Introduce word types for attributed spaces, so backends can distinguish

between emphasis _like_ _this_ and _like this_

[originally from svn r245]

5 files changed, 91 insertions, 11 deletions

diff --git a/bk_text.c b/bk_text.c
index 786c126..2b58c06 100644
--- a/bk_text.c
+++ b/bk_text.c
@@ -262,23 +262,42 @@ static void text_rdaddwc(rdstringc *rs, word *text, word *end) {
       case word_Emph:
       case word_Code:
       case word_WeakCode:
-	if (text->type == word_Emph)
+	if (text->type == word_Emph &&
+	    (text->aux == attr_First || text->aux == attr_Only))
 	    rdaddc(rs, '_');	       /* FIXME: configurability */
-	else if (text->type == word_Code)
+	else if (text->type == word_Code &&
+		 (text->aux == attr_First || text->aux == attr_Only))
 	    rdaddc(rs, '`');	       /* FIXME: configurability */
 	if (text_convert(text->text, &c))
 	    rdaddsc(rs, c);
 	else
 	    text_rdaddwc(rs, text->alt, NULL);
 	sfree(c);
-	if (text->type == word_Emph)
+	if (text->type == word_Emph &&
+	    (text->aux == attr_Last || text->aux == attr_Only))
 	    rdaddc(rs, '_');	       /* FIXME: configurability */
-	else if (text->type == word_Code)
+	else if (text->type == word_Code &&
+		 (text->aux == attr_Last || text->aux == attr_Only))
 	    rdaddc(rs, '\'');	       /* FIXME: configurability */
 	break;
 
       case word_WhiteSpace:
+      case word_EmphSpace:
+      case word_CodeSpace:
+      case word_WkCodeSpace:
+	if (text->type == word_EmphSpace &&
+	    (text->aux == attr_First || text->aux == attr_Only))
+	    rdaddc(rs, '_');	       /* FIXME: configurability */
+	else if (text->type == word_CodeSpace &&
+		 (text->aux == attr_First || text->aux == attr_Only))
+	    rdaddc(rs, '`');	       /* FIXME: configurability */
 	rdaddc(rs, ' ');
+	if (text->type == word_EmphSpace &&
+	    (text->aux == attr_Last || text->aux == attr_Only))
+	    rdaddc(rs, '_');	       /* FIXME: configurability */
+	else if (text->type == word_CodeSpace &&
+		 (text->aux == attr_Last || text->aux == attr_Only))
+	    rdaddc(rs, '\'');	       /* FIXME: configurability */
 	break;
     }
 }
@@ -308,13 +327,24 @@ static int text_width(word *text) {
       case word_Emph:
       case word_Code:
       case word_WeakCode:
-	return ((text->type == word_Emph || text->type == word_Code ? 2 : 0) +
+	return (((text->type == word_Emph ||
+		  text->type == word_Code)
+		 ? (text->aux == attr_Only ? 2 :
+		    text->aux == attr_Always ? 0 : 1)
+		 : 0) +
 		(text_convert(text->text, NULL) ?
 		 ustrlen(text->text) :
 		 text_width_list(text->alt)));
 
       case word_WhiteSpace:
-	return 1;
+      case word_EmphSpace:
+      case word_CodeSpace:
+      case word_WkCodeSpace:
+	return (((text->type == word_EmphSpace ||
+		  text->type == word_CodeSpace)
+		 ? (text->aux == attr_Only ? 2 :
+		    text->aux == attr_Always ? 0 : 1)
+		 : 0) + 1);
     }
     return 0;			       /* should never happen */
 }
diff --git a/buttress.h b/buttress.h
index b059322..1ce39d7 100644
--- a/buttress.h
+++ b/buttress.h
@@ -99,22 +99,41 @@ enum {
 struct word_Tag {
     word *next, *alt;
     int type;
+    int aux;
     wchar_t *text;
     filepos fpos;
 };
 enum {
+    /* ORDERING CONSTRAINT: these normal-word types ... */
     word_Normal,
     word_Emph,
     word_Code,			       /* monospaced; `quoted' in text */
     word_WeakCode,		       /* monospaced, normal in text */
+    /* ... must be in the same order as these space types ... */
+    word_WhiteSpace,		       /* text is NULL or ignorable */
+    word_EmphSpace,		       /* WhiteSpace when emphasised */
+    word_CodeSpace,		       /* WhiteSpace when code */
+    word_WkCodeSpace,		       /* WhiteSpace when weak code */
+    /* END ORDERING CONSTRAINT */
+    word_internal_endattrs,
     word_UpperXref,		       /* \K */
     word_LowerXref,		       /* \k */
     word_XrefEnd,		       /* (invisible; no text) */
     word_IndexRef,		       /* (always an invisible one) */
-    word_WhiteSpace,		       /* text is NULL or ignorable */
     word_HyperLink,		       /* (invisible) */
     word_HyperEnd		       /* (also invisible; no text) */
 };
+/* aux values for attributed words */
+enum {
+    attr_Only,			       /* a lone word with the attribute */
+    attr_First,			       /* the first of a series */
+    attr_Last,			       /* the last of a series */
+    attr_Always			       /* any other part of a series */
+};
+#define isattr(x) ( ( (x) > word_Normal && (x) < word_WhiteSpace ) || \
+                    ( (x) > word_WhiteSpace && (x) < word_internal_endattrs ) )
+#define sameattr(x,y) ( (x)-(y) == 0 || (x)-(y) == 4 || (x)-(y) == -4 )
+#define tospacestyle(x) ( (x) + 4 )
 
