Esempio n. 1
0
  public static String Resolve_file(boolean use_file_protocol, Io_url root_dir, String file) {
    String rv = file;

    // resolve relative urls; EX: "./a.js" -> "/xowa/wiki/simple.wikipedia.org/bin/script/a.js"
    if (String_.Has_at_bgn(rv, "./")) {
      // remove "./"
      rv = String_.Mid(rv, 2);

      if (use_file_protocol) rv = root_dir.To_http_file_str() + rv;
      else {
        // if fsys_url && wnt, replace "\" with "/"
        if (Op_sys.Cur().Tid_is_wnt())
          rv = String_.Replace(rv, Op_sys.Lnx.Fsys_dir_spr_str(), Op_sys.Wnt.Fsys_dir_spr_str());
        rv = root_dir.Xto_api() + Op_sys.Cur().Fsys_dir_spr_str() + rv;
      }
    }
    return rv;
  }
 private int Import_url_chk(
     byte[] rel_url_prefix,
     byte[] src,
     int src_len,
     int old_pos,
     int find_bgn,
     byte[] url_raw,
     Bry_bfr bfr) {
   if (find_bgn < Bry_import_len) return Bry_find_.Not_found;
   if (!Bry_.Match(src, find_bgn - Bry_import_len, find_bgn, Bry_import))
     return Bry_find_.Not_found;
   byte[] css_url = url_raw;
   int css_url_len = css_url.length;
   if (css_url_len > 0
       && css_url[0]
           == Byte_ascii
               .Slash) { // css_url starts with "/"; EX: "/page" or "//site/page" DATE:2014-02-03
     if (css_url_len > 1
         && css_url[1] != Byte_ascii.Slash) // skip if css_url starts with "//"; EX: "//site/page"
     css_url = Bry_.Add(rel_url_prefix, css_url); // "/w/a.css" -> "//en.wikipedia.org/w/a.css"
   }
   css_url =
       Bry_.Replace(
           css_url,
           Byte_ascii.Space,
           Byte_ascii
               .Underline); // NOTE: must replace spaces with underlines else download will fail;
                            // EX:https://it.wikivoyage.org/w/index.php?title=MediaWiki:Container e
                            // Infobox.css&action=raw&ctype=text/css; DATE:2015-03-08
   byte[] css_src_bry = Import_url_build(stylesheet_prefix, rel_url_prefix, css_url);
   String css_src_str = String_.new_u8(css_src_bry);
   download_wkr.Download_xrg()
       .Prog_fmt_hdr_(
           usr_dlg.Log_many(
               GRP_KEY, "logo.download", "downloading import for '~{0}'", css_src_str));
   byte[] css_trg_bry = download_wkr.Download_xrg().Exec_as_bry(css_src_str);
   if (css_trg_bry == null) {
     usr_dlg.Warn_many("", "", "could not import css: url=~{0}", css_src_str);
     return Bry_find_.Not_found; // css not found
   }
   bfr.Add_mid(src, old_pos, find_bgn - Bry_import_len).Add_byte_nl();
   bfr.Add(Bry_comment_bgn).Add(css_url).Add(Bry_comment_end).Add_byte_nl();
   if (Bry_find_.Find_fwd(css_url, Wikisource_dynimg_ttl) != -1)
     css_trg_bry =
         Bry_.Replace(
             css_trg_bry,
             Wikisource_dynimg_find,
             Wikisource_dynimg_repl); // FreedImg hack;
                                      // PAGE:en.s:Page:Notes_on_Osteology_of_Baptanodon._With_a_Description_of_a_New_Species.pdf/3 DATE:2014-09-06
   bfr.Add(css_trg_bry).Add_byte_nl();
   bfr.Add_byte_nl();
   int semic_pos = Bry_find_.Find_fwd(src, Byte_ascii.Semic, find_bgn + url_raw.length, src_len);
   return semic_pos + Int_.Const_dlm_len;
 }
 public static Fsdb_db_mgr new_src_bin_db_mgr(Xow_wiki wiki, String version) {
