From 934a9aeab34586b42ddaa9dfae6a03f9e0507e2c Mon Sep 17 00:00:00 2001 From: Michael R Sweet Date: Sun, 16 May 2004 18:25:20 +0000 Subject: [PATCH] Cleanup quicksort code to preserve the original order of nodes when they are equal. Fix mxmlIndexFind() implementation. Add tests. --- mxml-index.c | 191 ++++++++++++++++++++++++++++++++++++++++++++------- testmxml.c | 133 +++++++++++++++++++++++++++++++++-- 2 files changed, 293 insertions(+), 31 deletions(-) diff --git a/mxml-index.c b/mxml-index.c index 632a841..494d73c 100644 --- a/mxml-index.c +++ b/mxml-index.c @@ -1,5 +1,5 @@ /* - * "$Id: mxml-index.c,v 1.3 2004/05/16 13:45:56 mike Exp $" + * "$Id: mxml-index.c,v 1.4 2004/05/16 18:25:20 mike Exp $" * * Index support code for Mini-XML, a small XML-like file parsing library. * @@ -122,12 +122,24 @@ mxmlIndexFind(mxml_index_t *ind, /* I - Index to search */ last; /* Last entity in search */ +#ifdef DEBUG + printf("mxmlIndexFind(ind=%p, element=\"%s\", value=\"%s\")\n", + ind, element ? element : "(null)", value ? value : "(null)"); +#endif /* DEBUG */ + /* * Range check input... */ if (!ind || (!ind->attr && value)) + { +#ifdef DEBUG + puts(" returning NULL..."); + printf(" ind->attr=\"%s\"\n", ind->attr ? ind->attr : "(null)"); +#endif /* DEBUG */ + return (NULL); + } /* * If both element and value are NULL, just enumerate the nodes in the @@ -142,7 +154,14 @@ mxmlIndexFind(mxml_index_t *ind, /* I - Index to search */ */ if (!ind->num_nodes) + { +#ifdef DEBUG + puts(" returning NULL..."); + puts(" no nodes!"); +#endif /* DEBUG */ + return (NULL); + } /* * If cur_node == 0, then find the first matching node... @@ -157,20 +176,36 @@ mxmlIndexFind(mxml_index_t *ind, /* I - Index to search */ first = 0; last = ind->num_nodes - 1; - while (last > first) +#ifdef DEBUG + printf(" find first time, num_nodes=%d...\n", ind->num_nodes); +#endif /* DEBUG */ + + while ((last - first) > 1) { current = (first + last) / 2; +#ifdef DEBUG + printf(" first=%d, last=%d, current=%d\n", first, last, current); +#endif /* DEBUG */ + if ((diff = index_find(ind, element, value, ind->nodes[current])) == 0) { /* * Found a match, move back to find the first... */ +#ifdef DEBUG + puts(" match!"); +#endif /* DEBUG */ + while (current > 0 && !index_find(ind, element, value, ind->nodes[current - 1])) current --; +#ifdef DEBUG + printf(" returning first match=%d\n", current); +#endif /* DEBUG */ + /* * Return the first match and save the index to the next... */ @@ -183,31 +218,43 @@ mxmlIndexFind(mxml_index_t *ind, /* I - Index to search */ last = current; else first = current; + +#ifdef DEBUG + printf(" diff=%d\n", diff); +#endif /* DEBUG */ } /* * If we get this far, then we found exactly 0 or 1 matches... */ - current = (first + last) / 2; - ind->cur_node = ind->num_nodes; + for (current = first; current <= last; current ++) + if (!index_find(ind, element, value, ind->nodes[current])) + { + /* + * Found exactly one (or possibly two) match... + */ - if (!index_find(ind, element, value, ind->nodes[current])) - { - /* - * Found exactly one match... - */ +#ifdef DEBUG + printf(" returning only match %d...\n", current); +#endif /* DEBUG */ - return (ind->nodes[current]); - } - else - { - /* - * No matches... - */ + ind->cur_node = current + 1; - return (NULL); - } + return (ind->nodes[current]); + } + + /* + * No matches... + */ + + ind->cur_node = ind->num_nodes; + +#ifdef DEBUG + puts(" returning NULL..."); +#endif /* DEBUG */ + + return (NULL); } else if (ind->cur_node < ind->num_nodes && !index_find(ind, element, value, ind->nodes[ind->cur_node])) @@ -216,6 +263,10 @@ mxmlIndexFind(mxml_index_t *ind, /* I - Index to search */ * Return the next matching node... */ +#ifdef DEBUG + printf(" returning next match %d...\n", ind->cur_node); +#endif /* DEBUG */ + return (ind->nodes[ind->cur_node ++]); } @@ -225,6 +276,10 @@ mxmlIndexFind(mxml_index_t *ind, /* I - Index to search */ ind->cur_node = ind->num_nodes; +#ifdef DEBUG + puts(" returning NULL..."); +#endif /* DEBUG */ + return (NULL); } @@ -253,6 +308,11 @@ mxmlIndexNew(mxml_node_t *node, /* I - XML node tree */ * Range check input... */ +#ifdef DEBUG + printf("mxmlIndexNew(node=%p, element=\"%s\", attr=\"%s\")\n", + node, element ? element : "(null)", attr ? attr : "(null)"); +#endif /* DEBUG */ + if (!node) return (NULL); @@ -260,12 +320,22 @@ mxmlIndexNew(mxml_node_t *node, /* I - XML node tree */ * Create a new index... */ - if ((ind = calloc(1, sizeof(mxml_index_t))) != NULL) + if ((ind = calloc(1, sizeof(mxml_index_t))) == NULL) + { + mxml_error("Unable to allocate %d bytes for index - %s", + sizeof(mxml_index_t), strerror(errno)); return (NULL); + } - for (current = mxmlFindElement(node, node, element, attr, NULL, MXML_DESCEND); - current; - current = mxmlFindElement(current, node, element, attr, NULL, MXML_DESCEND)) + if (attr) + ind->attr = strdup(attr); + + if (!element && !attr) + current = node; + else + current = mxmlFindElement(node, node, element, attr, NULL, MXML_DESCEND); + + while (current) { if (ind->num_nodes >= ind->alloc_nodes) { @@ -293,15 +363,79 @@ mxmlIndexNew(mxml_node_t *node, /* I - XML node tree */ } ind->nodes[ind->num_nodes ++] = current; + + current = mxmlFindElement(current, node, element, attr, NULL, MXML_DESCEND); } /* * Sort nodes based upon the search criteria... */ +#ifdef DEBUG + { + int i; /* Looping var */ + + + printf("%d node(s) in index.\n\n", ind->num_nodes); + + if (attr) + { + printf("Node Address Element %s\n", attr); + puts("-------- -------- -------------- ------------------------------"); + + for (i = 0; i < ind->num_nodes; i ++) + printf("%8d %-8p %-14.14s %s\n", i, ind->nodes[i], + ind->nodes[i]->value.element.name, + mxmlElementGetAttr(ind->nodes[i], attr)); + } + else + { + puts("Node Address Element"); + puts("-------- -------- --------------"); + + for (i = 0; i < ind->num_nodes; i ++) + printf("%8d %-8p %s\n", i, ind->nodes[i], + ind->nodes[i]->value.element.name); + } + + putchar('\n'); + } +#endif /* DEBUG */ + if (ind->num_nodes > 1) index_sort(ind, 0, ind->num_nodes - 1); +#ifdef DEBUG + { + int i; /* Looping var */ + + + puts("After sorting:\n"); + + if (attr) + { + printf("Node Address Element %s\n", attr); + puts("-------- -------- -------------- ------------------------------"); + + for (i = 0; i < ind->num_nodes; i ++) + printf("%8d %-8p %-14.14s %s\n", i, ind->nodes[i], + ind->nodes[i]->value.element.name, + mxmlElementGetAttr(ind->nodes[i], attr)); + } + else + { + puts("Node Address Element"); + puts("-------- -------- --------------"); + + for (i = 0; i < ind->num_nodes; i ++) + printf("%8d %-8p %s\n", i, ind->nodes[i], + ind->nodes[i]->value.element.name); + } + + putchar('\n'); + } +#endif /* DEBUG */ + /* * Return the new index... */ @@ -321,6 +455,10 @@ mxmlIndexNew(mxml_node_t *node, /* I - XML node tree */ mxml_node_t * /* O - First node or NULL if there is none */ mxmlIndexReset(mxml_index_t *ind) /* I - Index to reset */ { +#ifdef DEBUG + printf("mxmlIndexReset(ind=%p)\n", ind); +#endif /* DEBUG */ + /* * Range check input... */ @@ -483,8 +621,11 @@ index_sort(mxml_index_t *ind, /* I - Index to sort */ * pivot node... */ - ind->nodes[left] = ind->nodes[tempr]; - ind->nodes[tempr] = pivot; + if (index_compare(ind, pivot, ind->nodes[tempr]) > 0) + { + ind->nodes[left] = ind->nodes[tempr]; + ind->nodes[tempr] = pivot; + } /* * Recursively sort the left and right partitions as needed... @@ -499,5 +640,5 @@ index_sort(mxml_index_t *ind, /* I - Index to sort */ /* - * End of "$Id: mxml-index.c,v 1.3 2004/05/16 13:45:56 mike Exp $". + * End of "$Id: mxml-index.c,v 1.4 2004/05/16 18:25:20 mike Exp $". */ diff --git a/testmxml.c b/testmxml.c index 92919db..76c1b53 100644 --- a/testmxml.c +++ b/testmxml.c @@ -1,5 +1,5 @@ /* - * "$Id: testmxml.c,v 1.16 2004/05/02 16:04:40 mike Exp $" + * "$Id: testmxml.c,v 1.17 2004/05/16 18:25:20 mike Exp $" * * Test program for Mini-XML, a small XML-like file parsing library. * @@ -51,6 +51,7 @@ main(int argc, /* I - Number of command-line args */ FILE *fp; /* File to read */ mxml_node_t *tree, /* XML tree */ *node; /* Node which should be in test.xml */ + mxml_index_t *ind; /* XML index */ char buffer[16384]; /* Save string */ static const char *types[] = /* Strings for node types */ { @@ -106,11 +107,11 @@ main(int argc, /* I - Number of command-line args */ mxmlNewReal(tree, 123.4f); mxmlNewText(tree, 1, "text"); - mxmlLoadString(tree, "string string string", + mxmlLoadString(tree, "string string string", MXML_NO_CALLBACK); - mxmlLoadString(tree, "1 2 3", + mxmlLoadString(tree, "1 2 3", MXML_INTEGER_CALLBACK); - mxmlLoadString(tree, "1.0 2.0 3.0", + mxmlLoadString(tree, "1.0 2.0 3.0", MXML_REAL_CALLBACK); mxmlLoadString(tree, "opaque opaque opaque", MXML_OPAQUE_CALLBACK); @@ -240,7 +241,127 @@ main(int argc, /* I - Number of command-line args */ mxmlDelete(tree); return (1); } - } + } + + /* + * Test indices... + */ + + ind = mxmlIndexNew(tree, NULL, NULL); + if (!ind) + { + fputs("ERROR: Unable to create index of all nodes!\n", stderr); + mxmlDelete(tree); + return (1); + } + + if (ind->num_nodes != 5) + { + fprintf(stderr, "ERROR: Index of all nodes contains %d " + "nodes; expected 5!\n", ind->num_nodes); + mxmlIndexDelete(ind); + mxmlDelete(tree); + return (1); + } + + mxmlIndexReset(ind); + if (!mxmlIndexFind(ind, "group", NULL)) + { + fputs("ERROR: mxmlIndexFind for \"group\" failed!\n", stderr); + mxmlIndexDelete(ind); + mxmlDelete(tree); + return (1); + } + + mxmlIndexDelete(ind); + + ind = mxmlIndexNew(tree, "group", NULL); + if (!ind) + { + fputs("ERROR: Unable to create index of groups!\n", stderr); + mxmlDelete(tree); + return (1); + } + + if (ind->num_nodes != 4) + { + fprintf(stderr, "ERROR: Index of groups contains %d " + "nodes; expected 4!\n", ind->num_nodes); + mxmlIndexDelete(ind); + mxmlDelete(tree); + return (1); + } + + mxmlIndexReset(ind); + if (!mxmlIndexEnum(ind)) + { + fputs("ERROR: mxmlIndexEnum failed!\n", stderr); + mxmlIndexDelete(ind); + mxmlDelete(tree); + return (1); + } + + mxmlIndexDelete(ind); + + ind = mxmlIndexNew(tree, NULL, "type"); + if (!ind) + { + fputs("ERROR: Unable to create index of type attributes!\n", stderr); + mxmlDelete(tree); + return (1); + } + + if (ind->num_nodes != 3) + { + fprintf(stderr, "ERROR: Index of type attributes contains %d " + "nodes; expected 3!\n", ind->num_nodes); + mxmlIndexDelete(ind); + mxmlDelete(tree); + return (1); + } + + mxmlIndexReset(ind); + if (!mxmlIndexFind(ind, NULL, "string")) + { + fputs("ERROR: mxmlIndexFind for \"string\" failed!\n", stderr); + mxmlIndexDelete(ind); + mxmlDelete(tree); + return (1); + } + + mxmlIndexDelete(ind); + + ind = mxmlIndexNew(tree, "group", "type"); + if (!ind) + { + fputs("ERROR: Unable to create index of elements and attributes!\n", stderr); + mxmlDelete(tree); + return (1); + } + + if (ind->num_nodes != 3) + { + fprintf(stderr, "ERROR: Index of elements and attributes contains %d " + "nodes; expected 3!\n", ind->num_nodes); + mxmlIndexDelete(ind); + mxmlDelete(tree); + return (1); + } + + mxmlIndexReset(ind); + if (!mxmlIndexFind(ind, "group", "string")) + { + fputs("ERROR: mxmlIndexFind for \"string\" failed!\n", stderr); + mxmlIndexDelete(ind); + mxmlDelete(tree); + return (1); + } + + mxmlIndexDelete(ind); + + /* + * Check the mxmlDelete() works properly... + */ for (i = 0; i < 8; i ++) { @@ -431,5 +552,5 @@ whitespace_cb(mxml_node_t *node, /* I - Element node */ /* - * End of "$Id: testmxml.c,v 1.16 2004/05/02 16:04:40 mike Exp $". + * End of "$Id: testmxml.c,v 1.17 2004/05/16 18:25:20 mike Exp $". */