" cached_file_contents.vim " @License: GPL (see http://www.gnu.org/licenses/gpl.txt) " @Last Change: 2010-01-03. " @Revision: 0.3.0 "exec vam#DefineAndBind('s:c','g:cache_dir_options','{}') if !exists('g:cache_dir_options') | let g:cache_dir_options = {} | endif | let s:c = g:cache_dir_options let s:c['cache_dir'] = get(s:c, 'cache_dir', expand('$HOME').'/.vim-cache') let s:c['scanned_files'] = get(s:c, 'scanned_files', {}) let s:scanned_files = s:c['scanned_files'] let s:define_cache_file = "let this_dir = s:c['cache_dir'].'/cached-file-contents' | let cache_file = expand(this_dir.'/'.substitute(string([func_as_string, a:file]),'[[\\]{}:/\\,''\"# ]\\+','_','g'))" " read a file, run function to extract contents and cache the result returned " by that function in memory. Optionally the result can be cached on disk as " because VimL can be slow! " " file : the file to be read " func: { 'func': function which will be called by funcref#Call " , 'version' : if this version changes cache will be invalidate automatically " , 'ftime_check': optional, default 1. if set to 0 cache isn't updated when file changes and file is in cache " } " " default: what to return if file doesn't exist " think twice about adding lines. This function is called many times. function! cached_file_contents#CachedFileContents(file, func, ...) abort let ignore_ftime = a:0 > 0 ? a:1 : 0 " using string for default so that is evaluated when needed only let use_file_cache = get(a:func, 'use_file_cache', 0) " simple kind of normalization. necessary when using file caching " this seems to be slower: " let file = fnamemodify(a:file, ':p') " simple kind of normalization. necessary when using file caching " / = assume its an absolute path " let file = a:file[0] == '/' ? a:file : expand(a:file, ':p') let file = a:file[0] == '/' ? a:file : fnamemodify(a:file, ':p') " simple kind of normalization. necessary when using file caching let func_as_string = string(a:func['func']) if (!has_key(s:scanned_files, func_as_string)) let s:scanned_files[func_as_string] = {} endif let dict = s:scanned_files[func_as_string] if use_file_cache && !has_key(dict, a:file) exec s:define_cache_file if filereadable(cache_file) let dict[file] = eval(readfile(cache_file,'b')[0]) endif endif if has_key(dict, a:file) let d = dict[a:file] if use_file_cache \ && (ignore_ftime || getftime(a:file) <= d['ftime']) \ && d['version'] == a:func['version'] return dict[a:file]['scan_result'] endif endif let scan_result = funcref#Call(a:func['func'], [a:file] ) let dict[a:file] = {"ftime": getftime(a:file), 'version': a:func['version'], "scan_result": scan_result } if use_file_cache if !exists('cache_file') | exec s:define_cache_file | endif if !isdirectory(this_dir) | call mkdir(this_dir,'p',0700) | endif call writefile([string(dict[a:file])], cache_file) endif return scan_result endfunction fun! cached_file_contents#ClearScanCache() let s:c['scanned_files'] = {} " Don't run rm -fr. Ask user to run it. It cache_dir may have been set to " $HOME ! (should nevere be the case but who knows echoe "run manually in your shell: rm -fr ".shellescape(s:c['cache_dir'])."/*" endf fun! cached_file_contents#Test() " usually you use a global option so that the function can be reused let my_interpreting_func = {'func' : funcref#Function('return len(readfile(ARGS[0]))'), 'version': 2, 'use_file_cache':1} let my_interpreting_func2 = {'func' : funcref#Function('return ARGS[0]') , 'version': 2, 'use_file_cache':1} let tmp = tempname() call writefile(['some text','2nd line'], tmp) let r = [ cached_file_contents#CachedFileContents(tmp, my_interpreting_func) \ , cached_file_contents#CachedFileContents(tmp, my_interpreting_func2) ] if r != [2, tmp] throw "test failed 1, got ".string(r) endif unlet r sleep 3 " now let's change contents call writefile(['some text','2nd line','3rd line'], tmp) let r = cached_file_contents#CachedFileContents(tmp, my_interpreting_func) if 3 != r throw "test failed 2, got ".string(r) endif echo "test passed" endf