X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/halibut/blobdiff_plain/87735b83ba32f7163ffaacd6f0061011eca8269f..944c4044572da20ed1fb9bc01085221856b53c1f:/bk_paper.c

diff --git a/bk_paper.c b/bk_paper.c
index c3a1bc0..e70ecf7 100644
--- a/bk_paper.c
+++ b/bk_paper.c
@@ -49,13 +49,10 @@
  * 
  *  - title pages
  * 
- *  - ability to import other Type 1 fonts
+ *  - ability to use Type 1 fonts without AFM files
  *     * we need to parse the font to extract its metrics
- *     * then we pass the font bodily to both PS and PDF so it can
- * 	 be included in the output file
  * 
  *  - character substitution for better typography?
- *     * fi, fl, ffi, ffl ligatures
  *     * use real ellipsis rather than ...
  *     * a hyphen in a word by itself might prefer to be an en-dash
  *     * (Americans might even want a convenient way to use an
@@ -141,20 +138,24 @@ enum {
     word_PageXref = word_NotWordType + 1
 };
 
+/* Flags for render_string() */
+#define RS_NOLIG	1
+
 static font_data *make_std_font(font_list *fontlist, char const *name);
 static void wrap_paragraph(para_data *pdata, word *words,
 			   int w, int i1, int i2, paper_conf *conf);
 static page_data *page_breaks(line_data *first, line_data *last,
 			      int page_height, int ncols, int headspace);
 static int render_string(page_data *page, font_data *font, int fontsize,
-			 int x, int y, wchar_t *str);
+			 int x, int y, wchar_t *str, unsigned flags);
 static int render_line(line_data *ldata, int left_x, int top_y,
 		       xref_dest *dest, keywordlist *keywords, indexdata *idx,
 		       paper_conf *conf);
 static void render_para(para_data *pdata, paper_conf *conf,
 			keywordlist *keywords, indexdata *idx,
 			paragraph *index_placeholder, page_data *index_page);
-static int string_width(font_data *font, wchar_t const *string, int *errs);
+static int string_width(font_data *font, wchar_t const *string, int *errs,
+			unsigned flags);
 static int paper_width_simple(para_data *pdata, word *text, paper_conf *conf);
 static para_data *code_paragraph(int indent, word *words, paper_conf *conf);
 static para_data *rule_paragraph(int indent, paper_conf *conf);
@@ -182,7 +183,7 @@ static int fonts_ok(wchar_t *string, ...)
     va_start(ap, string);
     while ( (font = va_arg(ap, font_data *)) != NULL) {
 	int errs;
-	(void) string_width(font, string, &errs);
+	(void) string_width(font, string, &errs, 0);
 	if (errs) {
 	    ret = FALSE;
 	    break;
@@ -1016,13 +1017,13 @@ void *paper_pre_backend(paragraph *sourceform, keywordlist *keywords,
 
 	    width = conf->pagenum_fontsize *
 		string_width(conf->fbase.fonts[FONT_NORMAL], page->number,
-			     NULL);
+			     NULL, 0);
 
 	    render_string(page, conf->fbase.fonts[FONT_NORMAL],
 			  conf->pagenum_fontsize,
 			  conf->left_margin + (conf->base_width - width)/2,
 			  conf->bottom_margin - conf->footer_distance,
-			  page->number);
+			  page->number, 0);
 	}
     }
 
@@ -1102,6 +1103,7 @@ static para_data *make_para_data(int ptype, int paux, int indent, int rmargin,
     pdata->rect_type = RECT_NONE;
     pdata->contents_entry = NULL;
     pdata->justification = JUST;
+    pdata->extraflags = 0;
 
     /*
      * Choose fonts for this paragraph.
@@ -1346,14 +1348,80 @@ static font_encoding *new_font_encoding(font_data *font)
     fe->free_pos = 0x21;
 
     for (i = 0; i < 256; i++) {
-	fe->vector[i] = NULL;
-	fe->indices[i] = -1;
+	fe->vector[i] = NOGLYPH;
 	fe->to_unicode[i] = 0xFFFF;
     }
 
     return fe;
 }
 
+static subfont_map_entry *encode_glyph_at(glyph g, wchar_t u,
+					  font_encoding *fe, int pos)
+{
+    subfont_map_entry *sme = snew(subfont_map_entry);
+
+    sme->subfont = fe;
+    sme->position = pos;
+    fe->vector[pos] = g;
+    fe->to_unicode[pos] = u;
+    add234(fe->font->subfont_map, sme);
+    return sme;
+}
+
+static int new_sfmap_cmp(void *a, void *b)
+{
+    glyph ga = *(glyph *)a;
+    subfont_map_entry *sb = b;
+    glyph gb = sb->subfont->vector[sb->position];
+
+    if (ga < gb) return -1;
+    if (ga > gb) return 1;
+    return 0;
+}
+
+static subfont_map_entry *encode_glyph(glyph g, wchar_t u, font_data *font)
+{
+    subfont_map_entry *sme;
+    int c;
+
+    sme = find234(font->subfont_map, &g, new_sfmap_cmp);
+    if (sme) return sme;
+
+    /*
+     * This character is not yet in a subfont. Assign one.
+     */
+    if (font->latest_subfont->free_pos >= 0x100)
+	font->latest_subfont = new_font_encoding(font);
+
+    c = font->latest_subfont->free_pos++;
+    if (font->latest_subfont->free_pos == 0x7F)
+	font->latest_subfont->free_pos = 0xA1;
+
+    return encode_glyph_at(g, u, font->latest_subfont, c);
+}
+
+static int sfmap_cmp(void *a, void *b)
+{
+    subfont_map_entry *sa = a, *sb = b;
+    glyph ga = sa->subfont->vector[sa->position];
+    glyph gb = sb->subfont->vector[sb->position];
+
+    if (ga < gb) return -1;
+    if (ga > gb) return 1;
+    return 0;
+}
+
+int width_cmp(void *a, void *b)
+{
+    glyph_width const *wa = a, *wb = b;
+
+    if (wa->glyph < wb->glyph)
+	return -1;
+    if (wa->glyph > wb->glyph)
+	return 1;
+    return 0;
+}
+
 int kern_cmp(void *a, void *b)
 {
     kern_pair const *ka = a, *kb = b;
@@ -1369,50 +1437,35 @@ int kern_cmp(void *a, void *b)
     return 0;
 }
 
-/* This wouldn't be necessary if C had closures. */
-static font_info *glyph_cmp_fi;
-
-static int glyph_cmp(void const *a, void const *b)
+int lig_cmp(void *a, void *b)
 {
-    return strcmp(glyph_cmp_fi->glyphs[*(unsigned short *)a],
-		  glyph_cmp_fi->glyphs[*(unsigned short *)b]);
-}
+    ligature const *la = a, *lb = b;
 
-/*
- * Set up the glyphsbyname index for a font.
- */
-void font_index_glyphs(font_info *fi) {
-    int i;
+    if (la->left < lb->left)
+	return -1;
+    if (la->left > lb->left)
+	return 1;
+    if (la->right < lb->right)
+	return -1;
+    if (la->right > lb->right)
+	return 1;
+    return 0;
+}
 
-    fi->glyphsbyname = snewn(fi->nglyphs, unsigned short);
-    for (i = 0; i < fi->nglyphs; i++)
-	fi->glyphsbyname[i] = i;
-    glyph_cmp_fi = fi;
-    qsort(fi->glyphsbyname, fi->nglyphs, sizeof(fi->glyphsbyname[0]),
-	  glyph_cmp);
+static int utoglyph(font_info const *fi, wchar_t u) {
+    return (u < 0 || u > 0xFFFF ? NOGLYPH : fi->bmp[u]);
 }
 
-int find_glyph(font_info *fi, char const *name) {
-    int i, j, k, r;
-
-    i = -1;
-    j = fi->nglyphs;
-    while (j-i > 1) {
-	k = (i + j) / 2;
-	r = strcmp(fi->glyphs[fi->glyphsbyname[k]], name);
-	if (r == 0)
-	    return fi->glyphsbyname[k];
-	else if (r > 0)
-	    j = k;
-	else
-	    i = k;
-    }
-    return -1;
+void listfonts(void) {
+    font_info const *fi;
+
+    init_std_fonts();
+    for (fi = all_fonts; fi; fi = fi->next)
+	printf("%s\n", fi->name);
 }
 
 static font_data *make_std_font(font_list *fontlist, char const *name)
 {
-    int nglyphs;
     font_info const *fi;
     font_data *f;
     font_encoding *fe;
@@ -1430,8 +1483,7 @@ static font_data *make_std_font(font_list *fontlist, char const *name)
 
     f->list = fontlist;
     f->info = fi;
-    nglyphs = f->info->nglyphs;
-    f->subfont_map = snewn(nglyphs, subfont_map_entry);
+    f->subfont_map = newtree234(sfmap_cmp);
 
     /*
      * Our first subfont will contain all of US-ASCII. This isn't
@@ -1444,34 +1496,33 @@ static font_data *make_std_font(font_list *fontlist, char const *name)
     fe->free_pos = 0xA1;	       /* only the top half is free */
     f->latest_subfont = fe;
 
-    for (i = 0; i < nglyphs; i++) {
-	wchar_t ucs;
-	ucs = ps_glyph_to_unicode(f->info->glyphs[i]);
-	if (ucs >= 0x20 && ucs <= 0x7E) {
-	    fe->vector[ucs] = f->info->glyphs[i];
-	    fe->indices[ucs] = i;
-	    fe->to_unicode[ucs] = ucs;
-	    f->subfont_map[i].subfont = fe;
-	    f->subfont_map[i].position = ucs;
-	} else {
-	    /*
-	     * This character is not yet assigned to a subfont.
-	     */
-	    f->subfont_map[i].subfont = NULL;
-	    f->subfont_map[i].position = 0;
-	}
+    for (i = 0x20; i <= 0x7E; i++) {
+	glyph g = utoglyph(fi, i);
+	if (g != NOGLYPH)
+	    encode_glyph_at(g, i, fe, i);
     }
 
     return f;
 }
 
 /* NB: arguments are glyph numbers from font->bmp. */
+int find_width(font_data *font, glyph index)
+{
+    glyph_width wantw;
+    glyph_width const *w;
+
+    wantw.glyph = index;
+    w = find234(font->info->widths, &wantw, NULL);
+    if (!w) return 0;
+    return w->width;
+}
+
 static int find_kern(font_data *font, int lindex, int rindex)
 {
     kern_pair wantkp;
     kern_pair const *kp;
 
-    if (lindex == 0xFFFF || rindex == 0xFFFF)
+    if (lindex == NOGLYPH || rindex == NOGLYPH)
 	return 0;
     wantkp.left = lindex;
     wantkp.right = rindex;
@@ -1481,27 +1532,48 @@ static int find_kern(font_data *font, int lindex, int rindex)
     return kp->kern;
 }
 
-static int string_width(font_data *font, wchar_t const *string, int *errs)
+static int find_lig(font_data *font, int lindex, int rindex)
+{
+    ligature wantlig;
+    ligature const *lig;
+
+    if (lindex == NOGLYPH || rindex == NOGLYPH)
+	return NOGLYPH;
+    wantlig.left = lindex;
+    wantlig.right = rindex;
+    lig = find234(font->info->ligs, &wantlig, NULL);
+    if (lig == NULL)
+	return NOGLYPH;
+    return lig->lig;
+}
+
+static int string_width(font_data *font, wchar_t const *string, int *errs,
+			unsigned flags)
 {
     int width = 0;
-    int index, oindex;
+    int nindex, index, oindex, lindex;
 
     if (errs)
 	*errs = 0;
 
-    oindex = 0xFFFF;
+    oindex = NOGLYPH;
+    index = utoglyph(font->info, *string);
     for (; *string; string++) {
-	index = (*string < 0 || *string > 0xFFFF ? 0xFFFF :
-		 font->info->bmp[*string]);
+	nindex = utoglyph(font->info, string[1]);
 
-	if (index == 0xFFFF) {
+	if (index == NOGLYPH) {
 	    if (errs)
 		*errs = 1;
 	} else {
-	    width += find_kern(font, oindex, index) +
-		font->info->widths[index];
+	    if (!(flags & RS_NOLIG) &&
+		(lindex = find_lig(font, index, nindex)) != NOGLYPH) {
+		index = lindex;
+		continue;
+	    }
+	    width += find_kern(font, oindex, index) + find_width(font, index);
 	}
 	oindex = index;
+	index = nindex;
     }
 
     return width;
@@ -1529,6 +1601,7 @@ static int paper_width_internal(void *vctx, word *word, int *nspaces)
     struct paper_width_ctx *ctx = (struct paper_width_ctx *)vctx;
     int style, type, findex, width, errs;
     wchar_t *str;
+    unsigned flags = 0;
 
     switch (word->type) {
       case word_HyperLink:
@@ -1548,6 +1621,8 @@ static int paper_width_internal(void *vctx, word *word, int *nspaces)
 	      style == word_Emph ? FONT_EMPH :
 	      FONT_CODE);
 
+    if (style == word_Code || style == word_WeakCode) flags |= RS_NOLIG;
+
     if (type == word_Normal) {
 	str = word->text;
     } else if (type == word_WhiteSpace) {
@@ -1564,7 +1639,7 @@ static int paper_width_internal(void *vctx, word *word, int *nspaces)
 	    str = ctx->conf->rquote;
     }
 
-    width = string_width(ctx->pdata->fonts[findex], str, &errs);
+    width = string_width(ctx->pdata->fonts[findex], str, &errs, flags);
 
     if (errs && word->alt)
 	return paper_width_list(vctx, word->alt, NULL, nspaces);
@@ -1584,7 +1659,7 @@ static int paper_width_simple(para_data *pdata, word *text, paper_conf *conf)
     ctx.pdata = pdata;
     ctx.minspacewidth =
 	(pdata->sizes[FONT_NORMAL] *
-	 string_width(pdata->fonts[FONT_NORMAL], L" ", NULL));
+	 string_width(pdata->fonts[FONT_NORMAL], L" ", NULL, 0));
     ctx.conf = conf;
 
     return paper_width_list(&ctx, text, NULL, NULL);
@@ -1612,7 +1687,7 @@ static void wrap_paragraph(para_data *pdata, word *words,
     }
 
     spacewidth = (pdata->sizes[FONT_NORMAL] *
-		  string_width(pdata->fonts[FONT_NORMAL], L" ", NULL));
+		  string_width(pdata->fonts[FONT_NORMAL], L" ", NULL, 0));
     if (spacewidth == 0) {
 	/*
 	 * A font without a space?! Disturbing. I hope this never
@@ -1640,7 +1715,6 @@ static void wrap_paragraph(para_data *pdata, word *words,
 
     for (p = wrapping; p; p = p->next) {
 	line_data *ldata;
-	word *wd;
 	int len, wid, spaces;
 
 	ldata = snew(line_data);
@@ -1665,7 +1739,6 @@ static void wrap_paragraph(para_data *pdata, word *words,
 	spaces = 0;
 	len = paper_width_list(&ctx, ldata->first, ldata->end, &spaces);
 	wid = (p == wrapping ? w - i1 : w - i2);
-	wd = ldata->first;
 
 	ldata->hshortfall = wid - len;
 	ldata->nspaces = spaces;
@@ -1934,52 +2007,40 @@ static void add_string_to_page(page_data *page, int x, int y,
  * Returns the updated x coordinate.
  */
 static int render_string(page_data *page, font_data *font, int fontsize,
-			 int x, int y, wchar_t *str)
+			 int x, int y, wchar_t *str, unsigned flags)
 {
     char *text;
-    int textpos, textwid, kern, glyph, oglyph;
+    int textpos, textwid, kern, nglyph, glyph, oglyph, lig;
     font_encoding *subfont = NULL, *sf;
+    subfont_map_entry *sme;
 
     text = snewn(1 + ustrlen(str), char);
     textpos = textwid = 0;
 
-    glyph = 0xFFFF;
+    glyph = NOGLYPH;
+    nglyph = utoglyph(font->info, *str);
     while (*str) {
 	oglyph = glyph;
-	glyph = (*str < 0 || *str > 0xFFFF ? 0xFFFF :
-		 font->info->bmp[*str]);
+	glyph = nglyph;
+	nglyph = utoglyph(font->info, str[1]);
 
-	if (glyph == 0xFFFF) {
+	if (glyph == NOGLYPH) {
 	    str++;
 	    continue;		       /* nothing more we can do here */
 	}
 
+	if (!(flags & RS_NOLIG) &&
+	    (lig = find_lig(font, glyph, nglyph)) != NOGLYPH) {
+	    nglyph = lig;
+	    str++;
+	    continue;
+	}
+
 	/*
 	 * Find which subfont this character is going in.
 	 */
-	sf = font->subfont_map[glyph].subfont;
-
-	if (!sf) {
-	    int c;
-
-	    /*
-	     * This character is not yet in a subfont. Assign one.
-	     */
-	    if (font->latest_subfont->free_pos >= 0x100)
-		font->latest_subfont = new_font_encoding(font);
-
-	    c = font->latest_subfont->free_pos++;
-	    if (font->latest_subfont->free_pos == 0x7F)
-		font->latest_subfont->free_pos = 0xA1;
-
-	    font->subfont_map[glyph].subfont = font->latest_subfont;
-	    font->subfont_map[glyph].position = c;
-	    font->latest_subfont->vector[c] = font->info->glyphs[glyph];
-	    font->latest_subfont->indices[c] = glyph;
-	    font->latest_subfont->to_unicode[c] = *str;
-
-	    sf = font->latest_subfont;
-	}
+	sme = encode_glyph(glyph, *str, font);
+	sf = sme->subfont;
 
