bibtex 0.99d-test1 from Oren, breaking .bbl lines only at whitespace

git-svn-id: svn://tug.org/texlive/trunk@17253 c570f23f-e606-0410-a88d-b1316a301751

5 files changed, 116 insertions, 46 deletions

diff --git a/Build/source/texk/web2c/ChangeLog b/Build/source/texk/web2c/ChangeLog
index 2fe43bf253e..5618b0fae0c 100644
--- a/Build/source/texk/web2c/ChangeLog
+++ b/Build/source/texk/web2c/ChangeLog
@@ -1,3 +1,13 @@
+2010-03-01  Karl Berry  <karl@tug.org>
+
+	* bibtex.web: 0.99d test 1 from Oren, breaking .bbl lines only at
+	whitespace.
+	* bibtex.ch: remove no-longer-needed workaround for \% break.
+
+	* tests/longline.tex,
+	* tests/longline.bib: new test files.
+	(Still working on their incorporation into make check.)
+
 2010-02-22  Karl Berry  <karl@tug.org>
 
 	* configure.ac: (web2_version): move to 2010/dev.
diff --git a/Build/source/texk/web2c/bibtex.ch b/Build/source/texk/web2c/bibtex.ch
index 9a3376af6be..a9d02437095 100644
--- a/Build/source/texk/web2c/bibtex.ch
+++ b/Build/source/texk/web2c/bibtex.ch
@@ -1101,24 +1101,24 @@ end;
 % ignored.  This is a special case of the general problem that we shouldn't
 % split macro invocations either -- however, the best way to avoid that is
 % not to split lines at all.
-@x [324] Check whether we're trying to break a \% combo.
-out_buf[end_ptr] := out_buf[max_print_line-1];  {save this character}
-out_buf[max_print_line-1] := comment;           {so \TeX\ does the thing right}
-out_buf_length := max_print_line;
-break_ptr := out_buf_length - 1;        {the `|-1|' allows for the restoration}
-output_bbl_line;                                {output what we can,}
-out_buf[max_print_line-1] := out_buf[end_ptr];  {restore this character}
-@y
-if out_buf[max_print_line-1] = comment then {assume \% combo here}
-  out_buf_length := max_print_line - 1
-else
-  out_buf_length := max_print_line;
-break_ptr := out_buf_length - 1;        {the `|-1|' allows for the restoration}
-out_buf[end_ptr] := out_buf[break_ptr];  {save this character}
-out_buf[break_ptr] := comment;           {so \TeX\ does the thing right}
-output_bbl_line;                                {output what we can,}
-out_buf[break_ptr] := out_buf[end_ptr];  {restore this character}
-@z
+% @x [324] Check whether we're trying to break a \% combo.
+% out_buf[end_ptr] := out_buf[max_print_line-1];  {save this character}
+% out_buf[max_print_line-1] := comment;           {so \TeX\ does the thing right}
+% out_buf_length := max_print_line;
+% break_ptr := out_buf_length - 1;        {the `|-1|' allows for the restoration}
+% output_bbl_line;                                {output what we can,}
+% out_buf[max_print_line-1] := out_buf[end_ptr];  {restore this character}
+% @y
+% if out_buf[max_print_line-1] = comment then {assume \% combo here}
+%   out_buf_length := max_print_line - 1
+% else
+%   out_buf_length := max_print_line;
+% break_ptr := out_buf_length - 1;        {the `|-1|' allows for the restoration}
+% out_buf[end_ptr] := out_buf[break_ptr];  {save this character}
+% out_buf[break_ptr] := comment;           {so \TeX\ does the thing right}
+% output_bbl_line;                                {output what we can,}
+% out_buf[break_ptr] := out_buf[end_ptr];  {restore this character}
+% @z
 
 @x [327] Add check for fieldinfo[] overflow.
     field_ptr := cite_ptr*num_fields + fn_info[ex_fn_loc];
diff --git a/Build/source/texk/web2c/bibtex.web b/Build/source/texk/web2c/bibtex.web
index e465a33c48d..614143d5e2c 100644
--- a/Build/source/texk/web2c/bibtex.web
+++ b/Build/source/texk/web2c/bibtex.web
@@ -1,4 +1,5 @@
-% This program is copyright (C) 1985 by Oren Patashnik; all rights reserved.
+% This program is copyright (C) 1985, 1988, 2010 by Oren Patashnik;
+% all rights reserved.
 % Copying of this file is authorized only if (1) you are Oren Patashnik, or if
 % (2) you make absolutely no changes to your copy. (The WEB system provides
 % for alterations via an auxiliary file; the master file should stay intact.)
@@ -32,8 +33,12 @@
 % Version 0.99c was released in February 1988.  It removed two begin-end pairs
 %       that, for convention only, surrounded entire modules, but that elicited
 %       label-related complaints from some compilers.
+% Version 0.99d was released in March 2010.  It made output lines breakable
+%       only at white_space (so that, for example, URLs would not be broken).
+%       Other known bugs (all minor) will be fixed in a subsequent release.
+%   THIS IS BibTeX 0.99d TEST VERSION 1;    PLEASE DO NOT RELEASE
 
-% Please report any bugs to Oren Patashnik (PATASHNIK@@SCORE.STANFORD.EDU)
+% Please report any bugs to biblio@@tug.org
 
 % Although considerable effort has been expended to make the BibTeX program
 % correct and reliable, no warranty is implied; the author disclaims any
@@ -131,7 +136,7 @@ style writers would be eternally grateful, if only they knew.
 The |banner| string defined here should be changed whenever \BibTeX\
 gets modified.
 
-@d banner=='This is BibTeX, Version 0.99c' {printed when the program starts}
+@d banner=='This is BibTeX, Version 0.99dtest' {printed when the program starts}
 
 
 @
@@ -7573,14 +7578,16 @@ This procedure adds to the output buffer the given string in
 length of the current string in |out_buf|, and thus also gives the
 location for the next character.  If there are enough characters
 present in the output buffer, it writes one or more lines out to the
-\.{.bbl} file.  It may break a line at any |white_space| character it
-likes, but if it does, it will add two |space|s to the next output
-line.
+\.{.bbl} file.  It breaks a line only at a |white_space| character,
+and when it does, it adds two |space|s to the next output line.
 
 @<Procedures and functions for style-file function execution@>=
 procedure add_out_pool (@!p_str : str_number);
+label loop1_exit,loop2_exit;
 var break_ptr : buf_pointer;    {the first character following the line break}
 @!end_ptr : buf_pointer;        {temporary end-of-buffer pointer}
+break_pt_found : boolean;       {a suitable |white_space| character}
+unbreakable_tail : boolean;     {because it contains no |white_space| character}
 begin
 p_ptr1 := str_start[p_str];
 p_ptr2 := str_start[p_str+1];
@@ -7594,7 +7601,8 @@ while (p_ptr1 < p_ptr2) do
     incr(out_buf_ptr);
     end;
 out_buf_length := out_buf_ptr;
-while (out_buf_length > max_print_line) do
+unbreakable_tail := false;
+while ((out_buf_length > max_print_line) and (not unbreakable_tail)) do
     @<Break that line@>;
 end;
 
@@ -7604,19 +7612,25 @@ Here we break the line by looking for a |white_space| character,
 backwards from |out_buf[max_print_line]| until
 |out_buf[min_print_line]|; we break at the |white_space| and indent
 the next line two |space|s.  The next module handles things when
-there's no |white_space| character to break at.
+there's no |white_space| character to break at.  (It seems that the
+annoyances to the average user of a warning message when there's an
+output line longer than |max_print_line| outweigh the benefits, so we
+don't issue such warnings in the current code.)
 
 @<Break that line@>=
 begin
 end_ptr := out_buf_length;
 out_buf_ptr := max_print_line;
+break_pt_found := false;
 while ((lex_class[out_buf[out_buf_ptr]] <> white_space) and
                                         (out_buf_ptr >= min_print_line)) do
     decr(out_buf_ptr);
 if (out_buf_ptr = min_print_line-1) then        {no |white_space| character}
-    @<Break that unbreakable line@>
+    @<Break that unbreakably long line@>        {(if |white_space| follows)}
 else
-    begin                                       {hit a |white_space| character}
+    break_pt_found := true;                     {hit a |white_space| character}
+if (break_pt_found) then
+    begin
     out_buf_length := out_buf_ptr;
     break_ptr := out_buf_length + 1;
     output_bbl_line;                    {output what we can}
@@ -7636,27 +7650,35 @@ end
 
