xref: /plan9/sys/src/cmd/postscript/text2post/text2post.c (revision 456a8764e4ea95d7aa2c2cf34e5112293070bc84)
1 #include <u.h>
2 #include <libc.h>
3 #include <ctype.h>
4 #include <bio.h>
5 #include <comments.h>
6 #include <path.h>
7 
8 #define UNKNOWNCHAR	"/sys/lib/postscript/prologues/pjw.char.ps"
9 
10 char	*optnames = "a:c:f:l:m:n:o:p:s:t:x:y:P:";
11 
12 double	aspectratio = 1.0;
13 Biobuf	inbuf, outbuf;
14 Biobuf	*bin, *bout;
15 int	char_no = 0;		/* character to be done on a line */
16 int	copies = 1;
17 int	formsperpage = 1;
18 int	in_string;		/* flag: we are inside a Postscript string */
19 int	landscape = 0;
20 int	line_no = 0;		/* line number on a page */
21 int	linesperpage = 66;
22 double	magnification = 1.0;
23 int	page_no = 0;		/* page number in a document */
24 int	pages_printed;
25 char	*passthrough = 0;
26 int	pointsize = 10;
27 int	spaces = 0;
28 int	tabs = 0;
29 double	xoffset = .25;
30 double	yoffset = .25;
31 
32 static int pplistmaxsize = 0;
33 
34 uchar *pplist = 0;		/* bitmap list for storing pages to print */
35 
36 struct strtab {
37 	int	size;
38 	char	*str;
39 	int	used;
40 };
41 
42 struct strtab charcode[256] = {
43 	{4, "\\000"}, {4, "\\001"}, {4, "\\002"}, {4, "\\003"},
44 	{4, "\\004"}, {4, "\\005"}, {4, "\\006"}, {4, "\\007"},
45 	{4, "\\010"}, {4, "\\011"}, {4, "\\012"}, {4, "\\013"},
46 	{4, "\\014"}, {4, "\\015"}, {4, "\\016"}, {4, "\\017"},
47 	{4, "\\020"}, {4, "\\021"}, {4, "\\022"}, {4, "\\023"},
48 	{4, "\\024"}, {4, "\\025"}, {4, "\\026"}, {4, "\\027"},
49 	{4, "\\030"}, {4, "\\031"}, {4, "\\032"}, {4, "\\033"},
50 	{4, "\\034"}, {4, "\\035"}, {4, "\\036"}, {4, "\\037"},
51 	{1, " "}, {1, "!"}, {1, "\""}, {1, "#"},
52 	{1, "$"}, {1, "%"}, {1, "&"}, {1, "'"},
53 	{2, "\\("}, {2, "\\)"}, {1, "*"}, {1, "+"},
54 	{1, ","}, {1, "-"}, {1, "."}, {1, "/"},
55 	{1, "0"}, {1, "1"}, {1, "2"}, {1, "3"},
56 	{1, "4"}, {1, "5"}, {1, "6"}, {1, "7"},
57 	{1, "8"}, {1, "9"}, {1, ":"}, {1, ";"},
58 	{1, "<"}, {1, "="}, {1, ">"}, {1, "?"},
59 	{1, "@"}, {1, "A"}, {1, "B"}, {1, "C"},
60 	{1, "D"}, {1, "E"}, {1, "F"}, {1, "G"},
61 	{1, "H"}, {1, "I"}, {1, "J"}, {1, "K"},
62 	{1, "L"}, {1, "M"}, {1, "N"}, {1, "O"},
63 	{1, "P"}, {1, "Q"}, {1, "R"}, {1, "S"},
64 	{1, "T"}, {1, "U"}, {1, "V"}, {1, "W"},
65 	{1, "X"}, {1, "Y"}, {1, "Z"}, {1, "["},
66 	{2, "\\\\"}, {1, "]"}, {1, "^"}, {1, "_"},
67 	{1, "`"}, {1, "a"}, {1, "b"}, {1, "c"},
68 	{1, "d"}, {1, "e"}, {1, "f"}, {1, "g"},
69 	{1, "h"}, {1, "i"}, {1, "j"}, {1, "k"},
70 	{1, "l"}, {1, "m"}, {1, "n"}, {1, "o"},
71 	{1, "p"}, {1, "q"}, {1, "r"}, {1, "s"},
72 	{1, "t"}, {1, "u"}, {1, "v"}, {1, "w"},
73 	{1, "x"}, {1, "y"}, {1, "z"}, {1, "{"},
74 	{1, "|"}, {1, "}"}, {1, "~"}, {4, "\\177"},
75 	{4, "\\200"}, {4, "\\201"}, {4, "\\202"}, {4, "\\203"},
76 	{4, "\\204"}, {4, "\\205"}, {4, "\\206"}, {4, "\\207"},
77 	{4, "\\210"}, {4, "\\211"}, {4, "\\212"}, {4, "\\213"},
78 	{4, "\\214"}, {4, "\\215"}, {4, "\\216"}, {4, "\\217"},
79 	{4, "\\220"}, {4, "\\221"}, {4, "\\222"}, {4, "\\223"},
80 	{4, "\\224"}, {4, "\\225"}, {4, "\\226"}, {4, "\\227"},
81 	{4, "\\230"}, {4, "\\231"}, {4, "\\232"}, {4, "\\233"},
82 	{4, "\\234"}, {4, "\\235"}, {4, "\\236"}, {4, "\\237"},
83 	{4, "\\240"}, {4, "\\241"}, {4, "\\242"}, {4, "\\243"},
84 	{4, "\\244"}, {4, "\\245"}, {4, "\\246"}, {4, "\\247"},
85 	{4, "\\250"}, {4, "\\251"}, {4, "\\252"}, {4, "\\253"},
86 	{4, "\\254"}, {4, "\\255"}, {4, "\\256"}, {4, "\\257"},
87 	{4, "\\260"}, {4, "\\261"}, {4, "\\262"}, {4, "\\263"},
88 	{4, "\\264"}, {4, "\\265"}, {4, "\\266"}, {4, "\\267"},
89 	{4, "\\270"}, {4, "\\271"}, {4, "\\272"}, {4, "\\273"},
90 	{4, "\\274"}, {4, "\\275"}, {4, "\\276"}, {4, "\\277"},
91 	{4, "\\300"}, {4, "\\301"}, {4, "\\302"}, {4, "\\303"},
92 	{4, "\\304"}, {4, "\\305"}, {4, "\\306"}, {4, "\\307"},
93 	{4, "\\310"}, {4, "\\311"}, {4, "\\312"}, {4, "\\313"},
94 	{4, "\\314"}, {4, "\\315"}, {4, "\\316"}, {4, "\\317"},
95 	{4, "\\320"}, {4, "\\321"}, {4, "\\322"}, {4, "\\323"},
96 	{4, "\\324"}, {4, "\\325"}, {4, "\\326"}, {4, "\\327"},
97 	{4, "\\330"}, {4, "\\331"}, {4, "\\332"}, {4, "\\333"},
98 	{4, "\\334"}, {4, "\\335"}, {4, "\\336"}, {4, "\\337"},
99 	{4, "\\340"}, {4, "\\341"}, {4, "\\342"}, {4, "\\343"},
100 	{4, "\\344"}, {4, "\\345"}, {4, "\\346"}, {4, "\\347"},
101 	{4, "\\350"}, {4, "\\351"}, {4, "\\352"}, {4, "\\353"},
102 	{4, "\\354"}, {4, "\\355"}, {4, "\\356"}, {4, "\\357"},
103 	{4, "\\360"}, {4, "\\361"}, {4, "\\362"}, {4, "\\363"},
104 	{4, "\\364"}, {4, "\\365"}, {4, "\\366"}, {4, "\\367"},
105 	{4, "\\370"}, {4, "\\371"}, {4, "\\372"}, {4, "\\373"},
106 	{4, "\\374"}, {4, "\\375"}, {4, "\\376"}, {4, "\\377"}
107 };
108 
109 #define FONTABSIZE 0x27
110 
111 struct strtab fontname[FONTABSIZE] = {
112 	{19, "LucidaSansUnicode00", 0},
113 	{19, "LucidaSansUnicode01", 0},
114 	{19, "LucidaSansUnicode02", 0},
115 	{19, "LucidaSansUnicode03", 0},
116 	{19, "LucidaSansUnicode04", 0},
117 	{19, "LucidaSansUnicode05", 0},
118 	{0, "", 0},
119 	{0, "", 0},
120 	{0, "", 0},
121 	{0, "", 0},
122 	{0, "", 0},
123 	{0, "", 0},
124 	{0, "", 0},
125 	{0, "", 0},
126 	{0, "", 0},
127 	{0, "", 0},
128 	{0, "", 0},
129 	{0, "", 0},
130 	{0, "", 0},
131 	{0, "", 0},
132 	{0, "", 0},
133 	{0, "", 0},
134 	{0, "", 0},
135 	{0, "", 0},
136 	{0, "", 0},
137 	{0, "", 0},
138 	{0, "", 0},
139 	{0, "", 0},
140 	{0, "", 0},
141 	{0, "", 0},
142 	{0, "", 0},
143 	{0, "", 0},
144 	{19, "LucidaSansUnicode20", 0},
145 	{19, "LucidaSansUnicode21", 0},
146 	{19, "LucidaSansUnicode22", 0},
147 	{0, "", 0},
148 	{19, "LucidaSansUnicode24", 0},
149 	{19, "LucidaSansUnicode25", 0},
150 	{7, "Courier", 0}
151 };
152 
153 /* This was taken from postprint */
154 
155 int
cat(char * filename)156 cat(char *filename) {
157 	int n;
158 	char buf[Bsize];
159 	Biobuf *bfile;
160 
161 	if ((bfile = Bopen(filename, OREAD)) == nil)
162 		return(1);
163 	while ((n = Bread(bfile, buf, Bsize)) > 0)
164 		if (Bwrite(bout, buf, n) != n)
165 			break;
166 	Bterm(bfile);
167 	if (n != 0)
168 		return(1);
169 	return(0);
170 }
171 
172 void
prologues(void)173 prologues(void) {
174 	char *ts;
175 	int tabstop;
176 
177 	Bprint(bout, "%s", CONFORMING);
