diff options
| author | Daniel Eklöf <daniel@ekloef.se> | 2025-02-05 12:47:06 +0000 |
|---|---|---|
| committer | Daniel Eklöf <daniel@ekloef.se> | 2025-02-05 12:47:06 +0000 |
| commit | dd01783f88902b8e33820c838fc4e6aba122b578 (patch) | |
| tree | 627a273b617d4c7dd1400f45c5e35db08e526830 /url-mode.c | |
| parent | box-drawings: handle architecture with soft-float (diff) | |
| parent | changelog: custom regex's (diff) | |
Merge branch 'regex-mode'
Closes #1386
Closes #1872
Diffstat (limited to 'url-mode.c')
| -rw-r--r-- | url-mode.c | 325 |
1 files changed, 114 insertions, 211 deletions
@@ -4,6 +4,7 @@ #include <string.h> #include <wctype.h> #include <unistd.h> +#include <regex.h> #include <sys/stat.h> #include <fcntl.h> @@ -66,12 +67,13 @@ spawn_url_launcher_with_token(struct terminal *term, return false; } + xassert(term->url_launch != NULL); bool ret = false; if (spawn_expand_template( - &term->conf->url.launch, 1, - (const char *[]){"url"}, - (const char *[]){url}, + term->url_launch, 2, + (const char *[]){"url", "match"}, + (const char *[]){url, url}, &argc, &argv)) { ret = spawn( @@ -83,6 +85,8 @@ spawn_url_launcher_with_token(struct terminal *term, free(argv); } + term->url_launch = NULL; + close(dev_null); return ret; } @@ -106,6 +110,8 @@ static bool spawn_url_launcher(struct seat *seat, struct terminal *term, const char *url, uint32_t serial) { + xassert(term->url_launch != NULL); + struct spawn_activation_context *ctx = xmalloc(sizeof(*ctx)); *ctx = (struct spawn_activation_context){ .term = term, @@ -291,243 +297,135 @@ urls_input(struct seat *seat, struct terminal *term, } } -static int -c32cmp_single(const void *_a, const void *_b) -{ - const char32_t *a = _a; - const char32_t *b = _b; - return *a - *b; -} +struct vline { + char *utf8; + size_t len; /* Length of utf8[] */ + size_t sz; /* utf8[] allocated size */ + struct coord *map; /* Maps utf8[ofs] to grid coordinates */ +}; static void -auto_detected(const struct terminal *term, enum url_action action, - url_list_t *urls) +regex_detected(const struct terminal *term, enum url_action action, + const regex_t *preg, url_list_t *urls) { - const struct config *conf = term->conf; - - const char32_t *uri_characters = conf->url.uri_characters; - if (uri_characters == NULL) - return; - - const size_t uri_characters_count = c32len(uri_characters); - if (uri_characters_count == 0) - return; - - size_t max_prot_len = conf->url.max_prot_len; - char32_t proto_chars[max_prot_len]; - struct coord proto_start[max_prot_len]; - size_t proto_char_count = 0; + /* + * Use regcomp()+regexec() to find patterns. + * + * Since we can't feed regexec() one character at a time, and + * since it doesn't accept wide characters, we need to build utf8 + * strings. + * + * Each string represents a logical line (i.e. handle line-wrap). + * To be able to map regex matches back to the grid, we store the + * grid coordinates of *each* character, in the line struct as + * well. This is offset based; utf8[ofs] has its grid coordinates + * in map[ofs. + */ - enum { - STATE_PROTOCOL, - STATE_URL, - } state = STATE_PROTOCOL; + /* There is *at most* term->rows logical lines */ + struct vline vlines[term->rows]; + size_t vline_idx = 0; - struct coord start = {-1, -1}; - char32_t url[term->cols * term->rows + 1]; - size_t len = 0; + memset(vlines, 0, sizeof(vlines)); + struct vline *vline = &vlines[vline_idx]; - ssize_t parenthesis = 0; - ssize_t brackets = 0; - ssize_t ltgts = 