Stream entities

src/main/c/error/error.c

@@ -21,6 +21,7 @@
 
 typedef enum hbe_errcode {
   HBE_INTERR_PEEK_OFFSET_GEQ_ZERO = 1,
+  HBE_INTERR_UNKNOWN_ENTITY_TYPE,
 
   HBE_CLI_TOO_MANY_OPTIONS = 17,
   HBE_CLI_INVALID_TAG_SET,

src/main/c/main.c

@@ -219,7 +219,7 @@ int main(int argc, char **argv) {
     hbe_fatal_set_autodelete(output_path);
   }
 
-  hbs_content(pipe);
+  hbs_content(config_stream, pipe);
 
   if (config_buffer) {
     output = hbu_fstreamout_create(output_path);

src/main/c/stream/comment.c

@@ -4,6 +4,7 @@
 #include "../util/hbchar.h"
 #include "../util/pipe.c"
 
+#include "./streamoptions.c"
 void hbs_comment(hbu_pipe_t pipe)
 {
   hbu_pipe_require(pipe, '<');

src/main/c/stream/content.c

@@ -5,13 +5,15 @@
 #include "../util/pipe.c"
 
 // Declare first before tag.c, as tag.c depends on it
-void hbs_content(hbu_pipe_t pipe);
+void hbs_content(hbs_options_t so, hbu_pipe_t pipe);
 
+#include "./streamoptions.c"
 #include "./tag.c"
 #include "./bang.c"
 #include "./comment.c"
+#include "./entity.c"
 
-void hbs_content(hbu_pipe_t pipe) {
+void hbs_content(hbs_options_t so, hbu_pipe_t pipe) {
   while (1) {
     hb_eod_char_t c = hbu_pipe_peek_eoi(pipe);
 
@@ -32,13 +34,12 @@ void hbs_content(hbu_pipe_t pipe) {
         hbs_bang(pipe);
 
       } else {
-        hbs_tag(pipe);
+        hbs_tag(so, pipe);
       }
       break;
 
     case '&':
-      // TODO
-      hbu_pipe_accept(pipe);
+      hbs_entity(so, pipe);
       break;
 
     default:

