gen_twitter_archive.c

#define STB_DEFINE
#include "stb.h"

char *header;

typedef struct
{
   char *id;
   char *html;
   char *commentary;
   char *category;
   int   rtcount;
   int   fvcount;
   float preference;
} Tweet;

enum
{
   O_reverse_chronological,
   O_chronological,
   O_retweets,
   O_likes,
   O_preference,
   O_commentary,

   O__count
};

Tweet *tweets;

void sort(int mode)
{
   int i,n;
   n = stb_arr_len(tweets);
   switch (mode) {
      case O_chronological:
      case O_commentary:
      case O_reverse_chronological: qsort(tweets, n, sizeof(tweets[0]), stb_qsort_strcmp(offsetof(Tweet,id))); break;
      case O_retweets:              qsort(tweets, n, sizeof(tweets[0]), stb_intcmp(offsetof(Tweet,rtcount))); break;
      case O_likes:                 qsort(tweets, n, sizeof(tweets[0]), stb_intcmp(offsetof(Tweet,fvcount))); break;
      case O_preference:            qsort(tweets, n, sizeof(tweets[0]), stb_floatcmp(offsetof(Tweet,preference))); break;
   }
   if (mode == O_chronological || mode == O_commentary)
      return;
   for (i=0; i < n/2; ++i) {
      Tweet t = tweets[i];
      tweets[i] = tweets[n-1-i];
      tweets[n-1-i] = t;
   }
}

char *tab_name[] =
{
   "Likes",
   "Retweets",
   "Chronological",
   "movie_goofs' top 20",
   "Commentary",
};

char *mode_name[] =
{
   "Date\xe2\x86\x91",
   "Date\xe2\x86\x93",
   "Retweets",
   "Likes",
   "movie_goofs' top 20",
   "Commentary",
};


char *mode_url[] =
{
   "index.html",
   "all_by_oldest.html",
   "all_by_retweets.html",
   "all_by_likes.html",
   "all_by_preferred.html",
   "commentary.html",
};

static int tab_for_mode_table[] = {
   0,0, 2,1,3, 4
};

int tab_for_mode(int m)
{
   return tab_for_mode_table[m];
}

int mode_for_tab(int t)
{
   int i;
   for (i=0; i < 6; ++i)
      if (tab_for_mode_table[i] == t)
         return i;
   return 0;
}

void write_file(int mode)
{
   int i,tab = tab_for_mode(mode);

   FILE *f;

   // someday!
   if (mode == O_commentary)
      return;
   
   f = fopen(mode_url[mode], "w");
   sort(mode);
   fputs(header, f);

   // write the tab header
   //fprintf(f, "<div width='100%%'>\n");
   fprintf(f, "<div align=center style='max-width:500px;margin:auto'>\n");
   fprintf(f, "<div align=center>Sort by</div>\n");
   fprintf(f, "<table cellspacing=0 cellpadding=2 border=1 width='100%%'><tr>\n");
   for (i=0; i < 4; ++i) {
      int m = mode_for_tab(i);
      if (i == 0 && tab == 0) // reverse mode for hotlink
         m = (O_chronological + O_reverse_chronological) - mode;
      // don't hotlink if it's the current tab
      if (i == tab && (i != 0))
         fprintf(f, "<td align=center>%s</td>\n", mode_name[m]);
      else 
         fprintf(f, "<td align=center><a style=\"text-decoration:none\" href=\"%s\">%s</a></td>\n", mode_url[m], mode_name[m]);
   }
   fprintf(f, "</tr></table>\n");
   //fprintf(f, "</div>\n");

   if (mode == O_preference)
      fprintf(f, "<p>My opinion is always changing, but at the moment, here's my twenty favorites.</p>\n");

   for (i=0; i < stb_arr_len(tweets); ++i) {
      if (mode == O_commentary) {
         if (tweets[i].commentary == NULL) continue;
         if (tweets[i].commentary[0] == 0) continue;
      }
      fprintf(f, "<p>%s\n", tweets[i].html);
      if (mode == O_commentary) {
         fprintf(f, "%s\n<hr><p>", tweets[i].commentary);
      }
      if (mode == O_preference)
         if (i >= 20)
            break;
   }
   fprintf(f, "</div>\n");

   fprintf(f, "</body>\n</html>\n");
   fclose(f);
}

Tweet *find_tweet(char *id)
{
   static Tweet dummy;
   int i;
   for (i=0; i < stb_arr_len(tweets); ++i)
      if (0==strcmp(id, tweets[i].id))
         return &tweets[i];
   return 0;
}

int main(int argc, char **argv)
{
   int i,n;

   // read the source files
   char **list = stb_stringfile("goof_tweets.csv", &n);
   for (i=0; i < n; ++i) {
      Tweet t = { 0 };
      char *s = list[i];
      char *a = s;
      char *b = strchr(a, ',') + 1;
      char *c = strchr(b, ',') + 1;
      b[-1] = 0;
      c[-1] = 0;
      // a = id
      // b = user_id
      // c = body text
      t.id = a;
      if (0==strcmp(t.id, "1036025365334917120"))
         continue;
      t.html = stb_file(stb_sprintf("embed/%s.html", a), 0);
      if (t.html)
         stb_arr_push(tweets, t);
      else
         printf("Missing html file for %s\n", t.id);
   }

   // not actually comma-separated since it was easier to parse in Go with spaces
   list = stb_stringfile("goof_stats.csv", &n);
   for (i=0; i < n; ++i) {
      char id[500];
      int fav,rt;
      char *s = list[i];
      if (3==sscanf(s, "%s %d %d", id, &fav, &rt)) {
         Tweet *t = find_tweet(id);
         if (t) {
            t->fvcount = fav;
            t->rtcount = rt;
         }
      }
   }

   list = stb_stringfile("goofs.csv", &n);
   for (i=2; i < n; ++i) {
      char *s = list[i];
      if (isdigit(s[0])) {
         Tweet *t;
         int k;
         char **tok = stb_tokens_quoted(s, ",", &k);
         t = find_tweet(tok[0]);
         if (t) {
            t->category = strdup(tok[2]);
            t->preference = (float) atof(tok[1]);
            t->commentary = strdup(tok[3]);
         }
         //printf("%s - %s - %s - %s\n", tok[0], tok[1], tok[2], tok[4]);
         free(tok);
      }
   }

   
   header = stb_file("header.html", 0);
   for (i=0; i < 6; ++i)
      write_file(i);
   return 0;
}