	{ 1, "" },
+	{ 2, "The quick" },
+	{ 3, "" },
+	{ 4, "  brown fox" },
+	{ 5, "" },
+	{ 6, "jumps over" },
+	{ 7, "the" },
+	{ 8, "lazy" },
+	{ 9, "" },
+	{ 10, "dog" },
+	{ 11, "" },
+};
+
+static const line_t lines_trim[] = {
+	{ 1, "" },
+	{ 2, "The quick" },
+	{ 3, "" },
+	{ 4, "brown fox" },
+	{ 5, "" },
+	{ 6, "jumps over" },
+	{ 7, "the" },
+	{ 8, "lazy" },
+	{ 9, "" },
+	{ 10, "dog" },
+	{ 11, "" },
+};
+
+static const line_t lines_no_empty[] = {
+	{ 2, "The quick" },
+	{ 3, "  " },
+	{ 4, "  brown fox  " },
+	{ 6, "jumps over" },
+	{ 7, "the" },
+	{ 8, "lazy" },
+	{ 10, "dog" },
+};
+
+static const line_t lines_no_empty_ltrim[] = {
+	{ 2, "The quick" },
+	{ 4, "brown fox  " },
+	{ 6, "jumps over" },
+	{ 7, "the" },
+	{ 8, "lazy" },
+	{ 10, "dog" },
+};
+
+static const line_t lines_no_empty_rtrim[] = {
+	{ 2, "The quick" },
+	{ 4, "  brown fox" },
+	{ 6, "jumps over" },
+	{ 7, "the" },
+	{ 8, "lazy" },
+	{ 10, "dog" },
+};
+
+static const line_t lines_no_empty_trim[] = {
+	{ 2, "The quick" },
+	{ 4, "brown fox" },
+	{ 6, "jumps over" },
+	{ 7, "the" },
+	{ 8, "lazy" },
+	{ 10, "dog" },
+};
+
+int main(void)
+{
+	run_test_case(lines_raw, 11, 0);
+	run_test_case(lines_ltrim, 11, ISTREAM_LINE_LTRIM);
+	run_test_case(lines_rtrim, 11, ISTREAM_LINE_RTRIM);
+	run_test_case(lines_trim, 11,
+		      ISTREAM_LINE_LTRIM | ISTREAM_LINE_RTRIM);
+
+	run_test_case(lines_no_empty, 7, ISTREAM_LINE_SKIP_EMPTY);
+	run_test_case(lines_no_empty_ltrim, 6,
+		      ISTREAM_LINE_SKIP_EMPTY | ISTREAM_LINE_LTRIM);
+	run_test_case(lines_no_empty_rtrim, 6,
+		      ISTREAM_LINE_SKIP_EMPTY | ISTREAM_LINE_RTRIM);
+	run_test_case(lines_no_empty_trim, 6,
+		      ISTREAM_LINE_SKIP_EMPTY | ISTREAM_LINE_LTRIM |
+		      ISTREAM_LINE_RTRIM);
+
+	return EXIT_SUCCESS;
+}
diff --git a/tests/get_line.txt b/tests/get_line.txt
new file mode 100644
index 0000000..a1994f0
--- /dev/null
+++ b/tests/get_line.txt
@@ -0,0 +1,11 @@
+

+The quick

+  

+  brown fox  

+

+jumps over

+the

+lazy

+

+dog

+

diff --git a/tests/test.h b/tests/test.h
index 7d38fd0..bc3ac92 100644
--- a/tests/test.h
+++ b/tests/test.h
@@ -13,6 +13,11 @@
 #include <stdio.h>
 #include <errno.h>
 
+#define STR(x) #x
+#define STRVALUE(x) STR(x)
+
+#define TEST_PATH STRVALUE(TESTPATH)
+
 #if defined(__GNUC__) || defined(__clang__)
 #	define ATTRIB_UNUSED __attribute__ ((unused))
 #else
diff --git a/tests/test_tar.h b/tests/test_tar.h
index 9ec2b12..0d5ccc1 100644
--- a/tests/test_tar.h
+++ b/tests/test_tar.h
@@ -11,9 +11,4 @@
 #include "tar.h"
 #include "test.h"
 
-#define STR(x) #x
-#define STRVALUE(x) STR(x)
-
-#define TEST_PATH STRVALUE(TESTPATH)
-
 #endif /* TEST_TAR_H */