0; + mbstate_t ps = {0}; for (int r = 0; r < term->rows; r++) { const struct row *row = grid_row_in_view(term->grid, r); for (int c = 0; c < term->cols; c++) { const struct cell *cell = &row->cells[c]; + const char32_t *wc = &cell->wc; + size_t wc_count = 1; - if (cell->wc >= CELL_SPACER) - continue; - - const char32_t *wcs = NULL; - size_t wc_count = 0; + /* Expand combining characters */ + if (wc[0] >= CELL_COMB_CHARS_LO && wc[0] <= CELL_COMB_CHARS_HI) { + const struct composed *composed = + composed_lookup(term->composed, wc[0] - CELL_COMB_CHARS_LO); + xassert(composed != NULL); - if (cell->wc >= CELL_COMB_CHARS_LO && cell->wc <= CELL_COMB_CHARS_HI) { - struct composed *composed = - composed_lookup(term->composed, cell->wc - CELL_COMB_CHARS_LO); - wcs = composed->chars; + wc = composed->chars; wc_count = composed->count; - } else { - wcs = &cell->wc; - wc_count = 1; } - for (size_t w_idx = 0; w_idx < wc_count; w_idx++) { - char32_t wc = wcs[w_idx]; - - switch (state) { - case STATE_PROTOCOL: - for (size_t i = 0; i < max_prot_len - 1; i++) { - proto_chars[i] = proto_chars[i + 1]; - proto_start[i] = proto_start[i + 1]; - } - - if (proto_char_count >= max_prot_len) - proto_char_count = max_prot_len - 1; - - proto_chars[max_prot_len - 1] = wc; - proto_start[max_prot_len - 1] = (struct coord){c, r}; - proto_char_count++; - - for (size_t i = 0; i < conf->url.prot_count; i++) { - size_t prot_len = c32len(conf->url.protocols[i]); - - if (proto_char_count < prot_len) - continue; - - const char32_t *proto = - &proto_chars[max_prot_len - prot_len]; + /* Convert wide character to utf8 */ + for (size_t i = 0; i < wc_count; i++) { + char buf[16]; + size_t char_len = c32rtomb(buf, wc[i], &ps); - if (c32ncasecmp(conf->url.protocols[i], proto, prot_len) == - 0) { - state = STATE_URL; - start = proto_start[max_prot_len - prot_len]; + if (char_len == (size_t)-1) + continue; - c32ncpy(url, proto, prot_len); - len = prot_len; + for (size_t j = 0; j < char_len; j++) { + const size_t requires_size = vline->len + char_len; - parenthesis = brackets = ltgts = 0; - break; + /* Need to grow? Remember to save at least one byte for terminator */ + if (vline->sz == 0 || requires_size > vline->sz - 1) { + const size_t new_size = requires_size * 2; + vline->utf8 = xreallocarray(vline->utf8, new_size, 1); + vline->map = xreallocarray(vline->map, new_size, sizeof(vline->map[0])); + vline->sz = new_size; } - } - break; - case STATE_URL: { - const char32_t *match = - bsearch(&wc, uri_characters, uri_characters_count, - sizeof(uri_characters[0]), &c32cmp_single); - - bool emit_url = false; - - if (match == NULL) { - /* - * Character is not a valid URI character. Emit - * the URL we've collected so far, *without* - * including _this_ character. - */ - emit_url = true; - } else { - xassert(*match == wc); - - switch (wc) { - default: - url[len++] = wc; - break; - - case U'(': - parenthesis++; - url[len++] = wc; - break; - - case U'[': - brackets++; - url[len++] = wc; - break; - - case U'<': - ltgts++; - url[len++] = wc; - break; - - case U')': - if (--parenthesis < 0) - emit_url = true; - else - url[len++] = wc; - break; - - case U']': - if (--brackets < 0) - emit_url = true; - else - url[len++] = wc; - break; + vline->utf8[vline->len + j] = buf[j]; + vline->map[vline->len + j] = (struct coord){c, term->grid->view + r}; + } - case U'>': - if (--ltgts < 0) - emit_url = true; - else - url[len++] = wc; - break; - } - } + vline->len += char_len; + } + } - if (c >= term->cols - 1 && row->linebreak) { - /* - * Endpoint is inclusive, and we'll be subtracting - * 1 from the column when emitting the URL. - */ - c++; - emit_url = true; - } + if (row->linebreak) { + if (vline->len > 0) { + vline->utf8[vline->len++] = '\0'; + ps = (mbstate_t){0}; - if (emit_url) { - struct coord end = {c, r}; + vline_idx++; + vline = &vlines[vline_idx]; + } + } + } - if (--end.col < 0) { - end.row--; - end.col = term->cols - 1; - } + for (size_t i = 0; i < ALEN(vlines); i++) { + const struct vline *v = &vlines[i]; + if (v->utf8 == NULL) + continue;; - /* Heuristic to remove trailing characters that - * are valid URL characters, but typically not at - * the end of the URL */ - bool done = false; - do { - switch (url[len - 1]) { - case U'.': - case U',': - case U':': - case U';': - case U'?': - case U'!': - case U'"': - case U'\'': - case U'%': - len--; - end.col--; - if (end.col < 0) { - end.row--; - end.col = term->cols - 1; - } - break; + const char *search_string = v->utf8; + while (true) { + regmatch_t matches[preg->re_nsub + 1]; + int r = regexec(preg, search_string, preg->re_nsub + 1, matches, 0); - default: - done = true; - break; - } - } while (!done); + if (r == REG_NOMATCH) + break; - url[len] = U'\0'; + const size_t mlen = matches[1].rm_eo - matches[1].rm_so; + const size_t start = &search_string[matches[1].rm_so] - v->utf8; + const size_t end = start + mlen; - start.row += term->grid->view; - end.row += term->grid->view; + LOG_DBG( + "regex match at row %d: %.*srow/col = %dx%d", + matches[1].rm_so, (int)mlen, &search_string[matches[1].rm_so], + v->map[start].row, v->map[start].col); - char *url_utf8 = ac32tombs(url); - if (url_utf8 != NULL) { - tll_push_back( - *urls, - ((struct url){.id = (uint64_t)rand() << 32 | rand(), - .url = url_utf8, - .range = - { - .start = start, - .end = end, - }, - .action = action, - .osc8 = false})); - } + tll_push_back( + *urls, + ((struct url){ + .id = (uint64_t)rand() << 32 | rand(), + .url = xstrndup(&v->utf8[start], mlen), + .range = { + .start = v->map[start], + .end = v->map[end - 1], /* Inclusive */ + }, + .action = action, + .osc8 = false})); - state = STATE_PROTOCOL; - len = 0; - parenthesis = brackets = ltgts = 0; - } - break; - } - } - } + search_string += matches[0].rm_eo; } + + free(v->utf8); + free(v->map); } } @@ -629,11 +527,13 @@ remove_overlapping(url_list_t *urls, int cols) } void -urls_collect(const struct terminal *term, enum url_action action, url_list_t *urls) +urls_collect(const struct terminal *term, enum url_action action, + const regex_t *preg, bool osc8, url_list_t *urls) { xassert(tll_length(term->urls) == 0); - osc8_uris(term, action, urls); - auto_detected(term, action, urls); + if (osc8) + osc8_uris(term, action, urls); + regex_detected(term, action, preg, urls); remove_overlapping(urls, term->grid->num_cols); } @@ -816,7 +716,7 @@ tag_cells_for_url(struct terminal *term, const struct url *url, bool value) } void -urls_render(struct terminal *term) +urls_render(struct terminal *term, const struct config_spawn_template *launch) { struct wl_window *win = term->window; @@ -851,6 +751,9 @@ urls_render(struct terminal *term) /* Snapshot the current grid */ term->url_grid_snapshot = grid_snapshot(term->grid); + /* Remember which launcher to use */ + term->url_launch = launch; + xassert(tll_length(win->urls) == 0); tll_foreach(win->term->urls, it) { struct wl_url url = {.url = &it->item}; |