   String domain_str = wiki.Domain_str();
   Fsdb_db_mgr rv = null;
   Io_url url = null;
   if (String_.Eq(version, "v1")) {
     url =
         wiki.Fsys_mgr()
             .File_dir()
             .OwnerDir()
             .GenSubDir(domain_str + "-prv"); // v1: EX: /xowa/file/en.wikipedia.org-prv/
     rv = new Fsdb_db_mgr__v1(url);
   } else if (String_.Eq(version, "v2")) {
     url = wiki.Fsys_mgr().Root_dir().GenSubDir("prv"); // v2: EX: /xowa/wiki/en.wikipedia.org/prv/
     rv = Fsdb_db_mgr_.new_detect(wiki, url, url); // note that v2 is prioritized over v1
   } else throw Err_.new_wo_type("fsdb.make:unknown fsdb_type", "version", version);
   if (rv == null)
     throw Err_.new_wo_type(
         "fsdb.make:source fsdb not found", "version", version, "url", url.Raw());
   return rv;
 }
 private void Download_exec(Xodb_tbl_oimg_xfer_itm fsdb) {
   Io_stream_rdr src_rdr = src_bin_mgr.Find_as_rdr(Xof_exec_tid.Tid_wiki_page, fsdb);
   try {
     if (src_rdr == Io_stream_rdr_.Noop) { // download failed
       ++exec_fail;
       usr_dlg.Warn_many(
           "",
           "",
           "failed: ttl=~{0}",
           String_.Format("[[File:{0}|{1}px]]", fsdb.Orig_ttl(), fsdb.Html_w()));
       Print_progress(fsdb);
     } else { // download passed
       long src_rdr_len = src_rdr.Len();
       int lnki_tier_id = fsdb.Lnki_tier_id();
       if (src_rdr_len > download_size_max && !Int_.In(lnki_tier_id, download_keep_tier_ids)) {
         usr_dlg.Warn_many(
             "",
             "",
             "skipped; ttl=~{0} w=~{1} size=~{2} tier=~{3}",
             fsdb.Orig_ttl(),
             fsdb.Lnki_w(),
             src_rdr_len,
             lnki_tier_id);
         return;
       }
       if (trg_bin_fil == null) // no trg_bin_fil
       Make_trg_bin_file(Bool_.Y, fsdb, src_rdr_len);
       else if (trg_bin_fil.Bin_len() + src_rdr_len
           > trg_bin_db_max) // or trg_bin_fil is out of space
       Make_trg_bin_file(Bool_.N, fsdb, src_rdr_len);
       else if (prv_lnki_tier_id != lnki_tier_id) { // or tier has changed
         if (prv_lnki_tier_id != -1 && !bin_db_mgr.Schema_is_1()) // do not increment dbs for v1
         Make_trg_bin_file(Bool_.Y, fsdb, src_rdr_len);
         prv_lnki_tier_id = lnki_tier_id;
       }
       trg_bin_updater.Save_bin(trg_mnt_itm, trg_atr_fil, trg_bin_fil, fsdb, src_rdr, src_rdr_len);
     }
   } finally {
     src_rdr.Rls();
   }
 }
 public byte[] Convert_to_local_urls(byte[] rel_url_prefix, byte[] src, List_adp list) {
   try {
     int src_len = src.length;
     int prv_pos = 0;
     Bry_bfr bfr = Bry_bfr_.New_w_size(src_len);
     Hash_adp img_hash = Hash_adp_bry.cs();
     while (true) {
       int url_pos = Bry_find_.Find_fwd(src, Bry_url, prv_pos);
       if (url_pos == Bry_find_.Not_found) {
         bfr.Add_mid(src, prv_pos, src_len);
         break;
       } // no more "url("; exit;
       int bgn_pos = url_pos + Bry_url_len; // set bgn_pos after "url("
       byte bgn_byte = src[bgn_pos];
       byte end_byte = Byte_ascii.Null;
       boolean quoted = true;
       switch (bgn_byte) { // find end_byte
         case Byte_ascii.Quote:
         case Byte_ascii.Apos: // quoted; end_byte is ' or "
           end_byte = bgn_byte;
