Use compressed _master-pixmap.txt.gz

diff --git a/pctb/Makefile b/pctb/Makefile
index 8223583a4cbc8fe8aff377be57eedbf51c18bfea..4da46770a04a54ef481e75bdf62418d30e6e1727 100644 (file)
--- a/pctb/Makefile
+++ b/pctb/Makefile
@@ -50,6 +50,6 @@ realclean: clean
        rm -f $(TARGETS)
        rm -f raw.tsv
        rm -f _pages.ppm _upload-*.html _commodmap.tsv
        rm -f $(TARGETS)
        rm -f raw.tsv
        rm -f _pages.ppm _upload-*.html _commodmap.tsv

-       rm -f _master-*.txt _local-*.txt
+       rm -f _master-*.txt _master-*.txt.gz _local-*.txt

        rm -f ./#pages#.ppm ./#upload-*#.html ./#commodmap#.tsv
        rm -f ./#master-*#.txt ./#local-*#.txt raw.tsv
        rm -f ./#pages#.ppm ./#upload-*#.html ./#commodmap#.tsv
        rm -f ./#master-*#.txt ./#local-*#.txt raw.tsv

diff --git a/pctb/README.files b/pctb/README.files
index b2384a1125b572defc47823dfa6eab3a3b550748..c8d49288e50f8f9c0046ee967f5d205c47dd45ed 100644 (file)
--- a/pctb/README.files
+++ b/pctb/README.files
@@ -43,8 +43,8 @@ The program reads and writes the following files:
    character set dictionary is missing the lowercase `y ' glyph.
    See README.charset.
 
    character set dictionary is missing the lowercase `y ' glyph.
    See README.charset.
 

- * _master-char*.txt  _local-char*.txt
-   _master-pixmap.txt _local-pixmap.txt
+ * _master-char*.txt      _local-char*.txt
+   _master-pixmap.txt.gz  _local-pixmap.txt

 
    Character set and image dictionaries.  For the semantics of the
    char* files README.charset.  There is not currently any accurate
 
    Character set and image dictionaries.  For the semantics of the
    char* files README.charset.  There is not currently any accurate


@@ -52,7 +52,7 @@ The program reads and writes the following files:
 
    _master-*.txt contain the centrally defined and approved data.
    They are downloaded automatically from the SC PCTB server and
 
    _master-*.txt contain the centrally defined and approved data.
    They are downloaded automatically from the SC PCTB server and

-   updated each run.  You can safely delete this file, if everything
+   updated each run.  You can safely delete these files, if everything

    is online, if you want to fetch a fresh copy.
 
    _local-*.txt are a local copy of your submissions, so that they
    is online, if you want to fetch a fresh copy.
 
    _local-*.txt are a local copy of your submissions, so that they

diff --git a/pctb/common.c b/pctb/common.c
index bc1cb8df749cb63ec2ef392300551c37262c6e16..2c2a085a30e9fdaa7ded64ffa7b10c82f31208ac 100644 (file)
--- a/pctb/common.c
+++ b/pctb/common.c
@@ -42,17 +42,68 @@ void *mrealloc(void *p, size_t sz) {
 
 
 FILE *dbfile;
 
 
 FILE *dbfile;

-static const char *path;
+static const char *basepath; /* as passed in by caller */
+static pid_t dbzcat;
+
+int dbfile_gzopen(const char *basepath_spec) {
+  assert(!dbfile);
+
+  basepath= basepath_spec;
+  //uncomppath= masprintf("%s (uncompressed)", basepath);
+
+  char *zpath= masprintf("%s.gz", basepath);
+  int zfd= open(zpath, O_RDONLY);
+  free(zpath);
+
+  if (zfd<0) { sysassert(errno==ENOENT); return 0; }
+
+  int pipefds[2];
+  sysassert(! pipe(pipefds) );
+
+  sysassert( (dbzcat=fork()) != -1 );
+  if (!dbzcat) {
+    sysassert( dup2(zfd,0)==0 );
+    sysassert( dup2(pipefds[1],1)==1 );
+    sysassert(! close(zfd) );
+    sysassert(! close(pipefds[0]) );
+    sysassert(! close(pipefds[1]) );
+    execlp("zcat","zcat",(char*)0);
+    sysassert(!"execlp zcat");
+  }
+  sysassert(! close(zfd) );
+  sysassert(! close(pipefds[1]) );
+  sysassert( dbfile= fdopen(pipefds[0], "r") );
+
+  return 1;
+}  

 
 int dbfile_open(const char *tpath) {
   assert(!dbfile);
 
 int dbfile_open(const char *tpath) {
   assert(!dbfile);

