X-Git-Url: http://andersk.mit.edu/gitweb/libyaml.git/blobdiff_plain/5a00d8fe664672493e14535cdca52d2fb69a0032..c45550e5cf265bf6877cad127faa9854851e291e:/src/yaml_private.h diff --git a/src/yaml_private.h b/src/yaml_private.h index efd1d43..eb72207 100644 --- a/src/yaml_private.h +++ b/src/yaml_private.h @@ -1,4 +1,3 @@ - #if HAVE_CONFIG_H #include #endif @@ -7,6 +6,7 @@ #include #include +#include /* * Memory management. @@ -66,6 +66,17 @@ yaml_parser_fetch_more_tokens(yaml_parser_t *parser); #define OUTPUT_RAW_BUFFER_SIZE (OUTPUT_BUFFER_SIZE*2+2) +/* + * The maximum size of a YAML input file. + * This used to be PTRDIFF_MAX, but that's not entirely portable + * because stdint.h isn't available on all platforms. + * It is not entirely clear why this isn't the maximum value + * that can fit into the parser->offset field. + */ + +#define MAX_FILE_SIZE (~(size_t)0 / 2) + + /* * The size of other stacks and queues. */ @@ -79,7 +90,7 @@ yaml_parser_fetch_more_tokens(yaml_parser_t *parser); */ #define BUFFER_INIT(context,buffer,size) \ - (((buffer).start = yaml_malloc(size)) ? \ + (((buffer).start = (yaml_char_t *)yaml_malloc(size)) ? \ ((buffer).last = (buffer).pointer = (buffer).start, \ (buffer).end = (buffer).start+(size), \ 1) : \ @@ -111,8 +122,15 @@ yaml_string_join( #define NULL_STRING { NULL, NULL, NULL } +#define STRING(string,length) { (string), (string)+(length), (string) } + +#define STRING_ASSIGN(value,string,length) \ + ((value).start = (string), \ + (value).end = (string)+(length), \ + (value).pointer = (string)) + #define STRING_INIT(context,string,size) \ - (((string).start = yaml_malloc(size)) ? \ + (((string).start = YAML_MALLOC(size)) ? \ ((string).pointer = (string).start, \ (string).end = (string).start+(size), \ memset((string).start, 0, (size)), \ @@ -125,9 +143,12 @@ yaml_string_join( (string).start = (string).pointer = (string).end = 0) #define STRING_EXTEND(context,string) \ - (((string).pointer+5 < (string).end) \ + ((((string).pointer+5 < (string).end) \ || yaml_string_extend(&(string).start, \ - &(string).pointer, &(string).end)) + &(string).pointer, &(string).end)) ? \ + 1 : \ + ((context)->error = YAML_MEMORY_ERROR, \ + 0)) #define CLEAR(context,string) \ ((string).pointer = (string).start, \ @@ -142,6 +163,253 @@ yaml_string_join( ((context)->error = YAML_MEMORY_ERROR, \ 0)) +/* + * String check operations. + */ + +/* + * Check the octet at the specified position. + */ + +#define CHECK_AT(string,octet,offset) \ + ((string).pointer[offset] == (yaml_char_t)(octet)) + +/* + * Check the current octet in the buffer. + */ + +#define CHECK(string,octet) CHECK_AT((string),(octet),0) + +/* + * Check if the character at the specified position is an alphabetical + * character, a digit, '_', or '-'. + */ + +#define IS_ALPHA_AT(string,offset) \ + (((string).pointer[offset] >= (yaml_char_t) '0' && \ + (string).pointer[offset] <= (yaml_char_t) '9') || \ + ((string).pointer[offset] >= (yaml_char_t) 'A' && \ + (string).pointer[offset] <= (yaml_char_t) 'Z') || \ + ((string).pointer[offset] >= (yaml_char_t) 'a' && \ + (string).pointer[offset] <= (yaml_char_t) 'z') || \ + (string).pointer[offset] == '_' || \ + (string).pointer[offset] == '-') + +#define IS_ALPHA(string) IS_ALPHA_AT((string),0) + +/* + * Check if the character at the specified position is a digit. + */ + +#define IS_DIGIT_AT(string,offset) \ + (((string).pointer[offset] >= (yaml_char_t) '0' && \ + (string).pointer[offset] <= (yaml_char_t) '9')) + +#define IS_DIGIT(string) IS_DIGIT_AT((string),0) + +/* + * Get the value of a digit. + */ + +#define AS_DIGIT_AT(string,offset) \ + ((string).pointer[offset] - (yaml_char_t) '0') + +#define AS_DIGIT(string) AS_DIGIT_AT((string),0) + +/* + * Check if the character at the specified position is a hex-digit. + */ + +#define IS_HEX_AT(string,offset) \ + (((string).pointer[offset] >= (yaml_char_t) '0' && \ + (string).pointer[offset] <= (yaml_char_t) '9') || \ + ((string).pointer[offset] >= (yaml_char_t) 'A' && \ + (string).pointer[offset] <= (yaml_char_t) 'F') || \ + ((string).pointer[offset] >= (yaml_char_t) 'a' && \ + (string).pointer[offset] <= (yaml_char_t) 'f')) + +#define IS_HEX(string) IS_HEX_AT((string),0) + +/* + * Get the value of a hex-digit. + */ + +#define AS_HEX_AT(string,offset) \ + (((string).pointer[offset] >= (yaml_char_t) 'A' && \ + (string).pointer[offset] <= (yaml_char_t) 'F') ? \ + ((string).pointer[offset] - (yaml_char_t) 'A' + 10) : \ + ((string).pointer[offset] >= (yaml_char_t) 'a' && \ + (string).pointer[offset] <= (yaml_char_t) 'f') ? \ + ((string).pointer[offset] - (yaml_char_t) 'a' + 10) : \ + ((string).pointer[offset] - (yaml_char_t) '0')) + +#define AS_HEX(string) AS_HEX_AT((string),0) + +/* + * Check if the character is ASCII. + */ + +#define IS_ASCII_AT(string,offset) \ + ((string).pointer[offset] <= (yaml_char_t) '\x7F') + +#define IS_ASCII(string) IS_ASCII_AT((string),0) + +/* + * Check if the character can be printed unescaped. + */ + +#define IS_PRINTABLE_AT(string,offset) \ + (((string).pointer[offset] == 0x0A) /* . == #x0A */ \ + || ((string).pointer[offset] >= 0x20 /* #x20 <= . <= #x7E */ \ + && (string).pointer[offset] <= 0x7E) \ + || ((string).pointer[offset] == 0xC2 /* #0xA0 <= . <= #xD7FF */ \ + && (string).pointer[offset+1] >= 0xA0) \ + || ((string).pointer[offset] > 0xC2 \ + && (string).pointer[offset] < 0xED) \ + || ((string).pointer[offset] == 0xED \ + && (string).pointer[offset+1] < 0xA0) \ + || ((string).pointer[offset] == 0xEE) \ + || ((string).pointer[offset] == 0xEF /* #xE000 <= . <= #xFFFD */ \ + && !((string).pointer[offset+1] == 0xBB /* && . != #xFEFF */ \ + && (string).pointer[offset+2] == 0xBF) \ + && !((string).pointer[offset+1] == 0xBF \ + && ((string).pointer[offset+2] == 0xBE \ + || (string).pointer[offset+2] == 0xBF)))) + +#define IS_PRINTABLE(string) IS_PRINTABLE_AT((string),0) + +/* + * Check if the character at the specified position is NUL. + */ + +#define IS_Z_AT(string,offset) CHECK_AT((string),'\0',(offset)) + +#define IS_Z(string) IS_Z_AT((string),0) + +/* + * Check if the character at the specified position is BOM. + */ + +#define IS_BOM_AT(string,offset) \ + (CHECK_AT((string),'\xEF',(offset)) \ + && CHECK_AT((string),'\xBB',(offset)+1) \ + && CHECK_AT((string),'\xBF',(offset)+2)) /* BOM (#xFEFF) */ + +#define IS_BOM(string) IS_BOM_AT(string,0) + +/* + * Check if the character at the specified position is space. + */ + +#define IS_SPACE_AT(string,offset) CHECK_AT((string),' ',(offset)) + +#define IS_SPACE(string) IS_SPACE_AT((string),0) + +/* + * Check if the character at the specified position is tab. + */ + +#define IS_TAB_AT(string,offset) CHECK_AT((string),'\t',(offset)) + +#define IS_TAB(string) IS_TAB_AT((string),0) + +/* + * Check if the character at the specified position is blank (space or tab). + */ + +#define IS_BLANK_AT(string,offset) \ + (IS_SPACE_AT((string),(offset)) || IS_TAB_AT((string),(offset))) + +#define IS_BLANK(string) IS_BLANK_AT((string),0) + +/* + * Check if the character at the specified position is a line break. + */ + +#define IS_BREAK_AT(string,offset) \ + (CHECK_AT((string),'\r',(offset)) /* CR (#xD)*/ \ + || CHECK_AT((string),'\n',(offset)) /* LF (#xA) */ \ + || (CHECK_AT((string),'\xC2',(offset)) \ + && CHECK_AT((string),'\x85',(offset)+1)) /* NEL (#x85) */ \ + || (CHECK_AT((string),'\xE2',(offset)) \ + && CHECK_AT((string),'\x80',(offset)+1) \ + && CHECK_AT((string),'\xA8',(offset)+2)) /* LS (#x2028) */ \ + || (CHECK_AT((string),'\xE2',(offset)) \ + && CHECK_AT((string),'\x80',(offset)+1) \ + && CHECK_AT((string),'\xA9',(offset)+2))) /* PS (#x2029) */ + +#define IS_BREAK(string) IS_BREAK_AT((string),0) + +#define IS_CRLF_AT(string,offset) \ + (CHECK_AT((string),'\r',(offset)) && CHECK_AT((string),'\n',(offset)+1)) + +#define IS_CRLF(string) IS_CRLF_AT((string),0) + +/* + * Check if the character is a line break or NUL. + */ + +#define IS_BREAKZ_AT(string,offset) \ + (IS_BREAK_AT((string),(offset)) || IS_Z_AT((string),(offset))) + +#define IS_BREAKZ(string) IS_BREAKZ_AT((string),0) + +/* + * Check if the character is a line break, space, or NUL. + */ + +#define IS_SPACEZ_AT(string,offset) \ + (IS_SPACE_AT((string),(offset)) || IS_BREAKZ_AT((string),(offset))) + +#define IS_SPACEZ(string) IS_SPACEZ_AT((string),0) + +/* + * Check if the character is a line break, space, tab, or NUL. + */ + +#define IS_BLANKZ_AT(string,offset) \ + (IS_BLANK_AT((string),(offset)) || IS_BREAKZ_AT((string),(offset))) + +#define IS_BLANKZ(string) IS_BLANKZ_AT((string),0) + +/* + * Determine the width of the character. + */ + +#define WIDTH_AT(string,offset) \ + (((string).pointer[offset] & 0x80) == 0x00 ? 1 : \ + ((string).pointer[offset] & 0xE0) == 0xC0 ? 2 : \ + ((string).pointer[offset] & 0xF0) == 0xE0 ? 3 : \ + ((string).pointer[offset] & 0xF8) == 0xF0 ? 4 : 0) + +#define WIDTH(string) WIDTH_AT((string),0) + +/* + * Move the string pointer to the next character. + */ + +#define MOVE(string) ((string).pointer += WIDTH((string))) + +/* + * Copy a character and move the pointers of both strings. + */ + +#define COPY(string_a,string_b) \ + ((*(string_b).pointer & 0x80) == 0x00 ? \ + (*((string_a).pointer++) = *((string_b).pointer++)) : \ + (*(string_b).pointer & 0xE0) == 0xC0 ? \ + (*((string_a).pointer++) = *((string_b).pointer++), \ + *((string_a).pointer++) = *((string_b).pointer++)) : \ + (*(string_b).pointer & 0xF0) == 0xE0 ? \ + (*((string_a).pointer++) = *((string_b).pointer++), \ + *((string_a).pointer++) = *((string_b).pointer++), \ + *((string_a).pointer++) = *((string_b).pointer++)) : \ + (*(string_b).pointer & 0xF8) == 0xF0 ? \ + (*((string_a).pointer++) = *((string_b).pointer++), \ + *((string_a).pointer++) = *((string_b).pointer++), \ + *((string_a).pointer++) = *((string_b).pointer++), \ + *((string_a).pointer++) = *((string_b).pointer++)) : 0) + /* * Stack and queue management. */ @@ -152,10 +420,10 @@ yaml_stack_extend(void **start, void **top, void **end); YAML_DECLARE(int) yaml_queue_extend(void **start, void **head, void **tail, void **end); -#define STACK_INIT(context,stack,size) \ - (((stack).start = yaml_malloc((size)*sizeof(*(stack).start))) ? \ +#define STACK_INIT(context,stack,type) \ + (((stack).start = (type)yaml_malloc(INITIAL_STACK_SIZE*sizeof(*(stack).start))) ? \ ((stack).top = (stack).start, \ - (stack).end = (stack).start+(size), \ + (stack).end = (stack).start+INITIAL_STACK_SIZE, \ 1) : \ ((context)->error = YAML_MEMORY_ERROR, \ 0)) @@ -167,6 +435,12 @@ yaml_queue_extend(void **start, void **head, void **tail, void **end); #define STACK_EMPTY(context,stack) \ ((stack).start == (stack).top) +#define STACK_LIMIT(context,stack,size) \ + ((stack).top - (stack).start < (size) ? \ + 1 : \ + ((context)->error = YAML_MEMORY_ERROR, \ + 0)) + #define PUSH(context,stack,value) \ (((stack).top != (stack).end \ || yaml_stack_extend((void **)&(stack).start, \ @@ -179,8 +453,8 @@ yaml_queue_extend(void **start, void **head, void **tail, void **end); #define POP(context,stack) \ (*(--(stack).top)) -#define QUEUE_INIT(context,queue,size) \ - (((queue).start = yaml_malloc((size)*sizeof(*(queue).start))) ? \ +#define QUEUE_INIT(context,queue,size,type) \ + (((queue).start = (type)yaml_malloc((size)*sizeof(*(queue).start))) ? \ ((queue).head = (queue).tail = (queue).start, \ (queue).end = (queue).start+(size), \ 1) : \ @@ -330,3 +604,81 @@ yaml_queue_extend(void **start, void **head, void **tail, void **end); #define MAPPING_END_EVENT_INIT(event,start_mark,end_mark) \ (EVENT_INIT((event),YAML_MAPPING_END_EVENT,(start_mark),(end_mark))) +/* + * Document initializer. + */ + +#define DOCUMENT_INIT(document,document_nodes_start,document_nodes_end, \ + document_version_directive,document_tag_directives_start, \ + document_tag_directives_end,document_start_implicit, \ + document_end_implicit,document_start_mark,document_end_mark) \ + (memset(&(document), 0, sizeof(yaml_document_t)), \ + (document).nodes.start = (document_nodes_start), \ + (document).nodes.end = (document_nodes_end), \ + (document).nodes.top = (document_nodes_start), \ + (document).version_directive = (document_version_directive), \ + (document).tag_directives.start = (document_tag_directives_start), \ + (document).tag_directives.end = (document_tag_directives_end), \ + (document).start_implicit = (document_start_implicit), \ + (document).end_implicit = (document_end_implicit), \ + (document).start_mark = (document_start_mark), \ + (document).end_mark = (document_end_mark)) + +/* + * Node initializers. + */ + +#define NODE_INIT(node,node_type,node_tag,node_start_mark,node_end_mark) \ + (memset(&(node), 0, sizeof(yaml_node_t)), \ + (node).type = (node_type), \ + (node).tag = (node_tag), \ + (node).start_mark = (node_start_mark), \ + (node).end_mark = (node_end_mark)) + +#define SCALAR_NODE_INIT(node,node_tag,node_value,node_length, \ + node_style,start_mark,end_mark) \ + (NODE_INIT((node),YAML_SCALAR_NODE,(node_tag),(start_mark),(end_mark)), \ + (node).data.scalar.value = (node_value), \ + (node).data.scalar.length = (node_length), \ + (node).data.scalar.style = (node_style)) + +#define SEQUENCE_NODE_INIT(node,node_tag,node_items_start,node_items_end, \ + node_style,start_mark,end_mark) \ + (NODE_INIT((node),YAML_SEQUENCE_NODE,(node_tag),(start_mark),(end_mark)), \ + (node).data.sequence.items.start = (node_items_start), \ + (node).data.sequence.items.end = (node_items_end), \ + (node).data.sequence.items.top = (node_items_start), \ + (node).data.sequence.style = (node_style)) + +#define MAPPING_NODE_INIT(node,node_tag,node_pairs_start,node_pairs_end, \ + node_style,start_mark,end_mark) \ + (NODE_INIT((node),YAML_MAPPING_NODE,(node_tag),(start_mark),(end_mark)), \ + (node).data.mapping.pairs.start = (node_pairs_start), \ + (node).data.mapping.pairs.end = (node_pairs_end), \ + (node).data.mapping.pairs.top = (node_pairs_start), \ + (node).data.mapping.style = (node_style)) + +/* Strict C compiler warning helpers */ + +#if defined(__clang__) || defined(__GNUC__) +# define HASATTRIBUTE_UNUSED +#endif +#ifdef HASATTRIBUTE_UNUSED +# define __attribute__unused__ __attribute__((__unused__)) +#else +# define __attribute__unused__ +#endif + +/* Shim arguments are arguments that must be included in your function, + * but serve no purpose inside. Silence compiler warnings. */ +#define SHIM(a) /*@unused@*/ a __attribute__unused__ + +/* UNUSED_PARAM() marks a shim argument in the body to silence compiler warnings */ +#ifdef __clang__ +# define UNUSED_PARAM(a) (void)(a); +#else +# define UNUSED_PARAM(a) /*@-noeffect*/if (0) (void)(a)/*@=noeffect*/; +#endif + +#define YAML_MALLOC_STATIC(type) (type*)yaml_malloc(sizeof(type)) +#define YAML_MALLOC(size) (yaml_char_t *)yaml_malloc(size)