Annotation of libwww/Library/src/HTAnchor.c, revision 1.9

1.1       timbl       1: /*     Hypertext "Anchor" Object                               HTAnchor.c
                      2: **     ==========================
                      3: **
                      4: ** An anchor represents a region of a hypertext document which is linked to
                      5: ** another anchor in the same or a different document.
                      6: **
                      7: ** History
                      8: **
                      9: **         Nov 1990  Written in Objective-C for the NeXT browser (TBL)
                     10: **     24-Oct-1991 (JFG), written in C, browser-independant 
                     11: **     21-Nov-1991 (JFG), first complete version
                     12: **
                     13: **     (c) Copyright CERN 1991 - See Copyright.html
                     14: */
                     15: 
                     16: #define HASH_SIZE 101          /* Arbitrary prime. Memory/speed tradeoff */
                     17: 
                     18: #include <ctype.h>
                     19: #include "tcp.h"
1.7       luotonen   20: #include "HTFormat.h"
1.1       timbl      21: #include "HTAnchor.h"
1.6       timbl      22: 
1.1       timbl      23: #include "HTUtils.h"
                     24: #include "HTParse.h"
1.6       timbl      25: #include "HTFWriter.h" /* for cache stuff */
1.1       timbl      26: 
                     27: typedef struct _HyperDoc Hyperdoc;
1.5       duns       28: #ifdef VMS
1.1       timbl      29: struct _HyperDoc {
                     30:        int junk;       /* VMS cannot handle pointers to undefined structs */
                     31: };
                     32: #endif
                     33: 
                     34: PRIVATE HTList **adult_table=0;  /* Point to table of lists of all parents */
                     35: 
                     36: /*                             Creation Methods
                     37: **                             ================
                     38: **
                     39: **     Do not use "new" by itself outside this module. In order to enforce
                     40: **     consistency, we insist that you furnish more information about the
                     41: **     anchor you are creating : use newWithParent or newWithAddress.
                     42: */
                     43: 
                     44: PRIVATE HTParentAnchor * HTParentAnchor_new
                     45:   NOARGS
                     46: {
                     47:   HTParentAnchor *newAnchor = 
                     48:     (HTParentAnchor *) calloc (1, sizeof (HTParentAnchor));  /* zero-filled */
                     49:   newAnchor->parent = newAnchor;
                     50:   return newAnchor;
                     51: }
                     52: 
                     53: PRIVATE HTChildAnchor * HTChildAnchor_new
                     54:   NOARGS
                     55: {
                     56:   return (HTChildAnchor *) calloc (1, sizeof (HTChildAnchor));  /* zero-filled */
                     57: }
                     58: 
                     59: 
                     60: /*     Case insensitive string comparison
                     61: **     ----------------------------------
                     62: ** On entry,
                     63: **     s       Points to one string, null terminated
                     64: **     t       points to the other.
                     65: ** On exit,
                     66: **     returns YES if the strings are equivalent ignoring case
                     67: **             NO if they differ in more than  their case.
                     68: */
                     69: 
                     70: PRIVATE BOOL equivalent
                     71:   ARGS2 (CONST char *,s, CONST char *,t)
                     72: {
                     73:   if (s && t) {  /* Make sure they point to something */
                     74:     for ( ; *s && *t ; s++, t++) {
                     75:         if (TOUPPER(*s) != TOUPPER(*t))
                     76:          return NO;
                     77:     }
                     78:     return TOUPPER(*s) == TOUPPER(*t);
                     79:   } else
                     80:     return s == t;  /* Two NULLs are equivalent, aren't they ? */
                     81: }
                     82: 
                     83: 
                     84: /*     Create new or find old sub-anchor
                     85: **     ---------------------------------
                     86: **
1.3       timbl      87: **     Me one is for a new anchor being edited into an existing
1.1       timbl      88: **     document. The parent anchor must already exist.
                     89: */
                     90: 
1.2       timbl      91: PUBLIC HTChildAnchor * HTAnchor_findChild
1.1       timbl      92:   ARGS2 (HTParentAnchor *,parent, CONST char *,tag)
                     93: {
                     94:   HTChildAnchor *child;
                     95:   HTList *kids;
                     96: 
                     97:   if (! parent) {
1.8       frystyk    98:     if (TRACE)
                     99:        fprintf(stderr, "HTAnchor_findChild called with NULL parent.\n");
1.1       timbl     100:     return NULL;
                    101:   }
1.7       luotonen  102:   if ((kids = parent->children)) {  /* parent has children : search them */
1.1       timbl     103:     if (tag && *tag) {         /* TBL */
1.7       luotonen  104:        while ((child = HTList_nextObject (kids))) {
1.1       timbl     105:            if (equivalent(child->tag, tag)) { /* Case sensitive 920226 */
1.4       timbl     106:                if (TRACE) fprintf (stderr,
1.1       timbl     107:               "Child anchor %p of parent %p with name `%s' already exists.\n",
1.4       timbl     108:                    (void*)child, (void*)parent, tag);
1.1       timbl     109:                return child;
                    110:            }
                    111:        }
                    112:      }  /*  end if tag is void */
                    113:   } else  /* parent doesn't have any children yet : create family */
                    114:     parent->children = HTList_new ();
                    115: 
                    116:   child = HTChildAnchor_new ();
                    117:   if (TRACE) fprintf(stderr, "new Anchor %p named `%s' is child of %p\n",
1.4       timbl     118:        (void*)child, (int)tag ? tag : (CONST char *)"" , (void*)parent); /* int for apollo */
1.1       timbl     119:   HTList_addObject (parent->children, child);
                    120:   child->parent = parent;
                    121:   StrAllocCopy(child->tag, tag);
                    122:   return child;
                    123: }
                    124: 
                    125: 
                    126: /*     Create or find a child anchor with a possible link
                    127: **     --------------------------------------------------
                    128: **
                    129: **     Create new anchor with a given parent and possibly
                    130: **     a name, and possibly a link to a _relatively_ named anchor.
                    131: **     (Code originally in ParseHTML.h)
                    132: */
                    133: PUBLIC HTChildAnchor * HTAnchor_findChildAndLink
                    134:   ARGS4(
                    135:        HTParentAnchor *,parent,        /* May not be 0 */
                    136:        CONST char *,tag,       /* May be "" or 0 */
                    137:        CONST char *,href,      /* May be "" or 0 */
                    138:        HTLinkType *,ltype      /* May be 0 */
                    139:        )
                    140: {
                    141:   HTChildAnchor * child = HTAnchor_findChild(parent, tag);
                    142:   if (href && *href) {
1.2       timbl     143:     char * relative_to = HTAnchor_address((HTAnchor *) parent);
                    144:     char * parsed_address = HTParse(href, relative_to, PARSE_ALL);
                    145:     HTAnchor * dest = HTAnchor_findAddress(parsed_address);
1.1       timbl     146:     HTAnchor_link((HTAnchor *) child, dest, ltype);
                    147:     free(parsed_address);
1.2       timbl     148:     free(relative_to);
1.1       timbl     149:   }
                    150:   return child;
                    151: }
                    152: 
                    153: 
                    154: /*     Create new or find old named anchor
                    155: **     -----------------------------------
                    156: **
1.3       timbl     157: **     Me one is for a reference which is found in a document, and might
1.1       timbl     158: **     not be already loaded.
                    159: **     Note: You are not guaranteed a new anchor -- you might get an old one,
                    160: **     like with fonts.
                    161: */
                    162: 
                    163: HTAnchor * HTAnchor_findAddress
                    164:   ARGS1 (CONST char *,address)
                    165: {
                    166:   char *tag = HTParse (address, "", PARSE_ANCHOR);  /* Anchor tag specified ? */
                    167: 
                    168:   /* If the address represents a sub-anchor, we recursively load its parent,
                    169:      then we create a child anchor within that document. */
                    170:   if (*tag) {
                    171:     char *docAddress = HTParse(address, "", PARSE_ACCESS | PARSE_HOST |
                    172:                                            PARSE_PATH | PARSE_PUNCTUATION);
                    173:     HTParentAnchor * foundParent =
                    174:       (HTParentAnchor *) HTAnchor_findAddress (docAddress);
                    175:     HTChildAnchor * foundAnchor = HTAnchor_findChild (foundParent, tag);
                    176:     free (docAddress);
                    177:     free (tag);
                    178:     return (HTAnchor *) foundAnchor;
                    179:   }
                    180:   
1.3       timbl     181:   else { /* If the address has no anchor tag, 
                    182:            check whether we have this node */
1.1       timbl     183:     int hash;
                    184:     CONST char *p;
                    185:     HTList * adults;
                    186:     HTList *grownups;
                    187:     HTParentAnchor * foundAnchor;
                    188: 
                    189:     free (tag);
                    190:     
                    191:     /* Select list from hash table */
1.2       timbl     192:     for(p=address, hash=0; *p; p++)
                    193:        hash = (hash * 3 + (*(unsigned char*)p))
                    194:         % HASH_SIZE;
1.1       timbl     195:     if (!adult_table)
                    196:         adult_table = (HTList**) calloc(HASH_SIZE, sizeof(HTList*));
                    197:     if (!adult_table[hash]) adult_table[hash] = HTList_new();
                    198:     adults = adult_table[hash];
                    199: 
                    200:     /* Search list for anchor */
                    201:     grownups = adults;
1.7       luotonen  202:     while ((foundAnchor = HTList_nextObject (grownups))) {
1.2       timbl     203:        if (equivalent(foundAnchor->address, address)) {
1.8       frystyk   204:        if (TRACE)
                    205:            fprintf(stderr, "Anchor %p with address `%s' already exists.\n",
                    206:                    (void*) foundAnchor, address);
1.1       timbl     207:        return (HTAnchor *) foundAnchor;
                    208:       }
                    209:     }
                    210:     
                    211:     /* Node not found : create new anchor */
                    212:     foundAnchor = HTParentAnchor_new ();
                    213:     if (TRACE) fprintf(stderr, "New anchor %p has hash %d and address `%s'\n",
1.8       frystyk   214:                       (void*)foundAnchor, hash, address);
1.1       timbl     215:     StrAllocCopy(foundAnchor->address, address);
                    216:     HTList_addObject (adults, foundAnchor);
                    217:     return (HTAnchor *) foundAnchor;
                    218:   }
                    219: }
                    220: 
                    221: 
                    222: /*     Delete an anchor and possibly related things (auto garbage collection)
                    223: **     --------------------------------------------
                    224: **
                    225: **     The anchor is only deleted if the corresponding document is not loaded.
                    226: **     All outgoing links from parent and children are deleted, and this anchor
                    227: **     is removed from the sources list of all its targets.
                    228: **     We also try to delete the targets whose documents are not loaded.
                    229: **     If this anchor's source list is empty, we delete it and its children.
                    230: */
                    231: 
                    232: PRIVATE void deleteLinks
1.3       timbl     233:   ARGS1 (HTAnchor *,me)
1.1       timbl     234: {
1.3       timbl     235:   if (! me)
1.1       timbl     236:     return;
                    237: 
                    238:   /* Recursively try to delete target anchors */
1.3       timbl     239:   if (me->mainLink.dest) {
                    240:     HTParentAnchor *parent = me->mainLink.dest->parent;
                    241:     HTList_removeObject (parent->sources, me);
1.1       timbl     242:     if (! parent->document)  /* Test here to avoid calling overhead */
                    243:       HTAnchor_delete (parent);
                    244:   }
1.3       timbl     245:   if (me->links) {  /* Extra destinations */
1.1       timbl     246:     HTLink *target;
1.7       luotonen  247:     while ((target = HTList_removeLastObject (me->links))) {
1.1       timbl     248:       HTParentAnchor *parent = target->dest->parent;
1.3       timbl     249:       HTList_removeObject (parent->sources, me);
1.1       timbl     250:       if (! parent->document)  /* Test here to avoid calling overhead */
                    251:        HTAnchor_delete (parent);
                    252:     }
                    253:   }
                    254: }
                    255: 
                    256: PUBLIC BOOL HTAnchor_delete
1.3       timbl     257:   ARGS1 (HTParentAnchor *,me)
1.1       timbl     258: {
                    259:   HTChildAnchor *child;
                    260: 
                    261:   /* Don't delete if document is loaded */
1.3       timbl     262:   if (me->document)
1.1       timbl     263:     return NO;
                    264: 
                    265:   /* Recursively try to delete target anchors */
1.3       timbl     266:   deleteLinks ((HTAnchor *) me);
1.1       timbl     267: 
1.3       timbl     268:   if (! HTList_isEmpty (me->sources)) {  /* There are still incoming links */
1.1       timbl     269:     /* Delete all outgoing links from children, if any */
1.3       timbl     270:     HTList *kids = me->children;
1.7       luotonen  271:     while ((child = HTList_nextObject (kids)))
1.1       timbl     272:       deleteLinks ((HTAnchor *) child);
                    273:     return NO;  /* Parent not deleted */
                    274:   }
                    275: 
                    276:   /* No more incoming links : kill everything */
                    277:   /* First, recursively delete children */
1.7       luotonen  278:   while ((child = HTList_removeLastObject (me->children))) {
1.1       timbl     279:     deleteLinks ((HTAnchor *) child);
                    280:     free (child->tag);
                    281:     free (child);
                    282:   }
                    283: 
                    284:   /* Now kill myself */
1.3       timbl     285:   HTList_delete (me->children);
                    286:   HTList_delete (me->sources);
                    287:   free (me->address);
1.1       timbl     288:   /* Devise a way to clean out the HTFormat if no longer needed (ref count?) */
1.3       timbl     289:   free (me);
1.6       timbl     290:   if (me->cacheItems) {
                    291:       HTCacheClear(me->cacheItems);
                    292:   }
1.1       timbl     293:   return YES;  /* Parent deleted */
                    294: }
                    295: 
                    296: 
                    297: /*             Move an anchor to the head of the list of its siblings
                    298: **             ------------------------------------------------------
                    299: **
                    300: **     This is to ensure that an anchor which might have already existed
                    301: **     is put in the correct order as we load the document.
                    302: */
                    303: 
                    304: void HTAnchor_makeLastChild
1.3       timbl     305:   ARGS1(HTChildAnchor *,me)
1.1       timbl     306: {
1.3       timbl     307:   if (me->parent != (HTParentAnchor *) me) {  /* Make sure it's a child */
                    308:     HTList * siblings = me->parent->children;
                    309:     HTList_removeObject (siblings, me);
                    310:     HTList_addObject (siblings, me);
1.1       timbl     311:   }
                    312: }
                    313: 
                    314: /*     Data access functions
                    315: **     ---------------------
                    316: */
                    317: 
                    318: PUBLIC HTParentAnchor * HTAnchor_parent
1.3       timbl     319:   ARGS1 (HTAnchor *,me)
1.1       timbl     320: {
1.3       timbl     321:   return me ? me->parent : NULL;
1.1       timbl     322: }
                    323: 
                    324: void HTAnchor_setDocument
1.3       timbl     325:   ARGS2 (HTParentAnchor *,me, HyperDoc *,doc)
1.1       timbl     326: {
1.3       timbl     327:   if (me)
                    328:     me->document = doc;
1.1       timbl     329: }
                    330: 
                    331: HyperDoc * HTAnchor_document
1.3       timbl     332:   ARGS1 (HTParentAnchor *,me)
1.1       timbl     333: {
1.3       timbl     334:   return me ? me->document : NULL;
1.1       timbl     335: }
                    336: 
                    337: 
                    338: /* We don't want code to change an address after anchor creation... yet ?
                    339: void HTAnchor_setAddress
1.3       timbl     340:   ARGS2 (HTAnchor *,me, char *,addr)
1.1       timbl     341: {
1.3       timbl     342:   if (me)
                    343:     StrAllocCopy (me->parent->address, addr);
1.1       timbl     344: }
                    345: */
                    346: 
                    347: char * HTAnchor_address
1.3       timbl     348:   ARGS1 (HTAnchor *,me)
1.1       timbl     349: {
                    350:   char *addr = NULL;
1.3       timbl     351:   if (me) {
                    352:     if (((HTParentAnchor *) me == me->parent) ||
                    353:        !((HTChildAnchor *) me)->tag) {  /* it's an adult or no tag */
                    354:       StrAllocCopy (addr, me->parent->address);
1.1       timbl     355:     }
                    356:     else {  /* it's a named child */
1.3       timbl     357:       addr = malloc (2 + strlen (me->parent->address)
                    358:                     + strlen (((HTChildAnchor *) me)->tag));
1.1       timbl     359:       if (addr == NULL) outofmem(__FILE__, "HTAnchor_address");
1.3       timbl     360:       sprintf (addr, "%s#%s", me->parent->address,
                    361:               ((HTChildAnchor *) me)->tag);
1.1       timbl     362:     }
                    363:   }
                    364:   return addr;
                    365: }
                    366: 
                    367: 
                    368: 
                    369: void HTAnchor_setFormat
1.3       timbl     370:   ARGS2 (HTParentAnchor *,me, HTFormat ,form)
1.1       timbl     371: {
1.3       timbl     372:   if (me)
                    373:     me->format = form;
1.1       timbl     374: }
                    375: 
1.2       timbl     376: HTFormat HTAnchor_format
1.3       timbl     377:   ARGS1 (HTParentAnchor *,me)
1.1       timbl     378: {
1.3       timbl     379:   return me ? me->format : NULL;
1.1       timbl     380: }
                    381: 
1.9     ! frystyk   382: void HTAnchor_clearIndex
        !           383:   ARGS1 (HTParentAnchor *,me)
        !           384: {
        !           385:   if (me)
        !           386:     me->isIndex = NO;
        !           387: }
1.1       timbl     388: 
                    389: void HTAnchor_setIndex
1.3       timbl     390:   ARGS1 (HTParentAnchor *,me)
1.1       timbl     391: {
1.3       timbl     392:   if (me)
                    393:     me->isIndex = YES;
1.1       timbl     394: }
                    395: 
                    396: BOOL HTAnchor_isIndex
1.3       timbl     397:   ARGS1 (HTParentAnchor *,me)
1.1       timbl     398: {
1.3       timbl     399:   return me ? me->isIndex : NO;
1.1       timbl     400: }
                    401: 
                    402: 
                    403: 
                    404: BOOL HTAnchor_hasChildren
1.3       timbl     405:   ARGS1 (HTParentAnchor *,me)
1.1       timbl     406: {
1.3       timbl     407:   return me ? ! HTList_isEmpty(me->children) : NO;
1.1       timbl     408: }
                    409: 
                    410: /*     Title handling
                    411: */
                    412: CONST char * HTAnchor_title
1.3       timbl     413:   ARGS1 (HTParentAnchor *,me)
1.1       timbl     414: {
1.3       timbl     415:   return me ? me->title : 0;
1.1       timbl     416: }
                    417: 
                    418: void HTAnchor_setTitle
1.3       timbl     419:   ARGS2(HTParentAnchor *,me, CONST char *,title)
1.1       timbl     420: {
1.3       timbl     421:   StrAllocCopy(me->title, title);
1.1       timbl     422: }
                    423: 
                    424: void HTAnchor_appendTitle
1.3       timbl     425:   ARGS2(HTParentAnchor *,me, CONST char *,title)
1.1       timbl     426: {
1.3       timbl     427:   StrAllocCat(me->title, title);
1.1       timbl     428: }
                    429: 
1.3       timbl     430: /*     Link me Anchor to another given one
1.1       timbl     431: **     -------------------------------------
                    432: */
                    433: 
                    434: BOOL HTAnchor_link
                    435:   ARGS3(HTAnchor *,source, HTAnchor *,destination, HTLinkType *,type)
                    436: {
                    437:   if (! (source && destination))
                    438:     return NO;  /* Can't link to/from non-existing anchor */
1.8       frystyk   439:   if (TRACE)
                    440:       fprintf(stderr, "Linking anchor %p to anchor %p\n",
                    441:              (void *) source, (void *) destination);
1.1       timbl     442:   if (! source->mainLink.dest) {
                    443:     source->mainLink.dest = destination;
                    444:     source->mainLink.type = type;
                    445:   } else {
                    446:     HTLink * newLink = (HTLink *) malloc (sizeof (HTLink));
                    447:     if (newLink == NULL) outofmem(__FILE__, "HTAnchor_link");
                    448:     newLink->dest = destination;
                    449:     newLink->type = type;
                    450:     if (! source->links)
                    451:       source->links = HTList_new ();
                    452:     HTList_addObject (source->links, newLink);
                    453:   }
                    454:   if (!destination->parent->sources)
                    455:     destination->parent->sources = HTList_new ();
                    456:   HTList_addObject (destination->parent->sources, source);
                    457:   return YES;  /* Success */
                    458: }
                    459: 
                    460: 
                    461: /*     Manipulation of links
                    462: **     ---------------------
                    463: */
                    464: 
                    465: HTAnchor * HTAnchor_followMainLink
1.3       timbl     466:   ARGS1 (HTAnchor *,me)
1.1       timbl     467: {
1.3       timbl     468:   return me->mainLink.dest;
1.1       timbl     469: }
                    470: 
                    471: HTAnchor * HTAnchor_followTypedLink
1.3       timbl     472:   ARGS2 (HTAnchor *,me, HTLinkType *,type)
1.1       timbl     473: {
1.3       timbl     474:   if (me->mainLink.type == type)
                    475:     return me->mainLink.dest;
                    476:   if (me->links) {
                    477:     HTList *links = me->links;
1.1       timbl     478:     HTLink *link;
1.7       luotonen  479:     while ((link = HTList_nextObject (links)))
1.1       timbl     480:       if (link->type == type)
                    481:        return link->dest;
                    482:   }
1.3       timbl     483:   return NULL;  /* No link of me type */
1.1       timbl     484: }
                    485: 
1.2       timbl     486: 
                    487: /*     Make main link
                    488: */
1.1       timbl     489: BOOL HTAnchor_makeMainLink
1.3       timbl     490:   ARGS2 (HTAnchor *,me, HTLink *,movingLink)
1.1       timbl     491: {
                    492:   /* Check that everything's OK */
1.3       timbl     493:   if (! (me && HTList_removeObject (me->links, movingLink)))
1.1       timbl     494:     return NO;  /* link not found or NULL anchor */
                    495:   else {
                    496:     /* First push current main link onto top of links list */
                    497:     HTLink *newLink = (HTLink*) malloc (sizeof (HTLink));
                    498:     if (newLink == NULL) outofmem(__FILE__, "HTAnchor_makeMainLink");
1.3       timbl     499:     memcpy (newLink, & me->mainLink, sizeof (HTLink));
                    500:     HTList_addObject (me->links, newLink);
1.1       timbl     501: 
                    502:     /* Now make movingLink the new main link, and free it */
1.3       timbl     503:     memcpy (& me->mainLink, movingLink, sizeof (HTLink));
1.1       timbl     504:     free (movingLink);
                    505:     return YES;
                    506:   }
1.2       timbl     507: }
                    508: 
                    509: 
                    510: /*     Methods List
                    511: **     ------------
                    512: */
                    513: 
1.3       timbl     514: PUBLIC HTList * HTAnchor_methods ARGS1(HTParentAnchor *, me)
1.2       timbl     515: {
1.3       timbl     516:     if (!me->methods) {
                    517:         me->methods = HTList_new();
1.2       timbl     518:     }
1.3       timbl     519:     return me->methods;
1.2       timbl     520: }
                    521: 
                    522: /*     Protocol
                    523: **     --------
                    524: */
                    525: 
1.3       timbl     526: PUBLIC void * HTAnchor_protocol ARGS1(HTParentAnchor *, me)
1.2       timbl     527: {
1.3       timbl     528:     return me->protocol;
1.2       timbl     529: }
                    530: 
1.3       timbl     531: PUBLIC void HTAnchor_setProtocol ARGS2(HTParentAnchor *, me,
1.2       timbl     532:        void*,  protocol)
                    533: {
1.3       timbl     534:     me->protocol = protocol;
1.2       timbl     535: }
                    536: 
                    537: /*     Physical Address
                    538: **     ----------------
                    539: */
                    540: 
1.3       timbl     541: PUBLIC char * HTAnchor_physical ARGS1(HTParentAnchor *, me)
1.2       timbl     542: {
1.3       timbl     543:     return me->physical;
1.2       timbl     544: }
                    545: 
1.3       timbl     546: PUBLIC void HTAnchor_setPhysical ARGS2(HTParentAnchor *, me,
1.2       timbl     547:        char *, physical)
                    548: {
1.9     ! frystyk   549:     if (!me || !physical) {
        !           550:        if (TRACE)
        !           551:            fprintf(stderr, "HTAnchor.... setPhysical, called with null argument\n");
        !           552:        return;
        !           553:     }
1.3       timbl     554:     StrAllocCopy(me->physical, physical);
1.1       timbl     555: }

Webmaster