inc/htmlPage.h File Reference

#include "dystring.h"

Include dependency graph for htmlPage.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Data Structures
struct	htmlStatus
struct	htmlCookie
struct	htmlAttribute
struct	htmlTag
struct	htmlFormVar
struct	htmlForm
struct	htmlPage
Functions
void	htmlStatusFree (struct htmlStatus **pStatus)
void	htmlStatusFreeList (struct htmlStatus **pList)
htmlStatus *	htmlStatusParse (char **pText)
void	htmlCookieFree (struct htmlCookie **pCookie)
void	htmlCookieFreeList (struct htmlCookie **pList)
htmlCookie *	htmlCookieFileRead (char *fileName)
void	htmlAttributeFree (struct htmlAttribute **pAttribute)
void	htmlAttributeFreeList (struct htmlAttribute **pList)
char *	htmlTagAttributeVal (struct htmlPage *page, struct htmlTag *tag, char *name, char *defaultVal)
char *	htmlTagAttributeNeeded (struct htmlPage *page, struct htmlTag *tag, char *name)
void	htmlTagFree (struct htmlTag **pTag)
void	htmlTagFreeList (struct htmlTag **pList)
void	htmlFormVarFree (struct htmlFormVar **pVar)
void	htmlFormVarFreeList (struct htmlFormVar **pList)
void	htmlFormVarPrint (struct htmlFormVar *var, FILE *f, char *prefix)
void	htmlFormFree (struct htmlForm **pForm)
void	htmlFormFreeList (struct htmlForm **pList)
void	htmlFormPrint (struct htmlForm *form, FILE *f)
char *	htmlFormCgiVars (struct htmlPage *page, struct htmlForm *form, char *buttonName, char *buttonVal, struct dyString *dyHeader)
htmlForm *	htmlFormGet (struct htmlPage *page, char *name)
htmlFormVar *	htmlFormVarGet (struct htmlForm *form, char *name)
void	htmlFormVarSet (struct htmlForm *form, char *name, char *val)
htmlFormVar *	htmlPageGetVar (struct htmlPage *page, struct htmlForm *form, char *name)
void	htmlPageSetVar (struct htmlPage *page, struct htmlForm *form, char *name, char *val)
void	htmlPageFree (struct htmlPage **pPage)
void	htmlPageFreeList (struct htmlPage **pList)
char *	htmlExpandUrl (char *base, char *url)
char *	htmlNextCrLfLine (char **pS)
slName *	htmlPageScanAttribute (struct htmlPage *page, char *tagName, char *attribute)
slName *	htmlPageLinks (struct htmlPage *page)
void	htmlPageFormOrAbort (struct htmlPage *page)
void	htmlPageValidateOrAbort (struct htmlPage *page)
char *	htmlSlurpWithCookies (char *url, struct htmlCookie *cookies)
htmlPage *	htmlPageParse (char *url, char *fullText)
htmlPage *	htmlPageParseOk (char *url, char *fullText)
htmlPage *	htmlPageParseNoHead (char *url, char *htmlText)
htmlPage *	htmlPageFromForm (struct htmlPage *origPage, struct htmlForm *form, char *buttonName, char *buttonVal)
htmlPage *	htmlPageGetWithCookies (char *url, struct htmlCookie *cookies)
htmlPage *	htmlPageGet (char *url)
htmlPage *	htmlPageForwarded (char *url, struct htmlCookie *cookies)
htmlPage *	htmlPageForwardedNoAbort (char *url, struct htmlCookie *cookies)

---

Function Documentation

void htmlAttributeFree

(

struct htmlAttribute **

pAttribute

)

Definition at line 124 of file htmlPage.c.

References freeMem(), freez(), htmlAttribute::name, and htmlAttribute::val.

Referenced by htmlAttributeFreeList().

{
struct htmlAttribute *att = *pAttribute;
if (att != NULL)
    {
    freeMem(att->name);
    freeMem(att->val);
    freez(pAttribute);
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlAttributeFreeList

(

struct htmlAttribute **

pList

)

Definition at line 136 of file htmlPage.c.

References htmlAttributeFree(), and htmlAttribute::next.

Referenced by htmlTagFree().

{
struct htmlAttribute *el, *next;

for (el = *pList; el != NULL; el = next)
    {
    next = el->next;
    htmlAttributeFree(&el);
    }
*pList = NULL;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct htmlCookie* htmlCookieFileRead

(

char *

fileName

)

[read]

Definition at line 80 of file htmlPage.c.

