Move istream_get_line function to libutil

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

4 files changed, 316 insertions, 4 deletions

diff --git a/lib/util/Makemodule.am b/lib/util/Makemodule.am
index a797930..1ca4802 100644
--- a/lib/util/Makemodule.am
+++ b/lib/util/Makemodule.am
@@ -1,6 +1,6 @@
 libutil_a_SOURCES = include/util/util.h include/util/str_table.h \
 	include/util/hash_table.h include/util/test.h include/util/rbtree.h \
-	include/util/array.h include/util/threadpool.h \
+	include/util/array.h include/util/threadpool.h include/util/parse.h \
 	include/util/w32threadwrap.h include/util/mempool.h \
 	lib/util/src/str_table.c lib/util/src/alloc.c lib/util/src/rbtree.c \
 	lib/util/src/array.c lib/util/src/xxhash.c lib/util/src/hash_table.c \
@@ -8,7 +8,8 @@ libutil_a_SOURCES = include/util/util.h include/util/str_table.h \
 	lib/util/src/is_memory_zero.c lib/util/src/mkdir_p.c \
 	lib/util/src/canonicalize_name.c lib/util/src/filename_sane.c \
 	lib/util/src/source_date_epoch.c lib/util/src/file_cmp.c \
-	lib/util/src/hex_decode.c lib/util/src/base64_decode.c
+	lib/util/src/hex_decode.c lib/util/src/base64_decode.c \
+	lib/util/src/get_line.c
 libutil_a_CFLAGS = $(AM_CFLAGS)
 libutil_a_CPPFLAGS = $(AM_CPPFLAGS)
 
@@ -76,10 +77,13 @@ test_hex_decode_LDADD = libutil.a libcompat.a
 test_base64_decode_SOURCES = lib/util/test/base64_decode.c
 test_base64_decode_LDADD = libutil.a libcompat.a
 
+test_get_line_SOURCES = lib/util/test/get_line.c
+test_get_line_LDADD = libutil.a libio.a libcompat.a
+
 LIBUTIL_TESTS = \
 	test_str_table test_rbtree test_xxhash test_threadpool test_ismemzero \
 	test_canonicalize_name test_filename_sane test_filename_sane_w32 \
-	test_sdate_epoch test_hex_decode test_base64_decode
+	test_sdate_epoch test_hex_decode test_base64_decode test_get_line
 
 check_PROGRAMS += $(LIBUTIL_TESTS)
 TESTS += $(LIBUTIL_TESTS)
diff --git a/lib/util/src/get_line.c b/lib/util/src/get_line.c
new file mode 100644
index 0000000..5a17b62
--- /dev/null
+++ b/lib/util/src/get_line.c
@@ -0,0 +1,130 @@
+/* SPDX-License-Identifier: LGPL-3.0-or-later */
+/*
+ * get_line.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "util/parse.h"
+#include "sqfs/io.h"
+#include "sqfs/error.h"
+#include "compat.h"
+
+#include <stdlib.h>
+#include <string.h>
+#include <ctype.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(sqfs_istream_t *strm, char **out,
+		     size_t *line_num, int flags)
+{
+	char *line = NULL, *new;
+	size_t line_len = 0;
+	int err;
+
+	for (;;) {
+		bool have_line = false;
+		size_t i, count, avail;
+		const sqfs_u8 *ptr;
+
+		err = strm->get_buffered_data(strm, &ptr, &avail, 0);
+		if (err < 0)
+			goto fail;
+		if (err > 0) {
+			if (line_len == 0)
+				goto out_eof;
+
+			line_len = trim(line, flags);
+			if (line_len > 0 ||!(flags & ISTREAM_LINE_SKIP_EMPTY))
+				break;
+
+			goto out_eof;
+		}
+
+		for (i = 0; i < avail; ++i) {
+			if (ptr[i] == '\n')
+				break;
+		}
+
+		if (i < avail) {
+			count = i++;
+			have_line = true;
+		} else {
+			count = i;
+		}
+
+		new = realloc(line, line_len + count + 1);
+		if (new == NULL) {
+			err = SQFS_ERROR_ALLOC;
+			goto fail;
+		}
+
+		line = new;
+		memcpy(line + line_len, ptr, count);
+		line_len += count;
+		line[line_len] = '\0';
+
+		strm->advance_buffer(strm, i);
+
+		if (have_line) {
+			if (line_len > 0 && line[line_len - 1] == '\r')
+				line[--line_len] = '\0';
+
+			line_len = trim(line, flags);
+			if (line_len > 0 || !(flags & ISTREAM_LINE_SKIP_EMPTY))
+				break;
+
+			free(line);
+			line = NULL;
+			*line_num += 1;
+		}
+	}
+
+	new = realloc(line, line_len + 1);
+	if (new != NULL)
+		line = new;
+
+	*out = line;
+	return 0;
+fail: {
+	os_error_t temp = get_os_error_state();
+	free(line);
+	*out = NULL;
+	set_os_error_state(temp);
+	return err;
+}
+out_eof:
+	free(line);
+	*out = NULL;
+	return 1;
+}
diff --git a/lib/util/test/get_line.c b/lib/util/test/get_line.c
new file mode 100644
index 0000000..6189d31
--- /dev/null
+++ b/lib/util/test/get_line.c
@@ -0,0 +1,178 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * get_line.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "util/parse.h"
+#include "io/mem.h"
+#include "util/test.h"
+
+typedef struct {
+	size_t line_num;
+	const char *str;
+} line_t;
+
+static void run_test_case(const char *raw, const line_t *lines, size_t count,
+			  int flags)
+{
+	size_t i, line_num, old_line_num;
+	sqfs_istream_t *fp;
+	char *line;
+	int ret;
+
+	fp = istream_memory_create("lines.txt", 2, raw, strlen(raw));
+	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_drop(fp);
+}
+
+static const char *file =
+"\r\n"
+"The quick\r\n"
+"  \r\n"
+"  brown fox  \r\n"
+"\r\n"
+"jumps over\r\n"
+"the\r\n"
+"lazy\r\n"
+"\r\n"
+"dog\r\n"
+"\r\n";
+
+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(int argc, char **argv)
+{
+	(void)argc; (void)argv;
+
+	run_test_case(file, lines_raw, 11, 0);
+	run_test_case(file, lines_ltrim, 11, ISTREAM_LINE_LTRIM);
+	run_test_case(file, lines_rtrim, 11, ISTREAM_LINE_RTRIM);
+	run_test_case(file, lines_trim, 11,
+		      ISTREAM_LINE_LTRIM | ISTREAM_LINE_RTRIM);
+
+	run_test_case(file, lines_no_empty, 7, ISTREAM_LINE_SKIP_EMPTY);
+	run_test_case(file, lines_no_empty_ltrim, 6,
+		      ISTREAM_LINE_SKIP_EMPTY | ISTREAM_LINE_LTRIM);
+	run_test_case(file, lines_no_empty_rtrim, 6,
+		      ISTREAM_LINE_SKIP_EMPTY | ISTREAM_LINE_RTRIM);
+	run_test_case(file, lines_no_empty_trim, 6,
+		      ISTREAM_LINE_SKIP_EMPTY | ISTREAM_LINE_LTRIM |
+		      ISTREAM_LINE_RTRIM);
+
+	return EXIT_SUCCESS;
+}
diff --git a/lib/util/test/str_table.c b/lib/util/test/str_table.c
index b146773..153d09e 100644
--- a/lib/util/test/str_table.c
+++ b/lib/util/test/str_table.c
@@ -7,8 +7,8 @@
 #include "config.h"
 
 #include "util/str_table.h"
+#include "util/parse.h"
 #include "compat.h"
-#include "io/istream.h"
 #include "util/test.h"
 #include "sqfs/io.h"