// Parse a link or the link portion of an image, or return a fallback.
static Reference ParseLinkDetails(Subject subj, bool supportPlaceholderBrackets)
{
int n;
int sps;
int endlabel, starturl, endurl, starttitle, endtitle, endall;
string url, title;
endlabel = subj.Position;
var c = peek_char(subj);
if (c == '(' &&
((sps = Scanner.scan_spacechars(subj.Buffer, subj.Position + 1, subj.Length)) > -1) &&
((n = Scanner.scan_link_url(subj.Buffer, subj.Position + 1 + sps, subj.Length)) > -1))
{
// try to parse an explicit link:
starturl = subj.Position + 1 + sps; // after (
endurl = starturl + n;
starttitle = endurl + Scanner.scan_spacechars(subj.Buffer, endurl, subj.Length);
// ensure there are spaces btw url and title
endtitle = (starttitle == endurl) ? starttitle :
starttitle + Scanner.scan_link_title(subj.Buffer, starttitle, subj.Length);
endall = endtitle + Scanner.scan_spacechars(subj.Buffer, endtitle, subj.Length);
if (endall < subj.Length && subj.Buffer[endall] == ')')
{
subj.Position = endall + 1;
url = subj.Buffer.Substring(starturl, endurl - starturl);
url = CleanUrl(url);
title = subj.Buffer.Substring(starttitle, endtitle - starttitle);
title = CleanTitle(title);
return(new Reference()
{
Title = title, Url = url
});
}
}
else if (c == '[')
{
var label = ParseReferenceLabel(subj);
if (label != null)
{
if (label.Value.Length == 0)
{
return(Reference.SelfReference);
}
var details = LookupReference(subj.DocumentData, label.Value);
if (details != null)
{
return(details);
}
// rollback the subject but return InvalidReference so that the caller knows not to
// parse 'foo' from [foo][bar].
subj.Position = endlabel;
return(Reference.InvalidReference);
}
}
// rollback the subject position because didn't match anything.
subj.Position = endlabel;
if (supportPlaceholderBrackets)
{
return(new Reference()
{
Url = subj.Buffer.Substring(subj.LastPendingInline.StartPosition, subj.Position - subj.LastPendingInline.StartPosition - 1),
IsPlaceholder = true
});
}
else
{
return(null);
}
}
// Process one line at a time, modifying a block.
// Returns 0 if successful. curptr is changed to point to
// the currently open block.
public static void IncorporateLine(LineInfo line, ref Block curptr)
{
var ln = line.Line;
Block last_matched_container;
// offset is the char position in the line
var offset = 0;
// column is the virtual position in the line that takes TAB expansion into account
var column = 0;
// the adjustment to the virtual position `column` that points to the number of spaces from the TAB that have not been included in any indent.
var remainingSpaces = 0;
// the char position of the first non-space char
int first_nonspace;
// the virtual position of the first non-space chart, that includes TAB expansion
int first_nonspace_column;
int matched;
int i;
ListData data;
bool all_matched = true;
Block cur = curptr;
var blank = false;
char curChar;
int indent;
// container starts at the document root.
var container = cur.Top;
// for each containing block, try to parse the associated line start.
// bail out on failure: container will point to the last matching block.
while (container.LastChild != null && container.LastChild.IsOpen)
{
container = container.LastChild;
FindFirstNonspace(ln, offset, column, out first_nonspace, out first_nonspace_column, out curChar);
indent = first_nonspace_column - column + remainingSpaces;
blank = curChar == '\n';
switch (container.Tag)
{
case BlockTag.BlockQuote:
{
if (indent <= 3 && curChar == '>')
{
AdvanceOffset(ln, indent + 1, true, ref offset, ref column, ref remainingSpaces);
AdvanceOptionalSpace(ln, ref offset, ref column, ref remainingSpaces);
}
else
{
all_matched = false;
}
break;
}
case BlockTag.ListItem:
{
if (indent >= container.ListData.MarkerOffset + container.ListData.Padding)
{
AdvanceOffset(ln, container.ListData.MarkerOffset + container.ListData.Padding, true, ref offset, ref column, ref remainingSpaces);
}
else if (blank && container.FirstChild != null)
{
// if container->first_child is NULL, then the opening line
// of the list item was blank after the list marker; in this
// case, we are done with the list item.
