44005f1 · typeable/hexml

git.iko.soy/typeable/hexml

on master help sign in

Switch to a parse table for looking up flags, very slight performance improvement

44005f1Neil Mitchell10 years ago

hexml.c+35−13

compare with master

unifiedsplit

@@ -225,6 +225,35 @@ node* node_nextChildBy(document* d, node* parent, node* prev, char* s, int slen)

225}

226

227

228+/////////////////////////////////////////////////////////////////////

229+// PARSE TABLE

230+

231+const char tag_name1 = 0x1;

232+const char tag_name = 0x2;

233+const char tag_space = 0x4;

234+

235+char parse_table[256];

236+

237+void init_parse_table()

238+{

239+ static bool done = 0;

240+ if (done) return;

241+ for (int i = 0; i < 256; i++)

242+ {

243+ bool name1 = i == ':' || i == '_' || (i >= 'a' && i <= 'z') || (i >= 'A' && i <= 'Z');

244+ bool name = name1 || i == '-' || (i >= '0' && i <= '9');

245+ bool space = i == ' ' || i == '\t' || i == '\r' || i == '\n';

246+ parse_table[i] = (name1 ? tag_name1 : 0) | (name ? tag_name : 0) | (space ? tag_space : 0);

247+ }

248+ done = 1;

249+}

250+

251+bool static inline is(char c, char tag) { return parse_table[c] & tag; }

252+bool static inline is_name1(char c) { return is(c, tag_name1); }

253+bool static inline is_name(char c) { return is(c, tag_name); }

254+bool static inline is_space(char c) { return is(c, tag_space); }

255+

256+

228/////////////////////////////////////////////////////////////////////

257/////////////////////////////////////////////////////////////////////

229// PARSER COMBINATORS

258// PARSER COMBINATORS

230

259

@@ -238,8 +267,6 @@ void static inline skip(document* d, int i){d->cursor += i;}

238char static inline peek(document* d){return peekAt(d, 0);}

267char static inline peek(document* d){return peekAt(d, 0);}

239char static inline get(document* d){char c = peek(d); skip(d, 1); return c;}

268char static inline get(document* d){char c = peek(d); skip(d, 1); return c;}

240

269

241−bool static inline is_space(char c){ return c == ' ' || c == '\t' || c == '\r' || c == '\n';}

242−

243// Remove whitespace characters from the cursor while they are still whitespace

270// Remove whitespace characters from the cursor while they are still whitespace

244void static inline trim(document* d)

271void static inline trim(document* d)

245{

272{

@@ -286,15 +313,6 @@ void attr_alloc(attr_buffer* b, int ask)

286 b->attrs = buf2;

313 b->attrs = buf2;

287}

314}

288

315

289−int static inline isName(char c)

290−{

291− return

292− (c >= 'a' && c <= 'z') ||

293− (c >= 'A' && c <= 'Z') ||

294− (c >= '0' && c <= '9') ||

295− c == '_' || c == '-';

296−}

297−

298// you now expect a name, perhaps preceeded by whitespace

316// you now expect a name, perhaps preceeded by whitespace

299// the name may be empty

317// the name may be empty

300str parse_name(document* d)

318str parse_name(document* d)

@@ -302,8 +320,11 @@ str parse_name(document* d)

302 trim(d);

320 trim(d);

303 str res;

321 str res;

304 res.start = d->cursor;

322 res.start = d->cursor;

323+ if (!is_name1(peek(d)))

324+ return start_length(d->cursor, 0);

325+ skip(d, 1);

305− while (isName(peek(d)))

326+ while (is_name(peek(d)))

306− get(d);

327+ skip(d, 1);

307 res.length = d->cursor - res.start;

328 res.length = d->cursor - res.start;

308 return res;

329 return res;

309}

330}

@@ -443,6 +464,7 @@ str parse_content(document* d)

443document* document_parse(char* s, int slen)

464document* document_parse(char* s, int slen)

444{

465{

445 if (slen == -1) slen = strlen(s);

466 if (slen == -1) slen = strlen(s);

467+ init_parse_table();

446

468

447 document* d = malloc(sizeof(document));

469 document* d = malloc(sizeof(document));

448 d->body = s;

470 d->body = s;