From b3c80b25d640d5641882482a2461980c90297274 Mon Sep 17 00:00:00 2001
From: simon <simon@cda61777-01e9-0310-a592-d414129be87e>
Date: Mon, 12 Aug 2002 11:24:28 +0000
Subject: [PATCH] Arrange for multiple \cfg, \IM and \BR paragraphs to not
 require blank-line separators (config directives visually _want_ to be in a
 single chunk).

git-svn-id: svn://svn.tartarus.org/sgt/halibut@1837 cda61777-01e9-0310-a592-d414129be87e
---
 input.c         | 50 ++++++++++++++++++++++++++++++++++++++++++++------
 inputs/test.but |  6 ++++--
 2 files changed, 48 insertions(+), 8 deletions(-)

diff --git a/input.c b/input.c
index 15e0a97..2e0dabf 100644
--- a/input.c
+++ b/input.c
@@ -528,11 +528,13 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 
     t.text = NULL;
     macros = newtree234(macrocmp);
+    already = FALSE;
 
     /*
      * Loop on each paragraph.
      */
     while (1) {
+	int start_cmd = c__invalid;
 	par.words = NULL;
 	par.keyword = NULL;
 	whptr = &par.words;
@@ -540,7 +542,10 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 	/*
 	 * Get a token.
 	 */
-	dtor(t), t = get_token(in);
+	if (!already) {
+	    dtor(t), t = get_token(in);
+	}
+	already = FALSE;
 	if (t.type == tok_eof)
 	    return;
 
@@ -620,19 +625,22 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		 */
 	      case c_A: needkw = 2; par.type = para_Appendix; break;
 	      case c_B: needkw = 2; par.type = para_Biblio; break;
-	      case c_BR: needkw = 1; par.type = para_BR; break;
+	      case c_BR: needkw = 1; par.type = para_BR;
+		start_cmd = c_BR; break;
 	      case c_C: needkw = 2; par.type = para_Chapter; break;
 	      case c_H: needkw = 2; par.type = para_Heading;
 		par.aux = 0;
 		break;
-	      case c_IM: needkw = 2; par.type = para_IM; break;
+	      case c_IM: needkw = 2; par.type = para_IM;
+		start_cmd = c_IM; break;
 	      case c_S: needkw = 2; par.type = para_Subsect;
 		par.aux = t.aux; break;
 	      case c_U: needkw = 32; par.type = para_UnnumberedChapter; break;
 		/* For \b and \n the keyword is optional */
 	      case c_b: needkw = 4; par.type = para_Bullet; break;
 	      case c_n: needkw = 4; par.type = para_NumberedList; break;
-	      case c_cfg: needkw = 8; par.type = para_Config; break;
+	      case c_cfg: needkw = 8; par.type = para_Config;
+		start_cmd = c_cfg; break;
 	      case c_copyright: needkw = 32; par.type = para_Copyright; break;
 	      case c_define: is_macro = TRUE; needkw = 1; break;
 		/* For \nocite the keyword is _everything_ */
@@ -709,12 +717,21 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 
 		/* Move to EOP in case of needkw==8 or 16 (no body) */
 		if (needkw & 24) {
-		    if (t.type != tok_eop && t.type != tok_eof) {
+		    /* We allow whitespace even when we expect no para body */
+		    while (t.type == tok_white)
+			dtor(t), t = get_token(in);
+		    if (t.type != tok_eop && t.type != tok_eof &&
+			(start_cmd == c__invalid ||
+			 t.type != tok_cmd || t.cmd != start_cmd)) {
 			error(err_bodyillegal, &t.pos);
 			/* Error recovery: eat the rest of the paragraph */
-			while (t.type != tok_eop && t.type != tok_eof)
+			while (t.type != tok_eop && t.type != tok_eof &&
+			       (start_cmd == c__invalid ||
+				t.type != tok_cmd || t.cmd != start_cmd))
 			    dtor(t), t = get_token(in);
 		    }
+		    if (t.type == tok_cmd)
+			already = TRUE;/* inhibit get_token at top of loop */
 		    addpara(par, ret);
 		    continue;	       /* next paragraph */
 		}
@@ -745,6 +762,14 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 	while (t.type != tok_eop && t.type != tok_eof) {
 	    iswhite = FALSE;
 	    already = FALSE;
+
+	    /* Handle implicit paragraph breaks after \IM, \BR etc */
+	    if (start_cmd != c__invalid &&
+		t.type == tok_cmd && t.cmd == start_cmd) {
+		already = TRUE;	       /* inhibit get_token at top of loop */
+		break;
+	    }
+
 	    if (t.type == tok_cmd && t.cmd == c__escaped) {
 		t.type = tok_word;     /* nice and simple */
 		t.aux = 0;	       /* even if `\-' - nonbreaking! */
@@ -765,6 +790,19 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		wd.aux = 0;
 		wd.fpos = t.pos;
 		wd.breaks = FALSE;
+
+		/*
+		 * Inhibit use of whitespace if it's (probably the
+		 * newline) before a repeat \IM / \BR type
+		 * directive.
+		 */
+		if (start_cmd != c__invalid) {
+		    dtor(t), t = get_token(in);
+		    already = TRUE;
+		    if (t.type == tok_cmd && t.cmd == start_cmd)
+			break;
+		}
+
 		if (indexing)
 		    rdadd(&indexstr, ' ');
 		if (!indexing || index_visible)
diff --git a/inputs/test.but b/inputs/test.but
index 5849e52..8ac6e86 100644
--- a/inputs/test.but
+++ b/inputs/test.but
@@ -19,7 +19,7 @@ a bit]
 
 \define{coopt} co\u00F6{-o}pt
 
-\versionid $Id: test.but,v 1.18 2002/08/05 10:31:33 simon Exp $
+\versionid $Id: test.but,v 1.19 2002/08/12 11:24:28 simon Exp $
 
 \C{ch\\ap} First chapter title; for similar wrapping reasons this
 chapter title will be ludicrously long. I wonder how much more
@@ -105,7 +105,8 @@ and here's a reference to it: \k{sub-sub}.
 \# \cfg{winhelp-topic}{Y5VQEXZQVJ} (uncomment this and it clashes)
 
 Here's an \i{appendix}, for no terribly good reason at all. See
-\k{book}.
+__\k{book}__ (please excuse those underscores, I'm testing
+whitespace).
 
 It also contains a \W{http://www.tartarus.org/}{hyperlink}.
 
@@ -117,6 +118,7 @@ It also contains a \W{http://www.tartarus.org/}{hyperlink}.
 the document even though there is no \cw{\\k} citing it.
 
 \BR{book} [SillyCitation]
+\BR{uncited} Badger.
 
 \nocite{nocite}
 
-- 
2.11.0