References AllocVar, cloneString(), errAbort(), lineFile::fileName, lineFileClose(), lineFileNextReal(), lineFileOpen(), lineFile::lineIx, nextWord(), skipLeadingSpaces(), slAddHead, slReverse(), and TRUE.

{
struct lineFile *lf = lineFileOpen(fileName, TRUE);
struct htmlCookie *list = NULL, *cookie;
char *line, *word;
while (lineFileNextReal(lf, &line))
    {
    word = nextWord(&line);
    line = skipLeadingSpaces(line);
    if (line == NULL)
        errAbort("Missing cookie value line %d of %s", lf->lineIx, lf->fileName);
    AllocVar(cookie);
    cookie->name = cloneString(word);
    cookie->value = cloneString(line);
    slAddHead(&list, cookie);
    }
lineFileClose(&lf);
slReverse(&list);
return list;
}

Here is the call graph for this function:

void htmlCookieFree

(

struct htmlCookie **

pCookie

)

Definition at line 52 of file htmlPage.c.

References htmlCookie::domain, htmlCookie::expires, freeMem(), freez(), htmlCookie::name, htmlCookie::path, and htmlCookie::value.

Referenced by htmlCookieFreeList().

{
struct htmlCookie *cookie = *pCookie;
if (cookie != NULL)
    {
    freeMem(cookie->name);
    freeMem(cookie->value);
    freeMem(cookie->domain);
    freeMem(cookie->path);
    freeMem(cookie->expires);
    freez(pCookie);
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlCookieFreeList

(

struct htmlCookie **

pList

)

Definition at line 67 of file htmlPage.c.

References htmlCookieFree(), and htmlCookie::next.

Referenced by htmlPageFree().

{
struct htmlCookie *el, *next;

for (el = *pList; el != NULL; el = next)
    {
    next = el->next;
    htmlCookieFree(&el);
    }
*pList = NULL;
}

Here is the call graph for this function:

Here is the caller graph for this function:

char* htmlExpandUrl	(	char *	base,
		char *	url
	)

Definition at line 1090 of file htmlPage.c.

References asciiEntityDecode(), cloneString(), cloneStringZ(), dyStringAppend(), dyStringAppendC(), dyStringAppendN(), dyStringCannibalize(), dyStringNew, expandRelativePath(), freez(), lastChar(), startsWith(), and dyString::string.

Referenced by htmlPageFromForm().

{
struct dyString *dy = NULL;
char *hostName, *pastHostName;

/* some mailto: have SGML char encoding, e.g a to hide from spambots */
url = cloneString(url); /* Clone because asciiEntityDecode may modify it. */
asciiEntityDecode(url, url, strlen(url));

/* In easiest case URL is actually absolute and begins with
 * protocol.  Just return clone of url. */
if (startsWith("http:", url) || startsWith("https:", url))
    return url;

/* If it's got a colon, but no http or https, then it's some
 * protocol we don't understand, like a mailto.  Just return NULL. */
if (strchr(url, ':') != NULL)
    {
    freez(&url);
    return NULL;
    }

/* Figure out first character past host name. Load up
 * return string with protocol (if any) and host name. */
dy = dyStringNew(256);
if (startsWith("http:", base) || startsWith("https:", base))
    hostName = (strchr(base, ':') + 3);
else
    hostName = base;
pastHostName = strchr(hostName, '/');
if (pastHostName == NULL)
    pastHostName = hostName + strlen(hostName);
dyStringAppendN(dy, base, pastHostName - base);

/* Add url to return string after host name. */
if (startsWith("/", url))       /* New URL is absolute, just append to hostName */
    {
    dyStringAppend(dy, url);
    }
else
    {
    char *curDir = pastHostName;
    char *endDir;
    if (curDir[0] == '/')
        curDir += 1;
    dyStringAppendC(dy, '/');
    endDir = strrchr(curDir, '/');
    if (endDir == NULL)
        endDir = curDir;
    if (startsWith("../", url))
        {
        char *dir = cloneStringZ(curDir, endDir-curDir);
        char *path = expandRelativePath(dir, url);
        if (path != NULL)
             {
             dyStringAppend(dy, path);
             }
        freez(&dir);
        freez(&path);
        }
    else
        {
        dyStringAppendN(dy, curDir, endDir-curDir);
        if (lastChar(dy->string) != '/')
            dyStringAppendC(dy, '/');
        dyStringAppend(dy, url);
        }
    }
freez(&url);
return dyStringCannibalize(&dy);
}

Here is the call graph for this function:

Here is the caller graph for this function:

char* htmlFormCgiVars	(	struct htmlPage *	page,
		struct htmlForm *	form,
		char *	buttonName,
		char *	buttonVal,
		struct dyString *	dyHeader
	)

Definition at line 1262 of file htmlPage.c.

References appendCgiVar(), appendMimeVar(), htmlFormVar::curVal, htmlPage::forms, isMimeEncoded(), htmlFormVar::name, newDyString(), htmlFormVar::next, safef(), sameWord, htmlFormVar::tagName, TRUE, htmlFormVar::type, and htmlForm::vars.

Referenced by htmlPageFromForm().

{
struct dyString *dy = newDyString(0);
struct htmlFormVar *var;
boolean isMime = isMimeEncoded(form);
int mimeParts = 0;
char boundary[256];

while(TRUE)
    {
    if (isMime)
        {
        /* choose a new string for the boundary */
        /* Set initial seed */
        int i = 0;
        safef(boundary,sizeof(boundary),"%s", "---------");
        srand( (unsigned)time( NULL ) );
        for(i=strlen(boundary);i<41;++i)
            {
            int r = (int) 26 * (rand() / (RAND_MAX + 1.0));
            boundary[i] = r+'A';
            }
        boundary[i] = 0;
        }

    if (form == NULL)
        form = page->forms;
    if (buttonName != NULL && !isMime)
        appendCgiVar(dy, buttonName, buttonVal);
    for (var = form->vars; var != NULL; var = var->next)
        {
        if (sameWord(var->tagName, "SELECT") || 
            sameWord(var->tagName, "TEXTAREA") || 
            (var->type != NULL &&
            ((sameWord(var->type, "RADIO") || sameWord(var->type, "TEXTBOX")
            || sameWord(var->type, "PASSWORD") || sameWord(var->type, "HIDDEN")
            || sameWord(var->type, "TEXT") || sameWord(var->type, "FILE")))))
            {
            char *val = var->curVal;
            if (val == NULL)
                val = "";
            if (isMime)
                {
                ++mimeParts;
                appendMimeVar(dy, var->name, val, var->type, boundary);
                }
            else            
                appendCgiVar(dy, var->name, val);
            }
        else if (var->type != NULL && sameWord(var->type, "CHECKBOX"))
            {
            if (var->curVal != NULL)
                {
                if (isMime)         
                    {
                    ++mimeParts;
                    appendMimeVar(dy, var->name, var->curVal, var->type, boundary);
                    }
                else        
                    appendCgiVar(dy, var->name, var->curVal);
                }
            }
        else if (isMime && buttonName && sameWord(buttonName,var->name))
            {
            ++mimeParts;
            appendMimeVar(dy, buttonName, buttonVal, NULL, boundary);
            }
        }
    if (isMime) 
        {
        ++mimeParts;
        appendMimeTerminus(dy,boundary);
        if (countOccurrences(boundary,strlen(boundary),dy->string,dy->stringSize) != mimeParts)
            { /* boundary was found in input! # occurrences not as expected */
            dyStringClear(dy);
            continue;  /* if at first you don't succeed, try another boundary string */
            }
        dyStringPrintf(dyHeader, "Content-type: multipart/form-data, boundary=%s\r\n",boundary);
        if (isMime && verboseLevel() == 2)
            {
            mustWrite(stderr, dyHeader->string, dyHeader->stringSize);
            mustWrite(stderr, dy->string, dy->stringSize);
            }
        }
    break;
    }   
    
return dyStringCannibalize(&dy);

}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlFormFree

(

struct htmlForm **

pForm

)

Definition at line 201 of file htmlPage.c.

References freez(), htmlFormVarFreeList(), and htmlForm::vars.

Referenced by htmlFormFreeList().

{
struct htmlForm *form = *pForm;
if (form != NULL)
    {
    htmlFormVarFreeList(&form->vars);
    freez(pForm);
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlFormFreeList

(

struct htmlForm **

pList

)

Definition at line 212 of file htmlPage.c.

References htmlFormFree(), and htmlForm::next.

Referenced by htmlPageFree().

{
struct htmlForm *el, *next;

for (el = *pList; el != NULL; el = next)
    {
    next = el->next;
    htmlFormFree(&el);
    }
*pList = NULL;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct htmlForm* htmlFormGet	(	struct htmlPage *	page,
		char *	name
	)			[read]

Definition at line 986 of file htmlPage.c.

References htmlPage::forms, htmlForm::name, htmlForm::next, and sameWord.