AdvanceOffset(ln, first_nonspace - offset, false, ref offset, ref column, ref remainingSpaces);
}
else
{
all_matched = false;
}
break;
}
case BlockTag.IndentedCode:
{
if (indent >= CODE_INDENT)
{
AdvanceOffset(ln, CODE_INDENT, true, ref offset, ref column, ref remainingSpaces);
}
else if (blank)
{
AdvanceOffset(ln, first_nonspace - offset, false, ref offset, ref column, ref remainingSpaces);
}
else
{
all_matched = false;
}
break;
}
case BlockTag.AtxHeading:
case BlockTag.SetextHeading:
{
// a heading can never contain more than one line
all_matched = false;
if (blank)
{
container.IsLastLineBlank = true;
}
break;
}
case BlockTag.FencedCode:
{
// -1 means we've seen closer
if (container.FencedCodeData.FenceLength == -1)
{
all_matched = false;
if (blank)
{
container.IsLastLineBlank = true;
}
}
else
{
// skip optional spaces of fence offset
i = container.FencedCodeData.FenceOffset;
while (i > 0 && ln[offset] == ' ')
{
offset++;
column++;
i--;
}
}
break;
}
case BlockTag.HtmlBlock:
{
// all other block types can accept blanks
if (blank && container.HtmlBlockType >= HtmlBlockType.InterruptingBlock)
{
container.IsLastLineBlank = true;
all_matched = false;
}
break;
}
case BlockTag.Paragraph:
{
if (blank)
{
container.IsLastLineBlank = true;
all_matched = false;
}
break;
}
}
if (!all_matched)
{
container = container.Parent; // back up to last matching block
break;
}
}
last_matched_container = container;
// check to see if we've hit 2nd blank line, break out of list:
if (blank && container.IsLastLineBlank)
{
BreakOutOfLists(ref container, line);
}
var maybeLazy = cur.Tag == BlockTag.Paragraph;
// unless last matched container is code block, try new container starts:
while (container.Tag != BlockTag.FencedCode &&
container.Tag != BlockTag.IndentedCode &&
container.Tag != BlockTag.HtmlBlock)
{
FindFirstNonspace(ln, offset, column, out first_nonspace, out first_nonspace_column, out curChar);
indent = first_nonspace_column - column;
blank = curChar == '\n';
var indented = indent >= CODE_INDENT;
if (!indented && curChar == '>')
{
AdvanceOffset(ln, first_nonspace + 1 - offset, false, ref offset, ref column, ref remainingSpaces);
AdvanceOptionalSpace(ln, ref offset, ref column, ref remainingSpaces);
container = CreateChildBlock(container, line, BlockTag.BlockQuote, first_nonspace);
}
else if (!indented && curChar == '#' && 0 != (matched = Scanner.scan_atx_heading_start(ln, first_nonspace, ln.Length, out i)))
{
AdvanceOffset(ln, first_nonspace + matched - offset, false, ref offset, ref column, ref remainingSpaces);
container = CreateChildBlock(container, line, BlockTag.AtxHeading, first_nonspace);
container.Heading = new HeadingData(i);
}
else if (!indented && (curChar == '`' || curChar == '~') && 0 != (matched = Scanner.scan_open_code_fence(ln, first_nonspace, ln.Length)))
{
container = CreateChildBlock(container, line, BlockTag.FencedCode, first_nonspace);
container.FencedCodeData = new FencedCodeData();
container.FencedCodeData.FenceChar = curChar;
container.FencedCodeData.FenceLength = matched;
container.FencedCodeData.FenceOffset = first_nonspace - offset;
AdvanceOffset(ln, first_nonspace + matched - offset, false, ref offset, ref column, ref remainingSpaces);
}
else if (!indented && curChar == '<' &&
(0 != (matched = (int)Scanner.scan_html_block_start(ln, first_nonspace, ln.Length)) ||
(container.Tag != BlockTag.Paragraph && 0 != (matched = (int)Scanner.scan_html_block_start_7(ln, first_nonspace, ln.Length)))
