[sgt/halibut] / main.c

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

#include <assert.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", xhtml_backend, xhtml_config_filename, 0x0002, 0},
    {"html", xhtml_backend, xhtml_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];

    /*
     * Set up initial (default) parameters.
     */
    infiles = mknewa(char *, argc);
    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++;
			    }
			    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 = CS_ASCII;

	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);

	mark_attr_ends(sourceform);

	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);

	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>
d7482997	6	#include <stdio.h>
	7	#include <stdlib.h>
	8	#include "halibut.h"
	9
	10	static void dbg_prtsource(paragraph *sourceform);
	11	static void dbg_prtwordlist(int level, word *w);
	12	static void dbg_prtkws(keywordlist *kws);
	13
43341922	14	static const struct pre_backend {
	15	void (func)(paragraph , keywordlist , indexdata *);
	16	int bitfield;
	17	} pre_backends[] = {
	18	{paper_pre_backend, 0x0001}
	19	};
	20
c8fb54d2	21	static const struct backend {
c8fb54d2	22	char *name;
43341922	23	void (func)(paragraph , keywordlist , indexdata , void *);
ba9c1487	24	paragraph (filename)(char *filename);
43341922	25	int bitfield, prebackend_bitfield;
c8fb54d2	26	} backends[] = {
43341922	27	{"text", text_backend, text_config_filename, 0x0001, 0},
	28	{"xhtml", xhtml_backend, xhtml_config_filename, 0x0002, 0},
	29	{"html", xhtml_backend, xhtml_config_filename, 0x0002, 0},
	30	{"hlp", whlp_backend, whlp_config_filename, 0x0004, 0},
	31	{"whlp", whlp_backend, whlp_config_filename, 0x0004, 0},
	32	{"winhelp", whlp_backend, whlp_config_filename, 0x0004, 0},
	33	{"man", man_backend, man_config_filename, 0x0008, 0},
	34	{"info", info_backend, info_config_filename, 0x0010, 0},
	35	{"ps", ps_backend, ps_config_filename, 0x0020, 0x0001},
	36	{"pdf", pdf_backend, pdf_config_filename, 0x0040, 0x0001},
c8fb54d2	37	};
c8fb54d2	38
d7482997	39	int main(int argc, char **argv) {
d7482997	40	char **infiles;
d7482997	41	int nfiles;
	42	int nogo;
	43	int errs;
	44	int reportcols;
	45	int debug;
43341922	46	int backendbits, prebackbits;
c8fb54d2	47	int k, b;
6a0b9d08	48	paragraph cfg, cfg_tail;
43341922	49	void *pre_backend_data[16];
d7482997	50
	51	/*
	52	* Set up initial (default) parameters.
	53	*/
	54	infiles = mknewa(char *, argc);
d7482997	55	nfiles = 0;
	56	nogo = errs = FALSE;
	57	reportcols = 0;
	58	debug = 0;
c8fb54d2	59	backendbits = 0;
6a0b9d08	60	cfg = cfg_tail = NULL;
d7482997	61
	62	if (argc == 1) {
	63	usage();
	64	exit(EXIT_SUCCESS);
	65	}
	66
	67	/*
	68	* Parse command line arguments.
	69	*/
	70	while (--argc) {
	71	char p = ++argv;
	72	if (*p == '-') {
	73	/*
	74	* An option.
	75	*/
	76	while (p && *++p) {
	77	char c = *p;
	78	switch (c) {
	79	case '-':
	80	/*
	81	* Long option.
	82	*/
	83	{
	84	char opt, val;
	85	opt = p++; /* opt will have _one_ leading - */
	86	while (p && p != '=')
	87	p++; /* find end of option */
	88	if (*p == '=') {
	89	*p++ = '\0';
	90	val = p;
	91	} else
	92	val = NULL;
c8fb54d2	93
	94	assert(opt[0] == '-');
	95	for (k = 0; k < (int)lenof(backends); k++)
	96	if (!strcmp(opt+1, backends[k].name)) {
	97	backendbits \|= backends[k].bitfield;
ba9c1487	98	if (val) {
	99	paragraph *p = backends[k].filename(val);
	100	assert(p);
	101	if (cfg_tail)
