path: root/Master/texmf-dist/scripts/texdoc/texdoclib-search.tlu

-- texdoclib-search.tlu: file searching functions for texdoc
--
-- The TeX Live Team, GPLv3, see texdoclib.tlu for details

-- Warning: Some functions here assume that M.init_databases() has been called.

-- dependencies
local kpse = require 'kpse'
local lfs = require 'lfs'
local texdoc = {
    const = require 'texdoclib-const',
    util = require 'texdoclib-util',
    alias = require 'texdoclib-alias',
    score = require 'texdoclib-score',
    config = require 'texdoclib-config',
}

-- shortcuts
local M = {}
local C = texdoc.const
local err_print = texdoc.util.err_print
local dbg_print = texdoc.util.dbg_print

-- shared by all functions in this file
local s_doclist -- the Doclist object to be populated by various functions
local s_meta -- {[normname] = meta, ...} (populated by init_tlp_database)
local vanilla -- is this a vanilla TL or a re-package one without tlpdb?

---------------------------   utility functions   ----------------------------

-- find the TeX Live root
local function get_tlroot()
    local tlroot = kpse.expand_path('$TEXMFROOT') -- it should be exisitng one
    get_tlroot = function() return tlroot end
    return tlroot
end

-- says if file has a known extension according to ext_list
-- (or known basename according to basename_list)
local function check_ext(file)
    file = file:lower()
    -- remove zipext if applicable
    file = texdoc.util.parse_zip(file)
    -- then do the normal thing
    for _, e in ipairs(texdoc.config.get_value('ext_list')) do
        if e == '*' then
            return true
        elseif (e == '') then
            if not file:find('.', 1, true) then
                return true
            end
        else
            local dot_e = '.' .. e
            if file:sub(-#dot_e) == dot_e then
                return true
            end
        end
    end
    -- is the basename good?
    for _, b in ipairs(texdoc.config.get_value('basename_list')) do
        if file:find('^' .. b .. '$') or file:find('^' .. b .. '%.') then
            return true
        end
    end
    return false
end

----------------------   docfile and doclist objects   -----------------------

--[[
doclist = {
    [1] = docfile1, [2] = docfiles2, ...,
    inv = {realpath1 = index1, ...}
}

The inv subtable is such that for all i
doclist.inv(doclist[i].realpath:lower()) == i
Paths are lowercased in order to avoid duplicates on windows.
--]]

local Doclist = {}
Doclist.__index = Doclist

-- create a new list of docfiles
function Doclist:new()
    local dl = {inv = {}}
    setmetatable(dl, self)
    return dl
end

-- add a docfile to a list
function Doclist:add(df)
    -- if no realpath information, unable to add
    -- (useful if vanilla == false)
    if not df.realpath then return end

    -- check the existence of the file
    if not lfs.isfile(df.realpath) then
        dbg_print('search', 'File %s not found.  Skipping.', df.realpath)
        return
    end

    -- add the docfile to the list
    local index = self.inv[df.realpath:lower()]
    if index then
        self[index]:mergein(df)
    else
        dbg_print('search', 'File %s found.', df.realpath)

        local newindex = #self + 1
        self[newindex] = df
        self.inv[df.realpath:lower()] = newindex
    end
end

-- stops a doclist
function Doclist:stop()
    self.inv = nil
end

--[[
docfile = {
    -- name and tree are mandatory
    name      = filename (used for scoring only)
    tree      = code of the tree, see below
    -- at least one of the following fields should exist
    matches   = {pattern1, pattern2, ...} or {}
    runtodoc  = true if there is a runfile -> docfile association
    tlptodoc  = true if there is a tlp name -> docfile association
    -- those are virtual members, see below
    realpath  = full path
    normname  = nomrmalised (path removed up to the 'doc' component)
    basename  = basename
    lang      = language tag from the catalogue metadata
    details   = details tag from the catalogue metadata
    quality   = 'good', 'bad', or 'killed' depending on score
    ext_pos   = position of the extension in ext_list
    -- set for elements of a list as a side effect of sort_doclist()
    score     = score
}
if tree > 1, this is the index of the tree in TEXDOCS
if tree = 0, then name is relative to TLROOT
tree = - 1 if and only if file is a sty file. Here name is absolute.
--]]

