Handle read errors more precisely

3 files changed, 39 insertions, 20 deletions

diff --git a/src/n3.c b/src/n3.c
index f8ce2b71..e7af476e 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -209,6 +209,11 @@ read_utf8_character(SerdReader* const reader,
   uint32_t   size     = 0;
   uint8_t    bytes[4] = {0, 0, 0, 0};
   SerdStatus st       = read_utf8_bytes(reader, bytes, &size, c);
+
+  if (!tolerate_status(reader, st)) {
+    return st;
+  }
+
   if (st) {
     const SerdStatus rst = push_bytes(reader, dest, replacement_char, 3);
     return rst ? rst : st;
@@ -341,12 +346,13 @@ read_STRING_LITERAL_LONG(SerdReader* const reader,
                          const uint8_t     q)
 {
   SerdStatus st = SERD_SUCCESS;
-
-  while (!(st && reader->strict)) {
+  while (tolerate_status(reader, st)) {
     const int c = peek_byte(reader);
     if (c == '\\') {
       skip_byte(reader, c);
       st = read_string_escape(reader, ref);
+    } else if (c == EOF) {
+      st = r_err(reader, SERD_NO_DATA, "unexpected end of file\n");
     } else if (c == q) {
       skip_byte(reader, q);
       const int q2 = eat_byte_safe(reader, peek_byte(reader));
@@ -365,14 +371,12 @@ read_STRING_LITERAL_LONG(SerdReader* const reader,
           st = read_character(reader, ref, (uint8_t)q2);
         }
       }
-    } else if (c == EOF) {
-      st = r_err(reader, SERD_BAD_SYNTAX, "end of file in long string\n");
     } else {
       st = read_character(reader, ref, (uint8_t)eat_byte_safe(reader, c));
     }
   }
 
-  return (st && reader->strict) ? st : SERD_SUCCESS;
+  return tolerate_status(reader, st) ? SERD_SUCCESS : st;
 }
 
 // STRING_LITERAL_QUOTE and STRING_LITERAL_SINGLE_QUOTE
@@ -384,7 +388,7 @@ read_STRING_LITERAL(SerdReader* const reader,
 {
   SerdStatus st = SERD_SUCCESS;
 
-  while (!st || !reader->strict) {
+  while (tolerate_status(reader, st)) {
     const int c = peek_byte(reader);
     switch (c) {
     case EOF:
@@ -405,7 +409,7 @@ read_STRING_LITERAL(SerdReader* const reader,
     }
   }
 
-  return st ? st : eat_byte_check(reader, q);
+  return tolerate_status(reader, st) ? SERD_SUCCESS : st;
 }
 
 static SerdStatus
@@ -792,15 +796,13 @@ read_IRIREF(SerdReader* const reader, SerdNode** const dest)
         }
       } else if (!(c & 0x80)) {
         st = push_byte(reader, *dest, c);
-      } else if (read_utf8_character(reader, *dest, (uint8_t)c)) {
-        if (reader->strict) {
-          return SERD_BAD_SYNTAX;
-        }
+      } else {
+        st = read_utf8_character(reader, *dest, (uint8_t)c);
       }
     }
   }
 
-  return st;
+  return tolerate_status(reader, st) ? SERD_SUCCESS : st;
 }
 
 static SerdStatus
@@ -1066,7 +1068,7 @@ read_BLANK_NODE_LABEL(SerdReader* const reader,
   // Adjust ID to avoid clashes with generated IDs if necessary
   st = adjust_blank_id(reader, buf);
 
-  return st;
+  return tolerate_status(reader, st) ? SERD_SUCCESS : st;
 }
 
 static SerdStatus
@@ -1102,7 +1104,8 @@ read_anon(SerdReader* const reader,
   ctx.subject = *dest;
   if (!empty) {
     bool ate_dot_in_list = false;
-    TRY_FAILING(st, read_predicateObjectList(reader, ctx, &ate_dot_in_list));
+    TRY(st, read_predicateObjectList(reader, ctx, &ate_dot_in_list));
+
     if (ate_dot_in_list) {
       return r_err(reader, SERD_BAD_SYNTAX, "'.' inside blank\n");
     }
@@ -1241,7 +1244,7 @@ read_objectList(SerdReader* const reader, ReadContext ctx, bool* const ate_dot)
       reader, SERD_BAD_SYNTAX, "syntax does not support abbreviation\n");
   }
 
-  while (!*ate_dot && eat_delim(reader, ',')) {
+  while (st <= SERD_FAILURE && !*ate_dot && eat_delim(reader, ',')) {
     st = read_object(reader, &ctx, true, ate_dot);
   }
 
@@ -1681,7 +1684,7 @@ read_turtleTrigDoc(SerdReader* const reader)
     const SerdStatus st              = read_n3_statement(reader);
 
     if (st > SERD_FAILURE) {
-      if (reader->strict) {
+      if (!tolerate_status(reader, st)) {
         serd_stack_pop_to(&reader->stack, orig_stack_size);
         return st;
       }
@@ -1738,9 +1741,7 @@ read_nquads_statement(SerdReader* const reader)
     TRY(st, eat_byte_check(reader, '.'));
   }
 
-  TRY(st, emit_statement(reader, ctx, ctx.object));
-
-  return SERD_SUCCESS;
+  return emit_statement(reader, ctx, ctx.object);
 }
 
 SerdStatus
diff --git a/src/reader.c b/src/reader.c
index 15c45202..f45e9e0d 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -53,6 +53,21 @@ genid_size(const SerdReader* const reader)
   return reader->bprefix_len + 1 + 10 + 1; // + "b" + UINT32_MAX + \0
 }
 
+bool
+tolerate_status(const SerdReader* const reader, const SerdStatus status)
+{
+  if (status == SERD_SUCCESS || status == SERD_FAILURE) {
+    return true;
+  }
+
+  if (status == SERD_BAD_STREAM || status == SERD_BAD_STACK ||
+      status == SERD_BAD_WRITE || status == SERD_NO_DATA) {
+    return false;
+  }
+
+  return !reader->strict;
+}
+
 SerdNode*
 blank_id(SerdReader* const reader)
 {
@@ -285,7 +300,7 @@ serd_reader_prepare(SerdReader* const reader)
   } else if (st == SERD_FAILURE) {
     reader->source.eof = true;
   } else {
-    r_err(reader, st, "read error: %s\n", strerror(errno));
+    r_err(reader, st, "error preparing read: %s\n", strerror(errno));
   }
   return st;
 }
diff --git a/src/reader.h b/src/reader.h
index 3f830ef4..fcbcb110 100644
--- a/src/reader.h
+++ b/src/reader.h
@@ -71,6 +71,9 @@ push_node(SerdReader*  reader,
 SERD_PURE_FUNC size_t
 genid_size(const SerdReader* reader);
 
+SERD_PURE_FUNC bool
+tolerate_status(const SerdReader* reader, SerdStatus status);
+
 SerdNode*
 blank_id(SerdReader* reader);