C语言解析XML的算法

Question

Is there any known algorithm that can detect and separate the tags from an xml txt file and store the content into another file with matching tag details?

I've tried to hard code it but it doesn't work for all tags. Tags such as "t" ... "/t" work but tags like "a href="http://example.com"" "/a" don't work.

void get_output(){

    int i=0;
    int j=0;
    int k=0;
    int l=0;
    int m=0;
    int n=0;

    printf("\n");

    for(i=0; i<1024; i++){
        k=0;
        for(j=0; j<strlen(tags[i]); j++){

            if(tags[i][j] == '<'||tags[i][j]=='>'){
                k++;
                if(k == 4){
                    for(l = 0; tags[i][l+1] != '>'; l++){
                        printf("%c",tolower(tags[i][l+1]));
                    }

                    printf(": ");

                    for(; tags[i][l+2] != '<'; l++){
                        printf("%c", tags[i][l+2]);
                    }

                    printf("\n");
                }   
            }
        }
    }
}

I'm also trying to avoid using 3rd party libraries too.

Answer 1

Your question asks about XML, but you've tagged this as HTML – note that these are rather different beasts.

In terms of its syntax, there's nothing special about XML, and you'd parse it just like you would any other syntax; there's no special algorithm.

That is you'd use a lexer such as flex to identify a stream of tokens such as < , </ , = , strings, quotes and so on, and then a parser generator such as bison to write down the syntactical rules, and code on top of that to turn correctly-formed syntax into useful data structures (that is, what does your program actually do when it has discovered an element start-tag such as <a href='urn:foo'> ?). This is perfectly doable, but it's a non-trivial project.

As part of that, you'll acquire a very close relationship to the XML spec , and you'd be well advised to assemble a lot of test-cases, the more pathological the better. There's a lot of fine detail, and plenty of subtleties, in that spec.

A few months ago, I was working on a project which aimed to extract a subset of the content of XML files. It wasn't a full parse of the file, but I, like you, wanted to keep it simple and avoid third party libraries. After about a week of solid work, and building on a fair amount of prior experience with yacc/bison-based parsers, I realised that I had actually ended up implementing most of a generic XML parser, which was clearly going to end up reasonably robust and functional, but which was still missing a couple of parsing corner-cases, and was going to be tedious to polish. I decided that using expat wasn't such a bad idea after all, so threw away my code and made significantly more rapid progress building on that work.

Note that parsing well-formed XML is a very different proposition from parsing (often very ill-formed) HTML. Because HTML barely conforms to a grammar at all, a parser for it would have to be significantly more ad-hoc; a bison-generated parser might have considerable difficulties, unless you put some effort into smart error-recovery. You might want to look at a C-based Markdown or Wiki parser for ideas. Or try googling for tagsoup c for library suggestions (there's a well-known Java parser for wild HTML called TagSoup , and similar things in other languages tend to give it a shout-out).

If doing this without a third-party library is an intellectual exercise, then it'll be a very instructive one, and an ambitious first parser project. If not, then you'd be very well advised to exploit the considerable effort that's gone into existing libraries.