-  path= tpath;
-  dbfile= fopen(path,"r");
-  if (dbfile) return 1;
-  sysassert(errno==ENOENT);
-  return 0;
+
+  basepath= tpath;
+
+  dbzcat= -1;
+  dbfile= fopen(tpath,"r");
+  if (!dbfile) { sysassert(errno==ENOENT); return 0; }
+  return 1;

 }  
 
 }  
 

+void dbfile_close(void) {
+  if (!dbfile) return;
+
+  sysassert(!ferror(dbfile));
+  sysassert(!fclose(dbfile));
+
+  if (dbzcat != -1) {
+    char *zcatstr= masprintf("zcat %s.gz", basepath);
+    waitpid_check_exitstatus(dbzcat,zcatstr,1);
+    free(zcatstr);
+    dbzcat= -1;
+  }
+
+  dbfile= 0;
+}
+

 #define dbassertgl(x) ((x) ? (void)0 : dbfile_assertfail(file,line,#x))
 
 void dbfile_getsline(char *lbuf, size_t lbufsz, const char *file, int line) {
 #define dbassertgl(x) ((x) ? (void)0 : dbfile_assertfail(file,line,#x))
 
 void dbfile_getsline(char *lbuf, size_t lbufsz, const char *file, int line) {


@@ -66,13 +117,6 @@ void dbfile_getsline(char *lbuf, size_t lbufsz, const char *file, int line) {
   lbuf[l]= 0;
 }
 
   lbuf[l]= 0;
 }
 

-void dbfile_close(void) {
-  if (!dbfile) return;
-  sysassert(!ferror(dbfile));
-  sysassert(!fclose(dbfile));
-  dbfile= 0;
-}
-

 int dbfile_vscanf(const char *fmt, va_list al) {
   int r= vfscanf(dbfile,fmt,al);
   sysassert(!ferror(dbfile));
 int dbfile_vscanf(const char *fmt, va_list al) {
   int r= vfscanf(dbfile,fmt,al);
   sysassert(!ferror(dbfile));


@@ -88,11 +132,16 @@ int dbfile_scanf(const char *fmt, ...) {
 }
 
 void dbfile_assertfail(const char *file, int line, const char *m) {
 }
 
 void dbfile_assertfail(const char *file, int line, const char *m) {

-  if (dbfile)
+  if (dbzcat)
+    fatal("Error in dictionary file %s.gz:\n"
+         " Requirement not met at %s:%d:\n"
+         " %s",
+         basepath, file,line, m);
+  else if (dbfile)

     fatal("Error in dictionary file %s at byte %ld:\n"
          " Requirement not met at %s:%d:\n"
          " %s",
     fatal("Error in dictionary file %s at byte %ld:\n"
          " Requirement not met at %s:%d:\n"
          " %s",

-         path,(long)ftell(dbfile), file,line, m);
+         basepath,(long)ftell(dbfile), file,line, m);

   else
     fatal("Semantic error in dictionaries:\n"
          " Requirement not met at %s:%d:\n"
   else
     fatal("Semantic error in dictionaries:\n"
          " Requirement not met at %s:%d:\n"

diff --git a/pctb/common.h b/pctb/common.h
index 210f23ac4834baa516d8b707025ca956e93e3083..92ba2daa13d65f27cdf97fe332c48c8ee3ef5108 100644 (file)
--- a/pctb/common.h
+++ b/pctb/common.h
@@ -41,9 +41,12 @@
 #include <stdlib.h>
 #include <unistd.h>
 #include <dirent.h>
 #include <stdlib.h>
 #include <unistd.h>
 #include <dirent.h>

-#include <pcre.h>

 #include <inttypes.h>
 
 #include <inttypes.h>
 

+#include <pcre.h>
+
+#include <fcntl.h>
+#include <unistd.h>

 #include <sys/wait.h>
 #include <sys/types.h>
 
 #include <sys/wait.h>
 #include <sys/types.h>
 


@@ -133,7 +136,7 @@ void fatal(const char *fmt, ...)       FMT(1,2) NORET;
 void sysassert_fail(const char *file, int line, const char *what)
    __attribute__((noreturn));
 
 void sysassert_fail(const char *file, int line, const char *what)
    __attribute__((noreturn));
 

-void waitpid_check_exitstatus(pid_t pid, const char *what);
+void waitpid_check_exitstatus(pid_t pid, const char *what, int sigpipeok);

 
 
 void *mmalloc(size_t sz);
 
 
 void *mmalloc(size_t sz);


@@ -145,7 +148,8 @@ void dbfile_assertfail(const char *file, int line, const char *m) NORET;
 
 FILE *dbfile;
 void dbfile_getsline(char *lbuf, size_t lbufsz, const char *file, int line);
 
 FILE *dbfile;
 void dbfile_getsline(char *lbuf, size_t lbufsz, const char *file, int line);

-int dbfile_open(const char *tpath); /* 0: ENOENT; 1: worked */
+int dbfile_open(const char *tpath);   /* 0: ENOENT; 1: worked */
+int dbfile_gzopen(const char *tpath); /* 0: ENOENT; 1: worked */

 void dbfile_close(void); /* idempotent */
 
 int dbfile_scanf(const char *fmt, ...) SCANFMT(1,2);
 void dbfile_close(void); /* idempotent */
 
 int dbfile_scanf(const char *fmt, ...) SCANFMT(1,2);

diff --git a/pctb/convert.c b/pctb/convert.c
index 8386c1609a0f68f85a0936a0df953b26c120f2c2..6bc9b2c8897d28ce7f88793a0d424ef131d3e76c 100644 (file)
--- a/pctb/convert.c
+++ b/pctb/convert.c
@@ -110,12 +110,13 @@ static void run_analysis(void) {
     EXECLP_HELPER("commod-results-processor", o_outmode_str, (char*)0);
   }
 
     EXECLP_HELPER("commod-results-processor", o_outmode_str, (char*)0);
   }
 

-  waitpid_check_exitstatus(processor, "output processor/uploader");
+  waitpid_check_exitstatus(processor, "output processor/uploader", 0);

   fclose(tf);
   progress_log("all complete.");
 }
 
   fclose(tf);
   progress_log("all complete.");
 }
 

-void fetch_with_rsync(const char *stem) {
+static void rsync_core(const char *stem, const char *suffix,
+                      const char *zopt) {

   pid_t fetcher;
 
   sysassert( (fetcher= fork()) != -1 );
   pid_t fetcher;
 
   sysassert( (fetcher= fork()) != -1 );


@@ -124,19 +125,23 @@ void fetch_with_rsync(const char *stem) {
     if (!rsync) rsync= "rsync";
   
     const char *src= getenv("YPPSC_PCTB_DICT_UPDATE");
     if (!rsync) rsync= "rsync";
   
     const char *src= getenv("YPPSC_PCTB_DICT_UPDATE");

-    char *remote= masprintf("%s/master-%s.txt", src, stem);
-    char *local= masprintf("_master-%s.txt", stem);
+    char *remote= masprintf("%s/master-%s.txt%s", src, stem, suffix);
+    char *local= masprintf("_master-%s.txt%s", stem, suffix);

     if (DEBUGP(rsync))
       fprintf(stderr,"executing rsync to fetch %s to %s\n",remote,local);
     if (DEBUGP(rsync))
       fprintf(stderr,"executing rsync to fetch %s to %s\n",remote,local);

-    execlp(rsync, "rsync",
-          DEBUGP(rsync) ? "-zvLt" : "-zLt",
-          "--",remote,local,(char*)0);
+    char *opts= masprintf("-Lt%s%s",
+                         zopt,
+                         DEBUGP(rsync) ? "v" : "");
+    execlp(rsync, "rsync",opts,"--",remote,local,(char*)0);

     sysassert(!"exec rsync failed");
   }
 
     sysassert(!"exec rsync failed");
   }
 

-  waitpid_check_exitstatus(fetcher, "rsync");
+  waitpid_check_exitstatus(fetcher, "rsync", 0);

 }
 
 }
 

+void fetch_with_rsync_gz(const char *stem) { rsync_core(stem,".gz",""); }
+void fetch_with_rsync(const char *stem) { rsync_core(stem,"","z"); }
+

 static void set_server(const char *envname, const char *defprotocol,
                       const char *defvalue, const char *defvalue_test,
                       const char *userspecified,
 static void set_server(const char *envname, const char *defprotocol,
                       const char *defvalue, const char *defvalue_test,
                       const char *userspecified,


@@ -411,7 +416,7 @@ void sysassert_fail(const char *file, int line, const char *what) {
   _exit(16);
 }
 
   _exit(16);
 }
 

-void waitpid_check_exitstatus(pid_t pid, const char *what) { 
+void waitpid_check_exitstatus(pid_t pid, const char *what, int sigpipeok) { 

   pid_t got;
   int st;
   for (;;) {
   pid_t got;
   int st;
   for (;;) {


@@ -426,8 +431,9 @@ void waitpid_check_exitstatus(pid_t pid, const char *what) {
       fatal("%s failed with nonzero exit status %d",
            what, WEXITSTATUS(st));
   } else if (WIFSIGNALED(st)) {
       fatal("%s failed with nonzero exit status %d",
            what, WEXITSTATUS(st));
   } else if (WIFSIGNALED(st)) {

-    fatal("%s died due to signal %s%s", what,
-         strsignal(WTERMSIG(st)), WCOREDUMP(st)?" (core dumped)":"");
+    if (!sigpipeok || WTERMSIG(st) != SIGPIPE)
+      fatal("%s died due to signal %s%s", what,
+           strsignal(WTERMSIG(st)), WCOREDUMP(st)?" (core dumped)":"");

   } else {
     fatal("%s gave strange wait status %d", what, st);
   }
   } else {
     fatal("%s gave strange wait status %d", what, st);
   }

diff --git a/pctb/convert.h b/pctb/convert.h
index ffa80ebf08062bf724b73567fe8fa39f0c9c01c8..0dc2255ddaea9c9a1cc214e04c3487c5255a145a 100644 (file)
--- a/pctb/convert.h
+++ b/pctb/convert.h
@@ -85,7 +85,8 @@ void analyse(FILE *tsv_output);
 /*----- from convert.c -----*/
 
 extern FILE *screenshot_file;
 /*----- from convert.c -----*/
 
 extern FILE *screenshot_file;

-extern void fetch_with_rsync(const char *stem);
+void fetch_with_rsync(const char *stem);
+void fetch_with_rsync_gz(const char *stem);

 
 void vwarning(const char *fmt, va_list) FMT(1,0);
 void warning(const char *fmt, ...)      FMT(1,2);
 
 void vwarning(const char *fmt, va_list) FMT(1,0);
 void warning(const char *fmt, ...)      FMT(1,2);

diff --git a/pctb/resolve.c b/pctb/resolve.c
index 19615f3e4ea251ab301cb496d28b09b7181203c3..e89d2054e91a0e076c685a03b6f1a2273d8415e2 100644 (file)
--- a/pctb/resolve.c
+++ b/pctb/resolve.c
@@ -83,7 +83,7 @@ void resolve_finish(void) {
   }
 
   if (r==0) {
   }
 
   if (r==0) {

-    waitpid_check_exitstatus(resolver_pid, "dictionary-manager");
+    waitpid_check_exitstatus(resolver_pid, "dictionary-manager", 0);

     fclose(resolver);
     close(resolver_done);
     resolver= 0;
     fclose(resolver);
     close(resolver_done);
     resolver= 0;

diff --git a/pctb/rgbimage.c b/pctb/rgbimage.c
index dfcd1808f25421ceb9223343c1d24618bbae2eef..81eddf9898e83427a83f7376bef93868e8a4dc56 100644 (file)
--- a/pctb/rgbimage.c
+++ b/pctb/rgbimage.c
@@ -50,9 +50,9 @@
 
 static int identify1(const RgbImage *base, Rect portion,
                    char result[MAXIMGIDENT], const char *what,
 
 static int identify1(const RgbImage *base, Rect portion,
                    char result[MAXIMGIDENT], const char *what,

-                   const char *which) {
+                   const char *which, int (*opener)(const char *fn)) {

   char *dbfile_name= masprintf("_%s-pixmap.txt",which);
   char *dbfile_name= masprintf("_%s-pixmap.txt",which);

-  if (!dbfile_open(dbfile_name))
+  if (!opener(dbfile_name))

     goto not_found;
 
 #define FGETSLINE (dbfile_getsline(result,MAXIMGIDENT,__FILE__,__LINE__))
     goto not_found;
 
 #define FGETSLINE (dbfile_getsline(result,MAXIMGIDENT,__FILE__,__LINE__))


@@ -111,8 +111,8 @@ static int identify1(const RgbImage *base, Rect portion,
 
 static int identify(const RgbImage *base, Rect portion,
                    char result[MAXIMGIDENT], const char *what) {
 
 static int identify(const RgbImage *base, Rect portion,
                    char result[MAXIMGIDENT], const char *what) {

-  return identify1(base,portion,result,what, "master") ||
-         identify1(base,portion,result,what, "local");
+  return identify1(base,portion,result,what, "master", dbfile_gzopen) ||
+         identify1(base,portion,result,what, "local",  dbfile_open);

 }
 
 void fwrite_ppmraw(FILE *f, const RgbImage *ri) {
 }
 
 void fwrite_ppmraw(FILE *f, const RgbImage *ri) {


@@ -147,7 +147,7 @@ void identify_rgbimage(const RgbImage *base, Rect portion,
 
   if (!synced) {
     if (o_flags & ff_dict_fetch)
 
   if (!synced) {
     if (o_flags & ff_dict_fetch)

-      fetch_with_rsync("pixmap");
+      fetch_with_rsync_gz("pixmap");

     synced++;
   }
   
     synced++;
   }