{
struct htmlForm *form;
for (form = page->forms; form != NULL; form = form->next)
    if (sameWord(form->name, name))
        break;
return form;
}

void htmlFormPrint	(	struct htmlForm *	form,
		FILE *	f
	)

Definition at line 977 of file htmlPage.c.

References htmlForm::action, htmlFormVarPrint(), htmlForm::method, htmlForm::name, htmlFormVar::next, and htmlForm::vars.

{
struct htmlFormVar *var;
fprintf(f, "%s\t%s\t%s\n", form->name, form->method, form->action);
for (var = form->vars; var != NULL; var = var->next)
    htmlFormVarPrint(var, f, "\t");
}

Here is the call graph for this function:

void htmlFormVarFree

(

struct htmlFormVar **

pVar

)

Definition at line 174 of file htmlPage.c.

References htmlFormVar::curVal, freeMem(), freez(), slFreeList(), htmlFormVar::tags, and htmlFormVar::values.

Referenced by htmlFormVarFreeList().

{
struct htmlFormVar *var = *pVar;
if (var != NULL)
    {
    freeMem(var->curVal);
    slFreeList(&var->values);
    slFreeList(&var->tags);
    freez(pVar);
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlFormVarFreeList

(

struct htmlFormVar **

pList

)

Definition at line 187 of file htmlPage.c.

References htmlFormVarFree(), and htmlFormVar::next.

Referenced by htmlFormFree().

{
struct htmlFormVar *el, *next;

for (el = *pList; el != NULL; el = next)
    {
    next = el->next;
    htmlFormVarFree(&el);
    }
*pList = NULL;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct htmlFormVar* htmlFormVarGet	(	struct htmlForm *	form,
		char *	name
	)			[read]

Definition at line 996 of file htmlPage.c.

References errAbort(), htmlFormVar::name, htmlFormVar::next, sameWord, and htmlForm::vars.

Referenced by htmlFormVarSet(), and htmlPageGetVar().

{
struct htmlFormVar *var;
if (form == NULL)
    errAbort("Null form passed to htmlFormVarGet");
for (var = form->vars; var != NULL; var = var->next)
    if (sameWord(var->name, name))
        break;
return var;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlFormVarPrint	(	struct htmlFormVar *	var,
		FILE *	f,
		char *	prefix
	)

Definition at line 966 of file htmlPage.c.

References htmlFormVar::curVal, naForNull(), htmlFormVar::name, slName::name, slName::next, htmlFormVar::tagName, htmlFormVar::type, and htmlFormVar::values.

Referenced by htmlFormPrint().

{
struct slName *val;
fprintf(f, "%s%s\t%s\t%s\t%s\n", prefix, var->name, var->tagName, 
        naForNull(var->type), 
        naForNull(var->curVal));
for (val = var->values; val != NULL; val = val->next)
     fprintf(f, "%s\t%s\n", prefix, val->name);
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlFormVarSet	(	struct htmlForm *	form,
		char *	name,
		char *	val
	)

Definition at line 1008 of file htmlPage.c.

References AllocVar, cloneString(), htmlFormVar::curVal, errAbort(), freez(), htmlFormVarGet(), slAddHead, and htmlForm::vars.

Referenced by htmlPageSetVar().

{
struct htmlFormVar *var;
if (form == NULL)
    errAbort("Null form passed to htmlFormVarSet");
var = htmlFormVarGet(form, name);
if (var == NULL)
    {
    AllocVar(var);
    var->type = "TEXT";
    var->tagName = "INPUT";
    var->name = name;
    slAddHead(&form->vars, var);
    }
freez(&var->curVal);
var->curVal = cloneString(val);
}

Here is the call graph for this function:

Here is the caller graph for this function:

char* htmlNextCrLfLine

(

char **

pS

)

Definition at line 328 of file htmlPage.c.

References verbose().

Referenced by htmlHeaderRead().

{
char *s = *pS, *e;
if (s == NULL || s[0] == 0)
    return NULL;
e = strchr(s, '\n');
if (e == NULL)
    verbose(1, "End of file in header\n");
else 
    {
    *e = 0;
    if (e == s || e[-1] != '\r')
        verbose(1, "Missing <CR> in header line\n");
    else
       e[-1] = 0;
    e += 1;
    }
*pS = e;
return s;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlPageFormOrAbort

(

struct htmlPage *

page

)

Definition at line 1755 of file htmlPage.c.

References errAbort(), and htmlPage::forms.

