Configurable strict parsing to tolerate invalid URIs.

git-svn-id: http://svn.drobilla.net/serd/trunk@480 490d8e77-9747-427b-9fa3-0b8f29cee8a0

5 files changed, 34 insertions, 4 deletions

diff --git a/NEWS b/NEWS
index 54781a7c..bc8ae2f8 100644
--- a/NEWS
+++ b/NEWS
@@ -1,11 +1,13 @@
-serd (0.21.0) unstable;
+serd (0.21.1) unstable;
 
   * Remove dependence on fmax() to avoid portability issues
   * Fix serd_reader_read_file() for URIs with escaped characters (spaces)
+  * Add serd_reader_set_strict() and -l (lax) option to serdi to tolerate
+    parsing URIs with escaped characters
   * Report errors for invalid IRI characters and missing terminators
   * Fix warnings when building with ISO C++ compilers
 
- -- David Robillard <d@drobilla.net>  Wed, 10 Dec 2014 14:29:19 -0500
+ -- David Robillard <d@drobilla.net>  Wed, 10 Dec 2014 15:49:04 -0500
 
 serd (0.20.0) stable;
 
diff --git a/serd/serd.h b/serd/serd.h
index 9f9257ec..acc1ee4f 100644
--- a/serd/serd.h
+++ b/serd/serd.h
@@ -689,6 +689,17 @@ serd_reader_new(SerdSyntax        syntax,
                 SerdEndSink       end_sink);
 
 /**
+   Enable or disable strict parsing.
+
+   The reader is non-strict (lax) by default, which will tolerate URIs with
+   invalid characters.  Setting strict will fail when parsing such files.  An
+   error is printed for invalid input in either case.
+*/
+SERD_API
+void
+serd_reader_set_strict(SerdReader* reader, bool strict);
+
+/**
    Set a function to be called when errors occur during reading.
 
    The `error_sink` will be called with `handle` as its first argument.  If
diff --git a/src/reader.c b/src/reader.c
index 80f1c2ed..8dddb568 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -85,6 +85,7 @@ struct SerdReaderImpl {
 	uint8_t           read_byte;  ///< 1-byte 'buffer' used when not paging
 	bool              from_file;  ///< True iff reading from `fd`
 	bool              paging;     ///< True iff reading a page at a time
+	bool              strict;     ///< True iff strict parsing
 	bool              eof;
 	bool              seen_genid;
 #ifdef SERD_STACK_CHECK
@@ -810,7 +811,10 @@ read_IRIREF(SerdReader* reader)
 					r_err(reader, SERD_ERR_BAD_SYNTAX,
 					      "invalid IRI character (escape %%%02X)\n", c, c);
 				}
-				return pop_node(reader, ref);
+				if (reader->strict) {
+					return pop_node(reader, ref);
+				}
+				push_byte(reader, ref, eat_byte_safe(reader, c));
 			} else {
 				push_byte(reader, ref, eat_byte_safe(reader, c));
 			}
@@ -1473,6 +1477,7 @@ serd_reader_new(SerdSyntax        syntax,
 	me->read_buf         = 0;
 	me->file_buf         = 0;
 	me->read_head        = 0;
+	me->strict           = false;
 	me->eof              = false;
 	me->seen_genid       = false;
 #ifdef SERD_STACK_CHECK
@@ -1489,6 +1494,13 @@ serd_reader_new(SerdSyntax        syntax,
 
 SERD_API
 void
+serd_reader_set_strict(SerdReader* reader, bool strict)
+{
+	reader->strict = strict;
+}
+
+SERD_API
+void
 serd_reader_set_error_sink(SerdReader*   reader,
                            SerdErrorSink error_sink,
                            void*         error_handle)
diff --git a/src/serdi.c b/src/serdi.c
index 3543580a..7568378d 100644
--- a/src/serdi.c
+++ b/src/serdi.c
@@ -50,6 +50,7 @@ print_usage(const char* name, bool error)
 	fprintf(os, "  -f           Keep full URIs in input (don't qualify).\n");
 	fprintf(os, "  -h           Display this help and exit.\n");
 	fprintf(os, "  -i SYNTAX    Input syntax (`turtle' or `ntriples').\n");
+	fprintf(os, "  -l           Lax (non-strict) parsing.\n");
 	fprintf(os, "  -o SYNTAX    Output syntax (`turtle' or `ntriples').\n");
 	fprintf(os, "  -p PREFIX    Add PREFIX to blank node IDs.\n");
 	fprintf(os, "  -q           Suppress all output except data.\n");
@@ -100,6 +101,7 @@ main(int argc, char** argv)
 	bool           bulk_read     = true;
 	bool           bulk_write    = false;
 	bool           full_uris     = false;
+	bool           lax           = false;
 	bool           quiet         = false;
 	const uint8_t* in_name       = NULL;
 	const uint8_t* add_prefix    = NULL;
@@ -119,6 +121,8 @@ main(int argc, char** argv)
 			full_uris = true;
 		} else if (argv[a][1] == 'h') {
 			return print_usage(argv[0], false);
+		} else if (argv[a][1] == 'l') {
+			lax = true;
 		} else if (argv[a][1] == 'q') {
 			quiet = true;
 		} else if (argv[a][1] == 'v') {
@@ -215,6 +219,7 @@ main(int argc, char** argv)
 		(SerdStatementSink)serd_writer_write_statement,
 		(SerdEndSink)serd_writer_end_anon);
 
+	serd_reader_set_strict(reader, !lax);
 	if (quiet) {
 		serd_reader_set_error_sink(reader, quiet_error_sink, NULL);
 		serd_writer_set_error_sink(writer, quiet_error_sink, NULL);
diff --git a/wscript b/wscript
index 5c60873c..ab7d3824 100644
--- a/wscript
+++ b/wscript
@@ -11,7 +11,7 @@ import waflib.extras.autowaf as autowaf
 # major increment <=> incompatible changes
 # minor increment <=> compatible changes (additions)
 # micro increment <=> no interface changes
-SERD_VERSION       = '0.21.0'
+SERD_VERSION       = '0.21.1'
 SERD_MAJOR_VERSION = '0'
 
 # Mandatory waf variables