| 1 | /* |
| 2 | * keywords.c: keep track of all cross-reference keywords |
| 3 | */ |
| 4 | |
| 5 | #include <stdio.h> |
| 6 | #include <stdlib.h> |
| 7 | #include <assert.h> |
| 8 | #include "halibut.h" |
| 9 | |
| 10 | static int kwcmp(void *av, void *bv) |
| 11 | { |
| 12 | const keyword *a = (const keyword *)av; |
| 13 | const keyword *b = (const keyword *)bv; |
| 14 | return ustrcmp(a->key, b->key); |
| 15 | } |
| 16 | |
| 17 | static int kwfind(void *av, void *bv) |
| 18 | { |
| 19 | wchar_t *a = (wchar_t *)av; |
| 20 | const keyword *b = (const keyword *)bv; |
| 21 | return ustrcmp(a, b->key); |
| 22 | } |
| 23 | |
| 24 | keyword *kw_lookup(keywordlist *kl, wchar_t *str) { |
| 25 | return find234(kl->keys, str, kwfind); |
| 26 | } |
| 27 | |
| 28 | /* |
| 29 | * This function reads through source form and collects the |
| 30 | * keywords. They get collected in a heap, sorted by Unicode |
| 31 | * collation, last at the top (so that we can Heapsort them when we |
| 32 | * finish). |
| 33 | */ |
| 34 | keywordlist *get_keywords(paragraph *source) { |
| 35 | int errors = FALSE; |
| 36 | keywordlist *kl = snew(keywordlist); |
| 37 | numberstate *n = number_init(); |
| 38 | int prevpara = para_NotParaType; |
| 39 | |
| 40 | number_cfg(n, source); |
| 41 | |
| 42 | kl->size = 0; |
| 43 | kl->keys = newtree234(kwcmp); |
| 44 | kl->nlooseends = kl->looseendssize = 0; |
| 45 | kl->looseends = NULL; |
| 46 | for (; source; source = source->next) { |
| 47 | wchar_t *p, *q; |
| 48 | p = q = source->keyword; |
| 49 | |
| 50 | /* |
| 51 | * Look for the section type override (`example', |
| 52 | * `question' or whatever - to replace `chapter' or |
| 53 | * `section' on a per-section basis). |
| 54 | */ |
| 55 | if (q) { |
| 56 | q = uadv(q); /* point q at the word beyond */ |
| 57 | if (!*q) q = NULL; |
| 58 | } |
| 59 | |
| 60 | /* |
| 61 | * Number the chapter / section / list-item / whatever. |
| 62 | * This also sets up the `parent', `child' and `sibling' |
| 63 | * links. |
| 64 | */ |
| 65 | source->kwtext = number_mktext(n, source, q, &prevpara, &errors); |
| 66 | |
| 67 | if (p && *p) { |
| 68 | if (source->kwtext || source->type == para_Biblio) { |
| 69 | keyword *kw, *ret; |
| 70 | |
| 71 | kw = snew(keyword); |
| 72 | kw->key = p; |
| 73 | kw->text = source->kwtext; |
| 74 | kw->para = source; |
| 75 | ret = add234(kl->keys, kw); |
| 76 | if (ret != kw) { |
| 77 | error(err_multikw, &source->fpos, &ret->para->fpos, p); |
| 78 | sfree(kw); |
| 79 | /* FIXME: what happens to kw->text? Does it leak? */ |
| 80 | } |
| 81 | } |
| 82 | } else { |
| 83 | if (kl->nlooseends >= kl->looseendssize) { |
| 84 | kl->looseendssize = kl->nlooseends + 32; |
| 85 | kl->looseends = sresize(kl->looseends, kl->looseendssize, |
| 86 | word *); |
| 87 | } |
| 88 | kl->looseends[kl->nlooseends++] = source->kwtext; |
| 89 | } |
| 90 | } |
| 91 | |
| 92 | number_free(n); |
| 93 | |
| 94 | if (errors) { |
| 95 | free_keywords(kl); |
| 96 | return NULL; |
| 97 | } |
| 98 | |
| 99 | return kl; |
| 100 | } |
| 101 | |
| 102 | void free_keywords(keywordlist *kl) { |
| 103 | keyword *kw; |
| 104 | while (kl->nlooseends) |
| 105 | free_word_list(kl->looseends[--kl->nlooseends]); |
| 106 | sfree(kl->looseends); |
| 107 | while ( (kw = index234(kl->keys, 0)) != NULL) { |
| 108 | delpos234(kl->keys, 0); |
| 109 | free_word_list(kw->text); |
| 110 | sfree(kw); |
| 111 | } |
| 112 | freetree234(kl->keys); |
| 113 | sfree(kl); |
| 114 | } |
| 115 | |
| 116 | void subst_keywords(paragraph *source, keywordlist *kl) { |
| 117 | for (; source; source = source->next) { |
| 118 | word *ptr; |
| 119 | for (ptr = source->words; ptr; ptr = ptr->next) { |
| 120 | if (ptr->type == word_UpperXref || |
| 121 | ptr->type == word_LowerXref) { |
| 122 | keyword *kw; |
| 123 | word **endptr, *close, *subst; |
| 124 | |
| 125 | kw = kw_lookup(kl, ptr->text); |
| 126 | if (!kw) { |
| 127 | error(err_nosuchkw, &ptr->fpos, ptr->text); |
| 128 | subst = NULL; |
| 129 | } else |
| 130 | subst = dup_word_list(kw->text); |
| 131 | |
| 132 | if (subst && ptr->type == word_LowerXref && |
| 133 | kw->para->type != para_Biblio && |
| 134 | kw->para->type != para_BiblioCited) |
| 135 | ustrlow(subst->text); |
| 136 | |
| 137 | close = snew(word); |
| 138 | close->text = NULL; |
| 139 | close->alt = NULL; |
| 140 | close->type = word_XrefEnd; |
| 141 | close->fpos = ptr->fpos; |
| 142 | |
| 143 | close->next = ptr->next; |
| 144 | ptr->next = subst; |
| 145 | |
| 146 | for (endptr = &ptr->next; *endptr; endptr = &(*endptr)->next) |
| 147 | (*endptr)->fpos = ptr->fpos; |
| 148 | |
| 149 | *endptr = close; |
| 150 | ptr = close; |
| 151 | } |
| 152 | } |
| 153 | } |
| 154 | } |