Hide fopen wrapper and use reader interface consistently

9 files changed, 115 insertions, 145 deletions

diff --git a/serd/serd.h b/serd/serd.h
index 2a79ff25..37a90750 100644
--- a/serd/serd.h
+++ b/serd/serd.h
@@ -915,12 +915,11 @@ serd_reader_set_default_graph(SerdReader*     reader,
                               const SerdNode* graph);
 
 /**
-   Read a file at a given `uri`.
+   Prepare to read from the file at a local file `uri`.
 */
 SERD_API
 SerdStatus
-serd_reader_read_file(SerdReader* reader,
-                      const char* uri);
+serd_reader_start_file(SerdReader* reader, const char* uri, bool bulk);
 
 /**
    Prepare to read from a stream.
@@ -972,14 +971,7 @@ serd_reader_read_document(SerdReader* reader);
 */
 SERD_API
 SerdStatus
-serd_reader_end_stream(SerdReader* reader);
-
-/**
-   Read `utf8`.
-*/
-SERD_API
-SerdStatus
-serd_reader_read_string(SerdReader* reader, const char* utf8);
+serd_reader_finish(SerdReader* reader);
 
 /**
    Free `reader`.
diff --git a/src/byte_source.c b/src/byte_source.c
index af01411e..2e2499f5 100644
--- a/src/byte_source.c
+++ b/src/byte_source.c
@@ -40,6 +40,7 @@ SerdStatus
 serd_byte_source_open_source(SerdByteSource*     source,
                              SerdReadFunc        read_func,
                              SerdStreamErrorFunc error_func,
+                             SerdStreamCloseFunc close_func,
                              void*               stream,
                              const char*         name,
                              size_t              page_size)
@@ -47,13 +48,14 @@ serd_byte_source_open_source(SerdByteSource*     source,
 	const Cursor cur = { name, 1, 1 };
 
 	memset(source, '\0', sizeof(*source));
+	source->read_func   = read_func;
+	source->error_func  = error_func;
+	source->close_func  = close_func;
 	source->stream      = stream;
-	source->from_stream = true;
 	source->page_size   = page_size;
 	source->buf_size    = page_size;
 	source->cur         = cur;
-	source->error_func  = error_func;
-	source->read_func   = read_func;
+	source->from_stream = true;
 
 	if (page_size > 1) {
 		source->file_buf = (uint8_t*)serd_bufalloc(page_size);
@@ -94,9 +96,14 @@ serd_byte_source_open_string(SerdByteSource* source, const char* utf8)
 SerdStatus
 serd_byte_source_close(SerdByteSource* source)
 {
+	SerdStatus st = SERD_SUCCESS;
+	if (source->close_func) {
+		st = source->close_func(source->stream) ? SERD_ERR_UNKNOWN
+		                                        : SERD_SUCCESS;
+	}
 	if (source->page_size > 1) {
 		free(source->file_buf);
 	}
 	memset(source, '\0', sizeof(*source));
-	return SERD_SUCCESS;
+	return st;
 }
diff --git a/src/byte_source.h b/src/byte_source.h
index dcf97bac..7a04db8d 100644
--- a/src/byte_source.h
+++ b/src/byte_source.h
@@ -23,6 +23,8 @@
 #include <stddef.h>
 #include <stdio.h>
 
+typedef int (*SerdStreamCloseFunc)(void*);
+
 typedef struct {
 	const char* filename;
 	unsigned    line;
@@ -32,6 +34,7 @@ typedef struct {
 typedef struct {
 	SerdReadFunc        read_func;    ///< Read function (e.g. fread)
 	SerdStreamErrorFunc error_func;   ///< Error function (e.g. ferror)
+	SerdStreamCloseFunc close_func;   ///< Function for closing stream
 	void*               stream;       ///< Stream (e.g. FILE)
 	size_t              page_size;    ///< Number of bytes to read at a time
 	size_t              buf_size;     ///< Number of bytes in file_buf
@@ -46,17 +49,13 @@ typedef struct {
 } SerdByteSource;
 
 SerdStatus
-serd_byte_source_open_file(SerdByteSource* source,
-                           FILE*           file,
-                           bool            bulk);
-
-SerdStatus
 serd_byte_source_open_string(SerdByteSource* source, const char* utf8);
 
 SerdStatus
 serd_byte_source_open_source(SerdByteSource*     source,
                              SerdReadFunc        read_func,
                              SerdStreamErrorFunc error_func,
+                             SerdStreamCloseFunc close_func,
                              void*               stream,
                              const char*         name,
                              size_t              page_size);
diff --git a/src/reader.c b/src/reader.c
index 3a72a3e4..7758384b 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -219,37 +219,6 @@ serd_reader_set_default_graph(SerdReader*     reader,
 	reader->default_graph = serd_node_copy(graph);
 }
 
