X-Git-Url: https://git.stderr.nl/gitweb?p=matthijs%2Fmaster-project%2Freport.git;a=blobdiff_plain;f=pret-lam.lua;h=f14b962ba2cd162a8f9743e73aeef8e30b695776;hp=94dd81085a9aa53617c262c12315a094864d82fe;hb=1108517b5c84c5fc7c8aa21ac60cd33e28487519;hpb=207cddf520024052b4da9040aa3a9636f30924db

diff --git a/pret-lam.lua b/pret-lam.lua
index 94dd810..f14b962 100644
--- a/pret-lam.lua
+++ b/pret-lam.lua
@@ -17,45 +17,129 @@ buffers.visualizers.lam.colors = {
     "prettyfour"
 }
 
-buffers.visualizers.lam.tokens = {
+-- Symbols that should have a different representation
+buffers.visualizers.lam.symbols = {
     [' '] = {repr = '\\obs '},
     ['_'] = {repr = '\\_'},
     ['->'] = {repr = '\\rightarrow'},
-    ['case'] = {style = 'bold'},
-    ['of'] = {style = 'bold'},
-    ['let'] = {style = 'bold'},
+    -- 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 = {
+    ['case'] = {},
+    ['of'] = {},
+    ['let'] = {},
+    ['in'] = {},
+}
+
+-- See if str starts with a symbol, and return the remaining string and that
+-- 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
+        -- Try to remove symbol from the start of str 
+        symbol, newstr = utf.match(str, "^(" .. symbol .. ")(.*)")
+        if symbol then
+            -- Return this tokens repr, or just the token if it has no
+            -- repr.
+            res = props.repr or symbol
+            -- Enclose the token in {\style .. }
+            if props.style then
+                res = "{\\" .. props.style ..  " " .. res ..  "}"
+            end
+            return res, newstr
+        end
+    end
+    -- No symbol found, just return the first character
+    return utf.match(str, "^(.)(.*)")
+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)
+        -- 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)
+    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
+
+-- 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
+    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 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,]+)$"
+        end
+    end
+    return word
+end
+
+function buffers.visualizers.lam.begin_of_buffer(type, name)
+    -- Initially allow subscripts using _ or just appending a number (later,
+    -- we will add extra patterns here.
+    submatches = {"^(.*)_([%a%d,]+)$", "^(.*[^%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.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
-        found = false
-        for tok,props in pairs(buffers.visualizers.lam.tokens) do
-            -- Try to remove tok from the start of str 
-            str, count = utf.gsub(str, "^" .. tok, "")
-            if count ~= 0 then
-                -- Replace the token with the repr, or just itself
-                result[#result+1] = props.repr or tok
-                -- Enclose the token in {\style .. }
-                if props.style then
-                    result[#result] = "{\\" .. props.style ..
-                                        " " .. result[#result] ..
-                                        "}"
-                end
-                -- We found a token, now start over
-                found = true
-                break
+        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
+                -- Make all keywords bold
+                word = "{\\bold " .. word ..  "}"
+            else
+                -- Process any subscripts in the word
+                word = do_subscripts(word)
             end
+        else
+            -- The next token is not a word, it must be a symbol
+            symbol, str = take_symbol(str)
         end
-        if not found then
-            result[#result+1] = utf.sub(str, 1, 1)
-            str = utf.sub(str, 2)
-        end
+
+        -- Append the resulting token
+        result[#result+1] = word or symbol
     end
+
     state = finish(state, result)
     buffers.flush_result(result,nested)
 end