1 -- filename : type-lam.lua
2 -- comment : Pretty printing of (extended) lambda calculus
3 -- author : Matthijs Kooijman, Universiteit Twente, NL
4 -- copyright: Matthijs Kooijman
7 local utf = unicode.utf8
9 local vis = buffers.newvisualizer("lam")
18 -- Symbols that should have a different representation
20 -- Note, the space we replace with is a Unicode non-breaking space
22 {symbol = ' ', repr = ' '},
23 {symbol = '_', repr = '\\_'},
24 {symbol = '->>', repr = '\\twoheadrightarrow'},
25 {symbol = '->', repr = '→'},
26 {symbol = '=>', repr = '⇒'},
27 -- The default * sits very high above the baseline, \ast (u+2217) looks
29 {symbol = '*', repr = '\\ast'},
30 {symbol = '~', repr = '\\HDLine[width=.20 * \\the\\textwidth]'},
31 {symbol = '|', repr = '\\char' .. utf.byte('|')},
32 -- Use ▶ and ∧ from our roman font, since Iwona doesn't have the glyph
33 {symbol = '▶', repr = '{\\rm{}▶}'},
34 --{symbol = '∧', repr = '{$∧$}'},
35 {symbol = '∧', repr = '{\\rm{}∧}'},
38 -- Keywords that should be bold
46 ['DEFAULT'] = {small = true},
52 -- Store the last line for each indent level
53 local indentlines = {}
55 function array_concat(a1, a2)
57 for k,v in ipairs(a2) do
64 -- See if str starts with a symbol, and return the remaining string and that
65 -- symbol. If no symbol from the table is matched, just returns the first
66 -- character. We can do a lookup directly, since symbols can be different in
67 -- length, so we just loop over all symbols, trying them in turn.
68 local function take_symbol(str)
69 for i,props in ipairs(symbols) do
70 -- Try to remove symbol from the start of str
71 symbol, newstr = utf.match(str, "^(" .. props.symbol .. ")(.*)")
73 -- Return this tokens repr, or just the token if it has no
75 res = props.repr or symbol
76 -- Enclose the token in {\style .. }
78 res = "{\\" .. props.style .. " " .. res .. "}"
83 -- No symbol found, just return the first character
84 return utf.match(str, "^(.)(.*)")
87 -- Take a single word from str, if posible. Returns the rest of the string and
89 local function take_word(str)
90 -- A word must always start with a-z (in particular, λ is not a valid
91 -- start of a word). A word must always end with a letter or a digit
92 res, newstr = utf.match(str, "^([a-zA-Z][%a%d%+%-%,_]*[%a%d]+)(.*)")
94 -- The above does not catch single letter words
95 res, newstr = utf.match(str, "^([a-zA-Z])(.*)")
97 return res, newstr or str
100 -- Tries to match each of the patterns and returns the captures of the first
101 -- matching pattern (up to 5 captures are supported). Returns nil when nothing
103 local function match_mul(str, patterns)
104 for i, pat in ipairs(patterns) do
105 a, b, c, d, e = utf.match(str, pat)
113 -- Find any subscripts in the given word and typeset them
114 local function do_subscripts(word)
115 base, sub = match_mul(res, submatches)
117 word = base .. "\\low{" .. sub .. "}"
118 -- After a word has been used as a base, allow subscripts
119 -- without _, even for non-numbers.
120 if not bases[base] then
121 -- Register that we've added this base
123 -- Add a patterns for this base. First, the base with a single
124 -- letter or number subscript.
125 submatches[#submatches+1] = "^(" .. base .. ")([%a%d])$"
126 -- Seconde, the base with a longer prefix that includes at least
127 -- one of +-, (to catch things like ri+1, but not return).
128 submatches[#submatches+1] = "^(" .. base .. ")([%a%d]*[%-%+,]+[%a%d%-%+,]*)$"
134 -- Do proper aligning for subsequent lines. For example, in
137 -- We replace the spaces in the second line with a skip with the same with as
138 -- "foo ", to align the | with the =.
139 -- For this, we keep a table "indentlines", which contains all previous lines
140 -- with smaller indent levels that are still "in scope" (e.g., have not yet
141 -- been followed by a line with a smaller indent level). For example:
147 -- After the last line, the table will contain:
148 -- { 0 = "line1", 2 = " line4", 4 = " line5"}
149 -- In other words, line3 is no longer in scope since it is "hidden" by
150 -- line4, and line is no longer in scope since it is replaced by line4.
151 local function do_indent(line)
152 newind, rest = utf.match(line, '^(%s*)(.*)')
154 -- Loop all the previous lines
155 for indent, unused in pairs(indentlines) do
156 if indent > #newind then
157 -- Remove any lines with a larger indent
158 indentlines[indent] = nil
159 elseif indent < #newind and indent > prev then
160 -- Find the last line (e.g, with the highest indent) with an
161 -- indent smaller than the new indent. This is the line from which
162 -- we need to copy the indent.
167 -- Always store this line, possibly overwriting a previous line with the
169 indentlines[#newind] = line
172 -- If there is a previous line with a smaller indent, make sure we
173 -- align with it. We do this by taking a prefix from that previous
174 -- line just as long as our indent. This gives us a bunch of
175 -- whitespace, with a few non-whitespace characters. We find out the
176 -- width of this prefix, and put whitespace just as wide as that
177 -- prefix before the current line, instead of the whitespace
178 -- characters that were there.
179 -- Doing this is slightly risky, since the prefix might contain
180 -- unfinished markup (e.g., \foo{bar without the closing }). We might
181 -- need to solve this later.
182 copyind = utf.sub(indentlines[prev], 1, #newind)
183 setwidth = "\\setwidthof{" .. copyind .. "}\\to\\pretlamalignwidth"
184 hskip = "\\hskip\\pretlamalignwidth"
185 return "{" .. setwidth .. hskip .. "}" .. rest
187 -- No previous line? Just return the unmodified line then
192 -- Mark the begin of a block of lambda formatted buffers or expressions. This
193 -- means that, until you call end_of_block again, the subscript bases are
194 -- shared. For example, if you have \lam{y1} some text \lam{yn} within a
195 -- single block, the yn will properly get subscripted. Be sure to call
196 -- end_of_block again!
198 -- Blocks can be partially nested, meaning that the block
199 -- won't be closed until end_of_block was called exactly as often as
200 -- begin_of_block. However, subscripts from the inner block can still
201 -- influence subscripts in the outer block.
202 function vis.begin_of_block()
203 vis.begin_of_display()
204 in_block = in_block + 1
207 -- Ends the current block
208 function vis.end_of_block()
209 in_block = in_block - 1
212 function vis.begin_of_display()
213 if in_block == 0 then
214 -- Initially allow subscripts using _ or just appending a number (later,
215 -- we will add extra patterns here.
216 submatches = {"^(%a*)_([%a%d,]+)$", "^(%a+)(%d[%d,]*)$"}
217 -- This stores all the bases we've encountered so far (to prevent
218 -- duplicates). For each of them there will be a pattern in submatches
226 -- Make things work for inline typeing (e.g., \type{}) as well.
227 vis.begin_of_inline = vis.begin_of_display
228 vis.end_of_inline = vis.end_of_display
230 function vis.flush_line(str,nested)
231 buffers.flush_result(vis.do_line(str, false), nested)
234 function vis.do_line(str, no_indent)
236 if not no_indent then
237 -- Allow ignore of the indentation stuff when we're calling ourselves
238 -- for a partial line.
244 local text, rest = utf.match(str, "^%-%-(.-)%-%-(.*)")
246 table.insert(result, '\\strikethrough{')
247 -- Recursively call ourselves to handle spaces gracefully.
248 result = array_concat(result, vis.do_line(text, true))
249 table.insert(result, '}')
250 -- Eat the processed characters
252 elseif utf.match(str, "^%-%-") then
253 table.insert(result, '{\\italic{--')
254 -- Recursively call ourselves to handle spaces gracefully.
255 result = array_concat(result, vis.do_line(utf.sub(str, 3), true))
256 table.insert(result, '}}')
257 -- Done with this line
260 -- See if the next token is a word
261 word, str = take_word(str)
263 if keywords[res] then
264 -- Make all keywords bold
265 word = "{\\bold " .. word .. "}"
266 if keywords[res].small then
267 word = "\\small" .. word -- Curlies were added above
270 -- Process any subscripts in the word
271 word = do_subscripts(word)
273 table.insert(result, word)
275 -- The next token is not a word, it must be a symbol
276 symbol, str = take_symbol(str)
277 table.insert(result, symbol)
285 -- vim: set sw=4 sts=4 expandtab ai: