[sgt/halibut] / bk_text.c

/*
 * text backend for Halibut
 */

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

typedef enum { LEFT, LEFTPLUS, CENTRE } alignment;
typedef struct {
    alignment align;
    int just_numbers;
    wchar_t underline;
    wchar_t *number_suffix;
} alignstruct;

typedef struct {
    int indent, indent_code;
    int listindentbefore, listindentafter;
    int width;
    alignstruct atitle, achapter, *asect;
    int nasect;
    int include_version_id;
    int indent_preambles;
    word bullet;
} textconfig;

static int text_convert(wchar_t *, char **);

static void text_heading(FILE *, word *, word *, word *, alignstruct, int,int);
static void text_rule(FILE *, int, int);
static void text_para(FILE *, word *, char *, word *, int, int, int);
static void text_codepara(FILE *, word *, int, int);
static void text_versionid(FILE *, word *);

static alignment utoalign(wchar_t *p) {
    if (!ustricmp(p, L"centre") || !ustricmp(p, L"center"))
	return CENTRE;
    if (!ustricmp(p, L"leftplus"))
	return LEFTPLUS;
    return LEFT;
}

static textconfig text_configure(paragraph *source) {
    textconfig ret;

    /*
     * Non-negotiables.
     */
    ret.bullet.next = NULL;
    ret.bullet.alt = NULL;
    ret.bullet.type = word_Normal;
    ret.atitle.just_numbers = FALSE;   /* ignored */

    /*
     * Defaults.
     */
    ret.indent = 7;
    ret.indent_code = 2;
    ret.listindentbefore = 1;
    ret.listindentafter = 3;
    ret.width = 68;
    ret.atitle.align = CENTRE;
    ret.atitle.underline = L'=';
    ret.achapter.align = LEFT;
    ret.achapter.just_numbers = FALSE;
    ret.achapter.number_suffix = ustrdup(L": ");
    ret.achapter.underline = L'-';
    ret.nasect = 1;
    ret.asect = mknewa(alignstruct, ret.nasect);
    ret.asect[0].align = LEFTPLUS;
    ret.asect[0].just_numbers = TRUE;
    ret.asect[0].number_suffix = ustrdup(L" ");
    ret.asect[0].underline = L'\0';
    ret.include_version_id = TRUE;
    ret.indent_preambles = FALSE;
    ret.bullet.text = ustrdup(L"-");

    for (; source; source = source->next) {
	if (source->type == para_Config) {
	    if (!ustricmp(source->keyword, L"text-indent")) {
		ret.indent = utoi(uadv(source->keyword));
	    } else if (!ustricmp(source->keyword, L"text-indent-code")) {
		ret.indent_code = utoi(uadv(source->keyword));
	    } else if (!ustricmp(source->keyword, L"text-width")) {
		ret.width = utoi(uadv(source->keyword));
	    } else if (!ustricmp(source->keyword, L"text-list-indent")) {
		ret.listindentbefore = utoi(uadv(source->keyword));
	    } else if (!ustricmp(source->keyword, L"text-listitem-indent")) {
		ret.listindentafter = utoi(uadv(source->keyword));
	    } else if (!ustricmp(source->keyword, L"text-chapter-align")) {
		ret.achapter.align = utoalign(uadv(source->keyword));
	    } else if (!ustricmp(source->keyword, L"text-chapter-underline")) {
		ret.achapter.underline = *uadv(source->keyword);
	    } else if (!ustricmp(source->keyword, L"text-chapter-numeric")) {
		ret.achapter.just_numbers = utob(uadv(source->keyword));
	    } else if (!ustricmp(source->keyword, L"text-chapter-suffix")) {
		ret.achapter.number_suffix = ustrdup(uadv(source->keyword));
	    } else if (!ustricmp(source->keyword, L"text-section-align")) {
		wchar_t *p = uadv(source->keyword);
		int n = 0;
		if (uisdigit(*p)) {
		    n = utoi(p);
		    p = uadv(p);
		}
		if (n >= ret.nasect) {
		    int i;
		    ret.asect = resize(ret.asect, n+1);
		    for (i = ret.nasect; i <= n; i++)
			ret.asect[i] = ret.asect[ret.nasect-1];
		    ret.nasect = n+1;
		}
		ret.asect[n].align = utoalign(p);
	    } else if (!ustricmp(source->keyword, L"text-section-underline")) {
		wchar_t *p = uadv(source->keyword);
		int n = 0;
		if (uisdigit(*p)) {
		    n = utoi(p);
		    p = uadv(p);
		}
		if (n >= ret.nasect) {
		    int i;
		    ret.asect = resize(ret.asect, n+1);
		    for (i = ret.nasect; i <= n; i++)
			ret.asect[i] = ret.asect[ret.nasect-1];
		    ret.nasect = n+1;
		}
		ret.asect[n].underline = *p;
	    } else if (!ustricmp(source->keyword, L"text-section-numeric")) {
		wchar_t *p = uadv(source->keyword);
		int n = 0;
		if (uisdigit(*p)) {
		    n = utoi(p);
		    p = uadv(p);
		}
		if (n >= ret.nasect) {
		    int i;
		    ret.asect = resize(ret.asect, n+1);
		    for (i = ret.nasect; i <= n; i++)
			ret.asect[i] = ret.asect[ret.nasect-1];
		    ret.nasect = n+1;
		}
		ret.asect[n].just_numbers = utob(p);
	    } else if (!ustricmp(source->keyword, L"text-section-suffix")) {
		wchar_t *p = uadv(source->keyword);
		int n = 0;
		if (uisdigit(*p)) {
		    n = utoi(p);
		    p = uadv(p);
		}
		if (n >= ret.nasect) {
		    int i;
		    ret.asect = resize(ret.asect, n+1);
		    for (i = ret.nasect; i <= n; i++)
			ret.asect[i] = ret.asect[ret.nasect-1];
		    ret.nasect = n+1;
		}
		ret.asect[n].number_suffix = ustrdup(p);
	    } else if (!ustricmp(source->keyword, L"text-title-align")) {
		ret.atitle.align = utoalign(uadv(source->keyword));
	    } else if (!ustricmp(source->keyword, L"text-title-underline")) {
		ret.atitle.underline = *uadv(source->keyword);
	    } else if (!ustricmp(source->keyword, L"text-versionid")) {
		ret.include_version_id = utob(uadv(source->keyword));
	    } else if (!ustricmp(source->keyword, L"text-indent-preamble")) {
		ret.indent_preambles = utob(uadv(source->keyword));
	    } else if (!ustricmp(source->keyword, L"text-bullet")) {
		ret.bullet.text = uadv(source->keyword);
	    }
	}
    }