-SerdStatus
-serd_reader_read_file(SerdReader* reader,
-                      const char* uri)
-{
-	char* const path = serd_file_uri_parse(uri, NULL);
-	if (!path) {
-		return SERD_ERR_BAD_ARG;
-	}
-
-	FILE* fd = serd_fopen(path, "rb");
-	if (!fd) {
-		serd_free(path);
-		return SERD_ERR_UNKNOWN;
-	}
-
-	SerdStatus st = serd_reader_start_stream(
-		reader, (SerdReadFunc)fread, (SerdStreamErrorFunc)ferror,
-		fd, path, SERD_PAGE_SIZE);
-
-	if (!st) {
-		st = serd_reader_read_document(reader);
-	}
-
-	const SerdStatus est = serd_reader_end_stream(reader);
-
-	fclose(fd);
-	free(path);
-
-	return st ? st : est;
-}
-
 static SerdStatus
 skip_bom(SerdReader* me)
 {
@@ -276,7 +245,31 @@ serd_reader_start_stream(SerdReader*         reader,
                          size_t              page_size)
 {
 	return serd_byte_source_open_source(
-		&reader->source, read_func, error_func, stream, name, page_size);
+		&reader->source, read_func, error_func, NULL, stream, name, page_size);
+}
+
+SerdStatus
+serd_reader_start_file(SerdReader* reader, const char* uri, bool bulk)
+{
+	char* const path = serd_file_uri_parse(uri, NULL);
+	if (!path) {
+		return SERD_ERR_BAD_ARG;
+	}
+
+	FILE* fd = serd_world_fopen(reader->world, path, "rb");
+	free(path);
+	if (!fd) {
+		return SERD_ERR_UNKNOWN;
+	}
+
+	return serd_byte_source_open_source(
+		&reader->source,
+		bulk ? (SerdReadFunc)fread : serd_file_read_byte,
+		(SerdStreamErrorFunc)ferror,
+		(SerdStreamCloseFunc)fclose,
+		fd,
+		uri,
+		bulk ? SERD_PAGE_SIZE : 1);
 }
 
 SerdStatus
@@ -318,18 +311,7 @@ serd_reader_read_chunk(SerdReader* reader)
 }
 
 SerdStatus
-serd_reader_end_stream(SerdReader* reader)
+serd_reader_finish(SerdReader* reader)
 {
 	return serd_byte_source_close(&reader->source);
 }
-
-SerdStatus
-serd_reader_read_string(SerdReader* reader, const char* utf8)
-{
-	serd_reader_start_string(reader, utf8);
-
-	const SerdStatus st  = serd_reader_read_document(reader);
-	const SerdStatus est = serd_byte_source_close(&reader->source);
-
-	return st ? st : est;
-}
diff --git a/src/serd_internal.h b/src/serd_internal.h
index 601af56c..f06c7051 100644
--- a/src/serd_internal.h
+++ b/src/serd_internal.h
@@ -29,7 +29,7 @@
 
 #include "world.h"
 
-#if defined(HAVE_POSIX_FADVISE) && defined(HAVE_FILENO)
+#if defined(HAVE_FILENO)
 #   include <fcntl.h>
 #endif
 
@@ -42,19 +42,20 @@
 #    define MIN(a, b) (((a) < (b)) ? (a) : (b))
 #endif
 
-static inline FILE*
-serd_fopen(const char* path, const char* mode)
+/** fread-like wrapper for getc (which is faster). */
+static inline size_t
+serd_file_read_byte(void* buf, size_t size, size_t nmemb, void* stream)
 {
-	FILE* fd = fopen(path, mode);
-	if (!fd) {
-		fprintf(stderr, "error: failed to open file %s (%s)\n",
-		        path, strerror(errno));
-		return NULL;
+	(void)size;
+	(void)nmemb;
+
+	const int c = getc((FILE*)stream);
+	if (c == EOF) {
+		*((uint8_t*)buf) = 0;
+		return 0;
 	}
-#if defined(HAVE_POSIX_FADVISE) && defined(HAVE_FILENO)
-	posix_fadvise(fileno(fd), 0, 0, POSIX_FADV_SEQUENTIAL);
-#endif
-	return fd;
+	*((uint8_t*)buf) = (uint8_t)c;
+	return 1;
 }
 
 static inline void*
diff --git a/src/serdi.c b/src/serdi.c
index eda73c2c..0d0bce04 100644
--- a/src/serdi.c
+++ b/src/serdi.c
@@ -27,6 +27,7 @@
 #include <string.h>
 