))
{
container = CreateChildBlock(container, line, BlockTag.HtmlBlock, first_nonspace);
container.HtmlBlockType = (HtmlBlockType)matched;
// note, we don't adjust offset because the tag is part of the text
}
else if (!indented && container.Tag == BlockTag.Paragraph && (curChar == '=' || curChar == '-') &&
0 != (matched = Scanner.scan_setext_heading_line(ln, first_nonspace, ln.Length)))
{
container.Tag = BlockTag.SetextHeading;
container.Heading = new HeadingData(matched);
AdvanceOffset(ln, ln.Length - 1 - offset, false, ref offset, ref column, ref remainingSpaces);
}
else if (!indented &&
!(container.Tag == BlockTag.Paragraph && !all_matched) &&
0 != (Scanner.scan_thematic_break(ln, first_nonspace, ln.Length)))
{
// it's only now that we know the line is not part of a setext heading:
container = CreateChildBlock(container, line, BlockTag.ThematicBreak, first_nonspace);
Finalize(container, line);
container = container.Parent;
AdvanceOffset(ln, ln.Length - 1 - offset, false, ref offset, ref column, ref remainingSpaces);
}
else if ((!indented || container.Tag == BlockTag.List) &&
0 != (matched = ParseListMarker(ln, first_nonspace, out data)))
{
// compute padding:
AdvanceOffset(ln, first_nonspace + matched - offset, false, ref offset, ref column, ref remainingSpaces);
var prevOffset = offset;
var prevColumn = column;
var prevRemainingSpaces = remainingSpaces;
while (column - prevColumn <= CODE_INDENT)
{
if (!AdvanceOptionalSpace(ln, ref offset, ref column, ref remainingSpaces))
{
break;
}
}
// i = number of spaces after marker, up to 5
if (column == prevColumn)
{
// no spaces at all
data.Padding = matched + 1;
}
else if (column - prevColumn > CODE_INDENT || ln[offset] == '\n')
{
data.Padding = matched + 1;
// too many (or none) spaces, ignoring everything but the first one
offset = prevOffset;
column = prevColumn;
remainingSpaces = prevRemainingSpaces;
AdvanceOptionalSpace(ln, ref offset, ref column, ref remainingSpaces);
}
else
{
data.Padding = matched + column - prevColumn;
}
// check container; if it's a list, see if this list item
// can continue the list; otherwise, create a list container.
data.MarkerOffset = indent;
if (container.Tag != BlockTag.List || !ListsMatch(container.ListData, data))
{
container = CreateChildBlock(container, line, BlockTag.List, first_nonspace);
container.ListData = data;
}
// add the list item
container = CreateChildBlock(container, line, BlockTag.ListItem, first_nonspace);
container.ListData = data;
}
else if (indented && !maybeLazy && !blank)
{
AdvanceOffset(ln, CODE_INDENT, true, ref offset, ref column, ref remainingSpaces);
container = CreateChildBlock(container, line, BlockTag.IndentedCode, offset);
}
else
{
break;
}
if (AcceptsLines(container.Tag))
{
// if it's a line container, it can't contain other containers
break;
}
maybeLazy = false;
}
// what remains at offset is a text line. add the text to the
// appropriate container.
FindFirstNonspace(ln, offset, column, out first_nonspace, out first_nonspace_column, out curChar);
indent = first_nonspace_column - column;
blank = curChar == '\n';
if (blank && container.LastChild != null)
{
container.LastChild.IsLastLineBlank = true;
}
// block quote lines are never blank as they start with >
// and we don't count blanks in fenced code for purposes of tight/loose
// lists or breaking out of lists. we also don't set last_line_blank
// on an empty list item.
container.IsLastLineBlank = (blank &&
container.Tag != BlockTag.BlockQuote &&
container.Tag != BlockTag.SetextHeading &&
container.Tag != BlockTag.FencedCode &&
!(container.Tag == BlockTag.ListItem &&
container.FirstChild == null &&
container.SourcePosition >= line.LineOffset));
Block cont = container;
while (cont.Parent != null)
{
cont.Parent.IsLastLineBlank = false;
cont = cont.Parent;
}
if (cur != last_matched_container &&
container == last_matched_container &&
!blank &&
cur.Tag == BlockTag.Paragraph &&
cur.StringContent.Length > 0)
{
AddLine(cur, line, ln, offset, remainingSpaces);
}
else
{ // not a lazy continuation
// finalize any blocks that were not matched and set cur to container:
while (cur != last_matched_container)
{
Finalize(cur, line);
cur = cur.Parent;
if (cur == null)
{
throw new CommonMarkException("Cannot finalize container block. Last matched container tag = " + last_matched_container.Tag);
}
}
if (container.Tag == BlockTag.IndentedCode)
{
AddLine(container, line, ln, offset, remainingSpaces);
}
else if (container.Tag == BlockTag.FencedCode)
{
if ((indent <= 3 &&
curChar == container.FencedCodeData.FenceChar) &&
(0 != Scanner.scan_close_code_fence(ln, first_nonspace, container.FencedCodeData.FenceLength, ln.Length)))
{
// if closing fence, set fence length to -1. it will be closed when the next line is processed.
container.FencedCodeData.FenceLength = -1;
}
else
{
AddLine(container, line, ln, offset, remainingSpaces);
}
}
else if (container.Tag == BlockTag.HtmlBlock)
{
AddLine(container, line, ln, offset, remainingSpaces);
if (Scanner.scan_html_block_end(container.HtmlBlockType, ln, first_nonspace, ln.Length))
{
Finalize(container, line);
container = container.Parent;
}
}
else if (blank)
{
// ??? do nothing
}
else if (container.Tag == BlockTag.AtxHeading)
{
int p = ln.Length - 1;
// trim trailing spaces
while (p >= 0 && (ln[p] == ' ' || ln[p] == '\n'))
{
p--;
}
// if string ends in #s, remove these:
while (p >= 0 && ln[p] == '#')
{
p--;
}
// there must be a space before the last hashtag
if (p < 0 || ln[p] != ' ')
{
p = ln.Length - 1;
}
AddLine(container, line, ln, first_nonspace, remainingSpaces, p - first_nonspace + 1);
Finalize(container, line);
container = container.Parent;
}
else if (AcceptsLines(container.Tag))
{
AddLine(container, line, ln, first_nonspace, remainingSpaces);
}
else if (container.Tag != BlockTag.ThematicBreak && container.Tag != BlockTag.SetextHeading)
{
// create paragraph container for line
container = CreateChildBlock(container, line, BlockTag.Paragraph, first_nonspace);
AddLine(container, line, ln, first_nonspace, remainingSpaces);
}
else
{
Utilities.Warning("Line {0} with container type {1} did not match any condition:\n\"{2}\"", line.LineNumber, container.Tag, ln);
}
curptr = container;
}
}
/// <summary>
/// Destructively unescape a string: remove backslashes before punctuation or symbol characters.
/// </summary>
/// <param name="url">The string data that will be changed by unescaping any punctuation or symbol characters.</param>
public static string Unescape(string url)
{
// remove backslashes before punctuation chars:
int searchPos = 0;
int lastPos = 0;
int match;
char c;
char[] search = new[] { '\\', '&' };
StringBuilder sb = null;
while ((searchPos = url.IndexOfAny(search, searchPos)) != -1)
{
c = url[searchPos];
if (c == '\\')
{
searchPos++;
if (url.Length == searchPos)
{
break;
}
c = url[searchPos];
if (Utilities.IsEscapableSymbol(c))
{
if (sb == null)
{
sb = new StringBuilder(url.Length);
}
sb.Append(url, lastPos, searchPos - lastPos - 1);
lastPos = searchPos;
}
}
else if (c == '&')
{
string namedEntity;
int numericEntity;
match = Scanner.scan_entity(url, searchPos, url.Length - searchPos, out namedEntity, out numericEntity);
if (match == 0)
{
searchPos++;
}
else
{
searchPos += match;
if (namedEntity != null)
{
var decoded = EntityDecoder.DecodeEntity(namedEntity);
if (decoded != null)
{
if (sb == null)
{
sb = new StringBuilder(url.Length);
}
sb.Append(url, lastPos, searchPos - match - lastPos);
sb.Append(decoded);
lastPos = searchPos;
}
}
else if (numericEntity > 0)
{
var decoded = EntityDecoder.DecodeEntity(numericEntity);
if (decoded != null)
{
if (sb == null)
{
sb = new StringBuilder(url.Length);
}
sb.Append(url, lastPos, searchPos - match - lastPos);
sb.Append(decoded);
}
else
{
if (sb == null)
{
sb = new StringBuilder(url.Length);
}
sb.Append(url, lastPos, searchPos - match - lastPos);
sb.Append('\uFFFD');
}
lastPos = searchPos;
}
}
}
}
if (sb == null)
{
return(url);
}
sb.Append(url, lastPos, url.Length - lastPos);
return(sb.ToString());
}
/// <summary>
/// Attempts to parse a list item marker (bullet or enumerated).
/// On success, returns length of the marker, and populates
/// data with the details. On failure, returns 0.
/// </summary>
/// <remarks>Original: int parse_list_marker(string ln, int pos, ref ListData dataptr)</remarks>
private static int ParseListMarker(string ln, int pos, out ListData data)
{
char c;
int startpos;
data = null;
var len = ln.Length;
startpos = pos;
c = ln[pos];
if (c == '+' || c == '•' || ((c == '*' || c == '-') && 0 == Scanner.scan_thematic_break(ln, pos, len)))
{
pos++;
if (pos == len || !Utilities.IsWhitespace(ln[pos]))
{
return(0);
}
data = new ListData();
data.BulletChar = c;
data.Start = 1;
}
else if (c >= '0' && c <= '9')
{
int start = c - '0';
while (pos < len - 1)
{
c = ln[++pos];
// We limit to 9 digits to avoid overflow, This also seems to be the limit for 'start' in some browsers.
if (c >= '0' && c <= '9' && start < 100000000)
{
start = start * 10 + (c - '0');
}
else
{
break;
}
}
if (pos >= len - 1 || (c != '.' && c != ')'))
{
return(0);
}
pos++;
if (pos == len || !Utilities.IsWhitespace(ln[pos]))
{
return(0);
}
data = new ListData();
data.ListType = ListType.Ordered;
data.BulletChar = '\0';
data.Start = start;
data.Delimiter = (c == '.' ? ListDelimiter.Period : ListDelimiter.Parenthesis);
}
else
{
return(0);
}
return(pos - startpos);
}
// Parse reference. Assumes string begins with '[' character.
// Modify refmap if a reference is encountered.
// Return 0 if no reference found, otherwise position of subject
// after reference is parsed.
public static int ParseReference(Subject subj)
{
string url = null;
string title = null;
int matchlen = 0;
int beforetitle;
var startPos = subj.Position;
// parse label:
var lab = ParseReferenceLabel(subj);
if (lab == null || lab.Value.Length > Reference.MaximumReferenceLabelLength)
{
goto INVALID;
}
// colon:
if (peek_char(subj) == ':')
{
subj.Position++;
}
else
{
goto INVALID;
}
// parse link url:
spnl(subj);
matchlen = Scanner.scan_link_url(subj.Buffer, subj.Position, subj.Length);
if (matchlen == 0)
{
goto INVALID;
}
url = subj.Buffer.Substring(subj.Position, matchlen);
url = CleanUrl(url);
subj.Position += matchlen;
// parse optional link_title
beforetitle = subj.Position;
spnl(subj);
matchlen = Scanner.scan_link_title(subj.Buffer, subj.Position, subj.Length);
if (matchlen > 0)
{
title = subj.Buffer.Substring(subj.Position, matchlen);
title = CleanTitle(title);
subj.Position += matchlen;
}
else
{
subj.Position = beforetitle;
title = string.Empty;
}
// parse final spaces and newline:
while (peek_char(subj) == ' ')
{
subj.Position++;
}
if (peek_char(subj) == '\n')
{
subj.Position++;
}
else if (peek_char(subj) != '\0')
{
goto INVALID;
}
// insert reference into refmap
AddReference(subj.ReferenceMap, lab.Value, url, title);
return(subj.Position);
INVALID:
subj.Position = startPos;
return(0);
}