lib/mime.c File Reference

#include "common.h"
#include "hash.h"
#include "linefile.h"
#include "cheapcgi.h"
#include "portable.h"
#include "errabort.h"
#include "mime.h"

Include dependency graph for mime.c:

Go to the source code of this file.

Defines
#define	MAXPARTSIZE   512*1024*1024
#define	MAXPARTLINESIZE   1024
#define	MAXDATASIZE   64LL*1024*1024*1024
#define	MAXBOUNDARY   72+5
Functions
static void	setEopMB (struct mimeBuf *b)
static void	setEodMB (struct mimeBuf *b)
static void	setBoundaryMB (struct mimeBuf *b, char *boundary)
static void	moreMimeBuf (struct mimeBuf *b)
static char	getcMB (struct mimeBuf *b)
static void	putBackMB (struct mimeBuf *b)
static char *	getLineMB (struct mimeBuf *b)
static void	getChunkMB (struct mimeBuf *b, char **address, int *size, boolean *hasZeros)
static void	readPartHeaderMB (struct mimeBuf *b, struct mimePart *p, char *altHeader)
mimeBuf *	initMimeBuf (int d)
char *	getMimeHeaderMainVal (char *header)
char *	getMimeHeaderFieldVal (char *header, char *field)
char *	getNewLineByType ()
mimePart *	parseMultiParts (struct mimeBuf *b, char *altHeader)
Variables
static char const	rcsid [] = "$Id: mime.c,v 1.13 2006/06/20 18:31:24 galt Exp $"
enum nlType	nlType = nlt_undet

---

Define Documentation

#define MAXBOUNDARY   72+5

Definition at line 27 of file mime.c.

Referenced by parseMultiParts().

#define MAXDATASIZE   64LL*1024*1024*1024

Definition at line 26 of file mime.c.

Referenced by parseMultiParts().

#define MAXPARTLINESIZE   1024

Definition at line 25 of file mime.c.

Referenced by getLineMB().

#define MAXPARTSIZE   512*1024*1024

Definition at line 24 of file mime.c.

Referenced by parseMultiParts().

---

Function Documentation

static void getChunkMB	(	struct mimeBuf *	b,
		char **	address,
		int *	size,
		boolean *	hasZeros
	)			[static]

Definition at line 198 of file mime.c.

References mimeBuf::eod, mimeBuf::eop, mimeBuf::i, and memMatch().

Referenced by parseMultiParts().

{
char *eoc = b->eop ? b->eop : b->eod; /* end of chunk */
//debug
//fprintf(stderr,"pre-getChunkMB dumpMB: ");
//dumpMB(b);  //debug
*address=b->i;
*size=eoc - b->i;
*hasZeros = (memMatch("", 1,*address, *size) != NULL);
b->i = eoc;
}

Here is the call graph for this function:

Here is the caller graph for this function:

static char getcMB

(

struct mimeBuf *

b

)

[static]

Definition at line 125 of file mime.c.

References mimeBuf::eod, mimeBuf::eoi, mimeBuf::eom, errAbort(), mimeBuf::i, and moreMimeBuf().

Referenced by getLineMB(), and parseMultiParts().

{
if (b->i >= b->eoi && b->eoi < b->eom)  /* at end of input */
    errAbort("getcMB error - requested input beyond end of MIME input.");
if (b->i >= b->eod && b->eoi == b->eom) /* at end of buffer */
    moreMimeBuf(b);
    
//fprintf(stderr,"b->buf:%lu b->i:%lu %c \n",
//    (unsigned long) b->buf,
//    (unsigned long) b->i,
//    *b->i
//    );
//fprintf(stderr,"%c",*b->i);
//fflush(stderr); 
return *b->i++;    
}

Here is the call graph for this function:

Here is the caller graph for this function:

static char* getLineMB

(

struct mimeBuf *

b

)

[static]

Definition at line 152 of file mime.c.

References cloneString(), errAbort(), getcMB(), MAXPARTLINESIZE, nlt_dos, nlt_mac, nlt_undet, nlt_unix, putBackMB(), and TRUE.

Referenced by parseMultiParts(), and readPartHeaderMB().