-- Docfile objects inherit members from Docfile
-- and have virtual members implemented by get_<member>() methods
-- for best cache performance, getters should not return nil
local Docfile = {}
function Docfile:__index(key)
    if Docfile[key] then return Docfile[key] end
    local getter = Docfile['get_' .. key]
    if getter then
        rawset(self, key, getter(self))
        return rawget(self, key)
    end
end

-- create a new docfile object using initialisation info
-- required fields: name, tree
function Docfile:new(info)
    local df = {}
    setmetatable(df, self)
    for k, v in pairs(info) do
        if k == 'pattern' then
            df.matches = {info.pattern}
        else
            df[k] = v
        end
    end
    return df
end

-- merge a second docfile object in, assuming it represents the same file
function Docfile:mergein(df)
    for k, v in pairs(df) do
        if k == 'matches' then
            for _, m in ipairs(df.matches or {}) do
                table.insert(self.matches, m)
            end
        else
            self[k] = v
        end
    end
end

-- return the full path to the file
local texdocs_tree_to_path  -- definition later
function Docfile:get_realpath()
    if self.tree > 0 then
        return texdocs_tree_to_path(self.tree, self.name)
    elseif self.tree == 0 then
        if vanilla then
            return get_tlroot() .. '/' .. self.name
        else
            return kpse.find_file(self.normname, 'TeX system documentation')
        end
    else
        return self.name
    end
end

-- normalise a name from the tlpdb (use for s_meta indexes)
local function reloc_tlpdb_path(name)
    return name:gsub('^texmf[^/]*/doc/', '')
end

-- return normalised name
function Docfile:get_normname()
    return (self.tree == 0) and reloc_tlpdb_path(self.name) or self.name
end

-- retrieve the lang from meta
function Docfile:get_lang()
    local meta = s_meta[self.normname]
    return meta and (meta.lang or false) or false
end

-- retrieve the details from meta
function Docfile:get_details()
    local meta = s_meta[self.normname]
    return meta and (meta.details or false) or false
end

-- return the base name
function Docfile:get_basename()
    return self.name:gsub('.*/', '')
end

-- for interface consistency, matches should always be a table, never nil
function Docfile:get_matches()
    return {}
end

-- from texdoclib-score.tlu
Docfile.get_quality = texdoc.score.docfile_quality

-- from texdoclib-score.tlu
function Docfile:get_ext_pos()
    return texdoc.score.ext_pos(self.basename)
end

-------------------   select results from TEXDOCS trees   --------------------

-- says if a file (with its path) matches a pattern
local function matches(pattern, file)
    if pattern.original then
        return file:lower():find(pattern.name:lower(), 1, true)
    else
        return texdoc.score.is_exact(file, pattern.name)
    end
end

-- return a docfile object if file "matches", nil otherwise
local function process_file(patlist, file, pathfile, code)
    local docfile
    local pattern
    for _, pattern in ipairs(patlist) do
        if matches(pattern, pathfile) then
            local info = {
                name = pathfile,
                tree = code,
                pattern = pattern,
            }
            if docfile then
                docfile:mergein(Docfile:new(info))
            else
                docfile = Docfile:new(info)
            end
        end
    end
    return docfile
end

-- scan a database
local function scan_db(patlist, code, lsr_db)
    for file, basename in pairs(lsr_db) do
        local df = process_file(patlist, basename, file, code)
        if df then s_doclist:add(df) end
    end
end

---------------------   manage TEXDOCS trees of the kpse   ----------------------

-- build a db from a ls-R file
local function init_lsr_db(root, shift)
    -- open the file
    local lsr = assert(io.open(root .. '/ls-R', 'r'))
    local _ = lsr:read('*line') -- throw away first line (comment)

