Public-Mirrors
/
markdown-it
mirror of https://github.com/markdown-it/markdown-it.git
1
0
Fork 0
Code Issues Releases Activity
Browse Source

lhead, fences, indent cache

pull/14/head
Vitaly Puzrin 10 years ago
parent
e9645b6e4e
commit
9c99ff47a1
3 changed files with 224 additions and 101 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 26
      index.js
  2. 270
      lib/lexer_block.js
  3. 25
      lib/renderer.js

26
index.js
View File

@ -9,7 +9,7 @@ var LexerInline = require('./lib/lexer_inline');
// Parser state class
//
function State(src, lexerBlock, lexerInline, renderer, options) {
var ch, s, start, pos, len;
var ch, s, start, pos, len, indent, indent_found;
// TODO: Temporary solution. Check if more effective possible,
// withous str change
@ -37,17 +37,33 @@ function State(src, lexerBlock, lexerInline, renderer, options) {
this.tokens = [];
this.bMarks = []; // lines begin/end markers for fast jumps
this.eMarks = []; //
this.bMarks = []; // line begin offsets for fast jumps
this.eMarks = []; // line end offsets for fast jumps
this.tShift = []; // indent for each line
// Generate markers.
s = this.src;
for(start = pos = 0, len = s.length; pos < len; pos++) {
indent = 0;
indent_found = false;
for(start = pos = indent = 0, len = s.length; pos < len; pos++) {
ch = s.charCodeAt(pos);
// TODO: check other spaces and tabs too or keep existing regexp replace ??
if (!indent_found && ch === 0x20/* space */) {
indent++;
}
if (!indent_found && ch !== 0x20/* space */) {
this.tShift.push(indent);
indent_found = true;
}
if (ch === 0x0D || ch === 0x0A) {
this.bMarks.push(start);
this.eMarks.push(pos);
indent_found = false;
indent = 0;
start = pos + 1;
}
if (ch === 0x0D && pos < len && s.charCodeAt(pos) === 0x0A) {
@ -58,6 +74,7 @@ function State(src, lexerBlock, lexerInline, renderer, options) {
if (ch !== 0x0D || ch !== 0x0A) {
this.bMarks.push(start);
this.eMarks.push(len);
this.tShift.push(indent);
}
// inline lexer variables
@ -71,7 +88,6 @@ function State(src, lexerBlock, lexerInline, renderer, options) {
// renderer
this.result = '';
}

270
lib/lexer_block.js
View File

@ -16,22 +16,12 @@ function isWhiteSpace(ch) {
// Check if line from `pos` is empty or contains spaces only
function isEmpty(state, line) {
var ch, pos = state.bMarks[line], max = state.src.length;
while (pos < max) {
ch = state.src.charCodeAt(pos++);
if (ch === 0x0A || ch === 0x0D) { return true; }
if (!isWhiteSpace(ch)) { return false; }
}
return true; // EOL reached
return state.bMarks[line] + state.tShift[line] >= state.eMarks[line];
}
// Return absolute position of char with default indent an given line,
// or -1 if no requested indent
function getIndent(state, line, indent) {
/*function getIndent(state, line, indent) {
var ch, pos, max;
if (line >= state.lineMax) { return -1; }
@ -41,7 +31,6 @@ function getIndent(state, line, indent) {
while (pos < max && indent > 0) {
ch = state.src.charCodeAt(pos++);
if (ch === 0x09) { indent -= 4; continue; }
if (isWhiteSpace(ch)) { indent--; continue; }
return -1;
}
@ -49,18 +38,38 @@ function getIndent(state, line, indent) {
if (indent > 0) { return -1; }
return pos;
}
}*/
// Skip empty lines, starting from `state.line`
// Seek first non empty line from given one and return it's number
function skipEmptyLines(state, from) {
while (from < state.lineMax) {
if (!isEmpty(state, from)) {
state.line = from;
return;
for (var max = state.lineMax; from < max; from++) {
if (!isEmpty(state, from)) { break; }
}
from++;
return from;
}
// Skip spaces from given position. Returns new position
function skipSpaces(state, pos) {
for (var max = state.src.length; pos < max; pos++) {
if (!isWhiteSpace(state.src.charCodeAt(pos))) { break; }
}
state.line = from;
return pos;
}
// Skip char codes from given position
function skipChars(state, pos, code) {
for (var max = state.src.length; pos < max; pos++) {
if (code !== state.src.charCodeAt(pos)) { break; }
}
return pos;
}
// Skip char codes reverse from given position
function skipCharsBack(state, pos, code, min) {
for (; pos >= min; pos--) {
if (code !== state.src.charCodeAt(pos)) { break; }
}
return pos;
}
@ -70,11 +79,11 @@ function skipEmptyLines(state, from) {
var rules = [];
// code
// code (4 spaced padded)
rules.push(function code(state, startLine, endLine, silent) {
var nextLine, last;
if (getIndent(state, startLine, 4) === -1) { return false; }
if (state.tShift[startLine] < 4) { return false; }
last = nextLine = startLine + 1;
@ -86,7 +95,7 @@ rules.push(function code(state, startLine, endLine, silent) {
}
continue;
}
if (getIndent(state, nextLine, 4) !== -1) {
if (state.tShift[nextLine] >= 4) {
nextLine++;
last = nextLine;
continue;
@ -107,18 +116,92 @@ rules.push(function code(state, startLine, endLine, silent) {
});
// heading
// fences (``` lang, ~~~ lang)
rules.push(function fences(state, startLine, endLine, silent) {
var marker, len, params, nextLine,
pos = state.bMarks[startLine] + state.tShift[startLine],
max = state.eMarks[startLine];
if (pos + 3 > max) { return false; }
marker = state.src.charCodeAt(pos);
if (marker !== 0x7E/* ~ */ && marker !== 0x60 /* ` */) {
return false;
}
// scan marker length
len = 1;
while (state.src.charCodeAt(++pos) === marker) {
len++;
}
if (len < 3) { return false; }
params = state.src.slice(pos, max).trim();
if (!/\S/.test(params)) { return false; }
// search end of block
nextLine = startLine;
do {
nextLine++;
if (nextLine > endLine) { return false; }
pos = state.bMarks[nextLine] + state.tShift[nextLine];
max = state.eMarks[nextLine];
if (pos + 3 > max) { continue; }
// check markers
if (state.src.charCodeAt(pos) !== marker &&
state.src.charCodeAt(pos + 1) !== marker &&
state.src.charCodeAt(pos + 2) !== marker) {
continue;
}
pos += 3;
// make sure tail has spaces only
//pos = pos < max ? skipSpaces(state, pos) : pos;
// stmd allow any combonation of markers and spaces in tail
if (pos < max) { continue; }
// found!
break;
} while (true);
if (silent) { return true; }
state.tokens.push({
type: 'fence',
params: params.split(/\s+/g),
startLine: startLine + 1,
endLine: nextLine
});
state.line = skipEmptyLines(state, nextLine + 1);
return true;
});
// heading (#, ##, ...)
rules.push(function heading(state, startLine, endLine, silent) {
var ch, level,
pos = state.bMarks[startLine],
max = state.eMarks[startLine];
max = state.eMarks[startLine],
start = pos;
ch = state.src.charCodeAt(pos);
pos += state.tShift[startLine];
// skip leading spaces
while (isWhiteSpace(ch) && pos < max) {
ch = state.src.charCodeAt(++pos);
}
if (pos >= max) { return false; }
ch = state.src.charCodeAt(pos);
if (ch !== 0x23/* # */ || pos >= max) { return false; }
@ -130,15 +213,10 @@ rules.push(function heading(state, startLine, endLine, silent) {
ch = state.src.charCodeAt(++pos);
}
if (!isWhiteSpace(ch) || pos >= max || level > 6) { return false; }
if (level > 6 || (pos < max && !isWhiteSpace(ch))) { return false; }
// skip spaces before heading text
ch = state.src.charCodeAt(++pos);
while (isWhiteSpace(ch) && pos < max) {
ch = state.src.charCodeAt(++pos);