{
char line[MAXPARTLINESIZE];
int i = 0;
char c = 0;
line[0]=0;
while(TRUE)
    {
    c =getcMB(b);
    if ((c == 0x0d) || (c == 0x0a))  /* CR or LF is end of line */
        break;
    line[i++] = c;
    if (i >= MAXPARTLINESIZE)
        errAbort("getLineMB error - MIME input header too long, "
                    "greater than %d chars",MAXPARTLINESIZE);
    }
line[i] = 0; /* terminate string */ 
if (nlType == nlt_undet)  /* determine newline type */
    {
    if (c == 0x0d)
        {
        nlType = nlt_mac;
        c = getcMB(b);
        if (c == 0x0a)
            nlType = nlt_dos;
        else
            putBackMB(b);
        }
    else
        {
        nlType = nlt_unix;
        }
    }
else if (nlType == nlt_dos)
    {
    if (c == 0x0d)
        getcMB(b); /* just waste the LF */
    else
        nlType = nlt_unix;
    }
return cloneString(line);
}

Here is the call graph for this function:

Here is the caller graph for this function:

char* getMimeHeaderFieldVal	(	char *	header,
		char *	field
	)

Definition at line 333 of file mime.c.

References cloneString(), errAbort(), and TRUE.

Referenced by cgiParseMultipart(), and parseMultiParts().

{
char value[1024]; 
char *fld = header;
int i = 0;
char *puncChars = ",;: \t\r\n"; /* punctuation chars */
while (TRUE)
    {
    fld = strstr(fld,field);
    if (!fld)
        return NULL;
    if (fld > header && strchr(puncChars,fld[-1]))
        {
        fld+=strlen(field);
        if (*fld == '=')
            {
            ++fld;
            break;
            }
        }    
    else
        {
        ++fld;
        }
    }   
if (*fld == '"')
    {
    puncChars = "\"";  /* quoted */
    ++fld;
    }
i=0;
while(TRUE)
    {
    char c = *fld++;
    if (c==0 || strchr(puncChars,c))
        break;
    value[i++] = c;
    if (i >= sizeof(value))
        errAbort("error: %s= value too long (>%lu) in MIME header Content-type:%s",field,(unsigned long)sizeof(value),header);
    }
value[i] = 0;    

return cloneString(value);

}

Here is the call graph for this function:

Here is the caller graph for this function:

char* getMimeHeaderMainVal

(

char *

header

)

Definition at line 306 of file mime.c.

References cloneString(), errAbort(), and TRUE.

Referenced by cgiParseMultipart().

{
char value[1024]; 
char *h = header;
int i = 0;
char *puncChars = ",;: \t\r\n"; /* punctuation chars */
i=0;
/* The header should have already been trimmed of leading and trailing spaces */
while(TRUE)
    {
    char c = *h++;
    if (c==0 || strchr(puncChars,c))
        break;
    value[i++] = c;
    if (i >= sizeof(value))
        errAbort("error: main value too long (>%lu) in MIME header Content-type:%s",(unsigned long)sizeof(value),header);
    }
value[i] = 0;    

return cloneString(value);

}

Here is the call graph for this function:

Here is the caller graph for this function:

char* getNewLineByType

(

)

Definition at line 382 of file mime.c.

References nlt_dos, nlt_mac, and nlt_unix.

Referenced by parseMultiParts().

{
switch (nlType)
    {
    case nlt_dos:
        //debug
        //fprintf(stderr,"nlType=nlt_dos\n");
        return "\x0d\x0a";
    case nlt_mac:
        //debug
        //fprintf(stderr,"nlType=nlt_mac\n");
        return "\x0d";
    case nlt_unix:
    default:
        //debug
        //fprintf(stderr,"nlType=nlt_unix\n");
        return "\x0a";
    }
}

Here is the caller graph for this function:

struct mimeBuf* initMimeBuf

(

int

d

)

[read]

Definition at line 289 of file mime.c.

References AllocA, mimeBuf::blen, mimeBuf::boundary, mimeBuf::buf, mimeBuf::d, mimeBuf::eod, mimeBuf::eoi, mimeBuf::eom, mimeBuf::i, and moreMimeBuf().

Referenced by cgiParseMultipart().