    return ret;
}

void text_backend(paragraph *sourceform, keywordlist *keywords,
		  indexdata *idx) {
    paragraph *p;
    textconfig conf;
    word *prefix, *body, *wp;
    word spaceword;
    FILE *fp;
    char *prefixextra;
    int nesting, nestindent;
    int indentb, indenta;

    IGNORE(keywords);		       /* we don't happen to need this */
    IGNORE(idx);		       /* or this */

    conf = text_configure(sourceform);

    /*
     * Determine the output file name, and open the output file
     *
     * FIXME: want configurable output file names here. For the
     * moment, we'll just call it `output.txt'.
     */
    fp = fopen("output.txt", "w");
    if (!fp) {
	error(err_cantopenw, "output.txt");
	return;
    }

    /* Do the title */
    for (p = sourceform; p; p = p->next)
	if (p->type == para_Title)
	    text_heading(fp, NULL, NULL, p->words,
			 conf.atitle, conf.indent, conf.width);

    /* Do the preamble and copyright */
    for (p = sourceform; p; p = p->next)
	if (p->type == para_Preamble)
	    text_para(fp, NULL, NULL, p->words,
		      conf.indent_preambles ? conf.indent : 0, 0,
		      conf.width + (conf.indent_preambles ? 0 : conf.indent));
    for (p = sourceform; p; p = p->next)
	if (p->type == para_Copyright)
	    text_para(fp, NULL, NULL, p->words,
		      conf.indent_preambles ? conf.indent : 0, 0,
		      conf.width + (conf.indent_preambles ? 0 : conf.indent));

    nestindent = conf.listindentbefore + conf.listindentafter;
    nesting = 0;

    /* Do the main document */
    for (p = sourceform; p; p = p->next) switch (p->type) {

      case para_QuotePush:
	nesting += 2;
	break;
      case para_QuotePop:
	nesting -= 2;
	assert(nesting >= 0);
	break;

      case para_LcontPush:
	nesting += nestindent;
	break;
      case para_LcontPop:
	nesting -= nestindent;
	assert(nesting >= 0);
	break;

	/*
	 * Things we ignore because we've already processed them or
	 * aren't going to touch them in this pass.
	 */
      case para_IM:
      case para_BR:
      case para_Biblio:		       /* only touch BiblioCited */
      case para_VersionID:
      case para_Copyright:
      case para_Preamble:
      case para_NoCite:
      case para_Title:
	break;

	/*
	 * Chapter titles.
	 */
      case para_Chapter:
      case para_Appendix:
      case para_UnnumberedChapter:
	text_heading(fp, p->kwtext, p->kwtext2, p->words,
		     conf.achapter, conf.indent, conf.width);
	break;

      case para_Heading:
      case para_Subsect:
	text_heading(fp, p->kwtext, p->kwtext2, p->words,
		     conf.asect[p->aux>=conf.nasect ? conf.nasect-1 : p->aux],
		     conf.indent, conf.width);
	break;

      case para_Rule:
	text_rule(fp, conf.indent + nesting, conf.width - nesting);
	break;

      case para_Normal:
      case para_DescribedThing:
      case para_Description:
      case para_BiblioCited:
      case para_Bullet:
      case para_NumberedList:
	if (p->type == para_Bullet) {
	    prefix = &conf.bullet;
	    prefixextra = NULL;
	    indentb = conf.listindentbefore;
	    indenta = conf.listindentafter;
	} else if (p->type == para_NumberedList) {
	    prefix = p->kwtext;
	    prefixextra = ".";	       /* FIXME: configurability */
	    indentb = conf.listindentbefore;
	    indenta = conf.listindentafter;
	} else if (p->type == para_Description) {
	    prefix = NULL;
	    prefixextra = NULL;
	    indentb = conf.listindentbefore;
	    indenta = conf.listindentafter;
	} else {
	    prefix = NULL;
	    prefixextra = NULL;
	    indentb = indenta = 0;
	}
	if (p->type == para_BiblioCited) {
	    body = dup_word_list(p->kwtext);
	    for (wp = body; wp->next; wp = wp->next);
	    wp->next = &spaceword;
	    spaceword.next = p->words;
	    spaceword.alt = NULL;
	    spaceword.type = word_WhiteSpace;
	    spaceword.text = NULL;
	} else {
	    wp = NULL;
	    body = p->words;
	}
	text_para(fp, prefix, prefixextra, body,
		  conf.indent + nesting + indentb, indenta,
		  conf.width - nesting - indentb - indenta);
	if (wp) {
	    wp->next = NULL;
	    free_word_list(body);
	}
	break;

      case para_Code:
	text_codepara(fp, p->words,
		      conf.indent + nesting + conf.indent_code,
		      conf.width - nesting - 2 * conf.indent_code);
	break;
    }

    /* Do the version ID */
    if (conf.include_version_id) {
	for (p = sourceform; p; p = p->next)
	    if (p->type == para_VersionID)
 		text_versionid(fp, p->words);
    }

    /*
     * Tidy up
     */
    fclose(fp);
    {
	int i;
	sfree(conf.achapter.number_suffix);
	for (i = 0; i < conf.nasect; i++)
	    sfree(conf.asect[i].number_suffix);
	sfree(conf.asect);
	sfree(conf.bullet.text);
    }
}

/*
 * Convert a wide string into a string of chars. If `result' is
 * non-NULL, mallocs the resulting string and stores a pointer to
 * it in `*result'. If `result' is NULL, merely checks whether all
 * characters in the string are feasible for the output character
 * set.
 *
 * Return is nonzero if all characters are OK. If not all
 * characters are OK but `result' is non-NULL, a result _will_
 * still be generated!
 */
static int text_convert(wchar_t *s, char **result) {
    /*
     * FIXME. Currently this is ISO8859-1 only.
     */
    int doing = (result != 0);
    int ok = TRUE;
    char *p = NULL;
    int plen = 0, psize = 0;

    for (; *s; s++) {
	wchar_t c = *s;
	char outc;

	if ((c >= 32 && c <= 126) ||
	    (c >= 160 && c <= 255)) {
	    /* Char is OK. */
	    outc = (char)c;
	} else {
	    /* Char is not OK. */
	    ok = FALSE;
	    outc = 0xBF;	       /* approximate the good old DEC `uh?' */
	}
	if (doing) {
	    if (plen >= psize) {
		psize = plen + 256;
		p = resize(p, psize);
	    }
	    p[plen++] = outc;
	}
    }
    if (doing) {
	p = resize(p, plen+1);
	p[plen] = '\0';
	*result = p;
    }
    return ok;
}

static void text_rdaddwc(rdstringc *rs, word *text, word *end) {
    char *c;

    for (; text && text != end; text = text->next) switch (text->type) {
      case word_HyperLink:
      case word_HyperEnd:
      case word_UpperXref:
      case word_LowerXref:
      case word_XrefEnd:
      case word_IndexRef:
	break;

      case word_Normal:
      case word_Emph:
      case word_Code:
      case word_WeakCode:
      case word_WhiteSpace:
      case word_EmphSpace:
      case word_CodeSpace:
      case word_WkCodeSpace:
      case word_Quote:
      case word_EmphQuote:
      case word_CodeQuote:
      case word_WkCodeQuote:
	assert(text->type != word_CodeQuote &&
	       text->type != word_WkCodeQuote);
	if (towordstyle(text->type) == word_Emph &&
	    (attraux(text->aux) == attr_First ||
	     attraux(text->aux) == attr_Only))
	    rdaddc(rs, '_');	       /* FIXME: configurability */
	else if (towordstyle(text->type) == word_Code &&
		 (attraux(text->aux) == attr_First ||
		  attraux(text->aux) == attr_Only))
	    rdaddc(rs, '`');	       /* FIXME: configurability */
	if (removeattr(text->type) == word_Normal) {
	    if (text_convert(text->text, &c))
		rdaddsc(rs, c);
	    else
		text_rdaddwc(rs, text->alt, NULL);
	    sfree(c);
	} else if (removeattr(text->type) == word_WhiteSpace) {
	    rdaddc(rs, ' ');
	} else if (removeattr(text->type) == word_Quote) {
	    rdaddc(rs, quoteaux(text->aux) == quote_Open ? '`' : '\'');
				       /* FIXME: configurability */
	}
	if (towordstyle(text->type) == word_Emph &&
	    (attraux(text->aux) == attr_Last ||
	     attraux(text->aux) == attr_Only))
	    rdaddc(rs, '_');	       /* FIXME: configurability */
	else if (towordstyle(text->type) == word_Code &&
		 (attraux(text->aux) == attr_Last ||
		  attraux(text->aux) == attr_Only))
	    rdaddc(rs, '\'');	       /* FIXME: configurability */
	break;
    }
}

static int text_width(word *);

static int text_width_list(word *text) {
    int w = 0;
    while (text) {
	w += text_width(text);
	text = text->next;
    }
    return w;
}

static int text_width(word *text) {
    switch (text->type) {
      case word_HyperLink:
      case word_HyperEnd:
      case word_UpperXref:
      case word_LowerXref:
      case word_XrefEnd:
      case word_IndexRef:
	return 0;

      case word_Normal:
      case word_Emph:
      case word_Code:
      case word_WeakCode:
	return (((text->type == word_Emph ||
		  text->type == word_Code)
		 ? (attraux(text->aux) == attr_Only ? 2 :
		    attraux(text->aux) == attr_Always ? 0 : 1)
		 : 0) +
		(text_convert(text->text, NULL) ?
		 ustrlen(text->text) :
		 text_width_list(text->alt)));

      case word_WhiteSpace:
      case word_EmphSpace:
      case word_CodeSpace:
      case word_WkCodeSpace:
      case word_Quote:
      case word_EmphQuote:
      case word_CodeQuote:
      case word_WkCodeQuote:
	assert(text->type != word_CodeQuote &&
	       text->type != word_WkCodeQuote);
	return (((towordstyle(text->type) == word_Emph ||
		  towordstyle(text->type) == word_Code)
		 ? (attraux(text->aux) == attr_Only ? 2 :
		    attraux(text->aux) == attr_Always ? 0 : 1)
		 : 0) + 1);
    }
    return 0;			       /* should never happen */
}

static void text_heading(FILE *fp, word *tprefix, word *nprefix, word *text,
			 alignstruct align, int indent, int width) {
    rdstringc t = { 0, 0, NULL };
    int margin, length;
    int firstlinewidth, wrapwidth;
    wrappedline *wrapping, *p;

    if (align.just_numbers && nprefix) {
	char *c;
	text_rdaddwc(&t, nprefix, NULL);
	if (text_convert(align.number_suffix, &c)) {
	    rdaddsc(&t, c);
	    sfree(c);
	}
    } else if (!align.just_numbers && tprefix) {
	char *c;
	text_rdaddwc(&t, tprefix, NULL);
	if (text_convert(align.number_suffix, &c)) {
	    rdaddsc(&t, c);
	    sfree(c);
	}
    }
    margin = length = (t.text ? strlen(t.text) : 0);

    if (align.align == LEFTPLUS) {
	margin = indent - margin;
	if (margin < 0) margin = 0;
	firstlinewidth = indent + width - margin - length;
	wrapwidth = width;
    } else if (align.align == LEFT || align.align == CENTRE) {
	margin = 0;
	firstlinewidth = indent + width - length;
	wrapwidth = indent + width;
    }

    wrapping = wrap_para(text, firstlinewidth, wrapwidth, text_width);
    for (p = wrapping; p; p = p->next) {
	text_rdaddwc(&t, p->begin, p->end);
	length = (t.text ? strlen(t.text) : 0);
	if (align.align == CENTRE) {
	    margin = (indent + width - length)/2;
	    if (margin < 0) margin = 0;
	}
	fprintf(fp, "%*s%s\n", margin, "", t.text);
	if (align.underline != L'\0') {
	    char *u, uc;
	    wchar_t uw[2];
	    uw[0] = align.underline; uw[1] = L'\0';
	    text_convert(uw, &u);
	    uc = u[0];
	    sfree(u);
	    fprintf(fp, "%*s", margin, "");
	    while (length--)
		putc(uc, fp);
	    putc('\n', fp);
	}
	if (align.align == LEFTPLUS)
	    margin = indent;
	else
	    margin = 0;
	sfree(t.text);
	t = empty_rdstringc;
    }
    wrap_free(wrapping);
    putc('\n', fp);

    sfree(t.text);
}

static void text_rule(FILE *fp, int indent, int width) {
    while (indent--) putc(' ', fp);
    while (width--) putc('-', fp);     /* FIXME: configurability! */
    putc('\n', fp);
    putc('\n', fp);
}

static void text_para(FILE *fp, word *prefix, char *prefixextra, word *text,
		      int indent, int extraindent, int width) {
    wrappedline *wrapping, *p;
    rdstringc pfx = { 0, 0, NULL };
    int e;
    int firstlinewidth = width;

    if (prefix) {
	text_rdaddwc(&pfx, prefix, NULL);
	if (prefixextra)
	    rdaddsc(&pfx, prefixextra);
	fprintf(fp, "%*s%s", indent, "", pfx.text);
	/* If the prefix is too long, shorten the first line to fit. */
	e = extraindent - strlen(pfx.text);
	if (e < 0) {
	    firstlinewidth += e;       /* this decreases it, since e < 0 */
	    if (firstlinewidth < 0) {
		e = indent + extraindent;
		firstlinewidth = width;
		fprintf(fp, "\n");
	    } else
		e = 0;
	}
	sfree(pfx.text);
    } else
	e = indent + extraindent;

    wrapping = wrap_para(text, firstlinewidth, width, text_width);
    for (p = wrapping; p; p = p->next) {
	rdstringc t = { 0, 0, NULL };
	text_rdaddwc(&t, p->begin, p->end);
	fprintf(fp, "%*s%s\n", e, "", t.text);
	e = indent + extraindent;
	sfree(t.text);
    }
    wrap_free(wrapping);
    putc('\n', fp);
}

static void text_codepara(FILE *fp, word *text, int indent, int width) {
    for (; text; text = text->next) if (text->type == word_WeakCode) {
	char *c;
	text_convert(text->text, &c);
	if (strlen(c) > (size_t)width) {
	    /* FIXME: warn */
	}
	fprintf(fp, "%*s%s\n", indent, "", c);
	sfree(c);
    }

    putc('\n', fp);
}

static void text_versionid(FILE *fp, word *text) {
    rdstringc t = { 0, 0, NULL };

    rdaddc(&t, '[');		       /* FIXME: configurability */
    text_rdaddwc(&t, text, NULL);
    rdaddc(&t, ']');		       /* FIXME: configurability */

