1 files changed, 45 insertions, 50 deletions

@@ -1,8 +1,8 @@
--- Pandoc River writer
+-- Pandoc Hapax writer
 -- it takes out all formatting, leaving only a river of text
--- running down the page: one word per line
+-- running down the page: one word per line, stripping all duplicates
 -- vim: fdm=marker
--- invoke with: pandoc -t river.lua
+-- invoke with: pandoc -t hapax.lua
 
 os.setlocale("en_US.UTF-8")
 
@@ -40,36 +40,22 @@ function flow(s)
 end
 
 function nude(s)
+    s = s:lower()
     -- Expand contractions
-    s = s:gsub("'%a+%s", function (x)
-        if x == "'ll" then
-            return " will "
-        elseif x == "'ve" then
-            return " have "
-        elseif x == "'re" then
-            return " are "
-        else
-            return x
-        end
-    end)
+    s = s:gsub("'ll", " will ")
+    s = s:gsub("'ve", " have ")
+    s = s:gsub("'re", " are ")
+    s = s:gsub("i'm", " i am ")
     s = s:gsub("it's", "it is")
     s = s:gsub("n't", " not ")
-    -- Get rid of quotes around words
-    s = s:gsub('"', ' ')
-    s = s:gsub("%s'", ' ')
-    s = s:gsub("'%s", ' ')
-    -- Remove HTML entities
-    s = s:gsub('&.-;', ' ')
-    s = s:gsub('%b<>', ' ')
-    -- Remove end-of-line backslashes
-    s = s:gsub('\\$', ' ')
-    -- Remove dashes (not hyphens)
+    s = s:gsub("&", " and ")
+    -- -- Remove dashes (not hyphens)
     s = s:gsub('%-[%-%s]+', ' ')
     -- Remove everything that is not letters or numbers
-    s = s:gsub('[%.!%?:;,%[%]%(%)<>]', ' ')
+    s = s:gsub('[^A-Za-z0-9/\'-]', ' ')
     -- Remove extra spaces
     s = s:gsub('%s+', ' ')
-    return s:lower()
+    return " "..s.." "
 end
 
 -- This function is called once for the whole document. Parameters:
@@ -89,71 +75,72 @@ function Doc(body, metadata, variables)
     end
     add(body)
     return hapax(flow(buffer))
+    -- return flow(buffer)
 end
 
 -- Remove all formatting {{{
 function Note(s)
-    return nude(s)
+    return s
 end
 
 function Blocksep()
     return "\n"
 end
 function Emph(s)
-    return nude(s)
+    return s
 end
 
 function Strong(s)
-    return nude(s)
+    return s
 end
 
 function Subscript(s)
-    return nude(s)
+    return s
 end
 
 function Superscript(s)
-    return nude(s)
+    return s
 end
 
 function SmallCaps(s)
-    return nude(s)
+    return s
 end
 
 function Strikeout(s)
-    return nude(s)
+    return s
 end
 
 function Code(s, attr)
-    return nude(s)
+    return s
 end
 
 function CodeBlock(s, attr)
-    return nude(s)
+    return s
 end
 
 function InlineMath(s)
-    return nude(s)
+    return s
 end
 
 function DisplayMath(s)
-    return nude(s)
+    return s
 end
 
 function Span(s, attr)
-    return nude(s)
+    return s
 end
 
 function Cite(s)
-    return nude(s)
+    return s
 end
 
 function Plain(s)
-    return nude(s)
+    return s
 end
 
 -- Links only include the link text
 function Link(s, src, tit)
-    return nude(s)
+    return s
 end
 
 -- Images have nothing to give us
@@ -162,16 +149,24 @@ function Image(s, src, tit)
     return "\n"
 end
 
+function RawBlock(s)
+    return s
+end
+
+function RawInline(s)
+    return s
+end
+
 function CaptionedImage(s, src, tit)
     return "\n"
 end
 
 function Str(s)
-    return nude(s)
+    return s
 end
 
 function Div(s, attr)
-    return nude(s)
+    return s
 end
 
 function Space(s)
@@ -183,15 +178,15 @@ function LineBreak()
 end
 
 function Para(s)
-    return nude(s)
+    return s
 end
 
 function Header(lev, s, attr)
-    return nude(s)
+    return s
 end
 
 function BlockQuote(s)
-    return nude(s)
+    return s
 end
 
 function HorizontalRule()
@@ -201,7 +196,7 @@ end
 function BulletList(items)
     local buffer = ""
     for _, item in pairs(items) do
-        buffer = buffer .. nude(item) .. "\n"
+        buffer = buffer .. " " .. item .. "\n"
     end
     return buffer .. "\n"
 end
@@ -209,7 +204,7 @@ end
 function OrderedList(items)
     local buffer = ""
     for _, item in pairs(items) do
-        buffer = buffer .. nude(item) .. "\n"
+        buffer = buffer .. " " .. item .. "\n"
     end
     return buffer .. "\n"
 end
@@ -218,7 +213,7 @@ function DefinitionList(items)
     local buffer = ""
     for _, item in pairs(items) do
         for k, v in pairs(item) do
-            buffer = buffer .. nude(k) .. "\n" .. nude(v) .. "\n"
+            buffer = buffer .. " " .. k .. "\n" .. v .. "\n"
         end
     end
     return buffer .. "\n"
@@ -227,7 +222,7 @@ end
 function Table(caption, aligns, widths, headers, rows)
     local buffer = ""
     local function add(s)
-        buffer = buffer .. nude(s) .. "\n"
+        buffer = buffer .. " " .. s .. "\n"
     end
     if caption ~= "" then
         add(caption)