Preliminary support for writing abbreviated Turtle.

git-svn-id: http://svn.drobilla.net/serd/trunk@22 490d8e77-9747-427b-9fa3-0b8f29cee8a0

4 files changed, 286 insertions, 52 deletions

diff --git a/serd/serd.h b/serd/serd.h
index 8efaab20..d350cf4c 100644
--- a/serd/serd.h
+++ b/serd/serd.h
@@ -244,10 +244,17 @@ serd_reader_free(SerdReader reader);
  * @{
  */
 
+typedef enum {
+	SERD_STYLE_ABBREVIATED = 1,
+	SERD_STYLE_ASCII       = 1 << 1,
+	SERD_STYLE_ESCAPE_
+} SerdStyle;
+
 /** Create a new RDF writer. */
 SERD_API
 SerdWriter
 serd_writer_new(SerdSyntax     syntax,
+                SerdStyle      style,
                 SerdNamespaces ns,
                 const SerdURI* base_uri,
                 SerdSink       sink,
@@ -258,12 +265,19 @@ SERD_API
 void
 serd_writer_free(SerdWriter writer);
 
-/** Set the base URI of writer. */
+/** Set the current output base URI. */
 SERD_API
 void
 serd_writer_set_base_uri(SerdWriter     writer,
                          const SerdURI* uri);
 
+/** Set the current output base URI. */
+SERD_API
+void
+serd_writer_set_prefix(SerdWriter        writer,
+                       const SerdString* name,
+                       const SerdString* uri);
+
 /** Write a statement. */
 SERD_API
 bool
@@ -278,6 +292,11 @@ serd_writer_write_statement(SerdWriter        writer,
                             const SerdString* object_datatype,
                             const SerdString* object_lang);
 
+/** Finish a write. */
+SERD_API
+void
+serd_writer_finish(SerdWriter writer);
+
 /** @} */
 
 /** @} */
diff --git a/src/serdi.c b/src/serdi.c
index 5db96388..a688b41b 100644
--- a/src/serdi.c
+++ b/src/serdi.c
@@ -90,6 +90,8 @@ event_prefix(void*             handle,
 	} else {
 		serd_namespaces_add(state->ns, name, uri_string);
 	}
+	serd_writer_set_prefix(state->writer, name, uri_string);
+
 	return true;
 }
 
@@ -133,13 +135,39 @@ file_sink(const void* buf, size_t len, void* stream)
 int
 main(int argc, char** argv)
 {
-	if (argc != 2 && argc != 3) {
+	if (argc < 2) {
 		return print_usage(argv[0], true);
 	}
 
-	const uint8_t* in_filename = (const uint8_t*)argv[1];
+	FILE*      in_fd         = NULL;
+	SerdSyntax output_syntax = SERD_NTRIPLES;
+
+	int a = 1;
+	for (; a < argc && argv[a][0] == '-'; ++a) {
+		if (argv[a][1] == '\0') {
+			in_fd = stdin;
+			break;
+		} else if (argv[a][1] == 'o') {
+			if (++a == argc) {
+				fprintf(stderr, "missing value for -i\n");
+				return 1;
+			}
+			if (!strcmp(argv[a], "turtle")) {
+				output_syntax = SERD_TURTLE;
+			} else if (!strcmp(argv[a], "ntriples")) {
+				output_syntax = SERD_NTRIPLES;
+			} else {
+				fprintf(stderr, "unknown output format `%s'\n",  argv[a]);
+			}
+		} else {
+			fprintf(stderr, "unknown option `%s'\n", argv[a]);
+			return print_usage(argv[0], true);
+		}
+	}
+	
+	const uint8_t* in_filename = (const uint8_t*)argv[a];
 
-	if (serd_uri_string_has_scheme(in_filename)) {
+	if (!in_fd && serd_uri_string_has_scheme(in_filename)) {
 		// Input is an absolute URI, ensure it's a file: URI and chop scheme
 		if (strncmp((const char*)in_filename, "file:", 5)) {
 			fprintf(stderr, "unsupported URI scheme `%s'\n", in_filename);
@@ -166,8 +194,11 @@ main(int argc, char** argv)
 
 	serd_uri_parse(base_uri_str->buf, &base_uri);
 
