diff options
Diffstat (limited to 'Master/texmf/scripts/texdoc/search.tlu')
| -rw-r--r-- | Master/texmf/scripts/texdoc/search.tlu | 364 |
1 files changed, 238 insertions, 126 deletions
diff --git a/Master/texmf/scripts/texdoc/search.tlu b/Master/texmf/scripts/texdoc/search.tlu index fcb044d5189..fa35cce2754 100644 --- a/Master/texmf/scripts/texdoc/search.tlu +++ b/Master/texmf/scripts/texdoc/search.tlu @@ -9,25 +9,115 @@ local L = {} load_env(L, { 'export_symbols', 'os', 'string', 'table', 'lfs', 'kpse', 'io', - 'ipairs', 'assert', 'tonumber', 'type', 'print', 'tostring', - 'err_print', 'win32_hook', 'parse_zip', - 'config', 'alias', 'C', + 'ipairs', 'assert', 'error', 'tostring', 'setmetatable', + 'deb_print', 'err_print', 'win32_hook', 'parse_zip', + 'get_patterns', 'sort_doclist', 'docfile_quality', + 'config', 'C', }) ----------------------------- the docfiles list ----------------------------- +-- shared by all functions in this file +local s_doclist --- shared by all functions below -local s_doc_files +----------------------- docfile and doclist objects ------------------------ --- structure of the s_docfiles variable --- s_docfiles = { --- [1] = docfile1, docfiles2, ..., +-- doclist = { +-- [1] = docfile1, [2] = docfiles2, ..., +-- inv = { realpath1 = index1, ... } -- } +-- +-- The inv subtable is such that for all i +-- doclist.inv(doclist[i].realpath) == i + +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) + local index = self.inv[df.realpath] + if index then + self[index]:mergein(df) + else + local newindex = #self + 1 + self[newindex] = df + self.inv[df.realpath] = newindex + end +end + +-- stops a doclist +function Doclist:stop() + self.inv = nil +end + -- docfile = { --- name = filename relative to tree, absolute if tree == nil, --- tree = number of the tree in doc_roots, --- exact = <boolean> does pattern match exactly, +-- name = filename (used for scoring only) +-- tree = code of the tree, see below +-- matches = {pattern1, pattern2, ...} -- } +-- if tree > 1, this is the index of the tree in TEXDOCS +-- if tree = 0, then name is relative to TLROOT (file found from tlpdb only) +-- tree = - 1 if and only if file is a sty file. Here name is absolute. + +local Docfile = {} +Docfile.__index = Docfile + +-- create a new docfile objet using initilisation info +-- fields : name (relative to tree), tree, pattern +function Docfile:new(info) + df = {} + setmetatable(df, self) + -- get realpath, tree and prepare name + df.tree = info.tree + if info.tree > 0 then + df.realpath = texdocs_tree_to_path(info.tree, info.name) + elseif info.tree == 0 then + error('Unimplemented') + df.realpath = get_tlroot()..'/'..info.name + info.name = string.gsub(info.name, '^texmf(-dist)?/doc/', '', 1) + elseif info.tree == -1 then + df.realpath = info.name + else + error('Internal error: bad tree number') + end + -- remove first component of name if at least two directory levels + if info.tree > -1 then + local name = string.match(info.name, '^..-/(.+/.+)$') + if name then + df.name = '/'..name + else + df.name = info.name + end + else + df.name = info.name + end + -- initialise the list of matches + if info.pattern then + df.matches = { info.pattern } + else + df.matches = {} + end + return df +end + +-- merge a second docfile objet into self +function Docfile:mergein(df) + if df.tree > self.tree then + self.name = df.name + self.tree = df.tree + end + for _, m in ipairs(df.matches) do + table.insert(self.matches, m) + end +end + +-- from score.tlu +Docfile.quality = docfile_quality ------------------ get results from TEXDOCS (à la kpse) ------------------- @@ -49,71 +139,57 @@ function get_texdocs () local kpse_texdocs = kpse.expand_var("$TEXDOCS") -- expand the path and turn it into a lua list local raw_doc_roots = string.explode(kpse.expand_braces(kpse_texdocs), sep) - err_print('Search paths:', 'debug3') - for i, dir in ipairs(raw_doc_roots) do - doc_roots[i] = {} + local max = #raw_doc_roots + 1 + for j, dir in ipairs(raw_doc_roots) do + local i = max - j + local dr = {} local n + -- get path, !! and // values dir, n = string.gsub (dir, '//$', '') - doc_roots[i].recursion_allowed = (n == 1) - doc_roots[i].path, n = string.gsub (dir, '^!!', '') - doc_roots[i].index_mandatory = (n == 1) - err_print(string.format('%s (index_mandatory=%s, recursion_allowed=%s)', - doc_roots[i].path, - tostring(doc_roots[i].index_mandatory), - tostring(doc_roots[i].recursion_allowed)), - 