X-Git-Url: https://git.stderr.nl/gitweb?p=matthijs%2Fmaster-project%2Freport.git;a=blobdiff_plain;f=pret-lam.lua;h=eb7d3530395529a0b6ce1b934a4aa6ce4b38c467;hp=15fcf9a3ac1fdc6343dd3e90077047d713921983;hb=3ff8ced65f6af69db754b35b263c7d6d9f19e61f;hpb=95b1a682e51c47d635d5bafe75bd7096e8d4ec18 diff --git a/pret-lam.lua b/pret-lam.lua index 15fcf9a..eb7d353 100644 --- a/pret-lam.lua +++ b/pret-lam.lua @@ -6,55 +6,134 @@ 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", "prettyfour" } -buffers.visualizers.lam.tokens = { +-- Symbols that should have a different representation +local 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'}, } -function buffers.visualizers.lam.flush_line(str,nested) + +-- Keywords that should be bold +local 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. +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 + -- 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. +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. +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 + return a, b, c, d, e + end + end + return nil +end + +-- Find any subscripts in the given word and typeset them +local function do_subscripts(word) + 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,]+)$" + end + end + return word +end + +function vis.begin_of_display() + -- Initially allow subscripts using _ or just appending a number (later, + -- we will add extra patterns here. + submatches = {"^(.*)_([%a%d,]+)$", "^(.*[^%d])(%d+)$"} + -- 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 vis.flush_line(str,nested) local result, state = { }, 0 local finish, change = buffers.finish_state, buffers.change_state -- Set the colorscheme, which is used by finish_state and change_state - buffers.currentcolors = buffers.visualizers.lam.colors - + buffers.currentcolors = 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 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