xml: comments

3 files changed, 277 insertions, 30 deletions

diff --git a/source/kit/xml.c b/source/kit/xml.c
index ed11b18..d815e9b 100644
--- a/source/kit/xml.c
+++ b/source/kit/xml.c
@@ -21,6 +21,38 @@ static kit_xml_intermediate_t kit_xml_parse_buf_(
   ib_t tag_text = ib_until(begin, SZ("<"));
   last          = ib_copy(tag_text);
 
+  for (;;) {
+    ib_t comment_open = ib_exact(last, SZ("<!--"));
+
+    if (comment_open.status != KIT_OK) {
+      ib_destroy(comment_open);
+      break;
+    }
+
+    ib_t comment_text  = ib_until(comment_open, SZ("-->"));
+    ib_t comment_close = ib_exact(comment_text, SZ("-->"));
+    ib_t next_text     = ib_until(comment_close, SZ("<"));
+
+    if (next_text.status == KIT_OK && next_text.data.size > 0) {
+      i64 n = tag_text.data.size;
+      DA_RESIZE(tag_text.data, n + next_text.data.size);
+
+      if (tag_text.data.size != n + next_text.data.size)
+        next_text.status = KIT_ERROR_BAD_ALLOC;
+      else
+        memcpy(tag_text.data.values + n, next_text.data.values,
+               next_text.data.size);
+    }
+
+    ib_destroy(last);
+    last = ib_copy(next_text);
+
+    ib_destroy(comment_open);
+    ib_destroy(comment_text);
+    ib_destroy(comment_close);
+    ib_destroy(next_text);
+  }
+
   DA_INIT(res.tags, 0, alloc);
 
   for (;;) {
@@ -39,7 +71,19 @@ static kit_xml_intermediate_t kit_xml_parse_buf_(
     xml_t tag;
     memset(&tag, 0, sizeof tag);
 
-    spaces        = ib_any(tag_open, SZ(" \t\r\n"));
+    ib_t decl_open = ib_exact(tag_open, SZ("?"));
+
+    ib_destroy(last);
+
+    if (decl_open.status == KIT_OK) {
+      tag.is_declaration = 1;
+      last               = ib_copy(decl_open);
+    } else
+      last = ib_copy(tag_open);
+
+    ib_destroy(decl_open);
+
+    spaces        = ib_any(last, SZ(" \t\r\n"));
     ib_t tag_name = ib_none(spaces, SZ(" \t\r\n/>"));
     ib_destroy(spaces);
 
@@ -50,7 +94,7 @@ static kit_xml_intermediate_t kit_xml_parse_buf_(
 
     for (;;) {
       spaces        = ib_any(last, SZ(" \t\r\n"));
-      ib_t property = ib_none(spaces, SZ(" \t\r\n=/>"));
+      ib_t property = ib_none(spaces, SZ(" \t\r\n=?/>"));
       ib_destroy(spaces);
 
       if (property.status != KIT_OK || property.data.size == 0) {
@@ -94,45 +138,92 @@ static kit_xml_intermediate_t kit_xml_parse_buf_(
       ib_destroy(value_text);
     }
 
-    spaces               = ib_any(last, SZ(" \t\r\n"));
-    ib_t tag_close       = ib_exact(spaces, SZ(">"));
-    ib_t tag_close_empty = ib_exact(spaces, SZ("/>"));
-    ib_destroy(spaces);
-
-    if (tag_close.status == KIT_OK) {
-      kit_xml_intermediate_t im = kit_xml_parse_buf_(tag_close,
-                                                     alloc);
-      tag.text                  = im.text;
-      tag.children              = im.tags;
+    spaces = ib_any(last, SZ(" \t\r\n"));
 
-      tagend_open = ib_exact(im.last, SZ("</"));
-      ib_destroy(im.last);
-      spaces           = ib_any(tagend_open, SZ(" \t\r\n"));
-      ib_t tagend_name = ib_exact(spaces, WRAP_STR(tag_name.data));
+    if (tag.is_declaration) {
+      ib_t tag_decl_close = ib_exact(spaces, SZ("?>"));
       ib_destroy(spaces);
-      spaces            = ib_any(tagend_name, SZ(" \t\r\n"));
-      ib_t tagend_close = ib_exact(spaces, SZ(">"));
-      ib_destroy(spaces);
-      ib_destroy(tagend_open);
-      ib_destroy(tagend_name);
-
-      ib_destroy(last);
-      last = tagend_close;
 
-    } else if (tag_close_empty.status == KIT_OK) {
       ib_destroy(last);
-      last = ib_copy(tag_close_empty);
+      last = tag_decl_close;
 
       DA_INIT(tag.text, 0, alloc);
       DA_INIT(tag.children, 0, alloc);
-    } else
-      last.status = KIT_ERROR_INTERNAL;
+    } else {
+      ib_t tag_close       = ib_exact(spaces, SZ(">"));
+      ib_t tag_close_empty = ib_exact(spaces, SZ("/>"));
+      ib_destroy(spaces);
+
+      if (tag_close.status == KIT_OK) {
+        kit_xml_intermediate_t im = kit_xml_parse_buf_(tag_close,
+                                                       alloc);
+        tag.text                  = im.text;
+        tag.children              = im.tags;
+
+        tagend_open = ib_exact(im.last, SZ("</"));
+        ib_destroy(im.last);
+        spaces           = ib_any(tagend_open, SZ(" \t\r\n"));
+        ib_t tagend_name = ib_exact(spaces, WRAP_STR(tag_name.data));
+        ib_destroy(spaces);
+        spaces            = ib_any(tagend_name, SZ(" \t\r\n"));
+        ib_t tagend_close = ib_exact(spaces, SZ(">"));
+        ib_destroy(spaces);
+        ib_destroy(tagend_open);
+        ib_destroy(tagend_name);
+
+        ib_destroy(last);
+        last = tagend_close;
+
+      } else if (tag_close_empty.status == KIT_OK) {
+        ib_destroy(last);
+        last = ib_copy(tag_close_empty);
+
+        DA_INIT(tag.text, 0, alloc);
+        DA_INIT(tag.children, 0, alloc);
+      } else
+        last.status = KIT_ERROR_INTERNAL;
+
+      ib_destroy(tag_close);
+      ib_destroy(tag_close_empty);
+    }
 
     ib_t tag_tail = ib_until(last, SZ("<"));
 
     ib_destroy(last);
     last = ib_copy(tag_tail);
 
+    for (;;) {
+      ib_t comment_open = ib_exact(last, SZ("<!--"));
+
+      if (comment_open.status != KIT_OK) {
+        ib_destroy(comment_open);
+        break;
+      }
+
+      ib_t comment_text  = ib_until(comment_open, SZ("-->"));
+      ib_t comment_close = ib_exact(comment_text, SZ("-->"));
+      ib_t next_text     = ib_until(comment_close, SZ("<"));
+
+      if (next_text.status == KIT_OK && next_text.data.size > 0) {
+        i64 n = tag_tail.data.size;
+        DA_RESIZE(tag_tail.data, n + next_text.data.size);
+
+        if (tag_tail.data.size != n + next_text.data.size)
+          next_text.status = KIT_ERROR_BAD_ALLOC;
+        else
+          memcpy(tag_tail.data.values + n, next_text.data.values,
+                 next_text.data.size);
+      }
+
+      ib_destroy(last);
+      last = ib_copy(next_text);
+
+      ib_destroy(comment_open);
+      ib_destroy(comment_text);
+      ib_destroy(comment_close);
+      ib_destroy(next_text);
+    }
+
     if (last.status == KIT_OK) {
       i64 n = res.tags.size;
       DA_RESIZE(res.tags, n + 1);
@@ -156,8 +247,6 @@ static kit_xml_intermediate_t kit_xml_parse_buf_(
 
     ib_destroy(tag_open);
     ib_destroy(tag_name);
-    ib_destroy(tag_close);
-    ib_destroy(tag_close_empty);
     ib_destroy(tag_tail);
   }
 
diff --git a/source/kit/xml.h b/source/kit/xml.h
index c33392b..00dde52 100644
--- a/source/kit/xml.h
+++ b/source/kit/xml.h
@@ -20,6 +20,7 @@ typedef KIT_DA(kit_xml_property_t) kit_da_xml_property_t;
 typedef KIT_DA(kit_xml_t) kit_da_xml_t;
 
 struct kit_xml_ {
+  i8                    is_declaration;
   kit_str_builder_t     tag;
   kit_str_builder_t     text;
   kit_str_builder_t     tail;
diff --git a/source/tests/xml.test.c b/source/tests/xml.test.c
index 52ff7c0..e477acb 100644
--- a/source/tests/xml.test.c
+++ b/source/tests/xml.test.c
@@ -10,6 +10,7 @@ TEST("xml parse tag") {
   REQUIRE_EQ(res.status, KIT_OK);
 
   if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 0);
     REQUIRE(AR_EQUAL(res.xml.tag, SZ("foo")));
     xml_destroy(&res.xml);
   }
@@ -17,6 +18,21 @@ TEST("xml parse tag") {
   is_destroy(is);
 }
 
+TEST("xml parse tag with dash") {
+  is_handle_t        is  = IS_WRAP_STRING(SZ("<foo-bar> </foo-bar>"));
+  xml_parse_result_t res = xml_parse(is, NULL);
+
+  REQUIRE_EQ(res.status, KIT_OK);
+
+  if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 0);
+    REQUIRE(AR_EQUAL(res.xml.tag, SZ("foo-bar")));
+    xml_destroy(&res.xml);
+  }
+
+  is_destroy(is);
+}
+
 TEST("xml parse tag not closed") {
   is_handle_t        is  = IS_WRAP_STRING(SZ("<foo>"));
   xml_parse_result_t res = xml_parse(is, NULL);
@@ -36,6 +52,7 @@ TEST("xml parse tag text") {
   REQUIRE_EQ(res.status, KIT_OK);
 
   if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 0);
     REQUIRE(AR_EQUAL(res.xml.tag, SZ("foo")));
     REQUIRE(AR_EQUAL(res.xml.text, SZ(" bar ")));
     xml_destroy(&res.xml);
@@ -51,6 +68,7 @@ TEST("xml parse empty tag") {
   REQUIRE_EQ(res.status, KIT_OK);
 
   if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 0);
     REQUIRE(AR_EQUAL(res.xml.tag, SZ("foo")));
     xml_destroy(&res.xml);
   }
@@ -143,4 +161,143 @@ TEST("xml parse child") {
   is_destroy(is);
 }
 
+TEST("xml parse declaration") {
+  is_handle_t        is  = IS_WRAP_STRING(SZ("<?foo ?>"));
+  xml_parse_result_t res = xml_parse(is, NULL);
+
+  REQUIRE_EQ(res.status, KIT_OK);
+
+  if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 1);
+    REQUIRE(AR_EQUAL(res.xml.tag, SZ("foo")));
+    xml_destroy(&res.xml);
+  }
+
+  is_destroy(is);
+}
+
+TEST("xml parse comment") {
+  is_handle_t        is  = IS_WRAP_STRING(SZ("<!-- foo --><bar />"));
+  xml_parse_result_t res = xml_parse(is, NULL);
+
+  REQUIRE_EQ(res.status, KIT_OK);
+
+  if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 0);
+    REQUIRE(AR_EQUAL(res.xml.tag, SZ("bar")));
+    xml_destroy(&res.xml);
+  }
+
+  is_destroy(is);
+}
+
+TEST("xml parse comment before text") {
+  is_handle_t is = IS_WRAP_STRING(SZ("<!-- foo --> bar <tag />"));
+  xml_parse_result_t res = xml_parse(is, NULL);
+
+  REQUIRE_EQ(res.status, KIT_OK);
+
+  if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 0);
+    REQUIRE(AR_EQUAL(res.xml.text, SZ(" bar ")));
+    xml_destroy(&res.xml);
+  }
+
+  is_destroy(is);
+}
+
+TEST("xml parse comment after text") {
+  is_handle_t is = IS_WRAP_STRING(SZ("foo <!-- bar --><tag />"));
+  xml_parse_result_t res = xml_parse(is, NULL);
+
+  REQUIRE_EQ(res.status, KIT_OK);
+
+  if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 0);
+    REQUIRE(AR_EQUAL(res.xml.text, SZ("foo ")));
+    xml_destroy(&res.xml);
+  }
+
+  is_destroy(is);
+}
+
+TEST("xml parse comment between text") {
+  is_handle_t is = IS_WRAP_STRING(
+      SZ("foo<!-- comment --> bar<tag />"));
+  xml_parse_result_t res = xml_parse(is, NULL);
+
+  REQUIRE_EQ(res.status, KIT_OK);
+
+  if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 0);
+    REQUIRE(AR_EQUAL(res.xml.text, SZ("foo bar")));
+    xml_destroy(&res.xml);
+  }
+
+  is_destroy(is);
+}
+
+TEST("xml parse comment tail") {
+  is_handle_t        is  = IS_WRAP_STRING(SZ("<foo /><!-- tail -->"));
+  xml_parse_result_t res = xml_parse(is, NULL);
+
+  REQUIRE_EQ(res.status, KIT_OK);
+
+  if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 0);
+    REQUIRE(AR_EQUAL(res.xml.tag, SZ("foo")));
+    xml_destroy(&res.xml);
+  }
+
+  is_destroy(is);
+}
+
+TEST("xml parse comment tail before text") {
+  is_handle_t is = IS_WRAP_STRING(SZ("<foo /><!-- tail --> bar"));
+  xml_parse_result_t res = xml_parse(is, NULL);
+
+  REQUIRE_EQ(res.status, KIT_OK);
+
+  if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 0);
+    REQUIRE(AR_EQUAL(res.xml.tag, SZ("foo")));
+    REQUIRE(AR_EQUAL(res.xml.tail, SZ(" bar")));
+    xml_destroy(&res.xml);
+  }
+
+  is_destroy(is);
+}
+
+TEST("xml parse comment tail after text") {
+  is_handle_t is = IS_WRAP_STRING(SZ("<foo /> bar <!-- tail -->"));
+  xml_parse_result_t res = xml_parse(is, NULL);
+
+  REQUIRE_EQ(res.status, KIT_OK);
+
+  if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 0);
+    REQUIRE(AR_EQUAL(res.xml.tag, SZ("foo")));
+    REQUIRE(AR_EQUAL(res.xml.tail, SZ(" bar ")));
+    xml_destroy(&res.xml);
+  }
+
+  is_destroy(is);
+}
+
+TEST("xml parse comment tail between text") {
+  is_handle_t is = IS_WRAP_STRING(SZ("<tag />foo<!-- tail --> bar"));
+  xml_parse_result_t res = xml_parse(is, NULL);
+
+  REQUIRE_EQ(res.status, KIT_OK);
+
+  if (res.status == KIT_OK) {
+    REQUIRE_EQ(res.xml.is_declaration, 0);
+    REQUIRE(AR_EQUAL(res.xml.tag, SZ("tag")));
+    REQUIRE(AR_EQUAL(res.xml.tail, SZ("foo bar")));
+    xml_destroy(&res.xml);
+  }
+
+  is_destroy(is);
+}
+
 #undef KIT_TEST_FILE