 #include "string_utils.h"
+#include "world.h"
 
 #define SERDI_ERROR(msg)       fprintf(stderr, "serdi: " msg);
 #define SERDI_ERRORF(fmt, ...) fprintf(stderr, "serdi: " fmt, __VA_ARGS__);
@@ -122,22 +123,6 @@ quiet_error_sink(void* handle, const SerdError* e)
 	return SERD_SUCCESS;
 }
 
-/** fread-like wrapper for getc (which is faster). */
-static size_t
-serd_file_read_byte(void* buf, size_t size, size_t nmemb, void* stream)
-{
-	(void)size;
-	(void)nmemb;
-
-	const int c = getc((FILE*)stream);
-	if (c == EOF) {
-		*((uint8_t*)buf) = 0;
-		return 0;
-	}
-	*((uint8_t*)buf) = (uint8_t)c;
-	return 1;
-}
-
 int
 main(int argc, char** argv)
 {
@@ -145,25 +130,23 @@ main(int argc, char** argv)
 		return print_usage(argv[0], true);
 	}
 
-	FILE*          in_fd         = NULL;
 	SerdSyntax     input_syntax  = (SerdSyntax)0;
 	SerdSyntax     output_syntax = (SerdSyntax)0;
-	bool           from_file     = true;
+	bool           from_string   = false;
+	bool           from_stdin    = false;
 	bool           ascii         = false;
 	bool           bulk_read     = true;
 	bool           bulk_write    = false;
 	bool           full_uris     = false;
 	bool           lax           = false;
 	bool           quiet         = false;
-	const char*    in_name       = NULL;
 	const char*    add_prefix    = NULL;
 	const char*    chop_prefix   = NULL;
 	const char*    root_uri      = NULL;
 	int            a             = 1;
 	for (; a < argc && argv[a][0] == '-'; ++a) {
 		if (argv[a][1] == '\0') {
-			in_name = (const char*)"(stdin)";
-			in_fd   = stdin;
+			from_stdin = true;
 			break;
 		} else if (argv[a][1] == 'a') {
 			ascii = true;
@@ -182,8 +165,7 @@ main(int argc, char** argv)
 		} else if (argv[a][1] == 'v') {
 			return print_version();
 		} else if (argv[a][1] == 's') {
-			in_name = (const char*)"(string)";
-			from_file = false;
+			from_string = true;
 			++a;
 			break;
 		} else if (argv[a][1] == 'i') {
@@ -229,22 +211,9 @@ main(int argc, char** argv)
 	_setmode(fileno(stdout), _O_BINARY);
 #endif
 
-	char*       input_path = NULL;
-	const char* input      = (const char*)argv[a++];
-	if (from_file) {
-		in_name = in_name ? in_name : input;
-		if (!in_fd) {
-			if (!strncmp(input, "file:", 5)) {
-				input_path = serd_file_uri_parse(input, NULL);
-				input      = input_path;
-			}
-			if (!input || !(in_fd = serd_fopen(input, "rb"))) {
-				return 1;
-			}
-		}
-	}
+	const char* input = (const char*)argv[a++];
 
-	if (!input_syntax && !(input_syntax = guess_syntax(in_name))) {
+	if (!input_syntax && !(input_syntax = guess_syntax(input))) {
 		input_syntax = SERD_TRIG;
 	}
 
@@ -260,7 +229,7 @@ main(int argc, char** argv)
 	if (a < argc) {  // Base URI given on command line
 		base = serd_node_new_uri_from_string(
 			(const char*)argv[a], NULL, &base_uri);
-	} else if (from_file && in_fd != stdin) {  // Use input file URI
+	} else if (!from_string && !from_stdin) {  // Use input file URI
 		base = serd_node_new_file_uri(input, NULL, &base_uri, true);
 	}
 
@@ -311,35 +280,32 @@ main(int argc, char** argv)
 	serd_node_free(root);
 
 	SerdStatus status = SERD_SUCCESS;
-	if (!from_file) {
+	if (from_string) {
 		status = serd_reader_start_string(reader, input);
+	} else if (from_stdin) {
+		status = serd_reader_start_stream(reader,
+		                                  serd_file_read_byte,
+		                                  (SerdStreamErrorFunc)ferror,
+		                                  stdin,
+		                                  "(stdin)",
+		                                  1);
 	} else {
-		status = serd_reader_start_stream(
-			reader,
-			bulk_read ? (SerdReadFunc)fread : serd_file_read_byte,
-			(SerdStreamErrorFunc)ferror,
-			in_fd,
-			in_name,
-			bulk_read ? SERD_PAGE_SIZE : 1);
+		status = serd_reader_start_file(reader, input, bulk_read);
 	}
 