 
 @
-If there's no |white_space| character to break the line at, we break
-it at |out_buf[max_print_line-1]|, append a |comment| character, and
-don't indent the next line.
+If there's no |white_space| character up through
+|out_buf[max_print_line]|, we instead break the line at the first
+following |white_space| character, if one exists.  And if, starting
+with that |white_space| character, there are multiple consecutive
+|white_space| characters, |out_buf_ptr| points to the last of them.
+If no |white_space| character exists, we haven't found a viable break
+point, so we don't break the line (yet).
 
-@<Break that unbreakable line@>=
+@<Break that unbreakably long line@>=
 begin
-out_buf[end_ptr] := out_buf[max_print_line-1];  {save this character}
-out_buf[max_print_line-1] := comment;           {so \TeX\ does the thing right}
-out_buf_length := max_print_line;
-break_ptr := out_buf_length - 1;        {the `|-1|' allows for the restoration}
-output_bbl_line;                                {output what we can,}
-out_buf[max_print_line-1] := out_buf[end_ptr];  {restore this character}
-out_buf_ptr := 0;
-tmp_ptr := break_ptr;
-while (tmp_ptr < end_ptr) do                    {and slide the rest down}
-    begin
-    out_buf[out_buf_ptr] := out_buf[tmp_ptr];
-    incr(out_buf_ptr);
-    incr(tmp_ptr);
+out_buf_ptr := max_print_line + 1;      {|break_pt_found| is still |false|}
+while (out_buf_ptr < end_ptr) do
+    if (lex_class[out_buf[out_buf_ptr]] <> white_space) then
+        incr(out_buf_ptr)
+    else
+        goto loop1_exit;
+loop1_exit:
+if (out_buf_ptr = end_ptr) then
+    unbreakable_tail := true            {because no |white_space| character}
+else                            {hit |white_space|, and |out_buf_ptr < end_ptr|}
+    begin
+    break_pt_found := true;
+    while (out_buf_ptr+1 < end_ptr) do  {look for more |white_space|}
+        if (lex_class[out_buf[out_buf_ptr+1]] = white_space) then
+            incr(out_buf_ptr)           {which then points to |white_space|}
+        else
+            goto loop2_exit;
+loop2_exit:
     end;
-out_buf_length := end_ptr - break_ptr;
 end
 
 
diff --git a/Build/source/texk/web2c/tests/longline.bib b/Build/source/texk/web2c/tests/longline.bib
new file mode 100644
index 00000000000..096fc495314
--- /dev/null
+++ b/Build/source/texk/web2c/tests/longline.bib
@@ -0,0 +1,24 @@
+@misc{long-bare,
+  howpublished = "http://long.example.org/1234567890a1234567890b1234567890c1234567890d1234567890e1234567890f1234567890g",
+}
+
+@misc{long-with-author,
+  author = "Long Author",
+  howpublished = "http://long.example.org/1234567890a1234567890b1234567890c1234567890d1234567890e1234567890f1234567890g",
+}
+
+@misc{long-full,
+  author = "Full Long Author",
+  title = "http://long.example.org/1234567890a1234567890b1234567890c1234567890d1234567890e1234567890f1234567890g",
+  note = "Some note.",
+}
+
+@misc{long-author,
+  author = "Really Long Aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaauthor",
+  title = "Titleeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee too",
+}
+
+@misc{normal,
+  author = "Normal Author",
+  howpublished = "http://example.org/",
+}
diff --git a/Build/source/texk/web2c/tests/longline.tex b/Build/source/texk/web2c/tests/longline.tex
new file mode 100644
index 00000000000..44c76395609
--- /dev/null
+++ b/Build/source/texk/web2c/tests/longline.tex
@@ -0,0 +1,14 @@
+% $Id$
+% Public domain.  Originally written 2010, Karl Berry.
+% Test for bibtex not breaking when longer than max_print_line.
+
+\input btxmac
+
+\let\mbox = \hbox
+
+\bibliography{longline}
+\bibliographystyle{plain}
+
+\nocite{*}
+
+\end