{
struct mimeBuf *b=AllocA(*b);
b->d = d;
b->boundary = NULL;
b->blen = 0;
b->eom = b->buf+MIMEBUFSIZE;
b->eoi = b->eom;
b->eod = b->eom;
b->i = b->eom;
moreMimeBuf(b);
return b;
}

Here is the call graph for this function:

Here is the caller graph for this function:

static void moreMimeBuf

(

struct mimeBuf *

b

)

[static]

Definition at line 93 of file mime.c.

References mimeBuf::blen, mimeBuf::buf, mimeBuf::d, mimeBuf::eoi, mimeBuf::eom, errnoAbort(), mimeBuf::i, setEodMB(), and setEopMB().

Referenced by getcMB(), initMimeBuf(), and parseMultiParts().

{
int bytesRead = 0, bytesToRead = 0;
if (b->blen > 1)
    {
    int r = b->eoi - b->i;
    memmove(b->buf, b->i, r);
    b->eoi = b->buf+r;
    }
else
    {
    b->eoi = b->buf;
    }
b->i = b->buf+0;
bytesToRead = b->eom - b->eoi;
while (bytesToRead > 0)
    {
    bytesRead = read(b->d, b->eoi, bytesToRead);
    if (bytesRead < 0)
        errnoAbort("moreMimeBuf: error reading MIME input descriptor");
    b->eoi += bytesRead;
    if (bytesRead == 0)
        break;
    bytesToRead = bytesToRead - bytesRead;
    }
setEopMB(b);
setEodMB(b);
//debug
//fprintf(stderr,"post-moreMime dumpMB: ");
//dumpMB(b);  //debug
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct mimePart* parseMultiParts	(	struct mimeBuf *	b,
		char *	altHeader
	)			[read]

Definition at line 403 of file mime.c.

References AllocA, mimePart::binary, mimeBuf::boundary, carefulClose(), cloneString(), mimePart::data, dyStringAppendN(), dyStringCannibalize(), dyStringNew, dyStringPrintf(), mimeBuf::eoi, mimeBuf::eom, errAbort(), FALSE, mimePart::fileName, tempName::forCgi, freeDyString(), freez(), getChunkMB(), getcMB(), getLineMB(), getMimeHeaderFieldVal(), getNewLineByType(), hashFindVal(), mimePart::hdr, mimeBuf::i, makeTempName(), MAXBOUNDARY, MAXDATASIZE, MAXPARTSIZE, moreMimeBuf(), mimePart::multi, mustOpen(), mustWrite(), needLargeMem(), newDyString(), nlt_dos, nlt_mac, nlt_unix, parseMultiParts(), readPartHeaderMB(), safef(), sameOk, sameString, setBoundaryMB(), setEopMB(), mimePart::size, slAddHead, slReverse(), startsWith(), dyString::string, dyString::stringSize, touppers(), and TRUE.

Referenced by cgiParseMultipart(), and parseMultiParts().

{ 
struct mimePart *p=AllocA(*p);
char *parentboundary = NULL, *boundary = NULL;
char *ct = NULL;
boolean autoBoundary = FALSE;


//debug
//fprintf(stderr,"altHeader=[%s]\n",altHeader);

if (sameOk(altHeader, "autoBoundary"))
    { /* process things with no explicit header.
       *  look for *MIME* \n\n-- */
    struct dyString *dy = dyStringNew(0);
    char *prevPrevLine = NULL;
    char *prevLine = NULL;
    char *line = NULL;
    boolean found = FALSE;
    autoBoundary = TRUE;
    while (TRUE)
        {
        if (b->i >= b->eoi && b->eoi < b->eom)  /* at end of input */
            break;
        line = getLineMB(b);
        if (line && startsWith("--",line) // && 
            //sameString(prevLine,"") && 
            //prevPrevLine &&
            //stringIn("MULTI",prevPrevLine) && 
            //stringIn("MIME",prevPrevLine) 
            )
            {
            found = TRUE;
            break;
            }
        freez(&prevPrevLine);
        prevPrevLine = prevLine;
        prevLine = line;
        if (prevPrevLine)
            touppers(prevPrevLine);
        }
    if (!found)
        errAbort("autoBoundary: No initial boundary found.");

    dyStringPrintf(dy, "CONTENT-TYPE:multipart/form-data; boundary=%s%s%s", 
        line+2, getNewLineByType(), getNewLineByType() );
    altHeader = dyStringCannibalize(&dy); 
    