}
if (pos >= max) { return false; }
pos = pos < max ? skipSpaces(state, pos) : pos;
// Now pos contains offset of first heared char
// Let's cut tails like ' ### ' from the end of string
@ -146,15 +224,15 @@ rules.push(function heading(state, startLine, endLine, silent) {
max--;
ch = state.src.charCodeAt(max);
while (isWhiteSpace(ch) && max > pos) {
while (max > start && isWhiteSpace(ch)) {
ch = state.src.charCodeAt(--max);
}
if (ch === 0x23/* # */) {
while (ch === 0x23/* # */ && max > pos) {
while (max > start && ch === 0x23/* # */) {
ch = state.src.charCodeAt(--max);
}
if (isWhiteSpace(ch)) {
while (isWhiteSpace(ch) && max > pos) {
while (max > start && isWhiteSpace(ch)) {
ch = state.src.charCodeAt(--max);
}
} else if (ch === 0x5C/* \ */) {
@ -165,81 +243,92 @@ rules.push(function heading(state, startLine, endLine, silent) {
if (silent) { return true; }
if (silent) {
return true;
}
state.tokens.push({ type: 'heading_open', level: level });
// only if header is not empty
if (pos < max) {
state.lexerInline.tokenize(state, pos, max);
}
state.tokens.push({ type: 'heading_close', level: level });
skipEmptyLines(state, ++startLine);
state.line = skipEmptyLines(state, ++startLine);
return true;
});
// lheading (---, ===)
rules.push(function lheading(state, startLine, endLine, silent) {
var marker, pos, mem, max,
next = startLine + 1;
if (next >= state.lineMax) { return false; }
// Scan next line
pos = state.bMarks[next] + state.tShift[next];
max = state.eMarks[next];
if (pos + 3 > max) { return false; }
marker = state.src.charCodeAt(pos);
if (marker !== 0x2D/* - */ && marker !== 0x3D/* = */) { return false; }
mem = pos;
pos = skipChars(state, pos, marker);
if (pos - mem < 3) { return false; }
pos = skipSpaces(state, pos);
if (pos < max) { return false; }
state.tokens.push({ type: 'heading_open', level: marker === 0x3D/* = */ ? 1 : 2 });
state.lexerInline.tokenize(state, state.bMarks[startLine], state.eMarks[startLine]);
state.tokens.push({ type: 'heading_close', level: marker === 0x3D/* = */ ? 1 : 2 });
state.line = skipEmptyLines(state, ++next);
return true;
});
// Horizontal rule
rules.push(function hr(state, startLine, endLine, silent) {
var ch, marker,
var marker, cnt, ch,
pos = state.bMarks[startLine],
space_max = pos + 3,
max = state.eMarks[startLine];
ch = state.src.charCodeAt(pos);
// should not have > 3 leading spaces
if (state.tShift[startLine] > 3) { return false; }
// quick test first char
if (!isWhiteSpace(ch) &&
ch !== 0x2A/* * */ &&
ch !== 0x2D/* - */ &&
ch !== 0x5F/* _ */) {
return false;
}
pos += state.tShift[startLine];
// skip up to 3 leading spaces
while (isWhiteSpace(ch) && pos < max && pos < space_max) {
pos++;
ch = state.src.charCodeAt(pos);
}
if (pos > max) { return false; }
marker = state.src.charCodeAt(pos++);
// Check hr marker
if (ch !== 0x2A/* * */ &&
ch !== 0x2D/* - */ &&
ch !== 0x5F/* _ */) {
if (marker !== 0x2A/* * */ &&
marker !== 0x2D/* - */ &&
marker !== 0x5F/* _ */) {
return false;
}
// remember marker type
marker = ch;
// markers can be mixed with spaces, but there should be at least 3 one
if (pos + 2 < max &&
state.src.charCodeAt(pos + 1) === marker &&
state.src.charCodeAt(pos + 2) === marker) {
// Style 1: ***, ---, ___
pos += 3;
} else if (pos + 4 < max &&
isWhiteSpace(state.src.charCodeAt(pos + 1)) &&
state.src.charCodeAt(pos + 2) === marker &&
isWhiteSpace(state.src.charCodeAt(pos + 3)) &&
state.src.charCodeAt(pos + 4) === marker) {
// Style 2: * * *, - - -, _ _ _
pos += 5;
} else {
return false;
}
// check that line tail has spaces only
while(pos < max) {
cnt = 1;
while (pos < max) {
ch = state.src.charCodeAt(pos++);
if (isWhiteSpace(ch)) {
return false;
}
if (ch !== marker && !isWhiteSpace(ch)) { return false; }
if (ch === marker) { cnt++; }
}
if (cnt < 3) { return false; }
if (silent) { return true; }
state.tokens.push({ type: 'hr' });
skipEmptyLines(state, ++startLine);
state.line = skipEmptyLines(state, ++startLine);
return true;
});
@ -252,9 +341,10 @@ rules.push(function paragraph(state, startLine, endLine) {
// jump line-by-line until empty one or EOF
while (nextLine < endLine && !isEmpty(state, nextLine)) {
// Force paragraph termination of next tag found
if (rules_named.fences(state, nextLine, endLine, true)) { break; }
if (rules_named.hr(state, nextLine, endLine, true)) { break; }
if (rules_named.heading(state, nextLine, endLine, true)) { break; }
//if (rules_named.lheading(state, nextLine, endLine, true)) { break; }
if (rules_named.lheading(state, nextLine, endLine, true)) { break; }
//if (rules_named.blockquote(state, nextLine, endLine, true)) { break; }
//if (rules_named.tag(state, nextLine, endLine, true)) { break; }
//if (rules_named.def(state, nextLine, endLine, true)) { break; }
@ -269,7 +359,7 @@ rules.push(function paragraph(state, startLine, endLine) {
);
state.tokens.push({ type: 'paragraph_close' });
skipEmptyLines(state, nextLine);
state.line = skipEmptyLines(state, nextLine);
return true;
});

25
lib/renderer.js
View File

@ -5,6 +5,12 @@ function escapeHTML(str) {
return str.replace(/&/g, '&amp;').replace(/</g, '&lt;').replace(/>/g, '&gt;');
}
var MD_UNESCAPE_RE = /\\([!"#$%&\'()*+,.\/:;<=>?@[\\\]^_`{|}~-])/g;
function unescapeMD(str) {
return str.replace(MD_UNESCAPE_RE, '$1');
}
function joinLines(state, begin, end) {
return state.src.slice(
state.bMarks[begin],
@ -16,10 +22,21 @@ var rules = {};
rules.code = function (state, token) {
// TODO: check if we need variable indent cut
var lines = joinLines(state, token.startLine, token.endLine).replace(/^ {4}/gm, '');
var content = joinLines(state, token.startLine, token.endLine).replace(/^ {4}/gm, '');
state.result += '<pre><code>' + escapeHTML(content) + '</code></pre>\n';
};
rules.fence = function (state, token) {
var content = joinLines(state, token.startLine, token.endLine);
var langMark = '';
if (token.params.length) {
langMark = ' class="language-' + escapeHTML(token.params[0]) + '"';
}
state.result += '<pre><code>' + escapeHTML(lines) + '</code></pre>\n';
state.result += '<pre><code' + langMark + '>' + escapeHTML(content) + '</code></pre>\n';
};
@ -45,7 +62,7 @@ rules.paragraph_close = function (state, token) {
rules.text = function (state, token) {
state.result += escapeHTML(state.src.slice(token.begin, token.end));
state.result += escapeHTML(unescapeMD(state.src.slice(token.begin, token.end)));
};

Write Preview
Loading…
Cancel
Save