 /*
  * error.c
@@ -231,6 +250,8 @@ char *rdtrimc(rdstringc *rs);
 
 int compare_wordlists(word *a, word *b);
 
+void mark_attr_ends(paragraph *sourceform);
+
 typedef struct tagWrappedLine wrappedline;
 struct tagWrappedLine {
     wrappedline *next;
diff --git a/input.c b/input.c
index 32eca3b..97d732e 100644
--- a/input.c
+++ b/input.c
@@ -417,7 +417,7 @@ static void read_file(paragraph ***ret, input *in, index *idx) {
     paragraph par;
     word wd, **whptr, **idximplicit;
     wchar_t utext[2], *wdtext;
-    int style;
+    int style, spcstyle;
     int already;
     int iswhite, seenwhite;
     int type;
@@ -618,6 +618,7 @@ static void read_file(paragraph ***ret, input *in, index *idx) {
 	 */
 	parsestk = stk_new();
 	style = word_Normal;
+	spcstyle = word_WhiteSpace;
 	indexing = FALSE;
 	seenwhite = TRUE;
 	while (t.type != tok_eop && t.type != tok_eof) {
@@ -630,7 +631,7 @@ static void read_file(paragraph ***ret, input *in, index *idx) {
 		if (whptr == &par.words)
 		    break;	       /* strip whitespace at start of para */
 		wd.text = NULL;
-		wd.type = word_WhiteSpace;
+		wd.type = spcstyle;
 		wd.alt = NULL;
 		wd.fpos = t.pos;
 		if (indexing)
@@ -672,8 +673,10 @@ static void read_file(paragraph ***ret, input *in, index *idx) {
 			whptr = sitem->whptr;
 			idximplicit = sitem->idximplicit;
 		    }
-		    if (sitem->type & stack_style)
+		    if (sitem->type & stack_style) {
 			style = word_Normal;
+			spcstyle = word_WhiteSpace;
+		    }
 		    if (sitem->type & stack_idx) {
 			indexword->text = ustrdup(indexstr.text);
 			if (index_downcase)
@@ -810,6 +813,7 @@ static void read_file(paragraph ***ret, input *in, index *idx) {
 				style = (t.cmd == c_c ? word_Code :
 					 t.cmd == c_cw ? word_WeakCode :
 					 word_Emph);
+				spcstyle = tospacestyle(style);
 				sitem->type |= stack_style;
 			    }
 			    dtor(t), t = get_token(in);
@@ -842,6 +846,7 @@ static void read_file(paragraph ***ret, input *in, index *idx) {
 			style = (type == c_c ? word_Code :
 				 type == c_cw ? word_WeakCode :
 				 word_Emph);
+			spcstyle = tospacestyle(style);
 			sitem = mknew(struct stack_item);
 			sitem->type = stack_style;
 			stk_push(parsestk, sitem);
@@ -874,6 +879,7 @@ static void read_file(paragraph ***ret, input *in, index *idx) {
 			    style = (t.cmd == c_c ? word_Code :
 				     t.cmd == c_cw ? word_WeakCode :
 				     word_Emph);
+			    spcstyle = tospacestyle(style);
 			    sitem->type |= stack_style;
 			}
 			dtor(t), t = get_token(in);
diff --git a/main.c b/main.c
index 34ce7d4..2827f0e 100644
--- a/main.c
+++ b/main.c
@@ -188,6 +188,8 @@ int main(int argc, char **argv) {
 	if (!sourceform)
 	    exit(EXIT_FAILURE);
 
+	mark_attr_ends(sourceform);
+
 	sfree(infiles);
 
 	keywords = get_keywords(sourceform);
diff --git a/misc.c b/misc.c
index 3d512c6..e68ed42 100644
--- a/misc.c
+++ b/misc.c
@@ -132,6 +132,26 @@ int compare_wordlists(word *a, word *b) {
 	return 0;
 }
 
+void mark_attr_ends(paragraph *sourceform) {
+    paragraph *p;
+    word *w, *wp;
+    for (p = sourceform; p; p = p->next) {
+	wp = NULL;
+	for (w = p->words; w; w = w->next) {
+	    if (isattr(w->type)) {
+		int before = (wp && isattr(wp->type) &&
+			      sameattr(wp->type, w->type));
+		int after = (w->next && isattr(w->next->type) &&
+			     sameattr(w->next->type, w->type));
+		w->aux = (before ?
+			  (after ? attr_Always : attr_Last) :
+			  (after ? attr_First : attr_Only));
+	    }
+	    wp = w;
+	}
+    }
+}
+
 wrappedline *wrap_para(word *text, int width, int subsequentwidth,
 		       int (*widthfn)(word *)) {
     wrappedline *head = NULL, **ptr = &head;
@@ -151,7 +171,8 @@ wrappedline *wrap_para(word *text, int width, int subsequentwidth,
 	thiswidth = lastgood = 0;
 	while (text) {
 	    thiswidth += widthfn(text);
-	    if (text->next && text->next->type == word_WhiteSpace) {
+	    if (text->next && (text->next->type == word_WhiteSpace ||
+			       text->next->type == word_EmphSpace)) {
 		if (thiswidth > width)
 		    break;
 		spc = text->next;