src/main/c/stream/entity.c

@@ -0,0 +1,160 @@
+#ifndef _HDR_HYPERBUILD_STREAM_ENTITY
+#define _HDR_HYPERBUILD_STREAM_ENTITY
+
+#include <inttypes.h>
+
+#include "../ext/nicehash/set/int32.h"
+
+#include "../util/hbchar.h"
+#include "../util/buffer.c"
+#include "../util/pipe.c"
+
+#include "../rule/entity/entityrefs.c"
+#include "../rule/char/lcalpha.c"
+#include "../rule/char/ucalpha.c"
+#include "../rule/char/digit.c"
+#include "../rule/char/hex.c"
+
+#define HBS_ENTITY_TYPE_NAME 1
+#define HBS_ENTITY_TYPE_DECIMAL 2
+#define HBS_ENTITY_TYPE_HEXADECIMAL 3
+
+#define HBS_ENTITY_MAX_ENTITY_LENGTH 10 // Don't try to parse malformed entities forever
+
+static void _hbs_entity_interr_unknown_entity(void) {
+  hbe_fatal(HBE_INTERR_UNKNOWN_ENTITY_TYPE, "INTERR $type is not a known entity type");
+}
+
+static void _hbs_entity_handle_error(hbs_options_t so, hbu_pipe_t pipe, int type, hbu_buffer_t entity_raw, int consumed_semicolon, hbe_errcode_t errcode, const char *reason) {
+  if (nh_set_int32_has(so->suppressed_errors, errcode)) {
+    switch (type) {
+    case -1:
+      hbu_pipe_write(pipe, '&');
+      break;
+
+    case HBS_ENTITY_TYPE_NAME:
+    case HBS_ENTITY_TYPE_DECIMAL:
+      hbu_pipe_write(pipe, '&');
+      // fall through
+
+    case HBS_ENTITY_TYPE_HEXADECIMAL:
+      hbu_pipe_write(pipe, '#');
+      break;
+
+    default:
+      _hbs_entity_interr_unknown_entity();
+    }
+
+    hbu_pipe_write_buffer(pipe, entity_raw);
+
+    if (consumed_semicolon) {
+      hbu_pipe_write(pipe, ';');
+    }
+
+    return;
+  }
+
+  hbu_pipe_error(pipe, errcode, reason);
+}
+
+void hbs_entity(hbs_options_t so, hbu_pipe_t pipe) {
+  hbu_pipe_require_skip(pipe, '&');
+
+  hb_char_t c = hbu_pipe_peek(pipe);
+
+  hbu_buffer_t entity_raw = hbu_buffer_create_size(HBS_ENTITY_MAX_ENTITY_LENGTH + 1);
+
+  int type = -1;
+
+  if (hbr_lcalpha_check(c) || hbr_ucalpha_check(c)) {
+    // Name-based entity
+    type = HBS_ENTITY_TYPE_NAME;
+
+  } else if (c == '#' && hbu_pipe_peek_offset(pipe, 2) == 'x') {
+    // Hexadecimal-based entity
+    // NOTE: Check before decimal-based
+    hbu_pipe_skip_amount(pipe, 2);
+    type = HBS_ENTITY_TYPE_HEXADECIMAL;
+
+  } else if (c == '#') {
+    // Decimal-based entity
+    hbu_pipe_skip(pipe);
+    type = HBS_ENTITY_TYPE_DECIMAL;
+
+  } else {
+    // Malformed entity
+    return _hbs_entity_handle_error(so, pipe, type, entity_raw, 0, HBE_PARSE_MALFORMED_ENTITY, "Invalid character after ampersand");
+  }
+
+  int under_max = 0;
+
+  for (int i = 0; i < HBS_ENTITY_MAX_ENTITY_LENGTH; i++) {
+    hb_char_t e = hbu_pipe_skip(pipe);
+
+    if (e == ';') {
+      under_max = 1;
+      break;
+    }
+
+    int well_formed = 0;
+
+    switch (type) {
+    case HBS_ENTITY_TYPE_NAME:
+      well_formed = hbr_lcalpha_check(e) || hbr_ucalpha_check(e);
+      break;
+
+    case HBS_ENTITY_TYPE_DECIMAL:
+      well_formed = hbr_digit_check(e);
+      break;
+
+    case HBS_ENTITY_TYPE_HEXADECIMAL:
+      well_formed = hbr_hex_check(e);
+      break;
+
+    default:
+      _hbs_entity_interr_unknown_entity();
+    }
+
+    if (!well_formed) {
+      return _hbs_entity_handle_error(so, pipe, type, entity_raw, 0, HBE_PARSE_MALFORMED_ENTITY, "Characters after ampersand don't form entity");
+    }
+
+    hbu_buffer_append(entity_raw, e);
+  }
+
+  if (!under_max) {
+    // Malformed entity (too long)
+    return _hbs_entity_handle_error(so, pipe, type, entity_raw, 0, HBE_PARSE_MALFORMED_ENTITY, "Malformed entity");
+  }
+
+  int valid = 1;
+  hb_char_t *entity_raw_u = hbu_buffer_underlying(entity_raw);
+  uintmax_t code_point;
+
+  switch (type) {
+  case HBS_ENTITY_TYPE_NAME:
+    valid = hbr_entityrefs_check(entity_raw_u);
+    if (valid && so->decode_entities) {
+      hbu_pipe_write(pipe, hbr_entityrefs_get(entity_raw_u));
+    }
+    break;
+
+  case HBS_ENTITY_TYPE_DECIMAL:
+  case HBS_ENTITY_TYPE_HEXADECIMAL:
+    code_point = strtoumax((char *) entity_raw_u, NULL, (type == HBS_ENTITY_TYPE_DECIMAL) ? 10 : 16);
+    valid = errno == 0 && code_point <= 0x10FFFF;
+    if (valid && so->decode_entities) {
+      valid = hbu_pipe_write_unicode(pipe, code_point);
+    }
+    break;
+
+  default:
+    _hbs_entity_interr_unknown_entity();
+  }
+
+  if (!valid) {
+    return _hbs_entity_handle_error(so, pipe, type, entity_raw, 1, HBE_PARSE_INVALID_ENTITY, "Invalid entity");
+  }
+}
+
+#endif // _HDR_HYPERBUILD_STREAM_ENTITY

src/main/c/stream/tag.c

@@ -12,13 +12,15 @@
 #include "./helper/tagname.c"
 #include "./helper/attr.c"
 #include "./helper/script.c"
+#include "./helper/style.c"
 
 // Declare first before content.c, as content.c depends on it
-void hbs_tag(hbu_pipe_t pipe);
+void hbs_tag(hbs_options_t so, hbu_pipe_t pipe);
 
+#include "./streamoptions.c"
 #include "./content.c"
 
-void hbs_tag(hbu_pipe_t pipe) {
+void hbs_tag(hbs_options_t so, hbu_pipe_t pipe) {
   int self_closing = 0;
 
   hbu_pipe_require(pipe, '<');
@@ -50,9 +52,12 @@ void hbs_tag(hbu_pipe_t pipe) {
   if (hbu_buffer_compare_lit(opening_name, "script") == 0) {
     // Script tag
     hbsh_script(pipe);
+  } else if (hbu_buffer_compare_lit(opening_name, "style") == 0) {
+    // Style tag
+    hbsh_style(pipe);
   } else {
     // Content
-    hbs_content(pipe);
+    hbs_content(so, pipe);
   }
 
   // Closing tag for non-void