1*2b14f697Sschwarze /* $OpenBSD: tag.c,v 1.38 2023/11/24 04:48:02 schwarze Exp $ */
2c0a657b3Sschwarze /*
3*2b14f697Sschwarze * Copyright (c) 2015, 2016, 2018, 2019, 2020, 2022, 2023
47fbd65bbSschwarze * Ingo Schwarze <schwarze@openbsd.org>
5c0a657b3Sschwarze *
6c0a657b3Sschwarze * Permission to use, copy, modify, and distribute this software for any
7c0a657b3Sschwarze * purpose with or without fee is hereby granted, provided that the above
8c0a657b3Sschwarze * copyright notice and this permission notice appear in all copies.
9c0a657b3Sschwarze *
10c0a657b3Sschwarze * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11c0a657b3Sschwarze * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12c0a657b3Sschwarze * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13c0a657b3Sschwarze * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14c0a657b3Sschwarze * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15c0a657b3Sschwarze * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16c0a657b3Sschwarze * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
170ac7e6ecSschwarze *
180ac7e6ecSschwarze * Functions to tag syntax tree nodes.
190ac7e6ecSschwarze * For internal use by mandoc(1) validation modules only.
20c0a657b3Sschwarze */
21c0a657b3Sschwarze #include <sys/types.h>
22c0a657b3Sschwarze
237d109111Sschwarze #include <assert.h>
243af42e7fSschwarze #include <limits.h>
25c0a657b3Sschwarze #include <stddef.h>
2624807e10Sschwarze #include <stdint.h>
27*2b14f697Sschwarze #include <stdio.h>
28c0a657b3Sschwarze #include <stdlib.h>
29c0a657b3Sschwarze #include <string.h>
30c0a657b3Sschwarze
31c0a657b3Sschwarze #include "mandoc_aux.h"
32c4b66caeSschwarze #include "mandoc_ohash.h"
33*2b14f697Sschwarze #include "mandoc.h"
340ac7e6ecSschwarze #include "roff.h"
359226948cSschwarze #include "mdoc.h"
366e2a0df9Sschwarze #include "roff_int.h"
37c0a657b3Sschwarze #include "tag.h"
38c0a657b3Sschwarze
39c0a657b3Sschwarze struct tag_entry {
400ac7e6ecSschwarze struct roff_node **nodes;
410ac7e6ecSschwarze size_t maxnodes;
420ac7e6ecSschwarze size_t nnodes;
4362615e18Sschwarze int prio;
44c0a657b3Sschwarze char s[];
45c0a657b3Sschwarze };
46c0a657b3Sschwarze
47a875cd0eSschwarze static void tag_move_href(struct roff_man *,
48a875cd0eSschwarze struct roff_node *, const char *);
499226948cSschwarze static void tag_move_id(struct roff_node *);
509226948cSschwarze
51c0a657b3Sschwarze static struct ohash tag_data;
52c0a657b3Sschwarze
53c0a657b3Sschwarze
54c0a657b3Sschwarze /*
550ac7e6ecSschwarze * Set up the ohash table to collect nodes
562ae7e873Sschwarze * where various marked-up terms are documented.
572ae7e873Sschwarze */
580ac7e6ecSschwarze void
tag_alloc(void)590ac7e6ecSschwarze tag_alloc(void)
600ac7e6ecSschwarze {
61c4b66caeSschwarze mandoc_ohash_init(&tag_data, 4, offsetof(struct tag_entry, s));
620ac7e6ecSschwarze }
632ae7e873Sschwarze
640ac7e6ecSschwarze void
tag_free(void)650ac7e6ecSschwarze tag_free(void)
660ac7e6ecSschwarze {
670ac7e6ecSschwarze struct tag_entry *entry;
680ac7e6ecSschwarze unsigned int slot;
690ac7e6ecSschwarze
704c8dba62Sschwarze if (tag_data.info.free == NULL)
714c8dba62Sschwarze return;
720ac7e6ecSschwarze entry = ohash_first(&tag_data, &slot);
730ac7e6ecSschwarze while (entry != NULL) {
740ac7e6ecSschwarze free(entry->nodes);
750ac7e6ecSschwarze free(entry);
760ac7e6ecSschwarze entry = ohash_next(&tag_data, &slot);
770ac7e6ecSschwarze }
780ac7e6ecSschwarze ohash_delete(&tag_data);
794c8dba62Sschwarze tag_data.info.free = NULL;
80c0a657b3Sschwarze }
81c0a657b3Sschwarze
82c0a657b3Sschwarze /*
830ac7e6ecSschwarze * Set a node where a term is defined,
847fbd65bbSschwarze * unless the term is already defined at a lower priority.