-	FILE* const in_fd  = fopen((const char*)in_filename,  "r");
-	FILE*       out_fd = stdout;
+	if (!in_fd) {
+		in_fd  = fopen((const char*)in_filename,  "r");
+	}
+	
+	FILE* out_fd = stdout;
 
 	if (!in_fd) {
 		fprintf(stderr, "failed to open file %s\n", in_filename);
@@ -175,8 +206,14 @@ main(int argc, char** argv)
 	}
 
 	SerdNamespaces ns = serd_namespaces_new();
+
+	SerdStyle output_style = (output_syntax == SERD_NTRIPLES)
+		? SERD_STYLE_ASCII
+		: SERD_STYLE_ABBREVIATED;
+
 	State state = {
-		serd_writer_new(SERD_NTRIPLES, ns, &base_uri, file_sink, out_fd),
+		serd_writer_new(output_syntax, output_style,
+		                ns, &base_uri, file_sink, out_fd),
 		ns, base_uri_str, base_uri
 	};
 
@@ -186,8 +223,11 @@ main(int argc, char** argv)
 	const bool success = serd_reader_read_file(reader, in_fd, in_filename);
 	serd_reader_free(reader);
 	fclose(in_fd);
-	serd_namespaces_free(state.ns);
+
+	serd_writer_finish(state.writer);
 	serd_writer_free(state.writer);
+
+	serd_namespaces_free(state.ns);
 	serd_string_free(state.base_uri_str);
 
 	if (success) {
diff --git a/src/writer.c b/src/writer.c
index ef10594b..684f79e2 100644
--- a/src/writer.c
+++ b/src/writer.c
@@ -21,16 +21,48 @@
 
 #include "serd/serd.h"
 
+typedef bool (*StatementWriter)(SerdWriter        writer,
+                                const SerdString* graph,
+                                const SerdString* subject,
+                                SerdNodeType      subject_type,
+                                const SerdString* predicate,
+                                SerdNodeType      predicate_type,
+                                const SerdString* object,
+                                SerdNodeType      object_type,
+                                const SerdString* object_datatype,
+                                const SerdString* object_lang);
+
+typedef bool (*NodeWriter)(SerdWriter        writer,
+                           SerdNodeType      type,
+                           const SerdString* str,
+                           const SerdString* datatype,
+                           const SerdString* lang);
+
 struct SerdWriterImpl {
-	SerdSyntax     syntax;
-	SerdNamespaces ns;
-	SerdURI        base_uri;
-	SerdSink       sink;
-	void*          stream;
+	SerdSyntax        syntax;
+	SerdStyle         style;
+	SerdNamespaces    ns;
+	SerdURI           base_uri;
+	SerdSink          sink;
+	void*             stream;
+	StatementWriter   write_statement;
+	NodeWriter        write_node;
+	const SerdString* prev_g;
+	const SerdString* prev_s;
+	const SerdString* prev_p;
+	const SerdString* prev_o;
+	unsigned          indent;
 };
 
+typedef enum {
+	WRITE_NORMAL,
+	WRITE_URI,
+	WRITE_STRING
+} WriteContext;
+	
 static bool
-serd_write_ascii(SerdWriter writer, const uint8_t* utf8, size_t n_bytes, const uint8_t esc)
+write_text(SerdWriter writer, WriteContext ctx,
+           const uint8_t* utf8, size_t n_bytes, uint8_t terminator)
 {
 	char escape[10] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
 	for (size_t i = 0; i < n_bytes;) {
@@ -40,12 +72,16 @@ serd_write_ascii(SerdWriter writer, const uint8_t* utf8, size_t n_bytes, const u
 		case '\n': writer->sink("\\n", 2, writer->stream); continue;
 		case '\r': writer->sink("\\r", 2, writer->stream); continue;
 		case '\t': writer->sink("\\t", 2, writer->stream); continue;
-		case '"':  if (esc == '"') { writer->sink("\\\"", 2, writer->stream); continue; }
+		case '"':
+			if (terminator == '"') {
+				writer->sink("\\\"", 2, writer->stream);
+				continue;
+			} // else fall-through
 		default: break;
 		}
 