    fprintf(fp, "%s\n", t.text);
    sfree(t.text);
}
Commit	Line	Data
d7482997	1	/*
	2	* text backend for Halibut
	3	*/
	4
	5	#include <stdio.h>
	6	#include <stdlib.h>
	7	#include <assert.h>
	8	#include "halibut.h"
	9
	10	typedef enum { LEFT, LEFTPLUS, CENTRE } alignment;
	11	typedef struct {
	12	alignment align;
	13	int just_numbers;
	14	wchar_t underline;
63223c78	15	wchar_t *number_suffix;
d7482997	16	} alignstruct;
	17
	18	typedef struct {
	19	int indent, indent_code;
	20	int listindentbefore, listindentafter;
	21	int width;
	22	alignstruct atitle, achapter, *asect;
	23	int nasect;
	24	int include_version_id;
	25	int indent_preambles;
	26	word bullet;
	27	} textconfig;
	28
	29	static int text_convert(wchar_t , char *);
	30
	31	static void text_heading(FILE , word , word , word , alignstruct, int,int);
	32	static void text_rule(FILE *, int, int);
	33	static void text_para(FILE , word , char , word , int, int, int);
	34	static void text_codepara(FILE , word , int, int);
	35	static void text_versionid(FILE , word );
	36
	37	static alignment utoalign(wchar_t *p) {
	38	if (!ustricmp(p, L"centre") \|\| !ustricmp(p, L"center"))
	39	return CENTRE;
	40	if (!ustricmp(p, L"leftplus"))
	41	return LEFTPLUS;
	42	return LEFT;
	43	}
	44
	45	static textconfig text_configure(paragraph *source) {
	46	textconfig ret;
	47
	48	/*
	49	* Non-negotiables.
	50	*/
	51	ret.bullet.next = NULL;
	52	ret.bullet.alt = NULL;
	53	ret.bullet.type = word_Normal;
	54	ret.atitle.just_numbers = FALSE; /* ignored */
	55
	56	/*
	57	* Defaults.
	58	*/
	59	ret.indent = 7;
	60	ret.indent_code = 2;
	61	ret.listindentbefore = 1;
	62	ret.listindentafter = 3;
	63	ret.width = 68;
	64	ret.atitle.align = CENTRE;
	65	ret.atitle.underline = L'=';
	66	ret.achapter.align = LEFT;
	67	ret.achapter.just_numbers = FALSE;
63223c78	68	ret.achapter.number_suffix = ustrdup(L": ");
d7482997	69	ret.achapter.underline = L'-';
	70	ret.nasect = 1;
	71	ret.asect = mknewa(alignstruct, ret.nasect);
	72	ret.asect[0].align = LEFTPLUS;
	73	ret.asect[0].just_numbers = TRUE;
63223c78	74	ret.asect[0].number_suffix = ustrdup(L" ");
d7482997	75	ret.asect[0].underline = L'\0';
	76	ret.include_version_id = TRUE;
	77	ret.indent_preambles = FALSE;
	78	ret.bullet.text = ustrdup(L"-");
	79
	80	for (; source; source = source->next) {
	81	if (source->type == para_Config) {
	82	if (!ustricmp(source->keyword, L"text-indent")) {
	83	ret.indent = utoi(uadv(source->keyword));
	84	} else if (!ustricmp(source->keyword, L"text-indent-code")) {
	85	ret.indent_code = utoi(uadv(source->keyword));
	86	} else if (!ustricmp(source->keyword, L"text-width")) {
	87	ret.width = utoi(uadv(source->keyword));
	88	} else if (!ustricmp(source->keyword, L"text-list-indent")) {
	89	ret.listindentbefore = utoi(uadv(source->keyword));
	90	} else if (!ustricmp(source->keyword, L"text-listitem-indent")) {
	91	ret.listindentafter = utoi(uadv(source->keyword));
	92	} else if (!ustricmp(source->keyword, L"text-chapter-align")) {
	93	ret.achapter.align = utoalign(uadv(source->keyword));
	94	} else if (!ustricmp(source->keyword, L"text-chapter-underline")) {
	95	ret.achapter.underline = *uadv(source->keyword);
	96	} else if (!ustricmp(source->keyword, L"text-chapter-numeric")) {
c83c6495	97	ret.achapter.just_numbers = utob(uadv(source->keyword));
63223c78	98	} else if (!ustricmp(source->keyword, L"text-chapter-suffix")) {
a0f2c111	99	ret.achapter.number_suffix = ustrdup(uadv(source->keyword));
d7482997	100	} else if (!ustricmp(source->keyword, L"text-section-align")) {
	101	wchar_t *p = uadv(source->keyword);
	102	int n = 0;
	103	if (uisdigit(*p)) {
	104	n = utoi(p);
	105	p = uadv(p);
	106	}
	107	if (n >= ret.nasect) {
	108	int i;
	109	ret.asect = resize(ret.asect, n+1);
	110	for (i = ret.nasect; i <= n; i++)
	111	ret.asect[i] = ret.asect[ret.nasect-1];
	112	ret.nasect = n+1;
	113	}
	114	ret.asect[n].align = utoalign(p);
	115	} else if (!ustricmp(source->keyword, L"text-section-underline")) {
	116	wchar_t *p = uadv(source->keyword);
	117	int n = 0;
	118	if (uisdigit(*p)) {
	119	n = utoi(p);
	120	p = uadv(p);
	121	}
	122	if (n >= ret.nasect) {
	123	int i;
	124	ret.asect = resize(ret.asect, n+1);
	125	for (i = ret.nasect; i <= n; i++)
	126	ret.asect[i] = ret.asect[ret.nasect-1];
	127	ret.nasect = n+1;
	128	}
	129	ret.asect[n].underline = *p;
	130	} else if (!ustricmp(source->keyword, L"text-section-numeric")) {
	131	wchar_t *p = uadv(source->keyword);
	132	int n = 0;
	133	if (uisdigit(*p)) {
	134	n = utoi(p);
	135	p = uadv(p);
	136	}
	137	if (n >= ret.nasect) {
	138	int i;
	139	ret.asect = resize(ret.asect, n+1);
	140	for (i = ret.nasect; i <= n; i++)
	141	ret.asect[i] = ret.asect[ret.nasect-1];
	142	ret.nasect = n+1;
	143	}
	144	ret.asect[n].just_numbers = utob(p);
63223c78	145	} else if (!ustricmp(source->keyword, L"text-section-suffix")) {
	146	wchar_t *p = uadv(source->keyword);
	147	int n = 0;
	148	if (uisdigit(*p)) {
	149	n = utoi(p);
	150	p = uadv(p);
	151	}
	152	if (n >= ret.nasect) {
	153	int i;
	154	ret.asect = resize(ret.asect, n+1);
	155	for (i = ret.nasect; i <= n; i++)
	156	ret.asect[i] = ret.asect[ret.nasect-1];
	157	ret.nasect = n+1;
	158	}
a0f2c111	159	ret.asect[n].number_suffix = ustrdup(p);
d7482997	160	} else if (!ustricmp(source->keyword, L"text-title-align")) {
	161	ret.atitle.align = utoalign(uadv(source->keyword));
	162	} else if (!ustricmp(source->keyword, L"text-title-underline")) {
	163	ret.atitle.underline = *uadv(source->keyword);
	164	} else if (!ustricmp(source->keyword, L"text-versionid")) {
	165	ret.include_version_id = utob(uadv(source->keyword));
	166	} else if (!ustricmp(source->keyword, L"text-indent-preamble")) {
	167	ret.indent_preambles = utob(uadv(source->keyword));
	168	} else if (!ustricmp(source->keyword, L"text-bullet")) {
	169	ret.bullet.text = uadv(source->keyword);
	170	}
	171	}
	172	}
	173
	174	return ret;
	175	}
	176
	177	void text_backend(paragraph sourceform, keywordlist keywords,
	178	indexdata *idx) {
	179	paragraph *p;
	180	textconfig conf;
	181	word prefix, body, *wp;
	182	word spaceword;
	183	FILE *fp;
	184	char *prefixextra;
7136a6c7	185	int nesting, nestindent;
d7482997	186	int indentb, indenta;
	187
	188	IGNORE(keywords); /* we don't happen to need this */
	189	IGNORE(idx); /* or this */
	190
	191	conf = text_configure(sourceform);
	192
	193	/*
	194	* Determine the output file name, and open the output file
	195	*
	196	* FIXME: want configurable output file names here. For the
	197	* moment, we'll just call it `output.txt'.
	198	*/
	199	fp = fopen("output.txt", "w");
	200	if (!fp) {
	201	error(err_cantopenw, "output.txt");
	202	return;
	203	}
	204
	205	/* Do the title */
	206	for (p = sourceform; p; p = p->next)
	207	if (p->type == para_Title)
	208	text_heading(fp, NULL, NULL, p->words,
	209	conf.atitle, conf.indent, conf.width);
	210
	211	/* Do the preamble and copyright */
	212	for (p = sourceform; p; p = p->next)
	213	if (p->type == para_Preamble)
	214	text_para(fp, NULL, NULL, p->words,
	215	conf.indent_preambles ? conf.indent : 0, 0,
	216	conf.width + (conf.indent_preambles ? 0 : conf.indent));
	217	for (p = sourceform; p; p = p->next)
	218	if (p->type == para_Copyright)
	219	text_para(fp, NULL, NULL, p->words,
	220	conf.indent_preambles ? conf.indent : 0, 0,
	221	conf.width + (conf.indent_preambles ? 0 : conf.indent));
	222
7136a6c7	223	nestindent = conf.listindentbefore + conf.listindentafter;
	224	nesting = 0;
	225
d7482997	226	/* Do the main document */
	227	for (p = sourceform; p; p = p->next) switch (p->type) {
	228
2614b01d	229	case para_QuotePush:
	230	nesting += 2;
	231	break;
	232	case para_QuotePop:
	233	nesting -= 2;
	234	assert(nesting >= 0);
	235	break;
	236
7136a6c7	237	case para_LcontPush:
2614b01d	238	nesting += nestindent;
7136a6c7	239	break;
7136a6c7	240	case para_LcontPop:
2614b01d	241	nesting -= nestindent;
2614b01d	242	assert(nesting >= 0);
7136a6c7	243	break;
7136a6c7	244
d7482997	245	/*
	246	* Things we ignore because we've already processed them or
	247	* aren't going to touch them in this pass.
	248	*/
	249	case para_IM:
	250	case para_BR:
	251	case para_Biblio: /* only touch BiblioCited */
	252	case para_VersionID:
	253	case para_Copyright:
	254	case para_Preamble:
	255	case para_NoCite:
	256	case para_Title:
	257	break;
	258
	259	/*
	260	* Chapter titles.
	261	*/
	262	case para_Chapter:
	263	case para_Appendix:
	264	case para_UnnumberedChapter:
	265	text_heading(fp, p->kwtext, p->kwtext2, p->words,
	266	conf.achapter, conf.indent, conf.width);
	267	break;
	268
	269	case para_Heading:
	270	case para_Subsect:
	271	text_heading(fp, p->kwtext, p->kwtext2, p->words,
	272	conf.asect[p->aux>=conf.nasect ? conf.nasect-1 : p->aux],
	273	conf.indent, conf.width);
	274	break;
	275
	276	case para_Rule:
2614b01d	277	text_rule(fp, conf.indent + nesting, conf.width - nesting);
d7482997	278	break;
	279
	280	case para_Normal:
7136a6c7	281	case para_DescribedThing:
7136a6c7	282	case para_Description:
d7482997	283	case para_BiblioCited:
	284	case para_Bullet:
	285	case para_NumberedList:
	286	if (p->type == para_Bullet) {
	287	prefix = &conf.bullet;
	288	prefixextra = NULL;
	289	indentb = conf.listindentbefore;
	290	indenta = conf.listindentafter;
	291	} else if (p->type == para_NumberedList) {
	292	prefix = p->kwtext;
	293	prefixextra = "."; /* FIXME: configurability */
	294	indentb = conf.listindentbefore;
	295	indenta = conf.listindentafter;
7136a6c7	296	} else if (p->type == para_Description) {
	297	prefix = NULL;
	298	prefixextra = NULL;
	299	indentb = conf.listindentbefore;
	300	indenta = conf.listindentafter;
d7482997	301	} else {
	302	prefix = NULL;
	303	prefixextra = NULL;
	304	indentb = indenta = 0;
	305	}
	306	if (p->type == para_BiblioCited) {
	307	body = dup_word_list(p->kwtext);
	308	for (wp = body; wp->next; wp = wp->next);
	309	wp->next = &spaceword;
	310	spaceword.next = p->words;
	311	spaceword.alt = NULL;
	312	spaceword.type = word_WhiteSpace;
	313	spaceword.text = NULL;
	314	} else {
	315	wp = NULL;
	316	body = p->words;
	317	}
	318	text_para(fp, prefix, prefixextra, body,
2614b01d	319	conf.indent + nesting + indentb, indenta,
2614b01d	320	conf.width - nesting - indentb - indenta);
d7482997	321	if (wp) {
	322	wp->next = NULL;
	323	free_word_list(body);
	324	}
	325	break;
	326
	327	case para_Code:
7136a6c7	328	text_codepara(fp, p->words,
2614b01d	329	conf.indent + nesting + conf.indent_code,
2614b01d	330	conf.width - nesting - 2 * conf.indent_code);
d7482997	331	break;
	332	}
	333
	334	/* Do the version ID */
	335	if (conf.include_version_id) {
	336	for (p = sourceform; p; p = p->next)
	337	if (p->type == para_VersionID)
	338	text_versionid(fp, p->words);
	339	}
	340
	341	/*
	342	* Tidy up
	343	*/
	344	fclose(fp);
677e18a2	345	{
	346	int i;
	347	sfree(conf.achapter.number_suffix);
	348	for (i = 0; i < conf.nasect; i++)
	349	sfree(conf.asect[i].number_suffix);
	350	sfree(conf.asect);
	351	sfree(conf.bullet.text);
	352	}
d7482997	353	}
	354
	355	/*
	356	* Convert a wide string into a string of chars. If `result' is
	357	* non-NULL, mallocs the resulting string and stores a pointer to
	358	* it in `*result'. If `result' is NULL, merely checks whether all
	359	* characters in the string are feasible for the output character
	360	* set.
	361	*
	362	* Return is nonzero if all characters are OK. If not all
	363	* characters are OK but `result' is non-NULL, a result _will_
	364	* still be generated!
	365	*/
	366	static int text_convert(wchar_t s, char *result) {
	367	/*
	368	* FIXME. Currently this is ISO8859-1 only.
	369	*/
	370	int doing = (result != 0);
	371	int ok = TRUE;
	372	char *p = NULL;
	373	int plen = 0, psize = 0;
	374
	375	for (; *s; s++) {
	376	wchar_t c = *s;
	377	char outc;
	378
	379	if ((c >= 32 && c <= 126) \|\|
	380	(c >= 160 && c <= 255)) {
	381	/* Char is OK. */
	382	outc = (char)c;
	383	} else {
	384	/* Char is not OK. */
	385	ok = FALSE;
	386	outc = 0xBF; /* approximate the good old DEC `uh?' */
	387	}
	388	if (doing) {
	389	if (plen >= psize) {
	390	psize = plen + 256;
	391	p = resize(p, psize);
	392	}
	393	p[plen++] = outc;
	394	}
	395	}
	396	if (doing) {
	397	p = resize(p, plen+1);
	398	p[plen] = '\0';
	399	*result = p;
	400	}
	401	return ok;
	402	}
	403
	404	static void text_rdaddwc(rdstringc rs, word text, word *end) {
	405	char *c;
	406
	407	for (; text && text != end; text = text->next) switch (text->type) {
	408	case word_HyperLink:
	409	case word_HyperEnd:
	410	case word_UpperXref:
	411	case word_LowerXref:
	412	case word_XrefEnd:
	413	case word_IndexRef:
	414	break;
	415
	416	case word_Normal:
417	case word_Emph:
418	case word_Code:
419	case word_WeakCode:
420	case word_WhiteSpace:
421	case word_EmphSpace:
422	case word_CodeSpace:
423	case word_WkCodeSpace:
424	case word_Quote:
425	case word_EmphQuote:
426	case word_CodeQuote:
427	case word_WkCodeQuote:
428	assert(text->type != word_CodeQuote &&
429	text->type != word_WkCodeQuote);
430	if (towordstyle(text->type) == word_Emph &&
431	(attraux(text->aux) == attr_First \|\|
432	attraux(text->aux) == attr_Only))
433	rdaddc(rs, '_'); /* FIXME: configurability */
434	else if (towordstyle(text->type) == word_Code &&
435	(attraux(text->aux) == attr_First \|\|
436	attraux(text->aux) == attr_Only))
437	rdaddc(rs, '`'); /* FIXME: configurability */
438	if (removeattr(text->type) == word_Normal) {
439	if (text_convert(text->text, &c))
440	rdaddsc(rs, c);
441	else
442	text_rdaddwc(rs, text->alt, NULL);
443	sfree(c);
444	} else if (removeattr(text->type) == word_WhiteSpace) {
445	rdaddc(rs, ' ');
446	} else if (removeattr(text->type) == word_Quote) {
447	rdaddc(rs, quoteaux(text->aux) == quote_Open ? '`' : '\'');
448	/* FIXME: configurability */
449	}
450	if (towordstyle(text->type) == word_Emph &&
451	(attraux(text->aux) == attr_Last \|\|
452	attraux(text->aux) == attr_Only))
453	rdaddc(rs, '_'); /* FIXME: configurability */
454	else if (towordstyle(text->type) == word_Code &&
455	(attraux(text->aux) == attr_Last \|\|
456	attraux(text->aux) == attr_Only))
457	rdaddc(rs, '\''); /* FIXME: configurability */
458	break;
459	}
460	}
461
462	static int text_width(word *);
463
464	static int text_width_list(word *text) {
465	int w = 0;
466	while (text) {
467	w += text_width(text);
468	text = text->next;
469	}
470	return w;
471	}
472
473	static int text_width(word *text) {
474	switch (text->type) {
475	case word_HyperLink:
476	case word_HyperEnd:
477	case word_UpperXref:
478	case word_LowerXref:
479	case word_XrefEnd:
480	case word_IndexRef:
481	return 0;
482
483	case word_Normal:
484	case word_Emph:
485	case word_Code:
486	case word_WeakCode:
487	return (((text->type == word_Emph \|\|