85c0a657b3Sschwarze */
86c0a657b3Sschwarze void
tag_put(const char * s,int prio,struct roff_node * n)870ac7e6ecSschwarze tag_put(const char *s, int prio, struct roff_node *n)
88c0a657b3Sschwarze {
89c0a657b3Sschwarze struct tag_entry *entry;
90c220f9cfSschwarze struct roff_node *nold;
91*2b14f697Sschwarze const char *se, *src;
92*2b14f697Sschwarze char *cpy;
93c4f5c5f1Sschwarze size_t len;
94c0a657b3Sschwarze unsigned int slot;
95*2b14f697Sschwarze int changed;
96c0a657b3Sschwarze
977d109111Sschwarze assert(prio <= TAG_FALLBACK);
983af42e7fSschwarze
997fbd65bbSschwarze /*
1007fbd65bbSschwarze * If the node is already tagged, the existing tag is
1017fbd65bbSschwarze * explicit and we are now about to add an implicit tag.
1027fbd65bbSschwarze * Don't do that; just skip implicit tagging if the author
1037fbd65bbSschwarze * specified an explicit tag.
1047fbd65bbSschwarze */
1057fbd65bbSschwarze
1067fbd65bbSschwarze if (n->flags & NODE_ID)
1077fbd65bbSschwarze return;
1087fbd65bbSschwarze
1097fbd65bbSschwarze /* Determine the implicit tag. */
1107fbd65bbSschwarze
111*2b14f697Sschwarze changed = 1;
1120ac7e6ecSschwarze if (s == NULL) {
1130ac7e6ecSschwarze if (n->child == NULL || n->child->type != ROFFT_TEXT)
1140ac7e6ecSschwarze return;
1150ac7e6ecSschwarze s = n->child->string;
116d55e8c17Sschwarze switch (s[0]) {
117d55e8c17Sschwarze case '-':
118d55e8c17Sschwarze s++;
119d55e8c17Sschwarze break;
120d55e8c17Sschwarze case '\\':
121d55e8c17Sschwarze switch (s[1]) {
122d55e8c17Sschwarze case '&':
123d55e8c17Sschwarze case '-':
124d55e8c17Sschwarze case 'e':
12552d50eacSschwarze s += 2;
126d55e8c17Sschwarze break;
127d55e8c17Sschwarze default:
128*2b14f697Sschwarze return;
129d55e8c17Sschwarze }
130d55e8c17Sschwarze break;
131d55e8c17Sschwarze default:
132*2b14f697Sschwarze changed = 0;
133d55e8c17Sschwarze break;
134d55e8c17Sschwarze }
1350ac7e6ecSschwarze }
1363af42e7fSschwarze
1373af42e7fSschwarze /*
138*2b14f697Sschwarze * Translate \- and ASCII_HYPH to plain '-'.
1391e876328Sschwarze * Skip whitespace and escapes and whatever follows,
1403af42e7fSschwarze * and if there is any, downgrade the priority.
1413af42e7fSschwarze */
1423af42e7fSschwarze
143*2b14f697Sschwarze cpy = mandoc_malloc(strlen(s) + 1);
144*2b14f697Sschwarze for (src = s, len = 0; *src != '\0'; src++, len++) {
145*2b14f697Sschwarze switch (*src) {
146*2b14f697Sschwarze case '\t':
147*2b14f697Sschwarze case ' ':
148*2b14f697Sschwarze changed = 1;
149*2b14f697Sschwarze break;
150*2b14f697Sschwarze case ASCII_HYPH:
151*2b14f697Sschwarze cpy[len] = '-';
152*2b14f697Sschwarze changed = 1;
153*2b14f697Sschwarze continue;
154*2b14f697Sschwarze case '\\':
155*2b14f697Sschwarze if (src[1] != '-')
156*2b14f697Sschwarze break;
157*2b14f697Sschwarze src++;
158*2b14f697Sschwarze changed = 1;
159*2b14f697Sschwarze /* FALLTHROUGH */
160*2b14f697Sschwarze default:
161*2b14f697Sschwarze cpy[len] = *src;
162*2b14f697Sschwarze continue;
163*2b14f697Sschwarze }
164*2b14f697Sschwarze break;
165*2b14f697Sschwarze }
1663af42e7fSschwarze if (len == 0)
167*2b14f697Sschwarze goto out;
168*2b14f697Sschwarze cpy[len] = '\0';
169730dc51fSschwarze
170*2b14f697Sschwarze if (*src != '\0' && prio < TAG_WEAK)
1717d109111Sschwarze prio = TAG_WEAK;
1723af42e7fSschwarze
173*2b14f697Sschwarze s = cpy;
174*2b14f697Sschwarze se = cpy + len;
1753af42e7fSschwarze slot = ohash_qlookupi(&tag_data, s, &se);
176c0a657b3Sschwarze entry = ohash_find(&tag_data, slot);
177730dc51fSschwarze
178730dc51fSschwarze /* Build a new entry. */
179730dc51fSschwarze
1800ac7e6ecSschwarze if (entry == NULL) {
1813af42e7fSschwarze entry = mandoc_malloc(sizeof(*entry) + len + 1);
182*2b14f697Sschwarze memcpy(entry->s, s, len + 1);
1830ac7e6ecSschwarze entry->nodes = NULL;
1840ac7e6ecSschwarze entry->maxnodes = entry->nnodes = 0;
185c0a657b3Sschwarze ohash_insert(&tag_data, slot, entry);
1860ac7e6ecSschwarze }
187730dc51fSschwarze
1883af42e7fSschwarze /*
1890ac7e6ecSschwarze * Lower priority numbers take precedence.
1900ac7e6ecSschwarze * If a better entry is already present, ignore the new one.
1913af42e7fSschwarze */
192ba3488f0Sschwarze
1930ac7e6ecSschwarze else if (entry->prio < prio)
194*2b14f697Sschwarze goto out;
1950ac7e6ecSschwarze
1960ac7e6ecSschwarze /*
1970ac7e6ecSschwarze * If the existing entry is worse, clear it.
1980ac7e6ecSschwarze * In addition, a tag with priority TAG_FALLBACK
1990ac7e6ecSschwarze * is only used if the tag occurs exactly once.
2000ac7e6ecSschwarze */
2010ac7e6ecSschwarze
2020ac7e6ecSschwarze else if (entry->prio > prio || prio == TAG_FALLBACK) {
203c220f9cfSschwarze while (entry->nnodes > 0) {
204c220f9cfSschwarze nold = entry->nodes[--entry->nnodes];
205c220f9cfSschwarze nold->flags &= ~NODE_ID;
206c220f9cfSschwarze free(nold->tag);
207c220f9cfSschwarze nold->tag = NULL;
208c220f9cfSschwarze }
2097d109111Sschwarze if (prio == TAG_FALLBACK) {
2107d109111Sschwarze entry->prio = TAG_DELETE;
211*2b14f697Sschwarze goto out;
212ba3488f0Sschwarze }
213730dc51fSschwarze }
214730dc51fSschwarze
2150ac7e6ecSschwarze /* Remember the new node. */
216730dc51fSschwarze
2170ac7e6ecSschwarze if (entry->maxnodes == entry->nnodes) {
2180ac7e6ecSschwarze entry->maxnodes += 4;
2190ac7e6ecSschwarze entry->nodes = mandoc_reallocarray(entry->nodes,
2200ac7e6ecSschwarze entry->maxnodes, sizeof(*entry->nodes));
221730dc51fSschwarze }
2220ac7e6ecSschwarze entry->nodes[entry->nnodes++] = n;
22362615e18Sschwarze entry->prio = prio;
2240ac7e6ecSschwarze n->flags |= NODE_ID;
225*2b14f697Sschwarze if (changed) {
226c220f9cfSschwarze assert(n->tag == NULL);
227c220f9cfSschwarze n->tag = mandoc_strndup(s, len);
2280ac7e6ecSschwarze }
229*2b14f697Sschwarze
230*2b14f697Sschwarze out:
231*2b14f697Sschwarze free(cpy);
232c0a657b3Sschwarze }
233c0a657b3Sschwarze
234beabc24cSschwarze int
tag_exists(const char * tag)235beabc24cSschwarze tag_exists(const char *tag)
236c0a657b3Sschwarze {
237beabc24cSschwarze return ohash_find(&tag_data, ohash_qlookup(&tag_data, tag)) != NULL;
2381a557d3cSschwarze }
2399226948cSschwarze
2409226948cSschwarze /*
2419226948cSschwarze * For in-line elements, move the link target
2429226948cSschwarze * to the enclosing paragraph when appropriate.
2439226948cSschwarze */
2449226948cSschwarze static void
tag_move_id(struct roff_node * n)2459226948cSschwarze tag_move_id(struct roff_node *n)
2469226948cSschwarze {
2479226948cSschwarze struct roff_node *np;
2489226948cSschwarze
2499226948cSschwarze np = n;
2509226948cSschwarze for (;;) {
2519226948cSschwarze if (np->prev != NULL)
2529226948cSschwarze np = np->prev;
2539226948cSschwarze else if ((np = np->parent) == NULL)
2549226948cSschwarze return;
2559226948cSschwarze switch (np->tok) {
2569226948cSschwarze case MDOC_It:
2579226948cSschwarze switch (np->parent->parent->norm->Bl.type) {
2589226948cSschwarze case LIST_column:
2599226948cSschwarze /* Target the ROFFT_BLOCK = <tr>. */
2609226948cSschwarze np = np->parent;
2619226948cSschwarze break;
2629226948cSschwarze case LIST_diag:
2639226948cSschwarze case LIST_hang:
2649226948cSschwarze case LIST_inset:
2659226948cSschwarze case LIST_ohang:
2669226948cSschwarze case LIST_tag:
2679226948cSschwarze /* Target the ROFFT_HEAD = <dt>. */
2689226948cSschwarze np = np->parent->head;
2699226948cSschwarze break;
2709226948cSschwarze default:
2719226948cSschwarze /* Target the ROFF_BODY = <li>. */
2729226948cSschwarze break;
2739226948cSschwarze }
2749226948cSschwarze /* FALLTHROUGH */
2759226948cSschwarze case MDOC_Pp: /* Target the ROFFT_ELEM = <p>. */
276c220f9cfSschwarze if (np->tag == NULL) {
277c220f9cfSschwarze np->tag = mandoc_strdup(n->tag == NULL ?
278c220f9cfSschwarze n->child->string : n->tag);
2799226948cSschwarze np->flags |= NODE_ID;
2809226948cSschwarze n->flags &= ~NODE_ID;
2819226948cSschwarze }
2829226948cSschwarze return;
2839226948cSschwarze case MDOC_Sh:
2849226948cSschwarze case MDOC_Ss:
2859226948cSschwarze case MDOC_Bd:
2869226948cSschwarze case MDOC_Bl:
2879226948cSschwarze case MDOC_D1:
2889226948cSschwarze case MDOC_Dl:
2899226948cSschwarze case MDOC_Rs:
2909226948cSschwarze /* Do not move past major blocks. */
2919226948cSschwarze return;
2929226948cSschwarze default:
2939226948cSschwarze /*
2949226948cSschwarze * Move past in-line content and partial
2959226948cSschwarze * blocks, for example .It Xo or .It Bq Er.
2969226948cSschwarze */
2979226948cSschwarze break;
2989226948cSschwarze }
2999226948cSschwarze }
3009226948cSschwarze }
3019226948cSschwarze
3029226948cSschwarze /*
303a875cd0eSschwarze * When a paragraph is tagged and starts with text,
304a875cd0eSschwarze * move the permalink to the first few words.
305a875cd0eSschwarze */
306a875cd0eSschwarze static void
tag_move_href(struct roff_man * man,struct roff_node * n,const char * tag)307a875cd0eSschwarze tag_move_href(struct roff_man *man, struct roff_node *n, const char *tag)
308a875cd0eSschwarze {
309a875cd0eSschwarze char *cp;
310a875cd0eSschwarze
311a875cd0eSschwarze if (n == NULL || n->type != ROFFT_TEXT ||
312a875cd0eSschwarze *n->string == '\0' || *n->string == ' ')
313a875cd0eSschwarze return;
314a875cd0eSschwarze
315a875cd0eSschwarze cp = n->string;
316a875cd0eSschwarze while (cp != NULL && cp - n->string < 5)
317a875cd0eSschwarze cp = strchr(cp + 1, ' ');
318a875cd0eSschwarze
319a875cd0eSschwarze /* If the first text node is longer, split it. */
320a875cd0eSschwarze
321a875cd0eSschwarze if (cp != NULL && cp[1] != '\0') {
322a875cd0eSschwarze man->last = n;
323a875cd0eSschwarze man->next = ROFF_NEXT_SIBLING;
324a875cd0eSschwarze roff_word_alloc(man, n->line,
325a875cd0eSschwarze n->pos + (cp - n->string), cp + 1);
326a875cd0eSschwarze man->last->flags = n->flags & ~NODE_LINE;
327a875cd0eSschwarze *cp = '\0';
328a875cd0eSschwarze }
329a875cd0eSschwarze
330a875cd0eSschwarze assert(n->tag == NULL);
331a875cd0eSschwarze n->tag = mandoc_strdup(tag);
332a875cd0eSschwarze n->flags |= NODE_HREF;
333a875cd0eSschwarze }
334a875cd0eSschwarze
335a875cd0eSschwarze /*
3369226948cSschwarze * When all tags have been set, decide where to put
3379226948cSschwarze * the associated permalinks, and maybe move some tags
3389226948cSschwarze * to the beginning of the respective paragraphs.
3399226948cSschwarze */
3409226948cSschwarze void
tag_postprocess(struct roff_man * man,struct roff_node * n)3416e2a0df9Sschwarze tag_postprocess(struct roff_man *man, struct roff_node *n)
3429226948cSschwarze {
3439226948cSschwarze if (n->flags & NODE_ID) {
3449226948cSschwarze switch (n->tok) {
3456e2a0df9Sschwarze case MDOC_Pp:
346a875cd0eSschwarze tag_move_href(man, n->next, n->tag);
3476e2a0df9Sschwarze break;
3489226948cSschwarze case MDOC_Bd:
349a875cd0eSschwarze case MDOC_D1:
350a875cd0eSschwarze case MDOC_Dl:
351a875cd0eSschwarze tag_move_href(man, n->child, n->tag);
352a875cd0eSschwarze break;
3539226948cSschwarze case MDOC_Bl:
3549226948cSschwarze /* XXX No permalink for now. */
3559226948cSschwarze break;
3569226948cSschwarze default:
3579226948cSschwarze if (n->type == ROFFT_ELEM || n->tok == MDOC_Fo)
3589226948cSschwarze tag_move_id(n);
3599226948cSschwarze if (n->tok != MDOC_Tg)
3609226948cSschwarze n->flags |= NODE_HREF;
361c220f9cfSschwarze else if ((n->flags & NODE_ID) == 0) {
3629226948cSschwarze n->flags |= NODE_NOPRT;
363c220f9cfSschwarze free(n->tag);
364c220f9cfSschwarze n->tag = NULL;
365c220f9cfSschwarze }
3669226948cSschwarze break;
3679226948cSschwarze }
3689226948cSschwarze }
3699226948cSschwarze for (n = n->child; n != NULL; n = n->next)
3706e2a0df9Sschwarze tag_postprocess(man, n);
3719226948cSschwarze }
372