[sgt/halibut] / main.c

/*
 * main.c: command line parsing and top level
 */

#include <assert.h>
#include <locale.h>
#include <stdio.h>
#include <stdlib.h>
#include "halibut.h"

static void dbg_prtsource(paragraph *sourceform);
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 *, void *);
    paragraph *(*filename)(char *filename);
    int bitfield, prebackend_bitfield;
} backends[] = {
    {"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) {
    char **infiles;
    int nfiles;
    int nogo;
    int errs;
    int reportcols;
    int debug;
    int backendbits, prebackbits;
    int k, b;
    paragraph *cfg, *cfg_tail;
    void *pre_backend_data[16];

    setlocale(LC_ALL, "");

    /*
     * Set up initial (default) parameters.
     */
    infiles = snewn(argc, char *);
    nfiles = 0;
    nogo = errs = FALSE;
    reportcols = 0;
    debug = 0;
    backendbits = 0;
    cfg = cfg_tail = NULL;

    if (argc == 1) {
	usage();
	exit(EXIT_SUCCESS);
    }

    /*
     * Parse command line arguments.
     */
    while (--argc) {
	char *p = *++argv;
	if (*p == '-') {
	    /*
	     * An option.
	     */
	    while (p && *++p) {
		char c = *p;
		switch (c) {
		  case '-':
		    /*
		     * Long option.
		     */
		    {
			char *opt, *val;
			opt = p++;     /* opt will have _one_ leading - */
			while (*p && *p != '=')
			    p++;	       /* find end of option */
			if (*p == '=') {
			    *p++ = '\0';
			    val = p;
			} else
			    val = NULL;

			assert(opt[0] == '-');
			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, "-help")) {
			    help();
			    nogo = TRUE;
			} else if (!strcmp(opt, "-version")) {
			    showversion();
			    nogo = TRUE;
			} else if (!strcmp(opt, "-licence") ||
				   !strcmp(opt, "-license")) {
			    licence();
			    nogo = TRUE;
			} else if (!strcmp(opt, "-precise")) {
			    reportcols = 1;
			} else {
			    errs = TRUE, error(err_nosuchopt, opt);
			}
		    }
		    p = NULL;
		    break;
		  case 'h':
		  case 'V':
		  case 'L':
		  case 'P':
		  case 'd':
		    /*
		     * Option requiring no parameter.
		     */
		    switch (c) {
		      case 'h':
			help();
			nogo = TRUE;
			break;
		      case 'V':
			showversion();
			nogo = TRUE;
			break;
		      case 'L':
			licence();
			nogo = TRUE;
			break;
		      case 'P':
			reportcols = 1;
			break;
		      case 'd':
			debug = TRUE;
			break;
		    }
		    break;
		  case 'C':
		    /*
		     * Option requiring parameter.
		     */
		    p++;
		    if (!*p && argc > 1)
			--argc, p = *++argv;
		    else if (!*p) {
			char opt[2];
			opt[0] = c;
			opt[1] = '\0';
			errs = TRUE, error(err_optnoarg, opt);
		    }
		    /*
		     * Now c is the option and p is the parameter.
		     */
		    switch (c) {
		      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';
				    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;
		  default:
		    /*
		     * Unrecognised option.
		     */
		    {
			char opt[2];
			opt[0] = c;
			opt[1] = '\0';
			errs = TRUE, error(err_nosuchopt, opt);
		    }
		}
	    }
	} else {
	    /*
	     * A non-option argument.
	     */
	    infiles[nfiles++] = p;
	}
    }

    if (errs)
	exit(EXIT_FAILURE);
    if (nogo)
	exit(EXIT_SUCCESS);

    /*
     * Do the work.
     */
    if (nfiles == 0) {
	error(err_noinput);
	usage();
	exit(EXIT_FAILURE);
    }

    {
	input in;
	paragraph *sourceform, *p;
	indexdata *idx;
	keywordlist *keywords;

	in.filenames = infiles;
	in.nfiles = nfiles;
	in.currfp = NULL;
	in.currindex = 0;
	in.npushback = in.pushbacksize = 0;
	in.pushback = NULL;
	in.reportcols = reportcols;
	in.stack = NULL;
	in.defcharset = charset_from_locale();

	idx = make_index();

	sourceform = read_input(&in, idx);
	if (!sourceform)
	    exit(EXIT_FAILURE);