178 	Bprint(bout, "%s %s\n", VERSION, PROGRAMVERSION);
179 	Bprint(bout, "%s %s\n", DOCUMENTFONTS, ATEND);
180 	Bprint(bout, "%s %s\n", PAGES, ATEND);
181 	Bprint(bout, "%s", ENDCOMMENTS);
182 
183 	if (cat(POSTPRINT))
184 		sysfatal("can't read %s: %r", POSTPRINT);
185 
186 	if (DOROUND)
187 		cat(ROUNDPAGE);
188 
189 	tabstop = 0;
190 	ts = getenv("tabstop");
191 	if(ts != nil)
192 		tabstop = strtol(ts, nil, 0);
193 	if(tabstop == 0)
194 		tabstop = 8;
195 	Bprint(bout, "/f {findfont pointsize scalefont setfont} bind def\n");
196 	Bprint(bout, "/tabwidth /Courier f (");
197 	while(tabstop--)
198 		Bputc(bout, 'n');
199 	Bprint(bout, ") stringwidth pop def\n");
200 	Bprint(bout, "/tab {tabwidth 0 ne {currentpoint 3 1 roll exch tabwidth mul add tabwidth\n");
201 	Bprint(bout, "\tdiv truncate tabwidth mul exch moveto} if} bind def\n");
202 	Bprint(bout, "/spacewidth /%s f ( ) stringwidth pop def\n", fontname[0].str);
203 	Bprint(bout, "/sp {spacewidth mul 0 rmoveto} bind def\n");
204 	Bprint(bout, "%s", ENDPROLOG);
205 	Bprint(bout, "%s", BEGINSETUP);
206 	Bprint(bout, "mark\n");
207 
208 	if (formsperpage > 1) {
209 		Bprint(bout, "%s %d\n", FORMSPERPAGE, formsperpage);
210 		Bprint(bout, "/formsperpage %d def\n", formsperpage);
211 	}
212 	if (aspectratio != 1) Bprint(bout, "/aspectratio %g def\n", aspectratio);
213 	if (copies != 1) Bprint(bout, "/#copies %d store\n", copies);
214 	if (landscape) Bprint(bout, "/landscape true def\n");
215 	if (magnification != 1) Bprint(bout, "/magnification %g def\n", magnification);
216 	if (pointsize != 10) Bprint(bout, "/pointsize %d def\n", pointsize);
217 	if (xoffset != .25) Bprint(bout, "/xoffset %g def\n", xoffset);
218 	if (yoffset != .25) Bprint(bout, "/yoffset %g def\n", yoffset);
219 	cat(ENCODINGDIR"/Latin1.enc");
220 	if (passthrough != 0) Bprint(bout, "%s\n", passthrough);
221 	Bprint(bout, "setup\n");
222 	if (formsperpage > 1) {
223 		cat(FORMFILE);
224 		Bprint(bout, "%d setupforms \n", formsperpage);
225 	}
226 	if (cat(UNKNOWNCHAR))
227 		fprint(2, "cannot open %s: %r\n", UNKNOWNCHAR);
228 	Bprint(bout, "%s", ENDSETUP);
229 }
230 
231 int
pageon(void)232 pageon(void) {
233 	if (pplist == 0 && page_no != 0)
234 		return(1);	/* no page list, print all pages */
235 	if (page_no/8 < pplistmaxsize && (pplist[page_no/8] & 1<<(page_no%8)))
236 		return(1);
237 	else
238 		return(0);
239 }
240 
241 void
startpage(void)242 startpage(void) {
243 	++char_no;
244 	++line_no;
245 	++page_no;
246 	if (pageon()) {
247 		++pages_printed;
248 		Bprint(bout, "%s %d %d\n", PAGE, page_no, pages_printed);
249 		Bprint(bout, "/saveobj save def\n");
250 		Bprint(bout, "mark\n");
251 		Bprint(bout, "%d pagesetup\n", pages_printed);
252 	}
253 }
254 
255 void
endpage(void)256 endpage(void) {
257 	line_no = 0;
258 	char_no = 0;
259 	if (pageon()) {
260 		Bprint(bout, "cleartomark\n");
261 		Bprint(bout, "showpage\n");
262 		Bprint(bout, "saveobj restore\n");
263 		Bprint(bout, "%s %d %d\n", ENDPAGE, page_no, pages_printed);
264 	}
265 }
266 
267 void
startstring(void)268 startstring(void) {
269 	if (!in_string) {
270 		if (pageon()) Bprint(bout, "(");
271 		in_string = 1;
272 	}
273 }
274 
275 void
endstring(void)276 endstring(void) {
277 	if (in_string) {
278 		if (pageon()) Bprint(bout, ") show ");
279 		in_string = 0;
280 	}
281 }
282 
283 void
prspace(void)284 prspace(void) {
285 	if (spaces) {
286 		endstring();
287 		if (pageon()) Bprint(bout, "%d sp ", spaces);
288 		spaces = 0;
289 	}
290 }
291 
292 void
prtab(void)293 prtab(void) {
294 	if (tabs) {
295 		endstring();
296 		if (pageon()) Bprint(bout, "%d tab ", tabs);
297 		tabs = 0;
298 	}
299 }
300 
301 void
txt2post(void)302 txt2post(void) {
303 	int lastfont = -1;
304 	int lastchar = -1;
305 	int thisfont, thischar;
306 	long r;
307 
308 	in_string = 0;
309 	char_no = 0;
310 	line_no = 0;
311 	page_no = 0;
312 	spaces = 0;
313 	fontname[0].used++;
314 	while ((r = Bgetrune(bin)) >= 0) {
315 		thischar = r & 0xff;
316 		thisfont = (r>>8) & 0xff;
317 
318 		if (line_no == 0 && char_no == 0)
319 			startpage();
320 
321 		if (line_no == 1 && char_no == 1) {
322 			if (pageon()) Bprint(bout, " /%s f\n", fontname[thisfont].str);
323 			lastfont = thisfont;
324 		}
325 
326 		switch (r) {
327 		case ' ':
328 			prtab();
329 			if (lastfont > 0) {
330 				spaces++;
331 				continue;
332 			}
333 			break;
334 		case '\n':
335 		case '\f':
336 			startstring();
337 			if (pageon()) Bprint(bout, ")l\n");
338 			char_no = 1;
339 			in_string = 0;
340 			spaces = 0;
341 			tabs = 0;
342 			if (++line_no > linesperpage || r == '\f')
343 				endpage();
344 			lastchar = -1;
345 			continue;
346 		case '\t':
347 			prspace();
348 			tabs++;
349 			char_no++;
350 			lastchar = -1;
351 			continue;
352 		case '\b':
353 			/* just toss out backspaces for now */
354 			if (lastchar != -1) {
355 				endstring();
356 				if (pageon()) Bprint(bout, "(%s) stringwidth pop neg 0 rmoveto ", charcode[lastchar].str);
357 			}
358 			char_no++;
359 			lastchar = -1;
360 			continue;
361 		}
362 
363 		/* do something if font is out of table range */
364 		if (thisfont>=FONTABSIZE || fontname[thisfont].size == 0) {
365 			prspace();
366 			prtab();
367 			endstring();
368 			Bprint(bout, "pw ");
369 			char_no++;
370 			lastchar = -1;
371 			continue;
372 		}
373 
374 		if (thisfont != lastfont) {
375 			endstring();
376 			if (pageon()) {
377 				Bprint(bout, "/%s f\n", fontname[thisfont].str);
378 			}
379 			fontname[thisfont].used++;
380 		}
381 		prspace();
382 		prtab();
383 		startstring();
384 		if (pageon()) Bprint(bout, "%s", charcode[thischar].str);
385 /*		if (pageon()) Bprint(bout, "%2.2x", thischar); /* try hex strings*/
386 		char_no++;
387 		lastchar = thischar;
388 		lastfont = thisfont;
389 	}
390 	if (line_no != 0 || char_no != 0) {
391 		if (char_no != 1) {
392 			fprint(2, "premature EOF: newline appended\n");
393 			startstring();
394 			if (pageon()) Bprint(bout, ")l\n");
395 		}
396 		endpage();
397 	}
398 }
399 
400 void
pagelist(char * list)401 pagelist(char *list) {
402 	char c;
403 	int n, state, start;
404 	unsigned m;
405 
406 	if (list == 0)
407 		return;
408 	start = 0;
409 	state = 1;
410 	while ((c=*list) != '\0') {
411 		n = 0;
412 		while (isdigit(c)) {
413 			n = n * 10 + c - '0';
414 			c = *++list;
415 		}
416 		switch (state) {
417 		case 1:
418 			start = n;
419 			/* fall through */
420 		case 2:
421 			if (n/8+1 > pplistmaxsize) {
422 				pplistmaxsize = n/8+1;
423 				if ((pplist = realloc(pplist, n/8+1)) == 0)
424 					sysfatal("malloc");
425 			}
426 			for (m=start; m<=n; m++)
427 				pplist[m/8] |= 1<<(m%8);
428 			break;
429 		}
430 		switch (c) {
431 		case '-':
432 			state = 2;
433 			list++;
434 			break;
435 		case ',':
436 			state = 1;
437 			list++;
438 			break;
439 		case '\0':
440 			break;
441 		}
442 	}
443 }
444 
445 void
finish(void)446 finish(void) {
447 	int i;
448 
449 	Bprint(bout, "%s", TRAILER);
450 	Bprint(bout, "done\n");
451 	Bprint(bout, "%s", DOCUMENTFONTS);
452 
453 	for (i=0; i<FONTABSIZE; i++)
454 		if (fontname[i].used)
455 			Bprint(bout, " %s", fontname[i].str);
456 	Bprint(bout, "\n");
457 
458 	Bprint(bout, "%s %d\n", PAGES, pages_printed);
459 }
460 
461 void
main(int argc,char * argv[])462 main(int argc, char *argv[]) {
463 	int i;
464 	char *t;
465 
466 	bin = &inbuf;
467 	bout = &outbuf;
468 	if (Binit(bout, 1, OWRITE) == Beof)
469 		sysfatal("Binit");
470 
471 	ARGBEGIN{
472 	case 'a':			/* aspect ratio */
473 		aspectratio = atof(ARGF());
474 		break;
475 	case 'c':			/* copies */
476 		copies = atoi(ARGF());
477 		break;
478 	case 'f':			/* primary font, for now */
479 		t = ARGF();
480 		fontname[0].str = malloc(strlen(t)+1);
481 		strcpy(fontname[0].str, t);
482 		break;
483 	case 'l':			/* lines per page */
484 		linesperpage = atoi(ARGF());
485 		break;
486 	case 'm':			/* magnification */
487 		magnification = atof(ARGF());
488 		break;
489 	case 'n':			/* forms per page */
490 		formsperpage = atoi(ARGF());
491 		break;
492 	case 'o':			/* output page list */
493 		pagelist(ARGF());
494 		break;
495 	case 'p':			/* landscape or portrait mode */
496 		if ( ARGF()[0] == 'l' )
497 			landscape = 1;
498 		else
499 			landscape = 0;
500 		break;
501 	case 's':			/* point size */
502 		pointsize = atoi(ARGF());
503 		break;
504 	case 'x':			/* shift things horizontally */
505 		xoffset = atof(ARGF());
506 		break;
507 
508 	case 'y':			/* and vertically on the page */
509 		yoffset = atof(ARGF());
510 		break;
511 	case 'P':			/* PostScript pass through */
512 		t = ARGF();
513 		i = strlen(t) + 1;
514 		passthrough = malloc(i);
515 		if (passthrough == 0)
516 			sysfatal("malloc");
517 		strncpy(passthrough, t, i);
518 		break;
519 	default:
520 		fprint(2, "unknown option %C\n", ARGC());
521 		break;
522 	}ARGEND;
523 
524 	prologues();
525 	if (argc <= 0) {
526 		if (Binit(bin, 0, OREAD) == Beof)
527 			sysfatal("cannot Binit stdin");
528 		txt2post();
529 	}
530 	for (i=0; i<argc; i++) {
531 		bin = Bopen(argv[i], OREAD);
532 		if (bin == nil) {
533 			fprint(2, "cannot open %s: %r\n", argv[i]);
534 			continue;
535 		}
536 		txt2post();
537 		Bterm(bin);
538 	}
539 	finish();
540 	exits("");
541 }
542