nokogiri 1.10.9 → 1.18.3

data/gumbo-parser/src/error.h

@@ -0,0 +1,152 @@
+#ifndef GUMBO_ERROR_H_
+#define GUMBO_ERROR_H_
+
+#include <stdint.h>
+
+#include "nokogiri_gumbo.h"
+#include "insertion_mode.h"
+#include "string_buffer.h"
+#include "token_type.h"
+#include "tokenizer_states.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct GumboInternalParser;
+
+typedef enum {
+  // Defined errors.
+  // https://html.spec.whatwg.org/multipage/parsing.html#parse-errors
+  GUMBO_ERR_ABRUPT_CLOSING_OF_EMPTY_COMMENT,
+  GUMBO_ERR_ABRUPT_DOCTYPE_PUBLIC_IDENTIFIER,
+  GUMBO_ERR_ABRUPT_DOCTYPE_SYSTEM_IDENTIFIER,
+  GUMBO_ERR_ABSENCE_OF_DIGITS_IN_NUMERIC_CHARACTER_REFERENCE,
+  GUMBO_ERR_CDATA_IN_HTML_CONTENT,
+  GUMBO_ERR_CHARACTER_REFERENCE_OUTSIDE_UNICODE_RANGE,
+  GUMBO_ERR_CONTROL_CHARACTER_IN_INPUT_STREAM,
+  GUMBO_ERR_CONTROL_CHARACTER_REFERENCE,
+  GUMBO_ERR_END_TAG_WITH_ATTRIBUTES,
+  GUMBO_ERR_DUPLICATE_ATTRIBUTE,
+  GUMBO_ERR_END_TAG_WITH_TRAILING_SOLIDUS,
+  GUMBO_ERR_EOF_BEFORE_TAG_NAME,
+  GUMBO_ERR_EOF_IN_CDATA,
+  GUMBO_ERR_EOF_IN_COMMENT,
+  GUMBO_ERR_EOF_IN_DOCTYPE,
+  GUMBO_ERR_EOF_IN_SCRIPT_HTML_COMMENT_LIKE_TEXT,
+  GUMBO_ERR_EOF_IN_TAG,
+  GUMBO_ERR_INCORRECTLY_CLOSED_COMMENT,
+  GUMBO_ERR_INCORRECTLY_OPENED_COMMENT,
+  GUMBO_ERR_INVALID_CHARACTER_SEQUENCE_AFTER_DOCTYPE_NAME,
+  GUMBO_ERR_INVALID_FIRST_CHARACTER_OF_TAG_NAME,
+  GUMBO_ERR_MISSING_ATTRIBUTE_VALUE,
+  GUMBO_ERR_MISSING_DOCTYPE_NAME,
+  GUMBO_ERR_MISSING_DOCTYPE_PUBLIC_IDENTIFIER,
+  GUMBO_ERR_MISSING_DOCTYPE_SYSTEM_IDENTIFIER,
+  GUMBO_ERR_MISSING_END_TAG_NAME,
+  GUMBO_ERR_MISSING_QUOTE_BEFORE_DOCTYPE_PUBLIC_IDENTIFIER,
+  GUMBO_ERR_MISSING_QUOTE_BEFORE_DOCTYPE_SYSTEM_IDENTIFIER,
+  GUMBO_ERR_MISSING_SEMICOLON_AFTER_CHARACTER_REFERENCE,
+  GUMBO_ERR_MISSING_WHITESPACE_AFTER_DOCTYPE_PUBLIC_KEYWORD,
+  GUMBO_ERR_MISSING_WHITESPACE_AFTER_DOCTYPE_SYSTEM_KEYWORD,
+  GUMBO_ERR_MISSING_WHITESPACE_BEFORE_DOCTYPE_NAME,
+  GUMBO_ERR_MISSING_WHITESPACE_BETWEEN_ATTRIBUTES,
+  GUMBO_ERR_MISSING_WHITESPACE_BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS,
+  GUMBO_ERR_NESTED_COMMENT,
+  GUMBO_ERR_NONCHARACTER_CHARACTER_REFERENCE,
+  GUMBO_ERR_NONCHARACTER_IN_INPUT_STREAM,
+  GUMBO_ERR_NON_VOID_HTML_ELEMENT_START_TAG_WITH_TRAILING_SOLIDUS,
+  GUMBO_ERR_NULL_CHARACTER_REFERENCE,
+  GUMBO_ERR_SURROGATE_CHARACTER_REFERENCE,
+  GUMBO_ERR_SURROGATE_IN_INPUT_STREAM,
+  GUMBO_ERR_UNEXPECTED_CHARACTER_AFTER_DOCTYPE_SYSTEM_IDENTIFIER,
+  GUMBO_ERR_UNEXPECTED_CHARACTER_IN_ATTRIBUTE_NAME,
+  GUMBO_ERR_UNEXPECTED_CHARACTER_IN_UNQUOTED_ATTRIBUTE_VALUE,
+  GUMBO_ERR_UNEXPECTED_EQUALS_SIGN_BEFORE_ATTRIBUTE_NAME,
+  GUMBO_ERR_UNEXPECTED_NULL_CHARACTER,
+  GUMBO_ERR_UNEXPECTED_QUESTION_MARK_INSTEAD_OF_TAG_NAME,
+  GUMBO_ERR_UNEXPECTED_SOLIDUS_IN_TAG,
+  GUMBO_ERR_UNKNOWN_NAMED_CHARACTER_REFERENCE,