    -- scan it
    local db = {}
    local maybe_dir, is_doc = true, false
    local current_dir
    local l = #shift
    while true do
        local line = lsr:read('*line')
        while line == '' do line, maybe_dir = lsr:read('*line'), true end
        if line == nil then break end -- EOF
        local dir_line = maybe_dir and line:match('^%./(.*):$')
        if dir_line then
            maybe_dir = false -- next line may not be a dir
            if string.sub(dir_line .. '/', 1, l) == shift then
                is_doc = true
                current_dir = dir_line:sub(l + 1)
                db[current_dir] = nil
            elseif is_doc then
                break -- we're exiting the ./doc (or shift) dir, so it's over
            end
        elseif is_doc then
            local file = line
            if current_dir ~= '' then
                file = current_dir .. '/' .. line
            end
            if check_ext(line) then db[file] = line end
        end
    end

    lsr:close()

    return db
end

-- build a db for a tree without ls-R index
local function init_tree_db(base, recurse)
    local db = {}
    local function init_tree_db_rec(dir)
        for file in lfs.dir(base .. '/' .. dir) do
            if file ~= '.' and file ~= '..' then
                local f = (dir == '') and file or dir .. '/' .. file
                if lfs.isdir(base..'/'..f) then
                    if recurse then init_tree_db_rec(f) end
                else
                    if check_ext(file) then db[f] = file end
                end
            end
        end
    end
    init_tree_db_rec('')
    return db
end

local init_texdocs_database, get_doclist_texdocs

do -- begin scope of doc_roots
local doc_roots

--[[
doc_roots[i] = {
    path    = initial path,
    db      = {[file1] = basename1, [file2] = basename2, ...},
}
--]]

-- populate the doc_roots filename databases
init_texdocs_database = function()
    -- find a ls-R file in a parent directory and return it or nil
    local function lsr_root(path)
        if not lfs.isdir (path) then return end
        local root, shift = path, ''
        if root:sub(-1) == '/' then root = root:sub(1, -2) end
        while root:find('/', 1, true) do
            if lfs.isfile(root .. '/ls-R') then
                return root, shift
            end
            local last_comp = root:match('^.*/(.*)$')
            -- /!\ cannot put last_comp in a regex: can contain special char
            root = root:sub(1, - (#last_comp + 2))
            shift = last_comp .. '/' .. shift
        end
    end

    doc_roots = {}
    local kpse_texdocs = kpse.expand_var('$TEXDOCS')

    -- expand the path and turn it into a lua list
    local raw_doc_roots = kpse.expand_braces(kpse_texdocs):explode(C.kpse_sep)
    local max = #raw_doc_roots + 1

    for j, dir in ipairs(raw_doc_roots) do
        local i = max - j
        local n
        local path, db

        -- get path, !! and // values
        dir, n = dir:gsub('//$', '')
        local recursion_allowed = (n == 1)
        local path, n = dir:gsub('^!!', '')
        local index_mandatory = (n == 1)
        dbg_print('texdocs',
            'texdocs[%d] = %s (index_mandatory=%s, recursion_allowed=%s)',
            i, path, tostring(index_mandatory), tostring(recursion_allowed))

        -- decide if we should use a ls-R index, the filesystem, or do nothing
        local root, shift = lsr_root(path)
        if root and shift and recursion_allowed then
            dbg_print('texdocs',
                'texdocs[%d] using index: %s (shift=%s)', i, root, shift)
            db = init_lsr_db(root, shift)
        elseif not index_mandatory and lfs.isdir(path) then
            dbg_print('texdocs',
                'texdocs[%d] using filesystem search', i)
            db = init_tree_db(path, recursion_allowed)
        end

        -- register this in docroots
        doc_roots[i] = {path=path, db=db}
    end
end

-- find docfiles in texdocs directories
get_doclist_texdocs = function(patlist)
    for code, dr in ipairs(doc_roots) do
        if dr.db then scan_db(patlist, code, dr.db) end
    end
end