    //debug
    //fprintf(stderr,"autoBoundary altHeader = [%s]\n",altHeader);
    //fflush(stderr); 

    freez(&prevPrevLine);           
    freez(&prevLine);       
    freez(&line);           
    }

//debug
//fprintf(stderr,"\n");
readPartHeaderMB(b,p,altHeader);

ct = hashFindVal(p->hdr,"content-type");  /* use lowercase key */
//debug
//fprintf(stderr,"ct from hash:%s\n",ct);
//fflush(stderr); 

if (ct && startsWith("multipart/",ct))
    {
    char bound[MAXBOUNDARY]; 
    char *bnd = NULL;
    struct mimePart *child = NULL;

    /* these 3 vars just for processing epilog chunk: */
    char *bp=NULL;
    int size=0;
    boolean hasZeros=FALSE;

    /* save */
    parentboundary = b->boundary;

    boundary = getMimeHeaderFieldVal(ct,"boundary");
    if (strlen(boundary) >= MAXBOUNDARY)
        errAbort("error: boundary= value too long in MIME header Content-type:%s",ct);
    safef(bound, sizeof(bound), "--%s",boundary);  /* do not prepend CRLF to boundary yet */
    freez(&boundary);
    boundary = cloneString(bound);
    //debug
    //fprintf(stderr,"initial boundary parsed:%s\n",boundary);
    //fflush(stderr); 

    if (!autoBoundary)
        {
        /* skip any extra "prolog" before the initial boundary marker */
        while (TRUE)
            {
            bnd = getLineMB(b);
            if (sameString(bnd,boundary)) 
               break;
            freez(&bnd);
            }
            //debug
            //fprintf(stderr,"initial boundary found:%s\n",bnd);
            //fflush(stderr); 
        freez(&bnd);
        }

    /* include crlf in the boundary so bodies won't have trailing a CRLF
     * this is done here so that in case there's no extra CRLF
     * between the header and the boundary, it will still work,
     * so we only prepend the CRLF to the boundary after initial found */
    safef(bound,sizeof(bound),"%s%s", getNewLineByType(), boundary);
    freez(&boundary);
    boundary=cloneString(bound);
    
    setBoundaryMB(b, boundary);

    while(TRUE)
        {
        int i = 0;
        char c1 = ' ', c2 = ' ';
        child = parseMultiParts(b,NULL);
        slAddHead(&p->multi,child);
        //call getLine, compare to boundary 
        /* skip extra initial boundary marker - it's moot anyway */
        freez(&bnd);
            //debug
            //fprintf(stderr,"post-parse pre-getLineMB dumpMB: ");
            //dumpMB(b);  //debug
        for (i=0;i<strlen(boundary);++i)
            bound[i] = getcMB(b);
        bound[i] = 0;    
        if (!sameString(bound,boundary))
            errAbort("expected boundary %s, but found %s in MIME",boundary,bound);
        //debug
        //fprintf(stderr,"\nfound boundary:%s\n",bound);
        //fflush(stderr); 
        c1 = getcMB(b);
        if (c1 == '-')
            {
            c2 = getcMB(b);
            if (c2 == '-')
                break;  /* last boundary found */
            else                    
                errAbort("expected -- after boundary %s, but found %c%c in MIME",boundary,c1,c2);
            }
        if (nlType == nlt_dos)
            c2 = getcMB(b);
        switch (nlType)
            {
            case nlt_dos:
                if (c1 == 0x0d && c2 == 0x0a)
                    break;
                else                
                    errAbort("expected CRLF after boundary %s, but found %c%c in MIME",boundary,c1,c2);
            case nlt_unix:
                if (c1 == 0x0a)
                    break;
                else                
                    errAbort("expected LF after boundary %s, but found %c in MIME",boundary,c1);
            case nlt_mac:
                if (c1 == 0x0d)
                    break;
                else                
                    errAbort("expected CR after boundary %s, but found %c in MIME",boundary,c1);
            default:
                    errAbort("unexpected nlType %d after boundary %s",nlType,boundary);
            }
        setEopMB(b);
        }       
    freez(&bnd);
    slReverse(&p->multi);
    /* restore */
    freez(&boundary);
    boundary = parentboundary;
        //debug
        //fprintf(stderr,"restoring parent boundary = %s\n",boundary);
    setBoundaryMB(b, boundary);

    /* dump any "epilog" that may be between the 
     * end of the child boundary and the parent boundary */
    getChunkMB(b, &bp, &size, &hasZeros);
    //debug
    //fprintf(stderr,"epilog size=%d\n",size);
           
    
    }
else
    {
    char *bp=NULL;
    int size=0;
    boolean hasZeros=FALSE;
    boolean toobig=FALSE;
    boolean asFile=FALSE;
    boolean convert=FALSE;
    FILE *f = NULL;
    struct dyString *dy=newDyString(1024);
    //debug
    //fprintf(stderr,"starting new part (non-multi), dumpMB: \n");
    //dumpMB(b);  //debug
    
    //debug
    //ct = hashFindVal(p->hdr,"content-transfer-encoding");  /* use lowercase key */
    //fprintf(stderr,"cte from hash:%s\n",ct);
        
    while(TRUE)
        {
        // break if eop, eod, eoi
        getChunkMB(b, &bp, &size, &hasZeros);
        //debug
        //fprintf(stderr,"bp=%lu size=%d, hasZeros=%d \n", 
        //    (unsigned long) bp,
        //    size,
        //    hasZeros);
        if (hasZeros)
            {
            p->binary=TRUE;
            }
        //if (hasZeros && !asFile)
        //    {
        //    convert=TRUE;
        //    }
        if (!asFile && p->size+size > MAXPARTSIZE)
            {
            toobig = TRUE;
            convert=TRUE;
            }
        if (convert)
            {
            struct tempName uploadedData;
            convert=FALSE;
            asFile = TRUE;
            makeTempName(&uploadedData, "hgSs", ".cgi");
            p->fileName=cloneString(uploadedData.forCgi);
            f = mustOpen(p->fileName,"w");
            mustWrite(f,dy->string,dy->stringSize);
            freeDyString(&dy);
            }
        if (asFile)
            {
            mustWrite(f,bp,size);
            }
        else
            {
            dyStringAppendN(dy,bp,size);
            }
        p->size+=size;
        if (p->size > MAXDATASIZE)
            errAbort("max data size allowable for upload in MIME exceeded %llu",(unsigned long long)MAXDATASIZE);
            
        
        if (b->eop && b->i == b->eop)  /* end of part */
            {
            break;
            }
        if (b->i == b->eoi && b->eoi < b->eom) /* end of data */
            {
            break;
            }
        moreMimeBuf(b);
        }
    if (dy)
        {
        p->data=needLargeMem(dy->stringSize+1);
        memcpy(p->data,dy->string,dy->stringSize);
        p->data[dy->stringSize] = 0;
        freeDyString(&dy);
        }
    if (f)
        carefulClose(&f);

    //debug
    //fprintf(stderr,"p->fileName=%s p->data=[%s]\n",p->fileName,p->data);

    }

return p;
}

Here is the call graph for this function:

Here is the caller graph for this function:

static void putBackMB

(

struct mimeBuf *

b

)

[static]

Definition at line 143 of file mime.c.

References mimeBuf::buf, errAbort(), and mimeBuf::i.

Referenced by getLineMB().

{
if (b->i == b->buf)  /* at beginning of buffer */
    errAbort("putBackMB error - requested pushback beyond beginning buffer.");
b->i--;    
}

Here is the call graph for this function:

Here is the caller graph for this function:

static void readPartHeaderMB	(	struct mimeBuf *	b,
		struct mimePart *	p,
		char *	altHeader
	)			[static]

Definition at line 212 of file mime.c.

References cloneString(), dyStringAppend(), dyStringClear, dyStringFree, dyStringNew, errAbort(), freez(), getLineMB(), hashAdd(), mimePart::hdr, lineFileClose(), lineFileNext(), lineFileOnString(), newHash(), nlt_undet, lineFile::nlType, sameString, dyString::string, tolowers(), trimSpaces(), and TRUE.