488	text->type == word_Code)
489	? (attraux(text->aux) == attr_Only ? 2 :
490	attraux(text->aux) == attr_Always ? 0 : 1)
491	: 0) +
492	(text_convert(text->text, NULL) ?
493	ustrlen(text->text) :
494	text_width_list(text->alt)));
495
496	case word_WhiteSpace:
497	case word_EmphSpace:
498	case word_CodeSpace:
499	case word_WkCodeSpace:
500	case word_Quote:
501	case word_EmphQuote:
502	case word_CodeQuote:
503	case word_WkCodeQuote:
504	assert(text->type != word_CodeQuote &&
505	text->type != word_WkCodeQuote);
506	return (((towordstyle(text->type) == word_Emph \|\|
507	towordstyle(text->type) == word_Code)
508	? (attraux(text->aux) == attr_Only ? 2 :
509	attraux(text->aux) == attr_Always ? 0 : 1)
510	: 0) + 1);
511	}
512	return 0; /* should never happen */
513	}
514
515	static void text_heading(FILE fp, word tprefix, word nprefix, word text,
516	alignstruct align, int indent, int width) {
517	rdstringc t = { 0, 0, NULL };
518	int margin, length;
519	int firstlinewidth, wrapwidth;
520	wrappedline wrapping, p;
521
522	if (align.just_numbers && nprefix) {
63223c78	523	char *c;
d7482997	524	text_rdaddwc(&t, nprefix, NULL);
63223c78	525	if (text_convert(align.number_suffix, &c)) {
	526	rdaddsc(&t, c);
	527	sfree(c);
	528	}
d7482997	529	} else if (!align.just_numbers && tprefix) {
63223c78	530	char *c;
d7482997	531	text_rdaddwc(&t, tprefix, NULL);
63223c78	532	if (text_convert(align.number_suffix, &c)) {
	533	rdaddsc(&t, c);
	534	sfree(c);
	535	}
d7482997	536	}
	537	margin = length = (t.text ? strlen(t.text) : 0);
	538
	539	if (align.align == LEFTPLUS) {
	540	margin = indent - margin;
	541	if (margin < 0) margin = 0;
	542	firstlinewidth = indent + width - margin - length;
	543	wrapwidth = width;
	544	} else if (align.align == LEFT \|\| align.align == CENTRE) {
	545	margin = 0;
	546	firstlinewidth = indent + width - length;
	547	wrapwidth = indent + width;
	548	}
	549
	550	wrapping = wrap_para(text, firstlinewidth, wrapwidth, text_width);
	551	for (p = wrapping; p; p = p->next) {
	552	text_rdaddwc(&t, p->begin, p->end);
	553	length = (t.text ? strlen(t.text) : 0);
	554	if (align.align == CENTRE) {
	555	margin = (indent + width - length)/2;
	556	if (margin < 0) margin = 0;
	557	}
	558	fprintf(fp, "%*s%s\n", margin, "", t.text);
	559	if (align.underline != L'\0') {
	560	char *u, uc;
	561	wchar_t uw[2];
	562	uw[0] = align.underline; uw[1] = L'\0';
	563	text_convert(uw, &u);
	564	uc = u[0];
	565	sfree(u);
	566	fprintf(fp, "%*s", margin, "");
	567	while (length--)
	568	putc(uc, fp);
	569	putc('\n', fp);
	570	}
	571	if (align.align == LEFTPLUS)
	572	margin = indent;
	573	else
	574	margin = 0;
	575	sfree(t.text);
	576	t = empty_rdstringc;
	577	}
	578	wrap_free(wrapping);
	579	putc('\n', fp);
	580
	581	sfree(t.text);
	582	}
	583
	584	static void text_rule(FILE *fp, int indent, int width) {
	585	while (indent--) putc(' ', fp);
	586	while (width--) putc('-', fp); /* FIXME: configurability! */
	587	putc('\n', fp);
	588	putc('\n', fp);
	589	}
	590
	591	static void text_para(FILE fp, word prefix, char prefixextra, word text,
	592	int indent, int extraindent, int width) {
	593	wrappedline wrapping, p;
	594	rdstringc pfx = { 0, 0, NULL };
	595	int e;
	596	int firstlinewidth = width;
	597
	598	if (prefix) {
	599	text_rdaddwc(&pfx, prefix, NULL);
600	if (prefixextra)
601	rdaddsc(&pfx, prefixextra);
602	fprintf(fp, "%*s%s", indent, "", pfx.text);
c83c6495	603	/* If the prefix is too long, shorten the first line to fit. */
d7482997	604	e = extraindent - strlen(pfx.text);
d7482997	605	if (e < 0) {
c83c6495	606	firstlinewidth += e; /* this decreases it, since e < 0 */
d7482997	607	if (firstlinewidth < 0) {
	608	e = indent + extraindent;
	609	firstlinewidth = width;
	610	fprintf(fp, "\n");
c83c6495	611	} else
c83c6495	612	e = 0;
d7482997	613	}
	614	sfree(pfx.text);
	615	} else
	616	e = indent + extraindent;
	617
	618	wrapping = wrap_para(text, firstlinewidth, width, text_width);
	619	for (p = wrapping; p; p = p->next) {
	620	rdstringc t = { 0, 0, NULL };
	621	text_rdaddwc(&t, p->begin, p->end);
	622	fprintf(fp, "%*s%s\n", e, "", t.text);
	623	e = indent + extraindent;
	624	sfree(t.text);
	625	}
	626	wrap_free(wrapping);
	627	putc('\n', fp);
	628	}
	629
	630	static void text_codepara(FILE fp, word text, int indent, int width) {
	631	for (; text; text = text->next) if (text->type == word_WeakCode) {
	632	char *c;
	633	text_convert(text->text, &c);
	634	if (strlen(c) > (size_t)width) {
	635	/* FIXME: warn */
	636	}
	637	fprintf(fp, "%*s%s\n", indent, "", c);
	638	sfree(c);
	639	}
	640
	641	putc('\n', fp);
	642	}
	643
	644	static void text_versionid(FILE fp, word text) {
	645	rdstringc t = { 0, 0, NULL };
	646
	647	rdaddc(&t, '['); /* FIXME: configurability */
	648	text_rdaddwc(&t, text, NULL);
	649	rdaddc(&t, ']'); /* FIXME: configurability */
	650
	651	fprintf(fp, "%s\n", t.text);
	652	sfree(t.text);
	653	}