-- return the real path from a texdocs tree number + relative path
texdocs_tree_to_path = function (tree, rel)
    return doc_roots[tree].path .. '/' .. rel
end

end -- end scope of doc_roots

---------------------------   look for sty files   ---------------------------

-- add doclist entries for sty files in patlist
local function get_doclist_sty(patlist)
    for _, pat in ipairs(patlist) do
        local file = kpse.find_file(pat.name)
        if file then
            local df = Docfile:new({
                name = file,
                tree = -1,
                pattern = pat,
            })
            s_doclist:add(df)
        end
    end
end

-------------------------------   use tlpdb   --------------------------------

-- tlpdb means TeX Live Package DataBase and tlp means TeX Live Package

-- return true if cache exists and is newer than original, false otherwise
local function good_cache(cache, ori)
    local cache_date = lfs.attributes(cache, 'modification')
    if not cache_date then return false end
    local ori_date = assert(lfs.attributes(ori, 'modification'))
    return cache_date > ori_date
end

-- make sure a given directory exists, or return nil plus an error string
local function mkdir_p(dir)
    if os.type == 'windows' and chgstrcp then
        if lfs.isdir(chgstrcp.syscptoutf8(dir)) then return true end
    else
        if lfs.isdir(dir) then return true end
    end
    local parent = texdoc.util.path_parent(dir)
    if parent then
        local ok, msg = mkdir_p(parent)
        if not ok then return nil, msg end
    end
    if os.type == 'windows' and chgstrcp then
        return lfs.mkdir(chgstrcp.syscptoutf8(dir))
    else
        return lfs.mkdir(dir)
    end
end

local print_out_tlpinfo, get_doclist_tlpdb
local get_tlpinfo_from_tlpdb, get_tlpinfo_from_cache, get_tlpinfo_from_dist

do -- begin scope of tlpinfo tables
local tlp_from_runfile -- {[runfile_basename] = {tlp1 = true, ...}, ...}
local tlp_doclist -- {[tlp_name] = {relname1, relname2, ...}, ...}

-- remove entries for tlp without any docfile
local function remove_useless_tlp()
    for tlp, doclist in pairs(tlp_doclist) do
        if #doclist == 0 then tlp_doclist[tlp] = nil end
    end
    for runfile, tlp_set in pairs(tlp_from_runfile) do
        for tlp in pairs(tlp_set) do
            if not tlp_doclist[tlp] then
                tlp_from_runfile[runfile][tlp] = nil
            end
        end
    end
end

-- populate tlpinfo tables using the given texlive.tlpdb
get_tlpinfo_from_tlpdb = function(filename)
    s_meta, tlp_from_runfile, tlp_doclist = {}, {}, {}
    local curr_tlp
    local state = 'none'
    for line in io.lines(filename) do
        if state == 'none' and line:find('^name ') then
            -- begin a new package
            curr_tlp = line:sub(6, -1):lower()
            tlp_doclist[curr_tlp] = {}
        elseif state == 'docfiles' then
            if not line:find('^ ') then
                state = 'none'
            else
                local file = line:match('^ ([^ ]*)')
                local meta = line:match('^ [^ ]* (.+)')
                local basename = file:match('([^/]*)$')
                if check_ext(basename) then
                    -- we've got a docfile here, add it
                    table.insert(tlp_doclist[curr_tlp], file)
                    if meta then
                        local details = meta:match('details="([^"]+)"')
                        local lang = meta:match('language="([^"]+)"')
                        s_meta[reloc_tlpdb_path(file)] = {
                            details = details,
                            lang = lang,
                        }
                    end
                end
            end
        elseif state == 'runfiles' then
            if not line:find('^ ') then
                state = 'none'
            else
                -- check for interesting runfiles
                local e = line:sub(-4, -1)
                if e == '.tex' or e == '.sty' or e == '.cls' then
                    local f = line:match('.*/(.*)%.')
                    tlp_from_runfile[f] = tlp_from_runfile[f] or {}
                    tlp_from_runfile[f][curr_tlp] = true
                end
            end
        end
        -- update state
        if line:find('^docfiles ') then
            state = 'docfiles'
        elseif line:find('^runfiles ') then
            state = 'runfiles'
        end
    end
    remove_useless_tlp()
end

-- print out data from tlpdb in dofile()-able form
print_out_tlpinfo = function(filename)
    local fh
    if os.type == 'windows' and chgstrcp then
        fh = assert(io.open(chgstrcp.syscptoutf8(filename), 'w'))
    else
        fh = assert(io.open(filename, 'w'))
    end
    local function printf(s, ...) fh:write(string.format(s, ...)) end

    -- s_meta
    printf('local s_meta = {\n')
    for k, v in pairs(s_meta) do
        printf('  [%q] = {', k)
        for i, j in pairs(v) do printf('[%q] = %q, ', i, j) end
        printf('},\n')
    end
    printf('}\n')

    -- tlp_from_runfile
    printf('local tlp_from_runfile = {\n')
    for k, v in pairs(tlp_from_runfile) do
        printf('  [%q] = {', k)
        for f in pairs(v) do printf('[%q]=true,', f) end
        printf('},\n')
    end
    printf('}\n')

    -- tlp_doclist
    printf('local tlp_doclist = {\n')
    for k, v in pairs(tlp_doclist) do
        printf('  [%q] = {\n', k)
        for _, f in ipairs(v) do printf('    %q,\n', f) end
        printf('  },\n')
    end
    printf('}\n')
    printf('return s_meta, tlp_from_runfile, tlp_doclist\n')

    fh:close()
end

-- get pre-hashed tlpdb info from a cache file
get_tlpinfo_from_cache = function(filename)
    s_meta, tlp_from_runfile, tlp_doclist = dofile(filename)
end

-- get pre-hashed tlpdb info from a pseudo-cache file
get_tlpinfo_from_dist = function()
    local f = kpse.find_file(C.data_tlpdb_name, 'texmfscripts')
    if not f then
        err_print('error', 'No texlive.tlpdb nor shipped tlpdb data found.')
        os.exit(C.exit_usage)
    end
    dbg_print('tlpdb', 'Getting data from shipped tlpdb data file ' .. f)
    s_meta, tlp_from_runfile, tlp_doclist = dofile(f)
end

-- get docfiles for pattern using specific tlpdb information
get_doclist_tlpdb = function(pattern)
    -- runfile to tlp to docfile
    if tlp_from_runfile[pattern] then
        for tlp in pairs(tlp_from_runfile[pattern]) do
            for _, file in ipairs(tlp_doclist[tlp]) do
                s_doclist:add(Docfile:new{
                    name = file,
                    tree = 0,
                    runtodoc = true,
                })
            end
        end
    end
    -- tlp name to docfile
    if tlp_doclist[pattern] then
        for _, file in ipairs(tlp_doclist[pattern]) do
            s_doclist:add(Docfile:new{
                name = file,
                tree = 0,
                tlptodoc = true,
            })
        end
    end
end

-- calculating Levenshtein distance by dynamic programming
-- cf. http://lua-users.org/lists/lua-l/2008-01/msg00095.html
function M.levenshtein(s, t)
    local s, t = tostring(s), tostring(t)
    if type(s) == 'string' and type(t) == 'string' then
        local m, n, d = #s, #t, {}
        for i = 0, m do d[i] = {[0] = i} end
        for j = 1, n do d[0][j] = j end
        for i = 1, m do
            for j = 1, n do
                local cost = s:sub(i,i) == t:sub(j,j) and 0 or 1
                d[i][j] = math.min(
                    d[i-1][j] + 1, d[i][j-1] + 1, d[i-1][j-1] + cost
                )
            end
        end
        return d[m][n]
    end
end

-- fuzzy search by using Levenshtein distance
function M.fuzzy_search(pattern)
    local tmp_d
    local min = math.huge
    local result = ''

    for p in pairs(tlp_doclist) do
        tmp_d = M.levenshtein(pattern, p)
        if tmp_d < min then
            min, result = tmp_d, p
        end
    end

    return result, min
end

end -- end scope of tlpinfo table

-- get tlpinfo tables initialised by whatever mean
local function init_tlp_database()
    -- we assume TEXMFVAR always consists of an element
    local cache_file = kpse.expand_var('$TEXMFVAR/' .. C.cache_name)

