path: root/rst_helpers.lua

#!/usr/bin/env texlua
--------------------------------------------------------------------------------
--         FILE:  rst_helpers.lua
--        USAGE:  ./rst_helpers.lua 
--  DESCRIPTION:  Complement to the reStructuredText parser
--       AUTHOR:  Philipp Gesang (Phg), <megas.kapaneus@gmail.com>
--      VERSION:  1.0
--      CREATED:  07/09/10 01:03:08 CEST
--------------------------------------------------------------------------------
--

local utf = unicode.utf8
local P, R, S, V, match 
    = lpeg.P, lpeg.R, lpeg.S, lpeg.V, lpeg.match

local C, Carg, Cb, Cc, Cg, Cmt, Cp, Cs, Ct 
    = lpeg.C, lpeg.Carg, lpeg.Cb, lpeg.Cc, lpeg.Cg, lpeg.Cmt, lpeg.Cp, lpeg.Cs, lpeg.Ct

local helpers = {}
helpers.table = {}
helpers.cell  = {}

local patterns  = {}

do
    local p = patterns
    p.dash   = P"-"
    p.equals = P"="
    p.plus   = P"+"
    p.bar    = P"|"
    p.eol    = P"\n"
    p.last   = -P(1)

    p.celldelim   = p.bar + p.plus
    p.cellcontent = (1 - p.celldelim)
    p.cell        = p.celldelim * C((1 - p.celldelim)^1) * #p.celldelim
    p.cell_line   = p.plus * p.dash^1 * #p.plus
    p.dashesonly  = p.dash^1 * p.last

    p.cells = P{
        [1] = "cells",
        cells = p.celldelim 
              * (C(V"in_cell")
               * (V"matchwidth" * C(V"in_cell")) ^1),

        in_cell = p.cellcontent^1
                + (p.dash - p.cellcontent)^1,

        matchwidth = Cmt(C(p.celldelim) * Carg(1), function(s,i,del, layout)
                         --print("~~~~~~~~~~~~~~ NEXT MATCH ~~~~~~~~~~~~~~")
                         local pos = 1
                         local lw = layout.widths
                         for n=1, #lw do
                             pos = pos + lw[n] + 1
                             --print(">",pos,i-1,i-1==pos)
                             if (i - 1) == pos then return true end
                         end
                         return false
                     end),
    }

    p.sep_line = p.plus * (p.dash^1   * p.plus)^1 * p.last
    p.sep_head = p.plus * (p.equals^1 * p.plus)^1 * p.last

    p.sep_part = ((1 - p.cell_line)^0 * p.cell_line) - p.sep_line

    p.new_row = p.sep_line + p.sep_head + p.sep_part

    p.whitespace = S" \t\v\r\n"^1
    p.strip = p.whitespace^0 * C((1 - (p.whitespace * p.last))^1) * p.whitespace^0 * p.last
end

function helpers.cell.create(raw, n_row, n_col, continue)
    local p = patterns
    local cell = {}
    cell.stripped = raw and p.strip:match(raw) or ""
    cell.content  = raw
    cell.width    = raw and utf.len(raw) or 0
    cell.bytes    = raw and #raw or 0
    cell.variant  = "normal" -- [normal|separator|y_continue|x_continue]
    cell.pos      = {}
    cell.pos.x    = n_col
    cell.pos.y    = n_row
    cell.span     = {}
    cell.span.x   = 1
    cell.span.y   = 1
    cell.parent   = nil
    return cell
end

function helpers.cell.get_x_span(content, layout, init)
    --print(#content,">"..content.."<")
    --print(layout.widths[init],">"..layout.slices[init].."<")
    local acc = 0
    local lw = layout.widths
    for n=init, #lw do
        acc = acc + lw[n] + 1
        --print(n, layout.slices[n], acc)
        if utf.len(content) + 1 == acc then 
            return n - init
        end
    end
    return false
end


-- Extending a cell by 1 cell horizontally.
function helpers.cell.add_x (cell)
    cell.span.x = cell.span.x + 1
end


local function set_layout (line)
    local p = patterns
    local layout = {}
    local slice = Ct((p.plus * C(p.dash^1) * #p.plus)^1)

    layout.widths = {}
    layout.slices = {}
    for n, elm in ipairs(slice:match(line)) do
        layout.widths[n] = #elm
        layout.slices[n] =  elm
    end
    return layout
end

function helpers.table.create(raw)
    local self = {}
    self.rows = {}

    local p = patterns

    self.resolve_parent = function(row, col)
        print(row, col)
        local cell = self.rows[row][col]
        local par_row, par_col
        if cell.parent then
            par_row, par_col = self.resolve_parent(cell.parent.y, cell.parent.x)
        else
            return par_row, par_col
        end
    end
    
    self.__init = function()
        local hc = helpers.cell
        self.layout = set_layout(raw[1])
        local rowcount = 0