	/*
	 * Append the config directives acquired from the command
	 * line.
	 */
	{
	    paragraph *end;

	    end = sourceform;
	    while (end && end->next)
		end = end->next;
	    assert(end);

	    end->next = cfg;
	}

	sfree(in.pushback);

	sfree(infiles);

	keywords = get_keywords(sourceform);
	if (!keywords)
	    exit(EXIT_FAILURE);
	gen_citations(sourceform, keywords);
	subst_keywords(sourceform, keywords);

	for (p = sourceform; p; p = p->next)
	    if (p->type == para_IM)
		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);
	    dbg_prtsource(sourceform);
	}

	/*
	 * 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)) {
		    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);
	free_keywords(keywords);
	cleanup_index(idx);
    }

    return 0;
}

static void dbg_prtsource(paragraph *sourceform) {
    /*
     * Output source form in debugging format.
     */

    paragraph *p;
    for (p = sourceform; p; p = p->next) {
	wchar_t *wp;
	printf("para %d ", p->type);
	if (p->keyword) {
	    wp = p->keyword;
	    while (*wp) {
		putchar('\"');
		for (; *wp; wp++)
		    putchar(*wp);
		putchar('\"');
		if (*++wp)
		    printf(", ");
	    }
	} else
	    printf("(no keyword)");
	printf(" {\n");
	dbg_prtwordlist(1, p->words);
	printf("}\n");
    }
}

static void dbg_prtkws(keywordlist *kws) {
    /*
     * Output keywords in debugging format.
     */

    int i;
    keyword *kw;

    for (i = 0; (kw = index234(kws->keys, i)) != NULL; i++) {
	wchar_t *wp;
	printf("keyword ");
	wp = kw->key;
	while (*wp) {
	    putchar('\"');
	    for (; *wp; wp++)
		putchar(*wp);
	    putchar('\"');
	    if (*++wp)
		printf(", ");
	}
	printf(" {\n");
	dbg_prtwordlist(1, kw->text);
	printf("}\n");
    }
}

static void dbg_prtwordlist(int level, word *w) {
    for (; w; w = w->next) {
	wchar_t *wp;
	printf("%*sword %d ", level*4, "", w->type);
	if (w->text) {
	    printf("\"");
	    for (wp = w->text; *wp; wp++)
		    putchar(*wp);
	    printf("\"");
	} else
	    printf("(no text)");
	if (w->alt) {
	    printf(" alt = {\n");
	    dbg_prtwordlist(level+1, w->alt);
	    printf("%*s}", level*4, "");
	}
	printf("\n");
    }
}
Commit	Line	Data
d7482997	1	/*
	2	* main.c: command line parsing and top level
	3	*/
	4
c8fb54d2	5	#include <assert.h>
7e976207	6	#include <locale.h>
d7482997	7	#include <stdio.h>
	8	#include <stdlib.h>
	9	#include "halibut.h"
	10
	11	static void dbg_prtsource(paragraph *sourceform);
	12	static void dbg_prtwordlist(int level, word *w);
	13	static void dbg_prtkws(keywordlist *kws);
	14
43341922	15	static const struct pre_backend {
	16	void (func)(paragraph , keywordlist , indexdata *);
	17	int bitfield;
	18	} pre_backends[] = {
	19	{paper_pre_backend, 0x0001}
	20	};
	21
c8fb54d2	22	static const struct backend {
c8fb54d2	23	char *name;
43341922	24	void (func)(paragraph , keywordlist , indexdata , void *);
ba9c1487	25	paragraph (filename)(char *filename);
43341922	26	int bitfield, prebackend_bitfield;
c8fb54d2	27	} backends[] = {
43341922	28	{"text", text_backend, text_config_filename, 0x0001, 0},
78c73085	29	{"xhtml", html_backend, html_config_filename, 0x0002, 0},
78c73085	30	{"html", html_backend, html_config_filename, 0x0002, 0},
43341922	31	{"hlp", whlp_backend, whlp_config_filename, 0x0004, 0},
	32	{"whlp", whlp_backend, whlp_config_filename, 0x0004, 0},
	33	{"winhelp", whlp_backend, whlp_config_filename, 0x0004, 0},
	34	{"man", man_backend, man_config_filename, 0x0008, 0},
	35	{"info", info_backend, info_config_filename, 0x0010, 0},
	36	{"ps", ps_backend, ps_config_filename, 0x0020, 0x0001},
	37	{"pdf", pdf_backend, pdf_config_filename, 0x0040, 0x0001},
c8fb54d2	38	};
c8fb54d2	39
d7482997	40	int main(int argc, char **argv) {
d7482997	41	char **infiles;
d7482997	42	int nfiles;
	43	int nogo;
	44	int errs;
	45	int reportcols;
	46	int debug;
43341922	47	int backendbits, prebackbits;
c8fb54d2	48	int k, b;
6a0b9d08	49	paragraph cfg, cfg_tail;
43341922	50	void *pre_backend_data[16];
d7482997	51
7e976207	52	setlocale(LC_ALL, "");
7e976207	53
d7482997	54	/*
	55	* Set up initial (default) parameters.
	56	*/
f1530049	57	infiles = snewn(argc, char *);
d7482997	58	nfiles = 0;
	59	nogo = errs = FALSE;
	60	reportcols = 0;
	61	debug = 0;
c8fb54d2	62	backendbits = 0;
6a0b9d08	63	cfg = cfg_tail = NULL;
d7482997	64
	65	if (argc == 1) {
	66	usage();
	67	exit(EXIT_SUCCESS);
	68	}
	69
	70	/*
	71	* Parse command line arguments.
	72	*/
	73	while (--argc) {
	74	char p = ++argv;
	75	if (*p == '-') {
	76	/*
	77	* An option.
	78	*/
	79	while (p && *++p) {
	80	char c = *p;
	81	switch (c) {
	82	case '-':
	83	/*
	84	* Long option.
	85	*/
	86	{
	87	char opt, val;
	88	opt = p++; /* opt will have _one_ leading - */
	89	while (p && p != '=')
	90	p++; /* find end of option */
	91	if (*p == '=') {
	92	*p++ = '\0';
	93	val = p;
	94	} else
	95	val = NULL;
c8fb54d2	96
	97	assert(opt[0] == '-');
	98	for (k = 0; k < (int)lenof(backends); k++)
	99	if (!strcmp(opt+1, backends[k].name)) {
	100	backendbits \|= backends[k].bitfield;
ba9c1487	101	if (val) {
	102	paragraph *p = backends[k].filename(val);
	103	assert(p);
	104	if (cfg_tail)
	105	cfg_tail->next = p;
	106	else
	107	cfg = p;
	108	while (p->next)
	109	p = p->next;
	110	cfg_tail = p;
	111	}
c8fb54d2	112	break;
	113	}
	114	if (k < (int)lenof(backends)) {
	115	/* do nothing */;
	116	} else if (!strcmp(opt, "-help")) {
d7482997	117	help();
	118	nogo = TRUE;
	119	} else if (!strcmp(opt, "-version")) {
	120	showversion();
	121	nogo = TRUE;
	122	} else if (!strcmp(opt, "-licence") \|\|
	123	!strcmp(opt, "-license")) {
	124	licence();
	125	nogo = TRUE;
d7482997	126	} else if (!strcmp(opt, "-precise")) {
	127	reportcols = 1;
	128	} else {
	129	errs = TRUE, error(err_nosuchopt, opt);
	130	}
	131	}
	132	p = NULL;
	133	break;
	134	case 'h':
	135	case 'V':
	136	case 'L':
	137	case 'P':
	138	case 'd':
	139	/*
	140	* Option requiring no parameter.
	141	*/
	142	switch (c) {
	143	case 'h':
	144	help();
	145	nogo = TRUE;
	146	break;
	147	case 'V':
	148	showversion();
	149	nogo = TRUE;
	150	break;
	151	case 'L':
	152	licence();
	153	nogo = TRUE;
	154	break;
	155	case 'P':
	156	reportcols = 1;
	157	break;
	158	case 'd':
	159	debug = TRUE;
	160	break;
	161	}
	162	break;
6a0b9d08	163	case 'C':
d7482997	164	/*
	165	* Option requiring parameter.
	166	*/
	167	p++;
	168	if (!*p && argc > 1)
	169	--argc, p = *++argv;
	170	else if (!*p) {
	171	char opt[2];
	172	opt[0] = c;
	173	opt[1] = '\0';
	174	errs = TRUE, error(err_optnoarg, opt);
	175	}
	176	/*
	177	* Now c is the option and p is the parameter.
	178	*/
	179	switch (c) {
6a0b9d08	180	case 'C':
	181	/*
	182	* -C means we split our argument up into
	183	* colon-separated chunks and assemble them
	184	* into a config paragraph.
	185	*/
	186	{
e4ea58f8	187	char s = dupstr(p), q, *r;
6a0b9d08	188	paragraph *para;
6a0b9d08	189
e4ea58f8	190	para = cmdline_cfg_new();
6a0b9d08	191
e4ea58f8	192	q = r = s;
6a0b9d08	193	while (*q) {
6a0b9d08	194	if (*q == ':') {
e4ea58f8	195	*r = '\0';
	196	cmdline_cfg_add(para, s);
	197	r = s;
6a0b9d08	198	} else {
	199	if (*q == '\\' && q[1])
	200	q++;
e4ea58f8	201	r++ = q;
6a0b9d08	202	}
	203	q++;
	204	}
57e17355	205	*r = '\0';
e4ea58f8	206	cmdline_cfg_add(para, s);
6a0b9d08	207
	208	if (cfg_tail)
	209	cfg_tail->next = para;
	210	else
	211	cfg = para;
	212	cfg_tail = para;
	213	}
d7482997	214	break;
	215	}
	216	p = NULL; /* prevent continued processing */
	217	break;
	218	default:
	219	/*
	220	* Unrecognised option.
	221	*/
	222	{
	223	char opt[2];
	224	opt[0] = c;
	225	opt[1] = '\0';
	226	errs = TRUE, error(err_nosuchopt, opt);
	227	}
	228	}
	229	}
	230	} else {
	231	/*
	232	* A non-option argument.
	233	*/
	234	infiles[nfiles++] = p;
	235	}
	236	}
	237
	238	if (errs)
	239	exit(EXIT_FAILURE);
	240	if (nogo)
	241	exit(EXIT_SUCCESS);
	242
	243	/*
	244	* Do the work.
	245	*/
	246	if (nfiles == 0) {
	247	error(err_noinput);
	248	usage();
	249	exit(EXIT_FAILURE);
	250	}
	251
	252	{
	253	input in;
	254	paragraph sourceform, p;
	255	indexdata *idx;
	256	keywordlist *keywords;
	257
	258	in.filenames = infiles;
	259	in.nfiles = nfiles;
	260	in.currfp = NULL;
	261	in.currindex = 0;
	262	in.npushback = in.pushbacksize = 0;
	263	in.pushback = NULL;
	264	in.reportcols = reportcols;
	265	in.stack = NULL;
e5cd393f	266	in.defcharset = charset_from_locale();
d7482997	267
	268	idx = make_index();
	269
	270	sourceform = read_input(&in, idx);
	271	if (!sourceform)
	272	exit(EXIT_FAILURE);
	273
6a0b9d08	274	/*
	275	* Append the config directives acquired from the command
	276	* line.
	277	*/
	278	{
	279	paragraph *end;
	280
	281	end = sourceform;
	282	while (end && end->next)
	283	end = end->next;
	284	assert(end);
	285
	286	end->next = cfg;
	287	}
	288
d7482997	289	sfree(in.pushback);
d7482997	290
d7482997	291	sfree(infiles);
	292
	293	keywords = get_keywords(sourceform);
	294	if (!keywords)