'debug3') + dr.recursion_allowed = (n == 1) + dr.path, n = string.gsub (dir, '^!!', '') + dr.index_mandatory = (n == 1) + deb_print('texdocs', string.format( + 'texdocs[%d] = %s (index_mandatory=%s, recursion_allowed=%s)', + i, dr.path, + tostring(dr.index_mandatory), + tostring(dr.recursion_allowed))) + -- decide if we should use a ls-R index, the filesystem, or do nothing + local root, shift = lsr_root(dr.path) + if root and shift and dr.recursion_allowed then + dr.lsr = root + dr.lsr_shift = shift + deb_print('texdocs', string.format( + 'texdocs[%d] using index: %s (shift=%s)', i, root, shift)) + elseif not dr.index_mandatory and lfs.isdir(dr.path) then + dr.searchfs = true + deb_print('texdocs', string.format( + 'texdocs[%d] using filesystem search', i)) + end + -- register this in docroots + doc_roots[i] = dr end end --- return the real path of a docfile -function real_path(docfile) - if docfile.tree == nil then return docfile.name end +-- return the real path from a texdocs tree number + relative path +function texdocs_tree_to_path(tree, rel) if doc_roots == nil then get_texdocs() end - return win32_hook(doc_roots[docfile.tree].path..'/'..docfile.name) + return win32_hook(doc_roots[tree].path..'/'..rel) end -- find docfiles in texdocs directories -function get_docfiles_texdocs (pattern) - s_docfiles = {} +function get_doclist_texdocs(patlist) if doc_roots == nil then get_texdocs() end - for code, doc_root in ipairs (doc_roots) do - root, shift = lsr_root (doc_root.path) - if root and shift and doc_root.recursion_allowed then - err_print("Looking in tree '"..doc_root.path - .."' using ls-R file'" ..root.."/ls-R'.", 'debug4') - scan_lsr(root, code, shift, pattern) - elseif (not doc_root.index_mandatory) - and lfs.isdir(doc_root.path) then - err_print("Looking in tree '"..doc_root.path - .."' using filesystem.", 'debug4') - scan_tree(code, doc_root.path, '', - pattern, doc_root.recursion_allowed) + for code, dr in ipairs(doc_roots) do + if dr.lsr then + scan_lsr(patlist, code, dr.lsr, dr.lsr_shift) + elseif dr.searchfs then + scan_tree(patlist, code, dr.path, '', dr.recursion_allowed) end end - return s_docfiles -end - --- merge two components of a path, taking care of empty components -function merge_path (a, b) - return ((a == '') or (b == '')) and a..b or a..'/'..b end --- scan a tree without ls-R file -function scan_tree (code, base, cwd, pattern, recurse) - err_print("Entering directory: "..cwd, 'debug4') - for file in lfs.dir(base..'/'..cwd) do - if file ~= '.' and file ~= '..' then - local f = (cwd == '') and file or cwd..'/'..file - if lfs.isdir(base..'/'..f) then - if recurse then scan_tree(code, base, f, pattern, recurse) end - else - local df = process_file(file, f, code, pattern, true) - if df then table.insert(s_docfiles, df) end - end - end - end - err_print("Leaving directory: "..cwd, 'debug4') -end +end -- scope of doc_roots -- find a ls-R file in a parent directory an return it or nil function lsr_root (path) @@ -131,10 +207,27 @@ function lsr_root (path) end end +-- scan a tree without ls-R file +function scan_tree (patlist, code, base, cwd, recurse) + deb_print('filesea', "Entering directory: "..cwd) + for file in lfs.dir(base..'/'..cwd) do + if file ~= '.' and file ~= '..' then + local f = (cwd == '') and file or cwd..'/'..file + if lfs.isdir(base..'/'..f) then + if recurse then scan_tree(patlist, code, base, f, recurse) end + else + local df = process_file(patlist, file, f, code, true) + if df then s_doclist:add(df) end + end + end + end + deb_print('filesea', "Leaving directory: "..cwd) +end + -- scan a ls-R file -function scan_lsr (cwd, code, shift, pattern) +function scan_lsr(patlist, code, cwd, shift) local is_dir = {} -- is_dir[path] = true iff path is a dir - local results = {} + local results = Doclist:new() local isdoc = false local current_dir local l = #shift @@ -152,106 +245,125 @@ function scan_lsr (cwd, code, shift, pattern) isdoc = true current_dir = string.sub (dir_line, l+1) is_dir[current_dir] = true - err_print('Scanning directory: '..current_dir, 'debug4') + deb_print('lsrsea', 'Scanning directory: '..current_dir) elseif isdoc then - err_print("Finished scanning: "..shift, 'debug4') + deb_print('lsrsea', "Finished scanning: "..shift) break -- we're exiting the ./doc (or shift) dir, so it's over end elseif isdoc then - local df = process_file( - line, merge_path(current_dir, line), code, pattern) - if df then table.insert(results, df) end + local df = process_file(patlist, line, current_dir..'