Changeset: c416436c479c for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=c416436c479c
Modified Files:
pathfinder/src/sqlhelpers/xmlshred/encoding.c
pathfinder/src/sqlhelpers/xmlshred/guides.c
pathfinder/src/sqlhelpers/xmlshred/include/encoding.h
pathfinder/src/sqlhelpers/xmlshred/include/guides.h
pathfinder/src/sqlhelpers/xmlshred/include/shred_helper.h
pathfinder/src/sqlhelpers/xmlshred/main.c
Branch: default
Log Message:
Various extensions to pfshred:
- Added support for arbitrarily long tagnames and textnodes.
- Extended guide information with unique name and uri ids
(cf., format options '%n' and '%u').
- Fix error when generating files without format option '%g'.
- Added new format options ('%X', '%U', '%N', '%r').
- Added option (-e) to disable string escaping.
- Added option (-x) to generate output for x100.
diffs (truncated from 770 to 300 lines):
diff --git a/pathfinder/src/sqlhelpers/xmlshred/encoding.c
b/pathfinder/src/sqlhelpers/xmlshred/encoding.c
--- a/pathfinder/src/sqlhelpers/xmlshred/encoding.c
+++ b/pathfinder/src/sqlhelpers/xmlshred/encoding.c
@@ -58,14 +58,20 @@
/* count the tags/text being truncated
* during the shredding process */
-unsigned int tag_stripped;
+unsigned int prefix_stripped;
+unsigned int name_stripped;
+unsigned int uri_stripped;
unsigned int text_stripped;
static shred_state_t shredstate;
/* Wrapper to ensure that guide information is only collected if needed */
-#define insert_guide_node_(u,n,p,k) ((shredstate.statistics) \
- ? insert_guide_node ((u),(n),(p),(k)) \
+#define insert_guide_node_(u,n,p,k) ((shredstate.statistics)
\
+ ? insert_guide_node (
\
+ (u),(n),
\
+ generate_uri_id ((xmlChar *) u),
\
+ generate_localname_id ((xmlChar *) n),
\
+ (p),(k))
\
: NULL)
#define guide_val_(g) ((shredstate.statistics) \
? (g)->guide \
@@ -78,20 +84,15 @@
#define BAILOUT(...) do { SHoops (SH_FATAL, __VA_ARGS__); \
free_hashtable (localname_hash); \
free_hashtable (uris_hash); \
+ free (buffer); \
exit (1); \
} while (0)
-/* id of empty namespace prefix */
-#define EMPTY_NS 0
-
-/* maximum localname/URI and text buffer sizes */
-#define TAG_SIZE 100
-#define BUF_SIZE 4096
-
/* buffer for XML node contents/values */
-static xmlChar buf[BUF_SIZE + 1];
-static int bufpos;
+static unsigned int bufsize;
+static xmlChar *buffer;
+static unsigned int bufpos;
static node_t stack[STACK_MAX];
static int level;
@@ -160,12 +161,18 @@
void
print_value (FILE *f, node_t tuple)
{
+ if (tuple.value)
+ fputs (tuple.value, f);
+}
+
+void
+print_value_escaped (FILE *f, node_t tuple)
+{
/* escape quotes in quoted string */
if (tuple.value) {
char *src = (char *) tuple.value;
unsigned int start = 0, end;
- putc ('"', f);
- for (end = 0; end < text_size && tuple.value[end]; end++)
+ for (end = 0; tuple.value[end]; end++)
/* escape quotes */
if (tuple.value[end] == '"') {
print_text (f, &src[start], end - start);
@@ -174,13 +181,9 @@
}
if (start < end)
print_text (f, &src[start], end - start);
- putc ('"', f);
}
}
-void (*print_localname) (FILE *, node_t);
-void (*print_uri) (FILE *, node_t);
-
void
print_localname_id (FILE *f, node_t tuple)
{
@@ -190,10 +193,16 @@
void
print_localname_str (FILE *f, node_t tuple)
{
- fprintf (f, "\"%s\"", (char *) tuple.localname);
+ fputs ((char *) tuple.localname, f);
}
void
+print_prefix_str (FILE *f, node_t tuple)
+{
+ fputs ((char *) tuple.prefix, f);
+}
+
+void
print_uri_id (FILE *f, node_t tuple)
{
fprintf (f, "%i", tuple.uri_id);
@@ -202,7 +211,7 @@
void
print_uri_str (FILE *f, node_t tuple)
{
- fprintf (f, "\"%s\"", (char *) tuple.uri);
+ fputs ((char *) tuple.uri, f);
}
/* implementation of formatting functions */
@@ -241,11 +250,20 @@
static void lambda_P (node_t n)
{ if (n.parent) lambda_E (*(n.parent)); }
+static void lambda_X (node_t n)
+{ if (n.uri_id != -1) print_prefix_str (out, n); }
+
static void lambda_n (node_t n)
-{ if (n.localname_id != -1) print_localname (out, n); }
+{ if (n.localname_id != -1) print_localname_id (out, n); }
+
+static void lambda_N (node_t n)
+{ if (n.localname_id != -1) print_localname_str (out, n); }
static void lambda_u (node_t n)
-{ if (n.uri_id != -1) print_uri (out, n); }
+{ if (n.uri_id != -1) print_uri_id (out, n); }
+
+static void lambda_U (node_t n)
+{ if (n.uri_id != -1) print_uri_str (out, n); }
static void lambda_d (node_t n)
{ print_number (out, n); }
@@ -253,8 +271,10 @@
static void lambda_r (node_t n)
{ fprintf (out, SSZFMT, n.root); }
+void (*print_textnode) (FILE *, node_t);
+
static void lambda_t (node_t n)
-{ print_value (out, n); }
+{ print_textnode (out, n); }
static void lambda_g (node_t n)
{ fprintf (out, SSZFMT, guide_val_ (n.guide)); }
@@ -275,7 +295,10 @@
, ['p'] = lambda_p
, ['P'] = lambda_P
, ['n'] = lambda_n
+ , ['X'] = lambda_X
+ , ['N'] = lambda_N
, ['u'] = lambda_u
+ , ['U'] = lambda_U
, ['d'] = lambda_d
, ['r'] = lambda_r
, ['t'] = lambda_t
@@ -381,11 +404,14 @@
static void
flush_node (kind_t kind,
- const xmlChar *URI, const xmlChar *localname,
+ const xmlChar *localname,
+ const xmlChar *prefix, const xmlChar *URI,
const xmlChar *value)
{
- char *short_value = NULL;
- int valStrLen = -1;
+ char *short_value = NULL,
+ *prefix_copy = NULL,
+ *localname_copy = NULL,
+ *URI_copy = NULL;
pre++;
rank++;
@@ -393,18 +419,30 @@
max_level = MAX(level, max_level);
- /* check if tagname is larger than TAG_SIZE characters */
- if (localname && xmlStrlen (localname) > TAG_SIZE)
- BAILOUT ("attribute local name `%s' exceeds %u characters",
- localname, TAG_SIZE);
+ /* check if prefix is larger than text_size characters */
+ if (text_size && prefix && (unsigned int) xmlStrlen (prefix) > text_size) {
+ prefix_stripped++;
+ prefix_copy = strndup ((char *) prefix, text_size);
+ } else if (prefix)
+ prefix_copy = strdup ((char *) prefix);
+
+ /* check if tagname is larger than text_size characters */
+ if (text_size && localname &&
+ (unsigned int) xmlStrlen (localname) > text_size) {
+ name_stripped++;
+ localname_copy = strndup ((char *) localname, text_size);
+ } else if (localname)
+ localname_copy = strdup ((char *) localname);
- if (URI && xmlStrlen (URI) > TAG_SIZE)
- BAILOUT ("namespace URI `%s' exceeds length of %u characters",
- URI, TAG_SIZE);
+ /* check if uri is larger than text_size characters */
+ if (text_size && URI && (unsigned int) xmlStrlen (URI) > text_size) {
+ name_stripped++;
+ URI_copy = strndup ((char *) URI, text_size);
+ } else if (URI)
+ URI_copy = strdup ((char *) URI);
/* check if value is larger than text_size characters */
- if (value && (valStrLen = xmlStrlen (value)) >= 0 &&
- (unsigned int) valStrLen > text_size) {
+ if (text_size && value && (unsigned int) xmlStrlen (value) > text_size) {
text_stripped++;
short_value = strndup ((char *) value, text_size);
}
@@ -422,9 +460,10 @@
, .children = 0
, .level = level
, .kind = kind
- , .localname = localname ? strdup ((char *) localname) : NULL
+ , .prefix = prefix_copy
+ , .localname = localname_copy
, .localname_id = generate_localname_id (localname)
- , .uri = URI ? strdup ((char *) URI) : NULL
+ , .uri = URI_copy
, .uri_id = generate_uri_id (URI)
, .value = short_value ? short_value : (char *) value
, .guide = insert_guide_node_ ((char *) URI, (char *) localname,
@@ -437,12 +476,16 @@
apply_fmt (stack[level]);
- if (stack[level].localname) {
- free (stack[level].localname);
+ if (prefix_copy) {
+ free (prefix_copy);
+ stack[level].prefix = NULL;
+ }
+ if (localname_copy) {
+ free (localname_copy);
stack[level].localname = NULL;
}
- if (stack[level].uri) {
- free (stack[level].uri);
+ if (URI_copy) {
+ free (URI_copy);
stack[level].uri = NULL;
}
if (short_value) {
@@ -458,23 +501,29 @@
static void
flush_buffer (void)
{
- if (buf[0]) {
- flush_node (text, NULL, NULL, buf);
+ if (bufpos) {
+ flush_node (text, NULL, NULL, NULL, buffer);
}
- buf[0] = '\0';
+ buffer[0] = '\0';
bufpos = 0;
}
static void
start_document (void *ctx)
{
+ char *doc_name;
+
(void) ctx;
root = pre;
- if (strlen (shredstate.doc_name) > MIN (FILENAME_MAX, text_size))
+ if (text_size && strlen (shredstate.doc_name) > text_size) {
text_stripped++;
+ doc_name = strndup (shredstate.doc_name, text_size);
+ } else
+ doc_name = strdup (shredstate.doc_name);
+
/* create a new node */
stack[level] = (node_t) {
@@ -488,15 +537,17 @@
, .children = 0
, .level = level
, .kind = doc
+ , .prefix = NULL
, .localname = NULL
, .localname_id = -1
, .uri = NULL
, .uri_id = -1
- , .value = strdup (shredstate.doc_name)
+ , .value = doc_name
, .guide = guide
};
/* extend guide counter */
_______________________________________________
Checkin-list mailing list
Checkin-list@monetdb.org
http://mail.monetdb.org/mailman/listinfo/checkin-list
