*/
#include <sys/types.h>
-#include <limits.h>
-#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
-#include <expat.h>
#include "pool.h"
#include "repo.h"
#include "tools_util.h"
#include "repo_rpmmd.h"
#include "chksum.h"
-
+#include "solv_xmlparser.h"
+#ifdef ENABLE_COMPLEX_DEPS
+#include "pool_parserpmrichdep.h"
+#endif
+#include "repodata_diskusage.h"
enum state {
STATE_START,
STATE_INSTALL_TIME,
/* product */
- STATE_SHORTNAME,
- STATE_DISTNAME, /* obsolete */
- STATE_DISTEDITION, /* obsolete */
- STATE_SOURCE,
- STATE_TYPE,
STATE_RELNOTESURL,
STATE_UPDATEURL,
STATE_OPTIONALURL,
STATE_FLAG,
- /* rpm-md dependencies inside the
- format tag */
+ /* rpm-md dependencies inside the format tag */
STATE_PROVIDES,
STATE_REQUIRES,
STATE_OBSOLETES,
NUMSTATES
};
-struct stateswitch {
- enum state from;
- char *ename;
- enum state to;
- int docontent;
-};
-
-static struct stateswitch stateswitches[] = {
- /** fake tag used to enclose 2 different xml files in one **/
+static struct solv_xmlparser_element stateswitches[] = {
+ /** fake tag used to enclose multiple xml files in one **/
{ STATE_START, "rpmmd", STATE_START, 0 },
- /** tags for different package data, we just ignore the tag **/
+ /** tags for different package data, just ignore them **/
+ { STATE_START, "patterns", STATE_START, 0 },
+ { STATE_START, "products", STATE_START, 0 },
{ STATE_START, "metadata", STATE_START, 0 },
{ STATE_START, "otherdata", STATE_START, 0 },
{ STATE_START, "filelists", STATE_START, 0 },
{ STATE_START, "patch", STATE_SOLVABLE, 0 },
{ STATE_START, "package", STATE_SOLVABLE, 0 },
+ { STATE_SOLVABLE, "format", STATE_SOLVABLE, 0 },
+
{ STATE_SOLVABLE, "name", STATE_NAME, 1 },
{ STATE_SOLVABLE, "arch", STATE_ARCH, 1 },
{ STATE_SOLVABLE, "version", STATE_VERSION, 0 },
/* product attributes */
/* note the product type is an attribute */
{ STATE_SOLVABLE, "release-notes-url", STATE_RELNOTESURL, 1 },
- { STATE_SOLVABLE, "update-url", STATE_UPDATEURL, 1 },
- { STATE_SOLVABLE, "optional-url", STATE_OPTIONALURL, 1 },
- { STATE_SOLVABLE, "flag", STATE_FLAG, 1 },
+ { STATE_SOLVABLE, "update-url", STATE_UPDATEURL, 1 },
+ { STATE_SOLVABLE, "optional-url", STATE_OPTIONALURL, 1 },
+ { STATE_SOLVABLE, "flag", STATE_FLAG, 1 },
- { STATE_SOLVABLE, "rpm:vendor", STATE_VENDOR, 1 },
- { STATE_SOLVABLE, "rpm:group", STATE_RPM_GROUP, 1 },
- { STATE_SOLVABLE, "rpm:license", STATE_RPM_LICENSE, 1 },
+ { STATE_SOLVABLE, "rpm:vendor", STATE_VENDOR, 1 },
+ { STATE_SOLVABLE, "rpm:group", STATE_RPM_GROUP, 1 },
+ { STATE_SOLVABLE, "rpm:license", STATE_RPM_LICENSE, 1 },
/* rpm-md dependencies */
- { STATE_SOLVABLE, "rpm:provides", STATE_PROVIDES, 0 },
- { STATE_SOLVABLE, "rpm:requires", STATE_REQUIRES, 0 },
- { STATE_SOLVABLE, "rpm:obsoletes", STATE_OBSOLETES, 0 },
- { STATE_SOLVABLE, "rpm:conflicts", STATE_CONFLICTS, 0 },
- { STATE_SOLVABLE, "rpm:recommends", STATE_RECOMMENDS , 0 },
- { STATE_SOLVABLE, "rpm:supplements", STATE_SUPPLEMENTS, 0 },
- { STATE_SOLVABLE, "rpm:suggests", STATE_SUGGESTS, 0 },
- { STATE_SOLVABLE, "rpm:enhances", STATE_ENHANCES, 0 },
- { STATE_SOLVABLE, "rpm:freshens", STATE_FRESHENS, 0 },
- { STATE_SOLVABLE, "rpm:sourcerpm", STATE_SOURCERPM, 1 },
- { STATE_SOLVABLE, "rpm:header-range", STATE_HEADERRANGE, 0 },
- { STATE_SOLVABLE, "file", STATE_FILE, 1 },
- { STATE_SOLVABLE, "changelog", STATE_CHANGELOG, 1 },
+ { STATE_SOLVABLE, "rpm:provides", STATE_PROVIDES, 0 },
+ { STATE_SOLVABLE, "rpm:requires", STATE_REQUIRES, 0 },
+ { STATE_SOLVABLE, "rpm:obsoletes", STATE_OBSOLETES, 0 },
+ { STATE_SOLVABLE, "rpm:conflicts", STATE_CONFLICTS, 0 },
+ { STATE_SOLVABLE, "rpm:recommends", STATE_RECOMMENDS , 0 },
+ { STATE_SOLVABLE, "rpm:supplements", STATE_SUPPLEMENTS, 0 },
+ { STATE_SOLVABLE, "rpm:suggests", STATE_SUGGESTS, 0 },
+ { STATE_SOLVABLE, "rpm:enhances", STATE_ENHANCES, 0 },
+ { STATE_SOLVABLE, "rpm:freshens", STATE_FRESHENS, 0 },
+ { STATE_SOLVABLE, "rpm:sourcerpm", STATE_SOURCERPM, 1 },
+ { STATE_SOLVABLE, "rpm:header-range", STATE_HEADERRANGE, 0 },
+ { STATE_SOLVABLE, "file", STATE_FILE, 1 },
+ { STATE_SOLVABLE, "changelog", STATE_CHANGELOG, 1 },
/* extended Novell/SUSE diskusage attributes (susedata.xml) */
{ STATE_DISKUSAGE, "dirs", STATE_DIRS, 0 },
Repo *repo;
Repodata *data;
char *kind;
- int depth;