-		if (in == esc) {
-			sprintf(escape, "\\u%04X", esc);
+		if (in == terminator) {
+			sprintf(escape, "\\u%04X", terminator);
 			writer->sink(escape, 6, writer->stream);
 			continue;
 		}
@@ -79,6 +115,13 @@ serd_write_ascii(SerdWriter writer, const uint8_t* utf8, size_t n_bytes, const u
 			return false;
 		}
 
+		if (!(writer->style & SERD_STYLE_ASCII)) {
+			// Write UTF-8 input directly to UTF-8 output
+			writer->sink(utf8, n_bytes, writer->stream);
+			i += n_bytes - 1;
+			continue;
+		}
+
 #define READ_BYTE() do { \
 			assert(i < n_bytes); \
 			in = utf8[i++] & 0x3f; \
@@ -106,11 +149,11 @@ serd_write_ascii(SerdWriter writer, const uint8_t* utf8, size_t n_bytes, const u
 }
 
 static bool
-serd_write_node(SerdWriter        writer,
-                SerdNodeType      type,
-                const SerdString* str,
-                const SerdString* datatype,
-                const SerdString* lang)
+write_node(SerdWriter        writer,
+           SerdNodeType      type,
+           const SerdString* str,
+           const SerdString* datatype,
+           const SerdString* lang)
 {
 	const SerdURI* base_uri = &writer->base_uri;
 	SerdNamespaces ns       = writer->ns;
@@ -123,14 +166,20 @@ serd_write_node(SerdWriter        writer,
 		writer->sink(str->buf, str->n_bytes - 1, writer->stream);
 		break;
 	case QNAME:
-		if (!serd_namespaces_expand(ns, str, &uri_prefix, &uri_suffix)) {
-			fprintf(stderr, "error: undefined namespace prefix `%s'\n", str->buf);
-			return false;
+		switch (writer->syntax) {
+		case SERD_NTRIPLES:
+			if (!serd_namespaces_expand(ns, str, &uri_prefix, &uri_suffix)) {
+				fprintf(stderr, "error: undefined namespace prefix `%s'\n", str->buf);
+				return false;
+			}
+			writer->sink("<", 1, writer->stream);
+			write_text(writer, WRITE_URI, uri_prefix.buf, uri_prefix.len, '>');
+			write_text(writer, WRITE_URI, uri_suffix.buf, uri_suffix.len, '>');
+			writer->sink(">", 1, writer->stream);
+			break;
+		case SERD_TURTLE:
+			writer->sink(str->buf, str->n_bytes - 1, writer->stream);
 		}
-		writer->sink("<", 1, writer->stream);
-		serd_write_ascii(writer, uri_prefix.buf, uri_prefix.len, '>');
-		serd_write_ascii(writer, uri_suffix.buf, uri_suffix.len, '>');
-		writer->sink(">", 1, writer->stream);
 		break;
 	case URI:
 		if (!serd_uri_string_has_scheme(str->buf)) {
@@ -146,27 +195,44 @@ serd_write_node(SerdWriter        writer,
 			}
 		} else {
 			writer->sink("<", 1, writer->stream);
-			serd_write_ascii(writer, str->buf, str->n_bytes - 1, '>');
+			write_text(writer, WRITE_URI, str->buf, str->n_bytes - 1, '>');
 			writer->sink(">", 1, writer->stream);
 			return true;
 		}
 		return false;
 	case LITERAL:
 		writer->sink("\"", 1, writer->stream);
-		serd_write_ascii(writer, str->buf, str->n_bytes - 1, '"');
+		write_text(writer, WRITE_STRING, str->buf, str->n_bytes - 1, '"');
 		writer->sink("\"", 1, writer->stream);
 		if (lang) {
 			writer->sink("@", 1, writer->stream);
 			writer->sink(lang->buf, lang->n_bytes - 1, writer->stream);
 		} else if (datatype) {
 			writer->sink("^^", 2, writer->stream);
-			serd_write_node(writer, URI, datatype, NULL, NULL);
+			write_node(writer, URI, datatype, NULL, NULL);
 		}
 		break;
 	}
 	return true;
 }
 
