CGI How-To/CGI HOW-TO.iso/chap5/5_3/input_c/parsehtm.c

C/C++ Source or Header | 1996-06-15 | 10.4 KB | 482 lines

#include <stdio.h>
#include <string.h>
#include "parsehtm.h"

static FILE *htmlFile = 0;

/* Global Dictionaries */
Dictionary endTags;
Dictionary handlerDict;

/*
 * Tag Handlers are functions that return void,
 * and take four arguements,
 * 1) the tag string, which they should alter to
 * the parsed text.
 * 2) an arg string, not containing any registered endTags
 * 3) any registered endTag
 * 4) a dictionary of data for the tag, from dictForTag()
 * The keys for tagDict are const char *,and the values are String
 *
 * The prototype for a handler is:
 * void handler(String ts,String as,String et,Dictionary td);
 */

/* This function allocates space for the global dictionaries
 * it must be called before any handlers or end tags are registered. */
void initializeHtmlParsingLibrary()
{
    endTags = dict_alloc();
    handlerDict = dict_alloc();
}

/*
 * parseHtml() provides the primary interface
 * to this library. It takes the file name
 * and returns a String containing the parsed html.
 */

String parseHtml(const char *fileName)
{
    String retVal;

    if(fileName && *fileName)
    {
        htmlFile = fopen(fileName,"r");
    }

    if(htmlFile)
    {
        retVal = mainHtmlParser((const char *) 0);
        fclose(htmlFile);
    }

    return retVal;
}

/*
 * mainHtmlParser is at the heart of the parsing code
 * This function loops over the file, stopping at the end
 * of the file, a stop character or a stop string.
 * It seperates the file into tags and plain text,
 * then calls the functions handleTag and handlePlainText
 * to parse the data.
 */
String mainHtmlParser(const char *stopStr)
{
    char c,stopChar;
    String tmpBuffer = 0;
    String mainBuffer = 0;
    int inTag = 0;

    /* Initialize the buffers. */
    tmpBuffer = string_alloc(64);
    mainBuffer = string_alloc(4096);

    /* If we got a stop string, see if
     * it is really a character. This allows
     * the function to have dual functionality off
     * of only one argument.
     */
    if(stopStr)
    {
        if(*stopStr && (strlen(stopStr) == 1))
        {
            stopChar = stopStr[0];
        }
        else
        {
            stopChar = '\0';
        }
    }

    /* Loop to the end of the file. */
    while(!feof(htmlFile))
    {
        /* Get a character at each step of the loop */
        c = fgetc(htmlFile);

        /* See what character this is*/

        if(c == '>')/* end of a tag */
        {
            if(!inTag)/*we are not in a tag, this is an error*/
            {
                exit(1);
            }

            /* Close the tag */
            inTag = 0;

            string_appendChar(tmpBuffer,c);

            /* See if this tag is the stop string */

            if(stopStr && !strcmp(tmpBuffer->string,stopStr))
            {
                return mainBuffer;
            }
            else
            {
                /* Handle the tag we just found */
                handleTag(tmpBuffer);

                /* Add the parsed text to the main buffer */
                string_appendString(mainBuffer,tmpBuffer->string);

                /* Reset the tmp buffer */
                string_empty(tmpBuffer);
            }
        }
        else if(c == '<')/* Start of a tag */
        {
            if(inTag)/*we are already in a tag, this is an error*/
            {
                exit(1);
            }

            inTag = 1;

            /* If we have some data, it is plain text,
                * add it to the main buffer.
                */
            if(tmpBuffer->string[0])
            {
                handlePlainText(tmpBuffer);

                /* Add the parsed text to the main buffer */
                string_appendString(mainBuffer,tmpBuffer->string);

                /* Reset the tmp buffer */
                string_empty(tmpBuffer);

            }

            /* Reset the tmp buffer */
            string_setStringValue(tmpBuffer,"<");
        }
        else if(c == stopChar)/* Found the stop char */
        {
            string_appendChar(tmpBuffer,c);

            /* Add the parsed text to the main buffer */
            string_appendString(mainBuffer,tmpBuffer->string);

            break;
        }
        else/* normal character */
        {
            string_appendChar(tmpBuffer,c);
        }
    }

    /* free the tmp buffer */
    string_free(tmpBuffer);

    /* Return the parsed text */
    return mainBuffer;
}

/*
 * Handle tag is called by mainHtmlParser.
 * This function looks for a tag handler and calls
 * it with the appropriate arguments.
 */
void handleTag(String tagString)
{
    Dictionary tagDict = 0;
    String argString = 0;
    String endTag = 0;
    void (*handler)();
    String tag;

    /* Create a dictionary for the tag
     * This will make it easier to access
     * information about the tag
     */
    tagDict = dictForTag(tagString);

    /* The tag dictionary should have a TAG value */
    if(tag = (String)dict_valueForKey(tagDict,"TAG"))
    {
        if(tag->string)
        {
            /* See if there is an endtag or a handler registered*/
            endTag = dict_valueForKey(endTags, tag->string);

            handler = dict_valueForKey(handlerDict, tag->string);
        }
    }

    /* If there is a handler, use it */
    if(handler)
    {
        /* If there is an end tag, parse up to it
         * be sure to remove the end tag from the body text.
         */
        if(endTag && endTag->string)
        {
            char * finalPointy = 0;

            argString = mainHtmlParser(endTag->string);

            /* Get rid of the end tag */

            if(endTag->string[0] == '<')/* this is a real tag */
            {
                finalPointy = strrchr(argString->string,'<');

                /* End the string at the final <, thus removing the end tag */
                if(finalPointy) *finalPointy = '\0';
            }
        }

        /* Call the handler */
        handler(tagString,argString,endTag,tagDict);
    }
    else /* No handler, treat this as plain text */
    {
        handlePlainText(tagString);
    }

    /* Clean up */
    dict_freeWithData(tagDict,string_free);
    string_free(argString);
}

void handlePlainText(String plainText)
{
    void (*handler)();

    /* See if there is a default handler */
    handler = dict_valueForKey(handlerDict,"DEFAULT");

    /* If there is a default handler, use it*/
    if(handler)
    {
        handler(plainText,0,0,0);
    }
}

/*
 * Convienence function that searches a c string for a tag
 * it returns a pointer into the string, after the tag.
 * This return value is used for followup calls.
 */

char * nextTag(char *start, String tagString)
{
    int inQuote = 0;

    if(start && *start)
    {
        /* Move past any starting delimiters */

        while((*start == ' ')||(*start == '<'))
        {
            start++;
        }

        string_empty(tagString);

        while(*start && (((*start != ' ')&&(*start != '>'))||inQuote))
        {
            if(*start == '"')/*Look for quotes, dont include them*/
            {
                if(inQuote) inQuote = 0;
                else inQuote=1;
            }
            else
            {
                string_appendChar(tagString,*start);
            }
            start++;
        }

        start++;/*Move to next character for subsequent next search*/
    }

    return start;
}

/*
 * dictForTag is a convience function
 * it takes a tag string, and returns
 * a dictionary of key value pairs.
 * One key should be TAG. All of the keys
 * are capitalized.
 */
Dictionary dictForTag(String tagString)
{
    Dictionary tagDict = 0;
    String tag;
    String tmpString;
    char *slide = (char *)0;
    char *key,*value;
    String dataToAdd;

    /* Make the dictionary */
    tagDict = dict_alloc();

    /* If there is no tag string, return the empty dict */
    if(!tagString || !(tagString->string)) return tagDict;

    /* Copy the tag string, since strtok will alter data */
    tag = string_alloc(strlen(tagString->string) + 1);
    string_setStringValue(tag, tagString->string);

    /* Make a temporary string, to help with our work */
    tmpString = string_alloc(strlen(tagString->string) + 1);

    /*Get the first token*/
    slide = nextTag(tag->string,tmpString);

    /* If the current token starts with a <, get rid of the < */
    if(tmpString->string && (tmpString->string[0] == '<'))
    {
        string_crop(tmpString,1);
    }

    /* Capitalize the tag string, and add it to the dict */
    if(tmpString->string)
    {
        string_toUpper(tmpString);

        dataToAdd = string_alloc(strlen(tmpString->string) +1);
        string_setStringValue(dataToAdd, tmpString->string);

        dict_setValueForKey(tagDict,"TAG", dataToAdd);
    }

    /* Loop over the tag string, breaking it into tokens */
    do
    {

        /*Get the next token*/
        slide = nextTag(slide,tmpString);

        /* If we got a token */
        if(tmpString->string)
        {
            /* See if it is binary or unary */
            if(value = strchr(tmpString->string,'='))
            {
                key = tmpString->string;

                *value = '\0'; /* set the = to a null */
                value++; /* Move value past the old = */

                string_toUpper(tmpString);/* Will only get the key */

                /* Clean up any trailing or leading quotes */
                if(value[strlen(value)-1] == '\"')
                {
                    value[strlen(value)-1] = '\0';
                }

                if(value[0] == '\"')
                {
                    value++;
                }

                /* Add the attribute to the tag dict */
                dataToAdd = string_alloc(strlen(value) +1);
                string_setStringValue(dataToAdd,value);

                dict_setValueForKey(tagDict, key, dataToAdd);
            }
            else
            {
                if(tmpString->string[strlen(tmpString->string)-1] == '\"')
                {
                    string_chop(tmpString,1);
                }

                if(tmpString->string[0] == '\"')
                {
                    string_crop(tmpString,1);
                }

                string_toUpper(tmpString);

                /* Add the unary attribute to the tag dict */

                dataToAdd = string_alloc(strlen(tmpString->string) +1);
                string_setStringValue(dataToAdd, tmpString->string);

                dict_setValueForKey(tagDict, tmpString->string, dataToAdd);
            }
        }

    }while (slide && *slide);

    string_free(tag);
    string_free(tmpString);

    return tagDict;
}

/*
 * stringForTagDict is provided for programmers
 * writing handler functions.
 * This function takes a tag dictionary
 * and returns the cooresponding tag string.
 * the string is allocated, and is the callers
 * responsibility.
 */
String stringForTagDict(Dictionary tagDict)
{
    String tagString;
    DictState iterator;
    const char *key = 0;
    String value = 0;

    /* Allocate the string */
    tagString = string_alloc(64);

    /* Make sure we have tag data */
    if(tagDict)
    {
        /* Add the tag to the string */
        value = dict_valueForKey(tagDict,"TAG");

        if(value && value->string)
        {
            string_appendString(tagString,"<");
            string_appendString(tagString,value->string);

            /* Loop over the dictionary, adding tag attributes */
            iterator = dict_initState(tagDict);

            while(dict_nextState(&iterator))
            {
                key = (const char *) iterator.curNode->key;
                value = (String) iterator.curNode->value;

                if(strcmp(key,"TAG") != 0)/* Ignore the tag */
                {
                    if(!strcmp(key,value->string))/* This is a unary key */
                    {
                        string_appendString(tagString," ");
                        string_appendString(tagString,key);
                    }
                    else /* Binary attribute */
                    {
                        string_appendString(tagString," ");
                        string_appendString(tagString,key);
                        string_appendString(tagString,"=\"");
                        string_appendString(tagString,value->string);
                        string_appendString(tagString,"\"");
                    }
                }
            }

            /* Close the tag string */
            string_appendString(tagString,">");
        }
    }
    /* Return the string */
    return tagString;
}