Move to Makefile and .gitignore
[enwiki-links-graph.git] / src / links-outgoing-to-incoming.c
diff --git a/src/links-outgoing-to-incoming.c b/src/links-outgoing-to-incoming.c
new file mode 100644 (file)
index 0000000..2fd853a
--- /dev/null
@@ -0,0 +1,92 @@
+#include <assert.h>
+#include <ctype.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/types.h>
+
+
+typedef __uint32_t art_id;
+
+
+int main()
+{
+       FILE *in_file;
+       FILE *out_file;
+
+       art_id titles;
+
+       art_id **linko;
+       art_id *linkos;
+
+       art_id **linki;
+       art_id *linkis;
+
+       art_id link_titles_done = 0;
+
+       art_id i;
+
+
+       /*
+        * Read all outgoing links into memory
+        */
+
+       in_file = fopen("links-outgoing.bin", "rb");
+
+       fread(&titles, sizeof(titles), 1, in_file);
+
+       linko = malloc(titles * sizeof(art_id*));
+       linkos = malloc(titles * sizeof(art_id));
+
+       for (i = 0; i < titles; i++) {
+               art_id j;
+
+               fread(&linkos[i], sizeof(linkos[i]), 1, in_file);
+
+               linko[i] = malloc(linkos[i] * sizeof(linko[i][0]));
+
+               for (j = 0; j < linkos[i]; j++) {
+                       fread(&linko[i][j], sizeof(linko[i][j]), 1, in_file);
+               }
+       }
+       fclose(in_file);
+
+       printf("Outgoing links read.\n");
+
+
+
+       linki = malloc(titles * sizeof(art_id*));
+       linkis = malloc(titles * sizeof(art_id));
+
+       for (i = 0; i < titles; i++) {
+               art_id j;
+
+               for (j = 0; j < linkos[i]; j++) {
+                       art_id x = linko[i][j];
+
+                       linkis[x]++;
+                       linki[x] = realloc(linki[x], linkis[x] * sizeof(linki[x][0]));
+
+                       linki[x][linkis[x] - 1] = i;
+               }
+       }
+
+       printf("Links turned upside down.\n");
+
+
+
+       out_file = fopen("links-incoming.bin", "wb");
+       fwrite(&titles, sizeof(titles), 1, out_file);
+       for (i = 0; i < titles; i++) {
+               art_id j;
+
+               fwrite(&linkis[i], sizeof(linkis[i]), 1, out_file);
+
+               for (j = 0; j < linkis[i]; j++) {
+                       fwrite(&linki[i][j], sizeof(linki[i][j]), 1, out_file);
+               }
+       }
+       fclose(out_file);
+
+       printf("Incoming links written.\n");
+}