        -- first get the correct horizontal spans
        local next_is_new_row = true
        for n, line in ipairs(raw) do
            -- caching the test results
            local sepline = p.sep_line:match(line)
            local sephead = p.sep_head:match(line)
            local seppart = p.sep_part:match(line)
            local newrow  = sepline or sephead or seppart

            if newrow and n > 1 then
                if sephead then
                    self.has_head = true
                end
                rowcount = rowcount + 1
            end

            if not (sepline  or
                    sephead) then
                local row = {}
                row.newrow = (newrow or next_is_new_row) and true or false
                next_is_new_row = false
                local sl = self.layout
                local splitted = {p.cells:match(line, 1, sl)}
                local pos_layout, pos_line = 1, 1
                local last = nil
                local ignore = {}
                ignore.n, ignore.parent = 0, nil
                while pos_layout <= #sl.slices do
                    --local splitpos = splitted[pos_line]
                    local splitpos = splitted[pos_layout]
                    local width_layout = sl.widths[pos_layout]
                    local span = 1
                    local this
                    if ignore.n > 0 then
                        ignore.n = ignore.n - 1
                        this = hc.create(false, n, pos_layout, true)
                        row[pos_layout] = this
                        this.parent = ignore.parent
                    else
                        local width_cell   = utf.len(splitpos)
                        if width_cell > width_layout then
                            --print(splitpos, width_cell, sl.slices[pos_layout], width_layout)
                            -- check the horizontal span
                            span = span + hc.get_x_span(splitpos, sl, pos_layout)
                        end
                        pos_line = pos_line + span
                        this = hc.create(splitpos, n, pos_layout, false)
                        if p.dashesonly:match(splitpos) then
                            this.variant = "separator"
                        end
                        this.span.x = span
                        last = this
                        row[pos_layout] = this
                        ignore.n = span - 1
                        ignore.parent = ignore.n > 0 and { y = rowcount, x = pos_layout } or nil
                    end
                    pos_layout = pos_layout + 1
                    --print(">",this.pos.y,this.pos.x,this.span.y,this.span.x)
                    --if this.span.y > 1 then
                        --print(">"..this.content.."<")
                    --end
                end
                self.rows[#self.rows+1] = row
                --print(#self.rows.."<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<")
                --for i,j in ipairs(row) do print(i,j.content) end
                --print(row.newrow)
            else -- separator
                next_is_new_row = true

            end
            --print(string.format("[t-%2s,%2s]> %s", n, m, cell))
        end
        --print("rc = "..rowcount, "nr = "..#self.rows)
        self.rowcount = rowcount

        if self.rowcount < #self.rows then -- some cells span vertically
            local has_partsep = function(row)
                for n, cell in ipairs(row) do
                    if cell.variant == "separator" then
                        return true
                    end
                end
                return false
            end

            local dont_do_next = false
            for nr,row in ipairs(self.rows) do
                local ps = has_partsep(row)

                row.ignore = ps and true or false
                if ps then
                    self.rows[nr+1].newrow = true
                end
                if (not row.newrow or ps) and not dont_do_next then
                    --print(">>>>>>>>>> NEXT <<<<<<<<<<<<")
                    --print(row.newrow and has_partsep(row))
                    -- continues last row either fully or partially
                    for nc, cell in ipairs(row) do
                        --print(nr, nc, cell.content)
                        if cell.content == false then
                            -- empty cell
                        elseif cell.variant ~= "separator" then
                            local par_row, par_col
                            if cell.parent then
                                par_row, par_col = self.resolve_parent(cell.parent.y, cell.parent.x)
                                --print("old parent: ", par_row, par_col)
                            else -- no previous span, setting parent to preceding row
                                par_row, par_col = nr -1, nc
                                --print("new parent: ", par_row, par_col)
                            end

                            print (nr, nc, cell.content, cell.parent, par_row, par_col)
                            local parent = self.rows[par_row][par_col]
                            --parent.span.y   = parent.span.y + 1
                            parent.content  = parent.content .. cell.content
                            parent.stripped = parent.stripped .. " " .. cell.stripped

                            if ps then
                                local successor = self.rows[nr+1][nc]
                                successor.parent = { x = par_col, y = par_row }
                                parent.span.y    = parent.span.y + 1
                                parent.content   = parent.content .. successor.content
                                parent.stripped  = parent.stripped .. " " .. successor.stripped
                                dont_do_next = true
                            else
                                dont_do_next = false
                            end
                            print(cell.content, parent.content)
                            --add_y(cell)
                        end
                    end
                elseif dont_do_next then
                    --row.ignore = true
                    dont_do_next = false
                end
            end
        end -- vertical span handler
    end -- end __init

    self.__init()
    return self
end
                

return helpers