{
if (page == NULL)
    errAbort("Can't validate NULL page");
if (page->forms == NULL)
    errAbort("No form found");
}

Here is the call graph for this function:

struct htmlPage* htmlPageForwarded	(	char *	url,
		struct htmlCookie *	cookies
	)			[read]

Definition at line 918 of file htmlPage.c.

References htmlPage::cookies, hashFindVal(), htmlPage::header, htmlPageFree(), and htmlPageGetWithCookies().

Referenced by htmlPageForwardedNoAbort().

{
struct htmlPage *page = htmlPageGetWithCookies(url, cookies);
int level, maxLevels = 7;
for (level = 0; level < maxLevels; ++level)
    {
    struct htmlPage *newPage;
    char *newUrl = hashFindVal(page->header, "Location:");
    if (newUrl == NULL)
        break;
    newPage = htmlPageGetWithCookies(newUrl, cookies);
    htmlPageFree(&page);
    page = newPage;
    }
return page;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct htmlPage* htmlPageForwardedNoAbort	(	char *	url,
		struct htmlCookie *	cookies
	)			[read]

Definition at line 938 of file htmlPage.c.

References htmlPage::cookies, errCatchEnd(), errCatchFree(), errCatchNew(), errCatchStart, errCatch::gotError, htmlPageForwarded(), errCatch::message, dyString::string, and warn().

{
struct errCatch *errCatch = errCatchNew();
struct htmlPage *page = NULL;
if (errCatchStart(errCatch))
    page = htmlPageForwarded(url, cookies);
errCatchEnd(errCatch);
if (errCatch->gotError)
    warn(errCatch->message->string);
errCatchFree(&errCatch);
return page;
}

Here is the call graph for this function:

void htmlPageFree

(

struct htmlPage **

pPage

)

Definition at line 225 of file htmlPage.c.

References htmlPage::cookies, htmlPage::forms, freeHashAndVals(), freez(), htmlPage::fullText, htmlPage::header, htmlCookieFreeList(), htmlFormFreeList(), htmlStatusFree(), htmlTagFreeList(), htmlPage::status, htmlPage::tags, and htmlPage::url.

Referenced by htmlPageForwarded(), htmlPageFreeList(), qaPageFromForm(), qaPageGet(), and qaStatusOnPage().

{
struct htmlPage *page = *pPage;
if (page != NULL)
    {
    freez(&page->url);
    htmlStatusFree(&page->status);
    freeHashAndVals(&page->header);
    htmlCookieFreeList(&page->cookies);
    freez(&page->fullText);
    htmlTagFreeList(&page->tags);
    htmlFormFreeList(&page->forms);
    freez(pPage);
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlPageFreeList

(

struct htmlPage **

pList

)

Definition at line 242 of file htmlPage.c.

References htmlPageFree(), and htmlPage::next.

{
struct htmlPage *el, *next;

for (el = *pList; el != NULL; el = next)
    {
    next = el->next;
    htmlPageFree(&el);
    }
*pList = NULL;
}

Here is the call graph for this function:

struct htmlPage* htmlPageFromForm	(	struct htmlPage *	origPage,
		struct htmlForm *	form,
		char *	buttonName,
		char *	buttonVal
	)			[read]

Definition at line 1356 of file htmlPage.c.

References htmlForm::action, cookieOutput(), htmlPage::cookies, dyStringAppend(), dyStringCannibalize(), dyStringFree, dyStringNew, dyStringPrintf(), FALSE, freez(), htmlExpandUrl(), htmlFormCgiVars(), htmlPageParse(), htmlForm::method, netOpenHttpExt(), netSlurpFile(), sameWord, dyString::string, dyString::stringSize, htmlPage::url, and verbose().

Referenced by qaPageFromForm().

{
struct htmlPage *newPage = NULL;
struct dyString *dyUrl = dyStringNew(0);
struct dyString *dyHeader = dyStringNew(0);
struct dyString *dyText = NULL;
char *url = htmlExpandUrl(origPage->url, form->action);
char *cgiVars = NULL;
int contentLength = 0;
int sd = -1;

dyStringAppend(dyUrl, url);
cookieOutput(dyHeader, origPage->cookies);
if (sameWord(form->method, "GET"))
    {
    cgiVars = htmlFormCgiVars(origPage, form, buttonName, buttonVal, dyHeader);
    dyStringAppend(dyUrl, "?");
    dyStringAppend(dyUrl, cgiVars);
    verbose(3, "GET %s\n", dyUrl->string);
    sd = netOpenHttpExt(dyUrl->string, form->method, FALSE);
    dyStringAppend(dyHeader, "\r\n");
    write(sd, dyHeader->string, dyHeader->stringSize);
    }
else if (sameWord(form->method, "POST"))
    {
    cgiVars = htmlFormCgiVars(origPage, form, buttonName, buttonVal, dyHeader);
    contentLength = strlen(cgiVars);
    verbose(3, "POST %s\n", dyUrl->string);
    sd = netOpenHttpExt(dyUrl->string, form->method, FALSE);
    dyStringPrintf(dyHeader, "Content-length: %d\r\n", contentLength);
    dyStringAppend(dyHeader, "\r\n");
    write(sd, dyHeader->string, dyHeader->stringSize);
    write(sd, cgiVars, contentLength);
    }
dyText = netSlurpFile(sd);
close(sd);
newPage = htmlPageParse(url, dyStringCannibalize(&dyText));
freez(&url);
dyStringFree(&dyUrl);
dyStringFree(&dyHeader);
freez(&cgiVars);
return newPage;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct htmlPage* htmlPageGet

(

char *

url

)

[read]

Definition at line 953 of file htmlPage.c.

References fileExists(), htmlPageGetWithCookies(), htmlPageParseNoHead(), and readInGulp().

Referenced by qaPageGet().

{
if (fileExists(url))
    {
    char *buf;
    readInGulp(url, &buf, NULL);
    return htmlPageParseNoHead(url, buf);
    }
else
    return htmlPageGetWithCookies(url, NULL);
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct htmlFormVar* htmlPageGetVar	(	struct htmlPage *	page,
		struct htmlForm *	form,
		char *	name
	)			[read]

Definition at line 1028 of file htmlPage.c.

References htmlPage::forms, and htmlFormVarGet().

{
if (form == NULL)
    form = page->forms;
return htmlFormVarGet(form, name);
}

Here is the call graph for this function:

struct htmlPage* htmlPageGetWithCookies	(	char *	url,
		struct htmlCookie *	cookies
	)			[read]

Definition at line 910 of file htmlPage.c.

References htmlPage::cookies, htmlPageParse(), and htmlSlurpWithCookies().

Referenced by htmlPageForwarded(), and htmlPageGet().

{
char *buf = htmlSlurpWithCookies(url, cookies);
return htmlPageParse(url, buf);
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct slName* htmlPageLinks

(

struct htmlPage *

page

)

[read]

Definition at line 1430 of file htmlPage.c.

References htmlPageScanAttribute().

Referenced by validateCgiUrls().

{
return htmlPageScanAttribute(page, NULL, "HREF");
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct htmlPage* htmlPageParse	(	char *	url,
		char *	fullText
	)			[read]

Definition at line 831 of file htmlPage.c.

References AllocVar, cloneLongString(), cloneString(), htmlPage::cookies, htmlPage::forms, freez(), htmlPage::fullText, hashAdd(), hashFindVal(), htmlPage::header, htmlHeaderRead(), htmlParseForms(), htmlStatusParse(), htmlTagScan(), htmlPage::htmlText, startsWith(), htmlStatus::status, htmlPage::status, htmlPage::tags, htmlPage::url, and warn().

Referenced by htmlPageFromForm(), htmlPageGetWithCookies(), and htmlPageParseOk().

{
struct htmlPage *page;
char *dupe = cloneLongString(fullText);
char *s = dupe;
struct htmlStatus *status = htmlStatusParse(&s);
char *contentType;

if (status == NULL)
    return NULL;

AllocVar(page);
page->url = cloneString(url);
page->fullText = fullText;
page->status = status;
page->header = htmlHeaderRead(&s, &page->cookies);
contentType = hashFindVal(page->header, "Content-Type:");
if (contentType == NULL)        
    {
    warn("No contentType, assuming text/html");
    contentType = cloneString("text/html");
    hashAdd(page->header, "Content-Type:", contentType);
    }
page->htmlText = fullText + (s - dupe);
if (startsWith("text/html", contentType))
    {
    page->tags = htmlTagScan(page->htmlText, s);
    page->forms = htmlParseForms(page, page->tags, NULL);
    }
freez(&dupe);
return page;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct htmlPage* htmlPageParseNoHead	(	char *	url,
		char *	htmlText
	)			[read]

Definition at line 865 of file htmlPage.c.

References AllocVar, cloneString(), htmlPage::forms, freez(), htmlPage::fullText, htmlParseForms(), htmlTagScan(), htmlPage::htmlText, htmlPage::tags, and htmlPage::url.

Referenced by htmlPageGet().

{
char *dupe = cloneString(htmlText);
struct htmlPage *page;
AllocVar(page);
page->url = cloneString(url);
page->fullText = page->htmlText = htmlText;
page->tags = htmlTagScan(page->htmlText, dupe);
page->forms = htmlParseForms(page, page->tags, NULL);
freez(&dupe);
return page;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct htmlPage* htmlPageParseOk	(	char *	url,
		char *	fullText
	)			[read]

Definition at line 879 of file htmlPage.c.

References errAbort(), htmlPageParse(), noWarnAbort(), htmlPage::status, and htmlStatus::status.

{
struct htmlPage *page = htmlPageParse(url, fullText);
if (page == NULL)
   noWarnAbort();
if (page->status->status != 200)
   errAbort("%s returned with status code %d", url, page->status->status);
return page;
}

Here is the call graph for this function:

struct slName* htmlPageScanAttribute	(	struct htmlPage *	page,
		char *	tagName,
		char *	attribute
	)			[read]

Definition at line 1403 of file htmlPage.c.

References htmlTag::attributes, htmlTag::name, htmlAttribute::name, htmlTag::next, htmlAttribute::next, sameWord, slAddHead, slNameNew, slReverse(), htmlPage::tags, and htmlAttribute::val.

Referenced by htmlPageLinks().

{
struct htmlTag *tag;
struct htmlAttribute *att;
struct slName *list = NULL, *el;

for (tag = page->tags; tag != NULL; tag = tag->next)
    {
    if (tagName == NULL || sameWord(tagName, tag->name))
        {
        for (att = tag->attributes; att != NULL; att = att->next)
            {
            if (sameWord(attribute, att->name))
                {
                el = slNameNew(att->val);
                slAddHead(&list, el);
                }
            }
        }
    }
slReverse(&list);
return list;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlPageSetVar	(	struct htmlPage *	page,
		struct htmlForm *	form,
		char *	name,
		char *	val
	)

Definition at line 1036 of file htmlPage.c.

References errAbort(), htmlPage::forms, and htmlFormVarSet().

{
if (page == NULL)
    errAbort("Null page passed to htmlPageSetVar");
if (form == NULL)
    form = page->forms;
if (form == NULL)
    errAbort("Null form in htmlPageSetVar");
htmlFormVarSet(form, name, val);
}

Here is the call graph for this function:

void htmlPageValidateOrAbort

(

struct htmlPage *

page

)

Definition at line 1764 of file htmlPage.c.

References ArraySize, checkExactlyOne(), errAbort(), FALSE, hashFindVal(), htmlPage::header, htmlTag::name, htmlTag::next, sameWord, startsWith(), htmlPage::tags, touppers(), TRUE, validateBody(), validateCgiUrls(), validateNestingTags(), and warn().

Referenced by qaPageFromForm(), and qaPageGet().

{
struct htmlTag *tag;
boolean gotTitle = FALSE;
char *contentType = NULL;

if (page == NULL)
    errAbort("Can't validate NULL page");
if (page->header != NULL)
    contentType = hashFindVal(page->header, "Content-Type:");
if (contentType == NULL || startsWith("text/html", contentType))
    {
    /* To simplify things upper case all tag names. */
    for (tag = page->tags; tag != NULL; tag = tag->next)
        touppers(tag->name);

    checkExactlyOne(page->tags, "BODY");

    /* Validate header, and make a suggestion or two */
    if ((tag = page->tags) == NULL)
        errAbort("No tags");
    if (!sameWord(tag->name, "HTML"))
        errAbort("Doesn't start with <HTML> tag");
    tag = tag->next;
    if (tag == NULL || !sameWord(tag->name, "HEAD"))
        warn("<HEAD> tag does not follow <HTML> tag");
    else
        {
        for (;;)
            {
            tag = tag->next;
            if (tag == NULL)
                errAbort("Missing </HEAD>");
            if (sameWord(tag->name, "TITLE"))
                gotTitle = TRUE;
            if (sameWord(tag->name, "/HEAD"))
                break;
            }
        if (!gotTitle)
            warn("No title in <HEAD>");
        validateNestingTags(page, page->tags, tag, headNesters, ArraySize(headNesters));
        tag = tag->next;
        }
    if (tag == NULL || !sameWord(tag->name, "BODY"))
        errAbort("<BODY> tag does not follow <HTML> tag");
    tag = validateBody(page, tag->next);
    if (tag == NULL || !sameWord(tag->name, "/HTML"))
        errAbort("Missing </HTML>");
    validateCgiUrls(page);
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

char* htmlSlurpWithCookies	(	char *	url,
		struct htmlCookie *	cookies
	)

Definition at line 890 of file htmlPage.c.

References cookieOutput(), dyStringAppend(), dyStringCannibalize(), dyStringFree, dyStringNew, FALSE, netOpenHttpExt(), netSlurpFile(), dyString::string, and dyString::stringSize.

Referenced by htmlPageGetWithCookies().

{
struct dyString *dyHeader = dyStringNew(0);
struct dyString *dyText;
int sd;

cookieOutput(dyHeader, cookies);
dyStringAppend(dyHeader, "\r\n");
sd = netOpenHttpExt(url, "GET", FALSE);
write(sd, dyHeader->string, dyHeader->stringSize);
dyText = netSlurpFile(sd);
close(sd);
dyStringFree(&dyHeader);
return dyStringCannibalize(&dyText);
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlStatusFree

(

struct htmlStatus **

pStatus

)

Definition at line 28 of file htmlPage.c.

References freeMem(), freez(), and htmlStatus::status.

Referenced by htmlPageFree(), and htmlStatusFreeList().

{
struct htmlStatus *status = *pStatus;
if (status != NULL)
    {
    freeMem(status->version);
    freez(pStatus);
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlStatusFreeList

(

struct htmlStatus **

pList

)

Definition at line 39 of file htmlPage.c.

References htmlStatusFree(), and htmlStatus::next.

{
struct htmlStatus *el, *next;

for (el = *pList; el != NULL; el = next)
    {
    next = el->next;
    htmlStatusFree(&el);
    }
*pList = NULL;
}

Here is the call graph for this function:

struct htmlStatus* htmlStatusParse

(

char **

pText

)

[read]

Definition at line 299 of file htmlPage.c.

References AllocVar, cloneStringZ(), skipLeadingSpaces(), skipToSpaces(), htmlStatus::status, and warn().

Referenced by htmlPageParse().

{
char *text = *pText;
char *end = strchr(text, '\n');
struct htmlStatus *status;
if (end != NULL)
   *pText = end+1;
else
   *pText = text + strlen(text);
end = skipToSpaces(text);
if (end == NULL)
    {
    warn("Short status line.");
    return NULL;
    }
AllocVar(status);
status->version = cloneStringZ(text, end-text);
end = skipLeadingSpaces(end);
if (!isdigit(end[0]))
    {
    warn("Not a number in status field");
    return NULL;
    }
status->status = atoi(end);
return status;
}

Here is the call graph for this function:

Here is the caller graph for this function:

char* htmlTagAttributeNeeded	(	struct htmlPage *	page,
		struct htmlTag *	tag,
		char *	name
	)

Definition at line 465 of file htmlPage.c.

References htmlTagAttributeVal(), tagWarn(), and htmlAttribute::val.

Referenced by formParseVars(), and htmlParseForms().

{
char *val = htmlTagAttributeVal(page, tag, name, NULL);
if (val == NULL)
    {
    tagWarn(page, tag, "Missing %s attribute", name);
    val = "n/a";
    }
return val;
}

Here is the call graph for this function:

Here is the caller graph for this function:

char* htmlTagAttributeVal	(	struct htmlPage *	page,
		struct htmlTag *	tag,
		char *	name,
		char *	defaultVal
	)

Definition at line 455 of file htmlPage.c.

References htmlTag::attributes, htmlAttributeFindVal(), and htmlAttribute::val.

Referenced by formParseVars(), htmlParseForms(), and htmlTagAttributeNeeded().

{
char *val = htmlAttributeFindVal(tag->attributes, name);
if (val == NULL)
    val = defaultVal;
return val;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlTagFree

(

struct htmlTag **

pTag

)

Definition at line 149 of file htmlPage.c.

References htmlTag::attributes, freeMem(), freez(), htmlAttributeFreeList(), and htmlTag::name.

Referenced by htmlTagFreeList().

{
struct htmlTag *tag = *pTag;
if (tag != NULL)
    {
    htmlAttributeFreeList(&tag->attributes);
    freeMem(tag->name);
    freez(pTag);
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void htmlTagFreeList

(

struct htmlTag **

pList

)

Definition at line 161 of file htmlPage.c.

References htmlTagFree(), and htmlTag::next.

Referenced by htmlPageFree().

{
struct htmlTag *el, *next;

for (el = *pList; el != NULL; el = next)
    {
    next = el->next;
    htmlTagFree(&el);
    }
*pList = NULL;
}

Here is the call graph for this function:

Here is the caller graph for this function:

---