extended punctuation

1 files changed, 49 insertions, 7 deletions

diff --git a/rst_parser.lua b/rst_parser.lua
index c489849..99265bd 100644
--- a/rst_parser.lua
+++ b/rst_parser.lua
@@ -581,7 +581,8 @@ local parser = P{
 -- Paragraphs * Inline Markup
 --------------------------------------------------------------------------------
 
-    paragraph = -(V"double_dot" + V"double_underscore") --  + V"bullet_indent") 
+    --paragraph = -(V"double_dot" + V"double_underscore") --  + V"bullet_indent") 
+    paragraph = -V"punctuation"
               * Cs((V"included_literal_block"
                   + V"enclosed_inline"
                   + V"inline_elements" 
@@ -752,22 +753,63 @@ local parser = P{
 
     groupchars = S"()[]{}",
 
+    --- Punctuation
+    -- Some of the following are used for markup as well as for punctuation.
+
     comma = P",",
     colon = P":",
     double_colon = V"colon" * V"colon",
     escaped_colon = V"backslash" * V"colon",
     dot = P".",
+    period = V"dot",
     double_dot = V"dot" * V"dot",
-    semicolon = P";",
-    questionmark = P"?",
-    exclamationmark = P"!",
-    punctuation = V"comma" + V"colon" + V"dot" + V"semicolon" +
-                  V"questionmark" + V"exclamationmark",--+ V"dash",
-
+    interpunct = P"·",
     underscore = P"_",
     double_underscore = V"underscore" * V"underscore",
     dash = P"-",
     double_dash = V"dash" * V"dash",
+    dashes = V"dash" + P"‒" + P"–" + P"—" + P"―",
+    hyphen = P"‐",
+    semicolon = P";",
+    questionmark = P"?",
+    exclamationmark = P"!",
+    inverted_exclamationmark = P"¡",
+    inverted_questionmark = P"¿",
+    interrobang = P"‽",
+
+    apostrophe = P"’" + P"'",
+    --brackets   = P"[ ], (",, { }, ⟨ ⟩ )
+    ellipsis   = P"…" + P"...",
+    guillemets = P"«" + P"»",
+    quotationmarks= P"‘" + P"’" + P"“" + P"”",
+    solidus= P"⁄",
+
+
+    punctuation = V"apostrophe"
+                + V"colon" 
+                + V"comma" 
+                + V"dashes"
+                + V"dot" 
+                + V"ellipsis"
+                + V"exclamationmark"
+                + V"guillemets"
+                + V"hyphen"
+                + V"interpunct"
+                + V"interrobang"
+                + V"questionmark" 
+                + V"quotationmarks"
+                + V"semicolon" 
+                + V"slash"
+                + V"solidus"
+                + V"underscore"
+                ,
+
+    -- These are treated separately as the might begin a paragraph (sigh!).
+    inverted_punctuation = V"inverted_exclamationmark"
+                         + V"inverted_questionmark",
+
+    -- End punctuation
+
     letter = R"az" + R"AZ",
 
     equals = P"=",