	102	cfg_tail->next = p;
	103	else
	104	cfg = p;
	105	while (p->next)
	106	p = p->next;
	107	cfg_tail = p;
	108	}
c8fb54d2	109	break;
	110	}
	111	if (k < (int)lenof(backends)) {
	112	/* do nothing */;
	113	} else if (!strcmp(opt, "-help")) {
d7482997	114	help();
	115	nogo = TRUE;
	116	} else if (!strcmp(opt, "-version")) {
	117	showversion();
	118	nogo = TRUE;
	119	} else if (!strcmp(opt, "-licence") \|\|
	120	!strcmp(opt, "-license")) {
	121	licence();
	122	nogo = TRUE;
d7482997	123	} else if (!strcmp(opt, "-precise")) {
	124	reportcols = 1;
	125	} else {
	126	errs = TRUE, error(err_nosuchopt, opt);
	127	}
	128	}
	129	p = NULL;
	130	break;
	131	case 'h':
	132	case 'V':
	133	case 'L':
	134	case 'P':
	135	case 'd':
	136	/*
	137	* Option requiring no parameter.
	138	*/
	139	switch (c) {
	140	case 'h':
	141	help();
	142	nogo = TRUE;
	143	break;
	144	case 'V':
	145	showversion();
	146	nogo = TRUE;
	147	break;
	148	case 'L':
	149	licence();
	150	nogo = TRUE;
	151	break;
	152	case 'P':
	153	reportcols = 1;
	154	break;
	155	case 'd':
	156	debug = TRUE;
	157	break;
	158	}
	159	break;
6a0b9d08	160	case 'C':
d7482997	161	/*
	162	* Option requiring parameter.
	163	*/
	164	p++;
	165	if (!*p && argc > 1)
	166	--argc, p = *++argv;
	167	else if (!*p) {
	168	char opt[2];
	169	opt[0] = c;
	170	opt[1] = '\0';
	171	errs = TRUE, error(err_optnoarg, opt);
	172	}
	173	/*
	174	* Now c is the option and p is the parameter.
	175	*/
	176	switch (c) {
6a0b9d08	177	case 'C':
	178	/*
	179	* -C means we split our argument up into
	180	* colon-separated chunks and assemble them
	181	* into a config paragraph.
	182	*/
	183	{
e4ea58f8	184	char s = dupstr(p), q, *r;
6a0b9d08	185	paragraph *para;
6a0b9d08	186
e4ea58f8	187	para = cmdline_cfg_new();
6a0b9d08	188
e4ea58f8	189	q = r = s;
6a0b9d08	190	while (*q) {
6a0b9d08	191	if (*q == ':') {
e4ea58f8	192	*r = '\0';
	193	cmdline_cfg_add(para, s);
	194	r = s;
6a0b9d08	195	} else {
	196	if (*q == '\\' && q[1])
	197	q++;
e4ea58f8	198	r++ = q;
6a0b9d08	199	}
	200	q++;
	201	}
e4ea58f8	202	cmdline_cfg_add(para, s);
6a0b9d08	203
	204	if (cfg_tail)
	205	cfg_tail->next = para;
	206	else
	207	cfg = para;
	208	cfg_tail = para;
	209	}
d7482997	210	break;
	211	}
	212	p = NULL; /* prevent continued processing */
	213	break;
	214	default:
	215	/*
	216	* Unrecognised option.
	217	*/
	218	{
	219	char opt[2];
	220	opt[0] = c;
	221	opt[1] = '\0';
	222	errs = TRUE, error(err_nosuchopt, opt);
	223	}
	224	}
	225	}
	226	} else {
	227	/*
	228	* A non-option argument.
	229	*/
	230	infiles[nfiles++] = p;
	231	}
	232	}
	233
	234	if (errs)
	235	exit(EXIT_FAILURE);
	236	if (nogo)
	237	exit(EXIT_SUCCESS);
	238
	239	/*
	240	* Do the work.
	241	*/
	242	if (nfiles == 0) {
	243	error(err_noinput);
	244	usage();
	245	exit(EXIT_FAILURE);
	246	}
	247
	248	{
	249	input in;
	250	paragraph sourceform, p;
	251	indexdata *idx;
	252	keywordlist *keywords;
	253
	254	in.filenames = infiles;
	255	in.nfiles = nfiles;
	256	in.currfp = NULL;
	257	in.currindex = 0;
	258	in.npushback = in.pushbacksize = 0;
	259	in.pushback = NULL;
	260	in.reportcols = reportcols;
	261	in.stack = NULL;
e34ba5c3	262	in.defcharset = CS_ASCII;
d7482997	263
	264	idx = make_index();
	265
	266	sourceform = read_input(&in, idx);
	267	if (!sourceform)
	268	exit(EXIT_FAILURE);
	269
6a0b9d08	270	/*
	271	* Append the config directives acquired from the command
	272	* line.
	273	*/
	274	{
	275	paragraph *end;
	276
	277	end = sourceform;