+static void
+serd_writer_write_delim(SerdWriter writer, const uint8_t delim)
+{
+	switch (delim) {
+	case 0:
+	case '\n':
+		break;
+	default:
+		writer->sink(" ", 1, writer->stream);
+		writer->sink(&delim, 1, writer->stream);
+	}
+	writer->sink("\n", 1, writer->stream);
+	for (unsigned i = 0; i < writer->indent; ++i) {
+		writer->sink("\t", 1, writer->stream);
+	}
+}
+
 SERD_API
 bool
 serd_writer_write_statement(SerdWriter        writer,
@@ -180,18 +246,98 @@ serd_writer_write_statement(SerdWriter        writer,
                             const SerdString* object_datatype,
                             const SerdString* object_lang)
 {
-	serd_write_node(writer, subject_type, subject, NULL, NULL);
+	return writer->write_statement(writer,
+		graph,
+		subject, subject_type,
+		predicate, predicate_type,
+		object, object_type, object_datatype, object_lang);
+}
+
+static bool
+serd_writer_write_statement_abbrev(SerdWriter        writer,
+                                   const SerdString* graph,
+                                   const SerdString* subject,
+                                   SerdNodeType      subject_type,
+                                   const SerdString* predicate,
+                                   SerdNodeType      predicate_type,
+                                   const SerdString* object,
+                                   SerdNodeType      object_type,
+                                   const SerdString* object_datatype,
+                                   const SerdString* object_lang)
+{
+	assert(subject && predicate && object);
+	if (subject == writer->prev_s) {
+		if (predicate == writer->prev_p) {
+			++writer->indent;
+			serd_writer_write_delim(writer, ',');
+			write_node(writer, object_type, object, object_datatype, object_lang);
+			--writer->indent;
+		} else {
+			serd_writer_write_delim(writer, ';');
+			write_node(writer, predicate_type, predicate, NULL, NULL);
+			writer->sink(" ", 1, writer->stream);
+			write_node(writer, object_type, object, object_datatype, object_lang);
+		}
+	} else {
+		if (writer->prev_s) {
+			--writer->indent;
+			serd_writer_write_delim(writer, '.');
+			serd_writer_write_delim(writer, '\n');
+		}
+		write_node(writer, subject_type, subject, NULL, NULL);
+		++writer->indent;
+		serd_writer_write_delim(writer, 0);
+
+		writer->sink(" ", 1, writer->stream);
+		write_node(writer, predicate_type, predicate, NULL, NULL);
+		writer->sink(" ", 1, writer->stream);
+
+		write_node(writer, object_type, object, object_datatype, object_lang);
+	}
+
+	writer->prev_g = graph;
+	writer->prev_s = subject;
+	writer->prev_p = predicate;
+	writer->prev_o = object;
+	return true;
+}
+
+SERD_API
+bool
+serd_writer_write_statement_flat(SerdWriter        writer,
+                                 const SerdString* graph,
+                                 const SerdString* subject,
+                                 SerdNodeType      subject_type,
+                                 const SerdString* predicate,
+                                 SerdNodeType      predicate_type,
+                                 const SerdString* object,
+                                 SerdNodeType      object_type,
+                                 const SerdString* object_datatype,
+                                 const SerdString* object_lang)
+{
+	assert(subject && predicate && object);
+	write_node(writer, subject_type, subject, NULL, NULL);
 	writer->sink(" ", 1, writer->stream);
-	serd_write_node(writer, predicate_type, predicate, NULL, NULL);
+	write_node(writer, predicate_type, predicate, NULL, NULL);
 	writer->sink(" ", 1, writer->stream);
-	serd_write_node(writer, object_type, object, object_datatype, object_lang);
+	write_node(writer, object_type, object, object_datatype, object_lang);
 	writer->sink(" .\n", 3, writer->stream);
 	return true;
 }
 
 SERD_API
+void
+serd_writer_finish(SerdWriter writer)
+{
+	if (writer->prev_s) {
+		writer->sink(" .\n", 3, writer->stream);
+	}
+}
+
+SERD_API
 SerdWriter
 serd_writer_new(SerdSyntax     syntax,
+                SerdStyle      style,
                 SerdNamespaces ns,
                 const SerdURI* base_uri,
                 SerdSink       sink,
@@ -199,10 +345,22 @@ serd_writer_new(SerdSyntax     syntax,
 {
 	SerdWriter writer = malloc(sizeof(struct SerdWriterImpl));
 	writer->syntax   = syntax;
+	writer->style    = style;
 	writer->ns       = ns;
 	writer->base_uri = *base_uri;
 	writer->sink     = sink;
 	writer->stream   = stream;
+	writer->prev_g   = 0;
+	writer->prev_s   = 0;
+	writer->prev_p   = 0;
+	writer->prev_o   = 0;
+	writer->indent   = 0;
+	writer->write_node = write_node;
+	if ((style & SERD_STYLE_ABBREVIATED)) {
+		writer->write_statement = serd_writer_write_statement_abbrev;
+	} else {
+		writer->write_statement = serd_writer_write_statement_flat;
+	}
 	return writer;
 }
 
