md2html.awk

1 #!/bin/awk -f

2 #

3 # by: Jesus Galan (yiyus) 2009

4 #

5 # Usage: md2html.awk file.md > file.html

6 # See: http://4l77.com/src/md2html.awk

7

8 function eschtml(t) {

9 gsub("&", "\\&", t);

10 gsub("<", "\\&lt;", t);

11 return t;

12 }

13

14 function oprint(t){

15 if(nr == 0)

16 print t;

17 else

18 otext = otext "\n" t;

19 }

20

21 function subref(id){

22 for(; nr > 0 && sub("<<" id, ref[id], otext); nr--);

23 if(nr == 0 && otext) {

24 print otext;

25 otext = "";

26 }

27 }

28

29 function nextil(t) {

30 if(!match(t, /[`<&\[*_\\-]|(\!\[)/))

31 return t;

32 t1 = substr(t, 1, RSTART - 1);

33 tag = substr(t, RSTART, RLENGTH);

34 t2 = substr(t, RSTART + RLENGTH);

35 if(ilcode && tag != "`")

36 return eschtml(t1 tag) nextil(t2);

37 # Backslash escaping

38 if(tag == "\\"){

39 if(match(t2, /^[\\`*_{}\[\]()#+\-\.!]/)){

40 tag = substr(t2, 1, 1);

41 t2 = substr(t2, 2);

42 }

43 return t1 tag nextil(t2);

44 }

45 # Dashes

46 if(tag == "-"){

47 if(sub(/^-/, "", t2))

48 tag = "&#8212;";

49 return t1 tag nextil(t2);

50 }

51 # Inline Code

52 if(tag == "`"){

53 if(sub(/^`/, "", t2)){

54 if(!match(t2, /``/))

55 return t1 "&#8221;" nextil(t2);

56 ilcode2 = !ilcode2;

57 }

58 else if(ilcode2)

59 return t1 tag nextil(t2);

60 tag = "<code>";

61 if(ilcode){

62 t1 = eschtml(t1);

63 tag = "</code>";

64 }

65 ilcode = !ilcode;

66 return t1 tag nextil(t2);

67 }

68 if(tag == "<"){

69 # Autolinks

70 if(match(t2, /^[^ ]+[\.@][^ ]+>/)){

71 url = eschtml(substr(t2, 1, RLENGTH - 1));

72 t2 = substr(t2, RLENGTH + 1);

73 linktext = url;

74 if(match(url, /@/) && !match(url, /^mailto:/))

75 url = "mailto:" url;

76 return t1 "<a href=\"" url "\">" linktext "</a>" nextil(t2);

77 }

78 # Html tags

79 if(match(t2, /^[A-Za-z\/!][^>]*>/)){

80 tag = tag substr(t2, RSTART, RLENGTH);

81 t2 = substr(t2, RLENGTH + 1);

82 return t1 tag nextil(t2);

83 }

84 return t1 "&lt;" nextil(t2);

85 }

86 # Html special entities

87 if(tag == "&"){

88 if(match(t2, /^#?[A-Za-z0-9]+;/)){

89 tag = tag substr(t2, RSTART, RLENGTH);

90 t2 = substr(t2, RLENGTH + 1);

91 return t1 tag nextil(t2);

92 }

93 return t1 "&amp;" nextil(t2);

94 }

95 # Images

96 if(tag == "!["){

97 if(!match(t2, /(\[.*\])|(\(.*\))/))

98 return t1 tag nextil(t2);

99 match(t2, /^[^\]]*/);

100 alt = substr(t2, 1, RLENGTH);

101 t2 = substr(t2, RLENGTH + 2);

102 if(match(t2, /^\(/)){

103 # Inline

104 sub(/^\(/, "", t2);

105 match(t2, /^[^\)]+/);

106 url = eschtml(substr(t2, 1, RLENGTH));

107 t2 = substr(t2, RLENGTH + 2);

108 title = "";

109 if(match(url, /[ ]+\".*\"[ ]*$/)) {

110 title = substr(url, RSTART, RLENGTH);

111 url = substr(url, 1, RSTART - 1);

112 match(title, /\".*\"/);

113 title = " title=\"" substr(title, RSTART + 1, RLENGTH - 2) "\"";

114 }

115 if(match(url, /^<.*>$/))

116 url = substr(url, 2, RLENGTH - 2);

117 return t1 "<img src=\"" url "\" alt=\"" alt "\"" title " />" nextil(t2);

118 }

119 else{

120 # Referenced

121 sub(/^ ?\[/, "", t2);

122 id = alt;

123 if(match(t2, /^[^\]]+/))

124 id = substr(t2, 1, RLENGTH);

125 t2 = substr(t2, RLENGTH + 2);

126 if(ref[id])

127 r = ref[id];

128 else{

129 r = "<<" id;

130 nr++;

131 }

132 return t1 "<img src=\"" r "\" alt=\"" alt "\" />" nextil(t2);

133 }

134 }

135 # Links

136 if(tag == "["){

137 if(!match(t2, /(\[.*\])|(\(.*\))/))

138 return t1 tag nextil(t2);

139 match(t2, /^[^\]]*(\[[^\]]*\][^\]]*)*/);

140 linktext = substr(t2, 1, RLENGTH);

141 t2 = substr(t2, RLENGTH + 2);

142 if(match(t2, /^\(/)){

143 # Inline

144 match(t2, /^[^\)]+(\([^\)]+\)[^\)]*)*/);

145 url = substr(t2, 2, RLENGTH - 1);

146 pt2 = substr(t2, RLENGTH + 2);

147 title = "";

148 if(match(url, /[ ]+\".*\"[ ]*$/)) {

149 title = substr(url, RSTART, RLENGTH);

150 url = substr(url, 1, RSTART - 1);

151 match(title, /\".*\"/);

152 title = " title=\"" substr(title, RSTART + 1, RLENGTH - 2) "\"";

153 }

154 if(match(url, /^<.*>$/))

155 url = substr(url, 2, RLENGTH - 2);

156 url = eschtml(url);

157 return t1 "<a href=\"" url "\"" title ">" nextil(linktext) "</a>" nextil(pt2);

158 }

159 else{

160 # Referenced

161 sub(/^ ?\[/, "", t2);

162 id = linktext;

163 if(match(t2, /^[^\]]+/))

164 id = substr(t2, 1, RLENGTH);

165 t2 = substr(t2, RLENGTH + 2);

166 if(ref[id])

167 r = ref[id];

168 else{

169 r = "<<" id;

170 nr++;

171 }

172 pt2 = t2;

173 return t1 "<a href=\"" r "\" />" nextil(linktext) "</a>" nextil(pt2);

174 }

175 }

176 # Emphasis

177 if(match(tag, /[*_]/)){

178 ntag = tag;

179 if(sub("^" tag, "", t2)){

180 if(stag[ns] == tag && match(t2, "^" tag))

181 t2 = tag t2;

182 else

183 ntag = tag tag

184 }

185 n = length(ntag);

186 tag = (n == 2) ? "strong" : "em";

187 if(match(t1, / $/) && match(t2, /^ /))

188 return t1 tag nextil(t2);

189 if(stag[ns] == ntag){

190 tag = "/" tag;

191 ns--;

192 }

193 else

194 stag[++ns] = ntag;

195 tag = "<" tag ">";

196 return t1 tag nextil(t2);

197 }

198 }

199

200 function inline(t) {

201 ilcode = 0;

202 ilcode2 = 0;

203 ns = 0;

204

205 return nextil(t);

206 }

207

208 function printp(tag) {

209 if(!match(text, /^[ ]*$/)){

210 text = inline(text);

211 if(tag != "")

212 oprint("<" tag ">" text "</" tag ">");

213 else

214 oprint(text);

215 }

216 text = "";

217 }

218

219 BEGIN {

220 blank = 0;

221 code = 0;

222 hr = 0;

223 html = 0;

224 nl = 0;

225 nr = 0;

226 otext = "";

227 text = "";

228 par = "p";

229 }

230

231 # References

232 !code && /^ *\[[^\]]*\]:[ ]+/ {

233 sub(/^ *\[/, "");

234 match($0, /\]/);

235 id = substr($0, 1, RSTART - 1);

236 sub(id "\\]:[ ]+", "");

237 title = "";

238 if(match($0, /\".*\"$/))

239 title = "\" title=\"" substr($0, RSTART + 1, RLENGTH - 2);

240 sub(/[ ]+\".*\"$/, "");

241 url = eschtml($0);

242 ref[id] = url title;

243

244 subref(id);

245 next;

246 }

247

248 # html

249 !html && /^<(address|blockquote|center|dir|div|dl|fieldset|form|h[1-6r]|\

250 isindex|menu|noframes|noscript|ol|p|pre|table|ul|!--)/ {

251 if(code)

252 oprint("</pre></code>");

253 for(; !text && block[nl] == "blockquote"; nl--)

254 oprint("</blockquote>");

255 match($0, /^<(address|blockquote|center|dir|div|dl|fieldset|form|h[1-6r]|\

256 isindex|menu|noframes|noscript|ol|p|pre|table|ul|!--)/);

257 htag = substr($0, 2, RLENGTH - 1);

258 if(!match($0, "(<\\/" htag ">)|((^<hr ?\\/?)|(--)>$)"))

259 html = 1;

260 if(html && match($0, /^<hr/))

261 hr = 1;

262 oprint($0);

263 next;

264 }

265

266 html && (/(^<\/(address|blockquote|center|dir|div|dl|fieldset|form|h[1-6r]|\

267 isindex|menu|noframes|noscript|ol|p|pre|table|ul).*)|(--)>$/ ||

268 (hr && />$/)) {

269 html = 0;