 	kern = find_kern(font, oglyph, glyph) * fontsize;
 
@@ -1997,8 +2058,8 @@ static int render_string(page_data *page, font_data *font, int fontsize,
 	    subfont = sf;
 	}
 
-	text[textpos++] = font->subfont_map[glyph].position;
-	textwid += font->info->widths[glyph] * fontsize;
+	text[textpos++] = sme->position;
+	textwid += find_width(font, glyph) * fontsize;
 
 	str++;
     }
@@ -2024,6 +2085,7 @@ static int render_text(page_data *page, para_data *pdata, line_data *ldata,
 	int style, type, findex, errs;
 	wchar_t *str;
 	xref_dest dest;
+	unsigned flags = 0;
 
 	switch (text->type) {
 	    /*
@@ -2150,11 +2212,14 @@ static int render_text(page_data *page, para_data *pdata, line_data *ldata,
 		  style == word_Emph ? FONT_EMPH :
 		  FONT_CODE);
 
+	if (style == word_Code || style == word_WeakCode) flags |= RS_NOLIG;
+	flags |= pdata->extraflags;
+
 	if (type == word_Normal) {
 	    str = text->text;
 	} else if (type == word_WhiteSpace) {
 	    x += pdata->sizes[findex] *
-		string_width(pdata->fonts[findex], L" ", NULL);
+		string_width(pdata->fonts[findex], L" ", NULL, 0);
 	    if (nspaces && findex != FONT_CODE) {
 		x += (*nspace+1) * shortfall / nspaces;
 		x -= *nspace * shortfall / nspaces;
@@ -2168,7 +2233,7 @@ static int render_text(page_data *page, para_data *pdata, line_data *ldata,
 		str = conf->rquote;
 	}
 
-	(void) string_width(pdata->fonts[findex], str, &errs);
+	(void) string_width(pdata->fonts[findex], str, &errs, flags);
 
 	if (errs && text->alt)
 	    x = render_text(page, pdata, ldata, x, y, text->alt, NULL,
@@ -2176,7 +2241,7 @@ static int render_text(page_data *page, para_data *pdata, line_data *ldata,
 			    conf);
 	else
 	    x = render_string(page, pdata->fonts[findex],
-			      pdata->sizes[findex], x, y, str);
+			      pdata->sizes[findex], x, y, str, flags);
 
 	if (*xr)
 	    (*xr)->rx = x;
@@ -2368,14 +2433,14 @@ static void render_para(para_data *pdata, paper_conf *conf,
 			      pdata->sizes[FONT_NORMAL],
 			      conf->left_margin + x,
 			      (conf->paper_height - conf->top_margin -
-			       pdata->last->ypos), L".");
+			       pdata->last->ypos), L".", 0);
 
 	render_string(pdata->last->page,
 		      pdata->fonts[FONT_NORMAL],
 		      pdata->sizes[FONT_NORMAL],
 		      conf->paper_width - conf->right_margin - wid,
 		      (conf->paper_height - conf->top_margin -
-		       pdata->last->ypos), num);
+		       pdata->last->ypos), num, 0);
     }
 
     /*
@@ -2422,6 +2487,7 @@ static para_data *code_paragraph(int indent, word *words, paper_conf *conf)
     pdata->rect_type = RECT_NONE;
     pdata->contents_entry = NULL;
     pdata->justification = LEFT;
+    pdata->extraflags = RS_NOLIG;
 
     for (; words; words = words->next) {
 	wchar_t *t, *e, *start;
@@ -2552,6 +2618,7 @@ static para_data *rule_paragraph(int indent, paper_conf *conf)
     pdata->rect_type = RECT_RULE;
     pdata->contents_entry = NULL;
     pdata->justification = LEFT;
+    pdata->extraflags = 0;
 
     standard_line_spacing(pdata, conf);