@@ -216,6 +374,21 @@ serd_writer_set_base_uri(SerdWriter     writer,
 
 SERD_API
 void
+serd_writer_set_prefix(SerdWriter        writer,
+                       const SerdString* name,
+                       const SerdString* uri)
+{
+	if (writer->syntax != SERD_NTRIPLES) {
+		writer->sink("@prefix ", 8, writer->stream); 
+		writer->sink(name->buf, name->n_bytes - 1, writer->stream);
+		writer->sink(": <", 3, writer->stream);
+		write_text(writer, WRITE_URI, uri->buf, uri->n_bytes - 1, '>');
+		writer->sink("> .\n", 4, writer->stream); 
+	}
+}
+
+SERD_API
+void
 serd_writer_free(SerdWriter writer)
 {
 	SerdWriter const me = (SerdWriter)writer;
diff --git a/wscript b/wscript
index f3bf2830..e8f18366 100644
--- a/wscript
+++ b/wscript
@@ -124,10 +124,20 @@ def test(ctx):
 		
 	autowaf.pre_test(ctx, APPNAME)
 
+	autowaf.run_tests(ctx, APPNAME,
+					  ['./serdi_static > /dev/null',
+					  './serdi_static ftp://example.org/unsupported.ttl > /dev/null'],
+					  1, name='serdi-fail')
+
+	autowaf.run_tests(ctx, APPNAME,
+					  ['./serdi_static file:../tests/manifest.ttl > /dev/null',
+					   './serdi_static ../tests/UTF-8.ttl > /dev/null'],
+					  0, name='serdi-succeed')
+
 	commands = []
 	for test in good_tests:
 		base_uri = 'http://www.w3.org/2001/sw/DataAccess/df1/' + test
-		commands = commands + [ './serdi_static ../%s \'%s\' > %s.out' % (test, base_uri, test) ]
+		commands += [ './serdi_static ../%s \'%s\' > %s.out' % (test, base_uri, test) ]
 
 	autowaf.run_tests(ctx, APPNAME, commands, 0, name='good')
 
@@ -145,24 +155,16 @@ def test(ctx):
 	
 	commands = []
 	for test in bad_tests:
-	    commands = commands + [ './serdi_static ../%s \'http://www.w3.org/2001/sw/DataAccess/df1/%s\' > %s.out' % (test, test, test) ]
+	    commands += [ './serdi_static ../%s \'http://www.w3.org/2001/sw/DataAccess/df1/%s\' > %s.out' % (test, test, test) ]
 
 	autowaf.run_tests(ctx, APPNAME, commands, 1, name='bad')
 
-	autowaf.run_tests(ctx, APPNAME,
-					  ['./serdi_static > /dev/null'],
-					  1, name='serdi-no-args')
-
-	autowaf.run_tests(ctx, APPNAME,
-					  ['./serdi_static file:../tests/manifest.ttl > /dev/null'],
-					  0, name='serdi-file-uri')
-
-	autowaf.run_tests(ctx, APPNAME,
-					  ['./serdi_static ftp://example.org/unsupported.ttl > /dev/null'],
-					  1, name='serdi-bad-uri')
-
-	autowaf.run_tests(ctx, APPNAME,
-					  ['./serdi_static ../tests/UTF-8.ttl > /dev/null'],
-					  0, name='utf8')
+#	commands = []
+#	for test in good_tests:
+#		out_filename = test + '.thru'
+#		commands += [ './serdi_static -o turtle ../%s \'%s\' | ./serdi_static - \'%s\' > %s.out' % (test, base_uri, base_uri, test) ]
+#		
+#	autowaf.run_tests(ctx, APPNAME, commands, 0, name='turtle-write')
+#
 
 	autowaf.post_test(ctx, APPNAME)