-	status = serd_reader_read_document(reader);
-
-	serd_reader_end_stream(reader);
+	if (!status) {
+		status = serd_reader_read_document(reader);
+	}
 
+	serd_reader_finish(reader);
 	serd_reader_free(reader);
 	serd_writer_finish(writer);
 	serd_writer_free(writer);
 	serd_env_free(env);
 	serd_node_free(base);
 	serd_world_free(world);
-	free(input_path);
-
-	if (from_file) {
-		fclose(in_fd);
-	}
 
-	if (fclose(out_fd)) {
+	if (fclose(stdout)) {
 		perror("serdi: write error");
 		status = SERD_ERR_UNKNOWN;
 	}
diff --git a/src/world.c b/src/world.c
index 0210760d..7ec483e8 100644
--- a/src/world.c
+++ b/src/world.c
@@ -14,12 +14,35 @@
   OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 */
 
+#define _POSIX_C_SOURCE 200809L /* for posix_fadvise */
+
 #include "serd_internal.h"
+#include "serd_config.h"
 
 #include <stdarg.h>
+#include <stdio.h>
+#if defined(HAVE_POSIX_FADVISE)
+#   include <fcntl.h>
+#endif
 
 #include "world.h"
 
+FILE*
+serd_world_fopen(SerdWorld* world, const char* path, const char* mode)
+{
+	FILE* fd = fopen(path, mode);
+	if (!fd) {
+		serd_world_errorf(world, SERD_ERR_INTERNAL,
+		                  "failed to open file %s (%s)\n",
+		                  path, strerror(errno));
+		return NULL;
+	}
+#if defined(HAVE_POSIX_FADVISE) && defined(HAVE_FILENO)
+	posix_fadvise(fileno(fd), 0, 0, POSIX_FADV_SEQUENTIAL);
+#endif
+	return fd;
+}
+
 SerdStatus
 serd_world_error(const SerdWorld* world, const SerdError* e)
 {
diff --git a/src/world.h b/src/world.h
index 7e451b0e..aeea70c2 100644
--- a/src/world.h
+++ b/src/world.h
@@ -24,6 +24,8 @@ struct SerdWorldImpl {
 	void*         error_handle;
 };
 
+FILE* serd_world_fopen(SerdWorld* world, const char* path, const char* mode);
+
 SerdStatus serd_world_error(const SerdWorld* world, const SerdError* e);
 
 SerdStatus
diff --git a/tests/serd_test.c b/tests/serd_test.c
index 76ba1966..be4afe19 100644
--- a/tests/serd_test.c
+++ b/tests/serd_test.c
@@ -659,8 +659,8 @@ static void
 test_reader(const char* path)
 {
 	SerdWorld*  world  = serd_world_new();
-	ReaderTest* rt     = (ReaderTest*)calloc(1, sizeof(ReaderTest));
-	SerdSink    sink   = { rt, NULL, NULL, test_sink, NULL };
+	ReaderTest  rt     = { 0, NULL };
+	SerdSink    sink   = { &rt, NULL, NULL, test_sink, NULL };
 	SerdReader* reader = serd_reader_new(world, SERD_TURTLE, &sink);
 	assert(reader);
 
@@ -670,20 +670,18 @@ test_reader(const char* path)
 	serd_reader_add_blank_prefix(reader, NULL);
 	serd_node_free(g);
 
-	assert(serd_reader_read_file(reader, "http://notafile"));
-	assert(serd_reader_read_file(reader, "file:///better/not/exist"));
-	assert(serd_reader_read_file(reader, "file://"));
+	assert(serd_reader_start_file(reader, "http://notafile", false));
+	assert(serd_reader_start_file(reader, "file://invalid", false));
+	assert(serd_reader_start_file(reader, "file:///nonexistant", false));
 
-	const SerdStatus st = serd_reader_read_file(reader, path);
-	assert(!st);
-	assert(rt->n_statements == 13);
-	assert(rt->graph && serd_node_get_string(rt->graph) &&
-	       !strcmp(serd_node_get_string(rt->graph), "http://example.org/"));
-
-	assert(serd_reader_read_string(reader, "This isn't Turtle at all."));
+	assert(!serd_reader_start_file(reader, path, true));
+	assert(!serd_reader_read_document(reader));
+	assert(rt.n_statements == 13);
+	assert(rt.graph && serd_node_get_string(rt.graph) &&
+	       !strcmp(serd_node_get_string(rt.graph), "http://example.org/"));
+	serd_reader_finish(reader);
 
 	serd_reader_free(reader);
-	free(rt);
 	serd_world_free(world);
 }