Normalize whitespaces in codespans

CHANGES.md

@@ -13,6 +13,8 @@
 
 - Updated colors for code fragments (@EmileTrotignon, #1023)
 - Fixed complexity of looking up `.odoc` files (@panglesd, #1075)
+- Normalize whitespaces in codespans (@gpetiot, #1085)
+  A newline followed by any whitespaces is normalized as one space character.
 
 ### Changed
 

src/parser/lexer.mll

@@ -609,13 +609,16 @@ and code_span buffer nesting_level start_offset input = parse
     { Buffer.add_char buffer c;
       code_span buffer nesting_level start_offset input lexbuf }
 
-  | newline newline
+  | newline horizontal_space* (newline horizontal_space*)+
     { warning
         input
         (Parse_error.not_allowed
           ~what:(Token.describe (`Blank_line "\n\n"))
           ~in_what:(Token.describe (`Code_span "")));
-      Buffer.add_char buffer '\n';
+      Buffer.add_char buffer ' ';
+      code_span buffer nesting_level start_offset input lexbuf }
+  | newline horizontal_space+
+    { Buffer.add_char buffer ' ';
       code_span buffer nesting_level start_offset input lexbuf }
 
   | eof

src/parser/test/test.ml

@@ -817,8 +817,7 @@ let%expect_test _ =
         {|
     ((output
       (((f.ml (1 0) (3 4))
-        (paragraph (((f.ml (1 0) (3 4)) (code_span  "foo\
-                                                   \nbar")))))))
+        (paragraph (((f.ml (1 0) (3 4)) (code_span "foo bar")))))))
      (warnings
       ( "File \"f.ml\", line 1, character 4 to line 3, character 0:\
        \nBlank line is not allowed in '[...]' (code)."))) |}]
@@ -829,8 +828,7 @@ let%expect_test _ =
         {|
     ((output
       (((f.ml (1 0) (3 4))
-        (paragraph (((f.ml (1 0) (3 4)) (code_span  "foo\
-                                                   \nbar")))))))
+        (paragraph (((f.ml (1 0) (3 4)) (code_span "foo bar")))))))
      (warnings
       ( "File \"f.ml\", line 1, character 4 to line 3, character 0:\
        \nBlank line is not allowed in '[...]' (code)."))) |}]

test/generators/cases/markup.mli

@@ -57,6 +57,10 @@
     Code can appear {b inside [other] markup}. Its display shouldn't be
     affected.
 
+    There is no differences between [a b] and [a
+    b].
+
+    Consecutive whitespaces not after a newline are conserved as they are: [a   b].
 
     {1 Links and references}
 

test/generators/html/Markup.html

@@ -123,6 +123,11 @@ <h6 id="subparagraph"><a href="#subparagraph" class="anchor"></a>
    <p>Code can appear <b>inside <code>other</code> markup</b>. Its display
      shouldn't be affected.
    </p>
+   <p>There is no differences between <code>a b</code> and <code>a b</code>.
+   </p>
+   <p>Consecutive whitespaces not after a newline are conserved as they
+     are: <code>a   b</code>.
+   </p>
    <h2 id="links-and-references">
     <a href="#links-and-references" class="anchor"></a>Links and references
    </h2>

test/generators/latex/Markup.tex

@@ -34,6 +34,10 @@ \subsection{Styling\label{styling}}%
 
 Code can appear \bold{inside \ocamlinlinecode{other} markup}. Its display shouldn't be affected.
 
+There is no differences between \ocamlinlinecode{a b} and \ocamlinlinecode{a b}.
+
+Consecutive whitespaces not after a newline are conserved as they are: \ocamlinlinecode{a   b}.
+
 \subsection{Links and references\label{links-and-references}}%
 This is a \href{\#}{link}\footnote{\url{\#}}. It sends you to the top of this page. Links can have markup inside them: \href{\#}{\bold{bold}}\footnote{\url{\#}}, \href{\#}{\emph{italics}}\footnote{\url{\#}}, \href{\#}{\emph{emphasis}}\footnote{\url{\#}}, \href{\#}{super\textsuperscript{script}}\footnote{\url{\#}}, \href{\#}{sub\textsubscript{script}}\footnote{\url{\#}}, and \href{\#}{\ocamlinlinecode{code}}\footnote{\url{\#}}. Links can also be nested \emph{\href{\#}{inside}\footnote{\url{\#}}} markup. Links cannot be nested inside each other. This link has no replacement text: \href{\#}{\#}\footnote{\url{\#}}. The text is filled in by odoc. This is a shorthand link: \href{\#}{\#}\footnote{\url{\#}}. The text is also filled in by odoc in this case.
 

test/generators/man/Markup.3o

@@ -85,6 +85,10 @@ It's possible for two markup elements to appear \fBnext\fR \fIto\fR each other a
 This is also true between non-code markup and code\.
 .sp 
 Code can appear \fBinside other markup\fR\. Its display shouldn't be affected\.
+.sp 
+There is no differences between a b and a b\.
+.sp 
+Consecutive whitespaces not after a newline are conserved as they are: a   b\.
 .nf 
 .sp 
 .in 3