    -- set vanilla and detect texlive.tlpdb to use
    local texlive_tlpdb = get_tlroot() .. '/tlpkg/texlive.tlpdb'
    vanilla = lfs.isfile(texlive_tlpdb)
    local tlpdb_found = vanilla

    local custom_texlive_tlpdb = texdoc.config.get_value('texlive_tlpdb')
    if custom_texlive_tlpdb then
        if lfs.isfile(custom_texlive_tlpdb) then
            texlive_tlpdb = custom_texlive_tlpdb
            tlpdb_found = true
        else
            err_print('warning',
                'Specified texlive.tlpdb does not exist: ' ..
                texdoc.util.w32_path(custom_texlive_tlpdb))
            err_print('warning',
                'Fallback to use the texlive.tlpdb in the distribution.')
        end
    end

    -- get tlpinfo
    if tlpdb_found then
        if good_cache(cache_file, texlive_tlpdb) then
            dbg_print('tlpdb', 'Using cached data from ' .. cache_file)
            get_tlpinfo_from_cache(cache_file)
        else
            dbg_print('tlpdb',
                'Getting data from tlpdb file ' .. texlive_tlpdb)
            get_tlpinfo_from_tlpdb(texlive_tlpdb)
            dbg_print('tlpdb', 'Writing data in cache file ' .. cache_file)
            local ok, msg = mkdir_p(texdoc.util.path_parent(cache_file))
            if not ok then
                err_print('warning',
                    'Failed to create cache file in %s:', cache_file)
                err_print('warning', msg)
            else
                print_out_tlpinfo(cache_file)
            end
        end
    else
        dbg_print('tlpdb', 'Using shipped tlpdb data.')
        get_tlpinfo_from_dist()
    end
end

-----------------------------   main function   ------------------------------

-- initialise the various databases (must be called first)
function M.init_databases()
    init_texdocs_database()
    init_tlp_database()
end

-- find docfiles according to pattern
function M.get_doclist(pattern, no_alias)
    dbg_print('search', 'Searching documents for pattern "%s"', pattern)

    -- separate sty patterns from the rest
    local function normal_vs_sty(list)
        local normal, sty = {}, {}
        for _, p in ipairs(list) do
            if p.name:lower():match('%.([^/.]*)$') == 'sty' then
                table.insert(sty, p)
            else
                table.insert(normal, p)
            end
        end
        return normal, sty
    end

    local function doc_search(pattern, no_alias)
        -- get patterns (inc. aliases)
        local patterns = texdoc.alias.get_patterns(pattern, no_alias)
        local normal, sty = normal_vs_sty(patterns)

        -- get results; _texdocs search comes after _tlpdb search so that
        -- files found by both will have the priority of the _texdocs tree.
        -- (https://puszcza.gnu.org.ua/bugs/?369)
        get_doclist_sty(sty)
        get_doclist_tlpdb(pattern)
        get_doclist_texdocs(normal)
    end

    -- initialise result list
    s_doclist = Doclist:new()

    -- 1. normal search with the input pattern
    doc_search(pattern, no_alias)

    -- 2. if no result, execute fuzzy search
    local fuzzy_level = texdoc.config.get_value('fuzzy_level')
    if not s_doclist[1] and fuzzy_level > 0 then
        local f_res, f_lev = M.fuzzy_search(pattern)
        if f_lev <= fuzzy_level then
            err_print('info', 'Fuzzy search result: ' .. f_res)
            dbg_print('search', 'Levenshtein distance: ' .. f_lev)
            pattern = f_res
            doc_search(pattern, no_alias)
        else
            dbg_print('search', 'Fuzzy search result: ' .. f_res)
            dbg_print('search', 'Levenshtein distance: ' .. f_lev)
        end
    end

    -- finally, sort results
    texdoc.score.sort_doclist(s_doclist, pattern)
    return s_doclist
end

return M

-- vim: ft=lua: