Skip to content
Snippets Groups Projects
Commit ece89370 authored by Marijn Haverbeke's avatar Marijn Haverbeke
Browse files

[markdown mode] Make strong/em delimiting more CommonMark-compliant 

Closes #4686
parent 5aa64cc4
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
mode/gfm/gfm.js
+ 0
− 1
View file @ ece89370
......@@ -113,7 +113,6 @@ CodeMirror.defineMode("gfm", function(config, modeConfig) {
};
var markdownConfig = {
underscoresBreakWords: false,
taskLists: true,
fencedCodeBlocks: '```',
strikethrough: true
......
mode/markdown/markdown.js
+ 30
− 43
View file @ ece89370
......@@ -35,10 +35,6 @@ CodeMirror.defineMode("markdown", function(cmCfg, modeCfg) {
if (modeCfg.maxBlockquoteDepth === undefined)
modeCfg.maxBlockquoteDepth = 0;
// Should underscores in words open/close em/strong?
if (modeCfg.underscoresBreakWords === undefined)
modeCfg.underscoresBreakWords = true;
// Use `fencedCodeBlocks` to configure fenced code blocks. false to
// disable, string to specify a precise regexp that the fence should
// match, and true to allow three or more backticks or tildes (as
......@@ -89,7 +85,8 @@ CodeMirror.defineMode("markdown", function(cmCfg, modeCfg) {
, setextHeaderRE = /^ *(?:\={1,}|-{1,})\s*$/
, textRE = /^[^#!\[\]*_\\<>` "'(~]+/
, fencedCodeRE = new RegExp("^(" + (modeCfg.fencedCodeBlocks === true ? "~~~+|```+" : modeCfg.fencedCodeBlocks) +
")[ \\t]*([\\w+#\-]*)");
")[ \\t]*([\\w+#\-]*)")
, punctuation = /[!\"#$%&\'()*+,\-\.\/:;<=>?@\[\\\]^_`{|}~—]/
function switchInline(stream, state, f) {
state.f = state.inline = f;
......@@ -375,9 +372,6 @@ CodeMirror.defineMode("markdown", function(cmCfg, modeCfg) {
return getType(state);
}
// Get sol() value now, before character is consumed
var sol = stream.sol();
var ch = stream.next();
// Matches link titles present on next line
......@@ -387,7 +381,7 @@ CodeMirror.defineMode("markdown", function(cmCfg, modeCfg) {
if (ch === '(') {
matchCh = ')';
}
matchCh = (matchCh+'').replace(/([.?*+^$[\]\\(){}|-])/g, "\\$1");
matchCh = (matchCh+'').replace(/([.?*+^\[\]\\(){}|-])/g, "\\$1");
var regex = '^\\s*(?:[^' + matchCh + '\\\\]+|\\\\\\\\|\\\\.)' + matchCh;
if (stream.match(new RegExp(regex), true)) {
return tokenTypes.linkHref;
......@@ -499,41 +493,34 @@ CodeMirror.defineMode("markdown", function(cmCfg, modeCfg) {
if (ch === '<' && stream.match(/^\/\w*?>/)) {
state.md_inside = false;
return "tag";
}
var ignoreUnderscore = false;
if (!modeCfg.underscoresBreakWords) {
if (ch === '_' && stream.peek() !== '_' && stream.match(/(\w)/, false)) {
var prevPos = stream.pos - 2;
if (prevPos >= 0) {
var prevCh = stream.string.charAt(prevPos);
if (prevCh !== '_' && prevCh.match(/(\w)/, false)) {
ignoreUnderscore = true;
}
}
} else if (ch === "*" || ch === "_") {
var len = 1, before = stream.pos == 1 ? " " : stream.string.charAt(stream.pos - 2)
while (len < 3 && stream.eat(ch)) len++
var after = stream.peek() || " "
// See http://spec.commonmark.org/0.27/#emphasis-and-strong-emphasis
var leftFlanking = !/\s/.test(after) && (!punctuation.test(after) || /\s/.test(before) || punctuation.test(before))
var rightFlanking = !/\s/.test(before) && (!punctuation.test(before) || /\s/.test(after) || punctuation.test(after))
var setEm = null, setStrong = null
if (len % 2) { // Em
if (!state.em && leftFlanking && (ch === "*" || !rightFlanking || punctuation.test(before)))
setEm = true
else if (state.em == ch && rightFlanking && (ch === "*" || !leftFlanking || punctuation.test(after)))
setEm = false
}
}
if (ch === '*' || (ch === '_' && !ignoreUnderscore)) {
if (sol && stream.peek() === ' ') {
// Do nothing, surrounded by newline and space
} else if (state.strong === ch && stream.eat(ch)) { // Remove STRONG
if (modeCfg.highlightFormatting) state.formatting = "strong";
var t = getType(state);
state.strong = false;
return t;
} else if (!state.strong && stream.eat(ch)) { // Add STRONG
state.strong = ch;
if (modeCfg.highlightFormatting) state.formatting = "strong";
return getType(state);
} else if (state.em === ch) { // Remove EM
if (modeCfg.highlightFormatting) state.formatting = "em";
var t = getType(state);
state.em = false;
return t;
} else if (!state.em) { // Add EM
state.em = ch;
if (modeCfg.highlightFormatting) state.formatting = "em";
return getType(state);
if (len > 1) { // Strong
if (!state.strong && leftFlanking && (ch === "*" || !rightFlanking || punctuation.test(before)))
setStrong = true
else if (state.strong == ch && rightFlanking && (ch === "*" || !leftFlanking || punctuation.test(after)))
setStrong = false
}
if (setStrong != null || setEm != null) {
if (modeCfg.highlightFormatting) state.formatting = setEm == null ? "strong" : setStrong == null ? "em" : "strong em"
if (setEm === true) state.em = ch
if (setStrong === true) state.strong = ch
var t = getType(state)
if (setEm === false) state.em = false
if (setStrong === false) state.strong = false
return t
}
} else if (ch === ' ') {
if (stream.eat('*') || stream.eat('_')) { // Probably surrounded by spaces
......
mode/markdown/test.js
+ 12
− 12
View file @ ece89370
......@@ -381,28 +381,28 @@
MT("listAsteriskFormatting",
"[variable-2 * ][variable-2&em *foo*][variable-2 bar]",
"[variable-2 * ][variable-2&strong **foo**][variable-2 bar]",
"[variable-2 * ][variable-2&strong **][variable-2&em&strong *foo**][variable-2&em *][variable-2 bar]",
"[variable-2 * ][variable-2&em&strong ***foo***][variable-2 bar]",
"[variable-2 * ][variable-2&comment `foo`][variable-2 bar]");
// Formatting in lists (+)
MT("listPlusFormatting",
"[variable-2 + ][variable-2&em *foo*][variable-2 bar]",
"[variable-2 + ][variable-2&strong **foo**][variable-2 bar]",
"[variable-2 + ][variable-2&strong **][variable-2&em&strong *foo**][variable-2&em *][variable-2 bar]",
"[variable-2 + ][variable-2&em&strong ***foo***][variable-2 bar]",
"[variable-2 + ][variable-2&comment `foo`][variable-2 bar]");
// Formatting in lists (-)
MT("listDashFormatting",
"[variable-2 - ][variable-2&em *foo*][variable-2 bar]",
"[variable-2 - ][variable-2&strong **foo**][variable-2 bar]",
"[variable-2 - ][variable-2&strong **][variable-2&em&strong *foo**][variable-2&em *][variable-2 bar]",
"[variable-2 - ][variable-2&em&strong ***foo***][variable-2 bar]",
"[variable-2 - ][variable-2&comment `foo`][variable-2 bar]");
// Formatting in lists (1.)
MT("listNumberFormatting",
"[variable-2 1. ][variable-2&em *foo*][variable-2 bar]",
"[variable-2 2. ][variable-2&strong **foo**][variable-2 bar]",
"[variable-2 3. ][variable-2&strong **][variable-2&em&strong *foo**][variable-2&em *][variable-2 bar]",
"[variable-2 3. ][variable-2&em&strong ***foo***][variable-2 bar]",
"[variable-2 4. ][variable-2&comment `foo`][variable-2 bar]");
// Paragraph lists
......@@ -605,7 +605,7 @@
"[image&image-marker !][image&image-alt-text&link [[][image-alt-text&strong&image&link **alt text**][image&image-alt-text&link ]]][string&url (http://link.to/image.jpg)]");
MT("imageEmStrong",
"[image&image-marker !][image&image-alt-text&link [[][image-alt-text&image&strong&link **][image&image-alt-text&em&strong&link *alt text**][image&image-alt-text&em&link *][image&image-alt-text&link ]]][string&url (http://link.to/image.jpg)]");
"[image&image-marker !][image&image-alt-text&link [[][image&image-alt-text&em&strong&link ***alt text***][image&image-alt-text&link ]]][string&url (http://link.to/image.jpg)]");
// Inline link with title
MT("linkTitle",
......@@ -629,7 +629,7 @@
// Inline link with EmStrong
MT("linkEmStrong",
"[link [[][link&strong **][link&em&strong *foo**][link&em *][link ]]][string&url (http://example.com/)] bar");
"[link [[][link&em&strong ***foo***][link ]]][string&url (http://example.com/)] bar");
// Image with title
MT("imageTitle",
......@@ -663,7 +663,7 @@
// Reference-style links with EmStrong
MT("linkReferenceEmStrong",
"[link [[][link&strong **][link&em&strong *foo**][link&em *][link ]]][string&url [[bar]]] hello");
"[link [[][link&em&strong ***foo***][link ]]][string&url [[bar]]] hello");
// Reference-style links with optional space separator (per documentation)
// "You can optionally use a space to separate the sets of brackets"
......@@ -757,7 +757,7 @@
"foo[em *bar*]hello");
MT("emInWordUnderscore",
"foo[em _bar_]hello");
"foo_bar_hello");
// Per documentation: "...surround an * or _ with spaces, it’ll be
// treated as a literal asterisk or underscore."
......@@ -766,11 +766,11 @@
"foo [em _bar _ hello_] world");
MT("emEscapedBySpaceOut",
"foo _ bar[em _hello_]world");
"foo _ bar [em _hello_] world");
MT("emEscapedByNewline",
"foo",
"_ bar[em _hello_]world");
"_ bar [em _hello_] world");
// Unclosed emphasis characters
// Instead of simply marking as EM / STRONG, it would be nice to have an
......@@ -791,14 +791,14 @@
"[em *foo][em&strong **bar*][strong hello**] world");
MT("emStrongUnderscore",
"[em _foo][em&strong __bar_][strong hello__] world");
"[em _foo ][em&strong __bar_][strong hello__] world");
// "...same character must be used to open and close an emphasis span.""
MT("emStrongMixed",
"[em _foo][em&strong **bar*hello__ world]");
MT("emStrongMixed",
"[em *foo][em&strong __bar_hello** world]");
"[em *foo ][em&strong __bar_hello** world]");
MT("linkWithNestedParens",
"[link [[foo]]][string&url (bar(baz))]")
......
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment