X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.db-test.git;a=blobdiff_plain;f=pctb%2Fconvert.c;h=e6e0102e6b02b69fb7031d52de2a43a80dc03e18;hp=b9325e88dcfacaf05be5ce374b6400e5fdae3915;hb=373a139d8c902f3c7af960d05c7665048cf4d5c0;hpb=6a3c0962283d32bc6e5f6c47c929baf37ddc642f

diff --git a/pctb/convert.c b/pctb/convert.c
index b9325e8..e6e0102 100644
--- a/pctb/convert.c
+++ b/pctb/convert.c
@@ -37,28 +37,38 @@ const char *get_libdir(void) { return "."; }
 
 
 enum mode {
-  mf_findwindow=      0001,
-  mf_screenshot=      0010,
-  mf_readscreenshot=  0020,
-  mf_analyse=         0100,
+  mf_findwindow=      00001,
+  mf_screenshot=      00010,
+  mf_readscreenshot=  00020,
+  mf_analyse=         00100,
+  mfm_special=        07000,
   
-  mode_findwindow=    0001,
-  mode_screenshot=    0011,
-  mode_analyse=       0120,
+  mode_findwindow=    00001,
+  mode_screenshot=    00011,
+  mode_analyse=       00120,
+  mode_showcharset=   01000,
 
-  mode_all=           0111,
+  mode_all=           00111,
 };
 
+enum outmodekind {
+  omk_unset, omk_upload, omk_str, omk_raw, omk_none
+};
+static enum outmodekind o_outmode_kind;
+static const char *o_outmode_str= 0;
+
 static enum mode o_mode= mode_all;
 static char *o_screenshot_fn;
-static int o_quiet;
-static const char *o_outputmode= "upload";
 static const char *o_serv_pctb, *o_serv_dict_fetch, *o_serv_dict_submit;
 
 const char *o_resolver= "./dictionary-manager";
 FILE *screenshot_file;
+const char *o_ocean, *o_pirate;
+int o_quiet;
 
-enum flags o_flags= ff_dict_fetch|ff_dict_submit|ff_dict_pirate;
+enum flags o_flags=
+   ff_charset_allowedit |
+   ff_dict_fetch|ff_dict_submit|ff_dict_pirate;
 
 static void vbadusage(const char *fmt, va_list) FMT(1,0) NORET;
 static void vbadusage(const char *fmt, va_list al) {
@@ -84,51 +94,59 @@ static void run_analysis(void) {
   analyse(tf);
 
   if (o_flags & ff_upload) {
-    if (o_flags & ff_singlepage)
+    if (npages<=1)
       fatal("Recognition successful, but refusing to upload partial data\n"
 	    " (--single-page specified).  Specify an output mode?");
   }
 
   sysassert( fseek(tf,0,SEEK_SET) == 0);
 
-  progress_log("processing results (--%s)...", o_outputmode);
+  progress_log("processing results (--%s)...", o_outmode_str);
   pid_t processor;
   sysassert( (processor= fork()) != -1 );
 
   if (!processor) {
     sysassert( dup2(fileno(tf),0) ==0 );
-    execlp("./yppsc-commod-processor", "yppsc-commod-processor",
-	   o_outputmode, (char*)0);
-    sysassert(!"execlp commod-processor failed");
+    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.");
 }
 
-void fetch_with_rsync(const char *stem) {
+static void rsync_core(const char *stem, const char *suffix,
+		       const char *zopt) {
   pid_t fetcher;
 
+  progress("Updating dictionary %s...",stem);
+
   sysassert( (fetcher= fork()) != -1 );
   if (!fetcher) {
     const char *rsync= getenv("YPPSC_PCTB_RSYNC");
     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);
-    execlp(rsync, "rsync",
-	   DEBUGP(rsync) ? "-vLt" : "-Lt",
-	   "--",remote,local,(char*)0);
+    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);
+    char *opts= masprintf("-Lt%s%s",
+			  zopt,
+			  DEBUGP(rsync) ? "v" : "");
+    execlp(rsync, "rsync",opts,"--",remote,local,(char*)0);
     sysassert(!"exec rsync failed");
   }
 
-  waitpid_check_exitstatus(fetcher, "dictionary-manager --update");
+  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 *userspecified,
+		       const char *defvalue, const char *defvalue_test,
+		       const char *userspecified,
 		       int enable) {
   const char *value;
   
@@ -138,6 +156,8 @@ static void set_server(const char *envname, const char *defprotocol,
     value= userspecified;
   else if ((value= getenv(envname)))
     ;
+  else if (o_flags & ff_testservers)
+    value= defvalue_test;
   else
     value= defvalue;
 
@@ -153,7 +173,10 @@ static void set_server(const char *envname, const char *defprotocol,
     /* rsync :: protocol specification - anyway, adding scheme:// won't help */
     goto ok;
 
-  value= masprintf("%s%s", defprotocol, value);
+  int vallen= strlen(value);
+
+  value= masprintf("%s%s%s", defprotocol, value,
+		   vallen && value[vallen-1]=='/' ? "" : "/");
 
  ok:
   sysassert(! setenv(envname,value,1) );
@@ -174,25 +197,34 @@ int main(int argc, char **argv) {
   while ((arg=*++argv)) {
     if (IS("--find-window-only"))      o_mode= mode_findwindow;
     else if (IS("--screenshot-only"))  o_mode= mode_screenshot;
+    else if (IS("--show-charset"))     o_mode= mode_showcharset;
     else if (IS("--analyse-only") ||
 	     IS("--same"))             o_mode= mode_analyse;
     else if (IS("--everything"))       o_mode= mode_all;
+    else if (IS("--find-island"))      o_flags |= ffs_printisland;
     else if (IS("--single-page"))      o_flags |= ff_singlepage;
     else if (IS("--quiet"))            o_quiet= 1;
-    else if (IS("--edit-charset"))     o_flags |= ff_editcharset;
+    else if (IS("--edit-charset"))     o_flags |= ff_charset_edit;
+    else if (IS("--no-edit-charset"))  o_flags &= ~(ffm_charset);
+    else if (IS("--test-servers"))     o_flags |= ff_testservers;
     else if (IS("--dict-local-only"))  o_flags &= ~ffs_dict;
     else if (IS("--dict-read-only"))   o_flags &= (~ffs_dict | ff_dict_fetch);
     else if (IS("--dict-anon"))        o_flags &= ~ff_dict_pirate;
     else if (IS("--dict-submit"))      o_flags |= ff_dict_fetch|ff_dict_submit;
-    else if (IS("--upload") ||
-	     IS("--arbitrage") ||
+    else if (IS("--raw-tsv"))          o_outmode_kind= omk_raw;
+    else if (IS("--upload"))           o_outmode_kind= omk_upload;
+    else if (IS("--arbitrage") ||
 	     IS("--tsv") ||
-	     IS("--best-prices"))      o_outputmode= arg+2;
-    else if (IS("--raw-tsv"))          o_outputmode= 0;
-    else if (IS("--screenshot-file"))  o_screenshot_fn= ARGVAL;
+	     IS("--best-prices"))      o_outmode_kind=omk_str,
+				       o_outmode_str=arg+2;
+
+    else if (IS("--screenshot-file")||
+	     IS("--screenshots-file")) o_screenshot_fn= ARGVAL;
     else if (IS("--pctb-server"))         o_serv_pctb=        ARGVAL;
     else if (IS("--dict-submit-server"))  o_serv_dict_submit= ARGVAL;
     else if (IS("--dict-update-server"))  o_serv_dict_fetch=  ARGVAL;
+    else if (IS("--ocean"))            o_ocean=  ARGVAL;
+    else if (IS("--pirate"))           o_pirate= ARGVAL;
 #define DF(f)					\
     else if (IS("-D" #f))			\
       debug_flags |= dbg_##f;
@@ -208,50 +240,98 @@ int main(int argc, char **argv) {
   }
 