+
+  // Encoding errors.
+  GUMBO_ERR_UTF8_INVALID,
+  GUMBO_ERR_UTF8_TRUNCATED,
+
+  // Generic parser error.
+  GUMBO_ERR_PARSER,
+} GumboErrorType;
+
+// Additional data for tokenizer errors.
+// This records the current state and codepoint encountered - this is usually
+// enough to reconstruct what went wrong and provide a friendly error message.
+typedef struct GumboInternalTokenizerError {
+  // The bad codepoint encountered.
+  int codepoint;
+
+  // The state that the tokenizer was in at the time.
+  GumboTokenizerEnum state;
+} GumboTokenizerError;
+
+// Additional data for parse errors.
+typedef struct GumboInternalParserError {
+  // The type of input token that resulted in this error.
+  GumboTokenType input_type;
+
+  // The HTML tag of the input token. TAG_UNKNOWN if this was not a tag token.
+  GumboTag input_tag;
+
+  // The HTML tag of the input token if it was nonstandard tag token. NULL otherwise.
+  char *input_name;
+
+  // The insertion mode that the parser was in at the time.
+  GumboInsertionMode parser_state;
+
+  // The tag stack at the point of the error. Note that this is an GumboVector
+  // of GumboTag's *stored by value* - cast the void* to an GumboTag directly to
+  // get at the tag. For nonstandard tags, this is a pointer to an owned char *
+  // containing the tag name.
+  GumboVector /* GumboTag */ tag_stack;
+} GumboParserError;
+
+// The overall error struct representing an error in decoding/tokenizing/parsing
+// the HTML. This contains an enumerated type flag, a source position, and then
+// a union of fields containing data specific to the error.
+struct GumboInternalError {
+  // The type of error.
+  GumboErrorType type;
+
+  // The position within the source file where the error occurred.
+  GumboSourcePosition position;
+
+  // The piece of text that caused the error.
+  GumboStringPiece original_text;
+
+  // Type-specific error information.
+  union {
+    // Tokenizer errors.
+    GumboTokenizerError tokenizer;
+
+    // Parser errors.
+    GumboParserError parser;
+  } v;
+};
+
+// Adds a new error to the parser's error list, and returns a pointer to it so
+// that clients can fill out the rest of its fields. May return NULL if we're
+// already over the max_errors field specified in GumboOptions.
+GumboError* gumbo_add_error(struct GumboInternalParser* parser);
+
+// Initializes the errors vector in the parser.
+void gumbo_init_errors(struct GumboInternalParser* errors);
+
+// Frees all the errors in the 'errors_' field of the parser.
+void gumbo_destroy_errors(struct GumboInternalParser* errors);
+
+// Frees the memory used for a single GumboError.
+void gumbo_error_destroy(GumboError* error);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif // GUMBO_ERROR_H_

data/gumbo-parser/src/foreign_attrs.c

@@ -0,0 +1,103 @@
+/* ANSI-C code produced by gperf version 3.1 */
+/* Command-line: gperf -m100 -n src/foreign_attrs.gperf  */
+/* Computed positions: -k'8-9' */
+/* Filtered by: gperf-filter.sed */
+
+#include "replacement.h"
+#include "macros.h"
+#include <string.h>
+
+#define TOTAL_KEYWORDS 11
+#define MIN_WORD_LENGTH 5
+#define MAX_WORD_LENGTH 13
+#define MIN_HASH_VALUE 0
+#define MAX_HASH_VALUE 10
+/* maximum key range = 11, duplicates = 0 */
+
+static inline unsigned int
+hash (register const char *str, register size_t len)
+{
+  static const unsigned char asso_values[] =
+    {
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11,  2,
+      11,  1, 11, 10,  4,  4, 11, 11,  3, 11,
+      11,  5,  3, 11,  0, 11,  2, 11, 11, 11,
+      11,  2, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+      11, 11, 11, 11, 11, 11
+    };
+  register unsigned int hval = 0;
+
+  switch (len)
+    {
+      default:
+        hval += asso_values[(unsigned char)str[8]];
+      /*FALLTHROUGH*/
+      case 8:
+        hval += asso_values[(unsigned char)str[7]];
+      /*FALLTHROUGH*/
+      case 7:
+      case 6:
+      case 5:
+        break;
+    }
+  return hval;
+}
+
+const ForeignAttrReplacement *
+gumbo_get_foreign_attr_replacement (register const char *str, register size_t len)
+{
+  static const unsigned char lengthtable[] =
+    {
+       5, 10, 13,  9, 13, 10, 11, 11, 10, 10,  8
+    };
+  static const ForeignAttrReplacement wordlist[] =
+    {
+      {"xmlns", "xmlns", GUMBO_ATTR_NAMESPACE_XMLNS},
+      {"xlink:href", "href", GUMBO_ATTR_NAMESPACE_XLINK},
+      {"xlink:arcrole", "arcrole", GUMBO_ATTR_NAMESPACE_XLINK},
+      {"xml:space", "space", GUMBO_ATTR_NAMESPACE_XML},
+      {"xlink:actuate", "actuate", GUMBO_ATTR_NAMESPACE_XLINK},
+      {"xlink:type", "type", GUMBO_ATTR_NAMESPACE_XLINK},
+      {"xlink:title", "title", GUMBO_ATTR_NAMESPACE_XLINK},
+      {"xmlns:xlink", "xlink", GUMBO_ATTR_NAMESPACE_XMLNS},
+      {"xlink:role", "role", GUMBO_ATTR_NAMESPACE_XLINK},
+      {"xlink:show", "show", GUMBO_ATTR_NAMESPACE_XLINK},
+      {"xml:lang", "lang", GUMBO_ATTR_NAMESPACE_XML}
+    };
+
+  if (len <= MAX_WORD_LENGTH && len >= MIN_WORD_LENGTH)
+    {
+      register unsigned int key = hash (str, len);
+
+      if (key <= MAX_HASH_VALUE)
+        if (len == lengthtable[key])
+          {
+            register const char *s = wordlist[key].from;
+
+            if (s && *str == *s && !memcmp (str + 1, s + 1, len - 1))
+              return &wordlist[key];
+          }
+    }
+  return 0;
+}

data/gumbo-parser/src/foreign_attrs.gperf

@@ -0,0 +1,27 @@
+%{
+#include "replacement.h"
+#include "macros.h"
+%}
+
+%struct-type
+%omit-struct-type
+%compare-lengths
+%readonly-tables
+%null-strings
+%includes
+%define lookup-function-name gumbo_get_foreign_attr_replacement
+%define slot-name from
+ForeignAttrReplacement;
+
+%%
+"xlink:actuate", "actuate", GUMBO_ATTR_NAMESPACE_XLINK
+"xlink:arcrole", "arcrole", GUMBO_ATTR_NAMESPACE_XLINK
+"xlink:href", "href", GUMBO_ATTR_NAMESPACE_XLINK
+"xlink:role", "role", GUMBO_ATTR_NAMESPACE_XLINK
+"xlink:show", "show", GUMBO_ATTR_NAMESPACE_XLINK
+"xlink:title", "title", GUMBO_ATTR_NAMESPACE_XLINK
+"xlink:type", "type", GUMBO_ATTR_NAMESPACE_XLINK
+"xml:lang", "lang", GUMBO_ATTR_NAMESPACE_XML
+"xml:space", "space", GUMBO_ATTR_NAMESPACE_XML
+"xmlns", "xmlns", GUMBO_ATTR_NAMESPACE_XMLNS
+"xmlns:xlink", "xlink", GUMBO_ATTR_NAMESPACE_XMLNS

data/gumbo-parser/src/insertion_mode.h

@@ -0,0 +1,33 @@
+#ifndef GUMBO_INSERTION_MODE_H_
+#define GUMBO_INSERTION_MODE_H_
+
+// https://html.spec.whatwg.org/multipage/parsing.html#insertion-mode
+// If new enum values are added, be sure to update the kTokenHandlers
+// dispatch table in parser.c.
+typedef enum {
+  GUMBO_INSERTION_MODE_INITIAL,
+  GUMBO_INSERTION_MODE_BEFORE_HTML,
+  GUMBO_INSERTION_MODE_BEFORE_HEAD,
+  GUMBO_INSERTION_MODE_IN_HEAD,
+  GUMBO_INSERTION_MODE_IN_HEAD_NOSCRIPT,
+  GUMBO_INSERTION_MODE_AFTER_HEAD,
+  GUMBO_INSERTION_MODE_IN_BODY,
+  GUMBO_INSERTION_MODE_TEXT,
+  GUMBO_INSERTION_MODE_IN_TABLE,
+  GUMBO_INSERTION_MODE_IN_TABLE_TEXT,
+  GUMBO_INSERTION_MODE_IN_CAPTION,
+  GUMBO_INSERTION_MODE_IN_COLUMN_GROUP,
+  GUMBO_INSERTION_MODE_IN_TABLE_BODY,
+  GUMBO_INSERTION_MODE_IN_ROW,
+  GUMBO_INSERTION_MODE_IN_CELL,
+  GUMBO_INSERTION_MODE_IN_SELECT,
+  GUMBO_INSERTION_MODE_IN_SELECT_IN_TABLE,
+  GUMBO_INSERTION_MODE_IN_TEMPLATE,
+  GUMBO_INSERTION_MODE_AFTER_BODY,
+  GUMBO_INSERTION_MODE_IN_FRAMESET,
+  GUMBO_INSERTION_MODE_AFTER_FRAMESET,
+  GUMBO_INSERTION_MODE_AFTER_AFTER_BODY,
+  GUMBO_INSERTION_MODE_AFTER_AFTER_FRAMESET
+} GumboInsertionMode;
+
+#endif // GUMBO_INSERTION_MODE_H_

data/gumbo-parser/src/macros.h

@@ -0,0 +1,91 @@
+#ifndef MACROS_H
+#define MACROS_H
+
+#if (!defined(__STDC_VERSION__) || !(__STDC_VERSION__ >= 199901L)) \
+    && !defined(_WIN32) && !defined(__cplusplus)
+# error C99 compiler required
+#endif
+
+#if defined(_WIN32)
+# define inline __inline
+# define __func__ __FUNCTION__
+#endif
+
+// Calculate the number of elements in an array.
+// The extra division on the third line is a trick to help prevent
+// passing a pointer to the first element of an array instead of a
+// reference to the array itself.
+#define ARRAY_COUNT(x) ( \
+    (sizeof(x) / sizeof((x)[0])) \
+    / ((size_t)(!(sizeof(x) % sizeof((x)[0])))) \
+)
+
+#ifdef NDEBUG
+    #define UNUSED_IF_NDEBUG(x) (void)(x)
+#else
+    #define UNUSED_IF_NDEBUG(x)
+#endif
+
+#ifdef __GNUC__
+    #define GNUC_AT_LEAST(major, minor) ( \
+        (__GNUC__ > major) \
+        || ((__GNUC__ == major) && (__GNUC_MINOR__ >= minor)) )
+#else
+    #define GNUC_AT_LEAST(major, minor) 0
+#endif
+
+#ifdef __has_attribute
+    #define HAS_ATTRIBUTE(x) __has_attribute(x)
+#else
+    #define HAS_ATTRIBUTE(x) 0
+#endif
+
+#if GNUC_AT_LEAST(3, 0) || HAS_ATTRIBUTE(unused) || defined(__TINYC__)
+    #define UNUSED __attribute__((__unused__))
+#else
+    #define UNUSED
+#endif
+
+#if GNUC_AT_LEAST(3, 0)
+    #define MALLOC __attribute__((__malloc__))
+    #define PRINTF(x) __attribute__((__format__(__printf__, (x), (x + 1))))
+    #define PURE __attribute__((__pure__))
+    #define CONST_FN __attribute__((__const__))
+#else
+    #define MALLOC
+    #define PRINTF(x)
+    #define PURE
+    #define CONST_FN
+#endif
+
+#define UNUSED_ARG(x) unused__ ## x UNUSED
+
+#if GNUC_AT_LEAST(3, 0) && defined(__OPTIMIZE__)
+    #define likely(x) __builtin_expect(!!(x), 1)
+    #define unlikely(x) __builtin_expect(!!(x), 0)
+#else
+    #define likely(x) (x)
+    #define unlikely(x) (x)
+#endif
+
+#if GNUC_AT_LEAST(3, 3) || HAS_ATTRIBUTE(nonnull)
+    #define NONNULL_ARGS __attribute__((__nonnull__))
+#else
+    #define NONNULL_ARGS
+#endif
+
+#if GNUC_AT_LEAST(3, 4) || HAS_ATTRIBUTE(warn_unused_result)
+    #define WARN_UNUSED_RESULT __attribute__((__warn_unused_result__))
+#else
+    #define WARN_UNUSED_RESULT
+#endif
+
+#if GNUC_AT_LEAST(5, 0) || HAS_ATTRIBUTE(returns_nonnull)
+    #define RETURNS_NONNULL __attribute__((__returns_nonnull__))
+#else
+    #define RETURNS_NONNULL
+#endif
+
+#define XMALLOC MALLOC RETURNS_NONNULL
+
+#endif // ndef MACROS_H