Referenced by parseMultiParts().

{
struct dyString *fullLine = dyStringNew(0);
char *key=NULL, *val=NULL;
struct lineFile *lf = NULL;
char *line = NULL;
char *lineAhead = NULL;
int size = 0;
p->hdr = newHash(3);
        //debug
        //fprintf(stderr,"headers dumpMB: ");
        //dumpMB(b);  //debug
if (altHeader)
    {
    lf = lineFileOnString("MIME Header", TRUE, altHeader);
    }
/* read ahead one line, skipping any leading blanks lines */   
do
    {
    if (altHeader)
        lineFileNext(lf, &lineAhead, &size);
    else
        lineAhead = getLineMB(b);
    } 
    while (sameString(lineAhead,""));

do
    {
    /* accumulate a full header line - some emailers split into mpl lines */
    dyStringClear(fullLine);
    do 
        {
        line = lineAhead;
        if (altHeader)
            lineFileNext(lf, &lineAhead, &size);
        else
            lineAhead = getLineMB(b);
        dyStringAppend(fullLine,line);    
        if (!altHeader) 
            freez(&line);
        } while (isspace(lineAhead[0]));
    line = fullLine->string;
    //fprintf(stderr,"found a line! [%s]\n",line);  //debug
    key = line;
    val = strchr(line,':');
    if (!val)
        errAbort("readPartHeaderMB error - header-line colon not found, line=[%s]",line);
    *val = 0;
    val++;
    key=trimSpaces(key);
    // since the hash is case-sensitive, convert to lower case for ease of matching
    tolowers(key);  
    val=trimSpaces(val);
    hashAdd(p->hdr,key,cloneString(val));
    
    //debug
    //fprintf(stderr,"MIME header: key=[%s], val=[%s]\n",key,val);
    //fflush(stderr); 
    
    } while (!sameString(lineAhead,""));
if (altHeader)
    {
    if (nlType == nlt_undet)
        nlType = lf->nlType;
    lineFileClose(&lf);
    }
else
    {
    freez(&lineAhead);
    }
dyStringFree(&fullLine);
    
}

Here is the call graph for this function:

Here is the caller graph for this function:

static void setBoundaryMB	(	struct mimeBuf *	b,
		char *	boundary
	)			[static]

Definition at line 53 of file mime.c.

References mimeBuf::blen, mimeBuf::boundary, setEodMB(), and setEopMB().

Referenced by parseMultiParts().

{
b->boundary = boundary;
b->blen = boundary ? strlen(b->boundary) : 0;
setEopMB(b);
setEodMB(b);
}

Here is the call graph for this function:

Here is the caller graph for this function:

static void setEodMB

(

struct mimeBuf *

b

)

[static]

Definition at line 40 of file mime.c.

References mimeBuf::blen, mimeBuf::eod, mimeBuf::eoi, and mimeBuf::eom.

Referenced by moreMimeBuf(), and setBoundaryMB().

{
if (b->blen > 1 && b->eoi == b->eom) 
    {
    b->eod = b->eoi - (b->blen-1);
    }
else
    {
    b->eod = b->eoi;
    }
}

Here is the caller graph for this function:

static void setEopMB

(

struct mimeBuf *

b

)

[static]

Definition at line 31 of file mime.c.

References mimeBuf::blen, mimeBuf::boundary, mimeBuf::eoi, mimeBuf::eop, mimeBuf::i, and memMatch().

Referenced by moreMimeBuf(), parseMultiParts(), and setBoundaryMB().

{
if (b->blen > 0)
    b->eop = memMatch(b->boundary, b->blen, b->i, b->eoi - b->i);
else
    b->eop = NULL;
}

Here is the call graph for this function:

Here is the caller graph for this function:

---

Variable Documentation

enum nlType nlType = nlt_undet

Definition at line 29 of file mime.c.

char const rcsid[] = "$Id: mime.c,v 1.13 2006/06/20 18:31:24 galt Exp $" [static]

Definition at line 19 of file mime.c.

---