Make pret-lam remember subscripted words.

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

diff --git a/pret-lam.lua b/pret-lam.lua
index b882196ed522070743b40959e36a51940d8f5aee..1011b19e1672e24fb3cb572eaff45e4a4ffa17d5 100644 (file)
--- a/pret-lam.lua
+++ b/pret-lam.lua
@@ -61,40 +61,93 @@ 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+)(.*)")
+        res, newstr = utf.match(str, "^(%a[%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)
+    for i, pat in ipairs(patterns) do
+        a, b, c, d, e = utf.match(str, pat)
+        if a then
+            return a, b, c, d, e
+        end
+    end
+    return nil
+end
+
+function buffers.visualizers.lam.begin_of_buffer()
+    -- Initially allow subscripts using _ or just appending a number (later,
+    -- we will add extra patterns here.
+    submatches = {"^(.*)_([%a%d,]+)$", "^(.*[^%d])(%d+)$"}
+    -- 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.end_of_buffer()
+    -- Reset submatches and bases, since flush_line can be called without
+    -- begin / end_of_buffer for \type.
+    buffers.visualizers.lam.begin_of_buffer()
+    bases = nil
+end
+
+function buffers.visualizers.lam.do_subscripts(word)
+    local match_mul = buffers.visualizers.lam.match_mul
+    base, sub = match_mul(res, submatches)
+    if sub then
+        word = base .. "\\low{" .. sub .. "}"
+        -- After a word has been used as a base, allow subscripts
+        -- without _, even for non-numbers.
+        if bases and 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,]+)$"
+        end
+    end
+    return word
+end
+
 function buffers.visualizers.lam.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
-
     while str ~= "" do
         local found = false
-        local res
+        local word, symbol
         -- See if the next token is a word
-        res, str = take_word(str)
-        if res then
+        word, str = take_word(str)
+        if word then
             if buffers.visualizers.lam.keywords[res] then
-                -- Make all keywors bold
-                res = "{\\bold " .. res ..  "}"
+                -- Make all keywords bold
+                word = "{\\bold " .. word ..  "}"
+            else
+                -- Process any subscripts in the word
+                word = do_subscripts(word)
             end
-            -- All other words are unmodified
         else
             -- The next token is not a word, it must be a symbol
-            res, str = take_symbol(str)
+            symbol, str = take_symbol(str)
         end
 
         -- Append the resulting token
-        result[#result+1] = res
+        result[#result+1] = word or symbol
     end
 
     state = finish(state, result)
     buffers.flush_result(result,nested)
 end
 
+-- Call end_of_buffer once to set up submatches (since \type doesn't call
+-- begin_of_buffer / end_of_buffer).
+buffers.visualizers.lam.end_of_buffer()
+
 -- vim: set sw=4 sts=4 expandtab ai: