lookup: Clean up code a bit

diff --git a/Makefile b/Makefile
index 960fcd7fbf7e4434a4869aa6c3e893a6f2c5adab..0dd478f8c6d369a2c7ef01929a23a80e641bf2cd 100644 (file)
--- a/Makefile
+++ b/Makefile
@@ -13,7 +13,7 @@ lookuptools: $(BINDIR)/lookup-incoming
 
 $(BINDIR)/lookup-incoming: $(SRCDIR)/lookup-incoming.c
        mkdir -p $(BINDIR)
-       $(GCC) -o $@ $<
+       $(GCC) -Wall -o $@ $<
 
 
 

diff --git a/src/lookup-incoming.c b/src/lookup-incoming.c
index 2d3fa9cc9bcaf2f9b12a6f2a1cea1539aa092e51..77ffda4f81266ed4d189f70baaea9a513b498bf3 100644 (file)
--- a/src/lookup-incoming.c
+++ b/src/lookup-incoming.c
@@ -6,9 +6,11 @@
 #include <sys/types.h>
 
 
+/*
+ * Bitfield
+ */
 #define ROUNDUP(x,y) ((((x)+(y)-1)/(y))*(y))
 
-
 typedef __uint32_t bitfield_type;
 #define BITS_PER_BITFIELD 32
 #define BITS_PER_BITFIELD_LOG 5
@@ -30,8 +32,11 @@ int ANY_BITS_SET(bitfield_type *set, size_t elems)
 }
 
 
-typedef __uint32_t art_id;
 
+/*
+ * Misc helpers
+ */
+typedef __uint32_t art_id;
 
 
 static int
@@ -41,24 +46,52 @@ cmpstring_p_pp(const void *p1, const void *p2)
 }
 
 
+void* load_file(char *path)
+{
+       FILE *in_file;
+       void *blob;
+       size_t blob_len;
+
+       in_file = fopen(path, "rb");
+
+       fseek(in_file, 0, SEEK_END);
+       blob_len = ftell(in_file);
+       rewind(in_file);
+       printf("  blob size: %zd bytes.\n", blob_len);
+
+       blob = malloc(blob_len);
+       if (!blob) {
+               printf("Failed to allocate memory.\n");
+               return NULL;
+       }
 
+       if (blob_len != fread(blob, 1, blob_len, in_file)) {
+               printf("Failed to read entire file in one go.\n");
+               fclose(in_file);
+               return NULL;
+       }
+       printf("  blob read :%zd bytes.\n", ftell(in_file));
 
+       fclose(in_file);
+       return blob;
+}
+
+
+
+/*
+ * Main
+ */
 int main(int argc, char **argv)
 {
-       FILE *in_file;
-       FILE *out_file;
+       art_id *link_blob;
+       art_id **linki;
+       art_id *linkis;
 
        char *title_blob;
-       size_t title_blob_len;
        char **title = NULL;
        art_id titles;
        art_id titles_read = 0;
 
-       art_id *link_blob;
-       size_t link_blob_len;
-       art_id **linki;
-       art_id *linkis;
-
        char **cur_title;
        art_id title_id;
 
@@ -79,29 +112,12 @@ int main(int argc, char **argv)
        }
 
 
-       /*
-        * Read all incoming links into memory
-        */
-
-       in_file = fopen("links-incoming.bin", "rb");
-       fseek(in_file, 0, SEEK_END);
-       link_blob_len = ftell(in_file);
-       rewind(in_file);
-       printf("Link blob size: %zd bytes.\n", link_blob_len);
-
-       link_blob = malloc(link_blob_len);
-       if (!link_blob) {
-               printf("Failed to allocate memory.\n");
-               return 1;
-       }
-
-       if (link_blob_len != fread(link_blob, 1, link_blob_len, in_file)) {
-               printf("Failed to read entire file in one go.\n");
+       /* Read all incoming links into memory */
+       link_blob = load_file("links-incoming.bin");
+       if (!link_blob)
                return 1;
-       }
-       printf("Link blob read (%zd bytes).\n", ftell(in_file));
-       fclose(in_file);
 
+       /* Parse blob */
        titles = *link_blob;
        link_blob++;
 
@@ -109,62 +125,39 @@ int main(int argc, char **argv)
        linkis = malloc(titles * sizeof(art_id));
 
        for (i = 0; i < titles; i++) {
-               art_id j;
-
                linkis[i] = *link_blob;
                link_blob++;
 
                linki[i] = link_blob;
                link_blob += linkis[i];
        }
-       printf("Incoming links prepared.\n");
-
-
-
-
-
-       /*
-        * Read all titles into memory
-        */
-
-
+       printf("Incoming links parsed: %d\n", titles);
 
-       in_file = fopen("titles-sorted.txt", "r");
-       fseek(in_file, 0, SEEK_END);
-       title_blob_len = ftell(in_file);
-       rewind(in_file);
-       printf("Title blob size: %zd bytes.\n", title_blob_len);
 
-       title_blob = malloc(title_blob_len + 1);
-       if (!title_blob) {
-               printf("Failed to allocate memory.\n");
+       /* Read all titles into memory */
+       title_blob = load_file("titles-sorted.txt");
+       if (!title_blob)
                return 1;
-       }
-
-       if (title_blob_len != fread(title_blob, 1, title_blob_len, in_file)) {
-               printf("Failed to read entire file in one go.\n");
-               return 1;
-       }
-       printf("Title blob read (%zd bytes).\n", ftell(in_file));
-       fclose(in_file);
-
-       title_blob[title_blob_len] = '\0';
 
+       /* Parse blob */
        title = malloc((titles + 1) * sizeof(title[0]));
 
        title[0] = title_blob;
        titles_read = 1;
-       while (title_blob = strchr(title_blob, '\n')) {
+       while ((title_blob = strchr(title_blob, '\n'))) {
                title_blob[0] = '\0';
                title_blob++;
+
+               if (title_blob[0] == '\0')
+                       /* Reached empty line at end of file */
+                       continue;
+
                title[titles_read] = title_blob;
                titles_read++;
        }
 
        /* Last title will be an empty string, as the file ends with a newline. */
-       printf("Titles prepared: %d of %d.\n", titles_read - 1, titles);
-
-
+       printf("Titles parsed: %d of %d.\n", titles_read, titles);
 
 
        /* Look up article ID */
@@ -190,13 +183,13 @@ int main(int argc, char **argv)
        #endif
 
 
-
-       printf("\n\n\n\nBuilding table of distances...\n\n");
+       /* Main part */
+       printf("\n\n\n\nFlooding graph...\n\n");
 
        titles_bitfield_elems = ROUNDUP(titles / BITS_PER_BITFIELD, BITS_PER_BITFIELD);
        titles_bitfield_bytes = titles_bitfield_elems * sizeof(bitfield_type);
 
-BENCHMARK:
+//BENCHMARK:
 
        titles_seen = calloc(titles_bitfield_bytes, 1);
        titles_prev_round = NULL;
@@ -235,11 +228,11 @@ BENCHMARK:
                }
        }
 
-       /* Result is in titles_prev_round */
+       /* Result is now in titles_prev_round */
 
        cur_dist -= 1;
 
-       printf("\n\n\nThe articles with the distance of %zd are:\n\n", cur_dist);
+       printf("\n\n\nThe articles with a distance of %d are:\n\n", cur_dist);
        for (i = 0; i < titles; i++) {
                if (BIT_ISSET(titles_prev_round, i)) {
                        printf("  %s\n", title[i]);