expected"))
next(x)
of xmlCData:
result = newCData(x.charData)
next(x)
of xmlEntity:
var u = entityToUtf8(x.entityName)
if u.len != 0: result = newText(u)
next(x)
of xmlEof: nil
proc parseHtml*(s: PStream, filename: string,
errors: var seq[string]): PXmlNode =
## parses the XML from stream `s` and returns a ``PXmlNode``. Every
## occured parsing error is added to the `errors` sequence.
var x: TXmlParser
open(x, s, filename, {reportComments})
next(x)
# skip the DOCTYPE:
if x.kind == xmlSpecial: next(x)
result = parse(x, errors)
if x.kind != xmlEof:
errors.add(errorMsg(x, "EOF expected"))
while x.kind != xmlEof:
var oldPos = x.bufpos # little hack to see if we made any progess
result.addNode(parse(x, errors))
if x.bufpos == oldPos:
# force progress!
next(x)
close(x)
proc parseHtml*(s: PStream): PXmlNode =
## parses the XTML from stream `s` and returns a ``PXmlNode``. All parsing
## errors are ignored.
var errors: seq[string] = @[]
result = parseHtml(s, "unknown_html_doc", errors)
proc loadHtml*(path: string, errors: var seq[string]): PXmlNode =
## Loads and parses HTML from file specified by ``path``, and returns
## a ``PXmlNode``. Every occured parsing error is added to
## the `errors` sequence.
var s = newFileStream(path, fmRead)
if s == nil: raise newException(EIO, "Unable to read file: " & path)
result = parseHtml(s, path, errors)
proc loadHtml*(path: string): PXmlNode =
## Loads and parses HTML from file specified by ``path``, and returns
## a ``PXmlNode``. All parsing errors are ignored.
var errors: seq[string] = @[]
result = loadHtml(path, errors)
when true:
nil
else:
proc checkHtmlAux(n: PXmlNode, errors: var seq[string]) =
nil
proc checkHtmlStructure*(n: PXmlNode, errors: var seq[string]) =
## checks the HTML structure after parsing for other errors like
## a ```` element within a ``
`` element.
if n == nil or n.htmlTag != tagHtml:
errors.add(" tag expected")
checkHtmlAux(n, errors)
when isMainModule:
import os
var errors: seq[string] = @[]
var x = loadHtml(paramStr(1), errors)
for e in items(errors): echo e
var f: TFile
if open(f, "test.txt", fmWrite):
f.write($x)
f.close()
else:
quit("cannot write test.txt")