           ++bgn_pos;
           break;
         default: // not quoted; end byte is ")"
           end_byte = Byte_ascii.Paren_end;
           quoted = false;
           break;
       }
       int end_pos = Bry_find_.Find_fwd(src, end_byte, bgn_pos, src_len);
       if (end_pos
           == Bry_find_.Not_found) { // unclosed "url("; exit since nothing else will be found
         usr_dlg.Warn_many(
             GRP_KEY,
             "parse.invalid_url.end_missing",
             "could not find end_sequence for 'url(': bgn='~{0}' end='~{1}'",
             prv_pos,
             String_.new_u8__by_len(src, prv_pos, prv_pos + 25));
         bfr.Add_mid(src, prv_pos, src_len);
         break;
       }
       if (end_pos - bgn_pos == 0) { // empty; "url()"; ignore
         usr_dlg.Warn_many(
             GRP_KEY,
             "parse.invalid_url.empty",
             "'url(' is empty: bgn='~{0}' end='~{1}'",
             prv_pos,
             String_.new_u8__by_len(src, prv_pos, prv_pos + 25));
         bfr.Add_mid(src, prv_pos, bgn_pos);
         prv_pos = bgn_pos;
         continue;
       }
       byte[] img_raw = Bry_.Mid(src, bgn_pos, end_pos);
       int img_raw_len = img_raw.length;
       if (Bry_.Has_at_bgn(img_raw, Bry_data_image, 0, img_raw_len)) { // base64
         bfr.Add_mid(src, prv_pos, end_pos); // nothing to download; just add entire String
         prv_pos = end_pos;
         continue;
       }
       int import_url_end =
           Import_url_chk(
               rel_url_prefix,
               src,
               src_len,
               prv_pos,
               url_pos,
               img_raw,
               bfr); // check for embedded stylesheets via @import tag
       if (import_url_end != Bry_find_.Not_found) {
         prv_pos = import_url_end;
         continue;
       }
       byte[] img_cleaned = Xob_url_fixer.Fix(wiki_domain, img_raw, img_raw_len);
       if (img_cleaned == null) { // could not clean img
         usr_dlg.Warn_many(
             GRP_KEY,
             "parse.invalid_url.clean_failed",
             "could not extract valid http src: bgn='~{0}' end='~{1}'",
             prv_pos,
             String_.new_u8(img_raw));
         bfr.Add_mid(src, prv_pos, bgn_pos);
         prv_pos = bgn_pos;
         continue;
       }
       if (!img_hash.Has(img_cleaned)) { // only add unique items for download;
         img_hash.Add_as_key_and_val(img_cleaned);
         list.Add(String_.new_u8(img_cleaned));
       }
       img_cleaned =
           Replace_invalid_chars(
               Bry_.Copy(
                   img_cleaned)); // NOTE: must call ByteAry.Copy else img_cleaned will change
                                  // *inside* hash
       bfr.Add_mid(src, prv_pos, bgn_pos);
       if (!quoted) bfr.Add_byte(Byte_ascii.Quote);
       bfr.Add(img_cleaned);
       if (!quoted) bfr.Add_byte(Byte_ascii.Quote);
       prv_pos = end_pos;
     }
     return bfr.To_bry_and_clear();
   } catch (Exception e) {
     usr_dlg.Warn_many(
         "",
         "",
         "failed to convert local_urls: ~{0} ~{1}",
         String_.new_u8(rel_url_prefix),
         Err_.Message_gplx_full(e));
     return src;
   }
 }
 String Replace_invalid_chars_str(String raw_str) {
