*/
#include <assert.h>
+#include <locale.h>
#include <stdio.h>
#include <stdlib.h>
#include "halibut.h"
static void dbg_prtwordlist(int level, word *w);
static void dbg_prtkws(keywordlist *kws);
+static const struct pre_backend {
+ void *(*func)(paragraph *, keywordlist *, indexdata *);
+ int bitfield;
+} pre_backends[] = {
+ {paper_pre_backend, 0x0001}
+};
+
static const struct backend {
char *name;
- void (*func)(paragraph *, keywordlist *, indexdata *);
- int bitfield;
+ void (*func)(paragraph *, keywordlist *, indexdata *, void *);
+ paragraph *(*filename)(char *filename);
+ int bitfield, prebackend_bitfield;
} backends[] = {
- {"text", text_backend, 0x0001},
- {"xhtml", xhtml_backend, 0x0002},
- {"html", xhtml_backend, 0x0002},
- {"hlp", whlp_backend, 0x0004},
- {"whlp", whlp_backend, 0x0004},
- {"winhelp", whlp_backend, 0x0004},
- {"man", man_backend, 0x0008},
+ {"text", text_backend, text_config_filename, 0x0001, 0},
+ {"xhtml", html_backend, html_config_filename, 0x0002, 0},
+ {"html", html_backend, html_config_filename, 0x0002, 0},
+ {"hlp", whlp_backend, whlp_config_filename, 0x0004, 0},
+ {"whlp", whlp_backend, whlp_config_filename, 0x0004, 0},
+ {"winhelp", whlp_backend, whlp_config_filename, 0x0004, 0},
+ {"man", man_backend, man_config_filename, 0x0008, 0},
+ {"info", info_backend, info_config_filename, 0x0010, 0},
+ {"ps", ps_backend, ps_config_filename, 0x0020, 0x0001},
+ {"pdf", pdf_backend, pdf_config_filename, 0x0040, 0x0001},
};
int main(int argc, char **argv) {
int nogo;
int errs;
int reportcols;
+ int list_fonts;
+ int input_charset;
int debug;
- int backendbits;
+ int backendbits, prebackbits;
int k, b;
+ paragraph *cfg, *cfg_tail;
+ void *pre_backend_data[16];
+
+ /*
+ * Use the specified locale everywhere. It'll be used for
+ * output of error messages, and as the default character set
+ * for input files if one is not explicitly specified.
+ *
+ * However, we need to use standard numeric formatting for
+ * output of things like PDF.
+ */
+ setlocale(LC_ALL, "");
+ setlocale(LC_NUMERIC, "C");
/*
* Set up initial (default) parameters.
*/
- infiles = mknewa(char *, argc);
+ infiles = snewn(argc, char *);
nfiles = 0;
nogo = errs = FALSE;
reportcols = 0;
+ list_fonts = 0;
+ input_charset = CS_ASCII;
debug = 0;
backendbits = 0;
+ cfg = cfg_tail = NULL;
if (argc == 1) {
usage();
*/
while (--argc) {
char *p = *++argv;
- if (*p == '-') {
+ if (*p == '-' && p[1]) {
/*
* An option.
*/
for (k = 0; k < (int)lenof(backends); k++)
if (!strcmp(opt+1, backends[k].name)) {
backendbits |= backends[k].bitfield;
+ if (val) {
+ paragraph *p = backends[k].filename(val);
+ assert(p);
+ if (cfg_tail)
+ cfg_tail->next = p;
+ else
+ cfg = p;
+ while (p->next)
+ p = p->next;
+ cfg_tail = p;
+ }
break;
}
if (k < (int)lenof(backends)) {
/* do nothing */;
+ } else if (!strcmp(opt, "-input-charset")) {
+ if (!val) {
+ errs = TRUE, error(err_optnoarg, opt);
+ } else {
+ int charset = charset_from_localenc(val);
+ if (charset == CS_NONE) {
+ errs = TRUE, error(err_cmdcharset, val);
+ } else {
+ input_charset = charset;
+ }
+ }
} else if (!strcmp(opt, "-help")) {
help();
nogo = TRUE;
!strcmp(opt, "-license")) {
licence();
nogo = TRUE;
+ } else if (!strcmp(opt, "-list-charsets")) {
+ listcharsets();
+ nogo = TRUE;
+ } else if (!strcmp(opt, "-list-fonts")) {
+ list_fonts = TRUE;
} else if (!strcmp(opt, "-precise")) {
reportcols = 1;
} else {
break;
}
break;
-#if 0
- case 'o':
+ case 'C':
/*
* Option requiring parameter.
*/
* Now c is the option and p is the parameter.
*/
switch (c) {
- case 'o':
- outfile = p;
+ case 'C':
+ /*
+ * -C means we split our argument up into
+ * colon-separated chunks and assemble them
+ * into a config paragraph.
+ */
+ {
+ char *s = dupstr(p), *q, *r;
+ paragraph *para;
+
+ para = cmdline_cfg_new();
+
+ q = r = s;
+ while (*q) {
+ if (*q == ':') {
+ *r = '\0';
+ /* XXX ad-hoc diagnostic */
+ if (!strcmp(s, "input-charset"))
+ error(err_futileopt, "Cinput-charset",
+ "; use --input-charset");
+ cmdline_cfg_add(para, s);
+ r = s;
+ } else {
+ if (*q == '\\' && q[1])
+ q++;
+ *r++ = *q;
+ }
+ q++;
+ }
+ *r = '\0';
+ cmdline_cfg_add(para, s);
+
+ if (cfg_tail)
+ cfg_tail->next = para;
+ else
+ cfg = para;
+ cfg_tail = para;
+ }
break;
}
p = NULL; /* prevent continued processing */
break;
-#endif
default:
/*
* Unrecognised option.
/*
* A non-option argument.
*/
- infiles[nfiles++] = p;
+ if (!strcmp(p, "-"))
+ infiles[nfiles++] = NULL; /* special case: read stdin */
+ else
+ infiles[nfiles++] = p;
}
}
/*
* Do the work.
*/
- if (nfiles == 0) {
+ if (nfiles == 0 && !list_fonts) {
error(err_noinput);
usage();
exit(EXIT_FAILURE);
in.pushback = NULL;
in.reportcols = reportcols;
in.stack = NULL;
+ in.defcharset = input_charset;
idx = make_index();
sourceform = read_input(&in, idx);
+ if (list_fonts) {
+ listfonts();
+ exit(EXIT_SUCCESS);
+ }
if (!sourceform)
exit(EXIT_FAILURE);
- sfree(in.pushback);
+ /*
+ * Append the config directives acquired from the command
+ * line.
+ */
+ {
+ paragraph *end;
- mark_attr_ends(sourceform);
+ end = sourceform;
+ while (end && end->next)
+ end = end->next;
+ assert(end);
+
+ end->next = cfg;
+ }
+
+ sfree(in.pushback);
sfree(infiles);
for (p = sourceform; p; p = p->next)
if (p->type == para_IM)
- index_merge(idx, TRUE, p->keyword, p->words);
+ index_merge(idx, TRUE, p->keyword, p->words, &p->fpos);
build_index(idx);
+ /*
+ * Set up attr_First / attr_Last / attr_Always, in the main
+ * document and in the index entries.
+ */
+ for (p = sourceform; p; p = p->next)
+ mark_attr_ends(p->words);
+ {
+ int i;
+ indexentry *entry;
+
+ for (i = 0; (entry = index234(idx->entries, i)) != NULL; i++)
+ mark_attr_ends(entry->text);
+ }
+
if (debug) {
index_debug(idx);
dbg_prtkws(keywords);
}
/*
+ * Select and run the pre-backends.
+ */
+ prebackbits = 0;
+ for (k = 0; k < (int)lenof(backends); k++)
+ if (backendbits == 0 || (backendbits & backends[k].bitfield))
+ prebackbits |= backends[k].prebackend_bitfield;
+ for (k = 0; k < (int)lenof(pre_backends); k++)
+ if (prebackbits & pre_backends[k].bitfield) {
+ assert(k < (int)lenof(pre_backend_data));
+ pre_backend_data[k] =
+ pre_backends[k].func(sourceform, keywords, idx);
+ }
+
+ /*
* Run the selected set of backends.
*/
for (k = b = 0; k < (int)lenof(backends); k++)
if (b != backends[k].bitfield) {
b = backends[k].bitfield;
- if (backendbits == 0 || (backendbits & b))
- backends[k].func(sourceform, keywords, idx);
+ if (backendbits == 0 || (backendbits & b)) {
+ void *pbd = NULL;
+ int pbb = backends[k].prebackend_bitfield;
+ int m;
+
+ for (m = 0; m < (int)lenof(pre_backends); m++)
+ if (pbb & pre_backends[m].bitfield) {
+ assert(m < (int)lenof(pre_backend_data));
+ pbd = pre_backend_data[m];
+ break;
+ }
+
+ backends[k].func(sourceform, keywords, idx, pbd);
+ }
}
free_para_list(sourceform);
printf("\"");
} else
printf("(no text)");
+ if (w->breaks)
+ printf(" [breaks]");
if (w->alt) {
printf(" alt = {\n");
dbg_prtwordlist(level+1, w->alt);
~mdw / sgt / halibut / blobdiff