Improve subscript handling in pret-lam.

[matthijs/master-project/report.git] / pret-lam.lua

diff --git a/pret-lam.lua b/pret-lam.lua
index c05f987bd71747b26b23c7a31b8775e4d8db1974..ea07811759ae40920b5ef4d6a4f8e174a3e3fa1f 100644 (file)
--- a/pret-lam.lua
+++ b/pret-lam.lua
@@ -6,11 +6,9 @@
 
 local utf = unicode.utf8
 
-if not buffers                 then buffers                 = { } end
-if not buffers.visualizers     then buffers.visualizers     = { } end
-if not buffers.visualizers.lam then buffers.visualizers.lam = { } end
+vis = buffers.newvisualizer("lam")
 
-buffers.visualizers.lam.colors = {
+local colors = {
     "prettytwo",
     "prettyone",
     "prettythree",
@@ -18,18 +16,19 @@ buffers.visualizers.lam.colors = {
 }
 
 -- Symbols that should have a different representation
-buffers.visualizers.lam.symbols = {
+local symbols = {
     [' '] = {repr = '\\obs '},
     ['_'] = {repr = '\\_'},
     ['->'] = {repr = '\\rightarrow'},
     -- The default * sits very high above the baseline, \ast (u+2217) looks
     -- better.
     ['*'] = {repr = '\\ast'},
+    ['~'] = {repr = '\\sim'},
 }
 
 
 -- Keywords that should be bold
-buffers.visualizers.lam.keywords = {
+local keywords = {
     ['case'] = {},
     ['of'] = {},
     ['let'] = {},
@@ -40,8 +39,8 @@ buffers.visualizers.lam.keywords = {
 -- symbol. If no symbol from the table is matched, just returns the first
 -- character.  We can do a lookup directly, since symbols can be different in
 -- length, so we just loop over all symbols, trying them in turn.
-function buffers.visualizers.lam.take_symbol(str)
-    for symbol,props in pairs(buffers.visualizers.lam.symbols) do
+local function take_symbol(str)
+    for symbol,props in pairs(symbols) do
         -- Try to remove symbol from the start of str 
         symbol, newstr = utf.match(str, "^(" .. symbol .. ")(.*)")
         if symbol then
@@ -61,15 +60,17 @@ end
 
 -- Take a single word from str, if posible. Returns the rest of the string and
 -- the word taken.
-function buffers.visualizers.lam.take_word(str)
-        res, newstr = utf.match(str, "^(%a[%a%d_]+)(.*)")
+local function take_word(str)
+        -- A word must always start with a-z (in particular, λ is not a valid
+        -- start of a word).
+        res, newstr = utf.match(str, "^([a-zA-Z][%a%d%+%-%,_]+)(.*)")
         return res, newstr or str
 end
 
 -- Tries to match each of the patterns and returns the captures of the first
 -- matching pattern (up to 5 captures are supported). Returns nil when nothing
 -- matches.
-function buffers.visualizers.lam.match_mul(str, patterns)
+local function match_mul(str, patterns)
     for i, pat in ipairs(patterns) do
         a, b, c, d, e = utf.match(str, pat)
         if a then
@@ -80,8 +81,7 @@ function buffers.visualizers.lam.match_mul(str, patterns)
 end
 
 -- Find any subscripts in the given word and typeset them
-function buffers.visualizers.lam.do_subscripts(word)
-    local match_mul = buffers.visualizers.lam.match_mul
+local function do_subscripts(word)
     base, sub = match_mul(res, submatches)
     if sub then
         word = base .. "\\low{" .. sub .. "}"
@@ -90,38 +90,43 @@ function buffers.visualizers.lam.do_subscripts(word)
         if not bases[base] then
             -- Register that we've added this base
             bases[base] = true
-            -- Add a pattern for this base
-            submatches[#submatches+1] = "^(" .. base .. ")([%a%d,]+)$"
+            -- Add a patterns for this base. First, the base with a single
+            -- letter or number subscript.
+            submatches[#submatches+1] = "^(" .. base .. ")([%a%d])$"
+            -- Seconde, the base with a longer prefix that includes at least
+            -- one of +-, (to catch things like ri+1, but not return).
+            submatches[#submatches+1] = "^(" .. base .. ")([%a%d]*[%-%+%,]+[%a%d%-%+%,]*)$"
         end
     end
     return word
 end
 
-function buffers.visualizers.lam.begin_of_buffer(type, name)
+function vis.begin_of_display()
     -- Initially allow subscripts using _ or just appending a number (later,
     -- we will add extra patterns here.
-    submatches = {"^(.*)_([%a%d,]+)$", "^(.*[^%d])(%d+)$"}
+    submatches = {"^(%a*)_([%a%d,]+)$", "^(%a+)(%d+)$"}
     -- This stores all the bases we've encountered so far (to prevent
     -- duplicates). For each of them there will be a pattern in submatches
     -- above.
     bases = {}
 end
 
-function buffers.visualizers.lam.flush_line(str,nested)
+-- Make things work for inline typeing (e.g., \type{}) as well.
+vis.begin_of_inline = vis.begin_of_display
+vis.end_of_inline = vis.end_of_display
+
+function vis.flush_line(str,nested)
     local result, state = { }, 0
     local finish, change = buffers.finish_state, buffers.change_state
-    local take_symbol = buffers.visualizers.lam.take_symbol
-    local take_word = buffers.visualizers.lam.take_word
-    local do_subscripts = buffers.visualizers.lam.do_subscripts
     -- Set the colorscheme, which is used by finish_state and change_state
-    buffers.currentcolors = buffers.visualizers.lam.colors
+    buffers.currentcolors = colors
     while str ~= "" do
         local found = false
         local word, symbol
         -- See if the next token is a word
         word, str = take_word(str)
         if word then
-            if buffers.visualizers.lam.keywords[res] then
+            if keywords[res] then
                 -- Make all keywords bold
                 word = "{\\bold " .. word ..  "}"
             else