   return String_.new_u8(Replace_invalid_chars(Bry_.new_u8(raw_str)));
 }
 @Override
 public void Cmd_bgn(Xob_bldr bldr) {
   wiki.Init_assert();
   this.poll_interval = poll_mgr.Poll_interval();
   this.bin_db_mgr = new Xob_bin_db_mgr(ns_ids);
   // src_bin_mgr
   if (src_bin_mgr__fsdb_version != null) {
     this.src_fsdb_wkr = Xof_bin_wkr__fsdb_sql.new_(wiki.File__mnt_mgr());
     src_bin_mgr.Wkrs__add(src_fsdb_wkr);
     src_fsdb_wkr.Mnt_mgr().Ctor_by_load(new_src_bin_db_mgr(wiki, src_bin_mgr__fsdb_version));
     src_fsdb_wkr.Mnt_mgr()
         .Mnts__get_main()
         .Txn_bgn(); // NOTE: txn on atr speeds up from 50 -> 300; DATE:2015-03-21
     if (src_bin_mgr__fsdb_skip_wkrs != null) {
       src_fsdb_wkr.Skip_mgr_init(
           src_fsdb_wkr.Mnt_mgr().Mnts__get_main().Cfg_mgr(), src_bin_mgr__fsdb_skip_wkrs);
     }
     if (src_bin_mgr__cache_enabled) {
       usr_dlg.Prog_many("", "", "src_bin_mgr.cache.bgn");
       src_fsdb_wkr.Mnt_mgr().Mnts__get_main().Atr_mgr().Db__core().Fil_cache_enabled_y_();
       usr_dlg.Prog_many("", "", "src_bin_mgr.cache.end");
     }
   }
   if (src_bin_mgr__wmf_enabled) {
     Xof_bin_wkr__http_wmf wmf_wkr = Xof_bin_wkr__http_wmf.new_(wiki);
     src_bin_mgr.Wkrs__add(wmf_wkr);
     wmf_wkr.Fail_timeout_(
         0); // 1000; NOTE: set Fail_timeout here; DATE:2014-06-21; NOTE: do not put in ctor, or
     // else will be 1st wkr; DATE:2014-06-28
   }
   // trg_mnt_itm
   this.trg_bin_db_max = app.Api_root().Bldr().Wiki().Import().File_db_max();
   Io_url trg_file_dir_v1 =
       String_.Eq(trg_bin_mgr__fsdb_version, "v1")
           ? wiki.Fsys_mgr().File_dir().GenNewNameOnly(wiki.Domain_str() + "-prv")
           : wiki.Fsys_mgr().File_dir(); // NOTE: convoluted way of setting trg to -prv if
   // trg_bin_mgr__fsdb_version_v1 is set; otherwise set to
   // "en.wikipedia.org" which will noop; DATE:2015-12-02
   Fsdb_db_mgr trg_db_mgr =
       Fsdb_db_mgr_.new_detect(wiki, wiki.Fsys_mgr().Root_dir(), trg_file_dir_v1);
   if (trg_db_mgr == null) trg_db_mgr = Fsdb_db_mgr__v2_bldr.Get_or_make(wiki, Bool_.Y);
   Fsm_mnt_mgr trg_mnt_mgr = new Fsm_mnt_mgr();
   trg_mnt_mgr.Ctor_by_load(trg_db_mgr);
   trg_mnt_mgr.Mnts__get_insert_idx_(
       Fsm_mnt_mgr
           .Mnt_idx_main); // NOTE: do not delete; mnt_mgr default to Mnt_idx_user; DATE:2014-04-25
   this.trg_mnt_itm = trg_mnt_mgr.Mnts__get_insert();
   Fsm_mnt_mgr.Patch(
       trg_mnt_itm.Cfg_mgr()
           .Tbl()); // NOTE: always patch again; fsdb_make may be run separately without lnki_temp;
   // DATE:2014-04-26
   this.trg_atr_fil = trg_mnt_itm.Atr_mgr().Db__core();
   this.trg_cfg_mgr = trg_mnt_itm.Cfg_mgr();
   bin_db_mgr.Init_by_mnt_mgr(trg_mnt_mgr);
   trg_atr_fil.Conn().Txn_bgn("bldr__fsdb_make__trg_atr_fil");
   if (!trg_atr_fil.Conn()
       .Eq(trg_cfg_mgr.Tbl().Conn())) // need to create txn for v1; DATE:2015-07-04
   trg_cfg_mgr.Tbl().Conn().Txn_bgn("bldr__fsdb_make__trg_cfg_fil");
   // bldr_db
   Xob_db_file bldr_db = Xob_db_file.New__file_make(wiki.Fsys_mgr().Root_dir());
   this.bldr_conn = bldr_db.Conn();
   this.bldr_cfg_tbl = bldr_db.Tbl__cfg(); // NOTE: cfg and atr is in same db; use it
   bldr_cfg_tbl.Conn().Txn_bgn("bldr__fsdb_make__bldr_cfg_tbl");
 }