   /* Consequential changes to options */
-  
-  if (!strcmp("upload",o_outputmode))
-    o_flags |= ffs_upload;
+
+  if (o_mode & mf_analyse) {
+    if (!o_outmode_kind) {
+      if (o_flags & ff_printisland) {
+	o_outmode_kind= omk_none;
+	o_flags |= ff_singlepage;
+      } else {
+	o_outmode_kind= omk_upload;
+      }
+    }
+
+    if (o_outmode_kind==omk_upload) {
+      o_flags |= ffs_upload;
+      o_outmode_str= "upload";
+    }
+  }
 
   /* Defaults */
   
   set_server("YPPSC_PCTB_PCTB",
-	     "http://",           "pctb.ilk.org",
-	     o_serv_pctb,         o_flags & (ff_needisland|ff_upload));
+	     "http://",          "pctb.ilk.org" /*pctb.crabdance.com*/,
+	                         "pctb.ilk.org",
+	     o_serv_pctb,        o_flags & (ff_needisland|ff_upload));
 	     
   set_server("YPPSC_PCTB_DICT_UPDATE",
-	     "rsync://",          "rsync.pctb.chiark.greenend.org.uk/pctb",
+	     "rsync://",         "rsync.pctb.chiark.greenend.org.uk/pctb",
+	                         "rsync.pctb.chiark.greenend.org.uk/pctb/test",
 	     o_serv_dict_fetch,   o_flags & ff_dict_fetch);
 
   set_server("YPPSC_PCTB_DICT_SUBMIT",
 	     "http://",           "dictup.pctb.chiark.greenend.org.uk",
+	                          "dictup.pctb.chiark.greenend.org.uk/test",
 	     o_serv_dict_submit,  o_flags & ff_dict_submit);
 
   if (!o_screenshot_fn)
-    o_screenshot_fn= masprintf("%s/#pages#.ppm",get_vardir());
+    o_screenshot_fn= masprintf("%s/_pages.ppm",get_vardir());
 
   /* Actually do the work */
 	     
+  canon_colour_prepare();
+  
   if (o_mode & mf_findwindow) {
     screenshot_startup();
     find_yppclient_window();
   }
+  if (!ocean)  ocean=  o_ocean;
+  if (!pirate) pirate= o_pirate;
+  
+  if (o_flags & ff_needisland)
+    if (!ocean)
+      badusage("need --ocean option when not using actual YPP client window"
+	       " (consider supplying --pirate too)");
+  if (ocean)
+    sysassert(! setenv("YPPSC_OCEAN",ocean,1) );
+  if (pirate && (o_flags & ff_dict_pirate))
+    sysassert(! setenv("YPPSC_PIRATE",pirate,1) );
+
+  switch (o_mode & mfm_special) {
+  case 0:                                      break;
+  case mode_showcharset:  ocr_showcharsets();  exit(0);
+  default:                                     abort();
+  }
+
   if (o_mode & mf_screenshot) {
     open_screenshot_file("w");
     if (o_flags & ff_singlepage) take_one_screenshot();
     else take_screenshots();
+    progress_log("OK for you to move the mouse now, and you can"
+		 " use the YPP client again.");
   }
   if (o_mode & mf_readscreenshot) {
+    if ((o_flags & ff_upload) && !(o_flags & ff_testservers))
+      badusage("must not reuse screenshots for upload to live PCTB database");
     open_screenshot_file("r");
     if (o_flags & ff_singlepage) read_one_screenshot();
     else read_screenshots();
   }
   if (o_mode & mf_analyse) {
-    find_islandname();
-    if (o_outputmode)
-      run_analysis();
-    else
-      analyse(stdout);
+    if (o_flags & ff_needisland) {
+      find_islandname();
+      if (o_flags & ff_printisland)
+	printf("%s, %s\n", archipelago, island);
+      sysassert(! setenv("YPPSC_ISLAND",island,1) );
+    }
+    switch (o_outmode_kind) {
+    case omk_upload: case omk_str:   run_analysis();        break;
+    case omk_raw:                    analyse(stdout);       break;
+    case omk_none:                                          break;
+    default: abort();
+    }
   }
+  progress_log("Finished.");
   return 0;
 }
 
@@ -339,7 +419,7 @@ void sysassert_fail(const char *file, int line, const char *what) {
   _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 (;;) {
@@ -354,8 +434,9 @@ 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 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);
   }