- enum state state;
- int statedepth;
- char *content;
- int lcontent;
- int acontent;
- int docontent;
Solvable *solvable;
Offset freshens;
- struct stateswitch *swtab[NUMSTATES];
- enum state sbtab[NUMSTATES];
+
+ struct solv_xmlparser xmlp;
struct joindata jd;
/* temporal to store attribute tag language */
const char *tmplang;
Id chksumtype;
Id handle;
- XML_Parser *parser;
- Id (*dirs)[3]; /* dirid, size, nfiles */
- int ndirs;
+ Queue diskusageq;
const char *language; /* default language */
Id langcache[ID_NUM_INTERNAL]; /* cache for the default language */
Id changelog_handle;
- /** Hash to maps checksums to solv */
- Stringpool cspool;
- /** Cache of known checksums to solvable id */
- Id *cscache;
- /* the current longest index in the table */
- int ncscache;
+ int extending; /* are we extending an existing solvable? */
+ int first; /* first solvable we added */
+ int cshash_filled; /* hash is filled with data */
+
+ Hashtable cshash; /* checksum hash -> offset into csdata */
+ Hashval cshashm; /* hash mask */
+ int ncshash; /* entries used */
+ unsigned char *csdata; /* [len, checksum, id] */
+ int ncsdata; /* used bytes */
};
static Id
return pd->langcache[tag];
}
-static int
-id3_cmp (const void *v1, const void *v2, void *dp)
-{
- Id *i1 = (Id*)v1;
- Id *i2 = (Id*)v2;
- return i1[0] - i2[0];
-}
-
-static void
-commit_diskusage (struct parsedata *pd, Id handle)
-{
- int i;
- Dirpool *dp = &pd->data->dirpool;
- /* Now sort in dirid order. This ensures that parents come before
- their children. */
- if (pd->ndirs > 1)
- solv_sort(pd->dirs, pd->ndirs, sizeof (pd->dirs[0]), id3_cmp, 0);
- /* Substract leaf numbers from all parents to make the numbers
- non-cumulative. This must be done post-order (i.e. all leafs
- adjusted before parents). We ensure this by starting at the end of
- the array moving to the start, hence seeing leafs before parents. */
- for (i = pd->ndirs; i--;)
- {
- Id p = dirpool_parent(dp, pd->dirs[i][0]);
- int j = i;
- for (; p; p = dirpool_parent(dp, p))
- {
- for (; j--;)
- if (pd->dirs[j][0] == p)
- break;
- if (j >= 0)
- {
- if (pd->dirs[j][1] < pd->dirs[i][1])
- pd->dirs[j][1] = 0;
- else
- pd->dirs[j][1] -= pd->dirs[i][1];
- if (pd->dirs[j][2] < pd->dirs[i][2])
- pd->dirs[j][2] = 0;
- else
- pd->dirs[j][2] -= pd->dirs[i][2];
- }
- else
- /* Haven't found this parent in the list, look further if
- we maybe find the parents parent. */
- j = i;
- }
- }
-#if 0
- char sbuf[1024];
- char *buf = sbuf;
- unsigned slen = sizeof (sbuf);
- for (i = 0; i < pd->ndirs; i++)
- {
- dir2str (attr, pd->dirs[i][0], &buf, &slen);
- fprintf (stderr, "have dir %d %d %d %s\n", pd->dirs[i][0], pd->dirs[i][1], pd->dirs[i][2], buf);
- }
- if (buf != sbuf)
- free (buf);
-#endif
- for (i = 0; i < pd->ndirs; i++)
- if (pd->dirs[i][1] || pd->dirs[i][2])
- {
- repodata_add_dirnumnum(pd->data, handle, SOLVABLE_DISKUSAGE, pd->dirs[i][0], pd->dirs[i][1], pd->dirs[i][2]);
- }
- pd->ndirs = 0;
-}
-
-
/*
* makeevr_atts
* parse 'epoch', 'ver' and 'rel', return evr Id
makeevr_atts(Pool *pool, struct parsedata *pd, const char **atts)
{
const char *e, *v, *r, *v2;
- char *c;
+ char *c, *space;
int l;
e = v = r = 0;
else if (!strcmp(*atts, "rel"))
r = atts[1];
}
- if (e && !strcmp(e, "0"))
+ if (e && (!*e || !strcmp(e, "0")))
e = 0;
if (v && !e)
{
l += strlen(v);
if (r)
l += strlen(r) + 1;
- if (l > pd->acontent)
- {
- pd->content = solv_realloc(pd->content, l + 256);
- pd->acontent = l + 256;
- }
- c = pd->content;
+ c = space = solv_xmlparser_contentspace(&pd->xmlp, l);
if (e)
{
strcpy(c, e);
c += strlen(c);
}
*c = 0;
- if (!*pd->content)
+ if (!*space)
return 0;
#if 0
- fprintf(stderr, "evr: %s\n", pd->content);
+ fprintf(stderr, "evr: %s\n", space);
#endif
- return pool_str2id(pool, pd->content, 1);
-}
-
-
-/*
- * find_attr
- * find value for xml attribute
- * I: txt, name of attribute
- * I: atts, list of key/value attributes
- * O: pointer to value of matching key, or NULL
- *
- */
-
-static inline const char *
-find_attr(const char *txt, const char **atts)
-{
- for (; *atts; atts += 2)
- {
- if (!strcmp(*atts, txt))
- return atts[1];
- }
- return 0;
+ return pool_str2id(pool, space, 1);
}
static unsigned int
adddep(Pool *pool, struct parsedata *pd, unsigned int olddeps, const char **atts, int isreq)