270 hr = 0;

271 oprint($0);

272 next;

273 }

274

275 html {

276 oprint($0);

277 next;

278 }

279

280 # List and quote blocks

281

282 # Remove indentation

283 {

284 for(nnl = 0; nnl < nl; nnl++)

285 if((match(block[nnl + 1], /[ou]l/) && !sub(/^( | )/, "")) || \

286 (block[nnl + 1] == "blockquote" && !sub(/^> ?/, "")))

287 break;

288 }

289 nnl < nl && !blank && text && ! /^ ? ? ?([*+-]|([0-9]+\.)+)( +| )/ { nnl = nl; }

290 # Quote blocks

291 {

292 while(sub(/^> /, ""))

293 nblock[++nnl] = "blockquote";

294 }

295 # Horizontal rules

296 { hr = 0; }

297 (blank || (!text && !code)) && /^ ? ? ?([-*_][ ]*)([-*_][ ]*)([-*_][ ]*)+$/ {

298 if(code){

299 oprint("</pre></code>");

300 code = 0;

301 }

302 blank = 0;

303 nnl = 0;

304 hr = 1;

305 }

306 # List items

307 block[nl] ~ /[ou]l/ && /^$/ {

308 blank = 1;

309 next;

310 }

311 { newli = 0; }

312 !hr && (nnl != nl || !text || block[nl] ~ /[ou]l/) && /^ ? ? ?[*+-]( +| )/ {

313 sub(/^ ? ? ?[*+-]( +| )/, "");

314 nnl++;

315 nblock[nnl] = "ul";

316 newli = 1;

317 }

318 (nnl != nl || !text || block[nl] ~ /[ou]l/) && /^ ? ? ?([0-9]+\.)+( +| )/ {

319 sub(/^ ? ? ?([0-9]+\.)+( +| )/, "");

320 nnl++;

321 nblock[nnl] = "ol";

322 newli = 1;

323 }

324 newli {

325 if(blank && nnl == nl && !par)

326 par = "p";

327 blank = 0;

328 printp(par);

329 if(nnl == nl && block[nl] == nblock[nl])

330 oprint("</li><li>");

331 }

332 blank && ! /^$/ {

333 if(match(block[nnl], /[ou]l/) && !par)

334 par = "p";

335 printp(par);

336 par = "p";

337 blank = 0;

338 }

339

340 # Close old blocks and open new ones

341 nnl != nl || nblock[nl] != block[nl] {

342 if(code){

343 oprint("</pre></code>");

344 code = 0;

345 }

346 printp(par);

347 b = (nnl > nl) ? nblock[nnl] : block[nnl];

348 par = (match(b, /[ou]l/)) ? "" : "p";

349 }

350 nnl < nl || (nnl == nl && nblock[nl] != block[nl]) {

351 for(; nl > nnl || (nnl == nl && pblock[nl] != block[nl]); nl--){

352 if(match(block[nl], /[ou]l/))

353 oprint("</li>");

354 oprint("</" block[nl] ">");

355 }

356 }

357 nnl > nl {

358 for(; nl < nnl; nl++){

359 block[nl + 1] = nblock[nl + 1];

360 oprint("<" block[nl + 1] ">");

361 if(match(block[nl + 1], /[ou]l/))

362 oprint("<li>");

363 }

364 }

365 hr {

366 oprint("<hr>");

367 next;

368 }

369

370 # Code blocks

371 code && /^$/ {

372 if(blanK)

373 oprint("");

374 blank = 1;

375 next;

376 }

377 !text && sub(/^( | )/, "") {

378 if(blanK)

379 oprint("");

380 blank = 0;

381 if(!code)

382 oprint("<code><pre>");

383 code = 1;

384 $0 = eschtml($0);

385 oprint($0);

386 next;

387 }

388 code {

389 oprint("</pre></code>");

390 code = 0;

391 }

392

393 # Setex-style Headers

394 text && /^=+$/ {printp("h1"); next;}

395 text && /^-+$/ {printp("h2"); next;}

396

397 # Atx-Style headers

398 /^#+/ && (!newli || par=="p" || /^##/) {

399 for(n = 0; n < 6 && sub(/^# */, ""); n++)

400 sub(/#$/, "");

401 par = "h" n;

402 }

403

404 # Paragraph

405 /^$/ {

406 printp(par);

407 par = "p";

408 next;

409 }

410

411 # Add text

412 { text = (text ? text " " : "") $0; }

413

414 END {

415 if(code){

416 oprint("</pre></code>");

417 code = 0;

418 }

419 printp(par);

420 for(; nl > 0; nl--){

421 if(match(block[nl], /[ou]l/))

422 oprint("</li>");

423 oprint("</" block[nl] ">");

424 }

425 gsub(/<<[^\"]*/, "", otext);

426 print(otext);

427 }