/'..line, code) + if df then results:add(df) end end end lsr:close() -- add non-directories to the list for _, df in ipairs(results) do if not is_dir[df.name] then - table.insert(s_docfiles, df) + s_doclist:add(df) end end end -end -- scope of doc_roots - ------------------------------- select results ------------------------------ - -- says if file has a 'good' extenstion according to ext_list -function check_ext(file, pattern) - local good_ext, exact_match = false, false +function check_ext(file) + file = string.lower(file) -- remove zipext if applicable file = parse_zip(file) -- then do the normal thing - local l, pat = string.len(pattern) + 1, pattern..'.' for _, e in ipairs(config.ext_list) do if e == '*' then - good_ext = true - if string.sub(file, 1, l) == pat then exact_match = true end + return true elseif (e == '') then - if not string.find(file, '.', 1, true) then good_ext = true end - if file == pattern then exact_match = true end + if not string.find(file, '.', 1, true) then + return true + end else - if string.sub(file, -string.len(e)) == e then good_ext = true end - if file == pattern..'.'..e then exact_match = true end + local dot_e = '.'..e + if string.sub(file, -string.len(dot_e)) == dot_e then + return true + end end end - return good_ext, exact_match + return false end --- return a docfile entry if it "matches", nil ortherwise -function process_file (file, pathfile, code, pattern) - err_print('Processing file: '..pathfile, 'debug5') - file = string.lower(file) - local base, ext = string.match(file, '^(.*)%.(.*)$') - if string.find(string.lower(pathfile), pattern, - 1, config.regex ~= 'regex') then - local good_ext, exact_match = check_ext(file, pattern) - err_print(string.format("File '%s' matches; good_ext=%s, exact=%s", - pathfile, tostring(good_ext), tostring(exact_match)), 'debug5') - if good_ext then - return { - name = pathfile, - tree = code, - exact = exact_match, - } +-- return a docfile object if file "matches", nil ortherwise +function process_file(patlist, file, pathfile, code) + deb_print('kpse', 'Processing file: '..pathfile) + local docfile + local pattern + for _, pattern in ipairs(patlist) do + if string.find(string.lower(pathfile), string.lower(pattern.name), + 1, config.mode ~= 'regex') then + local good_ext = check_ext(file) + deb_print('kpse', string.format( + "File '%s' matches '%s'; good_ext=%s", + pathfile, pattern.name, tostring(good_ext))) + if good_ext 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 end - return nil + return docfile end ------------------------- get results from elsewhere ------------------------ +---------------------------- look for sty files ---------------------------- --- for sty files, we obviously don't want to look in TEXDOCS... -function get_docfiles_sty (styname) - return {{ - name = kpse.find_file(styname) , - exact = true, - tree = nil, - }} +-- add doclist entries for sty files in patlist +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 ------------------------------ main function ------------------------------- -- find docfiles according to pattern -function get_docfiles(pattern) - local no_regex = (config.mode ~= 'regex') - -- apply aliases if relevant - if no_regex and config.alias_switch and alias[pattern] then - err_print (pattern.." aliased to "..alias[pattern], 'info') - pattern = alias[pattern] - end - -- search using the appropriate function - if string.match(string.lower(pattern), '%.([^/.]*)$') == 'sty' - and no_regex then - return get_docfiles_sty(pattern) - else - pattern = string.lower(pattern) - return get_docfiles_texdocs(pattern) +function get_doclist(pattern) + -- get patterns (inc. aliases) + local normal, sty = normal_vs_sty(get_patterns(pattern)) + -- initialise result list + s_doclist = Doclist:new() + -- get results + get_doclist_sty(sty) + get_doclist_texdocs(normal) + -- finally, sort results + sort_doclist(s_doclist, pattern) + return s_doclist +end + +-- separate sty patterns from the rest +function normal_vs_sty(list) + if config.mode == 'regex' then return list, {} end + local normal, sty = {}, {} + for _, p in ipairs(list) do + if string.match(string.lower(p.name), '%.([^/.]*)$') == 'sty' then + table.insert(sty, p) + else + table.insert(normal, p) + end end + return normal, sty end -- finally export a few symbols export_symbols(L, { - 'real_path', - 'get_docfiles', + 'get_doclist', }) |