| 1 | #include "cmark.h" |
| 2 | #include "utf8.h" |
| 3 | #include "parser.h" |
| 4 | #include "references.h" |
| 5 | #include "inlines.h" |
| 6 | #include "chunk.h" |
| 7 | |
| 8 | static void reference_free(cmark_reference_map *map, cmark_reference *ref) { |
| 9 | cmark_mem *mem = map->mem; |
| 10 | if (ref != NULL) { |
| 11 | mem->free(ref->label); |
| 12 | mem->free(ref->url); |
| 13 | mem->free(ref->title); |
| 14 | mem->free(ref); |
| 15 | } |
| 16 | } |
| 17 | |
| 18 | // normalize reference: collapse internal whitespace to single space, |
| 19 | // remove leading/trailing whitespace, case fold |
| 20 | // Return NULL if the reference name is actually empty (i.e. composed |
| 21 | // solely from whitespace) |
| 22 | static unsigned char *normalize_reference(cmark_mem *mem, cmark_chunk *ref) { |
| 23 | cmark_strbuf normalized = CMARK_BUF_INIT(mem); |
| 24 | unsigned char *result; |
| 25 | |
| 26 | if (ref == NULL) |
| 27 | return NULL; |
| 28 | |
| 29 | if (ref->len == 0) |
| 30 | return NULL; |
| 31 | |
| 32 | cmark_utf8proc_case_fold(&normalized, ref->data, ref->len); |
| 33 | cmark_strbuf_trim(&normalized); |
| 34 | cmark_strbuf_normalize_whitespace(&normalized); |
| 35 | |
| 36 | result = cmark_strbuf_detach(&normalized); |
| 37 | assert(result); |
| 38 | |
| 39 | if (result[0] == '\0') { |
| 40 | mem->free(result); |
| 41 | return NULL; |
| 42 | } |
| 43 | |
| 44 | return result; |
| 45 | } |
| 46 | |
| 47 | void cmark_reference_create(cmark_reference_map *map, cmark_chunk *label, |
| 48 | cmark_chunk *url, cmark_chunk *title) { |
| 49 | cmark_reference *ref; |
| 50 | unsigned char *reflabel = normalize_reference(map->mem, label); |
| 51 | |
| 52 | /* empty reference name, or composed from only whitespace */ |
| 53 | if (reflabel == NULL) |
| 54 | return; |
| 55 | |
| 56 | assert(map->sorted == NULL); |
| 57 | |
| 58 | ref = (cmark_reference *)map->mem->calloc(1, sizeof(*ref)); |
| 59 | ref->label = reflabel; |
| 60 | ref->url = cmark_clean_url(map->mem, url); |
| 61 | ref->title = cmark_clean_title(map->mem, title); |
| 62 | ref->age = map->size; |
| 63 | ref->next = map->refs; |
| 64 | |
| 65 | if (ref->url != NULL) |
| 66 | ref->size += strlen((char*)ref->url); |
| 67 | if (ref->title != NULL) |
| 68 | ref->size += strlen((char*)ref->title); |
| 69 | |
| 70 | map->refs = ref; |
| 71 | map->size++; |
| 72 | } |
| 73 | |
| 74 | static int |
| 75 | labelcmp(const unsigned char *a, const unsigned char *b) { |
| 76 | return strcmp((const char *)a, (const char *)b); |
| 77 | } |
| 78 | |
| 79 | static int |
| 80 | refcmp(const void *p1, const void *p2) { |
| 81 | cmark_reference *r1 = *(cmark_reference **)p1; |
| 82 | cmark_reference *r2 = *(cmark_reference **)p2; |
| 83 | int res = labelcmp(r1->label, r2->label); |
| 84 | return res ? res : ((int)r1->age - (int)r2->age); |
| 85 | } |
| 86 | |
| 87 | static int |
| 88 | refsearch(const void *label, const void *p2) { |
| 89 | cmark_reference *ref = *(cmark_reference **)p2; |
| 90 | return labelcmp((const unsigned char *)label, ref->label); |
| 91 | } |
| 92 | |
| 93 | static void sort_references(cmark_reference_map *map) { |
| 94 | unsigned int i = 0, last = 0, size = map->size; |
| 95 | cmark_reference *r = map->refs, **sorted = NULL; |
| 96 | |
| 97 | sorted = (cmark_reference **)map->mem->calloc(size, sizeof(cmark_reference *)); |
| 98 | while (r) { |
| 99 | sorted[i++] = r; |
| 100 | r = r->next; |
| 101 | } |
| 102 | |
| 103 | qsort(sorted, size, sizeof(cmark_reference *), refcmp); |
| 104 | |
| 105 | for (i = 1; i < size; i++) { |
| 106 | if (labelcmp(sorted[i]->label, sorted[last]->label) != 0) |
| 107 | sorted[++last] = sorted[i]; |
| 108 | } |
| 109 | map->sorted = sorted; |
| 110 | map->size = last + 1; |
| 111 | } |
| 112 | |
| 113 | // Returns reference if refmap contains a reference with matching |
| 114 | // label, otherwise NULL. |
| 115 | cmark_reference *cmark_reference_lookup(cmark_reference_map *map, |
| 116 | cmark_chunk *label) { |
| 117 | cmark_reference **ref = NULL; |
| 118 | cmark_reference *r = NULL; |
| 119 | unsigned char *norm; |
| 120 | |
| 121 | if (label->len < 1 || label->len > MAX_LINK_LABEL_LENGTH) |
| 122 | return NULL; |
| 123 | |
| 124 | if (map == NULL || !map->size) |
| 125 | return NULL; |
| 126 | |
| 127 | norm = normalize_reference(map->mem, label); |
| 128 | if (norm == NULL) |
| 129 | return NULL; |
| 130 | |
| 131 | if (!map->sorted) |
| 132 | sort_references(map); |
| 133 | |
| 134 | ref = (cmark_reference **)bsearch(norm, map->sorted, map->size, sizeof(cmark_reference *), |
| 135 | refsearch); |
| 136 | map->mem->free(norm); |
| 137 | |
| 138 | if (ref != NULL) { |
| 139 | r = ref[0]; |
| 140 | /* Check for expansion limit */ |
| 141 | if (map->max_ref_size && r->size > map->max_ref_size - map->ref_size) |
| 142 | return NULL; |
| 143 | map->ref_size += r->size; |
| 144 | } |
| 145 | |
| 146 | return r; |
| 147 | } |
| 148 | |
| 149 | void cmark_reference_map_free(cmark_reference_map *map) { |
| 150 | cmark_reference *ref; |
| 151 | |
| 152 | if (map == NULL) |
| 153 | return; |
| 154 | |
| 155 | ref = map->refs; |
| 156 | while (ref) { |
| 157 | cmark_reference *next = ref->next; |
| 158 | reference_free(map, ref); |
| 159 | ref = next; |
| 160 | } |
| 161 | |
| 162 | map->mem->free(map->sorted); |
| 163 | map->mem->free(map); |
| 164 | } |
| 165 | |
| 166 | cmark_reference_map *cmark_reference_map_new(cmark_mem *mem) { |
| 167 | cmark_reference_map *map = |
| 168 | (cmark_reference_map *)mem->calloc(1, sizeof(cmark_reference_map)); |
| 169 | map->mem = mem; |
| 170 | return map; |
| 171 | } |
| 172 | |