Refactor makefile; Add hapax preprocessor

7 files changed, 133 insertions, 51 deletions

@@ -0,0 +1,29 @@
+#!/bin/bash
+
+# Command-line variables
+searchQuery="$1";               # .html file to backlink
+outFile="$2";                   # .back file to create
+headerFile="$3";                # header information file
+shift 3;
+glob="$@";                      # where to search for backlinks
+
+# Find backlinkers
+echo -n "Back-linking \"$searchQuery\""
+cat "$headerFile" > "$outFile";
+grep -ql "$searchQuery" $glob >> "$outFile";
+
+# Change title of $outFile
+inText="`basename $searchQuery .html`.txt";
+title=`grep '^title:' "$inText" | cut -d' ' -f2-`;
+sed -i "s/_TITLE_/$title/" "$outFile";
+echo -n "."
+
+# Link to backlinks
+for file in `grep '.html' "$outFile"`; do
+    fText="`basename $file .html`.txt";
+    title=`grep '^title:' $fText | cut -d' ' -f2-`;
+    sed -i "s/^$file$/- [$title](&)/" "$outFile";
+    echo -n "."
+done
+
+echo "Done."
@@ -0,0 +1,17 @@
+-- Preprocessor for hapax.lua writer
+-- because for some damn reason, UTF-8 confuses things
+
+import Text.Pandoc.JSON
+import Data.Char (isAscii)
+
+main :: IO ()
+main = toJSONFilter unFancy
+
+unFancy :: Inline -> Inline
+unFancy (Str s) = Str $ map makeAscii s
+unFancy x       = x
+
+makeAscii :: Char -> Char
+makeAscii c
+    | isAscii c = c
+    | otherwise = ' '
@@ -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)
@@ -0,0 +1,26 @@
+#!/bin/bash
+
+file="$1";
+headFile="$2";
+shift 2;
+glob="$@";                       # *.hapax
+
+tempFile="${RANDOM}.tmp"
+
+echo -n "Linking \"$file\""
+# Begin
+cat "$headFile" > "$tempFile";
+echo -n "."
+# Link words to files they appear in
+for word in `sort "$file"`; do
+    f=`grep -liwq "^$word$" $glob`;
+    link="`basename $f .hapax`.html"
+    echo "[$word]($link)" >> "$tempFile";
+    echo -n "."
+done
+
+# Make the changes happen
+rm "$file"
+mv "$tempFile" "$file"
+
+echo "Done."
@@ -6,7 +6,7 @@
 function _lozenge() {
     var lozenge = document.getElementById('lozenge');
     // array with all files {{{
-    var files=["100-lines.html","about-the-author.html","about.html","about_author.html","abstract.html","amber-alert.html","and.html","angeltoabraham.html","apollo11.html","arspoetica.html","art.html","axe.html","big-dipper.html","boar.html","boy_bus.html","building.html","call-me-aural-pleasure.html","cereal.html","cold-wind.html","collage-instrument.html","common-titles.html","creation-myth.html","deadman.html","death-zone.html","deathstrumpet.html","dollywood.html","dream.html","early.html","elegyforanalternateself.html","epigraph.html","ex-machina.html","exasperated.html","father.html","feedingtheraven.html","finding-the-lion.html","fire.html","first-lines.html","found-typewriter-poem.html","hands.html","hapax.html","hard-game.html","hardware.html","howithappened.html","howtoread.html","hymnal.html","i-am.html","i-think-its-you.html","i-want-to-say.html","i-wanted-to-tell-you-something.html","in-bed.html","initial-conditions.html","january.html","joke.html","lappel-du-vide.html","largest-asteroid.html","last-bastion.html","last-passenger.html","leaf.html","leg.html","likingthings.html","listen.html","love-as-god.html","lovesong.html","man.html","manifesto_poetics.html","moon-drowning.html","moongone.html","mountain.html","movingsideways.html","music-433.html","no-nothing.html","notes.html","nothing-is-ever-over.html","on-genre-dimension.html","onformalpoetry.html","options.html","ouroboros_memory.html","paul.html","peaches.html","philosophy.html","phone.html","planks.html","plant.html","poetry-time.html","prelude.html","problems.html","process.html","proverbs.html","punch.html","purpose-dogs.html","question.html","real-writer.html","reports.html","riptide_memory.html","ronaldmcdonald.html","roughgloves.html","sapling.html","seasonal-affective-disorder.html","sense-of-it.html","serengeti.html","shed.html","shipwright.html","sixteenth-chapel.html","snow.html","something-simple.html","spittle.html","squirrel.html","stagnant.html","statements-frag.html","stayed-on-the-bus.html","stump.html","swansong-alt.html","swansong.html","swear.html","table_contents.html","tapestry.html","telemarketer.html","the-night-we-met.html","the-sea_the-beach.html","theoceanoverflowswithcamels.html","time-looks-up-to-the-sky.html","todaniel.html","toilet.html","toothpaste.html","treatise.html","underwear.html","walking-in-the-rain.html","wallpaper.html","weplayedthosegamestoo.html","what-we-are-made-of.html","when-im-sorry-i.html","window.html","words-irritable-reaching.html","words-meaning.html","worse-looking-over.html","writing.html","x-ray.html","yellow.html"]
+    var files=["100-lines.html","about-the-author.html","about.html","about_author.html","abstract.html","amber-alert.html","and.html","angeltoabraham.html","apollo11.html","arspoetica.html","art.html","axe.html","big-dipper.html","boar.html","boy_bus.html","building.html","call-me-aural-pleasure.html","cereal.html","cold-wind.html","collage-instrument.html","creation-myth.html","deadman.html","death-zone.html","deathstrumpet.html","dollywood.html","dream.html","early.html","elegyforanalternateself.html","epigraph.html","ex-machina.html","exasperated.html","father.html","feedingtheraven.html","finding-the-lion.html","fire.html","found-typewriter-poem.html","hands.html","hard-game.html","hardware.html","howithappened.html","howtoread.html","hymnal.html","i-am.html","i-think-its-you.html","i-want-to-say.html","i-wanted-to-tell-you-something.html","in-bed.html","initial-conditions.html","january.html","joke.html","lappel-du-vide.html","largest-asteroid.html","last-bastion.html","last-passenger.html","leaf.html","leg.html","likingthings.html","listen.html","love-as-god.html","lovesong.html","man.html","manifesto_poetics.html","moon-drowning.html","moongone.html","mountain.html","movingsideways.html","music-433.html","no-nothing.html","notes.html","nothing-is-ever-over.html","on-genre-dimension.html","onformalpoetry.html","options.html","ouroboros_memory.html","paul.html","peaches.html","philosophy.html","phone.html","planks.html","plant.html","poetry-time.html","prelude.html","problems.html","process.html","proverbs.html","punch.html","purpose-dogs.html","question.html","real-writer.html","reports.html","riptide_memory.html","ronaldmcdonald.html","roughgloves.html","sapling.html","seasonal-affective-disorder.html","sense-of-it.html","serengeti.html","shed.html","shipwright.html","sixteenth-chapel.html","snow.html","something-simple.html","spittle.html","squirrel.html","stagnant.html","statements-frag.html","stayed-on-the-bus.html","stump.html","swansong-alt.html","swansong.html","swear.html","table_contents.html","tapestry.html","telemarketer.html","the-night-we-met.html","the-sea_the-beach.html","theoceanoverflowswithcamels.html","time-looks-up-to-the-sky.html","todaniel.html","toilet.html","toothpaste.html","treatise.html","underwear.html","walking-in-the-rain.html","wallpaper.html","weplayedthosegamestoo.html","what-we-are-made-of.html","when-im-sorry-i.html","window.html","words-irritable-reaching.html","words-meaning.html","worse-looking-over.html","writing.html","x-ray.html","yellow.html"]
     // }}}
 
     var index = Math.floor(Math.random() * files.length);
@@ -0,0 +1,15 @@
+#!/bin/bash
+
+outFile="$1";
+shift;
+fileList="$@";
+
+echo -n "Updating \"lozenge.js\"..."
+
+list=`echo ${fileList[@]} |\
+      sed -e 's/\(\S\+.html\) \?/"\1",/g'\
+          -e 's/^\(.*\),$/var files=[\1]/'`
+
+sed -i "s/var files=.*/$list/" "$outFile";
+
+echo "Done."