{
- Id id, name, marker;
+ Id id, marker;
const char *n, *f, *k;
const char **a;
if (k)
{
int l = strlen(k) + 1 + strlen(n) + 1;
- if (l > pd->acontent)
- {
- pd->content = solv_realloc(pd->content, l + 256);
- pd->acontent = l + 256;
- }
- sprintf(pd->content, "%s:%s", k, n);
- name = pool_str2id(pool, pd->content, 1);
+ char *space = solv_xmlparser_contentspace(&pd->xmlp, l);
+ sprintf(space, "%s:%s", k, n);
+ id = pool_str2id(pool, space, 1);
+ }
+#ifdef ENABLE_COMPLEX_DEPS
+ else if (!f && n[0] == '(')
+ {
+ id = pool_parserpmrichdep(pool, n);
+ if (!id)
+ return olddeps;
}
+#endif
else
- name = pool_str2id(pool, (char *)n, 1);
+ id = pool_str2id(pool, (char *)n, 1);
if (f)
{
Id evr = makeevr_atts(pool, pd, atts);
if (!strcmp(f, flagtab[flags]))
break;
flags = flags < 6 ? flags + 1 : 0;
- id = pool_rel2id(pool, name, evr, flags, 1);
+ id = pool_rel2id(pool, id, evr, flags, 1);
}
- else
- id = name;
#if 0
- fprintf(stderr, "new dep %s%s%s\n", pool_id2str(pool, d), id2rel(pool, d), id2evr(pool, d));
+ fprintf(stderr, "new dep %s\n", pool_dep2str(pool, id));
#endif
return repo_addid_dep(pd->repo, olddeps, id, marker);
}
/*-----------------------------------------------*/
-/* XML callbacks */
-
-/*
- * startElement
- * XML callback
+/* checksum hash functions
+ *
+ * used to look up a solvable with the checksum for solvable extension purposes.
*
*/
-static void XMLCALL
-startElement(void *userData, const char *name, const char **atts)
+static void
+init_cshash(struct parsedata *pd)
{
- struct parsedata *pd = userData;
- Pool *pool = pd->pool;
- Solvable *s = pd->solvable;
- struct stateswitch *sw;
- const char *str;
- Id handle = pd->handle;
- const char *pkgid;
+}
- /* fprintf(stderr, "into %s, from %d, depth %d, statedepth %d\n", name, pd->state, pd->depth, pd->statedepth); */
+static void
+free_cshash(struct parsedata *pd)
+{
+ pd->cshash = solv_free(pd->cshash);
+ pd->ncshash = 0;
+ pd->cshashm = 0;
+ pd->csdata = solv_free(pd->csdata);
+ pd->ncsdata = 0;
+}
- if (pd->depth != pd->statedepth)
- {
- pd->depth++;
- return;
- }
+static inline Hashval
+hashkey(const unsigned char *key, int keyl)
+{
+ return key[0] << 24 | key[1] << 16 | key[2] << 8 | key[3];
+}
- if (pd->state == STATE_START && !strcmp(name, "patterns"))
- return;
- if (pd->state == STATE_START && !strcmp(name, "products"))
- return;
+static void
+rebuild_cshash(struct parsedata *pd)
+{
+ Hashval h, hh, hm;
+ Hashtable ht;
+ unsigned char *d, *de;
+
+ hm = pd->cshashm;
#if 0
- if (pd->state == STATE_START && !strcmp(name, "metadata"))
- return;
+ fprintf(stderr, "rebuild cshash with mask 0x%x\n", hm);
#endif
- if (pd->state == STATE_SOLVABLE && !strcmp(name, "format"))
- return;
+ solv_free(pd->cshash);
+ ht = pd->cshash = (Hashtable)solv_calloc(hm + 1, sizeof(Id));
+ d = pd->csdata;
+ de = d + pd->ncsdata;
+ while (d != de)
+ {
+ h = hashkey(d + 1, d[0] + 1) & hm;
+ hh = HASHCHAIN_START;
+ while (ht[h])
+ h = HASHCHAIN_NEXT(h, hh, hm);
+ ht[h] = d + 1 - pd->csdata;
+ d += 2 + d[0] + sizeof(Id);
+ }
+}
+
+static void
+put_in_cshash(struct parsedata *pd, const unsigned char *key, int keyl, Id id)
+{
+ Hashtable ht;
+ Hashval h, hh, hm;
+ unsigned char *d;
- pd->depth++;
- if (!pd->swtab[pd->state])
+ if (keyl < 4 || keyl > 256)
return;
- for (sw = pd->swtab[pd->state]; sw->from == pd->state; sw++)
- if (!strcmp(sw->ename, name))
- break;
- if (sw->from != pd->state)
+ ht = pd->cshash;
+ hm = pd->cshashm;
+ h = hashkey(key, keyl) & hm;
+ hh = HASHCHAIN_START;
+ if (ht)
{
-#if 0
- fprintf(stderr, "into unknown: %s\n", name);
-#endif
- return;
+ while (ht[h])
+ {
+ unsigned char *d = pd->csdata + ht[h];
+ if (d[-1] == keyl - 1 && !memcmp(key, d, keyl))
+ return; /* XXX: first id wins... */
+ h = HASHCHAIN_NEXT(h, hh, hm);
+ }
+ }
+ /* a new entry. put in csdata */
+ pd->csdata = solv_extend(pd->csdata, pd->ncsdata, 1 + keyl + sizeof(Id), 1, 4095);
+ d = pd->csdata + pd->ncsdata;
+ d[0] = keyl - 1;
+ memcpy(d + 1, key, keyl);
+ memcpy(d + 1 + keyl, &id, sizeof(Id));
+ pd->ncsdata += 1 + keyl + sizeof(Id);
+ if ((Hashval)++pd->ncshash * 2 > hm)
+ {
+ pd->cshashm = pd->cshashm ? (2 * pd->cshashm + 1) : 4095;
+ rebuild_cshash(pd);
}
- pd->state = sw->to;
- pd->docontent = sw->docontent;
- pd->statedepth = pd->depth;
- pd->lcontent = 0;
- *pd->content = 0;
+ else
+ ht[h] = pd->ncsdata - (keyl + sizeof(Id));
+}
- if (!s && pd->state != STATE_SOLVABLE)
+static Id
+lookup_cshash(struct parsedata *pd, const unsigned char *key, int keyl)
+{
+ Hashtable ht;
+ Hashval h, hh, hm;
+
+ if (keyl < 4 || keyl > 256)
+ return 0;
+ ht = pd->cshash;
+ if (!ht)
+ return 0;
+ hm = pd->cshashm;
+ h = hashkey(key, keyl) & hm;
+ hh = HASHCHAIN_START;
+ while (ht[h])
+ {
+ unsigned char *d = pd->csdata + ht[h];
+ if (d[-1] == keyl - 1 && !memcmp(key, d, keyl))
+ {
+ Id id;
+ memcpy(&id, d + keyl, sizeof(Id));
+ return id;
+ }
+ h = HASHCHAIN_NEXT(h, hh, hm);
+ }
+ return 0;
+}
+
+static void
+fill_cshash_from_repo(struct parsedata *pd)
+{
+ Dataiterator di;
+ /* setup join data */
+ dataiterator_init(&di, pd->pool, pd->repo, 0, SOLVABLE_CHECKSUM, 0, 0);
+ while (dataiterator_step(&di))
+ put_in_cshash(pd, (const unsigned char *)di.kv.str, solv_chksum_len(di.key->type), di.solvid);
+ dataiterator_free(&di);
+}
+
+static void
+fill_cshash_from_new_solvables(struct parsedata *pd)
+{
+ Pool *pool = pd->pool;
+ Id cstype = 0;
+ unsigned const char *cs;
+ int i;
+
+ for (i = pd->first; i < pool->nsolvables; i++)
+ {
+ if (pool->solvables[i].repo != pd->repo)
+ continue;
+ cs = repodata_lookup_bin_checksum_uninternalized(pd->data, i, SOLVABLE_CHECKSUM, &cstype);
+ if (cs)
+ put_in_cshash(pd, cs, solv_chksum_len(cstype), i);
+ }
+}
+
+/*-----------------------------------------------*/
+/* XML callbacks */
+
+/*
+ * startElement
+ */
+
+static void
+startElement(struct solv_xmlparser *xmlp, int state, const char *name, const char **atts)
+{
+ struct parsedata *pd = xmlp->userdata;
+ Pool *pool = pd->pool;
+ Solvable *s = pd->solvable;
+ Id handle = pd->handle;
+ const char *str;
+ const char *pkgid;
+
+ if (!s && state != STATE_SOLVABLE)
return;
- switch(pd->state)
+ switch(state)
{
case STATE_SOLVABLE:
pd->kind = 0;
a new solvable but just append the attributes to the existing
one.
*/
- if ((pkgid = find_attr("pkgid", atts)) != NULL)
+ pd->extending = 0;
+ if ((pkgid = solv_xmlparser_find_attr("pkgid", atts)) != NULL)
{
+ unsigned char chk[256];
+ int l;
+ const char *str = pkgid;
+ if (!pd->cshash_filled)
+ {
+ pd->cshash_filled = 1;
+ fill_cshash_from_new_solvables(pd);
+ }
+ handle = 0;
+ /* convert into bin checksum */
+ l = solv_hex2bin(&str, chk, sizeof(chk));
/* look at the checksum cache */
- Id index = stringpool_str2id(&pd->cspool, pkgid, 0);
- if (!index || index >= pd->ncscache || !pd->cscache[index])
+ if (l >= 4 && !pkgid[2 * l])
+ handle = lookup_cshash(pd, chk, l);
+#if 0
+ fprintf(stderr, "Lookup %s -> %d\n", pkgid, handle);
+#endif
+ if (!handle)
{
pool_debug(pool, SOLV_WARN, "the repository specifies extra information about package with checksum '%s', which does not exist in the repository.\n", pkgid);
- pd->solvable = 0;
pd->handle = 0;
+ pd->solvable = 0;
break;
}
- pd->solvable = pool_id2solvable(pool, pd->cscache[index]);
+ pd->extending = 1;
}
- else
+ else
{
/* this is a new package */
- pd->solvable = pool_id2solvable(pool, repo_add_solvable(pd->repo));
+ handle = repo_add_solvable(pd->repo);
+ if (!pd->first)
+ pd->first = handle;
pd->freshens = 0;
}
- pd->handle = pd->solvable - pool->solvables;
+ pd->handle = handle;
+ pd->solvable = pool_id2solvable(pool, handle);
+ if (pd->kind && pd->kind[1] == 'r')
+ {
+ /* products can have a type */
+ const char *type = solv_xmlparser_find_attr("type", atts);
+ if (type && *type)
+ repodata_set_str(pd->data, handle, PRODUCT_TYPE, type);
+ }
#if 0
fprintf(stderr, "package #%d\n", pd->solvable - pool->solvables);
#endif
break;
case STATE_VERSION:
+ if (pd->extending && s->evr)
+ break; /* ignore version tag repetition in extend data */
s->evr = makeevr_atts(pool, pd, atts);
break;
case STATE_PROVIDES:
case STATE_SUMMARY:
case STATE_CATEGORY:
case STATE_DESCRIPTION:
- pd->tmplang = join_dup(&pd->jd, find_attr("lang", atts));
+ pd->tmplang = join_dup(&pd->jd, solv_xmlparser_find_attr("lang", atts));
break;
case STATE_USERVISIBLE:
repodata_set_void(pd->data, handle, SOLVABLE_ISVISIBLE);
break;
case STATE_INCLUDESENTRY:
- str = find_attr("pattern", atts);
+ str = solv_xmlparser_find_attr("pattern", atts);
if (str)
- repodata_add_poolstr_array(pd->data, pd->handle, SOLVABLE_INCLUDES, join2(&pd->jd, "pattern", ":", str));
+ repodata_add_poolstr_array(pd->data, handle, SOLVABLE_INCLUDES, join2(&pd->jd, "pattern", ":", str));
break;
case STATE_EXTENDSENTRY:
- str = find_attr("pattern", atts);
+ str = solv_xmlparser_find_attr("pattern", atts);
if (str)
- repodata_add_poolstr_array(pd->data, pd->handle, SOLVABLE_EXTENDS, join2(&pd->jd, "pattern", ":", str));
+ repodata_add_poolstr_array(pd->data, handle, SOLVABLE_EXTENDS, join2(&pd->jd, "pattern", ":", str));
break;
case STATE_LOCATION:
- str = find_attr("href", atts);
+ str = solv_xmlparser_find_attr("href", atts);
if (str)
- repodata_set_location(pd->data, handle, 0, 0, str);
+ {
+ int medianr = 0;
+ const char *base = solv_xmlparser_find_attr("xml:base", atts);
+ if (base && !strncmp(base, "media:", 6))
+ {
+ /* check for the media number in the fragment */
+ int l = strlen(base);
+ while (l && base[l - 1] >= '0' && base[l - 1] <= '9')
+ l--;
+ if (l && base[l - 1] == '#' && base[l])
+ medianr = atoi(base + l);
+ }
+ repodata_set_location(pd->data, handle, medianr, 0, str);
+ if (base)
+ repodata_set_poolstr(pd->data, handle, SOLVABLE_MEDIABASE, base);
+ }
break;
case STATE_CHECKSUM:
- str = find_attr("type", atts);
+ str = solv_xmlparser_find_attr("type", atts);
pd->chksumtype = str && *str ? solv_chksum_str2type(str) : 0;
if (!pd->chksumtype)
- pd->ret = pool_error(pool, -1, "line %d: unknown checksum type: %s", (unsigned int)XML_GetCurrentLineNumber(*pd->parser), str ? str : "NULL");
+ pd->ret = pool_error(pool, -1, "line %d: unknown checksum type: %s", solv_xmlparser_lineno(xmlp), str ? str : "NULL");
break;
case STATE_TIME:
{
unsigned int t;
- str = find_attr("build", atts);
+ str = solv_xmlparser_find_attr("build", atts);
if (str && (t = atoi(str)) != 0)
repodata_set_num(pd->data, handle, SOLVABLE_BUILDTIME, t);
break;
}
case STATE_SIZE:
- if ((str = find_attr("installed", atts)) != 0)
+ if ((str = solv_xmlparser_find_attr("installed", atts)) != 0)
repodata_set_num(pd->data, handle, SOLVABLE_INSTALLSIZE, strtoull(str, 0, 10));
- if ((str = find_attr("package", atts)) != 0)
+ if ((str = solv_xmlparser_find_attr("package", atts)) != 0)
repodata_set_num(pd->data, handle, SOLVABLE_DOWNLOADSIZE, strtoull(str, 0, 10));
break;
case STATE_HEADERRANGE:
{
unsigned int end;
- str = find_attr("end", atts);
+ str = solv_xmlparser_find_attr("end", atts);
if (str && (end = atoi(str)) != 0)
repodata_set_num(pd->data, handle, SOLVABLE_HEADEREND, end);
break;
case STATE_DIR:
{
long filesz = 0, filenum = 0;
- Id dirid;
- if ((str = find_attr("name", atts)) != 0)
- dirid = repodata_str2dir(pd->data, str, 1);
- else
- {
+ Id did;
+
+ if ((str = solv_xmlparser_find_attr("name", atts)) == 0)
+ {
pd->ret = pool_error(pool, -1, "<dir .../> tag without 'name' attribute");
break;
- }
- if ((str = find_attr("size", atts)) != 0)
+ }
+ if (*str != '/')
+ {
+ if (s->arch == ARCH_SRC || s->arch == ARCH_NOSRC)
+ str = "/usr/src";
+ else
+ {
+ int l = strlen(str) + 2;
+ char *space = solv_xmlparser_contentspace(xmlp, l);
+ space[0] = '/';
+ memcpy(space + 1, str, l - 1);
+ str = space;
+ }
+ }
+ did = repodata_str2dir(pd->data, str, 1);
+ if ((str = solv_xmlparser_find_attr("size", atts)) != 0)
filesz = strtol(str, 0, 0);
- if ((str = find_attr("count", atts)) != 0)
+ if ((str = solv_xmlparser_find_attr("count", atts)) != 0)
filenum = strtol(str, 0, 0);
- pd->dirs = solv_extend(pd->dirs, pd->ndirs, 1, sizeof(pd->dirs[0]), 31);
- pd->dirs[pd->ndirs][0] = dirid;
- pd->dirs[pd->ndirs][1] = filesz;
- pd->dirs[pd->ndirs][2] = filenum;
- pd->ndirs++;
+ if (filesz || filenum)
+ {
+ queue_push(&pd->diskusageq, did);
+ queue_push2(&pd->diskusageq, filesz, filenum);
+ }
break;
}
case STATE_CHANGELOG:
pd->changelog_handle = repodata_new_handle(pd->data);
- if ((str = find_attr("date", atts)) != 0)
+ if ((str = solv_xmlparser_find_attr("date", atts)) != 0)
repodata_set_num(pd->data, pd->changelog_handle, SOLVABLE_CHANGELOG_TIME, strtoull(str, 0, 10));
- if ((str = find_attr("author", atts)) != 0)
+ if ((str = solv_xmlparser_find_attr("author", atts)) != 0)
repodata_set_str(pd->data, pd->changelog_handle, SOLVABLE_CHANGELOG_AUTHOR, str);
break;
default:
/*
* endElement
- * XML callback
- *
*/
-static void XMLCALL
-endElement(void *userData, const char *name)
+static void
+endElement(struct solv_xmlparser *xmlp, int state, char *content)
{
- struct parsedata *pd = userData;
+ struct parsedata *pd = xmlp->userdata;
Pool *pool = pd->pool;
Solvable *s = pd->solvable;
Repo *repo = pd->repo;
Id id;
char *p;
- if (pd->depth != pd->statedepth)
- {
- pd->depth--;
- /* printf("back from unknown %d %d %d\n", pd->state, pd->depth, pd->statedepth); */
- return;
- }
-
- /* ignore patterns & metadata */
- if (pd->state == STATE_START && !strcmp(name, "patterns"))
- return;
- if (pd->state == STATE_START && !strcmp(name, "products"))
- return;
-#if 0
- if (pd->state == STATE_START && !strcmp(name, "metadata"))
- return;
-#endif
- if (pd->state == STATE_SOLVABLE && !strcmp(name, "format"))
- return;
-
- pd->depth--;
- pd->statedepth--;
-
-
if (!s)
- {
- pd->state = pd->sbtab[pd->state];
- pd->docontent = 0;
- return;
- }
+ return;
- switch (pd->state)
+ switch (state)
{
case STATE_SOLVABLE:
+ if (pd->extending)
+ {
+ pd->solvable = 0;
+ break;
+ }
if (pd->kind && !s->name) /* add namespace in case of NULL name */
s->name = pool_str2id(pool, join2(&pd->jd, pd->kind, ":", 0), 1);
if (!s->arch)
s->evr = ID_EMPTY; /* some patterns have this */
if (s->name && s->arch != ARCH_SRC && s->arch != ARCH_NOSRC)
s->provides = repo_addid_dep(repo, s->provides, pool_rel2id(pool, s->name, s->evr, REL_EQ, 1), 0);
- s->supplements = repo_fix_supplements(repo, s->provides, s->supplements, pd->freshens);
- s->conflicts = repo_fix_conflicts(repo, s->conflicts);
+ repo_rewrite_suse_deps(s, pd->freshens);
pd->freshens = 0;
pd->kind = 0;
- pd->solvable = s = 0;
+ pd->solvable = 0;
break;
case STATE_NAME:
if (pd->kind)
- s->name = pool_str2id(pool, join2(&pd->jd, pd->kind, ":", pd->content), 1);
+ s->name = pool_str2id(pool, join2(&pd->jd, pd->kind, ":", content), 1);
else
- s->name = pool_str2id(pool, pd->content, 1);
+ s->name = pool_str2id(pool, content, 1);
break;
case STATE_ARCH:
- s->arch = pool_str2id(pool, pd->content, 1);
+ s->arch = pool_str2id(pool, content, 1);
break;
case STATE_VENDOR:
- s->vendor = pool_str2id(pool, pd->content, 1);
+ s->vendor = pool_str2id(pool, content, 1);
break;
case STATE_RPM_GROUP:
- repodata_set_poolstr(pd->data, handle, SOLVABLE_GROUP, pd->content);
+ repodata_set_poolstr(pd->data, handle, SOLVABLE_GROUP, content);
break;
case STATE_RPM_LICENSE:
- repodata_set_poolstr(pd->data, handle, SOLVABLE_LICENSE, pd->content);
+ repodata_set_poolstr(pd->data, handle, SOLVABLE_LICENSE, content);
break;
case STATE_CHECKSUM:
{
- Id index;
-
- if (!pd->chksumtype)
+ unsigned char chk[256];
+ int l = solv_chksum_len(pd->chksumtype);
+ const char *str = content;
+ if (!l || l > sizeof(chk))
break;
- if (strlen(pd->content) != 2 * solv_chksum_len(pd->chksumtype))
+ if (solv_hex2bin(&str, chk, l) != l || content[2 * l])
{
- pd->ret = pool_error(pool, -1, "line %d: invalid checksum length for %s", (unsigned int)XML_GetCurrentLineNumber(*pd->parser), solv_chksum_type2str(pd->chksumtype));
+ pd->ret = pool_error(pool, -1, "line %u: invalid %s checksum", solv_xmlparser_lineno(xmlp), solv_chksum_type2str(pd->chksumtype));
break;
}
- repodata_set_checksum(pd->data, handle, SOLVABLE_CHECKSUM, pd->chksumtype, pd->content);
- /* we save the checksum to solvable id relationship for extended
- metadata */
- index = stringpool_str2id(&pd->cspool, pd->content, 1 /* create it */);
- if (index >= pd->ncscache)
- {
- pd->cscache = solv_zextend(pd->cscache, pd->ncscache, index + 1 - pd->ncscache, sizeof(Id), 255);
- pd->ncscache = index + 1;
- }
- /* add the checksum to the cache */
- pd->cscache[index] = s - pool->solvables;
+ repodata_set_bin_checksum(pd->data, handle, SOLVABLE_CHECKSUM, pd->chksumtype, chk);
+ /* we save the checksum to solvable id relationship for extending metadata */
+ if (pd->cshash_filled)
+ put_in_cshash(pd, chk, l, s - pool->solvables);
break;
}
case STATE_FILE:
-#if 0
- id = pool_str2id(pool, pd->content, 1);
- s->provides = repo_addid_dep(repo, s->provides, id, SOLVABLE_FILEMARKER);
-#endif
- if ((p = strrchr(pd->content, '/')) != 0)
+ if ((p = strrchr(content, '/')) != 0)
{
*p++ = 0;
- if (pd->lastdir && !strcmp(pd->lastdirstr, pd->content))
+ if (pd->lastdir && !strcmp(pd->lastdirstr, content))
{
id = pd->lastdir;
}
else
{
- int l;
- id = repodata_str2dir(pd->data, pd->content, 1);
- l = strlen(pd->content) + 1;
- if (l > pd->lastdirstrl)
+ int l = p - content;
+ if (l + 1 > pd->lastdirstrl) /* + 1 for the possible leading / we need to insert */
{
pd->lastdirstrl = l + 128;
pd->lastdirstr = solv_realloc(pd->lastdirstr, pd->lastdirstrl);
}
- strcpy(pd->lastdirstr, pd->content);
+ if (content[0] != '/')
+ {
+ pd->lastdirstr[0] = '/';
+ memcpy(pd->lastdirstr + 1, content, l);
+ id = repodata_str2dir(pd->data, pd->lastdirstr, 1);
+ }
+ else
+ id = repodata_str2dir(pd->data, content, 1);
pd->lastdir = id;
+ memcpy(pd->lastdirstr, content, l);
}
}
else
{
- p = pd->content;
- id = 0;
+ p = content;
+ id = repodata_str2dir(pd->data, "/", 1);
}
- if (!id)
- id = repodata_str2dir(pd->data, "/", 1);
repodata_add_dirstr(pd->data, handle, SOLVABLE_FILELIST, id, p);
break;
case STATE_SUMMARY:
- repodata_set_str(pd->data, handle, langtag(pd, SOLVABLE_SUMMARY, pd->tmplang), pd->content);
+ repodata_set_str(pd->data, handle, langtag(pd, SOLVABLE_SUMMARY, pd->tmplang), content);
break;
case STATE_DESCRIPTION:
- set_description_author(pd->data, handle, pd->content, pd);
+ set_description_author(pd->data, handle, content, pd);
break;
case STATE_CATEGORY:
- repodata_set_str(pd->data, handle, langtag(pd, SOLVABLE_CATEGORY, pd->tmplang), pd->content);
+ repodata_set_str(pd->data, handle, langtag(pd, SOLVABLE_CATEGORY, pd->tmplang), content);
break;
case STATE_DISTRIBUTION:
- repodata_set_poolstr(pd->data, handle, SOLVABLE_DISTRIBUTION, pd->content);
+ repodata_set_poolstr(pd->data, handle, SOLVABLE_DISTRIBUTION, content);
break;
case STATE_URL:
- if (pd->content[0])
- repodata_set_str(pd->data, handle, SOLVABLE_URL, pd->content);
+ if (*content)
+ repodata_set_str(pd->data, handle, SOLVABLE_URL, content);
break;
case STATE_PACKAGER:
- if (pd->content[0])
- repodata_set_poolstr(pd->data, handle, SOLVABLE_PACKAGER, pd->content);
+ if (*content)
+ repodata_set_poolstr(pd->data, handle, SOLVABLE_PACKAGER, content);
break;
case STATE_SOURCERPM:
- if (pd->content[0])
- repodata_set_sourcepkg(pd->data, handle, pd->content);
+ if (*content)
+ repodata_set_sourcepkg(pd->data, handle, content);
break;
case STATE_RELNOTESURL:
- if (pd->content[0])
+ if (*content)
{
- repodata_add_poolstr_array(pd->data, pd->handle, PRODUCT_URL, pd->content);
- repodata_add_idarray(pd->data, pd->handle, PRODUCT_URL_TYPE, pool_str2id(pool, "releasenotes", 1));
+ repodata_add_poolstr_array(pd->data, handle, PRODUCT_URL, content);
+ repodata_add_idarray(pd->data, handle, PRODUCT_URL_TYPE, pool_str2id(pool, "releasenotes", 1));
}
break;
case STATE_UPDATEURL:
- if (pd->content[0])
+ if (*content)
{
- repodata_add_poolstr_array(pd->data, pd->handle, PRODUCT_URL, pd->content);
- repodata_add_idarray(pd->data, pd->handle, PRODUCT_URL_TYPE, pool_str2id(pool, "update", 1));
+ repodata_add_poolstr_array(pd->data, handle, PRODUCT_URL, content);
+ repodata_add_idarray(pd->data, handle, PRODUCT_URL_TYPE, pool_str2id(pool, "update", 1));
}
break;
case STATE_OPTIONALURL:
- if (pd->content[0])
+ if (*content)
{
- repodata_add_poolstr_array(pd->data, pd->handle, PRODUCT_URL, pd->content);
- repodata_add_idarray(pd->data, pd->handle, PRODUCT_URL_TYPE, pool_str2id(pool, "optional", 1));
+ repodata_add_poolstr_array(pd->data, handle, PRODUCT_URL, content);
+ repodata_add_idarray(pd->data, handle, PRODUCT_URL_TYPE, pool_str2id(pool, "optional", 1));
}
break;
case STATE_FLAG:
- if (pd->content[0])
- repodata_set_poolstr(pd->data, handle, PRODUCT_FLAGS, pd->content);
+ if (*content)
+ repodata_add_poolstr_array(pd->data, handle, PRODUCT_FLAGS, content);
break;
case STATE_EULA:
- if (pd->content[0])
- repodata_set_str(pd->data, handle, langtag(pd, SOLVABLE_EULA, pd->tmplang), pd->content);
+ if (*content)
+ repodata_set_str(pd->data, handle, langtag(pd, SOLVABLE_EULA, pd->tmplang), content);
break;
case STATE_KEYWORD:
- if (pd->content[0])
- repodata_add_poolstr_array(pd->data, pd->handle, SOLVABLE_KEYWORDS, pd->content);
+ if (*content)
+ repodata_add_poolstr_array(pd->data, handle, SOLVABLE_KEYWORDS, content);
break;
case STATE_DISKUSAGE:
- if (pd->ndirs)
- commit_diskusage(pd, pd->handle);
+ if (pd->diskusageq.count)
+ repodata_add_diskusage(pd->data, handle, &pd->diskusageq);
break;
case STATE_ORDER:
- if (pd->content[0])
- repodata_set_str(pd->data, pd->handle, SOLVABLE_ORDER, pd->content);
+ if (*content)
+ repodata_set_str(pd->data, handle, SOLVABLE_ORDER, content);
break;
case STATE_CHANGELOG:
- repodata_set_str(pd->data, pd->changelog_handle, SOLVABLE_CHANGELOG_DESCRIPTION, pd->content);
- repodata_add_flexarray(pd->data, pd->handle, SOLVABLE_CHANGELOG, pd->changelog_handle);
+ repodata_set_str(pd->data, pd->changelog_handle, SOLVABLE_CHANGELOG_TEXT, content);
+ repodata_add_flexarray(pd->data, handle, SOLVABLE_CHANGELOG, pd->changelog_handle);
pd->changelog_handle = 0;
break;
default:
break;
}
- pd->state = pd->sbtab[pd->state];
- pd->docontent = 0;
- /* fprintf(stderr, "back from known %d %d %d\n", pd->state, pd->depth, pd->statedepth); */
}
-
-/*
- * characterData
- * XML callback
- *
- */
-
-static void XMLCALL
-characterData(void *userData, const XML_Char *s, int len)
+static void
+errorCallback(struct solv_xmlparser *xmlp, const char *errstr, unsigned int line, unsigned int column)
{
- struct parsedata *pd = userData;
- int l;
- char *c;
-
- if (!pd->docontent)
- return;
- l = pd->lcontent + len + 1;
- if (l > pd->acontent)
- {
- pd->content = solv_realloc(pd->content, l + 256);
- pd->acontent = l + 256;
- }
- c = pd->content + pd->lcontent;
- pd->lcontent += len;
- while (len-- > 0)
- *c++ = *s++;
- *c = 0;
+ struct parsedata *pd = xmlp->userdata;
+ pd->ret = pool_error(pd->pool, -1, "repo_rpmmd: %s at line %u:%u", errstr, line, column);
}
/*-----------------------------------------------*/
-/* 'main' */
-
-#define BUFF_SIZE 8192
/*
* repo_add_rpmmd
{
Pool *pool = repo->pool;
struct parsedata pd;
- char buf[BUFF_SIZE];
- int i, l;
- struct stateswitch *sw;
Repodata *data;
unsigned int now;
- XML_Parser parser;
now = solv_timems(0);
data = repo_add_repodata(repo, flags);
memset(&pd, 0, sizeof(pd));
- for (i = 0, sw = stateswitches; sw->from != NUMSTATES; i++, sw++)
- {
- if (!pd.swtab[sw->from])
- pd.swtab[sw->from] = sw;
- pd.sbtab[sw->to] = sw->from;
- }
pd.pool = pool;
pd.repo = repo;
pd.data = data;
- pd.content = solv_malloc(256);
- pd.acontent = 256;
- pd.lcontent = 0;
pd.kind = 0;
pd.language = language && *language && strcmp(language, "en") != 0 ? language : 0;
+ queue_init(&pd.diskusageq);
- /* initialize the string pool where we will store
- the package checksums we know about, to get an Id
- we can use in a cache */
- stringpool_init_empty(&pd.cspool);
+ init_cshash(&pd);
if ((flags & REPO_EXTEND_SOLVABLES) != 0)
{
/* setup join data */
- Dataiterator di;
- dataiterator_init(&di, pool, repo, 0, SOLVABLE_CHECKSUM, 0, 0);
- while (dataiterator_step(&di))
- {
- const char *str;
- int index;
-
- if (!solv_chksum_len(di.key->type))
- continue;
- str = repodata_chk2str(di.data, di.key->type, (const unsigned char *)di.kv.str);
- index = stringpool_str2id(&pd.cspool, str, 1);
- if (index >= pd.ncscache)
- {
- pd.cscache = solv_zextend(pd.cscache, pd.ncscache, index + 1 - pd.ncscache, sizeof(Id), 255);
- pd.ncscache = index + 1;
- }
- pd.cscache[index] = di.solvid;
- }
- dataiterator_free(&di);
+ pd.cshash_filled = 1;
+ fill_cshash_from_repo(&pd);
}
- parser = XML_ParserCreate(NULL);
- XML_SetUserData(parser, &pd);
- pd.parser = &parser;
- XML_SetElementHandler(parser, startElement, endElement);
- XML_SetCharacterDataHandler(parser, characterData);
- for (;;)
- {
- l = fread(buf, 1, sizeof(buf), fp);
- if (XML_Parse(parser, buf, l, l == 0) == XML_STATUS_ERROR)
- {
- pd.ret = pool_error(pool, -1, "repo_rpmmd: %s at line %u:%u", XML_ErrorString(XML_GetErrorCode(parser)), (unsigned int)XML_GetCurrentLineNumber(parser), (unsigned int)XML_GetCurrentColumnNumber(parser));
- break;
- }
- if (l == 0)
- break;
- }
- XML_ParserFree(parser);
- solv_free(pd.content);
+ solv_xmlparser_init(&pd.xmlp, stateswitches, &pd, startElement, endElement, errorCallback);
+ solv_xmlparser_parse(&pd.xmlp, fp);
+ solv_xmlparser_free(&pd.xmlp);
+
solv_free(pd.lastdirstr);
join_freemem(&pd.jd);
- stringpool_free(&pd.cspool);
- solv_free(pd.cscache);
+ free_cshash(&pd);
repodata_free_dircache(data);
+ queue_free(&pd.diskusageq);
if (!(flags & REPO_NO_INTERNALIZE))
repodata_internalize(data);