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 [' '] = {repr = '\\obs '},
21 ['_'] = {repr = '\\_'},
22 ['->'] = {repr = '\\rightarrow'},
23 -- The default * sits very high above the baseline, \ast (u+2217) looks
25 ['*'] = {repr = '\\ast'},
26 ['~'] = {repr = '\\HDLine[width=.20 * \\the\\textwidth]'},
27 ['|'] = {repr = '\\char' .. utf.byte('|')},
28 ['$'] = {repr = '\\char' .. utf.byte('$')},
31 -- Keywords that should be bold
45 -- Store the last line for each indent level
46 local indentlines = {}
47 -- The amount of indent of the first line, which we will strip of all
48 -- subsequent lines. This allows the entire block to be indented as normal in
49 -- the tex source code.
52 -- See if str starts with a symbol, and return the remaining string and that
53 -- symbol. If no symbol from the table is matched, just returns the first
54 -- character. We can do a lookup directly, since symbols can be different in
55 -- length, so we just loop over all symbols, trying them in turn.
56 local function take_symbol(str)
57 for symbol,props in pairs(symbols) do
58 -- Try to remove symbol from the start of str
59 symbol, newstr = utf.match(str, "^(" .. symbol .. ")(.*)")
61 -- Return this tokens repr, or just the token if it has no
63 res = props.repr or symbol
64 -- Enclose the token in {\style .. }
66 res = "{\\" .. props.style .. " " .. res .. "}"
71 -- No symbol found, just return the first character
72 return utf.match(str, "^(.)(.*)")
75 -- Take a single word from str, if posible. Returns the rest of the string and
77 local function take_word(str)
78 -- A word must always start with a-z (in particular, λ is not a valid
80 res, newstr = utf.match(str, "^([a-zA-Z][%a%d%+%-%,_]+)(.*)")
81 return res, newstr or str
84 -- Tries to match each of the patterns and returns the captures of the first
85 -- matching pattern (up to 5 captures are supported). Returns nil when nothing
87 local function match_mul(str, patterns)
88 for i, pat in ipairs(patterns) do
89 a, b, c, d, e = utf.match(str, pat)
97 -- Find any subscripts in the given word and typeset them
98 local function do_subscripts(word)
99 base, sub = match_mul(res, submatches)
101 word = base .. "\\low{" .. sub .. "}"
102 -- After a word has been used as a base, allow subscripts
103 -- without _, even for non-numbers.
104 if not bases[base] then
105 -- Register that we've added this base
107 -- Add a patterns for this base. First, the base with a single
108 -- letter or number subscript.
109 submatches[#submatches+1] = "^(" .. base .. ")([%a%d])$"
110 -- Seconde, the base with a longer prefix that includes at least
111 -- one of +-, (to catch things like ri+1, but not return).
112 submatches[#submatches+1] = "^(" .. base .. ")([%a%d]*[%-%+%,]+[%a%d%-%+%,]*)$"
118 -- Do proper aligning for subsequent lines. For example, in
121 -- We replace the spaces in the second line with a skip with the same with as
122 -- "foo ", to align the | with the =.
123 -- For this, we keep a table "indentlines", which contains all previous lines
124 -- with smaller indent levels that are still "in scope" (e.g., have not yet
125 -- been followed by a line with a smaller indent level). For example:
131 -- After the last line, the table will contain:
132 -- { 0 = "line1", 2 = " line4", 4 = " line5"}
133 -- In other words, line3 is no longer in scope since it is "hidden" by
134 -- line4, and line is no longer in scope since it is replaced by line4.
135 local function do_indent(line)
136 newind, rest = utf.match(line, '^(%s*)(.*)')
138 -- Store the first line's indent
139 if not first_indent then
140 first_indent = utf.len(newind)
143 -- Strip the indent of the first line from this line's indent.
144 newind = utf.sub(newind, first_indent + 1)
145 -- Rebuild line, so we can still use it below
146 line = newind .. rest
149 -- Loop all the previous lines
150 for indent, unused in pairs(indentlines) do
151 if indent > #newind then
152 -- Remove any lines with a larger indent
153 indentlines[indent] = nil
154 elseif indent < #newind and indent > prev then
155 -- Find the last line (e.g, with the highest indent) with an
156 -- indent smaller than the new indent. This is the line from which
157 -- we need to copy the indent.
162 -- Always store this line, possibly overwriting a previous line with the
164 indentlines[#newind] = line
167 -- If there is a previous line with a smaller indent, make sure we
168 -- align with it. We do this by taking a prefix from that previous
169 -- line just as long as our indent. This gives us a bunch of
170 -- whitespace, with a few non-whitespace characters. We find out the
171 -- width of this prefix, and put whitespace just as wide as that
172 -- prefix before the current line, instead of the whitespace
173 -- characters that were there.
174 -- Doing this is slightly risky, since the prefix might contain
175 -- unfinished markup (e.g., \foo{bar without the closing }). We might
176 -- need to solve this later.
177 copyind = utf.sub(indentlines[prev], 1, #newind)
178 setwidth = "\\setwidthof{" .. copyind .. "}\\to\\pretlamalignwidth"
179 hskip = "\\hskip\\pretlamalignwidth"
180 return "{" .. setwidth .. hskip .. "}" .. rest
182 -- No previous line? Just return the unmodified line then
187 -- Mark the begin of a block of lambda formatted buffers or expressions. This
188 -- means that, until you call end_of_block again, the subscript bases are
189 -- shared. For example, if you have \lam{y1} some text \lam{yn} within a
190 -- single block, the yn will properly get subscripted. Be sure to call
191 -- end_of_block again!
193 -- Blocks can be partially nested, meaning that the block
194 -- won't be closed until end_of_block was called exactly as often as
195 -- begin_of_block. However, subscripts from the inner block can still
196 -- influence subscripts in the outer block.
197 function vis.begin_of_block()
198 vis.begin_of_display()
199 in_block = in_block + 1
202 -- Ends the current block
203 function vis.end_of_block()
204 in_block = in_block - 1
207 function vis.begin_of_display()
208 if in_block == 0 then
209 -- Initially allow subscripts using _ or just appending a number (later,
210 -- we will add extra patterns here.
211 submatches = {"^(%a*)_([%a%d,]+)$", "^(%a+)([%d,]+)$"}
212 -- This stores all the bases we've encountered so far (to prevent
213 -- duplicates). For each of them there will be a pattern in submatches
222 -- Make things work for inline typeing (e.g., \type{}) as well.
223 vis.begin_of_inline = vis.begin_of_display
224 vis.end_of_inline = vis.end_of_display
226 function vis.flush_line(str,nested)
227 local result, state = { }, 0
228 local finish, change = buffers.finish_state, buffers.change_state
230 -- Set the colorscheme, which is used by finish_state and change_state
231 buffers.currentcolors = colors
235 -- See if the next token is a word
236 word, str = take_word(str)
238 if keywords[res] then
239 -- Make all keywords bold
240 word = "{\\bold " .. word .. "}"
242 -- Process any subscripts in the word
243 word = do_subscripts(word)
246 -- The next token is not a word, it must be a symbol
247 symbol, str = take_symbol(str)
250 -- Append the resulting token
251 result[#result+1] = word or symbol
254 state = finish(state, result)
255 buffers.flush_result(result,nested)
258 -- vim: set sw=4 sts=4 expandtab ai: