W3cubDocs

/Nim

Module parsexml

This module implements a simple high performance XML / HTML parser. The only encoding that is supported is UTF-8. The parser has been designed to be somewhat error correcting, so that even most "wild HTML" found on the web can be parsed with it. Note: This parser does not check that each <tag> has a corresponding </tag>! These checks have do be implemented by the client code for various reasons:

Old HTML contains tags that have no end tag: <br> for example.
HTML tags are case insensitive, XML tags are case sensitive. Since this library can parse both, only the client knows which comparison is to be used.
Thus the checks would have been very difficult to implement properly with little benefit, especially since they are simple to implement in the client. The client should use the errorMsgExpected proc to generate a nice error message that fits the other error messages this library creates.

Example 1: Retrieve HTML title

The file examples/htmltitle.nim demonstrates how to use the XML parser to accomplish a simple task: To determine the title of an HTML document.

# Example program to show the parsexml module
# This program reads an HTML file and writes its title to stdout.
# Errors and whitespace are ignored.

import os, streams, parsexml, strutils

if paramCount() < 1:
  quit("Usage: htmltitle filename[.html]")

var filename = addFileExt(paramStr(1), "html")
var s = newFileStream(filename, fmRead)
if s == nil: quit("cannot open the file " & filename)
var x: XmlParser
open(x, s, filename)
while true:
  x.next()
  case x.kind
  of xmlElementStart:
    if cmpIgnoreCase(x.elementName, "title") == 0:
      var title = ""
      x.next()  # skip "<title>"
      while x.kind == xmlCharData:
        title.add(x.charData)
        x.next()
      if x.kind == xmlElementEnd and cmpIgnoreCase(x.elementName, "title") == 0:
        echo("Title: " & title)
        quit(0) # Success!
      else:
        echo(x.errorMsgExpected("/title"))
  
  of xmlEof: break # end of file reached
  else: discard # ignore other events

x.close()
quit("Could not determine title!")

Example 2: Retrieve all HTML links

The file examples/htmlrefs.nim demonstrates how to use the XML parser to accomplish another simple task: To determine all the links an HTML document contains.

# Example program to show the new parsexml module
# This program reads an HTML file and writes all its used links to stdout.
# Errors and whitespace are ignored.

import os, streams, parsexml, strutils

proc `=?=` (a, b: string): bool =
  # little trick: define our own comparator that ignores case
  return cmpIgnoreCase(a, b) == 0

if paramCount() < 1:
  quit("Usage: htmlrefs filename[.html]")

var links = 0 # count the number of links
var filename = addFileExt(paramStr(1), "html")
var s = newFileStream(filename, fmRead)
if s == nil: quit("cannot open the file " & filename)
var x: XmlParser
open(x, s, filename)
next(x) # get first event
block mainLoop:
  while true:
    case x.kind
    of xmlElementOpen:
      # the <a href = "xyz"> tag we are interested in always has an attribute,
      # thus we search for ``xmlElementOpen`` and not for ``xmlElementStart``
      if x.elementName =?= "a":
        x.next()
        if x.kind == xmlAttribute:
          if x.attrKey =?= "href":
            var link = x.attrValue
            inc(links)
            # skip until we have an ``xmlElementClose`` event
            while true:
              x.next()
              case x.kind
              of xmlEof: break mainLoop
              of xmlElementClose: break
              else: discard
            x.next() # skip ``xmlElementClose``
            # now we have the description for the ``a`` element
            var desc = ""
            while x.kind == xmlCharData:
              desc.add(x.charData)
              x.next()
            echo(desc & ": " & link)
      else:
        x.next()
    of xmlEof: break # end of file reached
    of xmlError:
      echo(errorMsg(x))
      x.next()
    else: x.next() # skip other events

echo($links & " link(s) found!")
x.close()

Imports

hashes, strutils, lexbase, streams, unicode, os

Types

XmlEventKind = enum xmlError, ## an error occurred during parsing xmlEof, ## end of file reached xmlCharData, ## character data xmlWhitespace, ## whitespace has been parsed xmlComment, ## a comment has been parsed xmlPI, ## processing instruction (``<?name something ?>``) xmlElementStart, ## ``<elem>`` xmlElementEnd, ## ``</elem>`` xmlElementOpen, ## ``<elem xmlAttribute, ## ``key = "value"`` pair xmlElementClose, ## ``>`` xmlCData, ## ``<![CDATA[`` ... data ... ``]]>`` xmlEntity, ## &entity; xmlSpecial ## ``<! ... data ... >``: enumation of all events that may occur when parsing
XmlErrorKind = enum errNone, ## no error errEndOfCDataExpected, ## ``]]>`` expected errNameExpected, ## name expected errSemicolonExpected, ## ``;`` expected errQmGtExpected, ## ``?>`` expected errGtExpected, ## ``>`` expected errEqExpected, ## ``=`` expected errQuoteExpected, ## ``"`` or ``'`` expected errEndOfCommentExpected ## ``-->`` expected: enumeration that lists all errors that can occur
XmlParseOption = enum reportWhitespace, ## report whitespace reportComments ## report comments: options for the XML parser
XmlParser = object of BaseLexer a, b, c: string kind: XmlEventKind err: XmlErrorKind state: ParserState cIsEmpty: bool filename: string options: set[XmlParseOption]: the parser object.

Procs

proc open(my: var XmlParser; input: Stream; filename: string; options: set[XmlParseOption] = {}) {...}{.raises: [Exception], tags: [ReadIOEffect].}: initializes the parser with an input stream. Filename is only used for nice error messages. The parser's behaviour can be controlled by the options parameter: If options contains reportWhitespace a whitespace token is reported as an xmlWhitespace event. If options contains reportComments a comment token is reported as an xmlComment event.
proc close(my: var XmlParser) {...}{.inline, raises: [Exception], tags: [].}: closes the parser my and its associated input stream.
proc kind(my: XmlParser): XmlEventKind {...}{.inline, raises: [], tags: [].}: returns the current event type for the XML parser
proc rawData(my: XmlParser): string {...}{.inline, raises: [], tags: [].}: returns the underlying 'data' string by reference. This is only used for speed hacks.
proc rawData2(my: XmlParser): string {...}{.inline, raises: [], tags: [].}: returns the underlying second 'data' string by reference. This is only used for speed hacks.
proc getColumn(my: XmlParser): int {...}{.inline, raises: [], tags: [].}: get the current column the parser has arrived at.
proc getLine(my: XmlParser): int {...}{.inline, raises: [], tags: [].}: get the current line the parser has arrived at.
proc getFilename(my: XmlParser): string {...}{.inline, raises: [], tags: [].}: get the filename of the file that the parser processes.
proc errorMsg(my: XmlParser): string {...}{.raises: [ValueError], tags: [].}: returns a helpful error message for the event xmlError
proc errorMsgExpected(my: XmlParser; tag: string): string {...}{.raises: [ValueError], tags: [].}: returns an error message "<tag> expected" in the same format as the other error messages
proc errorMsg(my: XmlParser; msg: string): string {...}{.raises: [ValueError], tags: [].}: returns an error message with text msg in the same format as the other error messages
proc next(my: var XmlParser) {...}{.raises: [Exception], tags: [ReadIOEffect].}: retrieves the first/next event. This controls the parser.

Templates

template charData(my: XmlParser): string: returns the character data for the events: xmlCharData, xmlWhitespace, xmlComment, xmlCData, xmlSpecial Raises an assertion in debug mode if my.kind is not one of those events. In release mode, this will not trigger an error but the value returned will not be valid.
template elementName(my: XmlParser): string: returns the element name for the events: xmlElementStart, xmlElementEnd, xmlElementOpen Raises an assertion in debug mode if my.kind is not one of those events. In release mode, this will not trigger an error but the value returned will not be valid.
template entityName(my: XmlParser): string: returns the entity name for the event: xmlEntity Raises an assertion in debug mode if my.kind is not xmlEntity. In release mode, this will not trigger an error but the value returned will not be valid.
template attrKey(my: XmlParser): string: returns the attribute key for the event xmlAttribute Raises an assertion in debug mode if my.kind is not xmlAttribute. In release mode, this will not trigger an error but the value returned will not be valid.
template attrValue(my: XmlParser): string: returns the attribute value for the event xmlAttribute Raises an assertion in debug mode if my.kind is not xmlAttribute. In release mode, this will not trigger an error but the value returned will not be valid.
template piName(my: XmlParser): string: returns the processing instruction name for the event xmlPI Raises an assertion in debug mode if my.kind is not xmlPI. In release mode, this will not trigger an error but the value returned will not be valid.
template piRest(my: XmlParser): string: returns the rest of the processing instruction for the event xmlPI Raises an assertion in debug mode if my.kind is not xmlPI. In release mode, this will not trigger an error but the value returned will not be valid.