	295	exit(EXIT_FAILURE);
	296	gen_citations(sourceform, keywords);
	297	subst_keywords(sourceform, keywords);
	298
	299	for (p = sourceform; p; p = p->next)
	300	if (p->type == para_IM)
f4551933	301	index_merge(idx, TRUE, p->keyword, p->words, &p->fpos);
d7482997	302
	303	build_index(idx);
	304
bb9e7835	305	/*
	306	* Set up attr_First / attr_Last / attr_Always, in the main
	307	* document and in the index entries.
	308	*/
	309	for (p = sourceform; p; p = p->next)
	310	mark_attr_ends(p->words);
	311	{
	312	int i;
	313	indexentry *entry;
	314
	315	for (i = 0; (entry = index234(idx->entries, i)) != NULL; i++)
	316	mark_attr_ends(entry->text);
	317	}
	318
d7482997	319	if (debug) {
	320	index_debug(idx);
	321	dbg_prtkws(keywords);
	322	dbg_prtsource(sourceform);
	323	}
	324
c8fb54d2	325	/*
43341922	326	* Select and run the pre-backends.
	327	*/
	328	prebackbits = 0;
	329	for (k = 0; k < (int)lenof(backends); k++)
	330	if (backendbits == 0 \|\| (backendbits & backends[k].bitfield))
	331	prebackbits \|= backends[k].prebackend_bitfield;
	332	for (k = 0; k < (int)lenof(pre_backends); k++)
	333	if (prebackbits & pre_backends[k].bitfield) {
	334	assert(k < (int)lenof(pre_backend_data));
	335	pre_backend_data[k] =
	336	pre_backends[k].func(sourceform, keywords, idx);
	337	}
	338
	339	/*
c8fb54d2	340	* Run the selected set of backends.
	341	*/
	342	for (k = b = 0; k < (int)lenof(backends); k++)
	343	if (b != backends[k].bitfield) {
	344	b = backends[k].bitfield;
43341922	345	if (backendbits == 0 \|\| (backendbits & b)) {
	346	void *pbd = NULL;
	347	int pbb = backends[k].prebackend_bitfield;
	348	int m;
	349
	350	for (m = 0; m < (int)lenof(pre_backends); m++)
	351	if (pbb & pre_backends[m].bitfield) {
	352	assert(m < (int)lenof(pre_backend_data));
	353	pbd = pre_backend_data[m];
	354	break;
	355	}
	356
	357	backends[k].func(sourceform, keywords, idx, pbd);
	358	}
c8fb54d2	359	}
d7482997	360
	361	free_para_list(sourceform);
	362	free_keywords(keywords);
	363	cleanup_index(idx);
	364	}
	365
	366	return 0;
	367	}
	368
	369	static void dbg_prtsource(paragraph *sourceform) {
	370	/*
	371	* Output source form in debugging format.
	372	*/
	373
	374	paragraph *p;
	375	for (p = sourceform; p; p = p->next) {
	376	wchar_t *wp;
	377	printf("para %d ", p->type);
	378	if (p->keyword) {
	379	wp = p->keyword;
	380	while (*wp) {
	381	putchar('\"');
	382	for (; *wp; wp++)
	383	putchar(*wp);
	384	putchar('\"');
	385	if (*++wp)
	386	printf(", ");
	387	}
	388	} else
	389	printf("(no keyword)");
	390	printf(" {\n");
	391	dbg_prtwordlist(1, p->words);
	392	printf("}\n");
	393	}
	394	}
	395
	396	static void dbg_prtkws(keywordlist *kws) {
	397	/*
	398	* Output keywords in debugging format.
	399	*/
	400
	401	int i;
	402	keyword *kw;
	403
	404	for (i = 0; (kw = index234(kws->keys, i)) != NULL; i++) {
	405	wchar_t *wp;
	406	printf("keyword ");
	407	wp = kw->key;
	408	while (*wp) {
	409	putchar('\"');
	410	for (; *wp; wp++)
	411	putchar(*wp);
	412	putchar('\"');
	413	if (*++wp)
	414	printf(", ");
	415	}
	416	printf(" {\n");
	417	dbg_prtwordlist(1, kw->text);
	418	printf("}\n");
	419	}
	420	}
	421
	422	static void dbg_prtwordlist(int level, word *w) {
	423	for (; w; w = w->next) {
424	wchar_t *wp;
425	printf("%sword %d ", level4, "", w->type);
426	if (w->text) {
427	printf("\"");
428	for (wp = w->text; *wp; wp++)
429	putchar(*wp);
430	printf("\"");
431	} else
432	printf("(no text)");
433	if (w->alt) {
434	printf(" alt = {\n");
435	dbg_prtwordlist(level+1, w->alt);
436	printf("%s}", level4, "");
437	}
438	printf("\n");
439	}
440	}