	278	while (end && end->next)
	279	end = end->next;
	280	assert(end);
	281
	282	end->next = cfg;
	283	}
	284
d7482997	285	sfree(in.pushback);
	286
	287	mark_attr_ends(sourceform);
	288
	289	sfree(infiles);
	290
	291	keywords = get_keywords(sourceform);
	292	if (!keywords)
	293	exit(EXIT_FAILURE);
	294	gen_citations(sourceform, keywords);
	295	subst_keywords(sourceform, keywords);
	296
	297	for (p = sourceform; p; p = p->next)
	298	if (p->type == para_IM)
f4551933	299	index_merge(idx, TRUE, p->keyword, p->words, &p->fpos);
d7482997	300
	301	build_index(idx);
	302
	303	if (debug) {
	304	index_debug(idx);
	305	dbg_prtkws(keywords);
	306	dbg_prtsource(sourceform);
	307	}
	308
c8fb54d2	309	/*
43341922	310	* Select and run the pre-backends.
	311	*/
	312	prebackbits = 0;
	313	for (k = 0; k < (int)lenof(backends); k++)
	314	if (backendbits == 0 \|\| (backendbits & backends[k].bitfield))
	315	prebackbits \|= backends[k].prebackend_bitfield;
	316	for (k = 0; k < (int)lenof(pre_backends); k++)
	317	if (prebackbits & pre_backends[k].bitfield) {
	318	assert(k < (int)lenof(pre_backend_data));
	319	pre_backend_data[k] =
	320	pre_backends[k].func(sourceform, keywords, idx);
	321	}
	322
	323	/*
c8fb54d2	324	* Run the selected set of backends.
	325	*/
	326	for (k = b = 0; k < (int)lenof(backends); k++)
	327	if (b != backends[k].bitfield) {
	328	b = backends[k].bitfield;
43341922	329	if (backendbits == 0 \|\| (backendbits & b)) {
	330	void *pbd = NULL;
	331	int pbb = backends[k].prebackend_bitfield;
	332	int m;
	333
	334	for (m = 0; m < (int)lenof(pre_backends); m++)
	335	if (pbb & pre_backends[m].bitfield) {
	336	assert(m < (int)lenof(pre_backend_data));
	337	pbd = pre_backend_data[m];
	338	break;
	339	}
	340
	341	backends[k].func(sourceform, keywords, idx, pbd);
	342	}
c8fb54d2	343	}
d7482997	344
	345	free_para_list(sourceform);
	346	free_keywords(keywords);
	347	cleanup_index(idx);
	348	}
	349
	350	return 0;
	351	}
	352
	353	static void dbg_prtsource(paragraph *sourceform) {
	354	/*
	355	* Output source form in debugging format.
	356	*/
	357
	358	paragraph *p;
	359	for (p = sourceform; p; p = p->next) {
	360	wchar_t *wp;
	361	printf("para %d ", p->type);
	362	if (p->keyword) {
	363	wp = p->keyword;
	364	while (*wp) {
	365	putchar('\"');
	366	for (; *wp; wp++)
	367	putchar(*wp);
	368	putchar('\"');
	369	if (*++wp)
	370	printf(", ");
	371	}
	372	} else
	373	printf("(no keyword)");
	374	printf(" {\n");
	375	dbg_prtwordlist(1, p->words);
	376	printf("}\n");
	377	}
	378	}
	379
	380	static void dbg_prtkws(keywordlist *kws) {
	381	/*
	382	* Output keywords in debugging format.
	383	*/
	384
	385	int i;
	386	keyword *kw;
	387
	388	for (i = 0; (kw = index234(kws->keys, i)) != NULL; i++) {
	389	wchar_t *wp;
	390	printf("keyword ");
	391	wp = kw->key;
	392	while (*wp) {
	393	putchar('\"');
	394	for (; *wp; wp++)
	395	putchar(*wp);
	396	putchar('\"');
	397	if (*++wp)
	398	printf(", ");
	399	}
	400	printf(" {\n");
	401	dbg_prtwordlist(1, kw->text);
	402	printf("}\n");
	403	}
	404	}
	405
	406	static void dbg_prtwordlist(int level, word *w) {
	407	for (; w; w = w->next) {
408	wchar_t *wp;
409	printf("%sword %d ", level4, "", w->type);
410	if (w->text) {
411	printf("\"");
412	for (wp = w->text; *wp; wp++)
413	putchar(*wp);
414	printf("\"");
415	} else
416	printf("(no text)");
417	if (w->alt) {
418	printf(" alt = {\n");
419	dbg_prtwordlist(level+1, w->alt);
420	printf("%s}", level4, "");
421	}
422	printf("\n");
423	}
424	}